<%BANNER%>

The Effect of Pain on Disruptive Behaviors in Nursing Home Residents with Dementia

Permanent Link: http://ufdc.ufl.edu/UFE0043905/00001

Material Information

Title: The Effect of Pain on Disruptive Behaviors in Nursing Home Residents with Dementia
Physical Description: 1 online resource (212 p.)
Language: english
Creator: Ahn, Hyo-Chol
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: behaviors -- dementia -- pain
Nursing -- Dissertations, Academic -- UF
Genre: Nursing Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: There are 1.5 million nursing home (NH) residents in the U.S. More than half of these residents have dementia. Forty to 80 percent of persons with dementia (PWDs) show disruptive behaviors. The health costs of demented Medicare beneficiaries aged 65 and older are three times higher than other Medicare beneficiaries in the same aged group, and about 30% cost of caring PWDs are directly attributed to the management of disruptive behaviors (Alzheimer's Association, 2011; Beeri, Werner, Davidson, & Noy, 2002). Disruptive behaviors are related to injuries, hospitalization, or death among PWDs (Aud, 2004; Beattie, Song, & LaGore, 2005; Doorn, et al., 2003; Kunik, et al., 2010; Matsuoka, Miyamoto, Ito, & Kurita, 2003; Suh, Yeon, Shah, & Lee, 2005). The purpose of this study is to examine the effect of pain on disruptive behaviors (wandering behaviors, aggressive behaviors, and agitated behaviors) among NH residents in Florida using Minimum Data Set (MDS) assessment data. The MDS coordinators who complete the MDS assessment document residents' health related characteristics based on reports from facility nursing staff who are familiar with residents, interviews with residents and family members, and a medical chart review. The unique admission or annual MDS data for non-comatose nursing home residents in Florida during calendar year 2009 (N = 56,577) are used for the analysis. On the average, subjects are 84 years of age (SD = + or - 7 years). The majority of subjects are female (67.7%), widowed (53.7%), high school graduates or less (72.4%), Caucasian (78.7%), and mildly or moderately cognitively impaired (54.8%). Subjects have a mean Charlson Comorbidity Index score of 3 (SD = + or - 2); take a mean of 11 medications (SD = + or - 5); and have a mean MDS-ADL impairment score of 19 (SD = + or - 6). The prevalence of pressure ulcer is 18.1%. The prevalence of disruptive behaviors among Florida NH residents with dementia is 9 - 30% (wandering behaviors: 9%; aggressive behaviors: 24%; and agitated behaviors: 30%), and the prevalence of pain is 37%. Pain is correlated with disruptive behaviors after controlling for the other factors in NH residents with dementia. In order to reduce disruptive behaviors which possibly bring about use of restraints, their underlying causes, such as pain, should be investigated and well managed. However, pain assessment in the cognitively impaired is challenging because they may have difficulties in remembering their pain, interpreting questions and information, or communicating with others (Cohen-Mansfield, 2004; Fisher, et al., 2002; Scherder, et al., 2005). However, the cognitive impaired can demonstrate a number of behaviors indicative of pain (Feldt, 2000; Herr, Coyne, et al., 2006; Warden, Hurley, & Volicer, 2003). Therefore, a comprehensive pain assessment that includes behavioral pain indicators should be developed, and pain should be well controlled to reduce these problematic disruptive behaviors.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Hyo-Chol Ahn.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Horgas, Ann L.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0043905:00001

Permanent Link: http://ufdc.ufl.edu/UFE0043905/00001

Material Information

Title: The Effect of Pain on Disruptive Behaviors in Nursing Home Residents with Dementia
Physical Description: 1 online resource (212 p.)
Language: english
Creator: Ahn, Hyo-Chol
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: behaviors -- dementia -- pain
Nursing -- Dissertations, Academic -- UF
Genre: Nursing Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: There are 1.5 million nursing home (NH) residents in the U.S. More than half of these residents have dementia. Forty to 80 percent of persons with dementia (PWDs) show disruptive behaviors. The health costs of demented Medicare beneficiaries aged 65 and older are three times higher than other Medicare beneficiaries in the same aged group, and about 30% cost of caring PWDs are directly attributed to the management of disruptive behaviors (Alzheimer's Association, 2011; Beeri, Werner, Davidson, & Noy, 2002). Disruptive behaviors are related to injuries, hospitalization, or death among PWDs (Aud, 2004; Beattie, Song, & LaGore, 2005; Doorn, et al., 2003; Kunik, et al., 2010; Matsuoka, Miyamoto, Ito, & Kurita, 2003; Suh, Yeon, Shah, & Lee, 2005). The purpose of this study is to examine the effect of pain on disruptive behaviors (wandering behaviors, aggressive behaviors, and agitated behaviors) among NH residents in Florida using Minimum Data Set (MDS) assessment data. The MDS coordinators who complete the MDS assessment document residents' health related characteristics based on reports from facility nursing staff who are familiar with residents, interviews with residents and family members, and a medical chart review. The unique admission or annual MDS data for non-comatose nursing home residents in Florida during calendar year 2009 (N = 56,577) are used for the analysis. On the average, subjects are 84 years of age (SD = + or - 7 years). The majority of subjects are female (67.7%), widowed (53.7%), high school graduates or less (72.4%), Caucasian (78.7%), and mildly or moderately cognitively impaired (54.8%). Subjects have a mean Charlson Comorbidity Index score of 3 (SD = + or - 2); take a mean of 11 medications (SD = + or - 5); and have a mean MDS-ADL impairment score of 19 (SD = + or - 6). The prevalence of pressure ulcer is 18.1%. The prevalence of disruptive behaviors among Florida NH residents with dementia is 9 - 30% (wandering behaviors: 9%; aggressive behaviors: 24%; and agitated behaviors: 30%), and the prevalence of pain is 37%. Pain is correlated with disruptive behaviors after controlling for the other factors in NH residents with dementia. In order to reduce disruptive behaviors which possibly bring about use of restraints, their underlying causes, such as pain, should be investigated and well managed. However, pain assessment in the cognitively impaired is challenging because they may have difficulties in remembering their pain, interpreting questions and information, or communicating with others (Cohen-Mansfield, 2004; Fisher, et al., 2002; Scherder, et al., 2005). However, the cognitive impaired can demonstrate a number of behaviors indicative of pain (Feldt, 2000; Herr, Coyne, et al., 2006; Warden, Hurley, & Volicer, 2003). Therefore, a comprehensive pain assessment that includes behavioral pain indicators should be developed, and pain should be well controlled to reduce these problematic disruptive behaviors.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Hyo-Chol Ahn.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Horgas, Ann L.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0043905:00001


This item has the following downloads:


Full Text

PAGE 1

1 THE EFFECT OF PAIN ON DISRUPTIVE BEHAVIORS IN NURSING HOME RESIDENTS WITH DEMENTIA By HYOCHOL AHN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 Hyochol Ahn

PAGE 3

3 To my sons: Dasol and Brian Ahn

PAGE 4

4 ACKNOWLEDGMENTS I thank my dissertation committee members (Drs. Horgas, Stechmiller, Neff, and Knapp) for their continued support, patien ce and guidance during this process. They shared their insight and helped me to finish this dissertation. In particular, Dr. Ann Horgas, my supervisory committee chair has given me enormous help in thinking and writing in a scien tific w ay Without her enormous efforts in guiding and assisting me along the way, this dissertation would not have been possible. I also give thanks to my family. Especially, my wife Eunyoung Choi, supported my decision to change the major from electric al and computer engineering to nursing, and endured all the hardship in the long process to earn Ph.D. degree in nursing. Also, presence of my two sons, Dasol and Brian, encouraged me to finish my study. They often sang a Korean song for me when I am tired : Dad, you can do it. We are here. My parents in Korea also help ed me financially and spiritually.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ......................... 10 LIST OF FIGURES ................................ ................................ ................................ ....................... 13 LIST OF ABBREVIATIONS ................................ ................................ ................................ ........ 14 ABSTRA CT ................................ ................................ ................................ ................................ ... 16 CHAPTER 1 BACKGROUND AND SIGNIFICANCE AND THEORETICAL FRAMEWORK ............ 18 Disruptive Behaviors in PWDs ................................ ................................ ............................... 18 Pain in PWDs ................................ ................................ ................................ .......................... 20 The Relationship between Pain and Disruptive Behaviors ................................ ..................... 21 Significance of the Proposed Study ................................ ................................ ........................ 21 Theoretical Framework ................................ ................................ ................................ ........... 23 The Original Theoretical Model ................................ ................................ ...................... 23 The Adapted Theoretical Model ................................ ................................ ...................... 24 Statement of Problem and Specific Aims ................................ ................................ ............... 24 Summary ................................ ................................ ................................ ................................ 28 2 REVIEW OF LITERATURE ................................ ................................ ................................ 32 Nursin g Homes ................................ ................................ ................................ ....................... 32 Nursing Homes in the United States ................................ ................................ ............... 32 Nursing Homes in Florida ................................ ................................ ............................... 32 The Nursing Home Minimum Data Set (MDS) ................................ ................................ ..... 32 Historical Context ................................ ................................ ................................ ............ 33 The Use of the MDS ................................ ................................ ................................ ........ 34 Clinical use ................................ ................................ ................................ ............... 34 Research use ................................ ................................ ................................ ............. 34 Quality Indicato rs Based on the MDS ................................ ................................ ............. 34 Scales Based on the MDS ................................ ................................ ................................ 36 Accuracy of the MDS ................................ ................................ ................................ ...... 37 Psychometric Evidence of the MDS Items Used in This Study ................................ ...... 39 Summary ................................ ................................ ................................ .......................... 40 Characteristics of NH Residents ................................ ................................ ............................. 41 Demographics of NH residents ................................ ................................ ....................... 41 Health Characteristics of NH Residents ................................ ................................ .......... 41 NH Residents with Dementia ................................ ................................ .......................... 42 Types of Disruptive Behaviors ................................ ................................ ............................... 43

PAGE 6

6 Wandering ................................ ................................ ................................ ....................... 43 Aggression ................................ ................................ ................................ ....................... 44 Agitation ................................ ................................ ................................ .......................... 45 Summary ................................ ................................ ................................ .......................... 46 Consequences of Disruptive Behaviors ................................ ................................ .................. 46 Consequences to PWDs ................................ ................................ ................................ ... 46 Injury ................................ ................................ ................................ ........................ 46 Hospitalization and death ................................ ................................ ......................... 47 Quality of life ................................ ................................ ................................ ........... 47 Consequences to NH Staff ................................ ................................ ............................... 47 Mana gements of Disruptive Behaviors ................................ ................................ .................. 48 Physical Restraint ................................ ................................ ................................ ............ 4 8 Pharmacological Intervention ................................ ................................ .......................... 50 Cholinesterase inhibitors ................................ ................................ .......................... 51 An tipsychotics ................................ ................................ ................................ .......... 51 Antidepressants ................................ ................................ ................................ ........ 51 Mood stabilizers ................................ ................................ ................................ ....... 52 Non Pharmacological Intervention ................................ ................................ ................. 53 Patient centered intervention ................................ ................................ .................... 53 Environment centered intervention ................................ ................................ .......... 54 Caregiver centered intervention ................................ ................................ ............... 56 Summary ................................ ................................ ................................ .......................... 57 Pain in Nursing Home Residents ................................ ................................ ............................ 57 Prevalence of Pain ................................ ................................ ................................ ........... 57 Under recognition of Pain ................................ ................................ ............................... 58 Pain Experience ................................ ................................ ................................ ............... 59 Under treatment of Pain ................................ ................................ ................................ .. 60 Consequences of Pain ................................ ................................ ................................ ...... 60 Summary ................................ ................................ ................................ .......................... 61 The Relationship between Pain and Disruptive Behaviors ................................ ..................... 61 Pain and Wandering ................................ ................................ ................................ ........ 61 Pain and Aggression ................................ ................................ ................................ ........ 63 Pain and Agitation ................................ ................................ ................................ ........... 66 Summary ................................ ................................ ................................ .......................... 68 The Relationship bet ween Physiological Factors and Disruptive Behaviors ......................... 69 The Relationship between Background Factors and Disruptive Behaviors ........................... 70 Cognitive Impairments and Disruptive Behaviors ................................ .......................... 70 General Health and Disruptive Behaviors ................................ ................................ ....... 72 Comorbidity and disruptive behaviors ................................ ................................ ..... 72 ADL impairments and disruptive behaviors ................................ ............................ 72 Sociodemographics and Disruptive Behaviors ................................ ................................ 73 Age and disruptive behaviors ................................ ................................ ................... 74 Gender and disruptive behaviors ................................ ................................ .............. 75 Marital status and disruptive behaviors ................................ ................................ .... 76 Education and disruptive behaviors ................................ ................................ ......... 76 Ethnicity and disruptive behaviors ................................ ................................ ........... 76

PAGE 7

7 Summary ................................ ................................ ................................ .......................... 77 The Relationship between Background Factors and Pain ................................ ...................... 77 Cognitive Impairments and Pain ................................ ................................ ..................... 78 General Health and Disruptive Behaviors ................................ ................................ ....... 78 Comorbidity and pain ................................ ................................ ............................... 78 Pressure ulcer and pain ................................ ................................ ............................. 79 Number of medications and pain ................................ ................................ ............. 79 ADL impairments and pain ................................ ................................ ...................... 80 Soci odemographics and Pain ................................ ................................ ........................... 80 Age and pain ................................ ................................ ................................ ............. 80 Gender and pain ................................ ................................ ................................ ....... 81 Marital status and pain ................................ ................................ ............................. 81 Educ ation and pain ................................ ................................ ................................ ... 81 Ethnicity and pain ................................ ................................ ................................ ..... 81 Summary ................................ ................................ ................................ .......................... 82 Summary ................................ ................................ ................................ ................................ 82 3 METHODS ................................ ................................ ................................ ............................. 94 Design ................................ ................................ ................................ ................................ ..... 94 Sample ................................ ................................ ................................ ................................ .... 94 Proposed Measures ................................ ................................ ................................ ................. 95 Pain ................................ ................................ ................................ ................................ .. 95 Secondary Proximal Factors ................................ ................................ ............................ 96 Hunger ................................ ................................ ................................ ...................... 96 Thirst ................................ ................................ ................................ ........................ 96 Elimination ................................ ................................ ................................ ............... 96 Background Factors ................................ ................................ ................................ ......... 97 Neurocognitive factor ................................ ................................ ............................... 97 Comorbidity ................................ ................................ ................................ ............. 98 Pressure ulcer ................................ ................................ ................................ ........... 99 Medication ................................ ................................ ................................ .............. 100 Activities of daily living ................................ ................................ ......................... 100 Sociodemographics ................................ ................................ ................................ 100 Disruptive Behaviors ................................ ................................ ................................ ..... 101 Wandering ................................ ................................ ................................ .............. 101 Aggression ................................ ................................ ................................ .............. 101 Agitation ................................ ................................ ................................ ................. 102 Procedures ................................ ................................ ................................ ............................. 103 Data Analysis ................................ ................................ ................................ ........................ 103 4 RESULTS ................................ ................................ ................................ ............................. 118 Sample Selection Process ................................ ................................ ................................ ..... 118 De scription of the Sample ................................ ................................ ................................ .... 118 Prevalence of Background Factors ................................ ................................ ................ 119 Prevalence of Proximal Factors ................................ ................................ ..................... 119

PAGE 8

8 Main Study Results ................................ ................................ ................................ ............... 119 Aim 1: P revalence of Disruptive Behaviors ................................ ................................ .. 119 Aim 2: The Effect of Pain Severity on the Frequency of Disruptive Behaviors ........... 120 Analysis approach ................................ ................................ ................................ .. 120 Bivariate relationships between background/proximal factors and disruptive behaviors ................................ ................................ ................................ ............. 122 The effect of pain severity on the frequency of wandering behaviors ................... 123 The effect of pain severity on the frequency of aggressive behaviors ................... 124 The effect of pain severity on the frequency of agitated behaviors ....................... 124 The statistical results of the model using imputation methods .............................. 125 Summary of the results of the Aim 2 ................................ ................................ ..... 125 Aim 3: Media ting Effect of Pain Severity on Disruptive Behaviors ............................. 126 Analysis approach ................................ ................................ ................................ .. 126 Bivariate analyses between background factors and pain severity or frequency of disruptive behaviors ................................ ................................ ....................... 128 The mediating effect of pain severity on the relationship between background factors and the frequency of wandering behaviors ................................ ............. 128 The mediating effect of pain severity on the relationship between background factors on the frequency of aggressive behaviors ................................ ............... 134 The mediating effect of pain severity on the relationship between background factors on the frequency of agitated behaviors ................................ ................... 134 Summary of the results of the aim 3 ................................ ................................ ....... 135 Summary of Findings ................................ ................................ ................................ .... 135 Study Results in Unrestrained Residents ................................ ................................ .............. 135 Prevalence of Background Factors in Unrestrained Residents ................................ ...... 136 Prevalence of Proximal Factors in Unrestrained Residents ................................ .......... 136 Prevalence of Disruptive Behaviors in Unrestrained Residents (Aim 1) ...................... 137 The Effect of Pain on Disr uptive Behaviors in Unrestrained Residents (Aim 2) ......... 137 Bivariate analyses between background/proximal factors and the frequency of disruptive behaviors in the unrestrained residents ................................ .............. 138 The effect of pain severity on the frequency of wandering behaviors in t he unrestrained residents ................................ ................................ ......................... 139 The effect of pain severity on the frequency of aggressive behaviors in the unrestrained residents ................................ ................................ ......................... 139 The effect of pain severity on the frequency of agitated behaviors in the unrestrained residents ................................ ................................ ......................... 140 The statistical results of the model using imputation methods .............................. 140 Summary of the results of the aim 2 among the unrestrained residents ................. 140 Mediating Effect of Pain in the Unrestrained Residents (Aim 3) ................................ 141 Bivariate analyses between background factors and pain severity or frequency of disruptive behaviors in the unrestrained residents ................................ ......... 141 Multivariate analyses for mediating effect of pain severity in the unrestrained residents ................................ ................................ ................................ .............. 142 Summary of the results of the aim 3 among the unrestrained residents ................. 142 Summary of Findings among the Unrestrained Residents ................................ ............ 142 5 DISCUSSION AND IMPLICATIONS ................................ ................................ ................ 176

PAGE 9

9 Summary of Results ................................ ................................ ................................ .............. 176 Prevalen ce of Pain in Cognitively Impaired Residents ................................ ........................ 176 Prevalence of Disruptive Behaviors in Cognitively Impaired Residents ............................. 178 The Relationship between Pain Severity and Disruptive Behaviors ................................ .... 178 No Mediating Effect of Pain Severity on the Disruptive Behaviors ................................ .... 180 Restraints Use on NH Residents with Dementia ................................ ................................ .. 181 Limitations ................................ ................................ ................................ ............................ 182 Implications ................................ ................................ ................................ .......................... 184 Implications for Nursing Practice ................................ ................................ .................. 184 Implications for Theory ................................ ................................ ................................ 185 Impl ications for Future Research ................................ ................................ .................. 186 Conclusion ................................ ................................ ................................ ............................ 187 REFERENCE LIST ................................ ................................ ................................ ..................... 189 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 212

PAGE 10

10 LIST OF TABLES Table p a ge 1 1 Conceptual, theoretical, and empirical indications for this study ................................ ...... 31 2 1 Characteristics of NHs in the United States ................................ ................................ ....... 86 2 2 Minimum Data Set items ................................ ................................ ................................ ... 88 2 3 Summary of psychometric properties on published MDS 2.0 items or scales on pain, cognition, and behavior ................................ ................................ ................................ ...... 90 2 4 Literature regarding the relationship between pain and disruptive behaviors in facility dwelling older adults ................................ ................................ ............................. 92 3 1 Concepts, variables, measures, operational definitions and their interpretation .............. 112 3 2 Charlson Comorbidity Index ................................ ................................ ............................ 115 3 3 Functional forms of variables ................................ ................................ .......................... 116 4 1 Sample description: Background factors (N=56,577) ................................ ..................... 145 4 2 Sample description: Proximal factors (N=56,577) ................................ .......................... 146 4 3 Prevalence o f disruptive behaviors among all the subjects (N=56,577) .......................... 147 4 4 Bivariate analysis between background factors, pain, and disruptive behaviors ............. 148 4 5 Bivariate analysis between proximal factors and the frequency of disruptive behaviors ................................ ................................ ................................ .......................... 150 4 6 Ordinal logistic regression for the background/proximal factors and the frequency of wandering behaviors (N = 43,676) ................................ ................................ .................. 151 4 7 Ordinal logistic regression for the background/proximal factors and the frequency of aggressive behaviors (N = 43,671) ................................ ................................ .................. 152 4 8 Ordinal logistic regression for the background/proximal factors and the frequency of agitated behaviors (N = 43,669) ................................ ................................ ....................... 153 4 9 Ordinal logistic regression for the background/proximal factors and the frequency of wandering behaviors after imputing ed ucation (N = 55,709) ................................ .......... 154 4 10 Ordinal logistic regression for the background/proximal factors and the frequency of aggressive beh aviors after imputing education (N = 55,704) ................................ .......... 155

PAGE 11

11 4 11 Ordinal logistic regression for the background/proximal factors and the fre quency of agitated behaviors after imputing education (N = 55,701) ................................ .............. 156 4 12 Mediator effect of pain severity on the relationship between background factors and wandering behaviors ................................ ................................ ................................ ........ 157 4 13 Mediator effect of pain severity on the relationship between background factors and aggressive behaviors ................................ ................................ ................................ ........ 158 4 14 Mediator effect of pain severity on the relationship between background factors and agitated behaviors ................................ ................................ ................................ ............ 159 4 15 Description of restraints use among 56,577 NH residents with dementia ....................... 160 4 16 Description of subsample without restraints: Background factors (N=12,991) .............. 161 4 17 Description of subsample without restraints: Proximal factors (N=12,991) ................... 162 4 18 Prevalence of disruptive behaviors in unrestrained residents (N=12,991) ...................... 163 4 19 Bivariate analysis between background factors, pain, and disruptive behaviors in unrestrained residents ................................ ................................ ................................ ....... 164 4 20 Bivariate analysis between proximal factors and disrupt ive behaviors in unrestrained residents ................................ ................................ ................................ ........................... 166 4 21 Ordinal logistic regression for the background/proximal factors and the frequency of wandering behaviors in unrestrained residents (N = 10,382) ................................ .......... 167 4 22 Ordinal logistic regression for the background and proxim al factors and the frequency of aggressive behaviors in unrestrained residents (N = 10,382) ..................... 168 4 23 Ordinal logistic regression for the background and proximal factors and the frequency of agitated behaviors in unrestrained residents (N = 10,381) ......................... 169 4 24 O rdinal logistic regression for the background/proximal factors and the frequency of wandering behaviors in unrestrained residents after imputing education (N = 12,774) .. 170 4 25 Ordinal logistic regression for the background/proximal factors and the frequency of aggressive behaviors in unrestrained residents after imputing education (N = 12,774) .. 171 4 26 Ordinal logistic regression for the background/proximal factors and the frequency of agitated behaviors in unrestrained residents after imputing education (N = 12,773) ...... 172 4 27 Mediator effect of pain severity on the relationship between background factors and wandering behaviors in unrestrained residents ................................ ................................ 173 4 28 Mediator effect of pain severity on the relationship between background factors and aggressive be haviors in unrestrained residents ................................ ................................ 174

PAGE 12

12 4 29 Mediator effect of pain severity on the relationship between background factors and agi tated behaviors in unrestrained residents ................................ ................................ .... 175

PAGE 13

13 LIST OF FIGURES Figure p a ge 1 1 The Need driven Dementia compromised Behavior (NDB) model ................................ .. 29 1 2 The adapted theoretical model for the current study ................................ ......................... 30 4 1 Sample selection process ................................ ................................ ................................ 144

PAGE 14

14 LIST OF ABBREVIATIONS ADLs Activities of Daily Living ASCII American Standard Code for Information Interchange CHSRA Center for Health Systems Research and Analysis CCI Charlson C omorbidity I ndex CI Confidence Interval CMAI Cohen Man s field Agitation Inventory CMS Centers for Medicare and Medicaid Service CNA Certified Nursing Assistant DNR Do Not Resuscitate DS DAT Discomfort Scale for Dementia of the Alzheimer's Type DV Dependent Variable IV Independent Variable MDB ABS MDS Aggression Behavior Scale MDB CBP MDS Challenging Behavior Profile MDS Minimum Data Set MDS CHESS MDS Change in Health, End stage disease and Signs and Symptoms MDS COGS MDS C ognition S cale MDS CPS MDS Cognitive Performance Scale MDS DBS MDS D iscomfort B ehavior S cale MDS DRS MDS D epression R ating S cale NDB Need driven Dementia compromised Behavior NH Nursing Home NOPPAIN Non

PAGE 15

15 OBRA Omnibus Budget Reconciliation Act OR Odds Ratio PADE Pain Assessment for the Dementing Elderly PPQ Proxy Pain Questionnaire PWD s Persons with Dementia RAI MH Resident Assessment Instrument Mental Health RAP Resident Assessment Protocol RAWS LTC Revised Algase wandering scale Long Term Care version RCT Randomized C ontrolled T rial ResDAC Research Data Assistance Center RMBPC Revised Memory and Behavior Problems Checklist SAGE Systematic Assessment and Geriatric drug use via Epidemiology VDS Verbal Descriptor Scale VIF Variance I nflation F actor

PAGE 16

16 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE EFFECT OF PAIN ON DISRUPTIVE BEHAVIORS IN NURSING HOME RESIDENTS W ITH DEMENTIA By Hyochol Ahn May 2012 Chair: Ann Horgas Major: Nursing Sciences There are 1.5 million nursing home (NH) residents in the U.S. More than half of these residents have dementia. Forty to 8 0 percent of persons with dementia (PWDs) show disruptive behavior s. The health costs of demented Medicare beneficiaries aged 65 and older are three times higher than other Medicare beneficiaries in the same aged group and about 30% cost of caring PWDs are directly attributed to the management of disr uptive behaviors Association, 2011; Beeri, Werner, Davidson, & Noy, 2002) D isruptive behaviors are related to injuries, hospitalization, or death among PWDs (Aud, 2004; Beattie, Song, & LaGore, 20 05; Doorn, et al., 2003; Kunik, et al., 2010; Matsuoka, Miyamoto, Ito, & Kurita, 2003; Suh, Yeon, Shah, & Lee, 2005) The purpose of this study is to examine the effect of pain on disruptive behavior s (wandering behaviors, aggressive behaviors, and agitated behaviors) among NH residents in Florida using Minimum Data Set (MDS) assessment data The MDS coordinators who complete the MDS assessment document based on reports from facility nursing staff who are fa miliar with residents interviews with residents and family members, and a medical chart review.

PAGE 17

17 The unique admission or annual MDS data for non comatose nursing home residents in Florida during calendar year 2009 (N = 56,577) are used for the analysis. O n the average, subjects are 84 years of age (SD = 7 years). The majority of subjects are female (67.7%), widowed (53.7%), high school graduates or less (72.4%), Caucasian (78.7%), and mildly or moderately cognitively impaired (54.8%). Subjects have a mea n Charlson Comorbidity Index score of 3 (SD = 2); take a mean of 11 medications (SD = 5); and have a mean MDS ADL impairment score of 19 (SD = 6). The prevalence of pressure ulcer is 18.1%. The prevalence of disruptive behaviors among Florida NH resi dents with dementia is 9 30% (wandering behaviors: 9%; aggressive behaviors: 24%; and agitated behaviors: 30%), and the prevalence of pain is 37%. Pain is correlated with disruptive behaviors after controlling for the other factors in NH residents with dementia. In order to reduce disruptive behaviors which possibly bring about use of restraints, their underlying causes, such as pain, should be investigated and well managed. However, pain assessment in the cognitively impaired is challenging because they may have difficulties in remembering their pain, interpreting questions and information, or communicating with others (Cohen Mansfield, 2004; Fisher, et al., 2002; Scherder, et al., 2005) However, the cognitive impaired can demonstrate a number of behaviors indicative of pain (Feldt, 2000; Herr, Coyne, et al., 2006; Warden, Hurley, & Volicer, 2003) Therefore, a c omprehensive pain assessment that includes behavioral pain indicators should be developed, and pain should be well controlled to reduce these problematic disruptive behaviors.

PAGE 18

18 CHAPTER 1 BACKGROUND A ND SIGNIFICANCE AND THEORETICAL FRAMEWOR K As of 2009, there are 16 ,000 nursing homes ( NHs ) 1.7 million beds, and 1. 4 million residents in the U nited S tates with approximately 90% of NH residents over 65 years old (Certification and Reporting Database, 2009) In Florida, there are 676 NHs, 81 887 beds, and 7 1, 657 residents (approximately 5% of U.S. NH residents) (Certification and Reporting Database, 2009) Dementia a syndrome of several progressive disorders that erase s memory and alter s affects more than half of NH residen ts (Ballard, O'Brien, et al., 2001; Bula & Wietlisbach, 2009; Magaziner, et al., 2000; N. Wu, Mor, & Roy, 2009) P ersons with dementia ( PWDs ) in NH s ha ve more decline in the ir activities of daily living ( ADL s) than their cognitive ly intact peers (Armstrong, Glenny, Stolee, & Berg, 2010; Carpenter, Hastie, Morris, Fries, & Ankri, 2006) and are dependent on NH staff for assistance with most o f the ir ADLs (McConnell, Pieper, Sloane, & Branch, 2002) The most high cost elders are severely disabled persons who need NH care. The cost of care for PWDs at NHs is three times higher than that of other persons with chronic disease at NHs (Harrow, Tennstedt, & McKinlay, 1995) The majority of PWDs showed disruptive behavior s during the progress of dementia and it significantly affect ed the high healthcare c ost for PWDs due to injuries, hospitalization, or even death associated with disruptive behaviors Disruptive B ehavior s in PWDs The majority of PWDs display disruptive behaviors reaching between 60 % to 90% in some groups (Ballard, O'Brien, et al., 2001; Brodaty, et al., 2001; Matsuoka, et al., 2003; Sloane, et al., 2007; Steinberg, et al., 2003; Testad, Aasland, & Aarsland, 2007; Zuidema, Derksen, Verhey, & Koopmans, 2007)

PAGE 19

19 atic behavior disturbing behaviors or refer to inappropriate, repetitive or dangerous behaviors that are disruptive to the living and wo rking environment in the NH (Allen Burge, Stevens, & Burgio, 1999; Kovach, Noonan, Schlidt, Reynolds, & Wells, 2006; Pieper, et al., 2011; Whall, Gillis, Yankou, Booth, & Beel Bates, 1992) Disruptive behaviors that accompany the cognitive a nd functional decline in dementia syndromes are common distressing to PWDs themselves, and are troublesome to NH staff and other residents (Leonard, Tinetti, Allore, & Drickamer, 2006; Talerico, Evans, & Strumpf, 2002; H. Z. Wu, Low, Xiao, & Brodaty, 2009) Disruptive behavior s cause hospitalization, injuries to PWDs and/or other residents, NH staff burnout, damage to propert y, a nd contribute to the high cost of long term care. Among many disruptive behavior s, three behaviors are most prominent in the current literature: wandering behaviors, aggressive behaviors, and agitated behaviors (Aalten, et al., 2006; Aud, 2004) Wandering is defined as a syndrome of dementia related locomotion behavior having a frequent, repetitive, temporally disordered and/or spatially disoriented nature that is manifested in lapping, random and/ or pacing patterns, some of which are associated with eloping, elop ement attempts or getting lost unless accompanied (Algase, Yao, Beel Bates, & Song, 2007) Wandering occurs in approximately 6 0% of PWDs 2007; Matsuoka, et al., 2003; Sink, Covinsky, Newcomer, & Yaffe, 2004) A ggression is defined as an overt act, involving the delivery of nox ious stimuli to (but not necessarily aimed at) another organism, object or self, which is clearly not accidental, and includes verbally or physically abusive and threatening behaviors (Nsman, Bucht, Eriksson, & Sandman, 1993 ; Patel & Hope, 1992; Ryden, 1988) Aggression occurs in about 42 % to 8 2 % of NH residents with cognitive impairments (Brodaty, et al., 2001; Chen, Borson, & S canlan, 2000; Kunik, et al., 2007;

PAGE 20

20 Schreiner, 2001) Agitation is defined as an unpleasant state of excitement experienced by the persons with dementia having excessive, inappropriate, repetitive, non specific, and observable nature (Kong, 2005) Agitati on occurs in about 40 % to 60% of NH residents with dementia (Ballard, Margallo Lana, et al., 2001; Margallo Lana, et al., 2001; Suh, 2004; Wood, et al., 2000) O ther disruptive behavior s discussed in the literature include repetitive vocalization, sexual disinhibition, delusions, and hallucinations. These behaviors occur less frequently (ranging from 9 % to 36 %) than wandering, aggression, and agitation (Palese, Menegazzo, Baulino, Pistrino, & Papparo tto, 2009; Sink, Holden, & Yaffe, 2005; Volicer, Bass, & Luther, 2007) Pain in PWDs Pain is common among NH residents af fecting 50 % to 80% of NH residents (Achterberg, et al., 2010; Black, et al., 2006; Horgas & Dunn, 2001; Jones, et al., 2006; Smalbrugge, Jongenelis, Pot, Beekman, & Eefsting, 2007; D. Weiner, Peterson, & Keefe, 1999; Zanocchi, et al., 2008; Zwakh alen, Koopmans, Geels, Berger, & Hamers, 2009) There is no empirical evidence that PWDs experience less pain but the prevalence of pain among PWDs is lower than their cognitively intact counterparts (American Geriatric Society, 2002) D etecting pain in PWDs is challen ging due to cognitive and communicative impairments (Horgas, Elliott, & Marsiske, 2009; Kunz, Mylius, Scharmann, Schepelman, & Lautenbacher, 2009; McAuliffe, Nay, O'Donnell, & Fetherstonhaugh, 2009) Pain self report, the gold standar d assessment in cognitively intact persons is questionable in PWDs because dementia impairs their ability to remember, interpret, and respond to pain. Thus, pain is often under reported in PWD, even when there is a probable cause for pain (Horgas, et al., 2009; Horgas & Tsai, 1998) These impairments may make it hard to approp riately articulate or convey their pain intensity (Horgas & Miller, 2008; Snow, et al., 20 09; Tait & Chibnall, 2008)

PAGE 21

21 Consequently, NH residents with dementia are less likely to have pain medication, and, when admi nistered, often receive lower dosages of pain medications than do other older adults (Chibnall & Tait, 2001; Horgas & Tsai, 1998; Morrison & Siu, 2000; Shega, et al., 2007) Sengupta and colleagues (2010) reported that 56% of non Caucasian NH residents with dementia lacked relevant pain treatment compared to 44% of non Caucasian cognitive ly intact NH residents using data from the 2004 National Nursing Home Survey. The Relationship between Pain and Disruptive B ehavior s Some recent studies suggest that pain may contribute to disruptive behavior s in PWDs. Dementia impairs cognitive and communicative abilit ies. Thus PWDs may express pain through disruptive behavior s b ecause they cannot properly verbalize their pain experience (Braun & Kunik, 2004; Desai & Grossberg, 2001; Scherder, et al., 2009; Shega, et al., 2007; Snow, et al., 2009) Cipher a nd Clifford (2004) reported pain effected disruptive behaviors among 234 residents living in eight long term care facilities in Texas. PW Ds gradually lose the ability to think, reason, remember, learn, speak, understand, and process information so that they are less likely to express pain in typical ways even when severe pain is present (Cunningham, McClean, & Kelly, 2010; Horgas & Elliott, 2004; Horgas, et al., 2009; Hor gas & Miller, 2008; Schmidt, et al., 2010) Disruptive behaviors may be one of the ways that PWDs express pain. Does pain severity contribute to disruptive behaviors in NH residents with dementia ? This question serves as the foundation for this study. Significance of the Proposed Study Disruptive behavior s are hard to control (Snowden, Sato, & Roy Byrne, 2003) so their prevention is preferable. The use of physical restraints in NHs has declined since the implementation of the Omnibus Budget Reconci liation Act of 1987 (OBRA ) which established regulatory guidelines for the use of chemical and physical restraints (Siegler, et al.,

PAGE 22

22 1997) ; however r estraints are still often used to manage disruptive behaviors in PWDs (Bourbonniere, Strumpf, Evans, & Maislin, 2003; Cotter, 2005; Evans & Cotter, 20 08; Gallinagha, 2002; Hamers, Gulpers, & Strik, 2004; Mace & Rabins, 2006) Provisions of the Nursing Home Reform Act (1987) which were included in OBRA stipulat ed that were to be imposed only to ensure the physical safety of the resident or that of other residents and only on the written orde (Guttman, Altman, & Karlan, 1999) These restraints have been shown to lead to functional disabilities, mobility problems, cognitive disturbances, multiple falls, incontinence, chronic constipation, pressure ulcers, loss of bone mass and muscle tone, skin abrasions, contractures, cardiac stress, lower extremity edema, disorganized behavior, sensory deprivation, and increased confusion and agitation (Flannery, 2003; Huabin, Lin, & Castle, 2011; W. W. Wang & Moyle, 2005) With person centered care, there is an increased concern about the loss of dignity, loss of autonomy, and suppression of overall wellbeing that is imposed by the use of physical or chemical restraints (Touhy, 2004) The use of restraints to limit the freedom of a PWD poses a major risk to personal dignity in addition to the potential for physical harm Disruptive behavior s can be interpreted as meaningful bodily expressions that describe needs or wishes that require addressing such as pain The use of restraints as a disruptive behavior management mechanism for PWDs violates the respe ct for autonomy. The better approach to manag ing disruptive behavior s is to control the ir possible cause, such as pain, rather than using restraints to control NH residents and thus reduce these behavior s. This study may reveal th at pain plays an important factor in disruptive behavior s If

PAGE 23

23 the pain is associated with disruptive behaviors, it would suggest that pain management might prevent or reduce disruptive behavior s of PWDs Theoretical Framework A theory is an organized, coherent, and syst ematic articulation of a set of statements related to significant questions in a discipline that are communicated in a meaningful whole. It is a symbolic depiction of aspects of reality that are discovered or invented for describing, explaining, predicting or prescribing responses, events, situation s, conditions, or relationships (Meleis, 2007) A theory guides the rationale for research, the selection of indepen dent and dependent variables by explaining the relationship s between them, the hypotheses which are and relevant instruments and suitable study participants, and selection of statistical analyses and interpretation of the results. Without a theoretical framework a researcher will be unable to understand the implications of the findings, and observations will be ended in isolated information. The O riginal T heoretic al M odel The need driven dementia compromised behavior (NDB) model (Algase, Yao, et al., 2007) is adapted to explain disruptive behaviors in relation to pain (Figure 1 1 ). The NDB model is the prevailing theory that explain s disruptive behaviors of PWDs. The NDB model hypothesizes that there are two main constructs that predict disruptive behaviors : background factors and proximal factors. Background factors, consisting of neurocognitive factor s general health, personal characteristics, sociodemographics, are hypothesized to identify PWDs who are likely to show behavioral symptoms under certain conditions Proximal factors, consisting of physiological and psychological needs and social / physical environment, are explained to represent the condition s under which disruptive behaviors occur. The NDB model posits that pain has a direct effect on

PAGE 24

24 disruptive behaviors and that pain mediates the relationship between background f actors (e.g., neurocognitive factor s general health, personal characteristics, and sociodemographics) and disruptive behaviors. Major propositions of the NDB model are: (1) Background factors constitute relatively stable, slowly changing features of PWDs; (2) Proximal factors are more dynamic characteristics of PWDs and of the environment; (3) Proximal factors are perceived in the context of existing background factors; (4) Both proximal and background factors directly affect the NDBs; and (5) Both b ackgro und and proximal factors interact or combine in some sequence to produce NDBs (Algase, Yao, et al., 2007) The A dapted T heoretical M odel The adapted theoretical model is shown in Figure 1 2. T he investigator of this study adapted the NDB model to study the effect of the pain on disruptive behavior s o f NH residents with dementia The adapted model will serve as the theoretical framework for this study. To study the effect of pain on disruptive behaviors the physiological need states among proximal factors will be investigated. All background factors except personal characteristics (e.g., personality and stress response) are included because none of the published studies reported the relationship between personal characteristics as defined in this theory, and pain. Three major disruptive behaviors a re chosen for this study: wandering behaviors, agitat ed behaviors, and aggressive behaviors. Ta ble 1 1 shows each construct, the theoretical definition, and empirical indications of measurement. Statement of Problem and Specific Aims The purpose of this study is to explore the relationship between pain severity and the frequency of disruptive behaviors in NH residents with dementia ( e.g., wandering behaviors, aggressive behaviors, and agitated behaviors). Such information would be of substantial benefit

PAGE 25

25 in understanding the effect of pain severity on disruptive behaviors, and may ide ntify potential new intervention approaches for managing these behaviors. The specific aims and hypotheses are described below: First aim To describe the prevalence of disruptive behaviors in NH residents with dementia. 1 A) To describe the prevalence of w andering behaviors as measured by MDS wandering frequency in NH residents with dementia. 1 B) To describe the prevalence of aggressive behaviors as measured by MDS Aggression Behavior Scale (MDS ABS) in NH residents with dementia. 1 C) To describe the preval ence of agitated behaviors as measured by MDS Challenging Behavior Profile (MDS CBP) agitation subscale in NH residents with dementia. Second aim To investigate the effect of pain severity on the frequency of disruptive behaviors in NH residents with deme ntia after controlling for the other background/proximal factors ( e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, hunger, thirst, bowel incontinence, and bladder incontinence ). 2A) To investigate the effect of pain severity on the frequency of wandering behaviors as measured by MDS wandering frequency in NH residents with dementia, after controlling for the other background/proximal factors 2B) To investig ate the effect of pain severity on the frequency of aggressive behaviors as measured by MDS Aggression Behavior Scale (MDS ABS) in NH residents with dementia, after controlling for the other background/proximal factors 2C) To investigate the effect of pai n severity on the frequency of agitated behaviors as measured by MDS challenging behavior profile (MDS CBP) agitation subscale in NH residents with dementia, after controlling for the other background/proximal factors

PAGE 26

26 Hypothesis for the second aim Among NH residents with dementia, pain severity will be significantly associated with increased frequency of disruptive behaviors, after controlling for the other background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of m edications, ADL impairments, age, gender, marital status, education, ethnicity, hunger, thirst, bowel incontinence, and bladder incontinence). 2A) Among NH residents with dementia, pain severity will be significantly associated with increased frequency of wandering behaviors as measured by MDS wandering frequency, after controlling for the other background/proximal factors 2B) Among NH residents with dementia, pain severity will be significantly associated with increased frequency of aggressive behaviors a s measured by MDS ABS, after controlling for the other background/proximal factors 2C) Among NH residents with dementia, pain severity will be significantly associated with increased frequency of agitated behaviors as measured by MDS CBP agitation subscale, after controlling for the other background/proximal factors Third aim. To evaluate whether pain severity mediates the effect of background factors ( e.g., cognitive impairments, comorbidity, p ressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, and ethnicity ) on the frequency of disruptive behaviors in NH residents with dementia 3A) To evaluate whether pain severity mediates the effect of background fa ctors on the frequency of wandering behaviors as measured by MDS wandering frequency in NH residents with dementia. 3B) To evaluate whether pain severity mediates th e effect of background factors on the frequency of aggressive behaviors as measured by MDS Aggression Behavior Scale (MDS ABS) in NH residents with dementia. 3C ) To evaluate whether pain severity mediates the effect of background factors on the frequency of

PAGE 27

27 agitated behaviors as measured by MDS challenging behavior profile (MDS CBP) agitation su bscale in NH residents with dementia. Hypothesis for the third aim. Among NH residents with dementia, p ain severity will mediate the effect of background f actors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairm ents, age, gender, marital status, education, and ethnicity) on the frequency of disruptive behaviors 3A) Among NH residents with dementia, p ain severity will mediate the effect of background factor on the frequency of wandering behaviors as measured by MDS wandering frequency in NH residents with dementia. 3B) Among NH residents with dementia, p ain severity will mediate the effect of background factors on the frequency of aggressive behaviors as measured by MDS Aggression Behavior Scale ( MDS ABS) in NH residents with dementia. 3C) Among NH residents with dementia, p ain severity will mediate the effect of background factors on the frequency of agitated behaviors as measured by MDS challenging behavior profile (MDS CBP) agitation subscale F ourth aim. To investigate those three aims in the unrestrained residents 4A) To describe the prevalence of disruptive behaviors in the unrestrained residents 4B) To investigate the effect of pain severity on the frequency of disruptive behaviors in the u nrestrained residents after controlling for the other background/proximal factors ( e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, hunger, thirst, bowel i ncontinence, and bladder incontinence ). 4 C) To evaluate whether pain severity mediates the effect of background factors ( e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, educatio n, and ethnicity ) on the frequency of disruptive behaviors in the unrestrained residents

PAGE 28

28 Hypothesis for the fourth aim. 4 A) Among the unrestrained residents disruptive behaviors will be more prevalent. 4 B) Among the unrestrained residents pain severity will be significantly associated with increased frequency of disruptive behaviors, after controlling for the other background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, m arital status, education, ethnicity, hunger, thirst, bowel incontinence, and bladder incontinence). 4C) Among the unstrained residents p ain severity will mediate the effect of background factors (e.g., cognitive impairments, comorbidity, pressure ulcer, n umber of medications, ADL impairments, age, gender, marital status, education, and ethnicity) on the frequency of disruptive behaviors. Summary Pain and disruptive behaviors are common and problematic in NH residents with dementia (Achterberg, et al., 2010; Leonard, et al., 2006; Whall, et al., 1992) Pain assessment in PWDs is challenging because of their cognitive and communicative impairments, and pain should be measured by the comprehensive approach including behavioral pain indicators (Horgas, et al., 2009; McAuliffe, et al., 2009) P ain in persons with cognitive impairments can be expressed via disruptive behaviors such as wandering, aggression, and agitation To reveal the relationship bet ween pain and disruptive behaviors, the NDB model (Algase, et al., 1996; Algase, Yao, et al., 2007) will be adapted and used as the theoretical framework for this study. A theoretical framework is necessary for a study to contribute to scien ce (Meleis, 2007)

PAGE 29

29 Figure 1 1. The Need driven Dementia compromised Behavior (NDB) model Proximal Factors Physiological need states : Pain, hunger, thirst, elimination Psychological need states: positive, negative Social and physical environment: Social interaction, staff mix and ratio, ambiance, crowding, complexity, ambient conditions Background Factors Neu rocognitive factor General health Personal characteristics: personality, stress response Sociodemographics Need driven Dementia compromised Behavior Wandering Aggression Problematic vocalizations Passivity

PAGE 30

30 Figure 1 2. The a dapted theoretical model for the current study Proximal Factors Primary p hysiological need state s : pain Secondary physiological need states: hunger, thirst, elimination Background Factors Neurocognitive factors General health : comorbidity, pressure ulcer, number of medications, ADL impairments S ociodemographics : age, gender, marital status, education race/ethnicity Disruptive B ehavior s W andering Aggression Agitation

PAGE 31

31 Table 1 1 Conceptu al, theoretical, and empirical i ndications for this study Conceptual Theoretical Empirical Indications Pain A distressing sensation in a particular part of the body MDS Pain severity scale Hunger A strong desire or need for food Complaints of Hunger (MDS item # K4B) Thirst A sensation of dryness in the mouth and throat related to a need or desire to drink Dehydration (MDS item # J1C) Elimination Bodily discharges including urine and feces Bow el continence (MDS item # H1A) and Bladder continence (MDS item # H1B) Neurocognitive factor The mental capacity of retaining and reviving facts, events, and impression, or of recalling or recognizing previous experiences MDS Cognitive Performance Scale (MDS CPS) General health comorbidity, pressure ulcer, number of medications, activities of daily living) Charlson Comorbidity Index, pressure ulcer (MDS item # M2), number of medications (MDS item # O1), and MDS ADL Long Form scale (MD S item # G1) Sociodemographics sociological and demographical measure relative to others Age (MDS item # AA3), gender (MDS item # AA2), marital status (MDS item # A5), education (MDS item # AB7), and race/ethnicity (MDS item # AA4) Wandering behaviors A syndrome of dementia related locomotion behavior having a frequent, repetitive, temporally disordered and/or spatially disoriented nature Wandering frequency ( MDS item # E4AA) Aggressive behaviors Physical or verbal harming others that is manifested i n hitting, kicking, and screaming MDS Aggression Behavior Scale (MDS ABS) Agitated behaviors An unpleasant state of excitement experienced by the persons with dementia having excessive, inappropriate, and repetitive nature MDS Challenging Behavior Profile (MDS CBP) agitation subscale MDS: Minimum Data Set

PAGE 32

32 CHAPTER 2 REVIEW OF LITERATURE Nursing Homes Nursing Homes i n t he United States As of 2009, there are 16 ,000 NHs, with 1.7 million beds, and 1.4 million residents (82% occupancy rate) in the U.S. (Certification and Reporting Database, 2009) Approximately 99 % of the se NHs are certified by Medicaid or Medicare, and 99% of NH residents reside in Medicaid or Medicare certified NHs. Approximately 62% of NHs are owned by proprietary o rganizations, 43% have 100 199 beds, and 68% are located in metropolitan areas. The characteristics of NHs in the U.S. are summarized in the Table 2 1 Nursing Homes i n Florida As of 2009, there are 676 NHs, 81,887 beds, and 71, 657 residents (87.5% occ upancy rate) in Florida (Certification an d Reporting Database, 2009) Approximately 71% of the NHs are owned by proprietary organizations, 76 % of the NHs are certified by Medicaid or Medicare (Shaping Long Term Care in America Project, 2007) The Nursing Home Minimum Data Set (MDS) The MDS assessment data is mandatory in all United States NHs certified to participate in Medicare and Medicaid and contains standardized physical, psychological and psycho social assessment data of NH residents. The MDS data includes a 284 item instrument which can be converted to categorical or continuous measures that generally meet the psychometric criteria of reliability and validity (Sgadari, et al., 1997; Shin & Scherer, 2009) The MDS data is a status (Morris, et al., 1990) The major sections and items in the MDS are described in the Table 2 2

PAGE 33

33 A full MDS assessment is completed on admission for every patient and annually thereafter. A shortened assessment is achieved on a quarterly basis. A trained clinical professional, typically a licensed registered nurse, completes the MDS assessment using data from interviews with facility nursing staff, residents and family members, and a medica l chart capacities, monitoring the quality of care provided to NH residents, and research purposes (Ouslander, 1994) Historical Context The first MDS was developed in the U.S. in 1988 in response to the United States which mandates resident assessment to formulate a (Hawes, et al., 1997) The MDS 1.0 was tes ted and modified through consultation with and suggestions from researchers and government regulators. The MDS 1.0 was nationally implemented in 1991, and the revised MDS 2.0 was nationally implemented in 1995. The MDS 2.0 had expanded stipulated standard assessment procedures for resource utilization and quality indicators. The newer version MDS 3.0 was nationally implemented in October 2010. Compared to previous versions, the MDS 3.0 mandated the resident interview on mental status, mood, pain, and health related quality of life and wider use of standardized assessment procedures, in an effort to increase measurement reliability and validity. The for mental status, mood, pain, behavior, and customary routines is increased, rang ing from .92 to .97 (Rahman & Applebaum, in p ress) Also, the MDS assessment time was reduced by 45% to an average of one hour by expanding standard assessment procedures to mental status and health related quality of life (MDS 2.0 did not stipulate the standard assessment procedure of those areas ).

PAGE 34

34 The Use o f the MDS The clinical and research use of MDS assessment data are described below C linical u se The MDS assessment data are used in screening functional status to identify potential problems and strengths of residents, and in monitoring qualit y of care. So far, MDS assessment improved quality of care in NHs by reducing restraints, decreasing dehydration, and increasing physical and cognitive function (Shin & Scherer, 2009) The standardized MDS assessment data are also used in comparing one NH with another and in monitoring the quality of care in NHs by providing the data from which measures of NH quality indicators can be created at minimal cost. All MDS data are conveyed through state public health agencies to a national database maintained by the Centers for Medicare and Medicaid Service s (CMS). All NHs certified by CMS are required to submit MDS data electronically to CMS on a quarterly basis. R esearch u se continuous observations by NH nursing staff, are widely used in behavioral studies of NHs because it prov ides comprehensive information on all United States NH residents. It is a comprehensive data o f over 1.5 million older adults who live in CMS certified NHs. The MDS has also shown acceptable reliability and validity (Frederiksen, Tariot, & De J onghe, 1996; Hawes, et al., 1995; Mor, 2004; Morris, et al., 1990; Sgadari, et al., 1997; Shin & Scherer, 2009; Snowden, et al., 1999; Snowden, et al., 2003; Stineman & Maislin, 2000; Y. Wang, Byers, & Velozo, 2008; Zhang, Paek, & Wan, 2009) Quality Indicators B ased o n the MDS There are quality indicator s based on the MDS, developed by researchers at the Center for Health Systems Research and Analysis (CHSRA) at the University of Wisconsin Madison,

PAGE 35

35 which indicate poor care practices or outcomes. Twenty four quality indicators c overing 11 health care domains were reviewed by national clinical panels involved in NH care and finalized after testing clinical validity and statistical robustness (Hutchinson, et al., 2010) including : Accidents: incidence of new fractures and prevalence of falls Behavioral and emotional patterns: prevalence of behavioral symptoms affecting others, prevalence of symptoms of depression, and prevalence of symptoms of depression without antidepressant therapy Clinical management: use of nine or more different medications Cognitive patterns: incidence of cognitive impairment Elimination and continence: prevalence of b ladder/bowel incontinence, prevalence of occasional bladder/bowel incontinence without a toileting plan, and prevalence of indwelling catheters, prevalence of fecal impaction Infection control: prevalence of urinary tract infections Nutrition and eating: p revalence of weight loss, prevalence of tube feeding, prevalence of dehydration Physical functioning: prevalence of bedfast residents, incidence of decline in late loss ADLs, incidence of decline in range of motion Psychotropic drug use: prevalence of anti psychotic use in the absence of psychotic and related conditions, prevalence of anti anxiety/hypnotic use, prevalence of hypnotic use more than two times in last week Quality of life: prevalence of daily physical restraints, prevalence of little or no acti vity Skin care: prevalence of stage 1 4 pressure ulcers The re is a report that qualify indicators are not valid measures for nursing quality in NHs (Hutchinson, et al., 2010) but t hese quality indicators are used to assist healthcare providers in identifying potential problem areas that would benefit from quality improvement programs and evaluating the effect of t hese programs, to help guide policy makers, and to serve as a source of consumer information (Centers f or Medicare & Medicaid Services, 2010) Many NHs undertake quality improvement programs based on the quality indicators derived from MDS assessment

PAGE 36

36 data (Rantz, et al., 2000) NH administrators and care providers identify and analyze the extent and impact of quality problems, infor m the developers of quality improvement initiatives, track response s to quality initiatives, and benchmark their facility's performance with national averages using q uality indicators (Grando, Rantz, & Maas, 2007; Hutchinson, et al., 2010) The regulatory state inspectors focus on identified quality problems based on quality indicator reports (Mor, et al., 2003) Quality indicators are used as a starting point for the inspection process of NHs. Consum ers or their families can use the MDS public quality data to help select a NH (Edgman Levitan & Cleary, 1996) NHs can be compared using num erous quality indicators based on MDS assessment data. S cales B ased on the MDS There are many MDS based scales that have been developed for research purposes: MDS Pain severity scale (Fries, Simon, Morris, Flodstrom, & Bookstein, 2001) MDS D epression R ating S cale (MDS DRS) (Burrows, Morris, & Simon, 2000) MDS A ggression B ehavior S cale (MDS ABS) (Perlman & Hirdes, 2008) MDS C hallenging B ehavior P rofile (MDS CBP) (Gerritsen, et al., 2008) MDS D iscomfort B ehavior S cale (MDS DBS) (Stevenson, Brown, Dahl, Ward, & Brown, 2006) MDS C ognitive P erformance S cale (MDS CPS) (Bula & Wietlisbach, 2009; Hartmaie r, Sloane, Guess, & Koch, 1994; Hartmaier, et al., 1995; McConnell, et al., 2002; Morris, et al., 1994; Stevenson, et al., 2006) MDS C ognition scale (MDS COGS) (H artmaier, et al., 1994; Zimmerman, et al., 2007) MDS C ognition (Casten, Lawton, Parmelee, & Kleban, 1998; Hawes, et al., 1995; Lawton, et al., 1998; Morris, et al., 1990; Sgadari, et al., 1997) MDS index of social engagement (Achterberg, et al., 2003; Hawes, et al., 1995; Sgadari, et al., 1 997) MDS A ctivities of D aily L iving scale (MDS ADL) (Frederiksen, et al., 1996; Williams, Li, Fries, & Warren, 1997)

PAGE 37

3 7 R esident A ssessment I nstrument M ental H ealth (RAI MH) (Hirdes, et al., 2002) and MDS C hange in H ealth, E nd s tage disease and Signs and S ymptoms (MDS CHESS) (Hirdes, Frijters, & Teare, 2003) These scales have been shown to have moderate to high validity and reliability coefficients and widely used in behavioral studi es for NH residents. Accuracy of the MDS In general, the MDS data have been reported to have acceptable reliability and validity coefficients Rahman and Applebaum (in press) reported that inter rater reliability coefficient ranges from 0.92 to 0.97 for MDS mental stat us, mood, pain, behavioral symptoms, and ADLs, and 0.75 to 0.89 for delirium. Wang and colleagues (2008) reported that Pearson reliability coefficient of MDS physical functioning items was 0.89 and that of MDS cognition items was 0.68 via Rasch analysis using MDS data colle cted by the Department of Veteran Affairs Austin Automation Center. Zhang and colleagues (2009) reported that the majority of clinical measures of MDS data exhibit good reliabili ty coefficients Parmelee and colleagues (2009) reported more than 75% of respondents rated MDS assessment data as accurate or very accurate, and only 2% of respondents rated MDS assessment data as inaccurate or very inaccurate using data from 289 directors of nursing, medical directors, MDS coordinators, nurse mana gers and other clinical management staff at 97 Veterans Affairs Medical Center nursing home care units nationwide. A detailed report for MDS items will be presented in a later section. There have been some concerns that the MDS may underreport or over repo rt some symptoms. Shin and Scherer (2009) stated that vision, health conditions, pain and falls were underreported in the MDS data, and intravenous medication, intake and output, and special therapies are over reported in MDS data. Bharucha and colleagues (2008) reported that MDS significantly underestimated verbally abusive symptoms ( p < .002), phy sically abusive

PAGE 38

38 symptoms ( p = .008), and socially inappropriate behaviors ( p = .016) compared with corresponding items from the Ryden Aggression Scale and the Cohen Mansfield Agitation Inventory (CMAI) for 15 NH residents with severe dementia. Lum and coll eagues (2005) reported that the agreement on ADL assessments between MDS and interview data was low to mo derate (Kappa = .25 to .52) using 3 385 ADL assessment data points from MDS data They stated that the MDS assessment data reported fewer ADL difficulties than interview data obtained from staff proxies and more ADL difficulties than interview data derived from residents. They tested the possible sources of bias using multinomial logit regression, and found that a large portion of the discrepancy in the MDS and interview assessments was accounted for not by the biases but by the confusion of the categorization of ADL items. Chu and colleagues (2004) using data from 895 residents in 33 community bas ed NHs, reported that 63 % of residents who reported daily pain or activity limiting pain on interview did not have daily or moderate to severe pain recorded on the MDS data Cohen Mansfield (2004) using data from 80 NH residents, reported that MDS pain ratings and the geriatricians' pain ratings were significantly correlated for those with mild or moderate cognitive impairment but not for those with severe impairment. Fisher and colleagues (2002) found that the MDS pain assessment and the Proxy Pain Questionnaire (PPQ) are not well correlated (pain frequency: r = .19, p = .18; pain intensity: r =.22, p = .11). Horgas and Margrett (2001) found that MDS assessments on cognition were significantly different from the Revised Mem ory and Behavior Problems Checklist (RMBPC) ( t = 7.3, p < 0.001) using data from 135 NH residents at one NH NH residents with dementia had significantly higher scores on the CPS than memory related behavior subscale of the RMBPC. Bates Jensen and colleagues (2005) reported that there was poor agreement between MDS bed mobility ratings and the research staff performance

PAGE 39

39 07, p = p < .001), and MDS over reported resident dependency in bed mobility using data from 197 long stay residents from 26 California NHs. Random errors and bias by the MDS coordinators who completed the measure not the characteristics of the MDS indicators, seem to be mostly responsible for the inaccuracy of the MDS assessment data (Shin & S cherer, 2009) Hendrix and colleagues (2003) reported that nurse administrators, who may have been trained to complete the MDS but have limited contact with the residents, may have completed the MDS so that MDS depression indicators do not reveal manifestations of depression in NH residents. They also raised concern s that the MDS completed by nurses who do not directly take care of residents may result in MDS data that do health status. Anderson and colleagues (2003) repor ted that nursing staff in many NH facilities do not receive enough training on how to evaluate residents for distressed mood or behavior al symptoms on the MDS. McCurren (2002) raised concerns that MDS coordinators were not knowledgeable about dealing with the psychological and beh avioral data may not accurately reflect those symptoms. Psychometric Evidence of the MDS I tems U sed in T his S tudy Psychometric studies on MDS 2.0 regarding the key elements of this research (e.g., pain, cognition, and disruptive behaviors) are summarized in the Table 2 3 The detailed reliability and validity coefficients for this study items will be described in the Chapter 3. Most studies on validity of th e MDS are reported using criterion validity, which is measured by the correlation between one instrument with other measures already held to be valid and taken as representative of the construct. Criterion validity coefficient ranges from zero to one, with higher values indicating higher criterion validity (Polit & Beck, 2004) The criterion

PAGE 40

40 validity measured by comparing the MDS with validated research instruments of pa in, cognition, ADL s and disruptive behaviors are acceptable, ranging from 0.41 to 0.77. The studies on reliability of the MDS used internal consistency reliability, inter rater reliability, and test retest reliability. Internal consistency reliability refers to the degree to which the subparts of an instrument are all measuring the same attribute or dimension (Polit & Beck, 2004) The most widely used me 0.7 indicates acceptable reliabil ity, and 0.8 or higher indicates good reliability (Polit, 2010) Inter rater reliability refers to the degree to which two raters assign the same ratings for an attribute being measured after independently retest reliability refers to assessment of the stability of an instrument by administering the same measures to a sample on two occasions, and comparing and correlating the scor es obtained on repeated administrations. The reliability of the MDS of pain, cognition, ADLs, and disruptive behaviors are acceptable, ranging from 0.61 to 0.97 (Casten, et al., 1998; Cohen Mansfield, Taylor, McConnell, & Horton, 1999; Fisher, et al., 2002; Fr ies, et al., 2001; Gerritsen, et al., 2008; Hawes, et al., 1995; Lawton, et al., 1998; Perlman & Hirdes, 2008; Sgadari, et al., 1997) Summary There are some concerns with the underreporting of symptoms of residents, mainly due to the variations in how MDS coordinators complete the tool. The M DS assessment data, however, has a number of advantages that counterbalance these limitations. First, researchers have access to large samples representing populations at state and national levels. Second, MDS assessment data have acceptable reliability an d validity for a study of pain, cognition, and disruptive behaviors. Considering the difficulty of consistent caregiving of PWDs in NHs and their

PAGE 41

41 cognitive and communicative inabilities, MDS assessment data based on regular and continuous observation by NH staff is considered a reliable instrument. Third, the use of the MDS for secondary analysis is efficient and economical, and can decrease time and money for research questions that might otherwise be tedious and expensive to investigate. Fourth, the longi tudinal nature of MDS assessment data allows for the examin ation of trends over time. The longitudinal nature of MDS assessment data, collected every three months or more often during a one year ge over time, lagged relationships or longitudinal relationships, and facilitates the use of more powerful statistical analysis techniques to describe both within and between person changes. Thus, this proposed study, examining cross sectional relationshi ps between pain and disruptive behaviors will be an important foundation for future research examining relationships over time. Characteristics of NH Residents Demographics of NH residents Approximately 90% of NH residents in the U.S. are over 65 years old 70% are female, 85% are Caucasian, 60% reside in for profit NH facilities, 70% reside in 100+ bed facilities, 80% reside in metropolitan areas, about half of them are widowed, and the most common residence before NH admission is an acute care hospital (3 5%) (National Center for Health Statistics, 2006) T hirty five out of 1,000 persons whose age is 65 years or more reside in NHs, and 139 out of 1,000 persons whose age is 85 years or over reside in NHs (National Center for Health Statistics, 2009) Health C haracteristics of NH R esidents Health characteristics of NH residents are described using Florida NH residents. The followings are the health characteristics in Florida NH residents: 41% have Do Not Resuscitate (DNR) orders, 44% have no or mild cognitive impairment (MDS CPS score=0 2), 38% have

PAGE 42

42 moderate cognitive impairment (MDS CPS score=3 4), 18% have severe cognitive impairment (MDS CPS score=5 6), 4% are bedfast, 8% walk independently, 54% have bowel incontinence, 59% have bladder incontinence, 18% have fall accidents 52% take antid epressants, and 25% take antipsychotics (Shaping Long Term Care in America Project, 2007) NH R esidents with D eme ntia Dementia caused by various diseases and conditions that result in damaged brain cells or neurons, affects more than one half of NH residents (Ballard, O'Brien, et al., 2001; Bula & Wietlisbach, 2009; Magaziner, et al., 2000; Margallo Lana, et al., 2001; N. Wu, et al., 2009) Commonly, PWDs need help with many activities of daily living, such as bathing, dressing, using the bathroom, or eating, so move into a NH where professional care is available 24 hours a day, seven days a week (Gaugler, et al., 2010; Kopetz, et al., 2000) In 2004, Medicare beneficiaries with dementia were eight times more likely to have a Medicare covered stay in NHs than other Medicare beneficiaries in the same age group (National Center for Health Statistics, 2006) NH residents with dementia incur high healthcare cost s (Harrow, et al., 1995) The cost of care is currently estimate d at $68,964 for a NH residents with dementia compared to $2 4,250 for a community dwelling PWD s Medicare and Medicaid spent much more for PWDs than othe r beneficiaries in the same age group. Medicare payments for PWDs are three times higher than average Medicare payments for other Medicare beneficiaries in the same age group. In 2004, Medicare spent $19,034 for a PWD compared to $6,720 for another non dem ented Medicare beneficiary in the same age group 2011) Medicaid payments for Medicare beneficiarie s aged 65 and older with dementia were more than nine times higher than Medicaid payments for other Medicare beneficiaries in the same age group. In 2004, Medicaid spent an average of $8,419 for a Medicare beneficiary aged

PAGE 43

43 65 and older with dementia compar ed with $915 for other Medicare beneficiaries in the same age group. Total payments for PWDs in 2011 are expected to be $183 billion, including $130 billion for Medicare and Medicaid The NH residents with dementia are older, more cognitively impaired, have greater medical comorbidity, and are more likely to exhibit disruptive behaviors than commu nity dwelling PWDs (Kopetz, et al., 2000) Disruptive behaviors occur in the majority of NH residents with dementia (Ballard, Margallo L ana, et al., 2001; Brodaty, et al., 2001; Margallo Lana, et al., 2001; Schonfeld, et al., 2007; Suh, 2004; Wood, et al., 2000) and o ne of the most t he persistent occurrence of disruptive behaviors and the difficulty of behavioral management (Gaugler, et al., 2010) Types of Disruptive Behaviors Three disruptive behaviors are most prominent in the current literature: wandering behaviors, agitated behaviors, and aggressive behaviors (Aalten, et al., 2006) Less common disruptive behaviors discussed in the literature include repetitive vocalization, sexual disinhibition, delusions, and hallucinations. These occur less frequently (ranging from 9 % to 36%) than wandering, aggression, and agitation (Palese, et al., 2009; Volicer, et al., 2007) PWDs who manifested disruptive behaviors had lower heath related qu ality of life (Cordner, Blass, Rabins, & Black, 2010) Wandering Wandering is defined as related locomotion behavior having a frequent, repetitive, temporally disordered and/or spatially disoriented nature that is manifested in lapping, random and/or pacing patterns, some of which are associated with eloping, elopement attempts or getting lost unless accompanied (Algase, Yao, et al., 2007) is

PAGE 44

44 with effects of dementia (Algase, Yao, et al., 2007; Lai & Arthur, 2003) Wandering has five aspects: persistent walking, repetitive walking, eloping behaviors, spatial disorientation, and negative outcomes (A lgase, Moore, Vandeweerd, & Gavin Dreschnack, 2007) The defining attributes of wandering are being frequent and repetitive, being purposeful, and having predictable pattern such as lapping or pacing (Algase, Son, et al., 2004; Algase, Yao, et al., 2007) Wandering behaviors in PWDs are very common and among the most st ressful patient behaviors for caregivers (Ata, et al., 2010; Hong & Song, 2009) The prevalence or incidence of wandering is 20 % to 67% as not ed in research studies (Lai & Arthur, 2003; Volicer, 2007) The (2007) warned that about 60% of PWDs will have wandering behaviors. Sconfeld and colleagues (2007) reported that the proportion of wandering was 20% among 15,092 residents with moderate or severe cognitive impairment from 134 NH facilities operated by t he Department of Veterans Affairs. Sink and colleagues (2004) wrote that the prevalence of wandering was 67% using data from 5,776 Medicare patients enrolled in the Medicare Alzheimer's Disease Demonstration and Evaluation study at eight sites across the United States between 1 989 and 1991. Holtzer and colleagues (2003) reported that prevalence of wandering was 39% to 57% using data from 236 PWDs in a cohort study with a follow up of five years. Matsuoka and colleagues (2003) reported the prevalence of wandering was 54% in 730 patients at 180 units that have specialized psychiatric beds for acute or long term care of PWDs. Hwang and colleagues (1997) reported that one half of PWDs who were consecutively admitted to the geropsychiatric ward had wandering behaviors. Aggression Aggression i s defined as an overt act, involving the delivery of noxious stimuli to (but not necessarily aimed at) another organism, object or self, which is clearly not accidental, and includes physically or verbally abusive behaviors and threatening behaviors (Nsman, et al.,

PAGE 45

45 1993; Patel & Hope, 1992; Perlman & Hirdes, 2008; Ryden, 1988) Three dimensions of aggression are : a normal/adaptive reaction, a violent reaction, and a functional reaction (Jansen, Dassen, & Moorer, 1997) Aggressive behaviors, commonly directed toward NH staff or caregivers, include verbally and physically abusive behaviors. Physically abusive behaviors include hitting, pushing, kicking, biting, spitting, grabbing, scratching, and throwing objects. Verbally abusive behaviors include disruptive vocalization, threatening, and cursing. Aggression occurs in about 21% to 82% of NH residents with cognitive impairments. Brodaty and colleagues (2001) reported that aggression occurred in 82% of 647 residents in 11 S y dney nursing homes. Kunik and col leagues (2007) reported that 75% of their sample manifested aggression measured by the Ryden Aggression Scale usin g data from 385 PWDs at the Veterans Affairs Medical Center in Houston, Texas. Voyer and colleagues (2005) reported that 21 % of 2,332 older adults in long term care facilities in the Quebec City area displayed physically aggressive behavior an d 21.5% displayed verbally aggressive behavior Schreiner (2001) reported that approximately one half of the sample manifested aggressive behavior during a two week study period involving 391 elderly nursing home residents with dementia in Japan. Chen and colleagues (2000) reported that aggression occurred in 64% of the sample of 125 heterogeneous minority elderly PWDs. Agitation Agitation is defined as an unpleasant state of excitement with an excessive, inappropriate, repetitive, non specific, and observable nature experienced by the PWD s (Kong, 2005) Agitated behaviors include irritability, restlessness, frustration, excessive anger, constant demands for attention and reassurance, repeated phy sical movements or questions. Associated attributes of agitation are inappropriate and excessive motor or vocal activities (Taft, 1989)

PAGE 46

46 Agitation o ccurs in about 48 % to 83% of NH residents with dementia. Suh (2004) reported th at 83% of 257 elderly NH residents with dementia in South Korea manifested agitated behaviors at least once a week Ballad and colleagues (2001) reported that 55% of subjects manifested agitated behaviors among 136 elderly NH residents with dementia. Margallo Lana (2001) repor ted that 48% of subjects exhibited agitation in a study using data from 137 elderly residents with dementia in three nursing homes in No rth East England. Wood and colleagues (2000) reported that agitation occurred in 67 % to 84% of the cases in a study observing 69 NH residents. Summary Disruptive behaviors including wandering, aggression, and agitation, are common in NH residents with dementia. Many PWDs display clinically significant disruptive behaviors during the course of their illness. These behaviors present negative consequences to NH staff and the residents themselves Consequences of Disruptive Behaviors Disruptive behaviors are problematic to both PWDs and NH staff. Disruptive behaviors are associated with injuries, hospitalization or decreased health related qualify of life among PWDs (Aud, 2004; Beattie, et al., 2005; Doorn, et al., 2003; Kunik, et al., 2010) T hey also contribute to staff stress and burnout (Allen Burge, et al., 1999; Norton, Allen, Snow, Hardin, & Burgio, 2010) Consequences to PWDs The consequences of disruptive behaviors to PWDs are described as follows: I njury Wandering PWDs can elope from the NH and get lost, enter physically unsafe areas and get injured or suffer from heat or cold exposure, drowning, or being struck by a car (Beattie, et

PAGE 47

47 al., 2005) NH residents with dementia are 74% more likely to sustain injuries associated with fall accidents relative risk (RR) = 1.74, 95% confidence interval (CI) [1.34, 2.25] (Doorn, et al., 2003) Aud (2004) reported that 62 elopement incidents in 50 NHs in Missouri between J anuary 1999 and June 2001 were recorded at the Missouri Department of Health and Senior Services, and 30% of PWDs who eloped had fallen and were injured. Kunik and colleagues (2010) found that aggressive PWDs showed an increased use of psy chotropic medications (0.2 persons/year to 0.41 persons/year, p < 0.04), and a 10 fold increase in injuries (0.02 persons/year to 0.21 persons/year, p < 0.0001). H ospitalization and death Matsuoka and colleagues (2003) indicated that wandering behaviors were strongly related to hospitalization among 730 inpatients from 180 units with specialized psychiatric beds for acute or long term care of dementia. Suh and colleagues (2005) reported that wandering PWDs were about two times more likely to die RR = 1.89 95% CI [ 1.18 3.02 ] Quality of life Disruptive behaviors are associated with decreased health related quality of life (Q o L) Cordner and colleagues (2010) indicated residents with dementia who had behavior problems had lower Q o L, p = .01, 95% CI [ 11.60, 1.30], among 119 residents at three NHs in Maryland. Wetzels and colleagues (2010) reported that agitation was a strong predictor of poor QoL among 288 individuals with dementia who reside in 14 special care units in 9 NHs. Consequences to NH S taff Disruptive behaviors have important clinical consequences such as i ncreased demands on staff resources and are associated with staff stress or burnout because wandering requires intensive staff supervision due to possible injury from falls or elopement (Kiely, Morris, & Algase, 2000) Physical aggression, including pushing, spitting, grabbing, kicking, and hitting,

PAGE 48

48 are associated with staff burnout and turnover (Allen Burge, et al., 1999) Norton and colleagues (2010) reported that certified nursing assistant burden was significantly associated with disruptive behaviors of NH residents with dementia R 2 = 0.246, p = 0.002. Managements of Disruptive Behaviors There are three appro aches to managing disruptive behaviors: physical restraint, pharmacological intervention, and nonpharmacological intervention. It is generally accepted that physical restraints should be avoided as much as possible and used only as a last resort to protect a PWD s safety. Most experts suggest that nonpharmacological intervention s should be used before pharmacological intervention s and medication should be judiciously used and time limited because of their side effects (Sink, et al., 2005; Tri pathi & Vibha, 2010) Physical R estraint Physical restraints defined as any device or equipment that inhibits mobility or change in position and is not easily removed by the person who is being restrained, are still employed in NHs, despite federal regulations limiting their use (Gallinagha, 2002; Hamers, et al., 2004) Bartels and colleagues (2003) reported that agitated NH residents with dementia had a high prevalence of physical restr aints (29%) compared to uncomplicated NH residents with dementia (19%) in 2,487 physically frail older residents from 109 long term care facilities. Kirkevold and colleagues (2003) stated t hat physical restraints were used on 67% of NH residents in Norway. NH residents with dementia are more likely to be physically restrained than cognitively intact NH residents. Huabin and colleagues (2011) reported that NH residents with dementia were more likely to be physically restrained than those residents without the condition (9.99% versus 3.91%, p < .001) using data from the 2004 National Nursing Home Survey. The application of physical restraints causes much physical harm and violates ethical principles of respect for autonomy and promotion of overall wellbeing.

PAGE 49

49 Physical restraints lead to functional disabilities or aggravated disruptive behaviors (Cotter, 2005) Restrained people often suffer from urinary incontinence and constipation, pressure ulcers, increased dependenc e in activities of daily living, and impaired muscle strength and balance. Other harms associated with restraint use are bruises, contractures, cardiac stress, lower extremity edema, and injuries such as fractures and falls (Gastmans & Milisen, 2006) Psychological harms, such as increased confusion and agitation, are also associated with physical restraints (Flannery, 2003) Capezuti and colleagues (2002) reported that use of side rails did not seem to significantly reduce the likelihood of falls or serious injuries in 463 residents from three nonprofit NHs. Wang and colleagues (2005) reported that there is no scientific evidence that physical restraints actually protect residents against injuries in a critical literature review published between 1992 and 2003 on the use of physical restraints on residents with dementia in long term care. Huabin and colleagues (2011) reported that the us e of trunk restraints was associated with a higher risk for falls (adjusted odds ratio = 1.66, p < .001) and fractures (adjusted odds ration = 2.77, p < .01) using data from the 2004 National Nursing Home Survey. Voyer and colleagues (2005) rep orted that residents with physical restraints were 43% more likely to manifest agitated behaviors (OR = 1.43) using data from 2,332 older adults in long term care facilities in the Quebec City area. With person centered care, there is an increased concern about the loss of dignity and loss of autonomy that is imposed by the use of physical restraints (Touhy, 2004) Even PWDs should be treate d as persons with dignity. Personal dignity cannot be abandoned through illness or disease. This value gives rise to the ethical norm that caregivers must prioritize respect for the dignity of PWDs. The use of physical restraints poses a big risk to person al dignity (Gastmans & Milisen, 2006)

PAGE 50

50 Disruptive behaviors could be interpreted as a kind of bodily autonomy (Algase, Yao, et al., 2007; Cohen Mansfield, 2000) They can b e considered as meaningful bodily expressions (Algase, Yao, et al., 2007) For example, a disruptive behavior might be a request for pain management Disruptive behaviors can be interpreted as a meaningful answer to extreme circumstances from somebody who once was a real person. The use of physical restraints as a disruptive behavior management mechanism for PWDs violates the respect for autonomy. Pharmacological I ntervention Despite federal regulations limiting medication use for behavioral symptoms and their potential side effect, pharmacological interventions using psychoactive medications are commonly used for NH residents with dementia in the setting of acute crisis (Cohen Mansfield & Jensen, 2008; Goldberg, 2002) The psychotropic medications are associated with decreased health related quality of life and increased risk for sedation extrapyramidal symptoms, and fall (Ballard, O'Brien, et al ., 2001; Sink, et al., 2005) T he most common pharmacological interventions are cholinesterase inhibitors, antipsychotics, antidepressants, and mood stabilizers but there is no standard of care (Hersch & Falzgraf, 2007; Sink, et al., 2005) Richter and the colleagues (in pr ess) reported that at least one psychotropic medication was prescribed to about three fourths of NH residents in Austria and more than half of NH residents in Germany to manage disruptive behaviors using data from 5,336 residents in 136 NHs. Tija and c olleagues (2010) reported that 53% residents took at least one psychotropic medication daily a mong 323 residents with advanced dementia from 22 NHs at Boston area. Petek Ster and Cedilnik Gorup (2011) described that 73% residents had a prescription of at least one psychotropic m edication among 2040 residents aged 65 years and older in 12 NHs in Slovenia between September 25 and November 30, 2006.

PAGE 51

51 C holinesterase inhibitors The cholinesterase inhibitors, such as donepezil, rivastigmine and galantamine, are used as a safe measure t o prevent disruptive behaviors. Rodda and colleagues (2009) suggested that cholinesterase inhibitors have limited efficacy, but can be an appropriate pharmacological intervention to manage disruptive behaviors using a systematic review of 14 randomized, placebo controlled trials. A ntipsychotics A ntipsychotics are somewhat effective in managing disruptive behaviors, but produce numerous side effects. Atypical antipsychotic medications ( e.g., clozapine, risperidone, olanzapine, quetiapine) produced fewer side effects compared to typical antipsychotics ( e.g., chlorpromazine, trifluoperazine, thioridazine, thiothixene, haloperidol, loxapine, and perphenazine), but are still associated with the de velopment of extrapyramidal signs, sedation, increased incidence of strokes, and higher mortality rates. Schneider and colleagues (2005) reported that PWDs have a 54% increase in death in antipsychotic medication users ( p = .02). Sink and colleagues (2005) found that atypical antipsychotics especially risperidone and olanzapine are better than other medications to manage disruptive behaviors, but also p ose increased risk of cerebrovascular events. Ballard and colleagues (2006) also found that atypical antipsychotics, particularly risperidone and olanzapine, reduce wandering behaviors, but have significant cardiovascular and motor side effects. A ntidepressants A ntidepressants, such as trazodone fluoxetine, sertraline, paroxetine, fluvoxamine, citalopram, are used to manage disruptive behaviors. However, these antidepressants can result

PAGE 52

52 in sedation, orthostatic hypotension, and gastrointestinal side effects that include nausea, diarrhea, and wei ght loss (Herrmann, 2001) Rojas Fernandez (200 3) suggested trazodone is useful to manage disruptive behaviors in PWDs in a study using 11 PWDs. Finkel and colleagues (2004) rep orted that sertraline is effective to reduce behavioral and psychological symptoms of dementia in a sub group of patients with moderate to severe symptoms of dementia, but diarrhea was significantly ( p < 0.05) more common in the donepezil and sertraline gr oup compared to the donepezil and placebo group using data from 24 patients who were treated with donepezil and sertraline and 120 patients with donepezil and placebo. Pollock and colleagues (2002) reported that patients treated with citalopram manifested significantly greater improvement in agitated or aggressive behaviors than those receiving placebo, but 52% of patients with citalopram dropped out of the study mainly due to lack of efficacy and adverse events, among 85 hospitalized PWDs with at least one moderate to severe aggression, agitation, hostility, suspic iousness, hallucinations, or delusions. Mood stabilizers Mood stabilizers, such as sodium valproate, divalproex sodium, and carbamazepine a re often used to manage disruptive behaviors. These medications are also grouped as anxiolytics or anticonvulsants. These medications can cause sedation nausea or vomiting, diarrhea, and unsteady gait (Herrmann, 2001) Sival and colleagues (2002) reported that t rea tment with sodium valproate (480mg/day) was not effective to control aggressive behavior compared to placebo among 42 PWDs at a psychogeriatric short stay ward at a psychiatric teaching hospital. Porsteinsson and colleagues (2001) showed that divalproex treatment reduced agitation, but not statistically significant ( p = 0.06), and more frequent adverse effects with divalproex than placebo (39% adv erse effects with divalproex versus 11% with placebo, p = .03) among 56 NH residents with dementia who have

PAGE 53

53 agitation. Olin and colleagues (2001) reported that carbamazepine was effect in reducing hostility in 21 PWDs who had not responded to antipsychotics but adverse events (mostly diarrhea and vomiting) were occurred in 44% of patients taking c arbamazepine Non P harmacological I ntervention There are three approaches to nonpharmacological interventions for manag ing disruptive behaviors in PWDs: patient centered intervention, environment centered intervention, and caregiver centered intervention (Remington, Abdallah, Melillo, & Flanagan, 2006) Patient centered intervention s include reminiscence therapy, pet therapy, and stimulated presence therapy. Environment centered interventions include music therapy and environmental modificati ons using landmarks, visual barriers or technology. Caregiver centered intervention s include simplification and structuring, multiple cueing, and reinforcement. These non pharmacological interventions are generally recommended to attempt before initiating drug therapy (Sin k, et al., 2005) P atient centered intervention Patient centered intervention refers to reality orientation intervention which helps PWDs be more aware of themselves in relation to the place and time. Reminiscence therapy, pet therapy, and stimulated p resence therapy are often used as patient centered interventions. Reminiscence therapy using memory aids, such as pictures or diaries is used to improve self esteem in the PWDs. Haight and colleagues (2006) reported that a life review or life storybook inter vention improved communication, positive mood, and cognition in 30 NH residents with dementia in Northern Ireland. Pet therapy is used as an effective approach to manage or reduce disruptive behaviors by providing sensory stimulation and comfort to PWDs. Kanamori and colleagues (2001) showed that pet therapy reduced disruptive behaviors in seven PWDs in a day care center. Libin an d

PAGE 54

54 Cohen Mansfield (2004) s howed that even robotic animal therapy reduced agitated behaviors in nine NH residents with dementia. Takanori (2004) found that a robo t pet offers direct reassurance and calming, the diversion of attention, and behavioral cuing. The robot pet might resist being taken out of the NH by whimpering convincingly when taken outside, or by verbally requesting that the PWD please take it back to NH. The pet could also relay information on to calm and reassure the PWD. Simulated presence therapy using personalized conversation on video or audio tape, can be used to reduce disruptive behaviors in PWDs. Camberg and colleagues (1999) reported that simulated presence consisting of playing a personal telephone conversation about valuable memories through a headset reduced agitation in 54 NH residents with dementia in nine nursing homes in Eastern Massachusetts and Southern New Hampshire. E nvironment centered interve ntion Environment centered intervention refers to alterations in the environment that compensate for impaired cognition and function. Environment centered interventions include music therapy and environmental modifications. Music has been shown to reduce d isruptive behaviors in PWDs (Goodall & Etters, 2005) Clark and colleagues (1998) music reduced aggression in 18 NH residents with dementia. Hick Moore (2005) also reported that reduction in agitated behaviors were achieved during the weeks of music therapy participated b y 13 subjects residing in a s pecial c are u nit over a four week period Sherratt and colleagues (2004) reported that live music significantly reduced disruptive behaviors in 24 PWDs. Svansdottir and Snaedal (2006) found that music therapy significantly reduced disruptive behaviors using data from 38 NH residents with dementia.

PAGE 55

55 Environmental modification, such as the use of landmarks visual barriers or technology, is based on the assumption that disruptive behaviors occur when PWDs experience either excessive or insufficient environmental demand in relation to th eir ability. When environmental demand exceeds ability, individuals experience excess disability, and function as if they are more impaired than they really are. When ability exceeds environmental demand, individuals are not challenged or stimu lated to use preserved abilities and may subsequently lose them at an accelerated rate (Calkins, 2004; Richards & Beck, 2004; Smith, Gerdner, Hall, & Buckwalter, 2004) There have been reports of particular environmental features preventing injury and creating a sense of well being for the resident (Fleming & Purandare, 2010) The general trend of environmental perspective is that the modification of the environment enables PWDs to move about in a safe, non intrusive manner within the structure (Day, Carreon, & Stump, 2000) Baskaya and colleagues (20 04) noted that landmarks, such as a large plant or a distinctive piece of furniture, are widely used in NH facilities to visually hint decision points of a building and differentiate hallways. The presence of landmarks using signs reduces confusion or a gitation level s when PWDs move around inside the NH. However, vision changes associated with aging and dementia may reduce the effectiveness of landmarks (Cronin Golomb, 1995) Roberts (1999) suggested that door visual barriers and taped grid patterns affixed to the floor in front of exit doorways reduces wandering behaviors in PWDs. Feliciano and colleagues (2004) reported that a cloth barrier reduced entry into an unsafe area in one 53 year old woman with moderate mental retardation, bipolar disorder, and probable dementia. Kincaid and Peacock (2003) reported that a wall mural painted over an exit door reduced wandering behaviors in 12 NH residents with dementia. Several other studies have shown that a taped grid pattern on the

PAGE 56

56 floor is effective in preventing wander ing behaviors of PWDs (Chafetz, 1990; Hewawasam, 1996) McShane and colleagues (1988) suggested that exit control and provision of the least restrictive environment using digital technology, such as electronic taggi ng or tracking devices, sensory pads, and alarms, can help prevent wandering. Algase and colleagues (2003) reported that StepWatch a biomechanical activity device that index wandering behaviors in NH residents with dementia was acceptable to NH staff because i t explained the most variance of all electronic devices. The PWDs wear a transmitter shaped like a watch, pager, or ankle bracelet and are located by central monitoring systems in NHs. Tracking devices on PWDs can provide the location of a PWD at any momen t, and report if they have fallen or gotten injured. Location aware systems can be used to prompt behaviors as have been shown by several demonstration identifi wearer behavior contingent on their location by proximity to a scanning device. Such devices might warn a NH staff if a PWD tries to get out of the NH when they touch exit door knobs. There are some concerns that electronic tagging or tracking devices breach autonomy and personal dignity (Robinson, et al., 2007) but these technology solutions are able to prote ct the PWDs from harmful situations while preserving their rights and dignity. C aregiver centered intervention Staff psycho education on how to interact with PWDs reduces disruptive behaviors in PWDs. Simplification and structuring, multiple cueing, and r einforcement are commonly used in NHs (M. F. Weiner & Teri, 2003) Simplification and structuring refers to reducing the number and complexity of demands and organizing tasks in simple ste Multiple cueing refers to using multiple stimuli to initiate or maintain required activities.

PAGE 57

57 Reinforcement involves the process of encouraging positive behaviors. These staff education modules are reported to reduce disrup tive behaviors (Lucero, 2002) Summary There is no definite intervention th at is universally effective in treating disruptive behaviors of PWDs. The management of these disruptive behaviors should use individualized systematic approaches to include patient related factors, caregiver related factors, and environment related factors (Ayalon, Gum, Feliciano, & Arean, 2006; Connor, et al., 2009; Herrmann, 2001; Kutsumi, Ito, Sugiura, Terabe, & Mikami, 2009) as well as using nonpharmacological interventions before resorting to pharmacological intervention s Most pharmacological interventions for disruptive behavior can result in significant side effects, so they should be used cautiously an d monitored by healthcare providers. Nonpharmacological interventions are the mainstay of management of disruptive behaviors in PWDs (Desai & Grossberg, 2001) including NH staff education on non confrontation, simplification, structuring, multiple cueing, repetition, guiding and demonstration, reinforcement, and reducing choices (Kutsumi, et al., 2009; M. F. Weiner & Teri, 2003) Also, physical restraint should be avoided as much as possible due to risks of physical and psychological harms. Pain in Nursing Home Residents Prevalence of P ain Pain is a complex phenomenon derived from sensory stimuli or neurologic injury and modified by individual memory, expectation, and emotions (American Geriatric Society, 2002) Pain is a common condition among NH residents. Using data from 12 3 NH residents with advanced dementia at three NHs in Maryland, Black and colleagues (2006) reported that 63% of these residents recognized pain. Jones and colleagues (2006) reported that 60 % of 2 033

PAGE 58

58 residents from 12 NHs located in both urban and r ural areas in Colorado conveyed pain Zanocchi and colleagues (2008) reported that chronic pain prevalence was 83% in 105 residents at two nursing homes in Italy. Cohen Mansfield (2004) reported that 34 % to 39% of residents suffere d from pain as indicated on MDS assessment data on 80 NH residents. Horgas and Dunn (2001) reported from a sample of 45 NH residents that 49% of them experienced pain in the past week. Achterberg and colleagues (2010) reported that the prevalence of pain varied 32% among 1,959 patients at 31 facilities in Italy, 43% in 2,295 patients at eight facilities in the Netherlands, and 57% in 5,761 patients at 64 facilities in Finland. Smalbrugge and colleagues (2007) stated that pain prevalence was 68% among 350 elderly NH patients from 14 Dutch NHs. Zwakhalen and colleagues (2009) wrote that 47% of NH residents with dementia experienced pain to some extent using data fr om 117 residents at three Dutch nursing home dementia special care units. Teno and colleagues (2004) reported that the prevalence of daily excruciating pain was 3.7% using 2,138,442 MDS assessment data points in the U.S. that were completed around April 1999. The prevalence of daily excruciating pain varied from 1.9 % (New York) to 6.8% (California). Under recognition of P ain Pain is underreported in NH residents with dementia (Cohen Mansfield, 2004; Horgas, et al., 2009; Parmelee, Smith, & Katz, 1993; Sengupta, et al., 2010) Sengupta and colleagues (2010) reported that 17% of NH residents with dementia showed pain while 29% of NH residents without dementia showed signs or symptoms of pain using data from the 2004 National Nursing Home Survey. Horgas and colleagues (2009) wrote that cognitively impaired elderly people self report less pain than cognitively i ntact elderly people, but behavioral pain indicators do not differ between cognitively impaired and intact groups in a sample of 64 cognitively intact and 62 cognitively impaired older adults. Cohen Mansfield (2004) re ported that persons with mild or moderate cognitive impairment were rated by the MDS as having more pain than those

PAGE 59

59 with severe impairment among 80 NH residents, and suggested this is because persons with severe cognitive impairment have a decreased abilit y to adequately communicate the nature and extent of the pain. Using data from 758 elderly institution residents, Parmelee and colleagues (1993) reported that PWDs were less likely to report pain, even though examination of possible physical causes of reported pain revealed no differences between pain reports of cognitively impaired versus intact individuals. In summary, pain assessment using self report instruments is not enough for NH residents with dementia. When using self report instruments to evaluate pain in NH residents, it is important to be aware of the presence of cognitive or communicative impairmen ts that may respond to pain is altered so that they are often unable to properly articulate or convey their pain intensity. More comprehensive pain assessment is needed when assessing pain in PWDs, and should include observational measures in addition to self reported pain scores. Pain E xperience PWDs do not experience less pain, but rather experience pain similar to cognitively intact individuals or even incr eased pain affect in some types of dementia (Gibson, Vou kelatos, Ames, Flicker, & Helme, 2001; Scherder, et al., 2003) Gibson and colleagues (2001) wrote that the subjective rating of evoked pain and the level of post stimulus cortical activation fol lowing noxious stimulation in PWDs were similar to those of cognitively intact older adults using data from 15 PWDs and 15 c ognitively intact older adults Scherder and colleagues (2003) reported that persons with vascular dement ia had increased pain as measured by the c olo red a nalogue s cale for pain affect and the f aces p ain s cale compared to cognitively intact persons using data from 20 patients with possible vascular dementia and 20 cognitively intact older adults who had compa rable pain conditions.

PAGE 60

60 Under treatment of P ain There is ample evidence that NH residents with dementia are inadequately treated for pain. Sengupta and colleagues (2010) reported that 56% of non Caucasian NH residents with dementia lacked relevant pain treatment compar ed to 44% of non Caucasian cognitively intact NH residents as reflected in data from the 2004 National Nursing Home Survey. Reynolds and colleagues (2008) also found that 56% of residents with severe impairment received pain medications, while 80% of residents with no cognitive impairment received pain medications ( p < 0.001) using data from 551 residents in six North Carolina NHs. Morrison and Siu (2000) reported that advanced dementia patients who had hip fractures received significantly less amount s of opioid analgesia compared to thei r cognitively intact counterparts using data from 59 cognitively intact elderly patients and 38 advanced dementia patients with hip fracture. Won and colleagues (1999) stated that NH residents with cognitive impairments were 44% more likely not to receive pain medications using data from 49, 971 NH residents OR = 1.44, 95% CI [1.29, 1.61] Horgas and Tsai (1998) found that NH residents with cognitive impairments were prescribed and administered significantly less analgesic medication, both in number and in dosage of pain drugs than their cognitively intact peers as seen in 339 NH residents from four nursing homes in Western Pennsylvania. Consequences of P ain Pai n management is closely related to health related quality of life in PWDs. Cordner and colleagues (2010) revealed that residents who received pain medication had a highe r health related quality of life using data from 119 NH residents with dementia from three NHs in Maryland p = .006, 95% CI [3.30 19.59] Adequate pain assessment and effective treatment using observational strategies improves health related quality of li fe in PWDs, reflecting an improvement in feelings and mood, enjoyment of activities, and social interaction. Won and

PAGE 61

61 colleagues (2006) found that long acting opioids improved functional status compared with short acting opioids without adverse events (i.e., constipation, delirium, dehydration, pneumonia, changes in cognition or mood statu s, increased risk of depression) in a longitudinal study of 10,372 NH residents with persistent pain adjusted hazard ratio = 1.85, 95% CI [1.05, 3.23] Lee and colleagues (2006) reported that residents who had more pain were more likely to have distressed moods and consequen tly a higher incidence of terminal restlessness based on 84 Summary P ain prevalence is high among NH residents with or without dementia ranging from 50 % to 80% However, NH residents with dementia have cognitive and communi cative impairments as symptom s of dementia thereby hampering effectively verbalization of their pain intensity. Thus, they did not have adequate pain treatment causing a decline in feelings and mood, functional status, enjoyment of activities, and social interaction. The Relationship between Pain and Disruptive Behaviors T he relationship between pain and disruptive behaviors has been examined in several published studies. The se studies are summarized in Table 2 4 and discussed in the following section Pain and W andering PWDs are more likely to wander due to many factors such as the need to address pain and other emotions. In a retrospective cohort study of 8,982 NH residents from Mississippi, Texas, and Vermont who had baseline and three month follow up MDS assessment data between January 1996 and December 1997, Kiely and colleagues (2000) demonstrated that NH residents who expressed pain were 65% more likely to develop wandering behaviors than their counterparts who did not using a logistic regression analysis, after controlling for pneumonia,

PAGE 62

62 short term memory, repetitive questions, long term memory, constipation, antipsychotic medication, ADL impairment, an d gender OR = 1.65, p = .02, 95% CI [1.08, 2.53]. The authors dichotomized the MDS wandering item as present (occurred at least once in the last 7 days) or absent (not exhibited in the last 7 days) and used this variables in a logistic regression analysis The researchers hypothesized that a resident with pain may move about the NH in an effort to relieve his or her pain or unsettled state, but such motivation may be unknown to observers and movement may be perceived as aimless wandering. This view is cons istent with the NDB model which hypothesizes that physiological and psychological need states may influence wandering behavior. T here are some limitations in th is study First, because Kiley and colleagues (2000) used the MDS 1.0, they were not able to measure pain appropriately. They measured pain by a dichotomized expression of sadness or pain (0 = no sad or pained expression 1 = sad or pained expression ) using MDS 1.0. The confounded expression of pain with expression of sadness may not be a valid measurement of pain. Sadness is not the same concept as pain. Second, the researchers recruited all NH residents from three states. Not all NH residents expressed pain via wandering. Wandering occurred mainly in persons with cognitively impairments, but not in all the older adult s or all NH residents (Algase, et al., 1996; Algase, Yao, et al., 2007) There is no reporting that wandering, as defined in the NDB model, is occurred in cognitively intact persons Thus, by including all NH residents and reporting the relationshi p between pain and wandering without controlling for their cogniti ve status the relationship between pain and wandering in persons with cognitive impairments may be underestimated

PAGE 63

63 Third, the researcher did not exclude or statistically control the use of physical or chemical restraint. Persons with these restraints cannot wander. Those residents are restrain ed should be appropriately controlled or excluded from study sample. Furthermore, this study may have present ed more meaningful results if an ordinal logistic regression was used instead of a binary logistic regression for multivariate analysis. Collapsing the ordinal wandering outcome into two categories and performing a binary logistic regression result s in the loss of valuable information. It is im portant to use ordinary logistic regression since the response categories of wandering have a 4 point ordered nature (no exhibition, 1 3 day exhibitions in last 7 days, 4 6 day exhibition in the last 7 days, and daily exhibition). By using an ordinal lo gistic regression analysis, it would be possible to demonstrate how the results change as the dependent variable moves to the next higher category. Pain and A ggression The relationship between pain and aggression is inconclusive as seen in two published st udies. In the first study, Volicer and colleagues (2009) conducted a study of pain and aggression in Dutch nursing home residents. The sample consisted of 929 residents older than 65 years old with dementia who were dependent in decision making and not comatose, and were collected from eight Dutch nursing homes and 10 residential homes within a 12 month time window for each facility separately, resulting in a time range from April 4, 2007, to December 1, 2008. The authors measured pain by a single MDS item (e.g., pain frequency) p hysical or sexually abused were threatened, screamed at, or cursed at o aggression versus one or more episodes for a logistic regression analysis. The authors reported that pain frequency was significantly correlated with frequency of verbal aggression using a bivariate correlation analysis

PAGE 64

64 ( r = .07, p = .028 ) However, in a logistic regression analysis, after controlling for hallucinations, delusions, ability to understand others, depression, and constipation pain frequency was not related to the frequency of verbal aggression ( OR = .73, p = .116 ) or the frequency of physi cal aggression ( OR = 0.69, p = .161 ) In the second study, Leonard and colleagues (2006) conducted a study of pain and ag gression in U.S. nursing home residents. The sample consisted of 103,344 NH residents in five large geographically diverse states (e.g., California, New York, Ohio, Pennsylvania, and Texas) who had at least one annual MDS assessment data completed during 2 002. Similar to the aforementioned study, the authors measured the frequency of aggression by a single MDS item, and dichotomized it as no aggression versus one or more episodes for a logistic regression analysis. The authors measured pain by a single MDS item (e.g., pain intensity), and reported that pain intensity was not significantly associated with the frequency of physical aggression in a binary logistic regression analyses, after controlling for depression, delusions, hallucinations, constipation, re spiratory tract infection, urinary tract infection, fever, and the amount of time participating in non ADL or treatment activities such as recreational activities. These two published studies used MDS assessment and large samples, but they have some limit ations. First, both of these studies did not appropriately examine the relationship between pain and aggression in NH residents with dementia. Leonard and colleagues (2006) measured the relationship between pain and aggression in all NH residents regardless of their cognitive status, in five states. Not all NH residents express pain via aggressive behaviors. The cognitively intact persons residing in these NHs are likely to verbalize their pain level or request pain medications when they felt pain. NH residents with dementia have cogn itive and communicative impairments that can cause them to express pain via aggressive behaviors (Algase, et al., 1996; Algase, Yao,

PAGE 65

65 et al., 2007) If the researchers excluded or statistically controlled for NH residents with out dementia, they may have found a relationship between pain and aggression. Volicer and colleagues (2009) measured pain frequency alone, and not pain frequency together with pain intensity The comprehensive measurements of pain frequency and intensity in the MDS assessment data would provide more meaningful outcomes. Single item measurement measuring global pain rating is acceptable, but using one item (e.g., pain frequency) from a compreh ensive pain scale is not a meaningful measurement (Youngblut & Casper, 1993) Second, these two studies used s ingle MDS items for measuring the frequency of aggressive behavior. In aggression, there are at least two components, verbal and physical aggressions. Using only one item from a comprehensive aggression scale is questionable because it does not examine the full range of aggressive behavior. T he use of the comprehensive MDS Aggression Behavior Scale (MDS ABS) (Perlman & Hirdes, 2008) put forward recently in 2008, helps to present more meaningful outcomes. T he MDS ABS is a sum score of four MDS items, and has a continuous format, ranging from zero to 12 with higher scores indicating greater severity of aggressive behaviors. The comprehensive MDS ABS has better psychometric evidence as compared to single MDS items an internal consistency reliability of 0.79 to 0.95, and a criterion validity coefficient of 0.72 with CMAI aggression subscale scores (Perlman & Hirdes, 2008) Third, in t hese two studies the researchers did not exclude or statistically control the use of physical or chemical restraint. Persons with these restraints cannot appropriately express aggression. Those residents who put restraints should be appropriately controll ed for reporting the meaningful outcomes.

PAGE 66

66 Fourth these two studies used logistic regression analyses after dichotomizing a single aggression item. These studies could have presented more accurate results using an ordinal logistic regression, rather than b inary logistic regression for multivariate analysis. Collapsing a single MDS aggression response into two categories and performing a logistic regression result s in the loss of valuable information. Pain and A gitation Several studies have identified pain as one of the most important contributing factors to agitation in NH residents with dementia. Buffum and colleagues (2001) reported tha t pain measured by a discomfort scale was significantly associated with the frequency of agitation as measured by the CMAI ( r = 50 p = .0 03 ) This study was conducted with 33 Veterans Affairs NH residents with dementia in an urban, university affiliated, 120 bed extended care facility within the Veterans Affairs Medical Center. Manfredi and colleagues (2003) demonstrated that low dose, long acting opioid treatment reduced the frequency of agitation as measured by the CMAI using a prospective longitudinal design mean ch ange in CMAI score: 6.4 ( 95% CI [ 10.96, 1.8] ) The sample consisted of 13 NH residents with dementia who were more than 85 years old. The authors administered a placebo for four weeks and a long acting opioid for the next four weeks. No new psychotropic medications were prescribed during the eight week study period, but necessary psychotropic medications which residents were taking when they entered the study were permitted during the study and their use was recorded. Villanueva (2003) reported that pain severity was significantly associated with the frequency of agitation in 40 residents from four long term care facilities, three skilled nursing facilities, and a locked dementia assisted living facility. Pain severity was measured by the P ain A ssessment for the D ementing E lderly (PADE), and the frequency of agitation was assessed by

PAGE 67

67 the CMAI. The PADE consists of 24 items and is composed of Part I (physical assessment; observable facial expression, breathing pattern, and posture), Part II (global assessment; allowing the caregiver to rate overall pain of the resident they are caring for), and Part III (functional assessment; ADLs). Part I in PADE (physical assessment; observable facial expression, breathing pattern, and posture) correlates significantly with the CMAI verbal agitation subscale ( r = .296, p < 0.01 ) and Part III in PADE (functional assessment; ADLs) is significantly associated with physically aggressive, verbal and physical agitation subscales of the CMAI ( r = .396, r = .398, r = .421, respectively ) Bartels and colleagues (2003) reported that the prevalence of pain was significantly associated with mixed agitation and depression in 2,487 physically frail older residents with dementia from 109 long term care facilities. Agitation was assessed by documenta tion of signs of agitated behaviors in medical record, and categorized 3 levels, with no agitation moderate agitation, and severe agitation. Pain was dichotomized based on medical chart review. Thirty four percent of PWDs with mixed agitation and depressi on exhibited pain, whereas only 23% of uncomplicated PWDs exhibited pain ( chi square = 36.20, p < .0001 ) Norton and colleagues (2010) used secondary analys is of prospective data in 169 residents from 10 nursing homes in Birmingham, Alabama, and reported mixed findings with regard to the relationship of pain and disruptive behaviors. A multiple regression analysis revealed that weekly resident pain intensity reported by certified nursing assistants (CNAs) were significantly associated with the frequency of disruptive behaviors as measured by the Revised Memory and Behavior Problem Checklist Nursing Home (RMBPC NH) after controlling for cognitive impairments, g ender, ethnicity, and ADL functional status = 0.183, p = .04 ) However, t he number of painful diagnoses and analgesic medication administration seen in the chart review

PAGE 68

68 were not significantly associated with disruptive behaviors as measured by subscales within the RMBPC NH The authors could not control data because it is a secondary analysis and there may be s ome variability due to different styles and skills among raters. In summary, these published studies showed a significant relationship between pain and agitation, but there are some gaps or limitations in the literature. First, the two published studies (Buffum, et al., 2001; Manfredi, et al., 2003) on the relationship between pain and agitation in the NH residents with dementia used small sample sizes, ranging from 13 to 33. Generalizability of findings may be limited by the small sample size. Also, in these two published studies, the researchers did not control for the use of restraints. Residents with res traints will manifest agitation different ly. Second, the other three published studies (Bartels, et al., 2003; Norton, et al., 2010; Villanueva, et al., 2003) included all NH residents and did not report the relationship between pain and agitation after controlling for cognitive status. Not all NH residents express pain via agitated behaviors; agitated behaviors are more common among NH residents with dementia. Furthermore, none of the studies used the MDS Challenging Behavior Profile (MDS CBP) agitation subscale or large scale data. The MDS CBP agitation subscale is a sum score of four MDS items, ranging from 0 to 7, with higher scores indicating more agitated behaviors. The MDS CBP agitation subscale has an internal consistency reliability of 0.70, inter rater reliability rank correlation coefficient of 0.50 with b ehavior rating s cale for psychogeriatric i npatients (Gerritsen, et al., 2008) Using the MDS CBP agitation subscale in a large sample size may provide more meaningful findings with increased generalizability. Summary There are only two published studies on the relationship between pain and disruptive behaviors in U.S. Nursing Home residents with dementia (Buffum, et al., 2001; Manfredi, et al.,

PAGE 69

69 2003) The other studies used all N H r esidents (not NH residents with dementia) or NH residents in other countries. In these two studies, the sample sizes were very small, ranging from 13 to 33, thereby limiting their gener alizability of findings. Most of the published studies on the relationship between pain and disruptive behaviors recruited NH residents, but did not focus on only NH residents with dementia or report their findings after controlling for the cognitive statu s of the participants. Not all NH residents express pain through disruptive behaviors such as wandering, aggression, and agitation. PWDs have cognitive and communicative impairments that can cause their unmet needs, such as pain management, to be expressed through disruptive behaviors (Algase, et al., 1996; Algase, Yao, et al., 2007) Most of the studies investigating the relationship between pain and wandering or aggression used a binary logistic regression for multivariate analysis, but wandering and aggression are measured by ordinal variables. Collapsing the ordinal wandering outcome or ordinal aggression outcome into two categories and performing a logistic regression may result in the loss of valuable information. Those published studies regarding the relationship between pain and wandering or aggression might present more meaningful results if an ordinal logistic regression, rathe r than logistic regression for multivariate analysis, was used. The Relationship between Physiological Factors and Disruptive Behaviors The study of the relationship between pain and disruptive behavior in NH residents with dementia can be complicated by multiple variables that potentially affect disruptive behaviors. The following is a brief review supporting the selection of physiological needs factors. The secondary physiological ne ed states which directly affect disruptive behaviors include hunger, thirst, and elimination (Algase, Yao, et al., 2007) There are a few studies on revealing the relat ionship between physiological need states. Burgio and colleagues (2001) reported that hunger would be a significant factor to disruptive behaviors T here is a n increase in disruptive

PAGE 70

70 vocalization arou nd the lunch time using data from 68 elderly NH residents. Nelson (1995) reported that disruptive behaviors was often stopped after providing l iquid from qualitative study. Sloane and colleagues (1999) reported that residents with urinary incontinence were two times likely to develop disruptive vocalization using 203 residents from 107 skilled nursing facilities. The Relationship between Background Factors and Disruptive Behaviors B ased on the NDB model (Algase, et al., 1996; Algase, Yao, et al., 2007) background factors are directly assoc iated with disruptive b ehaviors The following is a brief review of the relationship between b ackground factors and disruptive behaviors, including cognitive impairments general health (e.g., comorbidity, pressure ulcers, number of medication s ADL impairment s), and sociodemographics (e.g., age, gender, marital status, education, and ethnicity). Cognitive Impairments and Disruptive Behaviors Cognitive impairment positively affects disruptive behaviors. PWDs who are more cognitively impaired are more likely to exhibit higher levels of disruptive behaviors (Burgio, Park, Hardin & Sun, 2007; Holtzer, et al., 2003; Matsuoka, et al., 2003; Menon, et al., 2001; Norton, et al., 2010; Schonfeld, et al., 2007; Song & Algase, 2008; Vance, et al., 2003; Voyer, et al., 2005) Norton and colleagues (2010) reported that residents who were more cognitively impaired showing higher levels of disruptive behaviors measured by Revised Memory and Behavior Problem Checklist Nursing Home (RMBPC NH) using 169 residents from 10 nursing homes in Birmingham, Alabama. Sconfeld and colleagues (2007) reported that severe cognitively impaired NH residents were two tim es more likely to wandering (OR=2.33, 95% CI= 1.95 2.78) compared to NH residents with moderate cognitive impairments using 15,092 residents with moderate or severe cognitive impairment from 134 NH facilities operated by the Department of Veterans Affairs. Burgio and colleagues (2007) reported that severely cognitively

PAGE 71

71 impaired residents manifested more agitation than the moderately impaired group using 78 residents from 5 NHs. Voyer and colleagues (2005) reported t hat residents with mild moderate and severe cognitive impairment are more likely to display aggressive behaviors (OR = 2.87 and 3.77, respectively) using data from 2,332 older adults in long term care facilities in the Quebec City area. Vance and colleague s (2003) reported that more cognitive impaired residents exhibited more agitatio n using data from 123 residents from five nursing homes located in the Pittsburgh area. Holtzer and colleagues (2003) reported that wandering or agitation increased as a decrement in cognitive status using 236 PWDs in cohort study with follow up of 5 years. Matsuoka and colleagues (2003) reported the low cognition ability measured by low Mini Mental Status Examination score was related to aggression am ong 730 patients from 180 units that have specialized psychiatric beds for acute or long term care of PWDs. Song and colleagues (2008) reported that cognitive impairment measured by lower Mini Mental Status Examination score was significantly correlate d with five subscales of wandering, such as s patial disorientation, attention shift, negative outcomes, persistent walking, and specific patterns using 160 PWDs residing in 14 long term care facilities in South Korea. Menon and colleagues (2001) reported that aggression were more prevalent among those with greater severity of cognitive impairment ( p < 0.05) using 1101 residents with dementia from 59 NHs across Maryland. Beck and colleagues (1998) reported that MMSE score negatively correlated with disruptive behaviors using 45 nursing home residents with dementia from 3 NHs in Arkansas and Maryland. Sloane and colleagues (1998) reported that more severe cognit ive impairment were related with increased agitation levels using 3723 observations of resident behaviors in 53 Alzheimer's disease Special Care Units in NHs in Kansas, Maine, Mississippi, and South Dakota. Algase (1992) reported using 163 NH residents that wanderers had poorer performance than non

PAGE 72

72 wanderers on neurocognitive dimensions, includ ing memory, attention, language skills, visual spatial skills. General Health and Disruptive Behaviors General health, defined as o includ es comorbidity pressure ulcer, number of medications, and ADL impairments (Algase, Yao, et al., 2007) Comorbidity is negatively associated with disruptive behaviors, and ADL impairments showed mixed relationship with disruptive behaviors Th e relationship s between pressure ulcer s or number of medications and disruptive behaviors are not reported. The details are discussed below. Comorbidity and disruptive behaviors Studies reported a negative relationship between comorbidity and disruptive be haviors. Cohen Mansfield and Libin (2005) reported that number of medical diagnoses correlated negatively with physically agitated behavior ( r = 0.219, p = 0.002) among 175 elderly PWDs recruited from 11 nursing home facilities in Maryland. Beck and colleagues (1998) reported that the number of diagnoses correlated negatively with aggressive behaviors using 45 nursing home residents with dementia from 3 NHs in Arkansas and Maryland. ADL impair ments and disruptive behaviors Most of studies reported activities of daily living (ADL) impairment or dependence is positively associated with disruptive behaviors, but some studies reported the opposite relationship. Sconfeld and colleagues (2007) reported that NH residents who were dependent on ADL were two times more likely to wandering (OR = 2.36, 95% CI [ 1.78 3.14 ] ) compared to NH residents requiring no assistance among 15,092 residents with moderate or severe cognitive impairment from 134 N H facilities operated by the Department of Veterans Affairs. Cohen Mansfield and Libin (2005) reported that ADL impairment also positively correlated with verbal

PAGE 73

73 agitation ( r = 0.235, p = 0.001) among 175 elderly PWDs recruited from 11 nursing home facilities in Maryland. Menon and colleagues (2001) reported that physical aggressive behaviors were more frequent among residents with more limitations in physical functio ning ( p < 0.05) among 1 101 residents with dementia from 59 NHs across Maryland Beck and colleagues (1998) reported that ADL dependency positively correlated with vocal agitation among 45 nursing h ome residents with dementia from 3 NHs in Arkansas and Maryland. Sloane and colleagues (1998) reported that higher levels of dependency in ADLs were related with increased agitation levels using 3723 observations of resident behaviors in 53 Alzheime r's disease Special Care Units in NHs in Kansas, Maine, Mississippi, and South Dakota In contrast, Zeisel and colleagues ( 2003) reported that a resident having problems performing ADLs were more likely to have a lower aggression score on the Cohen Mansfield scale using data from 427 residents in 15 Alzheimer's special care units. Norton and colleagues (2010) reported that residents who were more impaired in ADL performance showed a lower rate of disruptive behavior measured by Revised Memory and Beh avior Problem Checklist Nursing Home (RMBPC NH) among 169 residents from 10 nursing homes in Birmingham, Alabama. Kiely and colleagues (2000) reported that NH residents with functional impairment were 72% less likely to develop wandering behaviors compared to NH resident without use of antipsychotic medication among 8982 NH residents from the states Mississippi, Texas, and Vermont. Sociodemographics and Disruptive Behaviors Sociodemographics are defined as sociological and demographical combined total measure (Algase, Yao, et al., 2007) Sociodemographics, including age, gender, marital status, education, ethnicity, are associated with disruptive behaviors in the NDB model (Algase, Yao, et al., 2007) However, in the published literature, all of the sociodemographic factors (e.g., age, gender, marital status,

PAGE 74

74 education, and race/ethnicity) showed equivocal or non significant relationship with disruptive behaviors The details were discussed below. A ge and d isruptive b ehaviors The relationship between age and disruptive behaviors is inconclusive. Three published studies (Schonfeld, et al., 2007 ; Song & Algase, 2008; Zeisel, et al., 2003) reported that age was positively related to disruptive behaviors, the other three studies (Beck, et al., 1 998; Brodaty, et al., 2001; Schreiner, Yamamoto, & Shiotani, 2000) reported that age was negatively related to disruptive behaviors and one study (Eustace, et al., 2001) reported that age was not rel ated with disruptive behaviors Sconfeld and colleagues (2007) reported that one unit change in age is related to1.02 (95% CI [ 1.01 1.03 ] ) greater odds of wandering behaviors among 15,092 residents with moderate or severe cognitive impairment f rom 134 NH facilities operated by the Department of Veterans Affairs. Song and Algase (2008) reported that a ge was significantly positively correlated with wandering factors, such as spatial disorientation, attention shift, and negative outcomes ( r = 29, r = .16 r = .23, respectively; p < .01) using data from 160 PWDs residing in 14 long term care facilities in South Korea. Zeisel and colleagues (2003) reported that age was significantly positively correlated with the expression of verbal using 427 residents in 15 Alzheimer's special care units. In contrast, Schreiner and colleagues (2000) reported that age was negatively associated with wandering ( p < .05) using data from 392 residents in 6 NHs in Ja pan. Brodaty (2001) reported that y ounger NH residents manifested more disru ptive behaviors among 647 residents from 11 Sidney nursing homes Beck and colleagues (1998) reported that age negatively correlated with disruptive behaviors ( p < .05) among 45 nursi ng home residents with dementia from 3 NHs in Arkansas and Maryland. Besides, Eustace and colleagues (2001) reported that age was not significantly associated with verbal aggression using 150 PWDs.

PAGE 75

75 G ender and disruptive behaviors The rela tionship between gender and disruptive behaviors is inconclusive Many studies reported that male gender showed more disruptive behaviors than female, but there are some studies reporting disruptive behaviors as being more common in female s Voyer and coll eagues (2005) reported that males are about two times more likely to display aggressive behaviors (OR = 2.13) using data from 2,332 older adults in long term care facilities in the Quebec City area Matsuoka and colleagues (2003) reported the mal e gender was significantly related to aggression using 730 patients from 180 units that have specialized psychiatric beds for acute or long term care of PWDs. Zeisel and colleagues (2003) reported that male residents are more likely to show aggression than female residents using data from 427 residents in 15 Alzheimer's special care units. Schreiner (2001) reported that male s were significantly more l ikely to exhibit physically aggressive behavior using 391 NH residents with dementia in Japan. Kiely and colleagues (2000) report ed that women were 39% less likely to develop wandering behaviors among 8 982 NH residents from the states of Mississippi, Texas, and Vermont. Menon and colleagues (2001) reported that v erbal ly aggressive behaviors were also more prevalent among males ( p < .05) among 1101 residents with dementia from 59 NHs across Maryland. Eustace and colleagues (2001) reported that m ale gender was significantly associated with verbal aggressi on ( p = .022) in a stepwise backward logistic regression analysis among 150 PWDs in Ireland. Beck and colleagues (1998) reported that male s had a greater frequency of disruptive behaviors ( p < .05) among 45 nursing home residents with dementia from three NHs in Arkansas and Maryland. Ott and colleagues (2000) reported that men were more likely than women to exhibit disruptive behaviors using Syste matic Assessment and Geriatric drug use via Epidemiology (SAGE) database, which contains data collected with the Minimum Data Set on a cross section of 28,367 NH residents with dementia in five US states.

PAGE 76

76 In contrast, Beck and colleagues (2011) reported that females were two times more likely to show agitation as measured by Cohen Mansfield Agitation Inventory using data from 138 residents of 17 NHs. Cohen Mansfield and Libin (2005) reported that females manifested more verbal agitation than males ( p < .001) among 175 elderly PWDs recruited from 11 nursing home facilities in Maryland. Burgio and colleagues (2000) reported that female residents displayed almost three times the amount of agi tation as male residents (35% versus 13% of total observation time, respectively) u sing data from 46 NH residents with clinically significant agitation. Vance and colleagues (2003) reported that female s exhibited more agitation than male s using 123 residents from five nursing homes located in the Pittsburgh area. M arital status and disruptive behaviors The relationship between marital status and disruptive behaviors is equivocal Menon and colleagues (2001) reported that p hysical ly aggressive behaviors were more frequent among residents who were not married before admission ( p < .001) among 1101 residents with dementia from 59 NHs across Maryland. On the contrary, Beck and colleagues (1998) reported that m arried participants were more likely to manifest aggressive behaviors using 45 nursing hom e residents with dementia from t hree NHs in Arkansas and Maryland. E ducation and disruptive behaviors The length of education is found to be somewhat negatively associated with disruptive behaviors but it is not statistically significant. Chen and colleagues (2000) reported that there the length of education had negative associations with disruptive behaviors ( p = .07) using data from 125 elderly patients with dementia E thnicity and disruptive behaviors The effect of e thnicity on disruptive behaviors is inconclusive. Sink and colleagues (2004) reported that African American PWDs were significantly more likely than Cauca sian to wander

PAGE 77

77 (OR=1.40, 95% CI: 1.08 1.81), and Latinos had a significantly higher likelihood of wandering than Caucasian s (OR=1.5 9, 95% CI : 1.21 2.26) using data from 5,776 Medicare patients enrolled in the Medicare Alzheimer's Disease Demonstration and Evaluation study at 8 sites across the United States between 1989 and 1991. However, Cohen and colleagues (1998) reported that African American dementia patients had significantly lower levels of agitation on the Cohen Mansfield Agitation Inventory than Caucasian dementia patients using 164 U.S. born African Americans, 54 African Caribbeans, an d 68 Caucasians NH residents with dementia Summary Among ten background factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, and ethnicity) the published litera tures showed inconclusive relationship s between background factors and disruptive behaviors with the except ion of cognitive impairment and comorbidity C ognitive impairments are positively associated with disruptive behaviors, and comorbidity is negativel y related to disruptive behaviors. The relationship between the other background factors (e.g., pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, and ethnicity) and disruptive behaviors is found to be equivocal or to have no published data The Relationship between Background Factors and Pain Based on the NDB model (Algase, et al., 1996) background factors are significantly associated with pain. The following is a brief review of the effects of background factors on pain, including neurocognitive factors, general health (e.g., comorbidity, pressure ulcers, number of medication s ADL impairment s), and sociodemographics (e.g., age, gender, marital status, education, and ethnicity).

PAGE 78

78 C ognitive Impairments and Pain Cognitive impairment s are negatively associated with pain report Reynolds and colleagues (2008) reported that pain prevalence decreased as cognitive abilities declined like pain prevalence of 34%, 31%, 24%, and 10%, respectively, for residents with no mild, moderate, and severe cognitive impairment ( p < .001) using 551 residents from six North Carolina NHs. Black and colleagues (2006) reported that those with pain reports had a significantly higher cognitive function ( t = 2.433, df = 119, p = .016) using 123 NH residents with advanced dementia from 3 NHS in Maryland. Teno and colleagues (2004) reported that persons who were more cognitively impaired were less likely to report daily excruciating pain using 2,138,442 MDS assessment data in the U.S. which were completed around April 1999. They reported that p ersons with the most severe cognitive impairment were 75% less likely to report daily excruciating pain than persons without any cognitive impairment. Fisher and colleagues (2002) reported that more cognitively impaired residents were less likely to report the presence of pain and higher frequency of pain using 57 residents with cognitively im pairments from three NHs in the greater Birmingham, Alabama area. General Health and Disruptive Behaviors General health is pressure ulcer, number of medications, and ADL impairments (Algase, Yao, et al., 2007) All factors regarding general health (e.g., comorbidity, pressure ulcer, number of medications, and ADL impairments) are reported to be pos itively correlated with pain. The details were discussed below. Comorbidity and pain Comorbidity is positively correlated with pain. Black and colleagues (2006) reported that those with pain reports had a significantly higher total number o f health problems than those

PAGE 79

79 with no pain reports ( t = 3.237, df =121, p = .002) among 123 NH residents with advanced dementia from three NHS in Maryland. Reports of pain were significantly related with musculoskeletal, hematologic/oncologic and circulatory disorders, aspiration and trauma or fractures. They showed that aspiration, peripheral vascular disease and musculoskeletal disorders remained significantly associated with pain by regres sion analysis. Shega and colleagues (2010) reported that comorbidity was positively correlated with pain report (odds ratio = 1.41, 95% CI [ 1.30 1.54 ] ) by logistic regression among cognitively i mpaired participants among 5,549 older adults. Schuler and colleagues (2004) reported that patient s with persistent pain tended to have more comorbidities using 55 patients of acute and rehabilitation wards of a German geriatric hospital. Pressure ulcer and pain P ressure ulcer is positively associated with pain Teno and colleagues (2004) reported that residents with pressure ulcer were 60% more likely to have daily excrucia ting pain than residents without pressure ulcer (adjusted odds ratio= 1.6) among 2,138,442 MDS assessment data in the U.S. which were completed around April 1999. Gunes (2008) found that 95% subjects reported pressure ulcer pain; of those, 60% reported constant pressure ulcer pain among 47 hospitalized patients admitted for neurologic al disorders. Nemeth and colleagues (2004) reported pain associated with ulcer using 20 older adults with venous ulcer Szor and Bourguignon (1999) reported that 42% reported they had a cont inuous pain due to pressure ulcer even when they were at rest in a sample of 32 adults. Number of medications and pain N umber of medication is positively associated with pain report. Schuler and colleagues (2004) reported chronic pain was significantly correlated with wider range of medication among 55 patients of acute and rehabilitation w ards of a German geriatric hospital

PAGE 80

80 ADL impairments and pain ADL impairments are positively associated with pain report. Boerlage and colleagues (2008) reported that pain was associated with limited ADL among 157 residents from three Dutch residential homes. Shega and colleagues (2010) reported that pain was associated with functional impairment (odds ratio = 1.74, p < 0.01) and lower self rated health (odds ratio = 2.35, p < 0.01) among cognitively impaired subjects using 5,549 older adults. Teno and colleagues (2004) reported that residents dependent in self care functions were 2.8 times more likely to have daily excruciating pain using 2,138,442 MDS assessment data in the U.S. which were completed around April 1999. Sociodemographics and Pain Sociodemograp hics are defined as sociological and demographical combined total measure (Algase, Yao, et al., 2007) Sociodemographics, including age, gender, marital status, education, ethnicity, are associated with disruptive behaviors (Algase, Yao, et al., 2007) Most of the socio demographics factors (e.g., age, marital status, and race/ethnicity) showed equivocal relationship with pain However, female gender was reported to be positively correlated with pain, and highest level of education was reported to be negatively associated with pain. The details were discussed below. A ge and pain The relationship between age and pain report is equivocal Shega and colleagues (2010) reported that age was negatively correlated wit h pain report (odds ratio=0.97, 95% CI [ 0.95 0.99 ] ) by logistic regression among cognitively impaired participants among 5,549 older adults. Teno and colleagues (2004) reported that daily excruciating pain was more prevalent in younger residents among 2,138,442 MDS assessment data in the U.S. which were completed around

PAGE 81

81 April 1999. They described that persons of age 80 and older were 50% less likely to report daily excruciating pain than those younger than 65 years old In contrast, Bruckenthal and colleagues (2009) reported pain incidenc e generally increase as the population ages. Stranjalis and colleagues (2011) reported that increased age was significantly associated with the presence of neck pain using 1000 person sample of Greek urban population. G ender and pain Female s are more likely to report pain. Shega and colleagues (2010) reported that pain was negatively correlated with being a male using 5,549 older adults. Stranjalis and colleagues (2011) reported that f emale was significantly associated with the presence of neck pa in using 1 000 Greek urban population. Marital status and pain The relationship between m arried status and pain is equivocal. Shega and colleagues (2010) reported that pain was negatively correlat ed with being married using 5,549 older adults. In contrast, Stranjalis and colleagues (2011) reported that being married was significantly associated with the presence of neck pain among 1 000 G reek urban population. Education and pain Higher education is negatively associated with pain report. Shega and colleagues (2010) reported that pain was negatively correlated with highest level of education among 5,549 older adults. Ethni city and pain The relationship between ethnicity and pain is inconclusive. Horgas and colleagues (2008) reported that Blacks and Whites did not differ significantly in intensity ( t = 1.14, df = 44, p = .26) or duration of self reported pain (chi square = 3.68, df = 3, p = .30), or in the number of pain

PAGE 82

82 locations reported ( t = 1.12, df = 67, p = .23) among 115 older adults. Golightly and Cominick (2005) reported that African American s with osteoarthritis reported more severe pain than Caucasians among 202 African American and Caucasian veterans. Teno and colleagues (2004) reported that African Americans were 25% less likely to report daily excruciating pain using 2,138,442 MDS assessment data in the U.S. which were completed aroun d April 1999. Rahim Williams and colleagues (2007) reported that African American and Hispanic subjects showed lower pain tolerances than non Hispanic White Americans in 63 African American, 61 Hispanic and 82 non Hispanic white participants. Green and colleagues (2003) reported that racial and ethnic minorities tend to be undertreated for pain when compared with non Hispanic Whites using a selective literature review performed by expe rts in pain. Riley and colleagues (2002) reported that African American patients reported significantly higher levels of pain unp leasantness, emotional response to pain, and pain behavior, but not pain intensity than Caucasians using 1084 Cau ca sian and 473 African American with chronic pain. Summary Among 10 background factors, five background factors (e.g., comorbidity, pressure u lcer, number of medications, ADL impairments, and female gender ) are positively associated with pain. Cognitive impairments and education level are negatively correlated with pain report. The relationship the other background factors (e.g., age, marital st atus, and ethnicity) and pain is reported to be equivocal. Summary The review of the literature supports the need for and importance for further study on the relationship between pain and disruptive behaviors in PWDs. First, the one published article regar ding between pain and wandering (Kiely, et al., 2000) has limited findings due to inappropriate pain measurement and sample. Pain was measured using a sad or painful

PAGE 83

83 expression. The painful expression can be considered a valid pain indicator, but confound ing the expression of pain with expression of sadness may not be a valid measurement of pain. The authors did not report the reliability or validity of a sad or pained expression as a measurement of pain, and sadness is not the same concept as pain. Also, they used data from all NH residents, without controll ing for cognitive status. Not all NH residents expressed pain via wandering. Wandering behavior was observed mainly in NH residents with dementia. So, a study o f the relationship between pain and wandering using the appropriate pain measurement and sample in this case NH residents with dementia, is needed. Second, there are two studies on the relationship between pain and aggression (Leonard, et al., 2006; Volicer, et al., 2009) but their fin dings are also limited. Leonard and colleagues measured pain using only pain intensity and aggression frequency by means of a single MDS aggression item, and used data from all NH residents, not just NH residents with dementia. Volicer and colleagues measu red pain using only pain frequency and aggression frequency b y means of a single MDS aggression item, and recruited Dutch Nursing Home residents with dementia. Comprehensive pain measurements using both pain frequency and pain intensity have better psychom etric evidences than either MDS pain frequency or intensity item. Also, t he newly developed comprehensive MDS ABS (Perlman & Hirdes, 2008) has better psychometric evidence compared to a single MDS item for measuring aggression A study on the relationship between pain and aggression using appropriate measurement and involving U.S. Nursing Home res idents with dementia is needed. Third, only two studies (Buffum, et al., 2001; Manfredi, et al., 2003) reported a significant relationship between pain and agitation in Nur sing Home residents with dementia The other studies used all NH residents (not NH residents with dementia) or NH residents in other

PAGE 84

84 countries. However, even these two studies used small sample sizes (N = 13 and 33). A small sample size limits the generali zability of the findings (Polit & Beck, 2004) A study using large sample is needed to validate these findings. By using the MDS CBP agitation subscale (Gerritsen, et al., 2008) one gains access to state or nation wide large samples of federally mandated MDS assessment data. A large sample size will be used in this study. Fourth, none of these published studies regarding the relationship between pain and disruptive behaviors controll ed for the use of physical or chemical restraint. Residents with restraints cannot display disruptive behaviors in the same way In particular, restrai ned residents cannot wander. Therefore, acquiring more accurate estimates of disruptive behaviors requires controllin g for restraint use. Fifth, studies on the relationship between pain and wandering or aggression used a binary logistic regression for mult ivariate analysis. The published studies on the relationship between pain and wandering would present more meaningful results if an ordinal logistic regression analysis rather than binary logistic regression, were used for multivariate analysis, because t he levels of measurement for wandering and agitation are ordinal. Sixth the high cost of care for the cognitively impaired, especially in nursing homes, also supports the need to study the relationship between pain and disruptive behaviors in NH resident s with dementia The cost of caring for PWDs in nursing homes is much higher than that of their cognitively intact counterparts Health Statistics, 2006) One of t he reasons for this is the difficulty in the management of disruptive behaviors of NH residents with dementia (Ayalon, et al., 2006; Connor, et al., 2009; Herrmann, 2001; K utsumi, et al., 2009) If the relationship between pain and disruptive behavior

PAGE 85

85 is validated, then appropriate pain management can be utilized to reduce or avert these behaviors and possibly reduce healthcare costs for this population. Seventh, t he lack of definite treatment of disruptive behaviors also supports the need to investigate the relationship between pain and disruptive behaviors. Disruptive behaviors are hard to control because there is no universally effective treatment (Ayalon, et al., 2006; Connor, et al., 2009; Herrmann, 2001; Kutsumi, et al., 2009) Physical and chemical restraints are often used to man age disruptive behaviors. However, restraints of either kind cause significant side effects, and are not recommended as the first choice of intervention. Pain could be a reason for disruptive behaviors in PWDs and pain management could be a solution for ad dressing the problem. Finally, t he need to investigate this relationship is supported by poor pain management of NH residents with dementia NH residents with dementia do not get appropriate management of pain (Morrison & Siu, 2000; Reynolds, et al., 2008; Sengupta, et al., 2010) The under recognition and under treatment of pain in this population unders core the possibilities of the high prevalence of disruptive behaviors.

PAGE 86

86 Table 2 1 Characteristics of NHs in the United States Characteristics NHs Beds NH Residents Number of nursing staff per 100 beds Number of registered nurses per 100 beds Number of l icensed practical nurses per 100 beds Number of c ertified nursing assistants per 100 beds Total 16,100 1,730,000 1,492,200 917,400 (53) 119,500 (7) 184,600 (11) 600,800 (35) Type of Ownership Proprietary 9,900 (61.5%) 1,074,20 0 918,000 548,200 (51) 66,200 (6) 114,700 (11) 359,900 (34) Voluntary nonprofit 5,000 (30.8%) 503,600 440,300 279,200 (55) 40,000 (8) 52,700 (10) 182,500 (36) Government and other 1,200 (7.7%) 152,200 133,900 89,900 (59) 13,300 (9) 17,200 (11) 58,400 (38) Number of Beds Fewer than 50 beds 2,200 (13.9%) 75,800 62,200 43,600 (58) 6,700 (9) 8,300 (11) 27,900 (37) 50 99 beds 6,000 (37.3%) 454,700 422,600 240,800 (53) 32,300 (7) 47,100 (10) 156,200 (34) 100 199 beds 6,800 (42.5%) 903,100 788,500 478,800 (53) 60,200 (7) 98,600 (11) 314,900 (35) 200 beds or more 1,000 (6.2%) 296,400 218,900 154,300 (52) 20,300 (7) 30,600 (10) 101,900 (34)

PAGE 87

87 Table 2 1. Continued Characteristics NHs Beds NH Residents Numbe r of nursing staff per 100 beds Number of registered nurses per 100 beds Number of l icensed practical nurses per 100 beds Number of c ertified nursing assistants per 100 beds Geographic R egion Northeast 2,800 (17.4%) 381,500 331,300 214,900 (56) 32,900 (9) 41,700 (11) 138,200 (36) Midwest 5,300 (33.0%) 526,600 448,000 263,700 (50) 35,200 (7) 49,300 (9) 174,700 (33) South 5,400 (33.6%) 585,600 501,500 311,100 (53) 33,500 (6) 69,200 (12) 205,000 (35) West 2,600 (16.0%) 236,200 211,400 127,800 (54) 18,000 (8) 24,400 (10) 82,900 (35) Location Metropolitan statistical area 10,900 (67.7%) 1,290,900 1,127,800 686,900 (53) 91,200 (7) 138,800 (11) 449,400 (35) Micropolitan statistical area 2,600 (16.2%) 242,200 202,000 129,800 (54) 15,300 (6) 26,900 (11) 84,300 (35) Other location 2,600 (16.0%) 196,900 162,400 100,700 (51) 13,000 (76) 18,800 (10) 67,100 (34) Excerpted from National Center for Health statistics 2006 from http://www.cdc.gov/nchs/data/nnhsd/nursinghomefacilities2006.pdf

PAGE 88

88 T able 2 2 Minimum Data Set i tems Section number Section name Items Section A. Identification and background information Room number, assessment reference date, reentry date, admitted from at entry, marital status, medical record number, payment source for NH stay, reasons for assessme nt, responsibility/legal guardian, advance directives Section AA. Identification information Resident name, gender, birth date, race/ethnicity, social security number, resident Medicare number or comparable number, facility state number, facility federal number, resident Medicaid number, primary reason for assessment, special assessment code Section AB. Demographic information Date of entry, admitted from at entry, lived alone prior to entry, zip code of prior primary residence, residential history, lifetime occupation, highest level education completed, language, mental health history, condition related to mental retardation, date background information completed Section AC. Customary routine Cycle of daily events, eating patterns, activi ties of daily living, involvement patterns Section B. Cognitive patterns Comatose, memory, memory/recall ability, cognitive skills for daily decision making, indicators of delirium, change in cognitive status Section C. Communication/ hearing patterns Hearing, communication devices/techniques, modes of expression, making self understood, speech clarity, ability to understand others, change in communication/hearing Section D. Vision patterns Vision, visual limitation/difficulties, visual appliances Section E. Mood and behavior patterns Indicators of depression or anxiety or sad mood, mood persistence, change in mood, behavioral symptoms, change in behavioral symptoms Section F. Psychosocial well being Sense of initiative, unsettled relationships, past roles Section G. Physical functioning and structural problems Activities of daily living (ADLs) self performance, ADL support, bathing, test for balance, functional limitation in range of motion, modes of locomotion, modes of transfe r, task segmentation, ADL functional rehabilitation potential, change in ADL function Section H. Continence in last 14 days Continence self control, bowel elimination patterns, appliances and programs, change in urinary continence Section I. Disease diagnoses Diseases, infection Section J. Health conditions Problem conditions, pain symptoms, pain site, accidents, stability of conditions

PAGE 89

89 Table 2 2 Continued Section number Section name Items Section K. Oral/nutritional status Oral problems, height and weight, weight change, nutritional problems, nutritional approaches, parental or enteral intake Section L. Oral/dental Status Oral status and disease prevention Section M. Skin condition Ulcer, type of ulcer, history of resolved ulcers, other skin problems or lesions present, skin treatments, foot problems and care Section N. Activity pursuit patterns Time awake, average time involved in activities, preferred activity settings, general activity preferences, prefers change in daily rout ine Section O. Medications Number of medications, new medications, injections, days received the medications Section P. Special treatment and procedures Special treatments or programs, intervention programs for mood or behavior, nursing rehabilitation/restorative care, devices and restraints, hospital days, emergency room visits, physician visits, physician orders, abnormal lab values Section Q. Discharge potential and overall status Discharge potential, overall change in care needs Section R. Assessment information Participation in assessment, discharge disposition Section T. Supplement special treatments and procedures Special treatments and procedures, walking when most self confident, Section V. Resident Assessment Pr otocol (RAP) Summary Location and date of RAP assessment documentation Section W. Additional items collected Influenza and pneumococcal vaccination

PAGE 90

90 Table 2 3 Summary of psychometric properties on published MDS 2.0 items or scales on pain, cognition, and behavior Source Number of residents Number of NHs Validity Reliability MDS Pain (Fries, et al., 2001) 95 25 Criterion validity: 0.70 against Visual Analogue Scale Inter rater reliability coefficient of 0.73 (Cohen Mansfield, et al., 1999) 80 8 coefficient of 0.69 0.88 (Fisher, et al., 2002) 57 3 No significant relationship with Proxy Pain Questionnaire developed by researchers Test retest reliability coefficient of 0.84 0.87 MDS Cognitive Performance Scale (MDS CPS) (Bula & Wietlisbach, 2009) 401 Criterion validity: 0.52 against MMSE (Morris, et al., 1994) 2,172 Criterion validity: high against MMSE and Test for Severe Impairment, Construct validity: reflected two different samples Test retest reliability: high (Hartmaier, et al., 1994) 200 8 Criterion validity: 0.41 0.76 against Global Deterioration Scale, high against MMSE (Gruber Baldini, Zimmerman, Mortimore, & Magaziner, 2000) 1,939 59 Criterion validity: 0.68 against MMSE, 0.66 against Psychogeriatric Dependency Rating Scale, Orientation Scale (Cohen Mansfield, et al., 1999) 290 1 Criterion validity: 0.71 0.75 against MMSE, 0.75 0.77 against Global Deterioration Scale (Snowden, et al., 1999) 140 Criterion validity: 0.45 against MMSE (Lawton, et al., 1998) Criterion validity: 0.45 against Mattis Dementia Rating Scale, 0.7 against Global Deterioration Scale (Horgas & Margrett, 2001) 135 1 Criterion validity: Significant for nondepressed sample, but not for depressed sample against RMBPC

PAGE 91

91 Table 2 3. Continued Source Number of residents Number of NHs Validity Reliability MDS Activities of Daily Living (Lawton, et al., 1998) 513 1 Criterion validity: 0.58 0.79 against Physical Self Maintenance Scale Reliability coefficient of 0 .97, kappa coefficient of 0.61 (Hawes, et al., 1995) 123 13 Reliability coefficient of 0.92 (Sgadari, et al., 1997) 123 13 Reliability coefficient of 0.92 MDS wandering (Casten, et al., 1998) Inter rater r eliability coefficient of 0.95 (Sgadari, et al., 1997) 123 13 Reliability coefficient of 0.63 (Lawton, et al., 1998) Criterion validity: 0.51 0.54 MDS Aggression Behavior Scale (Perlman & Hirdes, 2008) 652 4 Criterion validity: 0.72 with Cohen Mansfield Agitation Inventory aggression subscale scores coefficient of 0.79 to 0.95 MDS Agitation Scale (Gerritsen, et al., 2008) 656 4 Criterion va rank correlation coefficient of 0.50 with Behavior Rating Scale for Psychogeriatric Inpatients coefficient of 0.70, inter rater reliability of 0.61 Note: MMSE : Mini Mental State Examination; NH : nursing home; RMBPC : Revised Memory and Behavior Problem Checklist The detailed reliability and validity coefficients for this study items will be described in the Chapter 3.

PAGE 92

92 Table 2 4 Literature regarding the relationship between pain and disruptive behaviors in fac ility dwelling older adults Source Sample/setting /study period Findings Pain and Wandering (Kiely, et al., 2000) 8,982 NH residents from Mississippi, Texas, and Vermont on MDS assessment data between January, 1996 and December, 1997 NH residents who expressed sadness or pain in MDS assessment data were 65% more likely to develop wandering behavior than their counterparts who did not express sadness or pain in MDS data using a logistic regression analysis, after controlling for pneumonia, short term memory, repetitive questions, long term memory, constipation, antipsychotic medication, ADL impairment, and gender, OR = 1.65, p = .02, 95% CI [1.08, 2.53]. Pain and aggression (Volicer, et al., 2009) MDS assessment data of 929 residents with dementia from 8 Dutch nursing homes and 10 residential homes between April 4, 2007 and December 1, 2008 Pain frequency was significantly correlated with frequency of verbal aggression using MDS assessment data in a bivariate correlation analysis ( r = .07, p = .028 ) In a multivariate analysis, pain frequency was not significantly associated with frequency of aggression, after controlling for hallucinations, delusions, ability to understand others, depression, and co nstipation. (Leonard, et al., 2006) 103,344 NH residents in 5 states who had at least one annual MDS assessment data completed during 2 002 Pain intensity was not significantly associated with frequency of physical aggression using MDS assessment data in a multivariate logistic regression analysis, after controlling for depression, delusions, hallucinations, constipation, respiratory tract infection, urinary tract infection, fever, and the amount of time participating in non ADL or treatment activities such as recreational activities. Pain and agitation (Buffum, et al., 2001) 33 Veterans Affairs NH residents with dementia from one Veterans Affairs Medical Center unknown study period Pain severity measured by a discomfort scale was significantly associated with the presence of agitation, as measured by modified Cohen Mansfi eld Agitation Inventory ( r = .50, p = .0 03 ) (Manfredi, et al., 2003) 13 NH residents with dementia in one nursing home who were more than 85 years old conducted from January 1999 to January 2001 Low dose, long acting op ioid treatment reduced the frequency of agitation measured by the CMAI using a prospective longitudinal design over 4 weeks, mean change in CMAI score: 6.4, 95% CI [ 10.96, 1.8].

PAGE 93

93 Table 2 4 Continued Source Sample/setting Findings (Villanueva, et al., 2003) 40 residents from four long term care facilities, three skilled nursing facilities, and a locked dementia assisted living facility unknown study period Pain measured by the pain assess ment for the dementing elderly (PADE) was significantly associated with frequency of agitation as measured by the CMAI. The physical assessment section of PADE correlates significantly with the CMAI verbal agitation subscale ( r = .296, p < 0.01 ) and the g lobal assessment section of PADE is significantly associated with physically aggressive, verbal and physical agitation subscales of the CMAI ( r = 0.396, r = 0.398, r = 0.421, respectively ; p < .01 ) (Bartels, et al., 2003) 2,487 physically frail older residents from 109 long term care facilities unknown study period The prevalence of pain was sign ificantly associated with mixed agitation and depression in NH residents with dementia by documentation on the medical record. Thirty four percent of PWDs with mixed agitation and depression exhibited pain, whereas only 23% of uncomplicated PWDs exhibited pain, chi square=36.20, p < .0001. (Norton, et al., 2010) 169 residents from 10 NHs in Birmingham, Alabama unknown study period Certified nursing assistant (CNA) reported weekly resident pain intensity were significantly associated with disruptive behaviors as measured by subscales within the Revised Memory and Behavior Problem Chec klist Nursing Home (RMBPC NH) in a multiple regression analysis, after controlling for cognitive impairments, gender, ethnicity, and ADL functional status = .183, p = 0.04 ) The number of painful diagnoses and analgesic medication administration by the chart review was not significantly associated with disruptive behaviors as measured by subscales within the RMBPC NH

PAGE 94

94 CHAPTER 3 METHODS Design This is a secondary analysis of the Minimum Data Set (MDS 2.0) assessment data on long term care from the state of Florida. An exploratory cross sectional design is used for this study. Cross sectional data, the first MDS comprehensive assessment data for each NH residents during the 12 month time frame, are used to explore the relationship between pain and disruptive behaviors. Sample Participants in this study are residents with dementia in Medicare or Medicaid certified NHs in the state of Florida betwe en January 1, 2009 and December 31, 2009 who ha ve a MDS comprehensive assessment on file. The MDS comprehensive assessment data are used for this study because this is the only version of the MDS that includes all the needed variables. These are shortened assessment data that do not include some variables for this study such as complaints of hunger ( MDS item # K4B), diseases ( MDS ite m # I1), infection ( MDS item # I2), other d iagnoses ( MDS item # I3), and education ( MDS item # AB7) and therefore are excluded MDS assessment data from Florida NHs are acquired from the Centers for Medicare & Medicaid Services (CMS). All of the U.S. nurs ing home MDS assessment is collected quarterly by CMS, and is stored in American Standard Code for Information Interchange ( ASCII) format text file. The Research Data Assistance Center (ResDAC), based at the University of Minnesota, is the data coordinatin g agency that pre reviews proposals for MDS use and helps distribute MDS data from CMS to researchers. There is a fee for its use ($ 1,000/state/year). CMS stipulated that the MDS data users should: ensure that data will be used only for the specified pur poses in the data use agreement; submit the executive summary which states the

PAGE 95

95 data management plan and lists the key personnel of the study being proposed; submit the study protocol which includes background and importance, research questions and methods, analysis plan and schedule; plan on implementing safeguards to prevent unauthorized use of data; return or destroy data by a specified date; and not release information that would permit the identification of the subjects (Research Data Assistance Center [ResDAC], 2010) Selection criteria are applied to ascerta in data from NH residents older than 65 years old MDS item # I1Q = yes or # I1U = yes). Data from comatose residents (MDS item # B1 = 1) are excluded. Also, the statistical analyses for NH residents without restraints are reported separately because residents with restraints can not exhibit disruptive behaviors Proposed Measure s Measures, o perational d efinitions and the interpretations of each measure are summarized in Table 3 1. Pain Th e MDS pain severity scale (Fries, et al., 2001) combin ing both pain frequency and pain intensity ( MDS item # J2A and J2B respectively ) is used to assess pain severity in NH residents with dementia MDS pain severity scale (Fries, et al., 2001) is recorded on a 4 point ordinal sc ale, 0 ( no pain ), 1 ( mild pain ), 2 ( moderate pain ), and 3 ( excruciating pain ). Mild pain accounts for pain which is less than daily in MDS pain frequency. Moderate pain accounts for daily pain in MDS pain frequency and mild or moderate pain in MDS pain int ensity. Excruciating pain accounts for daily pain in MDS pain frequency and horrible or excruciating pain in MDS pain intensity. MDS pain items (frequency and intensity) code the highest level of pain present in the last seven days. Pain frequency is recor ded on a three point ordinal scale, with 0 ( no pain ) 1 ( pain less than daily ), and 2 ( pain daily ). Pain intensity is also coded on a three

PAGE 96

96 point ordinal scale, with 1 ( mild pain ) 2 ( moderate pain ), and 3 ( horrible or excruciating pain ). f report is reflected in the MDS pain items if residents can self report and staff report. Otherwise, the staff who complete the MDS assessment document pain symptoms based on reports from fac ility nursing staff that takes care of the residents. The MDS pain severity scale has been reported to have an inter rater reliability coefficient of 0.73 and higher, and kappa coefficient of 0.70 with a Visual Analogue Scale in a study involving 95 U.S. nursing home residents at 25 Medicare certified skilled nursing facilities in Massachusetts (Fries, et al., 2001) The ori ginal MDS pain intensity and frequency is reported to have an internal consistency of 0.69 0.88 and test retest reliability of 0.84 0.87 (Cohen Mansfield, et al., 1999; Fisher, et al., 2002) Secondary P roximal F actors The measurements of secondary proximal factors (e.g., hunger, thirst, and elimination) are described belo w. H unger The MDS hunger item ( MDS item # K4B) is regular or repetitive complaints of hunger by a dichotom ized variable, with 0 ( no ) and 1 ( yes ). The MDS hunger item has a reliability coefficient of 0.69 (Hawes, et al., 1995) T hirst The MDS dehydration status ( MDS item # J1C) is recorded on a dichotomous scale, 0 (no) and 1 (yes). The MDS dehydration item has a reliability coefficient of 0.49 (Hawes, et al., 1995) E limination The MDS bowel in continence and bladder in continence items ( MDS item # H1A and H1B) are 5 point ordinal scale, with

PAGE 97

97 0 ( continent ) 1 ( usually continent ), 2 ( occasionally incontinent ), 3 ( frequently incontinent ), and 4 ( incontinent ). The MDS continence items have a reliability coefficient of 0.91 to 0.99 and an inter rater reliability coefficient of 0.61 (Casten, et al., 1998; Hawes, et al., 1995) Background F actors The measurements of the background factors (e.g., neurocognitive fact or comorbidity, pressure ulcer, medication, ADL impairments and sociodemographics) are described below. Neurocognitive factor The MDS cognitive performance scale (MDS CPS) (Morris, et al., 1994) is used to measure memory and cognition. It is a 7 point ordinal scale, with 0 ( intact ) 1 ( borderline intact ), 2 ( mild impairment ), 3 ( moderate impairment ), 4 ( moderate severe impairment ), 5 ( severe impairment ), and 6 ( very severe impair ment ) ( Carpenter, et al., 2006; McConnell, Branch, Sloane, & Pieper, 2003; McConnell, et al., 2002) The MDS CPS score is calculated using five MDS items: comatose ( MDS item # B1), short term memory ( MDS item # B2A), cognitive skills or daily decision making ( MDS item # B4), making oneself understood ( MDS item # C4), and self performance in eating ( MDS item # G1HA). Comatose ( MDS item # B1) is recorded dichotomously (0 = no 1 = yes ) S hort term memory ( MDS item # B2A) is recorded dichotomously (0 = memory OK 1 = memory problematic ). Cognitive skills or daily decision making ( MDS item # B4) is recorded 4 point ordinal scale, with 0 ( independent ) 1 ( modified independence ) 2 ( moderately impaired ), and 3 ( severely impaired ) M aking oneself understood ( MDS item # C4) is recorded 4 point ordinal scale, with 0 ( understood ) 1( usually understood ), 2 ( sometimes understood ), and 3 ( rarely or never understood ). S elf performance in eating ( MDS item # G1HA) is recorded 5 point ordinal scale, with 0 ( independent ) 1 ( supervision ), 2 ( limited assistance ), 3 ( extensive assistance ), and 4 ( total dependence )

PAGE 98

98 The MDS CPS is widely used in research, and has high test retest reliability, a criterion validity coefficient of 0.45 0.75 with Mini M ental State Examination, a criterion validity coefficient of 0.41 0.77 against Global Deterioration Scale, a criterion validity coefficient of 0.66 against Psychogeriatric Dependency Rating Scale, a criterion validity coefficient of 0.45 against Mattis Dementia Rating Scale, a sensitivity of 0.94, and a specificity of 0.94 (Bula & Wietlisbach, 2009; Cohen Mansfield, et al., 1999; Gruber Baldini, et al., 2000; Hartmaier, et al., 1994; Hartmaier, et al., 1995; Lawton, et al., 1998; McConnell, et al., 2003; McConnell, et al., 20 02; Morris, et al., 1994; Snowden, et al., 1999) C omorbidity The Charlson c omorbidity i ndex (CCI) (Charlson, Pomp ei, Ales, & MacKenzie, 1987) is used to assess comorbidity using MDS disease, infections, and other diagnoses ( MDS item # I1, I2, I3, respectively). The CCI is widely used in research (Diederichs, Berger, & Bartels, 2011; Norton, et al., 2010) and has acceptable reliability an d validity (Charlson, Szatrowski, Peterson, & Gol d, 1994) This index is a sum score of 19 weighted categories of comorbidity to assess the impact of comorbid diseases on one year mortality using a medical chart review. The CCI ranges 0 to 37, with higher scores indicating greater comorbidity. Weighti ng of clinical conditions for Charlson C omorbidity I ndex is presented in T able 3 2 Weight one is assigned for myocardial infarct, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia chronic pulmonary disease, connecti ve tissue disease, ulcer disease, mild liver disease, diabetes; Weight two is assigned for hemiplegia, moderate or severe renal disease, diabetes with end organ damage, any tumor, leukemia, lymphoma; Weight three is assigned for moderate or severe liver di sease; Weight six is assigned for metastatic solid tumor and AIDS.

PAGE 99

99 Among 19 diseases for the Charles comorbidity index, 1 3 diseases are counted in the MDS disease or infection item (MDS item # I1 I2, respectively ) : myocardial infarct (MDS item # I1 D ) congestive heart failure (MDS item # I1 F ) peripheral vascular disease (MDS item # I1 J ) cerebrovascular disease (MDS item # I1 T ) dementia (MDS item # I1 Q or I1U ) chronic pulmonary disease (MDS item # I1 II ) diabetes (MDS item # I1 A ) hemiplegia (MDS item # I1 V ) moderate or severe renal disease (MDS item # I1 QQ ) diabetes with end organ damage (MDS item # I1KK) tumors (MDS item # I1PP) mild chronic liver disease or cirrhosis (MDS item # I2K) AIDS (MDS item # I2D) Each disease is recorded by dichotomized variable, with 0 ( no ) and 1 ( yes ). However, six diseases are identified using ICD 9 codes in the MDS other diagnoses (MDS item # I3): connective tissue disease ( ICD 9 code #7100, 7101, 7104, 7140, 7141, 7142, 71481, 5171, and 725 ) peptic ulcer disease ( IC D 9 code #531, 532, 533, and 534 ) leukemia ( ICD 9 code #204, 205, 206, 207, and 208 ) lymphoma ( ICD 9 code # 200, 201, and 202 ) moderate or severe liver disease ( ICD 9 code #5722, 5723, 5724, and 5728 ) and metastatic solid tumor ( ICD 9 code #196, 197, 198, 1990, and 1991 ) The MDS disease / infection item ha s a reliability coefficient of 0.74 (Hawes, et al., 1995) P ressure ulcer MDS pressure ulcer (MDS item # M2 ) is pressure ulcer MDS pres sure ulcer is a 5 point ordinal scale, 0 (none) to 4 (stage 4), and has been reported to have a reliability coefficient of 0.62 (Hawes, et al., 1995)

PAGE 100

100 M edication The number of medications ( MDS item # O1) is recorded by its actual number of different medications used in the last 7 days The MDS medication items have a reliability coefficient of 0.73 (Hawes, et al., 1995) A ctivities of daily living MDS Activities of Daily Living Long Form (MDS ADL Long Form ) (Morris, Fries, & Morris, 1999) is used to measure ADL levels of each resident. MDS ADL Long Form is calculated using 7 MDS items : self performance of bed mobility ( MDS item # G1AA) transfer ( MDS item # G1BA) locomotion on unit ( MDS item # G1EA) dressing ( MDS item # G1GA) eating ( MDS item # G1HA) toile use ( MDS item # G1IA) and personal hygiene ( MDS item # G1JA) Each of these individual items is recorded on a 6 point ordinal scale, with 0 ( independent ), 1 ( supervision ), 2 ( limited assistance ), 3 ( extensive assistance ), 4 ( total dependence ), and 8( activity did not occur during last 7 days ). The MDS ADL Long Form scores are activity did not occur (Morris, et al., 1999) MDS ADL Long Form can range from 0 to 28 with higher scores indicating more dependence of ADLs Th e MDS ADL Long Form is widely used in research, and has been reported to have a reliability coefficient of 0.92 0.97, an inter rater reliability coefficient of 0.61 0.95, and a criterion validity coefficient of 0.58 0.79 against Physical Self Maintenance Scale (Hawes, et a l., 1995; Lawton, et al., 1998; Sgadari, et al., 1997) S ociodemographics Age ( MDS item # AA3), g ender ( MDS item # AA2), marital status ( MDS item # A5), ed ucation ( MDS item # AB7), and race/ ethnicity ( MDS item # AA4) are captured from MDS assessment data

PAGE 101

101 Disruptive B ehaviors The measurements of disruptive behaviors (e.g., wandering, aggression, and agitation) are described as follows W andering MDS wandering frequency item ( MDS item # E4A) is used to measure the frequency of wandering. Wandering frequency is recorded by staff observation of t he frequency of movement with no rational purpose in the last 7 days It is recorded on a 4 point ordinal scale, with 0 ( no wandering ) 1 ( wandering occurred 1 to 3 days in last 7 days ), 2 ( wandering occurred 4 to 6 days in last 7 days ), and 3 ( wandering daily ). The w andering item has been reported to have a reliability coefficient of 0.63, an inter rater reliability of 0.95, and a concurrent validity of 0.51 to 0.54 (Casten, et al., 1998; Hawes, et al., 1995; Lawton, et al., 1998; Sgadari, et al., 1997) A ggression The MDS Ag gression behavior scale (MDS ABS) is used to measure the frequency of aggressive behaviors. MDS ABS is a sum score of four MDS items: verbally abusive behavioral symptoms ( MDS item # E4BA), physically abusive behavioral symptoms ( MDS item # E4CA ), socially inappropriate/disruptive behavioral symptoms ( MDS item # E4DA), and resisting care ( MDS item # E4EA). Each of these individual items are recorded the frequency of these behavioral symptoms in last 7 days on a 4 point ordinal scale, with 0 ( no symptoms ), 1 ( symptoms occurred 1 to 3 days in last 7 days ), 2 ( symptoms occurred 4 to 6 days in last 7 days ), and 3 ( symptoms daily ). MDS ABS can range from 0 to 12, with higher scores indicating more frequency of aggressive behaviors. If needed, aggressive behaviors ( MDS ABS scores) can be collapsed into four groups for statistical analysis none (MDS ABS = 0), moderate (MDS ABS = 1 2), severe (MDS ABS = 3 5), and very severe (ABS= 6 12), based on an established algorithm (Perlman & Hirdes, 2008) The MDS ABS has been reported to have an internal

PAGE 102

102 consistency reliability of 0.79 to 0.95, and a criterion validity coefficient of 0.72 with CMAI aggression subscale scores (Perlman & Hirdes, 2008) A gitation The MDS Challenging Behavior Profile (MDS CBP) agitation subscale is used to assess the frequency of agitated behaviors. MDS CBP agitation scores are calculated using four MDS items: periods of restlessness ( MDS item # B5D), repetitive physical movements ( MDS item # E1N) wandering ( MDS item # E4AA), and socially inappropriate/disruptive behavior ( MDS item # E4DA). Periods of restlessness (MDS item # B5D) is recorded the frequency of this behavioral symptom by 3 point ordinal scale, with 0 (behavior not present), 1 (behavior present, not of recent onset), and 2 (new onset or worsening behavior). Repetitive physical movements (MDS item # E1N) is recorded the frequency of this behavioral symptom in last 7 days by 3 point ordinal scale, with 0 (behavior not exhibited in last 30 days), 1 (behavior exhibited up to five days a week), and 2 (behavior exhibited daily or almost daily). W andering (MDS item # E4AA) and socially inappropriate behavior (MDS item # E4DA) are recorded the frequency of these behavioral symptoms in last 7 days by 4 point ordinal scale, with 0 ( no behavior exhibition ), 1 ( behavior occurred 1 to 3 days in last 7 days ), 2 ( behavior occurred 4 to 6 days in last 7 days ), and 3 ( behavior daily ). The MDS CBP agitation scores are calculated by summing these four items after the scale of three items are changed. Periods of restlessness (MDS item # B5D) is modified from 3 point scale (0 to 2) to 2 point scale (0 and 1). Wandering (MDS item # E4AA) and soc ially inappropriate/disruptive behavior (MDS item # E4DA) are modified from 4 point scale (0 to 3) to 3 point scale (0 to 2). MDS CBP agitation can range from 0 to 7, with higher scores indicating more frequency of agitated behaviors.

PAGE 103

103 MDS CBP agitation su bscale has been reported to have an internal consistency reliability of 0.70, inter rank correlation coefficient of 0.50 with Behavior Rating Scale for Psychogeriatric Inpatients (Gerritsen, et al., 2008) The overall MDS CBP has an internal consistency reliability of 0.54 to 0.78, inter rater reliability of 0.53 (Gerritsen, et al., 2008) Procedures Approval for the study is obtained from the University of Florida Health Science Center Institutional Review Boar d. The researcher work ed with the Research Data Assistance Center (ResDAC) to obtain the Florida 2009 MDS data from the Centers for Medicare & Medicaid Services (CMS) Subject records are assigned an identification number, and no personal identification in formation, such as name, address and social security number, are obtained from CMS. Once obtained, the MDS assessment dataset are stored on network storage drives at the University of Florida Health Science Center, accessible only with a password via an en crypted desktop computer. Data Analysis The ASCII format MDS text file is reformatted to a SPSS data file. The SPSS program is used for storing the database file, analyzing the data, and computing the statistical tests in this study. Results are considered statistically significant with a p value of less than 0.05. First aim To describe the prevalence of disruptive behaviors in NH residents with dementia. 1A) To describe the prevalence of wandering behaviors as measured by MDS wandering frequency in NH re sidents with dementia. 1B) To describe the prevalence of aggressive behaviors as measured by MDS Aggression Behavior Scale (MDS ABS) in NH residents with dementia. 1C) To describe the prevalence of agitated behaviors as measured by

PAGE 104

104 MDS Challenging Behavior Profile (MDS CBP) agitation subscale in NH residents with dementia. MDS wandering frequency is recorded as an ordinal form, with 0 ( no wandering ), 1 ( wandering occurred 1 3 days in last 7 days ), 2 ( wandering occurred 4 6 days in last 7 days ), and 3 ( wande ring daily ). Frequency dat a are presented with a number and percentage for wandering behaviors in NH residents with dementia MDS ABS is recorded as a continuous form, ranging from 0 to 12, with higher scores indicating more frequency of aggressive behavior s MDS CBP agitation subscale is recorded as a continuous variable ranging from 0 to 7, with higher scores indicating more frequency of agitated behaviors. Descriptive statistics, including means standard deviations, and ranges, are presented for aggression and agitation in NH residents with dementia Second aim To investigate the effect of pain severity on the frequency of disruptive behaviors in NH residents with dementia, after controlling for the other background/proximal factors ( e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, hunger, thirst, bowel incontinence, and bladder incontinence ). 2A) To investigate the effect of pain severity on the frequency of wandering behaviors as measured by MDS wandering frequency in NH residents with dementia, after controlling for the other background/proximal factors 2B) To investigate the effect of pain severity on the frequency of aggressive behaviors as measured by MDS Aggression Behavior Scale (MDS ABS) in NH residents with dementia, after controlling for the other background/proximal factors 2C) To investigate the effect of pain severity on the frequency of agitated behaviors as measured by MDS challenging behavior profile (MDS CBP) agitation

PAGE 105

105 subscale in NH residents with dementia, after controlling for the other background/proximal factors Hypothesis for the second aim. Among NH residents with dementia, pa in severity will be significantly associated with increased frequency of disruptive behaviors, after controlling for the other background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, ag e, gender, marital status, education, ethnicity, hunger, thirst, bowel incontinence, and bladder incontinence). 2A) Among NH residents with dementia, pain severity will be significantly associated with increased frequency of wandering behaviors as measured by MDS wandering frequency, after controlling for the other background/proximal factors. 2B) Among NH residents with dementia, pain severity will be significantly associated with increased frequency of aggressive behaviors as measured by MDS ABS, after co ntrolling for the other background/proximal factors 2C) Among NH residents with dementia, pain severity will be significantly associated with increased frequency of agitated behaviors as measured by MDS CBP agitation subscale, after controlling for the ot her background/proximal factors Univariate and bivariate analyses are conducted first. Descriptive data are present ed, including mean s standard deviation s and ranges for continuous variables. Frequency dat a such as number and percentage, are present ed for categorical variable s Bivariate analyses are conducted based on the functional form of variables using independent t test, one way independent ANOVA, Pearson correlation, Mann Whitney U test (nonparametric version of independent t test) Kruskal Wallis test (nonparametric version of one way independent ANOVA) or correlation (nonparametric version of Pearson correlation) F unctional forms of variable are summarized in the Table 3 3

PAGE 106

106 Multivariate analyses are conducted based on t he functional form of the dependent variables. First, m ultiple ordinal least squares regression analysis is used to evaluate the effect of pain on aggression and agitation, because MDS ABS and MDS CBP agitation subscales are recorded as continuous forms, w ith higher scores indicating more aggressive and agitated behaviors. Among several kinds of multiple regression, h ierarchical multiple regression analysis is used because it is the statistical method to control covariates based on the theoretical model (Field, 2005; Polit, 2010) Covariates are entered first and the most important independent variable are entered last. For this study, pain severity score is entered in block 2, after the other background/ proximal factors ( e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, hunger, thirst, bowel incontinence, and bladder incontinence ) are entered in block 1 Nominal independent variables are recoded into dummy variables before executing multiple regression analyses If ordinal independent variables have sufficient categories, the distributions are reasonably close to being normal distribution and hav e a linear relationship with other variables, then ordinal variables are used without reco ding into dummy variables Otherwise, ordinal independent variables are also reco ded into dummy variables Influential o utliers are identified. Influential outliers that can have a strong impact on the regression solution are diagnosed and addressed To detect outliers and influential cases, standardized residuals of regression model and DfBeta in SPSS output table are examined. Standardized residuals greater than 1.9 6 are determined as outliers and a mong these outliers, i nfluential cases whose DfBeta s greater than one are addressed (Field, 2005) Multicollinearity is a problem that can occur when independent variables are highly interco rrelated, and can yield misleading regression results. To detect multicollinearity, the

PAGE 107

107 collinearity statistics are also examined. In the multicollinearity evaluation step, a variance inflation factor (VIF) less than 10 can be assumed to have no multicolli nearity (Field, 2005) If the VIF value is greater than 10, a collinearity diagnostic table will be generated and examined to detect problematic variable from model. Variables whose v ariance proportion s are greater than 0.5 can be considered as problematic variable and these variables are deleted except one variable from the model and regression will be rerun. After identifying influential outliers and multicollinearity, the assumptions of multiple regression analysis are e xamined: (a) linear relationship between continuous variables through scatterplot, null plot, and partial plot, (b) normality of the error term through histogram of residuals and Shapiro Wilk test, (c) constant variance of the error (homoscedeaticity) thro ugh null plot, and (d) independence of the error terms through Durbin Watson test. The linear relationship, not curvillinearity, is checked using scatter plots, null plots, and partial plots, if independent variables are metric. H istogram of standardized r esiduals and a Shapiro Wilk test are conducted to check the normality of error terms. If the p value of the Shapiro Wilk test is greater than .05, normal distribution will be assumed. The homoscadesiticity constant variance of the error, is examined by a null plot. Null plot will be expected to have equal variance across all predicted values and no pattern of increasing or decreasing residuals. The independence of the error terms is checked by the Durbin Watson test. T he Durbin Watson test of two represent s independence of error terms and being close to 0 or 4 represents dependence of error terms. After checking assumptions, hierarchical m ultiple regression analysis is used to estimate the adjusted R square and to calculate the F ratio, and the p value is used as the test of statistical significance. Also, information on the changes to R 2 and the significance of the changes are presented. The adjusted R 2 refers to amount of variance in the dependent variable that is

PAGE 108

108 accounted for by the independent variabl es. F ratio and p value refers to the statistical significance of the model. Standardized b eta refers to strength and direction of the relationship between independent variable and dependent variable, and indicates the number of units by standard deviation that the DV is expected to change for one standard deviation change in a predictor when the effects of all other predictors are held constant. Second, ordinal logistic regression analysis is used to evaluate the effect of pain on wandering, because MDS wa ndering frequency is recorded as an ordinal variable. Before conducting ordinal logistic regression, the assumptions for ordinal logistic regression are evaluated (empty cells between categorical variables, influential outliers and multicollinearity). Empt y or small cells are checked by executing a crosstab between categorical independent variables and the dependent variables. Influential outliers and multicollinearity that can have a strong impact on the regression solution are diagnosed and addressed. After evaluating the assumptions for logistic regression analysis, o rdinal logistic regression analysis is interpreted, using the 2 log likelihood, the Nagelkerke R square, odds ratio, and the p value to identify the statistical significance First, t he s tatistical significance of the logistic regression model is determined by the 2 log likelihood ( 2LL) If 2LL is statistically significant, then the ordinal logistic model is assumed to fit significantly better than null model which i s model with no pred ictors. Then, using the Nagelkerke R square the variance of the dependent variable will be evaluated to be explained by predictors. The Nagelkerke R square refers to the variance of the dependent variable which can be explained by independent variables, s uch as R square in a multiple regression analysis Finally, how the individual coefficients predict the dependent variable will be interpreted using odds ratio and the p value. For every unit increase in predictor, the odds of the being one unit higher in dependent variable will be evaluated by odds ratio.

PAGE 109

109 Third aim. To evaluate whether pain severity mediates the effect of background factors ( e.g., cognitive impairments, comorbid ity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, and ethnicity ) on the frequency of disruptive behaviors in NH residents with dementia. 3A) To evaluate whether pain severity mediates the effect of backgro und factors on the frequency of wandering behaviors as measured by MDS wandering frequency in NH residents with dementia. 3B) To evaluate whether pain severity mediates th e effect of background factors on the frequency of aggressive behaviors as measured b y MDS Aggression Behavior Scale (MDS ABS) in NH residents with dementia. 3C ) To evaluate whether pain severity mediates the effect of background factors on the frequency of agitated behaviors as measured by MDS challenging behavior profile (MDS CBP) agitat ion subscale in NH residents with dementia. Hypothesis for the third aim. Among NH residents with dementia, pain severity will mediate the effect of background factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL i mpairments, age, gender, marital status, education, and ethnicity) on the frequency of disruptive behaviors. 3A) Among NH residents with dementia, pain severity will mediate the effect of background factor on the frequency of wandering behaviors as measure d by MDS wandering frequency in NH residents with dementia. 3B) Among NH residents with dementia, pain severity will mediate the effect of background factors on the frequency of aggressive behaviors as measured by MDS Aggression Behavior Scale (MDS ABS) in NH residents with dementia. 3C) Among NH residents with dementia, pain severity will mediate the effect of background factors on the frequency of agitated behaviors as measured by MDS challenging behavior profile (MDS CBP) agitation subscale.

PAGE 110

110 Univariate a nd bivariate analyses are conducted first Descriptive data are present ed, including mean s standard deviation s and ranges for continuous variables. Frequency dat a, such as number and percentage, are present ed f or categorical variables. Bivariate analyses are conducted based on the functional form of variables. Functional form of variable is summarized in the Table 3 3 Path analysis using ordinal logistic regression and multiple regression analysis are executed to evaluate whether pain severity mediate s the effect of background factors on the frequency of disruptive behaviors For analysis of the mediator effect, three regressions analyses are performed. These include (1) the regression between background factor (predictor) and disruptive behavior (outcome); (2) the regression between background factor (predictor) and pain (mediator); and (3) the regression among background factor and pain as independent variables with disruptive behavior as a dependent variable (Baron & Kenny, 1986; Meyers, Gamst, & Guarino, 2006) In step one, the relatio nship between background factor (predictor) and disruptive behaviors (outcome) is estimated If the regression between background factor (predi ctor) and disruptive behavior (outcome) is statistically significant then the second step will be taken. In step two, the relationship between background factor (predictor) and pain (mediator) is estimated If the relationship between background factor (p redictor) and pain (mediator) is statistically significant, then the third step will be done. In step three, the regression among background factor and pain (independent variables) and disruptive behaviors (dependent variable) is estimat ed. If the relation ship between background factors (predictor) and disruptive behaviors (outcome) is not statistically significant and the relationship between pain (mediator) and disruptive behaviors (outcome) is significant, the mediator relationship is deemed to exist.

PAGE 111

111 Th ere should be non significant relationship between background factor (predictor) and disruptive behavior (outcome) when pain (mediator) is in the regression model, if the mediating effect is existed (Bennett, 2000; Holmbeck, 1997) Fourth aim. To investigate those three aims in the unstrained residents 4A) To describe the prevalence of disruptive behaviors in the unstrained residents 4B) To investigate the effect of pain severit y on the frequency of disruptive behaviors in the unrestrained residents after controlling for the other background/proximal factors ( e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital st atus, education, ethnicity, hunger, thirst, bowel incontinence, and bladder incontinence ). 4 C) To evaluate whether pain severity mediates the effect of background factors ( e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, and ethnicity ) on the frequency of disruptive behaviors in the unstrained residents Hypothesis for the fourth aim. 4A) Among the unstrained residents disruptive behaviors will be more prevalent. 4B) A mong the unstrained residents pain severity will be significantly associated with increased frequency of disruptive behaviors, after controlling for the other background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, hunger, thirst, bowel incontinence, and bladder incontinence). 4C) Among the unstrained residents pain severity will mediate the effect of background factors (e.g., cognitiv e impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, and ethnicity) on the frequency of disruptive behaviors. The statistical analyses for the previous three aims will be reported separately using data among the un restrained residents

PAGE 112

112 Table 3 1 Concepts, variables, measures, operational d e finitions and their i nterpretation Variables Instrument MDS items Range/interpretation Reliability/validity Pain MDS Pain severity Scale Calculated from two MDS pain items: pain frequency (MDS item # J2A) and pain intensity (MDS item # J2B) 0 (no pain) 1 (mild pain), 2 (moderate pain), 3 (excruciating pain) Inter rater reliability coefficient of 0.73, internal consistency coefficient of 0.69 0.88, test retest reliability coefficient of 0.84 0.87, criterion validity coefficient of 0.70 with Visual Analogue Scale (Cohen Mansfield, et al., 1999; Fisher, et al., 2002; Fries, et al., 2001) Hunger Complaints of Hunger Complaints of Hunger (MDS item # K4B) 0 (No), 1 (Yes) Reliability coefficient of 0.69 (Hawes, et al., 1995) Thirst Fluid intake Dehydration (MDS item # J1C) 0 (No), 1 (yes) Reliability coefficient of 0.49 (Hawes, et al., 1995) Bowel elimination Bowel continence Bowel continence (MDS item # H1A) 0 (Continent), 1 (usually continent), 2 (occasionally incontinent), 3 (frequently incontinent), 4 (incontinent) Correlation coefficient of 0.91 0.99, kappa coefficient of 0.61 (Casten, et al., 1998; Hawes, et al., 1995) Bladder elimination Bladder continence Bladder continence (MDS item # H1B) 0 (Continent), 1 (usually continent), 2 (occasionally incontinent), 3 (frequently incontinent), 4 (incontinent) Correlation coefficient of 0.91 0.99, kappa coefficient of 0.61 (Casten, et al., 1998; Hawes, et al., 1995)

PAGE 113

113 Table 3 1. Continued Variables Instrument MDS items Range/interpretation Reliability/validity Neurocogniti ve factor MDS Cognitive performance scale (CPS) Calculated from 5 MDS items: comatose (MDS item # B1), short term memory ( MDS item # B2A), cognitive skills or daily decision making ( MDS item # B4), making oneself understood ( MDS item # C4), and self performance in eating ( MDS item # G1HA). 0 (intact) 1 (borderline intact) 2 (mild impairment) 3 (moderate impairment) 4 (moderate severe impairment) 5 (severe impairment) 6 (very severe impairment) High test retest reliability, criterion validity coefficient of 0.45 0.75 with Mini M ental State Examination, c riterion validity coefficient of 0.41 0.77 against Global Deterioration Scale, criterion validity coefficient of 0.66 against Psychogeriatric Dependency Rating Scale, criterion validity coefficient of 0.45 against Mattis Dementia Rating Scale, sensitivity of 0.94, and specificity of 0.94 (Bula & Wietlisbach, 2009; Cohen Mansfield, et al., 1999; Gruber Baldini, et al., 2000; Hartmaier, et al., 1994; Hartmaier, et al., 1995; Lawton, et al., 1998; McConnell, et al., 2003; McConnell, et al., 2002; Morris, et al., 1994; Snowden, et al., 1999) Comorbidi ty Charlson Comorbidity Index Charlson C omorbidity I ndex: Calculated from the list of specific diseases or infections or other current diagnoses (MDS item # I1 or I2 or I3) 0 37, with higher scores indicating greater comorbidity MDS items for calculating Charlson comorbidity index: Reliability coefficient of 0.74 (Hawes, et al., 1995) Pressure ulcer Pressure ulcer Pressure ulcer (MDS item # M2) 0 (none) 1 (stage 1) 2 (stage 2) 3 (stage 3) 4 (stage 4) Reliability coefficient of 0.62 (Hawes, et al., 1995) Medications The number of medications The number of medications (MDS item # O1) Actual number of different medications used in the last 7 days, with higher number indicating more medications Reliability coefficient of 0.73 (Hawes, et al., 1995)

PAGE 114

114 Table 3 1. Continued Variables Instrument MDS items Range/interpretation Reliability/validity ADL MDS ADL Long Form scale The sum score of 7 MDS items: self performance of bed mobility, transfer, locomotion on unit, eating, toile use, and personal hygiene (MDS item # G1AA G1BA, G1EA, G1GA, G1HA, G1IA, and G1JA, respectively) 0 28 with higher scores indicating more ADL dependence Reliability coefficient of 0.92 0.97, Inter rater reliability coefficient of 0.61 0.95, criterion validity coefficient of 0.58 0.79 again st Physical Self Maintenance Scale (Hawes, et al., 1995; Lawton, et al., 1998; Sgadari, et al., 1997) Sociodemogr aphics Sociodemogr aphics Age (MDS item # AA3), gender (MDS item # AA2), marital status (MDS item # A5), education (MDS item # AB7), and race/ethnicity (MDS item # AA4) Wandering behaviors Wandering frequency Wandering frequency ( MDS item # E4AA): The frequency of movement with no rational purpose 0 (no wandering), 1 (wandering occurred 1 3 days in last 7 days), 2 (wandering occurred 4 6 days in last 7 days), 3 (wandering daily) Reliability coefficient of 0.63, inter rater reliability coefficient of 0.95, concurrent validity coefficient of 0.51 0.54 (Casten, et al., 1998; Hawes, et al., 1995; Lawton, et al., 1998; Sgadari, et al., 1997) Aggressive behaviors MDS Aggression behavior scale (MDS ABS) The sum score of 4 MDS items: verbally abusive behavioral symptoms ( MDS item # E4BA), physically abusive behavioral symptoms ( MDS item # E4CA ), socially inappropriate behavioral symptoms ( MDS item # E4DA), resists care ( MDS item # E4EA) 0 12, with higher scores indicating more aggressive behaviors Internal consistency reliability of 0.79 to 0.95, criterion validity coefficient of 0.72 with Cohen Mansfield Agitation Inventory aggression subscale scores (Perlma n & Hirdes, 2008) Agitated behaviors MDS Challenging Behavior Profile (MDS CBP) agitation subscale The sum score of 4 MDS items: periods of restlessness (MDS item # B5D), repetitive physical movements (MDS item # E1N), wandering (MDS item # E4AA), socially inappropriate behavior (MDS item # E4DA) 0 7, with higher scores indicating more agitated behaviors Internal consistency reliability of 0.70, inter rank correlation coefficient of 0.50 with Behavior Rating S cale for Psychogeriatric Inpatients (Gerritsen, et al., 2008)

PAGE 115

115 Table 3 2 Charlson Comorbidity Index Weight Clinical Conditions MDS item number ICD 9 Code 1 myocardial infarct MDS item # I1D congestive heart failure MDS item # I1F peripheral vascular disease MDS item # I1J cerebrovascular disease MDS item # I1T dementia MDS item # I1Q, I1U chronic pulmonary disease MDS item # I1II connective tissue disease MDS item # I3 7100, 7101, 7104, 7140, 7141, 7142, 71481, 5171, 725 peptic ulcer disease MDS item # I3 531, 532, 533, 534 mild chronic liver disease or cirrhosis MDS item # I2K diabetes MDS item # I1A 2 hemiplegia MDS item # I1V moderate or severe renal disease MDS item # I1QQ diabetes with end organ damage MDS item # I1KK tumors MDS item # I1PP leukemia MDS item # I3 204, 205, 206, 207, 208 lymphoma MDS item # I3 200, 201, 202 3 moderate or severe liver disease MDS item # I3 5722, 5723, 5724, 5728 6 metastatic solid tumor MDS item # I3 196, 197, 198, 1990, 1991 AIDS MDS item # I2D

PAGE 116

116 Table 3 3 Functional forms of v ariables Variables MDS item Range/interpretation Functional form Proximal Factors Pain MDS Pain severity scale 0 (no pain), 1 (mild pain), 2 (moderate pain), 3 (excruciating pain) Ordinal Hunger Complaints of Hunger (MDS item # K4B) 0 (No), 1 (Yes) Dichotomous categorical Thirst Dehydration (MDS item # J1C) 0 (No), 1 (yes) Dichotomous categorical Elimination Bowel continence (MDS item # H1A) 0 (Continent), 1 (usually continent), 2 (occasionally incontinent), 3 (frequently incontinent), 4 (incontinent) Ordinal Bladder continence (MDS item # H1B) 0 (Continent), 1 (usually continent), 2 (occasionally incontinent), 3 (frequently incontinent), 4 (incontinent) Ordinal Background Factors Neurocognitive factor MDS Cognitive performance scale (MDS CPS) 0 (intact) 1 (borderline intact) 2 (mild impairment) 3 (moderate impairment) 4 (moderate severe impairment) 5 (severe impairment) 6 (very severe impairment) O rdinal Comorbidity Charlson Comorbidity Index 0 37, with higher scores indicating greater comorbidity Continuous Pressure ulcer Pressure ulcer (MDS item # M2A) 0 (none) 1 (stage 1) 2 (stage 2) 3 (stage 3) 4 (stage 4) Ordinal Medications The number of medications (MDS item # O1) Actual number of different medications used in the last 7 days, with higher number indicating more medications Continuous ADL MDS ADL Long Form scale 0 28, with higher scores indicating more ADL dependence Continuous Age Age Actual age in years Continuous

PAGE 117

117 Table 3 3 Continued Variables MDS item Range/interpretation Functional form Gender Gender (MDS item # AA2) 1 (male) 2 (female) Dichotomous categorical Marital status Marital status (MDS item # A5) 1 ( never married ) 2 (married ) 3 (widowed ) 4 (separated ) 5 ( divorced ) Nominal (5 groups) Education Education (MDS item # AB7) 1 (K 11 grade ) 2 (high school ) 3 (some college ) ) 5 ( graduate degree ) Ordinal Ethnicity Ethnicity (MDS item # AA4) 1 (Indian/Asian ) 2 (African American ) 3 (Hispanic ) 4 (Caucasian ) Nominal (4 groups) Dependent Variables Wandering behaviors MDS Wandering frequency ( MDS item # E4AA) 0 (no wandering), 1 (1 3 days in last 7 days), 2 (4 6 days in last 7 days), 3 (wandering daily) Ordinal Aggressive behaviors MDS Aggression behavior scale (MDS ABS) 0 12, with higher scores indicating higher frequency Continuous Agitated behaviors MDS Challenging Behavior Profile (MDS CBP) agitation subscale 0 7, with higher scores indicating more frequency of agitated behaviors Continuous

PAGE 118

118 CHAPTER 4 RESULTS The purpose of this study was to explore the relationship between pain and disruptive behaviors (e.g., wandering, aggressive, and agitated behaviors) among NH residents with dementia T his study ha d four specific aims: (1) to describe the prevalence of disruptive behaviors in NH residents with dementia (2) to investigate t he effect of pain severity on the frequency of disruptive behaviors after controlling for the other background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnic ity, hunger, thirst, bowel incontinence, and bladder incontinence) (3) to evaluate the mediator effect of pain severity on the relationship between background factors (e.g., cognitive impairments comorbidity, pressure ulcer, number of medication s ADL im pairment, age, gender, marital status, education, and race/ethnicity ) on the frequency of disruptive behaviors, and (4) to investigate those previous three aims in the unrestrained residents Sample Selection Process The sample selection process is summari zed in Figure 4 1. First, a dmission or annual MDS data for non comatose NH residents in Florida during calendar year 2009 was obtained from Centers for Medicare and Medicaid Service (N=197,097 cases) Then, cases in which there is no diagnosis of dementia (N=130,955 cases), cases in which residents are less than 65 years old (N=2,840 cases), and duplicate cases (N= 6,725 duplicate cases ) were excluded This selection process yield s 56,577 unique cases for the analyses. Description of the Sample In this section, d escriptive statistics for the study variables are presented Analyses are organized according to the theoretical model, and presented as background factors (e.g.,

PAGE 119

119 cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity), proximal factors (e.g., pain, hunger, thirst, bowel incontinence, and bladder incontinence ), and disruptive behaviors (e.g., wandering, aggressive, and agitated behaviors) Means and standard deviations are presented for continuous variables and frequencies are presented for categorical variables. Prevalence of Background Factors B ackground factors among the study sample (N=5 6,577 ) are summarized in Table 4 1 On the average, subjects are 84 years of age (SD = 7 years) The majority of subjects are female (67.7%), widowed (53.7%), high school graduates or less (7 2 4 %), Caucasian (78.7%) and mildly or moderately cognitively impaired (54.8%) Subjects ha ve a m ean C harlson C omorbidity I ndex score of 3 (SD = 2); take a mean of 11 m edications (SD = 5); and have a mean MDS ADL impairment score of 19 (SD = 6) The prevalence of pressure ulcer is 18.1% Prevalence of Proximal Factors A description of p roximal factors (e.g., pain, hunger, thirst, bowel and bladder incontinence) among the study sample (N=56,577) is summarized in Table 4 2. The prevalence of proximal f actors is as follows : pain (36.9%), hunger ( .1% ), thirst ( .1% ), bowel incontinence ( 57.7% ) and bladder incontinence ( 64.4% ) Main Study Results Aim 1 : Prevalence of Disruptive Behaviors The first aim of this study is to describe the prevalence of disruptive behaviors (e.g., wandering, aggressive, and agitated behaviors) in NH residents with dementia The prevalence of disruptive behaviors is : wandering be haviors ( 9 .0 % ), aggressive behaviors (24. 4 %) and agitated behaviors ( 30.4 % ) (Table 4 3 )

PAGE 120

120 Aim 2: The Effect of Pain Severity on the Frequency of Disruptive Behaviors The second aim of this study is to investigate the effect of pain severity on the frequency of disruptive behaviors in NH residents with dementia, after controlling for the other background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications ADL impairments, age, gender, marital status, education, ethnicity, hunger, thirst, bowel incontinence, and bladder incontinence). The hypothesis was that pain severity would be significantly associated with increased frequency of disruptive behaviors wh en the effects of the other background/proximal factors are held constant. Analy sis approach In order to control for the other background/proximal factors, t he ordinal logistic regression analysis was used to evaluate the effect of pain severity on the frequency of disruptive behaviors ( e.g., wandering, aggressive, and agitated behaviors). Initially, ordinal logistic regression was planned to evaluate the effect of pain severity on the frequency of wandering behaviors, and multiple regression analysi s was planned to evaluate the effect of pain severity on the frequency of ag gressive and agitated behaviors ; the dependent variable wandering is expressed as an ordinal variable, and aggressive and agitated behaviors are expressed as continuous variable s However, both aggressive and agitated behaviors were severely positively skewed, and assumptions for multiple regression were violated (e.g., homoscadesiticity of error term, normal distribution of error term, and independence of error term). Also, none of the transformations available to solve positively skewed data (e.g., logarithmic transformation, square root transformation, inverse transformation, and square transformation) resolve d the normality distribution problem. Therefore, aggressive behaviors (M DS ABS scores) were collapsed into four groups : none (MDS ABS = 0), moderate (MDS ABS = 1 2), severe (MDS ABS = 3 5), and very severe (ABS= 6 12), based on an established algorithm (Perlman &

PAGE 121

121 Hirdes, 20 08) Also, agitated behaviors ( MDS CBP agitation scores) were collapsed into four groups : 0, 1, 2, 3 with higher numbers indicating more frequent agitated behaviors Since the dependent variable s were expressed in an ordinal format, ordinal logic regression was used to evaluate the effect of pain severity on the frequency of all three disruptive behaviors, after controlling for the other background/proximal factors After conducting b ivariate analyses between background/ proximal factors and disru ptive behaviors the assumptions for ordinal logistic regression were evaluated ( e.g., empty cells between categorical variables, influential outliers and multicollinearity). There are no influential outliers or multicollinearity. H unger and /or thirst var iables had empty cells in a crosstab between wandering and aggression and were subsequently excluded from the analyses. The analytical steps used to conduct and evaluate the o rdinal logistic regression analysis are outlined below. First, the statistical significance of the logistic regression model is determined by the 2 log likelihood ( 2LL). If 2LL is statistically significant, then the ordinal logistic model is assumed to fit significantly better than the null model with no pre dictors. Then, the Nagelkerke R square, is used to evaluate the variance of the dependent variable that is explained by predictors. The Nagelkerke R square refers to the variance of the dependent variable which can be explained by independent variables, an d is interpreted like an R square in a multiple regression analysis. Finally, the extent to which individual coefficients predict the dependent variable is interpreted using odds ratio and the p value. For every unit increase in the value of the predictor, the odds of the being one unit higher in the dependent variable is evaluated by odds ratio. Most of the variables have very littl e missing values (less than 1%). O ne covariate highest level of education ha d 22% missing values. For the logistic regression, listwise deletion method

PAGE 122

122 is used to handle missing values. M ultiple approaches were investigated to estimate missing values. For ordinal level variables, substitution with the subgroup mode was considered a s an acceptable approach (Polit, 2010) Thus subgroup mean (mode) substitution method is used in order to compare the precision and trends in the results. Missing values on education a re replaced using mode for ethnicity. The missing values on education for Indian/Asian and Caucasian are imputed as high school graduates, those for African American and Hispanic are imputed as less than high school graduates. U sing the same independent va riables in analysis with different dependent variables carries the risk of inflating the Type I error. To keep the overall risk of a Type I error to the 5% level, p value for the each regression analysis is set at .017 (.05/3; divided by the number of test s conducted) (Polit, 2010) B ivariate relationships between background/proximal factors and disruptive behaviors The bivariate analyses between background/proximal factors (e.g., cogni tive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, pain, hunger, thirst, bowel incontinence, and bladder incontinence) and the frequency of disruptive behaviors are summ arized in Table 4 4 and Table 4 5. For the relationship between background factors and the frequency of disruptive behaviors, all the background factors except education level are significantly related to all three disruptive behaviors. Among background fa ctors, o nly cognitive impairments are positively correlated with disruptive behaviors. Most of the background factors (e.g., Charlson Comorbidity Index, pressure ulcer, number of medications, and age) are negatively correlated with the frequency of disrupt ive behaviors. For gender, male is correlated with higher frequency of disruptive behaviors than female. For marital status, separated residents are correlated with higher frequency of disruptive behaviors the other residents. For ethnicity, African Americ an s are

PAGE 123

123 correlated with more frequent wandering or aggressive behaviors than the other ethnicity groups and Hispanics are associated with more frequent agitated behaviors For the relationship between proximal factors and the frequency of disruptive beha viors, p ain severity is negatively correlated with all three disruptive behaviors ( correlation = .0 90 .03 8 correlation = .0 43 respectively; p < .001 ), but the other proximal factors (e.g., hunger, thirst, bowel incontinence, and bladder incontinence ) show non significant or equivocal relationships with the various disruptive behaviors examined The effect of pain severity on the frequency of wandering behaviors Ordinal logistic regression analysis was used to evaluate the effect of pain severity on the frequency of wandering behaviors, after controlling for the other background/proximal factors T he 2 log likelihood, the Nagelkerke R square, odds rati o, and the p value was used to evaluate the statistical significance. The results indicated that t he ordinal logistic model fitted significantly better than the null model with no predictors ( 2 log likelihood = 29,283.252, Chi square = 4,753.104, p = .000). Nagelkerke R square showed that 19% of the variance of the wandering behaviors could be explained by these background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, pain, bowel incontinence, and bladder incontinence) Pain severity is negatively associated with the frequency of wandering behaviors, after controlling for the other background/proximal factors (OR=.75, p =.000) (Tab le 4 6). For every unit increase in MDS Pain severity score (e.g., none to mild pain, mild to moderate pain, or moderate to excruciating pain), the odds of the being one unit higher frequency in wandering behaviors (e.g., none to 1 3 days exhibition, 1 3 days exhibition to 4 6 days exhibition, or 4

PAGE 124

124 6 exhibition to daily wandering) is 25% decreased, after controlling for the other background/proximal factors in the model The effect of pain severity on the frequency of aggressive behaviors Ordinal logistic regression analysis was used to evaluate the effect of pain severity on the frequency of aggressive behaviors after controlling for the other background/proximal factors The results indicated that t h e ordinal logistic regression model fitted significantly better than the null model with no predictors ( 2 log likelihood = 60,707.294, Chi square = 2824.689, p = .000). Nagelkerke R square showed that 8.2% of the variance of the aggressive behaviors could be explained by these background / proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, pain, hunger, bowel incontinence, and bladder incontinen ce) Pain severity is positively associated with the frequency of aggressive behaviors, after controlling for the other background/proximal factors (OR=1.05, p =.010) (Table 4 7) For every unit increase in MDS Pain severity score (e.g., none to mild pain mild to moderate pain, or moderate to excruciating pain) the odds of the being one unit higher frequency in aggressive behaviors (e.g., none to moderate, moderate to severe, or severe to very severe frequent behaviors) is increased by 5 %, after controll ing for the other background/proximal factors in the model The effect of pain severity on the frequency of agitated behaviors Ordinal logistic regression analysis was also used to evaluate the effect of pain severity on the frequency of agitated behaviors, after controlling for the other background/proximal factors The ordinal logistic regression model fitted significantly better than null model with no predictors ( 2 log l ikelihood = 76,952.644, Chi square = 5,173.731, p = .000). Nagelkerke R square showed that 13% of the variance of the aggressive behaviors could be explained by these

PAGE 125

125 background/ proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, nu mber of medications, ADL impairments, age, gender, marital status, education, ethnicity, pain, hunger, thirst, bowel incontinence, and bladder incontinence) Pain severity is positively associated with the frequency of agitated behaviors, after controlling for the other background/proximal factors (OR=1.06, p =.000) (Table 4 8). For every unit increase in MDS Pain severity score (e.g., none to mild pain, mild to moderate pain, or moderate to excruciating pain), the odds of the being one unit higher frequenc y in agitated behaviors is increased by 6%, after controlling for the other background/proximal factors. The statistical results of the model using imputation methods To determine whether the extent of missing data on the education variable and the listwise deletion influenced the study results, the analyses were rerun using subgroup mean (mode) substitution These results are summarized in Table 4 9 4 10 and 4 11. The statistical results are almost identical between the model using listwise deleti on and the model using subgroup mean (mode) substitution. In both models, pain severity is negatively associated with the frequency of wandering behaviors but positively associated with the frequency of aggressive and agitated behaviors after controlling for the other background/proximal factors. Summary of the results of the Aim 2 Pain severity is negatively associated with the frequency of disruptive behaviors in .090 .038; p < .001) In multivariate ordi nal logistic regression, pain severity is negatively associated with the frequency of wandering behaviors (OR=.75, p =.000) but positively associated with the frequency of aggressive and agitated behaviors (OR=1.05, p =.010; OR=1.06, p =.000; respectively), after controlling for the other background/proximal factors The hypothes is that pain severity would be significantly associated with increased frequency of wandering behaviors is not supported, but the hypothes is

PAGE 126

126 that pain severity would be significantly associated with increased frequency of aggressive or agitated behaviors is supported. Aim 3 : Mediating Effect of Pain Severity on Disruptive Behaviors The third aim of this study is to evaluate whether pain severity mediates the effe ct of background factors (e.g., neurocognitive factors, comorbidity, pressure ulcer, number of medication s ADL impairment, age, gender, marital status, education, and race/ethnicity) on the frequency of disruptive behaviors (e.g., wandering, aggression, and agitation). The hypothesis was that pain severity would mediate the relationship between background factors and disruptive behaviors. Analysis approach Path analysis using a series of ordinal logistic regression analyses was used to evaluate this medi ation model Initially, path analysis using ordinal logistic regression was planned to evaluate the mediating effect of pain on the frequency of wandering behaviors, and path analysis using multiple regression was planned for the models with aggressive and agitated behaviors because the dependent variable wandering is expressed as an ordinal variable and aggressive and agitated behaviors are expressed in a continuous variable However, both aggressive and agitated behaviors were severely positively skewed, and assumptions for multiple regression were violated ( e.g., homoscadesiticity of error term, normal distribution of error term, independence of error term). Also, none of the transformations available to solve positively skewe d data (e.g., logarithmic transformation, square root transformation, inverse transformation, and square transformation) resolve d the normality distribution problem. Therefore, aggressive and agitated behaviors were collapsed into four groups as in the pr evious analysis for the second aim Since the dependent variable is expressed as an ordinal variable ordinal logic regression is used for

PAGE 127

127 path analysis to evaluate the mediati ng effect of pain severity on the relationship between background factors and th e frequency of all three disruptive behaviors. After conducting bivariate analysis between background factors and pain severity or the frequency of disruptive behaviors, the assumptions for ordinal logistic regression are evaluated ( e.g., empty cells between categorical variables, influential outliers and multicollinearity). There are no violations of the assumptions. To evaluate the potential role of pain as a mediator, three ordinal logistic regressions analyses are performed following t he approach outlined by Baron and Kenny (1986) These include (1) the regression between background factor (predictor) and disruptive behavior (outcome) (2) the regression between background factor (predictor) and pain (mediator) and (3) the regression a mong background factor and pain as independent variables with disruptive behavior as a dependent variable (Baron & Kenny, 1986; Holmbeck, 1997) In step one, the relationship between background factor (predictor) and disruptive behaviors (outcome) is estimated. If the regression between background factor (predictor) and disruptive behavior (outcome) is statistically significant, then the second step will be taken. In step two, the relationship between background factor (predicto r) and pain (mediator) is estimated. If the relationship between background factor (predictor) and pain (mediator) is also statistically significant then the third step can be done. In step three, the regression among background factor and pain (independent variables) and disruptive behaviors (dependent variable) is estimated. If the relationship between background factors (predictor) and disruptive behaviors (outcome) is not statisti cally significant and the relationship between pain (mediator) and disruptive behaviors (outcome) is significant ( p < .05), the mediator relationship is deemed to exist. There should be non significant relationship between background factors (predictor) a nd disruptive

PAGE 128

128 behaviors (outcome) when pain (mediator) is in the regression model (Bennett, 2000; Holmbeck, 1997) During the se three steps, the next step cannot be done if the pre vious step does not show significant result U sing the same independent variables in analysis with different dependent variables carries the risk of inflating the Type I error. To keep overall risk of a Type I error to the 5% level, p value for the each re gression analysis is set at .017 (.05/3; divided by the number of tests conducted) (Polit, 2010) B ivariate analyses between background factors and pai n severity or frequency of disruptive behaviors The bivariate analyses between background factors and pain severity are summarized in Table 4 4 For the relationship between background factors and the pain severity, a ll the background factors are significa ntly correlated with pain severity except Charlson Comorbidity Index. Pain severity is positively correlated with pressure ulcer, number of medications, and education level. However, pain severity is negatively correlated with cognitive impairments, ADL i mpairments, and age. For gender, female is correlated with higher pain severity than male. For marital status, married or widowed residents are correlated with higher pain severity than never married residents. For ethnicity, Caucasian residents are correlated with higher seve rity of pain than non Caucasian. For the relationship between background factors and the frequency of disruptive behaviors and between pain intensity and the frequency of disruptive behaviors refer to bivariate analyses in the aim 2 (page 124) The mediating e ffect of pain severity on the r elationship between background factors and the frequency of wandering b ehaviors The mediating effect of pain severity on the relationship between background factors and the frequency of wander ing behaviors is summarized in Table 4 12

PAGE 129

129 Cognitive impairments and wandering The mediating effect of pain severity on the relationship between cognitive impairments on the frequency of wandering behaviors is examined using a series of ordinal logistic regression analyses (Holmbeck, 1997) The first step of path analysis, regression analysis of cognitive impairment ( predictor ) on wandering ( outcome ) is statistical ly significant ( OR= 1.32, p =. 000 ) The second step of path analysis, regression analysis of cognitive impairment ( predictor ) on pain severity ( mediat or ) is statistically significant ( OR= .73, p =. 000 ) The third step of path analysis, regression analysis of cognitive impairment and pain severity ( predictor and mediator ) on wandering ( outcome ) is statistically significant ( cognitive impairment: OR= 1.27, p =. 000; pain severity : OR= .64, p =. 000 ) T he relationship between cognitive impairments (predictor) and wandering behaviors (outcome) is still significant, after controlling for pain severity (mediator) There should be non significant relationship between cognitive impairments (predictor) and wan dering (outcome) when pain severity ( mediator ) is in the regression model if the mediating effect exist s (Bennett, 2000; Holmbeck, 1997) Therefore there is no mediator effect of pain severity on the relationship between cogniti ve impairment and the frequency of wandering behaviors. Charlson comorbidity index and wandering The mediating effect of pain severity on the relationship between Charlson Comorbidit y Index ( CCI ) and the frequency of wandering behaviors is examined u sing a series of ordinal logistic regression. The first step of path analysis, regression analysis of CCI (predictor) on wandering (outcome) is statistically significant (OR= .77, p =.000). The second of path analysis, regression analysis of CCI (predictor) on pain severity (mediator) is not statistically significant (OR= 1.00, p =.400). The third step of path analysis regression analysis of CCI and pain severity (predictor and med iator ) on wandering ( outcome ) c annot be executed because the second step show s a non significant relationship (Bennett, 2000;

PAGE 130

130 Holmbeck, 1997) There is no mediator effect of pain severity on the relationship between CCI and the frequency of wandering behaviors. Pressure ulcer and wandering. The mediating effect of pain severity on the relationship between pressure ulcer on the frequency of wandering behaviors is examined using a s eries of ordinal logistic regression. The first step of path analysis, regression analysis of pressure (predictor) on wandering (outcome), is statistically significant (OR= 54 p =.000). The second step of path analysis, regression analysis of pressure ul cer (predictor) on pain severity (mediator) is statistically significant (OR= 1. 24 p =. 0 00). The third step of path analysis, regression analysis of pressure ulcer and pain severity (predictor and mediator) on wandering (outcome) is statistically significant ( pressure ulcer: OR= .56, p =. 000; pain severity : OR= .60, p =. 000 ) T he relationship between pressure ulcer (predictor) and wandering behaviors (outcome) is still significant, after controlling for pain severity (mediator) There should be no n significant relationship between pressure ulcer (predictor) and wandering (outcome) when pain severity (mediator) is in the regression model if the mediating effect exist s (Bennett, 2000; Holmbeck, 1997) Therefore there is no mediator effect of pain severity on the relationship between pressure ulcer and the frequency of wandering behaviors. Number of medications and wandering. The med iating effect of pain severity on the relationship between number of medications and the frequency of wandering is examined. The first step of path analysis, regression analysis of number of medications (predictor) on wandering (outcome) is statistically significant (OR= 94 p =.000). The second step of path analysis, regression analysis of number of medications (predictor) on pain severity (mediator) is statistically significant (OR= 1. 12 p =. 0 00). The third step of path analysis, regres sion analysis of number of medications and pain severity (predictor and mediator) on wandering (outcome) is

PAGE 131

131 statistically significant ( number of medications: OR= .95, p =. 000; pain severity : OR= .62, p =. 000 ) T he relationship between number of medication s (predictor) and wandering behaviors (outcome) is still significant, after controlling for pain severity (mediator) There should be non significant relationship between number of medications (predictor) and wandering (outcome) when pain severity (mediator) is in the regression model if the mediating effect exists (Bennett, 2000; Holmbeck, 1997) Therefore there is no mediator effect of pain severity on the relationship b etween pressure ulcer and the frequency of wandering behaviors. ADL impairments and wandering. The mediating effect of pain severity on the relationship between AD L i mpairments and the frequency of wandering is examined. The first step of path analysis, regression analysis of ADL impairments (predictor) on wandering (outcome) is statistically significant (OR= 92 p =.000). The second step of path analysis, regression analysis of ADL impairments (predictor) on pain severi ty (mediator) is not statistically significant (OR= 1.00, p =.304). The third step of path analysis, regression analysis of ADL impairments and pain severity (predictor and mediator) on wand ering (outcome) cannot be executed because the second step show s non significant relationship (Bennett, 2000; Holmbeck, 1997) There is no mediator effect of pain severity on the relationship between ADL impairments and the frequency of wanderi ng behaviors. Age and wandering. The mediating effect of pain severity on the relationship between age and the frequency of wandering is examined. The first step of path analysis, regression analysis of age (predictor) on wandering (outcome) is statistically significant (OR= 98 p =.000). The second step of path analysis, regression analysis of age (predictor) on pain severity (mediator) is statistically significant (OR= 99 p =. 0 00). The third step of path analysis, regression analysis of ag e and pain severity (predictor and mediator) on wandering (outcome) is statistically

PAGE 132

132 significant ( age: OR= .98, p =. 000; pain severity : OR= .56, p =. 000 ) T he relationship between age (predictor) and wandering behaviors (outcome) is still significant, after controlling for pain severity (mediator) There should be non significant relationship between age (predictor) and wandering (outcome) when pain severity (mediator) is in the regression model if the mediating effect exists (Bennett, 2000; Holmbeck, 1997) Therefore there is no mediator effect of pain severity on the relationship between age and the frequency of wandering behaviors. Gender and wandering. The mediating effect of pain severity on the relationship between gender and the frequency of wandering is examined. The first step of path analysis, regression analysis of gender (predictor) on wandering (outcome) is statistically significant ( male: OR= 1.24 p =.000). The second step of path analysis, regression analysis of gender (predictor) on pain severity (mediator) is statistically significant ( male: OR= 80 p =. 0 00). The third step of path analysis, regression analysis of gender and pain severity (predictor and mediator) on wandering (outcome) is statistically significant ( male: OR= 1.20, p =. 000; pain severity : OR= .57, p =. 000 ) T he relationship between gender (predictor) and wandering behaviors (outcome) is still significant, after controlling for pain severity (mediator) There should be non significant relationship between gender (predictor) and wandering (outcome) when pain severity (mediator) is in the regression model if the mediating effect exists (Bennett, 2000; Holmbeck, 1997) Therefore there is no mediator effect of pain severity on the relationship between gender and the frequency of wandering behaviors. Marital status and wandering The mediating effect of pain severity on the relationship between marital status on the frequency of wandering behaviors. Marital status is dichotomized for the analysis, married versus not married. The first step of path analysis, regression an alysis of marital status (predictor) on wandering (outcome) is statistically significant ( married: OR= 1.19

PAGE 133

133 p =.000). The second step of path analysis, regression analysis of married status (predictor) on pain severity (mediator) is statistically significant ( married: OR= 1.06 p =. 0 0 2 ). The third step of path analysis, regression analysis of marital status and pain severity (predictor and mediator) on wandering (outcome) is statistically significant ( married: OR= 1.21, p =. 000; pai n severity : OR= .57, p =. 000 ) T he relationship between marital status (predictor) and wandering behaviors (outcome) is still significant, after controlling for pain severity (mediator) There should be non significant relationship between marital status (predictor) and wandering (outcome) when pain severity (mediator) is in the regression model if the mediating effect exists (Bennett, 2000; Holmbeck, 1997) Therefore there is no mediator effect of pain severity on the relationship between marital status and the frequency of wandering behaviors. Education level and wandering behaviors. T he mediating effect of pain severity on the relationship between educational level and the frequency of wandering is examined. The first step of path analysis, regression analysis of level of highest education (predictor) on wandering (outcome) is not statistically significant (OR= 1.00 p =. 855 ). The next step of path analysis cannot be executed because the first step show s non significant relationship (Bennett, 2000; Holmbeck, 1997) There is no mediator ef fect of pain severity on the relationship between level of education and the frequency of wandering behaviors Ethnicity and wandering behaviors. The mediating effect of pain severity on the relationship between ethnicity on the frequency of wandering behaviors is examined using a series of ordinal logistic regression. Ethnicity is dichotomized for the analysis, minority and Caucasian. The first step of path analysis, regression analysis of ethnicity (predictor) on wandering (outcome) is stat istically significant ( minority: OR= .89 p =.0 01 ). The second step of path analysis, regression analysis of ethnicity (predictor) on pain severity (mediator) is

PAGE 134

134 statistically significant ( minority: OR= .62 p =. 0 0 0 ). The third step of path analysis, regression analysis of ethnicity and pain severity (predictor and mediator) on wandering (outcome) is statistically significant ( minority: OR= .82, p =. 000; pain severity : OR= .56, p =. 000 ) T he relationship between ethnicity (pred ictor) and wandering behaviors (outcome) is still significant, after controlling for pain severity (mediator) There should be non significant relationship between ethnicity (predictor) and wandering (outcome) when pain severity (mediator) is in the regres sion model if the mediating effect exists (Bennett, 2000; Holmbeck, 1997) Therefore there is no mediator effect of pain severity on the relationship between ethnicity and the frequency of wandering behaviors. In summary, there is no evidence of a mediating effect of pain severity on the relationship between any of the background factors and the frequency of wandering behaviors (Table 4 12 ) The mediating effect of pain severity on the relationship between background factors on the frequency of agg ressive behaviors The same analytical approach was used to examine t he mediating effect of pain severity on the relationship between background factors and the frequency of aggressive behaviors ( Table 4 1 3 ) T here is no evidence of a mediating effect of pa in severity on the relationship s between background factors and the frequency of aggressive behaviors. Refer to Table 4 1 3 for details on the mediating effect of pain severity on the relationship between any of the background factors and the frequency of aggressive behaviors. The mediating effect of pain severity on the relationship between background factors on the frequency of agitated behaviors The same analytical approach was used to examine the mediating effect of pain severity on the relationship between background factors and the frequency of ag itated behaviors (Table 4 14 ). T here is no evidence of a mediating effect of pain severity on the relationships between any of the background factors and the frequency of agitated behaviors. Refer to Table 4 1 4 for details

PAGE 135

135 on the mediating effect of pain severity on the relationship between background factors and the frequency of agitated behaviors. Summary of the result s of the a im 3 There is no mediating effect of pain severity on the relationship between background factors and the frequency of disruptive behaviors. The hypothes i s that pain severity would mediate the effect of background factors on the frequency of disruptive behaviors is not supported. Summary of Findings For the first aim the prevalence of disruptive behaviors ranges from 9 30% (wandering: 9.0%, aggression: 24.4%, and agitation: 30.4%) For the second aim, pain severity is negatively associated with the frequency of wandering behaviors (OR =.75, p =.000), but positively associated with the frequency of aggressive and agitated behaviors (OR=1.05, p =.010; OR=1.06, p =.000; respectively), after controlling for the other background/proximal factors The hypothes is that pain severity would be si gnificantly associated with increased frequency of wandering behaviors is not supported, but the hypothes is that pain severity would be significantly associated with increased frequency of aggressive or agitated behaviors is supported. For the third aim, no mediating effect of pain severity is found The hypothes is that pain severity would mediate the effect of background factors on the frequency of disruptive behaviors is not supported. Study Results in Unrestrained Res idents All of the prior results rega rding the relationship between pain severity and the frequency of disruptive behaviors are examined without controlling for the use of physical or chemical restraint s However, r esidents with restraints may not exhibit disruptive behaviors even when they have pain. Therefore, the study aims were tested again in the unrestrained residents (N =

PAGE 136

136 12,991), after excluding residents who are restrained Means and standard deviations are presented for continuous variables and f requencies are presented for categorical variables. The number and percentage of restrained residents among 56,577 NH residents with dementia are summarized in Table 4 15. There are 43,585 residents (77%) who had physical restraints or psychoactive medicat ions during the last 7 days P hysical restraints were used in 15,300 residents (27%) such as trunk restraints, limb restraints, restraint chair, and bed rails. Also, t here are 39,141 residents (69.2%) had psychoactive medications such as antipsychotics, antianxiety, antidepressant, and hypnotics. Prevalence of Background Factors in Unrestrained Residents The prevalence of b ackground factors (e.g., age, gender, marital status, education, ethnicity, cognition, Charlson Comorbidity Index score, pressure ulcer, number of medications, and ADL impairments) among unrestrained residents (N = 12,991) is summarized in Table 4 1 6 Subjects in the subsample have a mean age of 85 years (SD = 7 years). The majority of subjects in the subsample are female (66.7%), widowed (54.4%), high school graduates or less ( 72.1 %), Caucasian (77.2%) and mildly or moderately cognitively impaired (5 3.1 %) Subjects in the subsample have an average Charlson Comorbidity Index score of 3 (SD = 2); take a mean of 9 medicati ons (SD = 5); and have an average MDS ADL impairment score of 19 (SD = 7). The prevalence of pressure ulcer in the unrestrained residents is 20.7%. Prevalence of Proximal Factors in Unrestrained Residents The prevalence of p roximal factors among unrestrained residents (N = 12,991) is summarized in Table 4 1 7 The prevalence of proximal factors in the unrestrained residents is as follows : pain (35.1%), hunger (.1%), thirst (< .1%), bowel incontinence (61.8%), and bladder incontinence (67.3%).

PAGE 137

137 Preva lence of Disruptive Behaviors in Unrestrained Residents (Aim 1) The prevalence of disruptive behaviors among unrestrained residents is as follows : wandering behaviors (5.0%), aggressive behaviors ( 1 4 .2 %), and agitated behaviors ( 16.9 %) (Table 4 1 8 ) In contrast to the hypothesis that disruptive behaviors would be more prevalent among the unrestrained residents, t he prevalence of these behaviors is lower than it is among the total sample. The prevalence of wandering decreased from 9% to 5%, aggressive behaviors decreased from 24.4% to 14.2%, and agitated behaviors decreased from 30.4% to 16.9% in the subsample The Effect of Pain on Disruptive Behaviors in Unrestrained Residents (Aim 2) Among unrestrained residents the effect of pain severity on the f requency of disruptive behaviors, after controlling for the other background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, hunger, thir st, bowel incontinence, and bladder incontinence ), were examined using ordinal logistic regression. The hypothesis was that pain severity would be significantly associated with increased frequency of disruptive behaviors, after controlling for the other ba ckground/proximal factors. After conducting bivariate analyses, the assumptions for ordinal logistic regression were evaluated ( e.g., empty cells between categorical variables, influential outliers and multicollinearity). Hunger and thirst were excluded from the ordinal logistic analysis because there are empty cells in a crosstab between these variables and disruptive behaviors. There are no influential outliers or multicollinearity problems. For the logistic regression, listwise deletion metho d was used to handle missing values. A lso subgroup mean (mode) substitution method was used in order to compare the precision and trends in the results Missing values on education were replaced using the mode separately for ethnicity. The missing values on education for Indian/Asian and Caucasian are i mputed as high school

PAGE 138

138 graduates; those for African American and Hispanic are imputed as less than high school graduates. To keep overall risk of a Type I error to the 5% level, p value for the each regressio n analysis is set at .017 (.05/3; divided by the number of tests conducted) (Polit, 2010) B ivariate analyses between background/proximal factors and the frequency of disruptive behavi ors in the unrestrained residents The bivariate analyse s between background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, pain, hunger, thirst, bowel incontinence, and bladder incontinence) and the frequency of disruptive behaviors are summarized in Table 4 1 9 and Table 4 20 For the relation ship between background factors and the frequency of disruptive behaviors, only cognitive impairments are positively correlated with disruptive behaviors. Charlson Comorbidity Index, pressure ulcer, and number of medications are negatively correlated with disruptive behaviors. The other background factors (e.g., ADL impairments, age, gender, marital status, education, ethnicity, pain, hunger, thirst, bowel incontinence, and bladder incontinence) are not significantly correlated with at least one of disrupti ve behaviors. For the relationship between proximal factors and the frequency of disruptive behaviors, pain severity is negatively correlated with the frequency of all three disruptive behaviors .078, aggression: Spear .031, .035, respectively; p < .001), but the other proximal factors (e.g., hunger, thirst, bowel incontinence, and bladder incontinence) show non significant or equivocal relationships with the var ious disruptive behaviors examined This result is similar to the previous results using all the subjects.

PAGE 139

139 The effect of pain severity on the frequency of wandering behaviors in the unrestrained residents In the multivariate ordinal logistic regression p a in severity is shown to be negatively associated with the frequency of wandering behaviors (OR = .68, p = .000) after controlling for the other background/proximal factors (Table 4 2 1 ). For every unit increase in MDS Pain severity score (e.g., none to mild pain, mild to moderate pain, or moderate to excruciating pain), the odds of the being one unit higher frequency in wandering behaviors (e.g., none to 1 3 days exhibition, 1 3 days exhibition to 4 6 days exhibition, or 4 6 exhibition to daily w andering) is 32% decreased, after controlling for the other background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, bowel incontinence and bladder incontinence). The effect of pain severity on the frequency of aggressive behaviors in the unrestrained residents In the multivariate ordinal logistic regression, pain severity is shown to be positively associated with the frequency of aggres sive behaviors (OR = 1 10 p = .0 32 ), after controlling for the other background/proximal factors (Table 4 2 2 ). For every unit increase in MDS Pain severity score (e.g., none to mild pain, mild to moderate pain, or moderate to excruciating pain), the odds of the being one unit higher frequency in aggressive behaviors (e.g., none to moderate, moderate to severe, o r severe to very severe frequent behaviors) is increased by 10%, after controlling for the other background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, edu cation, ethnicity, pain, bowel incontinence, and bladder incontinence).

PAGE 140

140 The effect of pain severity on the frequency of agitated behaviors in the unrestrained residents In the multivariate ordinal logistic regression pain severity is shown to be negativel y associated with the frequency of agitated behaviors (OR = 1 1 1 p = .0 16 ), after controlling for the other background/proximal factors (Table 4 2 3 ). For every unit increase in MDS Pain severity score (e.g., none to mild pain, mild to moderate pain, or mo derate to excruciating pain) the odds of the being one unit higher frequency in agitated behaviors is increased by 1 1 %, after controlling for the other background/proximal factors (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medica tions, ADL impairments, age, gender, marital status, education, ethnicity, pain, bowel incontinence, and bladder incontinence). The statistical results of the model using imputation methods The statistical results after imputing education using subgroup me an (mode) substitution are summarized in Table 4 2 4 4 2 5 and 4 2 6 The statistical results are essentially unchanged between the model s using listwise deletion and the model s using subgroup mean (mode) substitution. In both models, pain severity is negatively associated with the frequency of wandering behaviors but positively associated with the frequency of aggressive and agitated behaviors after controlling for the other background/proximal factors. Summary of the results of the aim 2 among the u nrestrained residents Pain severity is negatively associated with the frequency of disruptive behaviors in .078 .031; p < .001). In multivariate ordinal logistic regression, pain severity is negatively assoc iated with the frequency of wandering behaviors (OR = .68, p = .000), but positively associated with the frequency of aggressive and agitated behaviors (OR = 1.10, p = .032; OR = 1.11, p = .016; respectively), after controlling for the other background/proximal factors. The hypothesis that pain severity would be significantly

PAGE 141

141 associated with increased frequency of wandering behaviors is not supported, but the hypothesis that pain severity would be significantly associated with increased frequency of aggressive or agitated behaviors is supported. Mediating Effect of Pain in the Unrestrained Residents ( Aim 3) Among unrestrained residents the mediating effect of pain severity on the relationship between background factors and the frequency of disruptive behaviors was examined using p ath analysis composed of a series of ordinal logistic regression analyses. The hypothesis was that pain severity would mediate the effect of background factors on the frequency of disrup tive behaviors. After conducting bivariate analysis between background factors and pain severity or the frequency of disruptive behaviors, the assumptions for ordinal logistic regression (empty cells between categorical variables, influential outliers and multicollinearity) are evaluated. There are no violations of the assumptions. To keep overall risk of a Type I error to the 5% level, p value for the each regression analysis is set at .017 (.05/3; divided by the number of tests conducted) (Polit, 2010) B ivariate analyses between background factors and pain severity or frequency of disruptive behaviors in the unrestrained residents The bivariate analys es between background factors and pain severity are summarized in Table 4 1 9 For the relationship between background factors and pain severity, a ll the background factors are significantly correlated with pain severity. Pain severity is positively correlated with pressure ulcer, number of medications, and education level. However, pain severity is negatively correlated with cognitive impairments, C harlson C omorbidity I ndex ADL impairments, and age. For gender, female is correlated with hig her severity of pain than male For marital status, married or widowed or divorced residents are more highly correlated with

PAGE 142

142 higher pain severity than never married residents. For ethnicity, Caucasian residents are correlated with higher severity of pai n than African American residents For the relationship between background factors and the frequency of disruptive behaviors and between pain intensity and the frequency of disruptive behaviors, refer to bivariate analyses in the aim 2 (page 139) M ultiva riate analyses for mediating effect of pain severity in the unrestrained residents The mediating effects of pain on the relationship between background factors on disruptive behaviors (wandering, aggression, and agitation) are examined using a series of or dinal logistic regression analyses (Baron & Kenny, 1986; Holmbeck, 1997) T he mediating effect s of pain on the relationship between background factors and disruptive behaviors (e.g., wandering, aggressive, and agitated behaviors) among unrestrained residents are summarized in the Table 4 2 7 Table 4 2 8 and Table 4 2 9 Sum mary of the results of the aim 3 among the unrestrained residents There is no mediating effect of pain severity on the relationship between background factors and the frequency of disruptive behaviors in unrestrained residents The hypothesis that pain severity would mediate the effect of background factors on the frequency of disruptive behaviors in unrestrained residents is not supported. Summary of Findings among the Unrestrained Residents The results using data from unrestrained residents are similar to t he o nes among all the NH residents except th at the overall prevalence of disruptive behaviors is lower in the subsample without restraints First, the prevalence of wandering decreased from 9% to 5%, aggressive behaviors decreased from 24.4% to 14.2%, and agi tated behaviors decreased from 30.4% to 16.9% in the subsample. The hypothesis that disruptive behaviors would be more prevalent among the unrestrained residents is not supported. Second, p ain severity is still negatively

PAGE 143

14 3 associated with the frequency of w andering behaviors ( OR = .68, p = .000 ), but positively associated with the frequency of aggressive and agitated behaviors ( OR = 1.10, p = .032; OR = 1.11, p = .016; respectively ), after controlling for the other background/proximal factors The hypothesis that pain severity would be significantly associated with increased frequency of wandering behaviors is not supported, but the hypothesis that pain severity would be significantly associated with increased frequency of aggressive or agitated behaviors is supported. Third there are still no mediating effect s of pain severity on the relationship between background factors and disruptive behaviors. The hypothes is that pain severity would mediate the effect of background factors on the frequency of disruptive behaviors is not supported.

PAGE 144

144 Figure 4 1. Sample selection process Admission or annual MDS data for non comatose residents in Florida during calendar year 2009 (N=197,097 cases) Residents with dementia (N=66,142 cases) Residents with dementia and age over 65 years (N=63,302 cases) Final sample: N=56,577 unique cases N=130,955 cases excluded: no diagnosis of dementia N=2,840 cases excluded: age less than 65 years old N= 6 ,725 duplicate cases excluded: 5,243 residents ha d 2 records ( 5,243 cases) 594 residents ha d 3 records (1, 188 cases) 83 residents ha d 4 records (2 49 cases) 10 residents ha d 5 records ( 4 0 cases) 1 resident ha d 6 records ( 5 cases)

PAGE 145

145 Table 4 1. Sample description : Background factors (N=56,577) Characteristics Possible Range N n (%) M ( SD) Actual Range Age (years) 56,577 84 (7) 65 109 Gender 56,566 Male 18,265 (32.3) Female 38,301 (67.7) Marital status 55,920 Never married 3,994 (7.1) Married 16,937 (30.3) Widowed 30,038 (53.7) Separated 659 (1.2) Divorced 4,292 (7.7) Education 44,295 K 11 grades 10,916 (24.6) High school 21,176 (47.8) Some college 7,296 (16.5) 3,189 (7.2) Graduate degree 1,718 (3.9) Ethnicity 56,392 Indian /Asian 291 (. 5 ) African American 5,944 (10.5) Hispanic 5,791 (10.3) Caucasian 44,366 (78.7) MDS C ognitive P erformance S cale (MDS CPS) 0 6 56,543 Intact 3,446 (6.1) Borderline intact 3,195 (5.7) Mild impairment 9,314 (16.5) Moderate impairment 21,657 (38.3) Moderate severe impairment 7,650 (13.5) Severe impairment 5,980 (10.6) Very severe impairment 5,301 (9.4) Charlson Comorbidity Index 0 37 56,562 3 (2) 1 16 Pressure ulcer 0 4 56,576 None 46,349 (81.9) Stage I 4,351 (7.7) Stage II 3,307 (5.8) Stage III 649 (1.1) Stage IV 1,920 (3.4) Number of medications 56,576 11 (5) 0 90 MDS ADL impairment score 0 28 56,577 19 (6) 0 28

PAGE 146

146 Table 4 2 Sample description : Proximal factors (N=56,577) Characteristics Possible Range N n (%) MDS Pain 0 3 56,568 No pain 35,710 (63.1) Mild pain 15,139 (26.8) Moderate pain 5,227 (9.2) Excruciating pain 492 (.9) Hunger 0 1 56,573 No 56,525 (99.1) Yes 48 (.1) Thirst 0 1 56,576 No 56,538 (99.1) Yes 38 (.1) Bowel incontinence 0 4 56,574 Continent 20,543 (36.3) Usually continent 3,405 (6.0) Occasionally incontinent 3,460 (6.1) Frequently incontinent 6,476 (11.4) Incontinent 22,690 (40.1) Bladder incontinence 0 4 56,577 Continent 16,660 (29.4) Usually continent 3,490 (6.2) Occasionally incontinent 4,591 (8.1) Frequently incontinent 9,501 (16.8) Incontinent 22,335 (39.5)

PAGE 147

147 Table 4 3 Prevalence of disruptive behaviors among all the subjects (N=56,577) Characteristics Possible Range N n (%) M (SD) Actual Range Wandering 0 3 56,573 No wandering 51,463 (91.0) 1 3 days in 7 days 2,637 (4.7) 4 6 days in 7 days 994 (1.8) Wandering daily 1,479 (2.6) Aggression 0 12 56,572 .54 (1.27) 0 12 None 42,764 (75.6) Moderate 9 667 (17.1) Severe 3 390 (6.0) Very severe 751 (1.3) Agitation 0 7 56,570 .61 (1.12) 0 7 0 39,345 (69.6) 1 7 575 (13.4) 2 5 075 (9.0) 3 7 4 575 (8.0)

PAGE 148

148 Table 4 4 Bivariate analysis between background factors, pain and disruptive behaviors Variables Pain Wandering Aggression Agitation Cognitive Impairments r s = 219 p = .000 N=56,5 34 r s = .138 p = .000 N=56,540 r s = .202 p = .000 N=56,539 r s = .280 p = .000 N=56,537 Charlson Comorbidity Index r s = 007 p = .111 N=56,5 53 r s = .102 p = .000 N=56,558 r s = .023 p = .000 N=56,557 r s = .085 p = .000 N=56,555 Pressure ulcer r s = .116 p = .000 N=56,5 67 r s = .094 p = .000 N=56,572 r s = .031 p = .000 N=56,571 r s = .036 p = .000 N=56,569 Number of medication r s = 251 p = .000 N=56,5 67 r s = .085 p = .000 N=56,572 r s = .018 p = .000 N=56,571 r s = .037 p = .000 N=56,569 ADL Impairments r s = 017 p = .000 N=56,5 68 r s = .187 p = .000 N=56,573 r s = .044 p = .000 N=56,572 r s = .017 p = .000 N=56,570 Age r s = 027 p = .000 N=56,5 68 r s = .047 p = .000 N=56,573 r s = .032 p = .000 N=56,572 r s = .040 p = .000 N=56,570 Gender U = 3 68,448,792 p =.000 N=56,5 57 U = 343,233,726 p =.000 N=56,562 U = 333,909,726 p =.000 N=56,561 U = 336,350,737 p =.000 N=56,559 Male Median=.00 M (SD)=.43 (.67) Median=.00 M (SD)=.18(.59) Median=.00 M (SD)=.63(1.36) Median=.00 M (SD)=.67(1.17) Female Median=.00 M (SD)=.50(.71) Median=.00 M (SD)=.15(.56) Median=.00 M (SD)=.50(1.21) Median=.00 M (SD)=.58(1.08) Marital status H (4) = 114.237 p = .000 N=55,91 1 H (4) = 41.964 p = .000 N=55,916 H (4) = 81.886 p = .000 N=55,915 H (4) = 98.511 p = .000 N=55,913 Never married Median=.00 M (SD)=.37(.64) Median=.00 M (SD)=.14(.53) Median=.00 M (SD)=.63(1.35) Median=.00 M (SD)=.59(1.09) Married Median=.00 M (SD)=.49 (.71) Median=.00 M (SD)=.18(.61) Median=.00 M (SD)=.58(1.35) Median=.00 M (SD)=.68(1.19) Widowed Median=.00 M (SD)=.49(.70) Median=.00 M (SD)=.15(.55) Median=.00 M (SD)=.50(1.20) Median=.00 M (SD)=.56(1.07)

PAGE 149

149 Table 4 4. Continued Variables Pain Wandering Aggression Agitation Separated Median=.00 M (SD)=.42(.64) Median=.00 M (SD)=.20(.64) Median=.00 M (SD)=.67(1.35) Median=.00 M (SD)=.70(1.24) Divorced Median=.00 M (SD)=.48(.71) Median=.00 M (SD)=.16(.57) Median=.00 M (SD)=.58(1.28) Median=.00 M (SD)=.60(1.10) Education r s = .042 p = .000 N = 44,289 0 r s = .003 p = .491 N=44,292 r s = .004 p = .436 N=44,291 r s = .015 p = .002 N=44,290 Ethnicity H (3) = 477.957 p = .004 N=56,38 3 H (3) = 13.390 p = .004 N=56,388 H (3) = 36.174 p = .000 N=56,387 H (3) = 153.789 p = .000 N=56,385 Indian/Asian Median=.00 M (SD)=.36(.68) Median=.14 M (SD)=.00(.56) Median=.00 M (SD)=.38(.85) Median=.00 M (SD)=.46(.90) African American Median=.00 M (SD)=.35(.60) Median=.00 M (SD)=.16(.58) Median=.00 M (SD)=.59(1.33) Median=.00 M (SD)=.52(1.05) Hispanic Median=.00 M (SD)=.36(.60) Median=.00 M (SD)=.14(.53) Median=.00 M (SD)=.44(1.07) Median=.00 M (SD)=.69(1.12) Caucasian Median=.00 M (SD)=.51(.72) Median=.00 M (SD)=.16(.58) Median=.00 M (SD)=.55(1.28) Median=.00 M (SD)=.61(1.12) r s U : Mann Whitney U score, H : Kruskal Wallis H score

PAGE 150

150 Table 4 5 Bivariate analysis between proximal factors and the frequency of disruptive behaviors Variables Wandering Aggression Agitation Pain r s = .090 p = .000 N=56,564 r s = .038 p = .000 N=56,563 r s = .043 p = .000 N=56,561 Hunger U = 1,435,408 p =.160 N=56,569 U = 1,653,618.50 p =.000 N=56,568 U = 1,634,866.50 p =.002 N=56,566 No n/a Median=.00 M (SD)=.54(1.27) Median=.60 M (SD)=.60(1.11) Yes n/a Median=.00 M (SD)=.98(1.38) Median=.00 M (SD)=1.21(1.52) Thirst U = 977,056 p =.052 N=56,572 U = 1,102,549.50 p =.707 N=56,571 U = 1,037,880.50 p =.921 N=56,569 No n/a n/a n/a Yes n/a n/a n/a Bowel continence r s = .048 p = .000 N=56,570 r s = .112 p = .000 N=56,569 r s = .088 p = .000 N=56,567 Bladder continence r s = .017 p = .000 N=56,573 r s = .113 p = .000 N=56,572 r s = .095 p = .000 N=56,570 r s U : Mann Whitney U score

PAGE 151

151 Table 4 6 Ordinal logistic regression for the background / proximal factors and the frequency of wandering behaviors (N = 43,676) Independent Variables B S E B p 95% C I OR Background factors Cognitive impairments .687 .017 .000 0 .654 0 .721 1.988 Charlson Comorbidity Index .181 .014 .000 0 .207 0 .154 0.834 Pressure ulcer .347 .033 .000 0 .412 0 .281 0.707 Number of medications .012 .004 .005 0 .021 0 .004 0.988 ADL impairments .146 .004 .000 0 .154 0 .139 0.864 A ge .025 .003 .000 0 .030 0 .020 0.975 Gender ( f emale) .339 .041 .000 0 .418 0 .259 0.712 Married status Never married .417 .079 .000 0 .571 0 .262 0.659 Divorced .201 .074 .006 0 .345 0 .056 0.818 Widowed .073 .044 .096 0 .159 0 .013 0.930 Separated .157 .143 .273 0 .124 0 .438 1.170 Married (Reference) 0 Education .005 .018 .794 0 .041 0 .031 0.995 Race/Ethnicity Indian/Asian .785 .300 .009 1.372 0 .198 0.456 African American .215 .064 .001 0 .340 0 .089 0.807 Hispanic .201 .067 .003 0 .331 0 .070 0.818 Caucasian (Reference) 0 Proximal factors Pain .292 .032 .000 0 .354 0 .229 0.747 Bowel incontinence .086 .018 .000 0 .122 0 .051 0.918 Bladder incontinence .139 .018 .000 0 .104 0 .174 1.149 CI : confidence interval for B, OR : o dds ratio

PAGE 152

152 Table 4 7. Ordinal logistic regression for the background/ proximal factors and the frequency of aggressive behaviors (N = 43,671 ) Independent Variables B S E B p 95% C I OR Background factors Cognitive impairments .379 .010 .000 0 .359 0 .398 1.461 Charlson Comorbidity Index .035 .007 .000 0 .049 0 .021 0.966 Pressure ulcer .082 .013 .000 0 .107 0 .056 0.921 Number of medications .018 .003 .000 0 .012 0 .023 1.018 ADL impairments .038 .003 .000 0 .043 0 .033 0.963 A ge .010 .002 .000 0 .013 0 .007 0.990 Gender (female) .345 .026 .000 0 .397 0 .294 0.708 Married status Never married .215 .047 .000 0 .123 0 .307 1.240 Divorced .238 .046 .000 0 .147 0 .329 1.269 Widowed .070 .029 .015 0 .014 0 .126 1.073 Separated .372 .097 .000 0 .181 0 .562 1.451 Married (Reference) 0 Education .014 .012 .253 0 .037 0 .010 0.986 Race/ethnicity Indian/Asian .472 .179 .008 0 .822 0 .122 0.624 African American .115 .040 .004 0 .193 0 .036 0.891 Hispanic .457 .043 .000 0 .541 0 .373 0.633 Caucasian (Reference) 0 Proximal factors Pain .045 .017 .010 0 .011 0 .078 1.046 Hunger .906 .314 .004 0 .291 1.521 2.474 Bowel incontinence .058 .011 .000 0 .037 0 .079 1.060 Bladder incontinence .089 .010 .000 0 .070 0 .109 1.093 CI: confidence interval for B, OR: odds ratio

PAGE 153

153 Table 4 8 Ordinal logistic regression for the background/ proximal factors and the frequency of agitated behaviors (N = 43,669) Independent Variables B S E B p 95% C I OR Background factors Cognitive impairments .540 .009 .000 0 .521 0 .558 1.716 Charlson Comorbidity Index .100 .007 .000 0 .114 0 .086 0.905 Pressure ulcer .088 .012 .000 0 .112 0 .064 0.916 Number of medications .017 .002 .000 0 .012 0 .022 1.017 ADL impairments .057 .002 .000 0 .062 0 .053 0.945 A ge .014 .002 .000 0 .017 0 .011 0.986 Gender (female) .312 .024 .000 0 .360 0 .265 0.732 Married status Never married .058 .045 .197 0 .145 0 .030 0.944 Divorced .039 .043 .375 0 .047 0 .124 1.040 Widowed .050 .026 .057 0 .102 0 .001 0.951 Separated .065 .094 .491 0 .119 0 .249 1.067 Married (Reference) 0 Education .018 .011 .103 0 .039 0 .004 0.982 Race/ethnicity Indian/Asian .587 .167 .000 0 .914 0 .260 0.556 African American .356 .039 .000 0 .433 0 .279 0.700 Hispanic .178 .035 .000 0 .108 0 .247 1.195 Caucasian (Reference) 0 Proximal factors Pain .062 .016 .000 0 .031 0 .093 1.064 Hunger .800 .305 .009 0 .201 1.398 2.226 Thirst .285 .403 .480 1.074 0 .505 0.752 Bowel incontinence .001 .010 .883 0 .021 0 .018 0.999 Bladder incontinence .081 .009 .000 0 .063 0 .099 1.084 CI: confidence interval for B, OR: odds ratio

PAGE 154

154 Table 4 9. Ordinal logistic regression for the background/proximal factors and the frequency of wandering behaviors after imputing education (N = 55 709 ) Independent Variables B SE B p 95% CI OR Background factors Cognitive impairments .636 .015 .000 0 .607 0 .666 1.889 Charlson Comorbidity Index .189 .012 .000 0 .212 0 .166 0.828 Pressure ulcer .330 .031 .000 0 .391 0 .268 0.719 Number of medications .008 .004 .029 0 .016 0 .001 0.992 ADL impairments .146 .003 .000 0 .152 0 .140 0.864 Age .021 .002 .000 0 .025 0 .016 0.979 Gender (female) .301 .036 .000 0 .372 0 .230 0.740 Married status Never married .434 .068 .000 0 .568 0 .300 0.648 Divorced .242 .064 .000 0 .368 0 .117 0.785 Widowed .100 .039 .010 0 .176 0 .024 0.905 Separated .084 .133 .529 0 .177 0 .345 1.088 Married (Reference) 0 Education (imputed) .025 .018 .173 0 .011 0 .061 1.025 Race/ethnicity Indian/Asian .474 .237 .046 0 .939 0 .009 0.623 African American .176 .055 .001 0 .284 0 .068 0.839 Hispanic .099 .057 .082 0 .210 0 .013 0.906 Caucasian (Reference) 0 Proximal factors Pain .220 .029 .000 0 .277 0 .164 0.803 Bowel incontinence .053 .016 .001 0 .084 0 .021 0.948 Bladder incontinence .156 .016 .000 0 .124 0 .188 1.169 CI: confidence interval for B, OR: odds ratio

PAGE 155

155 Table 4 10. Ordinal logistic regression for the background/proximal factors and the frequency of aggressive behaviors after imputing education (N = 55,704) Independent Variables B SE B p 95% CI OR Background factors Cognitive impairments .340 .009 .000 0 .323 0 .357 1.405 Charlson Comorbidity Index .029 .006 .000 0 .041 0 .016 0.971 Pressure ulcer .074 .012 .000 0 .098 0 .050 0.929 Number of medications .018 .002 .000 0 .013 0 .022 1.018 ADL impairments .041 .002 .000 0 .045 0 .037 0.960 Age .008 .001 .000 0 .011 0 .006 0.992 Gender (female) .330 .023 .000 0 .376 0 .284 0.719 Married status Never married .205 .041 .000 0 .125 0 .285 1.228 Divorced .195 .040 .000 0 .116 0 .274 1.215 Widowed .052 .025 .042 0 .002 0 .102 1.053 Separated .364 .087 .000 0 .193 0 .535 1.439 Married (Reference) 0 Education (imputed) .013 .012 .283 0 .035 0 .010 0.987 Race/ethnicity Indian/Asian .335 .145 .021 0 .620 0 .051 0.715 African American .115 .034 .001 0 .183 0 .048 0.891 Hispanic .331 .036 .000 0 .402 0 .260 0.718 Caucasian (Reference) 0 Proximal factors Pain .056 .016 .000 0 .025 0 .086 1.058 Hunger .890 .285 .002 0 .332 1.449 2.435 Bowel incontinence .070 .010 .000 0 .051 0 .089 1.073 Bladder incontinence .086 .009 .000 0 .068 0 .104 1.090 CI: confidence interval for B, OR: odds ratio

PAGE 156

156 Table 4 11. Ordinal logistic regression for the background/proximal factors and the frequency of agitated behaviors after imputing education (N = 55,701) Independent Variables B SE B p 95% CI OR Background factors Cognitive impairments .505 .008 .000 0 .489 0 .521 1.657 Charlson Comorbidity Index .103 .006 .000 0 .115 0 .091 0.902 Pressure ulcer .074 .011 .000 0 .097 0 .052 0.929 Number of medications .021 .002 .000 0 .016 0 .025 1.021 ADL impairments .056 .002 .000 0 .060 0 .052 0.946 Age .014 .001 .000 0 .016 0 .011 0.986 Gender (female) .281 .022 .000 0 .324 0 .237 0.755 Married status Never married .088 .039 .026 0 .165 0 .011 0.916 Divorced .022 .038 .557 0 .097 0 .052 0.978 Widowed .059 .024 .012 0 .105 0 .013 0.943 Separated .030 .086 .730 0 .139 0 .198 1.030 Married (Reference) 0 Education (imputed) .003 .011 .756 0 .018 0 .025 1.003 Race/ethnicity Indian/Asian .468 .139 .001 0 .741 0 .195 0.626 African American .317 .034 .000 0 .384 0 .251 0.728 Hispanic .215 .031 .000 0 .154 0 .276 1.240 Caucasian (Reference) 0 Proximal factors Pain .093 .015 .000 0 .064 0 .122 1.097 Hunger 1.078 .272 .000 0 .545 1.610 2.939 Thirst .091 .375 .809 0 .827 0 .645 0.913 Bowel incontinence .009 .009 .298 0 .008 0 .027 1.009 Bladder incontinence .077 .009 .000 0 .061 0 .094 1.080 CI: confidence interval for B, OR: odds ratio

PAGE 157

157 Table 4 12 Mediator effect of pain severity on the relationship between background factors and wandering behaviors Mediator, Predictor, Outcome Step Independent variable Dependent variable OR p Pain, cogniti ve impairments, wandering 1 Cogniti ve impairments Wandering 1.32 .000 2 Cogniti ve impairments Pain .73 .000 3 Cogniti ve impairments Wandering 1.27 .000 Pain Wandering .64 .000 Pain, CCI wandering 1 C C I Wandering .77 .000 2 CCI Pain 1.00 .400 3 CCI Wandering n/a Pain Wandering n/a Pain, pressure ulcer, wandering 1 Pressure ulcer Wandering .54 .000 2 Pressure ulcer Pain 1.24 .000 3 Pressure ulcer Wandering .56 .000 Pain Wandering .60 .000 Pain, number of medications wandering 1 Number of medications Wandering .94 .000 2 Number of medications Pain 1.12 .000 3 Number of medications Wandering .95 .000 Pain Wandering .62 .000 Pain, ADL impairment wandering 1 ADL impairments Wandering .92 .000 2 ADL impairments Pain 1.00 .304 3 ADL impairments Wandering n/a Pain Wandering n/a Pain, age, wandering 1 Age Wandering .98 .000 2 Age Pain .99 .000 3 Age Wandering .98 .000 Pain Wandering .56 .000 Pain, gender, wandering 1 Gender (male) Wandering 1.24 .000 2 Gender (male) Pain .80 .000 3 Gender (male) Wandering 1.20 .000 Pain Wandering .57 .000 Pain, marital status, wandering 1 Marital status (married) Wandering 1.19 .000 2 Marital status (married) Pain 1.06 .002 3 Marital status (married) Wandering 1.21 .000 Pain Wandering .57 .000 Pain, education, wandering 1 Education Wandering 1.00 .855 2 Education Pain n/a 3 Education Wandering n/a Pain Wandering n/a Pain, ethnicity, wandering 1 Ethnicity (minority) Wandering .89 .001 2 Ethnicity (minority) Pain .62 .000 3 Ethnicity (minority) Wandering .82 .000 Pain Wandering .56 .000 CCI : Charlson Comorbidity Index OR : o dds ratio

PAGE 158

158 Table 4 1 3 Mediator effect of pain severity on the relationship between background factors and aggressive behaviors Mediator, Predictor, Outcome Step Independent variable Dependent variable OR p Pain, cogniti ve impairments, aggression 1 Cogniti ve impairment Aggression 1.34 .000 2 Cogniti ve impairment Pain .73 .000 3 Cogniti ve impairment Aggression 1.34 .000 Pain Aggression 1.01 .633 Pain, CCI aggression 1 CCI Aggression .97 .000 2 CCI Pain 1.00 .400 3 CCI Aggression n/a Pain Aggression n/a Pain, pressure ulcer, aggression 1 Pressure ulcer Aggression .93 .000 2 Pressure ulcer Pain 1.24 .000 3 Pressure ulcer Aggression .94 .000 Pain Aggression .89 .000 Pain, number of medications aggression 1 Number of medications Aggression .99 .000 2 Number of medications Pain 1.12 .000 3 Number of medications Aggression .99 .005 Pain Aggression .89 .000 Pain, ADL impairment aggression 1 ADL impairments Aggression 1.01 .000 2 ADL impairments Pain 1.00 .304 3 ADL impairments Aggression n/a Pain Aggression n/a Pain, age, aggression 1 Age Aggression .99 .000 2 Age Pain .99 .000 3 Age Aggression .99 .000 Pain Aggression .88 .000 Pain, gender, aggression 1 Gender (male) Aggression 1.26 .000 2 Gender (male) Pain .80 .000 3 Gender (male) Aggression 1.25 .000 Pain Aggression .89 .000 Pain, marital status, aggression 1 Marital status (married) Aggression 1.06 .011 2 Marital status (married) Pain 1.06 .002 3 Marital status (married) Aggression 1.06 .009 Pain Aggression .88 .000 Pain, education, aggression 1 Education Aggression 1.00 .674 2 Education Pain n/a 3 Education Aggression n/a Pain Aggression n/a Pain, ethnicity, aggression 1 Ethnicity (minority) Aggression .93 .004 2 Ethnicity (minority) Pain .62 .000 3 Ethnicity (minority) Aggression .91 .000 Pain Aggression .88 .000 CCI: Charlson Comorbidity Index, OR: odds ratio

PAGE 159

159 Table 4 1 4 Mediator effect of pain severity on the relationship between background factors and agitated behaviors Mediator, Predictor, Outcome Step Independent variable Dependent variable OR p Pain, cogniti ve impairments agitation 1 Cognitive impairments Agitation 1.46 .000 2 Cognitive impairments Pain .73 .000 3 Cognitive impairments Agitation 1.47 .000 Pain Agitation 1.04 .006 Pain, CCI, agitation 1 CCI Agitation .90 .000 2 CCI Pain 1.00 .400 3 CCI Agitation n/a Pain Agitation n/a Pain, pressure ulcer, agitation 1 Pressure ulcer Agitation .92 .000 2 Pressure ulcer Pain 1.24 .000 3 Pressure ulcer Agitation .93 .000 Pain Agitation .88 .000 Pain, number of medications, agitation 1 Number of medications Agitation .98 .000 2 Number of medications Pain 1.12 .000 3 Number of medications Agitation .99 .000 Pain Agitation .89 .000 Pain, ADL impairments agitation 1 ADL impairments Agitation 1.01 .000 2 ADL impairments Pain 1.00 .304 3 ADL impairments Agitation n/a Pain Agitation n/a Pain, age, agitation 1 Age Agitation .99 .000 2 Age Pain .99 .000 3 Age Agitation .99 .000 Pain Agitation .87 .000 Pain, gender, agitation 1 Gender (male) Agitation 1.19 .000 2 Gender (male) Pain .80 .000 3 Gender (male) Agitation 1.18 .000 Pain Agitation .88 .000 Pain, marital status, agitation 1 Marital status (married) Agitation 1.19 .000 2 Marital status (married) Pain 1.06 .002 3 Marital status (married) Agitation 1.20 .000 Pain Agitation .87 .000 Pain, education, agitation 1 Education Agitation .98 .071 2 Education Pain n/a 3 Education Agitation n/a Pain Agitation n/a Pain, ethnicity, agitation 1 Ethnicity (minority) Agitation 1.04 .081 2 Ethnicity (minority) Pain n/a 3 Ethnicity (minority) Agitation n/a Pain Agitation n/a CCI: Charlson Comorbidity Index, OR: odds ratio

PAGE 160

160 Table 4 15 Description of r estraints use among 56,577 NH residents with dementia Restraints Type Number Percent Total restraints use 43,585 77.0 Physical restraints use 15,300 27.0 B ed rails 1 3 ,3 54 2 3 .6 Trunk restraints 1,753 3.1 Restraints chair 784 1.4 Limb restraints 73 0.1 Psychoactive medications use 39,141 69.2 Antidepressant 27,098 47.9 Antipsychotics 15,356 27.1 Antianxiety 14,370 25.4 Hypnotics 6,147 10.9 Some residents had more than one restraint. Therefore, summation of the number of the residents who have a restraint is greater than total number of residents who have restraints.

PAGE 161

161 Table 4 1 6 Description of subsample without restraints: Background factors (N=12,991) Characteristics Possible Range N n (%) M (SD) Actual Range Age (years) 12,991 85 (7) 65 109 Gender 12,988 Male 4,326 (33.3) Female 8,662 (66.7) Marital status 12,820 Never married 952 (7.4) Married 3,869 (30.2) Widowed 6,978 (54.4) Separated 134 (1.0) Divorced 887 (6.9) Education 10,552 K 11 grades 2,510 (23.8) High school 5,096 (48.3) Some college 1,705 (16.2) 805 (7.6) Graduate degree 436 (4.1) Ethnicity 12,956 Indian/Asian 78 (.6) African American 1,838 (14.2) Hispanic 1,041 (8.0) Caucasian 9,999 (77.2) MDS C ognitive P erformance S cale (MDS CPS) 0 6 12,981 3.13 (1.65) 0 6 Intact 956 ( 7.4 ) Borderline intact 843 ( 6.5 ) Mild impairment 2 498 ( 19.2 ) Moderate impairment 4 403 ( 33.9 ) Moderate severe impairment 1 519 ( 11.7 ) Severe impairment 1 070 ( 8.2 ) Very severe impairment 1 692 ( 13.0 ) Charlson Comorbidity Index 0 37 12,991 3 (2) 1 13 Pressure ulcer 0 4 12 991 None 10 302 ( 79.3 ) Stage I 1 066 ( 8.2 ) Stage II 879 ( 6.8 ) Stage III 183 ( 1.4 ) Stage IV 561 ( 4.3 ) Number of medications 12,991 9 (5) 0 90 MDS ADL impairment score 0 28 12,991 19 (7) 0 28

PAGE 162

162 Table 4 1 7 Description of subsample without restraints: Proximal factors (N=12,991) Characteristics Possible Range N n (%) MDS Pain 0 3 12,986 No pain 8,432 (64.9) Mild pain 3,466 (26.7) Moderate pain 983 (7.6) Excruciating pain 105 (.8) Hunger 0 1 12,989 No 12,980 (99.9) Yes 9 (.1) Thirst 0 1 12,991 No 12,985 (100) Yes 6 (<.0) Bowel incontinence 0 4 12,990 Continent 4,966 (38.2) Usually continent 712 (5.5) Occasionally incontinent 756 (5.8) Frequently incontinent 1,338 (10.3) Incontinent 5,218 (40.2) Bladder incontinence 0 4 12,991 Continent 4,250 (32.7) Usually continent 767 (5.9) Occasionally incontinent 994 (7.7) Frequently incontinent 1,981 (15.2) Incontinent 4,999 (38.5)

PAGE 163

163 Table 4 1 8 Prevalence of disruptive behaviors in unrestrained residents (N=12,991) Characteristics Possible Range N n (%) M (SD) Actual Range Wandering 0 3 12,990 No wandering 12,338 (95.0) 1 3 days in 7 days 344 (2.6) 4 6 days in 7 days 137 (1.1) Wandering daily 171 (1.3) Aggression 0 12 12,990 .27 (.83) 0 12 None 11,141 (85.8) Moderate (1,2) 1,452 (11.2) Severe (3 5) 337 (2.6) Very severe (6 12) 60 (.5) Agitation 0 7 12,989 .29 (.74) 0 7 0 10,788 (83.1) 1 1,227 (9.4) 2 606 (4.7) 3 368 (2.8)

PAGE 164

164 Table 4 1 9 Bivariate analysis between background factors pain and disruptive behaviors in unrestrained residents Variables Pain Wandering Aggression Agitation Cognitive Impairments r s = .238 p = .000 N=12,976 r s = .088 p = .000 N=12,980 r s = .126 p = .000 N=12,980 r s = .188 p = .000 N=12,979 Charlson Comorbidity Index r s = .020 p = .022 N=12,984 r s = .087 p = .000 N=12,988 r s = .030 p = .001 N=12,988 r s = .068 p = .000 N=12,987 Pressure ulcer r s = .130 p = .000 N=12,986 r s = .077 p = .000 N=12,990 r s = .026 p = .003 N=12,990 r s = .040 p = .000 N=12,989 Number of medication r s = .248 p = .000 N=12,986 r s = .087 p = .000 N=12,990 r s = .076 p = .000 N=12,990 r s = .084 p = .000 N=12,989 ADL Impairments r s = .046 p = .000 N=12,986 r s = .171 p = .000 N=12,990 r s = .004 p = .634 N=12,990 r s = .024 p = .006 N=12,989 Age r s = .026 p = .003 N=12,986 r s = .017 p = .052 N=12,990 r s = .022 p = .011 N=12,990 r s = .000 p = .958 N=12,989 Gender U = 19,125,204 p =.017 N=12,983 U = 18,637,022 p =.204 N=12,987 U = 18,547,333.50 p =.127 N=12,987 U = 18,504,319 p =.084 N=12,986 Male Median=.00 M (SD)=.42(.64) n/a n/a n/a Female Median=.00 M (SD)=.46(.68) n/a n/a n/a Marital status H (4) = 22.049 p = .000 N=12,815 H (4) = 3.660 p = .454 N=12,819 H (4) = 9.630 p = .047 N=12,819 H (4) = 5.437 p = .245 N=12,818 Never married Median=.00 M (SD)=.35(.62) n/a n/a n/a Married Median=.00 M (SD)=.45(.67) n/a n/a n/a Widowed Median=.00 M (SD)=.45(.67) n/a n/a n/a

PAGE 165

165 Table 4 1 9 Continued Variables Pain Wandering Aggression Agitation Separated Median=.00 M (SD)=.40(.66) n/a n/a n/a Divorced Median=.00 M (SD)=.45(.70) n/a n/a n/a Education r s = .019 p = .048 N=10,548 r s = .006 p = .554 N=10,552 r s = .002 p = .814 N=10,552 r s = .011 p = .249 N=10,551 Ethnicity H (3) = 81.434 p = .000 N=12,951 H (3) = 3.474 p = .324 N=12,955 H (3) = 9.116 p = .028 N=12,955 H (3) = 19.044 p = .000 N=12,954 Indian/Asian Median=.00 M (SD)=.42(.80) n/a Median=.00 M (SD)=.32(.90) Median=.00 M (SD)=.32(.73) African American Median=.00 M (SD)=.33(.58) n/a Median=.00 M (SD)=.25(.85) Median=.00 M (SD)=.23(.68) Hispanic Median=.00 M (SD)=.35(.60) n/a Median=.00 M (SD)=.21(.67) Median=.00 M (SD)=.25(.66) Caucasian Median=.00 M (SD)=.47(.69) n/a Median=.00 M (SD)=.28(.84) Median=.00 M (SD)=.30(.76) r s U : Mann Whitney U score, H : Kruskal Wallis H score

PAGE 166

166 Table 4 20 Bivariate analysis between proximal factors and disruptive behaviors in unrestrained residents Variables Wandering Aggression Agitation Pain r s = .078 p = .000 N= 12,985 r s = .0 31 p = .000 N= 12,985 r s = .035 p = .000 N= 12,984 Hunger U = 74,886 p =.000 N= 12,988 U = 69,559.50 p =.102 N= 12,988 U = 69,442.50 p =.133 N= 12,987 No Median=.00 M (SD)=.09(.42) n/a n/a Yes Median=.00 M (SD)=.56(1.01) n/a n/a Thirst U = 36,996 p =.573 N= 12,990 U = 39,455.50 p =.928 N= 12,990 U = 3 8 ,353.50 p =.921 N= 12,989 No n/a n/a n/a Yes n/a n/a n/a Bowel continence r s = .063 p = .000 N= 12,989 r s = .063 p = .000 N= 12,989 r s = .035 p = .000 N= 12,988 Bladder continence r s = .030 p = .001 N= 12,990 r s = .062 p = .000 N= 12,990 r s = .044 p = .000 N= 12,989 r s U : Mann Whitney U score

PAGE 167

167 Table 4 2 1 Ordinal logistic regression for the background / proximal factors and the frequency of wandering behaviors in unrestrained residents (N = 10,382) Independent Variables B SE B p 95% CI OR Background factors Cognitive impairments .674 .046 .000 0 .585 0 .764 1.962 Charlson Comorbidity Index .217 .039 .000 0 .292 0 .141 0.805 Pressure ulcer .256 .083 .002 0 .419 0 .094 0.774 Number of medications .034 .013 .009 0 .059 0 .008 0.967 ADL impairments .145 .009 .000 0 .163 0 .126 0.865 A ge .015 .007 .038 0 .028 0 .001 0.985 Gender (female) .305 .108 .005 0 .517 0 .093 0.737 Marital status Never married .624 .227 .006 1.068 0 .180 0.536 Divorced .154 .201 .444 0 .548 0 .240 0.857 Widowed .014 .117 .905 0 .244 0 .216 0.986 Separated .198 .411 .630 0 .607 1.002 1.219 Married (Reference) 0 Education .048 .047 .312 0 .045 0 .141 1.049 Race/ethnicity Indian/Asian .330 .744 .657 1.788 1.128 0.719 African American .040 .156 .797 0 .345 0 .265 0.961 Hispanic .091 .202 .651 0 .487 0 .304 0.913 Caucasian (Reference) 0 Proximal factors Pain .385 .096 .000 0 .574 0 .196 0.680 Bowel incontinence .164 .049 .001 0 .261 0 .068 0.849 Bladder incontinence .130 .046 .005 0 .039 0 .221 1.139 CI : confidence interval for B, OR : o dds ratio

PAGE 168

168 Table 4 2 2 Ordinal logis t ic regression for the background and proximal factors and the frequency of aggressive behaviors in unrestrained residents (N = 10,382) Independent Variables B SE B p 95% CI OR Background factors Cognitive impairments .320 .024 .000 0 .272 0 .367 1.377 Charlson Comorbidity Index .027 .019 .151 0 .064 0 .010 0.973 Pressure ulcer .026 .031 .407 0 .086 0 .035 0.974 Number of medications .022 .007 .003 0 .036 0 .007 0.978 ADL impairments .051 .006 .000 0 .064 0 .038 0.950 A ge .008 .004 .059 0 .000 0 .017 1.008 Gender .282 .067 .000 0 .413 0 .151 0.754 Marital status Never married .252 .117 .032 0 .022 0 .482 1.287 Divorced .403 .117 .001 0 .173 0 .634 1.496 Widowed .067 .074 .362 0 .077 0 .212 1.069 Separated .132 .285 .642 0 .426 0 .690 1.141 Married (Reference) 0 Education .004 .030 .880 0 .053 0 .062 1.004 Race/ethnicity Indian/Asian .079 .361 .827 0 .629 0 .786 1.082 African American .276 .096 .004 0 .464 0 .089 0.759 Hispanic .475 .126 .000 0 .722 0 .229 0.622 Caucasian (Reference) 0 Proximal factors Pain .098 .046 .032 0 .008 0 .187 1.103 Bowel incontinence .086 .028 .002 0 .032 0 .141 1.090 Bladder incontinence .044 .025 .072 0 .004 0 .093 1.045 CI: confidence interval for B, OR: odds ratio

PAGE 169

169 Table 4 2 3 Ordinal logistic regression for the background and proximal factors and the frequency of agitated behaviors in unrestrained residents (N = 10,38 1 ) Independent Variables B SE B p 95% CI OR Background factors Cognitive impairments .504 .024 .000 0 .458 0 .550 1.655 Charlson Comorbidity Index .061 .018 .001 0 .096 0 .026 0.941 Pressure ulcer .059 .030 .047 0 .117 0 .001 0.943 Number of medications .019 .007 .005 0 .032 0 .006 0.981 ADL impairments .068 .006 .000 0 .080 0 .057 0.934 Age .000 .004 .925 0 .008 0 .007 1.000 Gender (female) .305 .062 .000 0 .426 0 .184 0.737 Marital status Never married .026 .112 .815 0 .246 0 .193 0.974 Divorced .096 .113 .394 0 .125 0 .317 1.101 Widowed .029 .067 .667 0 .159 0 .102 0.971 Separated .066 .269 .807 0 .594 0 .462 0.936 Married (Reference) 0 Education .030 .027 .270 0 .023 0 .083 1.030 Race/ethnicity Indian/Asian .119 .354 .738 0 .812 0 .575 0.888 African American .328 .091 .000 0 .506 0 .151 0.720 Hispanic .234 .108 .030 0 .446 0 .022 0.791 Caucasian (Reference) 0 Proximal factors Pain .102 .042 .016 0 .019 0 .185 1.107 Bowel incontinence .028 .026 .275 0 .078 0 .022 0.972 Bladder incontinence .057 .023 .013 0 .012 0 .102 1.059 CI: confidence interval for B, OR: odds ratio

PAGE 170

170 Table 4 2 4 Ordinal logistic regression for the background/proximal factors and the frequency of wandering behaviors in unrestrained residents after imputing education (N = 12,774) Independent Variables B SE B p 95% CI OR Background factors Cognitive impairments .620 .041 .000 0 .539 0 .701 1.859 Charlson Comorbidity Index .217 .035 .000 0 .284 0 .149 0.805 Pressure ulcer .220 .077 .004 0 .371 0 .069 0.803 Number of medications .024 .011 .035 0 .046 0 .002 0.976 ADL impairments .153 .008 .000 0 .170 0 .137 0.858 Age .014 .006 .028 0 .026 0 .001 0.986 Gender (female) .208 .099 .036 0 .403 0 .014 0.812 Marital status Never married .520 .194 .007 0 .901 0 .140 0.595 Divorced .145 .177 .414 0 .493 0 .203 0.865 Widowed .043 .106 .687 0 .251 0 .165 0.958 Separated .056 .399 .888 0 .727 0 .839 1.058 Married (Reference) 0 Education (imputed) .092 .047 .051 0 .000 0 .184 1.096 Race/ethnicity Indian/Asian .142 .616 .818 1.349 1.066 0.868 African American .064 .138 .640 0 .334 0 .205 0.938 Hispanic .139 .179 .439 0 .491 0 .213 0.870 Caucasian (Reference) 0 Proximal factors Pain .312 .090 .001 0 .488 0 .136 0.732 Bowel incontinence .130 .045 .004 0 .218 0 .042 0.878 Bladder incontinence .188 .044 .000 0 .103 0 .274 1.207 CI: confidence interval for B, OR: odds ratio

PAGE 171

171 Table 4 2 5 Ordinal logistic regression for the background/proximal factors and the frequency of aggressive behaviors in unrestrained residents after imputing education (N = 12,774) Independent Variables B SE B p 95% CI OR Background factors Cognitive impairments .260 .022 .000 0 .218 0 .302 1.297 Charlson Comorbidity Index .017 .016 .299 0 .049 0 .015 0.983 Pressure ulcer .018 .029 .543 0 .074 0 .039 0.982 Number of medications .027 .007 .000 0 .040 0 .015 0.973 ADL impairments .054 .006 .000 0 .065 0 .043 0.947 Age .009 .004 .013 0 .002 0 .017 1.009 Gender (female) .288 .061 .000 0 .407 0 .170 0.750 Marital status Never married .193 .105 .067 0 .013 0 .399 1.213 Divorced .405 .104 .000 0 .202 0 .608 1.499 Widowed .102 .066 .122 0 .027 0 .231 1.107 Separated .145 .255 .570 0 .355 0 .644 1.156 Married (Reference) 0 Education (imputed) .003 .029 .920 0 .054 0 .060 1.003 Race/ethnicity Indian/Asian .100 .313 .749 0 .513 0 .713 1.105 African American .186 .082 .023 0 .346 0 .025 0.830 Hispanic .347 .106 .001 0 .554 0 .140 0.707 Caucasian (Reference) 0 Proximal factors Pain .094 .043 .028 0 .010 0 .177 1.099 Bowel incontinence .096 .026 .000 0 .046 0 .146 1.101 Bladder incontinence .044 .023 .054 0 .001 0 .090 1.045 CI: confidence interval for B, OR: odds ratio

PAGE 172

172 Table 4 2 6 Ordinal logistic regression for the background/proximal factors and the frequency of agitated behaviors in unrestrained residents after imputing education (N = 12,773) Independent Variables B SE B p 95% CI OR Background factors Cognitive impairments .450 .021 .000 0 .409 0 .492 1.568 Charlson Comorbidity Index .067 .016 .000 0 .098 0 .035 0.935 Pressure ulcer .046 .028 .102 0 .100 0 .009 0.955 Number of medications .015 .006 .014 0 .027 0 .003 0.985 ADL impairments .068 .005 .000 0 .078 0 .058 0.934 Age .002 .004 .572 0 .009 0 .005 0.998 Gender (female) .286 .057 .000 0 .397 0 .175 0.751 Marital status Never married .013 .101 .895 0 .211 0 .184 0.987 Divorced .114 .101 .258 0 .083 0 .311 1.121 Widowed .001 .060 .987 0 .119 0 .118 0.999 Separated .129 .253 .611 0 .624 0 .367 0.879 Married (Reference) 0 Education (imputed) .056 .027 .036 0 .004 0 .108 1.058 Race/ethnicity Indian/Asian .062 .294 .833 0 .513 0 .637 1.064 African American .342 .081 .000 0 .500 0 .184 0.710 Hispanic .170 .094 .071 0 .354 0 .015 0.844 Caucasian (Reference) 0 Proximal factors Pain .126 .040 .002 0 .048 0 .204 1.134 Bowel incontinence .012 .024 .624 0 .058 0 .035 0.988 Bladder incontinence .061 .022 .005 0 .019 0 .104 1.063 CI: confidence interval for B, OR: odds ratio

PAGE 173

173 Table 4 2 7 Mediator effect of pain severity on the relationship between background factors and wandering behaviors in unrestrained residents Mediator, Predictor, Outcome Step Independent variable Dependent variable OR p Pain, cognition, wandering 1 Cognition Wandering 1.32 .000 2 Cognition Pain .73 .000 3 Cognition Wandering 1.27 .000 Pain Wandering .64 .000 Pain, CCI wandering 1 CCI Wandering .77 .000 2 CCI Pain 1.00 .400 3 CCI Wandering n/a Pain Wandering n/a Pain, pressure ulcer, wandering 1 Pressure ulcer Wandering .54 .000 2 Pressure ulcer Pain 1.24 .000 3 Pressure ulcer Wandering .56 .000 Pain Wandering .60 .000 Pain, number of medications, wandering 1 Number of medications Wandering .94 .000 2 Number of medications Pain 1.12 .000 3 Number of medications Wandering .95 .000 Pain Wandering .62 .000 Pain, ADL impairments wandering 1 ADL Wandering .92 .000 2 ADL Pain 1.00 .304 3 ADL Wandering n/a Pain Wandering n/a Pain, age, wandering 1 Age Wandering .98 .000 2 Age Pain .99 .000 3 Age Wandering .98 .000 Pain Wandering .56 .000 Pain, gender, wandering 1 Gender (male) Wandering 1.24 .000 2 Gender (male) Pain .80 .000 3 Gender (male) Wandering 1.20 .000 Pain Wandering .57 .000 Pain, marital status, wandering 1 Marital status (married) Wandering 1.19 .000 2 Marital status (married) Pain 1.06 .002 3 Marital status (married) Wandering 1.21 .000 Pain Wandering .57 .000 Pain, education, wandering 1 Education Wandering 1.00 .855 2 Education Pain n/a 3 Education Wandering n/a Pain Wandering n/a Pain, ethnicity, wandering 1 Ethnicity (minority) Wandering .89 .001 2 Ethnicity (minority) Pain .62 .000 3 Ethnicity (minority) Wandering .82 .000 Ethnicity (minority) Wandering .56 .000 CCI: Charlson Comorbidity Index OR : odds ratio

PAGE 174

174 Table 4 2 8 Mediator effect of pain severity on the relationship between background factors and aggressive behaviors in unrestrained residents Mediator, Predictor, Outcome Step Independent variable Dependent variable OR p Pain, cognition, aggression 1 Cognition Aggression 1.34 .000 2 Cognition Pain .73 .000 3 Cognition Aggression 1.34 .000 Pain Aggression 1.01 .633 Pain, CCI aggression 1 CCI Aggression .97 .000 2 CCI Pain 1.00 .400 3 CCI Aggression n/a Pain Aggression n/a Pain, pressure ulcer, aggression 1 Pressure ulcer Aggression .93 .000 2 Pressure ulcer Pain 1.24 .000 3 Pressure ulcer Aggression .94 .000 Pain Aggression .89 .000 Pain, number of medications, aggression 1 Number of medications Aggression .99 .000 2 Number of medications Pain 1.12 .000 3 Number of medications Aggression .99 .005 Pain Aggression .89 .000 Pain, ADL impairments aggression 1 ADL Aggression 1.01 .000 2 ADL Pain 1.00 .304 3 ADL Aggression n/a Pain Aggression n/a Pain, age, aggression 1 Age Aggression .99 .000 2 Age Pain .99 .000 3 Age Aggression .99 .000 Pain Aggression .88 .000 Pain, gender, aggression 1 Gender (male) Aggression 1.26 .000 2 Gender (male) Pain .80 .000 3 Gender (male) Aggression 1.25 .000 Pain Aggression .89 .000 Pain, marital status, aggression 1 Marital status (married) Aggression 1.06 .011 2 Marital status (married) Pain 1.06 .002 3 Marital status (married) Aggression 1.06 .009 Pain Aggression .88 .000 Pain, education, aggression 1 Education Aggression 1.00 .674 2 Education Pain n/a 3 Education Aggression n/a Pain Aggression n/a Pain, ethnicity, aggression 1 Ethnicity (minority) Aggression .93 .004 2 Ethnicity (minority) Pain .62 .000 3 Ethnicity (minority) Aggression .91 .000 Ethnicity (minority) Aggression .88 .000 CCI: Charlson Comorbidity Index, OR: odds ratio

PAGE 175

175 Table 4 2 9 Mediator effect of pain severity on the relationship between background factors and agitated behaviors in unrestrained residents Mediator, Predictor, Outcome Step Independent variable Dependent variable OR p Pain, cognition, agitation 1 Cognition Agitation 1.46 .000 2 Cognition Pain .73 .000 3 Cognition Agitation 1.47 .000 Pain Agitation 1.04 .006 Pain, CCI agitation 1 CCI Agitation .90 .000 2 CCI Pain 1.00 .400 3 CCI Agitation n/a Pain Agitation n/a Pain, pressure ulcer, agitation 1 Pressure ulcer Agitation .92 .000 2 Pressure ulcer Pain 1.24 .000 3 Pressure ulcer Agitation .93 .000 Pain Agitation .88 .000 Pain, number of medications, agitation 1 Number of medications Agitation .98 .000 2 Number of medications Pain 1.12 .000 3 Number of medications Agitation .99 .000 Pain Agitation .89 .000 Pain, ADL impairments agitation 1 ADL Agitation 1.01 .000 2 ADL Pain 1.00 .304 3 ADL Agitation n/a Pain Agitation n/a Pain, age, agitation 1 Age Agitation .99 .000 2 Age Pain .99 .000 3 Age Agitation .99 .000 Pain Agitation .87 .000 Pain, gender, agitation 1 Gender (male) Agitation 1.19 .000 2 Gender (male) Pain .80 .000 3 Gender (male) Agitation 1.18 .000 Pain Agitation .88 .000 Pain, marital status, agitation 1 Marital status (married) Agitation 1.19 .000 2 Marital status (married) Pain 1.06 .002 3 Marital status (married) Agitation 1.20 .000 Pain Agitation .87 .000 Pain, education, agitation 1 Education Agitation .98 .071 2 Education Pain n/a 3 Education Agitation n/a Pain Agitation n/a Pain, ethnicity, agitation 1 Ethnicity (minority) Agitation 1.04 .081 2 Ethnicity (minority) Pain n/a 3 Ethnicity (minority) Agitation n/a Pain Agitation n/a CCI: Charlson Comorbidity Index, OR: odds ratio

PAGE 176

176 CHAPTER 5 DISCUSSION AND IMPLI CATIONS Summary of Results This study examined the relationship between pain severity and the frequency of disruptive behaviors (e.g., wandering, aggressive, and agitated behaviors). Among F lorida NH residents with dementia, the prevalence of pain is 37 % and the prevalence of disruptive behaviors is 9 30% (wandering behaviors: 9%; aggressive behaviors: 24% ; and agitated behaviors: 30%). More severe p ain is associated with fewer wandering b ehaviors, but is associated with more frequent aggressive and agitated behaviors, after controlling for the other background/proximal factors. There are no mediating effects of pain severity on the relationship between any of background factors (e.g., cogn itive impairment comorbidity, pressure ulcer, number of medications, ADL impairment, age, gender, marital status, education, and race/ethnicity) and the frequency of disruptive behaviors. Among the unrestrained NH residents, the results are similar to the ones among all the NH residents except that the overall prevalence of disruptive behaviors is lower. In the following section s the findings are discussed in detail. Prevalence of Pain in Cognitively Impaired Residents In this study, about one of the third of the NH residents with dementia were reported as experiencing pain. Th e prevalence of pain in this study is lower than that of the published studies using direct pain assessment, but similar to that of studies using MDS assessment data. A number of studies reported that 50% to 80% of NH residents had pain using complete interview or direct observation of behavioral pain indicators (Black, et al., 2006; Horgas & Dunn, 2001; Jones, et al., 2006; Smalbrugge, et al., 2007; D. Weiner, et al., 1999; Zanocchi, et al., 2008; Zwakhalen, et al., 2009) These rates are lower when MDS assessment data is used (Cohen Mansfield, 2004) and are consistent with the results of this study (approximately 30%)

PAGE 177

177 In the MDS assessment data, pain items (frequency and intensity) are coded based on self reports from NH residents and/ or facility nursing staff or informal caregivers P ain assessme nt using self report from residents with dementia and/or caregivers who do no t have enough training to observe signs of pain may incorrectly describe pain in cognitively impaired residents (Hadjistavropoulos & Craig, 2002; Horgas & Ell iott, 2004; Horgas, et al., 2009) P to remember, interpret, and respond to pain is altered so that they are often unable to properly articulate or convey their pain experience. Als o, nursing staff or informal caregivers often do not consider behavioral pain indicators when they evaluate pain in cognitively impaired residents, so their self report can underestimate pain for these residents (Cadogan, Schnelle, Yamamoto Mitani, Cabrera, & Simmons, 2004; Eritz & Hadjistavropoulos, 2010; Horgas & Dunn, 2001) There is also a consistent relationship between pain and cognitive status in the literature. In this study, correlation = .219, p =.000). This result is consistent with the findings in the literature that pain is less likely to be reported by cognitively impaired residents than among those who are intact (Cohen Mansfield, 2004; Horgas, et al., 2009; Sengupta, et al., 2010) Pain assessment in cognitively impaired adults is challenging because their pain perception and expression is different from those in the cognitively intact (Cohen Mansfield, 2004; Fisher, et al., 2002; Scherder, et al., 2005) Cognitively impaired re sidents may have difficulties in remembering their pain, interpreting questions and information, or communicating with others, but usually have no difficulties in demonstrating a number of behaviors indicative of pain, such as facial grimacing, rubbing, re stlessness, and moaning (Herr, Coyne, et al., 2006; Horgas, et al., 2009; Warden, et al., 2 003) Therefore, a comprehensive pain scale that includes behavioral pain indicators (e.g., facial expressions, verbalizations or vocalizations, body movements, changes in

PAGE 178

178 interpersonal interactions, changes in activity patterns or routines, and mental status changes ) would better recognize pain in cognitively impaired residents (American Geriatric Society, 2002; Horgas, Nichols, Schapson, & Vietes, 2007) Prevalence of Disruptive Behaviors in Cognitively Impaired Residents In this study, about one to three out of 10 NH residents with dementia were reported as e xp ressing disruptive behaviors (e.g., wandering, aggressive, and agitated behaviors) The prevalence of disruptive behaviors in this study is lower than that of the published studies Several studies reported that 40% to 8 0% of NH residents had disruptive behaviors using comprehensive caregiver interview or direct observation of behaviors (Ballard, Margallo Lana, et al., 2001; Brodaty, et al., 2001; Chen, et al., 2000; Holtzer, et al., 2003; Hwang, et al., 1997; Kunik, et al., 2007; Margallo Lana, et al., 2001; Matsuoka, et al. 2003; Schreiner, 2001; Sink, et al., 2004; Suh, 2004; Wood, et al., 2000) R andom errors and bias es by MDS coordinators may be responsible for this low frequency of disruptive behaviors on the MDS assessment data (Bharucha, et al., 2008; Shin & Scherer, 2009) MDS coordinators may receive insufficient training on how to evaluate residents for behavioral symptoms, and have limited contact with the residents In addition, MDS coordinators may have concerns that a high prevalence of disruptive behaviors might reputations (Anderson, et al., 2003; Hendrix, et al., 2003) One of the CMS quality indicators, b ehavioral and em otional patterns is composed of the prevalence or incidence of these disruptive behaviors. The Relationship between Pain Severity and Disruptive Behaviors Most of the published literature suggested that there is a positive relationship between pain and d isruptive behaviors (Aalten, et al., 2006; Bartels, et al., 2003; Buffum, et al., 2001; Kiely, et al., 2000; Manfredi, et al., 2003; Norton, et al., 2010; Villanueva, et al., 2003) However, the

PAGE 179

179 results of this study suggest that the relationship between pain and disruptive behaviors depends on the type of behaviors examined The direction of the relationship between these variab les depends on whether the disruptive behaviors are accompanied by locomotion behaviors. Pa in is positively correlated with disruptive behaviors that do not involve locomotion (e.g., aggression and agitation) but negatively related to disruptive behaviors that are accompanied by locomotion (e.g., wandering) That is, t he residents who experience more severe pain are more likely to display aggression and agitation, and less likely to wander. The finding that pain and aggressi ve or agitated behaviors are pos itively linked in NH residents with dementia is consistent with other published reports. Buffum and colleagues (2001) reported that pain was positively related to agitation ( r = 50 p = .0 03 ) using a bivariate correlation analysis in 33 Veterans Aff airs NH residents with dementia, and Manfredi and colleagues (2003) demonstrated that opioid treatment for pain reduced agitation in 13 NH residents wi th dementia who were more than 85 years old (mean change in CMAI score: 6.4, 95% CI [ 10.96, 1.8]) Both of the se studies have small sample size. Thus, the results of this study using a large sample from all the residents with dementia in the state of Fl orida substantiate the positive relationship between pain and non locomotive disruptive behaviors from these previous findings. In contrast, t he finding on the relationship between pain and wandering behavior in this study is opposite to the finding s prese nted in the literature review. Kiley and the colleagues (2000) used MDS assessment data from 8,982 NH re sidents and reported that NH residents who expressed sadness or pain in MDS assessment data were 65% more likely to develop wandering behav ior than their counterparts who did not express sadness or pain using a logistic regression analysis ( OR = 1.65, p = .02 ) after controlling for p neumonia, short term memory, repetitive

PAGE 180

180 questions, long term memory, constipation, antipsychotic medication, ADL impairment, and gender This dissertation study has several strengths compared to the published study. Our study measure d pain more appropriate ly using the MDS pain severity scale (Fries, et al., 2001) combining both pa i n frequency and pain intensity, while Kiley and the collegues (2000) measured pain by a dichotomized expression of sadness or pain S adness is not typically considered an indicator of pain, a nd its inclusion may have confounded pain and depression or mood disorder. Also the 1 2 covariates (e.g., cognitive impairments, comorbidity, pressure ulcer, number of medications, ADL impairments, age, gender, marital status, education, ethnicity, bowel incontinence, and bladder incontinence) are controlled in this study based on the NDB model, the prevailing theory that explains disruptive behaviors However Kiley and the colleagues controlled only 8 covariates ( pneumonia, short term memory, repetitive questions, long term memory, constipation, antipsychotic medication, ADL impairment, and gender ) and there was no clear theoretical basis for their ch oice of covariates. Kiley and the colleagues did not control for comorbidity age marital status, ethnicity, incontinence, and other factors which may influence the results. Among th e disruptive behaviors studied d isruptive behaviors that are accompanied by locomotion (e.g., w andering ) are negatively correlated with pain. Wandering is a locom o tion behavior (Algase, et al., 1996 ; Algase, Yao, et al., 2007) and this movement increase s pain (Hadjistavropoulos, LaChapelle, MacLeod, Snider, & Craig, 2000; Horgas, et al., 2009) Therefore this negative relationship may reflect the fact that residents are reluctant to wander or move around when they have pain No Mediating Effect of Pain Severity on the Disruptive Behaviors In the NDB model, Algase and colleagues (1996) conceptualized pain (a proximal factor) as a mediator of the relationship between background factors and dementia compromised

PAGE 181

181 behavior s and depicted pain as a mediator in the conceptual map. Background factors are hypot hesized to directly affect pain (a proximal factor), and then pain affected dementia compromised behaviors. However, a mediating effect is not found in this study T he effect of background factors (predictor) on disruptive behaviors (outcome) is not significantly diminished in the presence of pain (mediator) in the regression model. Thus, the results of this study do not support the theoretical relationships posited in the NDB model. The revised NDB model (Algase, Yao, et al., 2007) posits that pain (a proximal factor ) serves as either mediator or moderator in the relationship between background factors and dementia compromised behaviors. Moderation would indicate that the effect of background factors on disruptive behaviors depends on the level of pain. That is, the relationship between background factors and disruptive behaviors differs between the residents who have a severe pain and the ones who do not have For example, for mildly or moderately cognitively impaired residents, there is no effect of pain on disruptive behaviors; however, for severely cognitively impaired residents, residents with severe pain have more frequent disruptive behaviors than residents w ithout pain. Restraints Use on NH Residents with Dementia Conceptually, it only makes sense to investigate influences on disruptive behaviors among residents who are able to display such behaviors; that is, those who are not restrained. Thus, the analyses were repeated in this subsample. In contrast to the hypothesis that un re strained residents would display more frequent disruptive behaviors, these residents exhibited approximately 50% less frequen t disruptive behaviors. The prevalence of wandering decreased from 9% to 5%, aggressive behaviors decreased from 24.4% to 14.2%, and agitated behaviors decreased from 30.4% to 16.9% in the subsample. The decrease of disruptive behaviors among unrestrained residents may reflect that restraints are still often used as a tool to manage disruptive behaviors.

PAGE 182

182 W e can speculate that if residents display disruptive behaviors, they are restrained ; o therwise restrained. Therefore, unrestrained residents had less frequent disruptive behavior. Also, in this study, residents with disruptive behaviors are significantly more likely to be 2 = 330.56, df =1, p = .000; 2 = 945.74, df =1, p 2 = 1451.91, df =1, p = .000). T he decreased prevalence of disruptive behaviors in unrestrained residents did not essentially change the statistical results. Pain severity is still positively associated with the frequency of disr uptive behaviors except wandering behaviors and does not mediate the relationship between background factors and disruptive behaviors. The re is a high prevalence of restraint use in this study A lmost 80% residents with dementia were restrained physicall y or chemically: physical restraints (e.g., trunk restraints, limb restraints, restraint chair, and bed rails) are used in 27%, and psychoactive medications (e.g., antipsychotics, antianxiety, antidepressant, and hypnotics ) are used in 69% of NH residents with dementia This finding is consistent with the published literature The published studies reported that physical restraints were used in about 30% to 70% of NH residents with dementia (Bartels, et al., 20 03; Kirkevold, et al., 2003) and psychoactive medications were used in 50% to 75 % of these residents (Petek Ster & Cedilnik Gorup, 2011; Richter, et al., in press; Tjia, et al., 2010) Restraints have many side ef fects, such as increased risk of stroke, death, and so forth (Capezuti, et al., 2002; Gastmans & Milisen, 2006; Huabin, et al., 2011; Schneider, et al., 2005; Sink, et al., 2005) The appropriate management of pain can be used to reduce the use of restraints because pain severity is positively correlated with more frequent disruptive behaviors. Limitations Several limitation s of this study should be noted. First, this study is inherently limited by secondary analysis of federally mandated MDS assessment data, which contains critical

PAGE 183

183 information for CMS certification of all Medicaid or Medicare certified NHs. MDS assessment data may have some variability du e to different styles and skills of MDS coordinators in each facility, and MDS coordinators might underreport pain or disruptive behaviors because of their belief that a high prevalence of pain or disruptive behaviors might elicit more attention from the C MS survey team during their annual visits This would affect the certification or financial support from Medicaid and Medicare (Schnelle, Wood, Schnelle, & Simmons, 2001) Second, t he role of pain medications is not con sidered in this study. Pain medications are usually given when nursing staff note resident s pain is measured at the highest level during the last 7 days and behavioral symptoms are measured as the frequency in the last 7 days. The highest level of pain could have been managed by pain medicati ons during the last 7 days, but it is not possible to discern this in the MDS assessment data. However, similar to our study, most of the literature reported the relationship between highest level of pain and the frequency of behavioral symptoms during the observation period without controlling for pain medications (Bartels, et al., 2003; Buffum, et al., 2001; Kiely, et al., 2000; Leonard, et al., 2006; Norton, et al., 2010; Villanueva, et al., 2003) Third this study design is descriptive and cross sectional As such this study is not able to examine causal relationships between pain and disruptive behaviors. Fourth some variables in the MDS assessment data are dichotomous or ordinal meas urements with small ranges so they cannot describe the relative magnitude of the differences. For example, hunger and thirst in MDS assessment data are given dichotomous categorical measurements, and are very positively skewed (prevalence of hunger and thi rst : 0 .1%). Therefore, these variables are not included in some analyses. Also, pain and wandering

PAGE 184

184 are measured as an ordinal format with four levels. If these were measured as continuous format the outcomes might be more informative. Fifth the C harlson C omorbidity I ndex (CCI) computed from 19 diagnosed medical conditions, provide d in our study S ix conditions (connective tissue disease, peptic ulcer disease, leukemia, lymphoma, moderate or sev ere liver disease, and metastatic solid tumor) had to be identified using ICD 9 codes in the MDS other diagnoses item, but these conditions in most of the MDS data Therefore, the possible range of CCI is 0 to 37, but actual range of CCI in our study is restricted (Range = 0 to 16 ) Sixth, the partial mediation effects were not estimated in this study. The dependent variables were not continuous format and path coeffi cients are not b weights of multiple regression. Therefore, we used traditional full mediating effect model a n d e x a m i n e d w hether pain fully mediated the relationship between background factors and disruptive behaviors. Finally, the NDB model was adapted to guide this study, but this study did not include all the variables in the NDB model. In particular, the relationship between social/ physical environmental factors of the NH facilities and disruptive behaviors was not examined (e.g. social interaction, staff mix and ratio, ambiance, crowding, complexity, and ambient conditions). Therefore, the variance of disruptive behaviors which can be explained by background and proximal factors in this study was relatively small ( Nagelkerke R s quare: wandering 19%, aggression 8.2%, and agitation 13%). Implications Implications for Nursing Practice Pain severity can c ontribute to disruptive b ehaviors in residents with dementia (American Geriatric Society, 2002; Bartels, et al., 2003; Buffum, et al., 2001; Horgas & Elliott, 2004; Kiely,

PAGE 185

185 et al., 2000; Norton, et al., 2010; Shega, Hougham, Stocking, Cox Hayley, & Sachs, 2005; Snow, et al., 2009; Villanueva, et al., 2003; Zieber, Hagen, Armstrong Esther, & Aho, 2005) H ealth care providers should consider pain as a potential cause of disruptive behaviors, especially disruptive behaviors which are n ot related to locomotion A comprehensive pain assessment, which includes behaviors indicative of pain should be used to assess pain in residents with cognitive impairments Currently, some pain measures, such as the A ssessment of D iscomfort in D ementia p rotocol (Kovach, Weissman, Griffie, Matson, & Muchka, 1999) Dolopus II (Pautex, Herrmann, Michon, Giannakopoulos, & Gold, 2007) P ain A ssessment C hecklist for S eniors with L imited A bility to C ommunicate (Fuchs Lacelle & Hadjistavropoulos, 2004) includes behavioral pain indicators However, these pain assessment tools are not widely used because of the difficulty of training and scoring these pain measures, or limited psychometric evidence (Herr, Bjoro, & Decker, 2006) A thorough pain measure ment tool which is easy to administer and ha s strong psychometric evidences, is needed. A ppropriate pain management could help decrease disruptive behaviors and consequently reduce extra cost s brought about by the need for constant supervision and the use of other resources, including physical and c hemical restraints (Allen Burge, et al., 1999; Kiely, et al., 2000; Norton, et al., 2010) For example, scheduled pain medication s instead of as needed antipsychotics or antianxiety medications, would be a better option to manage disruptive behavior s (Elliott & Horgas, 2009; Ma nfredi, et al., 2003; Reynolds, et al., 2008) Implications for Theory The NDB model was adapted to guide this study. The study purpose is not to validate all the propositions in the NDB model, but to explore the specific relationship between pain and disruptive behaviors using that model. The NDB model hypothesizes that there are two main constructs that predict disruptive behaviors: background factors and proximal factors. The model

PAGE 186

186 also posits that both background and proximal factors interact or combine in some sequence to produce NDBs (Algase, Yao, et al., 2007) We tested the hypotheses that pain is positively related to disruptive behaviors and pain mediates the relationship between background facto rs and disruptive behaviors which were deduced from one of propositions in the NDB model We found that there are positive relationship s between pain and disruptive behaviors except wandering behaviors, but did not find a mediating effect of pain on the r elationship between background factors and disruptive behaviors. One can hypothesize that pain moderates the relationship between background f actors and disruptive behaviors using the other proposition (pain as a moderator) in the revised NDB model This w ill be tested in future research. Implications for Future Research Findings from this study can be a foundation for future research. First, studies using specialized in depth measures and prospective design are needed to validate these individual based findings. The Pain Assessment in Advanced Dementia ( PAINAD) (Warden, et al., 2003) the Revised Algase W andering S cale Long Term Care v ersion (RAWS LTC) (Algase, Beattie, et al., 2004) and the Cohen Mansfield Agitation Inventory (Cohen Mansfield & Billig, 1986) are easy to administer and require little training. Also, randomized controlled trials (RCTs) can be used to compare comprehensive pain management and usual pain manag ement with regard to the frequency of disruptive behaviors. This type of study can provide evidence for causal relationships between pain management and disruptive behaviors and support changes in clinical practice. Second, more expanded studies could be i nvestigated using MDS assessment data. For instance, the role of pain medications on disruptive behaviors warrants investigation (Horgas, et al., 2009) This can be studied using MDS data and medication profile data. A lso, we can

PAGE 187

187 examine the moderating effect of pain on the relationship between background factors and disruptive behaviors. In addition, w e can develop a comprehensive pain assessment measure using MDS items that includes six behavioral pain indicators (e.g., facial expressions, verbalizations or vocalizations, b ody movements, changes in interpersonal interactions, changes in activity patterns or routines, and mental status changes) in the American Geriatrics Society guidelines for pain (2002) These efforts will facilitate the use of MDS assessment data, since the MDS was intended, in part, as a tool for behavioral research in NH residents. Third, future research would include the longitudinal MDS assessment data to examine trends over time. The longitudinal nature of MDS assessment data, collected every three months or more often, provides an opportunity to describe change over time, lagged relationships or longitudinal relationships, and facilitates the use of more powerful statistical analysis techniques to describe both within and between person changes (Meyers, et al., 2006) Furthermore, future studies should include measurements of social and physical environmental factors of disruptive behaviors in NH residents with dementia (e.g., staff mix and ratio, ambiance, crowding, complexity, ambient conditions ). These domains are one of the proximal factors of disruptive behaviors ( Algase, et al., 1996; Algase, Yao, et al., 2007) and the inclusion of these factors may shed more light on the ways in which NH residents express pain through disruptive behaviors. Conclusion In conclusion, pain exacerbated disruptive behaviors that ar e not locomotion based In order to reduce these disruptive behaviors which can lead to use of restraints, their underlying causes, such as pain, should be investigated and well managed. However, pain assessment in cognitively impaired residents can be challenging. Pain assessment based on self report from residents or nursing staff, is not a reliable assessment method in cognitively impaired adults

PAGE 188

188 Comprehensive pain assessment that includes behavioral pain indicators should be developed further and p ain should be well controlled to reduce these problematic disruptive behaviors.

PAGE 189

189 REFERE N CE LI S T Aalten, P., van Valen, E., de Vugt, M. E., Lousberg, R., Jolles, J., & Verhey, F. R. (2006). Awareness and behavioral problems in dementia pa tients: A prospective study. International Psychogeriatrics, 18 (1), 3 17. Achterberg, W., Gambassi, G., Finne Soveri, H., Liperoti, R., Noro, A., Frijters, D. H. M., et al. (2010). Pain in European long term care facilities: Cross national study in Finland Italy and the Netherlands. Pain, 148 (1), 70 74. Achterberg, W., Pot, A. M., Kerkstra, A., Ooms, M., Muller, M., & Ribbe, M. (2003). The effect of depression on social engagement in newly admitted Dutch nursing home residents. Gerontologist, 43 (2), 213 21 8. Algase, D. L. (1992). Cognitive discriminants of wandering among nursing home residents [Abstract]. Nursing Research, 41 78 81. Algase, D. L., Beattie, E. R., Leitsch, S. A., & Beel Bates, C. A. (2003). Biomechanical activity devices to index wandering behavior in dementia. American Journal of Alzheimer's Disease and Other Dementias, 18 (2), 85 92. Algase, D. L., Beattie, E. R., Song, J. A., Milke, D., Duffield, C., & Cowan, B. (2004). Validation of the Algase wandering scale (version 2) in a cross cultu ral sample. Aging & Mental Health, 8 (2), 133 142. Algase, D. L., Beck, C., Kolanowski, A., Whall, A., Berent, S., & Richards, K., et al. (1996). Need driven dementia compromised behavior: An alternative view of disruptive behavior. American Journal of Alzheimer's Disease and Other Dementias, 11 (6), 10 19. Algase, D. L., Moore, D. H., Vandeweerd, C., & Gavin Dreschnack, D. J. (2007). Mapping the maze of terms and definitions in dementia related wandering. Aging & Mental Health, 11 (6), 686 698. Algase, D. L., Son, G., Beattie, E., Song, J. A., Leitsch, S., & Yao, L. (2004). The interrelatedness of wandering and wayfinding in a community sample of persons with dementia. Dementia and Geriatric Cognitive Disorders, 17 (3), 231 239. Algase, D. L., Yao, L., Beel Bates, C. A., & Song, J. (2007). Theoretical models of wandering. In A. L. Nelson & D. L. Algase (Eds.), Evidence based protocols for managing wandering behaviors (pp. 19 51). New York: Springer Publishing Company. Allen Burge, R., Stevens, A. B., & Burgi o, L. D. (1999). Effective behavioral interventions for decreasing dementia related challenging behavior in nursing homes. International Journal of Geriatric Psychiatry, 14 (3), 213 228. preventing it. Retrieved February 12, 2011, from http://www.alz.org/national/documents/topicsheet_wandering.pdf

PAGE 190

190 18, 2011, from http://www.alz.org/downloads/Facts_Figures_2011.pdf American Geriatric Society (2002). Clinical practice guidelines: The management of persistent pain in older persons. Journal of the American Geriatrics Society, 50 (Suppl. 6), 205 224. Anderson, R. L., Buckwalter, K. C., Buchanan, R. J., Maas, M. L., & Imhof, S. L. (2003). Validity and reliability of the Minimum Data Set depression rating scale (MDSDRS) for older adults in nursing homes. Age and Ageing, 32 (4), 435 438. Armstrong, J., Glenny, C., Stolee, P., & Berg, K. (2010). A comparison of two assessment systems in predicting fun ctional outcomes of older rehabilitation patients. Age and Ageing, 39 (3), 394 399. Ata, T., Terada, S., Yokota, O., Ishihara, T., Fujisawa, Y., Sasaki, K., et al. (2010). Wandering and fecal smearing in people with dementia. International Psychogeriatrics, 22 (3), 493 500. Aud, M. A. (2004). Dangerous wandering: Elopements of older adults with dementia from long term care facilities. American Journal of Alzheimer's Disease and Other Dementias, 19 (6), 361 368. Ayalon, L., Gum, A. M., Feliciano, L., & Arean, P A. (2006). Effectiveness of nonpharmacological interventions for the management of neuropsychiatric symptoms in patients with dementia: a systematic review. Archives of Internal Medicine, 166 (20), 2182 2188. Ballard, C., Margallo Lana, M., Fossey, J., Re ichelt, K., Myint, P., Potkins, D., et al. (2001). A 1 year follow up study of behavioral and psychological symptoms in dementia among people in care environments [Abstract]. Journal of Clinical Psychiatry, 62 (8), 631 636. Ballard, C., O'Brien, J., James, I., Mynt, P., Lana, M., Potkins, D., et al. (2001). Quality of life for people with dementia living in residential and nursing home care: The impact of performance on activities of daily living, behavioral and psychological symptoms, language skills, and p sychotropic drugs. International Psychogeriatrics, 13 (1), 93 106. Ballard, C., Waite, J., & Birks, J. (2006). Atypical antipsychotics for aggression and psychosis in Alzheimer's disease. Cochrane Database of Systematic: Reviews, 1, CDC0034786 Baron, R. M. & Kenny, D. A. (1986). The moderator mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51 (6), 1173 1182. Bartels, S. J., Horn, S. D., Smout R. J., Dums, A. R., Flaherty, E., Jones, J. K., et al. (2003). Agitation and depression in frail nursing home elderly patients with dementia: Treatment characteristics and service use. American Journal of Geriatric Psychiatry, 11 (2), 231 238.

PAGE 191

191 Baskaya, A. Wilson, C., & Ozcan, Y. Z. (2004). Wayfinding in an unfamiliar environment: Different spatial settings of two polyclinics. Environment and Behavior, 36 (6), 839 867. Bates Jensen, B. M., Simmons, S. F., Schnelle, J. F., & Alessi, C. (2005). Evaluating the accuracy of Minimum Data Set bed mobility ratings against independent performance assessments: Systematic error and directions for improvement. The Gerontologist, 45 (6), 731 738. Beattie, E. R., Song, J., & LaGore, S. (2005). A comparison of wandering beh avior in nursing homes and assisted living facilities. Research and Theory for Nursing Practice, 19 (2), 181 196. Beck, C., Frank, L., Chumbler, N. R., O'Sullivan, P., Vogelpohl, T. S., Rasin, J., et al. (1998). Correlates of disruptive behavior in severely cognitively impaired nursing home residents. Gerontologist, 38 (2), 189 198. Beck, C., Richards, K., Lambert, C., Doan, R., Landes, R. D., Whall, A., et al. (2011). Factors associated with problematic vocalizations in Nursing Home residents with dementia. Gerontologist Bennett, J. A. (2000). Mediator and moderator variables in nursing research: Conceptual and statistical differences. Research in Nursing & Health, 23 (5), 415 420. Bharucha, A. J., Vasilescu, M., Dew, M. A., Begley, A., Stevens, S., Degenholt z, H., et al. (2008). Prevalence of behavioral symptoms: Comparison of the Minimum Data Set assessments with research instruments. Journal of the American Medical Directors Association, 9 (4), 244 250. Black, B. S., Finucane, T., Baker, A., Loreck, D., Blas s, D., Fogarty, L., et al. (2006). Health problems and correlates of pain in nursing home residents with advanced dementia. Alzheimer Disease and Associated Disorders, 20 (4), 283 290. Boerlage, A. A., van Dijk, M., Stronks, D. L., de Wit, R., & van der Rij t, C. C. (2008). Pain prevalence and characteristics in three Dutch residential homes. European Journal of Pain, 12 (7), 910 916. Bourbonniere, M., Strumpf, N. E., Evans, L. K., & Maislin, G. (2003). Organizational characteristics and restraint use for hospitalized nursing home residents. Journal of the American Geriatrics Society, 51 (8), 1079 1084. Braun, U. K., & Kunik, M. E. (2004). Behavioral disturbances in dementia: Finding t he cause(s). Geriatrics, 59 (12), 32 34. Brodaty, H., Draper, B., Saab, D., Low, L. F., Richards, V., Paton, H., et al. (2001). Psychosis, depression and behavioural disturbances in Sydney nursing home residents: Prevalence and predictors. International Journal of Geriatric Psychiatry, 16 504 512.

PAGE 192

192 Bruckenthal, P., Reid, M. C., & Reisner, L. (2009). Special issues in the management of chronic pain in older adults. Pain Medicine, 10 (Suppl. 2), 67 78. Buffum, M. D., Miaskowski, C., Sands, L., & Brod, M. (2 001). A pilot study of the relationship between discomfort and agitation in patients with dementia. Geriatric Nursing, 22 (2), 80 85. Bula, C. J., & Wietlisbach, V. (2009). Use of the cognitive performance scale (CPS) to detect cognitive impairment in the a cute care setting: Concurrent and predictive validity. Brain Research Bulletin, 80 (4 5), 173 178. Burgio, L. D., Butler, F. R., Roth, D. L., Hardin, J. M., Hsu, C. C., & Ung, K. (2000). Agitation in nursing home residents: the role of gender and social con text. Int Psychogeriatr, 12 (4), 495 511. Burgio, L. D., Park, N. S., Hardin, J. M., & Sun, F. (2007). A longitudinal examination of agitation and resident characteristics in the nursing home. Gerontologist, 47 (5), 642 649. Burgio, L. D., Scilley, K., Hardi n, J. M., & Hsu, C. (2001). Temporal patterns of disruptive vocalization in elderly nursing home residents. International Journal of Geriatric Psychiatry, 16 (4), 378 386. Burrows, A. B., Morris, J. N., & Simon, S. E., et al. (2000). Development of an Minim um Data Set based depression rating scale for use in nursing homes. Age and Ageing, 29 165 172. Cadogan, M. P., Schnelle, J. F., Yamamoto Mitani, N., Cabrera, G., & Simmons, S. F. (2004). A Minimum Data Set prevalence of pain quality indicator: Is it accu rate and does it reflect differences in care processes? Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, 59 (3), 281 285. Calkins, M. P. (2004). Articulating environmental press in environments for people with dementia. Alzheime r's Care Quarterly, 5 (2), 165 172. Camberg, L., Woods, P., Ooi, W. L., Hurley, A., Volicer, L., Ashley, J., et al. (1999). Evaluation of Simulated Presence: A personalized approach to enhance well being in persons with Alzheimer's disease. Journal of the A merican Geriatrics Society, 47 (4), 446 452. Capezuti, E., Maislin, G., Strumpf, N., & Evans, L. K. (2002). Side rail use and bed related fall outcomes among nursing home residents. Journal of the American Geriatrics Society, 50 (1), 90 96. Carpenter, G. I., Hastie, C. L., Morris, J. N., Fries, B. E., & Ankri, J. (2006). Measuring change in activities of daily living in nursing home residents with moderate to severe cognitive impairment. BMC Geriatrics, 6 1 8. Casten, R., Lawton, M. P., Parmelee, P. A., & Kl eban, M. H. (1998). Psychometric characteristics of the minimum data set I: Confirmatory factor analysis. Journal of the American Geriatrics Society, 46 (6), 726 735.

PAGE 193

193 Centers for Medicare & Medicaid Services. (2010). MDS Quality indicator report. Retrieve d March 4, 2011, from https:// www.cms.gov/MDSPubQIandResRep/03_qireports.asp Certification and Reporting Database. (2009). Annual numbers of nursing homes, beds, and residents are based on a 15 month OSCAR reporting cycle. Retrieved February 8, 2011, from http://www.cdc.gov/nchs/data/hus/hus2009tables/Table119.pdf Chafetz, P. K. (1990). Two dimensional grid is i neffective against demented patients' exiting through glass doors. Psychology and Aging, 5 (1), 146 147. Charlson, M., Pompei, P., Ales, K. L., & MacKenzie, C. R. (1987). A new method of classifying prognostic comorbidity in longitudinal studies: Developmen t and validation. Journal of Chronic Diseases, 40 (5), 373 383. Charlson, M., Szatrowski, T. P., Peterson, J., & Gold, J. (1994). Validation of a combined comorbidity index. Journal of Clinical Epidemiology, 47 (11), 1245 1251. Chen, J. C., Borson, S., & Scanlan, J. M. (2000). Stage specific prevalence of behavioral symptoms in Alzheimer's disease in a multi ethnic community sample. American Journal of Geriatric Psychiatry, 8 (2), 123 133. Chibnall, J. T., & Tait, R. C. (2001). Pain assessment in cognitive ly impaired and unimpaired older adults: A comparison of four scales. Pain, 92 (1 2), 173 186. Chu, L., Schnelle, J. F., Cadogan, M. P., & Simmons, S. F. (2004). Using the Minimum Data Set to select nursing home residents for interview about pain. Journal o f the American Geriatrics Society, 52 (12), 2057 2061. Cipher, D. J., & Clifford, P. A. (2004). Dementia, pain, depression, behavioral disturbances, and ADLs: Toward a comprehensive conceptualization of quality of life in long term care. International Journ al of Geriatric Psychiatry, 19 (8), 741 748. Clark, M. E., Lipe, A. W., & Bilbrey, M. (1998). Use of music to decrease aggressive behaviors in people with dementia. Journal of Gerontological Nursing, 24 (7), 10 17. Cohen Mansfield, J. (2000). Theoretical fra mework for behavioral problems in dementia [Abstract]. Alzheimer's Care Quarterly, 1 (4), 8 21. Cohen Mansfield, J. (2004). The adequacy of the minimum data set assessment of pain in cognitively impaired nursing home residents. Journal of Pain and Symptom M anagement, 27 (4), 343 351. Cohen Mansfield, J., & Billig, N. (1986). Agitated behaviors in the elderly. I. A conceptual review. Journal of the American Geriatrics Society, 34 (10), 711 721. Cohen Mansfield, J., & Jensen, B. (2008). Assessment and treatment approaches for behavioral disturbances associated with dementia in the nursing home: self reports of physicians' practices. J Am Med Dir Assoc, 9 (6), 406 413.

PAGE 194

194 Cohen Mansfield, J., & Libin, A. (2005). Verbal and physical non aggressive agitated behaviors in elderly persons with dementia: Robustness of syndromes. Journal of Psychiatric Research, 39 (3), 325 332. Cohen Mansfield, J., Taylor, L., McConnell, D., & Horton, D. (1999). Estimating the cognitive ability of nursing home residents from the minimum data set [Abstract]. Outcomes Management for Nursing Practice, 3 (1), 43 46. Cohen, C. I., Hyland, K., & Magai, C. (1998). Interracial and intraracial difference s in neuropsychiatric symptoms, sociodemography, and treatment among nursing home patients with dementia. Gerontologist, 38 (3), 353 361. Connor, K., McNeese Smith, D., van Servellen, G., Chang, B., Lee, M., Cheng, E., et al. (2009). Insight into dementia c are management using social behavioral theory and mixed methods. Nursing Research, 58 (5), 348 358. Cordner, Z., Blass, D. M., Rabins, P. V., & Black, B. S. (2010). Quality of life in nursing home residents with advanced dementia. Journal of the American Ge riatrics Society, 58 (12), 2394 2400. Cotter, V. T. (2005). Restraint free care in older adults with dementia. The Keio Journal of Medicine, 54 (2), 80 84. Cronin Golomb, A. (1995). Vision in Alzheimer's disease. Gerontologist, 135 (3), 370 376. Cunningham, C ., McClean, W., & Kelly, F. (2010). The assessment and management of pain in people with dementia in care homes. Nursing Older People, 22 (7), 29 35. Day, K., Carreon, D., & Stump, C. (2000). The therapeutic design of environments for people with dementia: a review of the empirical research. Gerontologist, 40 (4), 397 416. Desai, A. K., & Grossberg, G. T. (2001). Recognition and Management of Behavioral Disturbances in Dementia. The Primary Care Companion to The Journal of Clinical Psychiatry, 3 (3), 93 109. D iederichs, C., Berger, K., & Bartels, D. B. (2011). The measurement of multiple chronic diseases: A systematic review on existing multimorbidity indices. Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, 66 (3), 301 311. Doorn, C. V., Gruber Baldini, A. L., Zimmerman, S., Hebel, J. R., Port, C. L., Baumgarten, M., et al. (2003). Dementia as a risk factor for falls and fall Injuries among Nursing Home residents. Journal of the American Geriatrics Society, 51 (9), 1213 1218. Edgman Le vitan, S., & Cleary, P. D. (1996). What information do consumers want and need? Health Affairs, 15 42 56. Elliott, A. F., & Horgas, A. L. (2009). Effects of an analgesic trial in reducing pain behaviors in community dwelling older adults with dementia. Nursing Research, 58 (2), 140 145.

PAGE 195

195 Eritz, H., & Hadjistavropoulos, T. (2010). Do informal caregivers consider nonverbal behavior when they assess pain in people with severe dementia? The Journal of Pain, 12 (3), 331 339. Eustace, A., Kidd, N., Greene, E., Fa llon, C., Bhrain, S. N., Cunningham, C., et al. (2001). Verbal aggression in Alzheimer's disease. Clinical, functional and neuropsychological correlates. International Journal of Geriatric Psychiatry, 16 (9), 858 861. Evans, L. K., & Cotter, V. T. (2008). A voiding restraints in patients with dementia: Understanding, prevention, and management are the keys. The American Journal of Nursing, 108 (3), 40 49. Feldt, K. S. (2000). The checklist of nonverbal pain indicators (CNPI). Pain Management Nursing, 1 (1), 13 21. Feliciano, L., Vore, J., LeBlanc, L. A., & Baker, J. C. (2004). Decreasing entry into a restricted area using a visual barrier. Journal of Applied Behavior Analysis, 37 (1), 107 110. Field, A. (2005). Discovering statistics using SPSS London: Sage Publ ications. Finkel, S. I., Mintzer, J. E., Dysken, M., Krishnan, K. R., Burt, T., & McRae, T. (2004). A randomized, placebo controlled study of the efficacy and safety of sertraline in the treatment of the behavioral manifestations of Alzheimer's disease in outpatients treated with donepezil. Int J Geriatr Psychiatry, 19 (1), 9 18. Fisher, S. E., Burgio, L. D., Thorn, B. E., Allen Burge, R., Gerstle, J., Roth, D. L., et al. (2002). Pain assessment and management in cognitively impaired nursing home residents: Association of certified nursing assistant pain report, Minimum Data Set pain report, and analgesic medication use. Journal of the American Geriatrics Society, 50 (1), 152 156. Flannery, R. B., Jr. (2003). Restraint procedures and dementia sufferers with ps ychological trauma. American Journal of Alzheimer's Disease and Other Dementias, 18 (4), 227 230. Fleming, R., & Purandare, N. (2010). Long term care for people with dementia: environmental design guidelines. International Psychogeriatrics, 22 (7), 1084 1096 Frederiksen, K., Tariot, P., & De Jonghe, E. (1996). Minimum Data Set Plus (MDS+) scores compared with scores from five rating scales. Journal of the American Geriatrics Society, 44 (3), 305 309. Fries, B. E., Simon, S. E., Morris, J. N., Flodstrom, C., & Bookstein, F. L. (2001). Pain in U.S. nursing homes: Validating a pain scale for the minimum data set. Gerontologist, 41 (2), 173 179. Fuchs Lacelle, S., & Hadjistavropoulos, T. (2004). Development and preliminary validation of the pain assessment checklis t for seniors with limited ability to communicate (PACSLAC). Pain Management Nursing, 5 (1), 37 49.

PAGE 196

196 Gallinagha, R., Nevinb, R., Ilroyc, D. K., Mitchellb, F., Campbellb, L., Ludwickd, R., et al. (2002). The use of physical restraints as a safety measure in t he care of older people in four rehabilitation wards: Findings from an exploratory study. International Journal of Nursing Studies, 39 (2), 147 156. Gastmans, C., & Milisen, K. (2006). Use of physical restraint in nursing homes: Clinical ethical considerati ons. Journal of Medical Ethics, 32 148 152. Gaugler, J. E., Wall, M. M., Kane, R. L., Menk, J. S., Sarsour, K., Johnston, J. A., et al. (2010). The effects of incident and persistent behavioral problems on change in caregiver burden and nursing home admis sion of persons with dementia. Medical Care, 48 (10), 875 883. Gerritsen, D. L., Achterberg, W. P., Steverink, N., Pot, A. M., Frijters, D. H., & Ribbe, M. W. (2008). The MDS challenging behavior profile for long term care. Aging & Mental Health, 12 (1), 116 123. Gibson, S. J., Voukelatos, X., Ames, D., Flicker, L., & Helme, R. D. (2001). An examination of pain perception and cerebral event related potentials following carbon dioxide laser stimulation in patients with Alzheimer's disease and age matched co ntrol volunteers. Pain Research & Management, 6 (3), 126 132. Goldberg, R. J. (2002). Management of behavioral complications of dementia. Medicine and Health, Rhode Island, 85 (9), 281 285. Golightly, Y. M., & Dominick, K. L. (2005). Racial variations in sel f reported osteoarthritis symptom severity among veterans. Aging Clinical and Experimental Research, 17 (4), 264 269. Goodall, D., & Etters, L. (2005). The therapeutic use of music on agitated behavior in those with dementia. Holistic Nursing Practice, 19 (6 ), 258 262. Grando, V. T., Rantz, M. J., & Maas, M. (2007). Nursing home staff's views on quality improvement interventions: A follow up study. Journal of Gerontological Nursing, 33 (1), 40 47. Green, C. R., Anderson, K. O., Baker, T. A., Campbell, L. C., Decker, S., Fillingim, R. B., et al. (2003). The unequal burden of pain: Confronting racial and ethnic disparities in pain. Pain Medicine, 4 (3), 277 294. Gruber Baldini, A. L., Zimmerman, S. I., Mortimore, E., & Magaziner, J. (2000). The validity of the minimum data set in measuring the cognitive impairment of persons admitted to nursing homes. Journal of the American Geriatrics Society, 48 (12), 1601 1606. Gunes, U. Y. (2008). A descriptive study of pressure ulcer pain. Ostomy Wound Management, 54 (2), 56 61.

PAGE 197

197 Guttman, R., Altman, R. D., & Karlan, M. S. (1999). Report of the Council on Scientific Affairs. Use of restraints for patients in nursing homes. Archives of Family Medicine, 8 (2), 101 105. Ha djistavropoulos, T., & Craig, K. D. (2002). A theoretical framework for understanding self report and observational measures of pain: A communications model. Behaviour Research and Therapy, 40 (5), 551 570. Hadjistavropoulos, T., LaChapelle, D. L., MacLeod, F. K., Snider, B., & Craig, K. D. (2000). Measuring movement exacerbated pain in cognitively impaired frail elders [Abstract]. Clinical Journal of Pain, 16 (1), 54 63. Haight, B. K., Gibson, F., & Michel, Y. (2006). The Northern Ireland life review/life st orybook project for people with dementia. Alzheimer's & Dementia, 2 (1), 56 58. Hamers, J. P. H., Gulpers, M. J. M., & Strik, W. (2004). Use of physical restraints with cognitively impaired nursing home residents. Journal of Advanced Nursing, 45 (3), 246 251 Harrow, B. S., Tennstedt, S. L., & McKinlay, J. B. (1995). How costly is it to care for disabled elders in a community setting? Gerontologist, 35 (6), 803 813. Hartmaier, S. L., Sloane, P. D., Guess, H. A., & Koch, G. G. (1994). The MDS cognition scale: A valid instrument for identifying and staging nursing home residents with dementia using the minimum data set. Journal of the American Geriatrics Society, 42 (11), 1212 1213. Hartmaier, S. L., Sloane, P. D., Guess, H. A., Koch, G. G., Mitchell, C. M., & Phi llips, C. D. (1995). Validation of the Minimum Data Set cognitive performance scale: Agreement with the Mini Mental State Examination. Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, 50A (2), 128 133. Hawes, C., Morris, J. N., P hillips, C. D., Fries, B. E., Murphy, K., & Mor, V. (1997). Development of the nursing home resident assessment instrument in the USA. Age and Ageing, 26 (Suppl. 2), 19 25. Hawes, C., Morris, J. N., Phillips, C. D., Mor, V., Fries, B. E., & Nonemaker, S. (1 995). Reliability estimates for the Minimum Data Set for nursing home resident assessment and care screening (MDS). Gerontologist, 35 (2), 172 178. Hendrix, C. C., Sakauye, K. M., Karabatsos, G., & Daigle, D. (2003). The use of the Minimum Data Set to ident ify depression in the elderly. Journal of the American Medical Directors Association, 4 (6), 308 312. Herr, K., Bjoro, K., & Decker, S. (2006). Tools for assessment of pain in nonverbal older adults with dementia: A state of the science review. Journal of P ain and Symptom Management, 31 (2), 170 192.

PAGE 198

198 Herr, K., Coyne, P. J., Key, T., Manworren, R., McCaffery, M., Merkel, S., et al. (2006). Pain assessment in the nonverbal patient: Position statement with clinical practice recommendations. Pain Management Nursi ng, 7 (2), 44 52. Herrmann, N. (2001). Recommendations for the management of behavioral and psychological symptoms of dementia. Canadian Journal Neurological Sciences, 28 (Suppl. 1), 96 107. Hersch, E. C., & Falzgraf, S. (2007). Management of the behavioral and psychological symptoms of dementia. Journal of Clinical Interventions in Aging, 2 (4), 611 621. Hewawasam, L. C. (1996). The use of two dimensional grid patterns to limit hazardous ambulation in elderly patients with Alzheimer's disease. Nursing Times Research, 1 (3), 217 227. Hicks Moore, S. L. (2005). Relaxing music at mealtime in nursing homes: Effects of agitated patients with dementia. Journal of Gerontological Nursing, 31 (12), 26 32. Hirdes, J. P., Frijters, D. H., & Teare, G. F. (2003). The MDS CH ESS scale: A new measure to predict mortality in institutionalized older people. Journal of the American Geriatrics Society, 51 (1), 96 100. Hirdes, J. P., Smith, T. F., Rabinowitz, T., Yamauchi, K., Perez, E., Telegdi, N. C., et al. (2002). The Resident As sessment Instrument Mental Health (RAI MH): Inter rater reliability and convergent validity. Journal of Behavioral Health Services and Research, 29 (4), 419 432. Holmbeck, G. N. (1997). Toward terminological, conceptual, and statistical clarity in the study of mediators and moderators: Examples from the child clinical and pediatric psychology literatures. Journal of Consulting and Clinical Psychology, 65 (4), 599 610. Holtzer, R., Tang, M. X., Devanand, D. P., Albert, S. M., Wegesin, D. J., Marder, K., et al. (2003). Psychopathological features in Alzheimer's disease: Course and relationship with cognitive status. Journal of the American Geriatrics Society, 51 (7), 953 9 60. Hong, G. R., & Song, J. A. (2009). Relationship between familiar environment and wandering behaviour among Korean elders with dementia. Journal of Clinical Nursing, 18 (9), 1365 1373. Horgas, A., & Dunn, K. (2001). Pain in nursing home residents. Compar ison of residents' self report and nursing assistants' perceptions. Incongruencies exist in resident and caregiver reports of pain; therefore, pain management education is needed to prevent suffering. Journal of Gerontological Nursing, 27 (3), 44 53. Horgas A., & Elliott, A. F. (2004). Pain assessment and management in persons with dementia. The Nursing Clinics of North America, 39 (3), 593 606. Horgas, A., Elliott, A. F., & Marsiske, M. (2009). Pain assessment in persons with dementia: Relationship between self report and behavioral observation. Journal of the American Geriatrics Society, 57 (1), 126 132.

PAGE 199

199 Horgas, A., & Margrett, J. A. (2001). Measuring behavioral and mood disruptions in nursing home residents using the Minimum Data Set. Outcomes Management fo r Nursing Practice, 5 (1), 28 35. Horgas, A., & Miller, L. (2008). Pain assessment in people with dementia. The American Journal of Nursing, 108 (7), 62 70. Horgas, A., Nichols, A. L., Schapson, C. A., & Vietes, K. (2007). Assessing pain in persons with deme ntia: Relationships among the non communicative patient's pain assessment instrument, self report, and behavioral observations. Pain Management Nursing, 8 (2), 77 85. Horgas, A., & Tsai, P. F. (1998). Analgesic drug prescription and use in cognitively impai red nursing home residents. Nursing Research, 47 (4), 235 242. Horgas, A., Yoon, S. L., Nichols, A. L., & Marsiske, M. (2008). The relationship between pain and functional disability in Black and White older adults. Research in Nursing & Health, 31 (4), 341 354. Huabin, L., Lin, M., & Castle, N. (2011). Physical restraint use and falls in nursing homes: A comparison between residents with and without dementia. American Journal of Alzheimer's Disease and Other Dementias, 26 (1), 44 50. Hutchinson, A. M., Milke, D. L., Maisey, S., Johnson, C., Squires, J. E., Teare, G., et al. (2010). The resident assessment instrument Minimum Data Set 2.0 quality indicators: A systematic review. BMC Health Services Research, 10 Hwang, J. P., Yang, C. H., Tsai, S. J., & Liu, K. M. (1997). Behavioural disturbances in psychiatric inpatients with dementia of the Alzheimer's type in Taiwan. International Journal of Geriatric Psychiatry, 12 (9), 902 906. Jansen, G., Dassen, T., & Moorer, P. (1997). The perception of aggression. Scandin avian Journal of Caring Sciences, 11 (1), 51 55. Jones, K. R., Fink, R. M., Clark, L., Hutt, E., Vojir, C. P., & Mellis, B. K. (2006). Nursing home resident barriers to effective pain management: Why nursing home residents may not seek pain medication. Jour nal of the American Medical Directors Association, 7 (Suppl. 3), 21 28. Kanamori, M., Suzuki, M., Yamamoto, K., Kanda, M., Matsui, Y., Kojima, E., et al. (2001). A day care program and evaluation of animal assisted therapy (AAT) for the elderly with senile dementia. American Journal of Alzheimer's Disease and Other Dementias, 16 (4), 234 239. Kiely, D. K., Morris, J. N., & Algase, D. L. (2000). Resident characteristics associated with wandering in nursing homes. International Journal of Geriatric Psychiatry, 15 (11), 1013 1020.

PAGE 200

200 Kincaid, C., & Peacock, J. R. (2003). The effect of a wall mural on decreasing four types of door testing behaviors. Journal of Applied Gerontology, 22 (1), 76 88. Kirkevold, ., Laake, K., & Engedal, K. (2003). Use of constraints and sur veillance in Norwegian wards for the elderly. International Journal of Geriatric Psychiatry, 18 (6), 491 497. Kong, E. H. (2005). Agitation in dementia: Concept clarification. Journal of Advanced Nursing, 52 (5), 526 536. Kopetz, S., Steele, C. D., Brandt, J ., Baker, A., Kronberg, M., Galik, E., et al. (2000). Characteristics and outcomes of dementia residents in an assisted living facility. International Journal of Geriatric Psychiatry, 15 (7), 586 593. Kovach, C. R., Noonan, P. E., Schlidt, A. M., Reynolds, S., & Wells, T. (2006). The serial trial intervention: An innovative approach to meeting needs of individuals with dementia. Journal of Gerontological Nursing, 32 (4), 18 25. Kovach, C. R., Weissman, D. E., Griffie, J., Matson, S., & Muchka, S. (1999). Asse ssment and treatment of discomfort for people with late stage dementia. Journal of Pain and Symptom Management, 18 (6), 412 419. Kunik, M. E., Snow, A. L., Davila, J. A., McNeese, T., Steele, A. B., Balasubramanyam, V., et al. (2010). Consequences of aggressive behavior in patients with dementia. The Journal of Neuropsychiatry and Clinical Neurosciences, 22 (1), 40 47. Kunik, M. E., Walgama, J. P., Snow, A. L., Davila, J. A., Schulz, P. E., Steele, A. B., et al. (2007). Documentation, assessment, and treatment of aggression in patients with newly diagnosed dementia. Alzheimer Disease and Associated Disorders, 21 (2), 115 121. Kunz, M., Mylius, V., Scharmann, S., Schepelman, K., & Lautenbacher, S. (2009). Influence of dementia on multiple components of pain. European Journal of Pain, 13 (3), 317 325. Kutsumi, M., Ito, M., Sugiura, K., Terabe, M., & Mikami, H. (2009). Management of behavio ral and psychological symptoms of dementia in long term care facilities in Japan. Psychogeriatrics, 9 (4), 186 195. Lai, C. K., & Arthur, D. G. (2003). Wandering behaviour in people with dementia. Journal of Advanced Nursing, 44 (2), 173 182. Lawton, M. P., Casten, R., Parmelee, P. A., Haitsma, K. V., Corn, J., & Kleban, M. H. (1998). Psychometric characteristics of the Minimum Data Set II: Validity. Journal of the American Geriatrics Society, 46 (6), 736 744. Lee, F. P., Leppa, C., & Schepp, K. (2006). Using the Minimum Data Set to determine predictors of terminal restlessness among nursing home residents. Journal of Nursing Research, 14 (4), 286 296.

PAGE 201

201 Leonard, R., Tinetti, M. E., Allore, H. G., & Drickamer, M. A. (2006). Potentially mo difiable resident characteristics that are associated with physical or verbal aggression among nursing home residents with dementia. Archives of Internal Medicine, 166 (12), 1295 1300. Libin, A., & Cohen Mansfield, J. (2004). Therapeutic robocat for nursing home residents with dementia: Preliminary inquiry. American Journal of Alzheimer's Disease and Other Dementias, 19 (2), 111 116. Lucero, M. (2002). Intervention strategies for exit seeking wandering behavior in dementia residents. American Journal of Alzhe imer's Disease and Other Dementias, 17 (5), 277 280. Lum, T. Y., Lin, W. C., & Kane, R. L. (2005). Use of proxy respondents and accuracy of Minimum Data Set assessments of activities of daily living. Journals of Gerontology. Series A, Biological Sciences an d Medical Sciences, 60 (5), 654 659. Mace, N. L., & Rabins, P. V. (2006). The 36 hour day: A family guide to caring for people with Alzheimer Disease, other dementias, and memory loss in later life (4th ed.). Baltimore: The Johns Hopkins University Press. M agaziner, J., German, P., Zimmerman, S. I., Hebel, J. R., Burton, L., Gruber Baldini, A. L., et al. (2000). The prevalence of dementia in a statewide sample of new nursing home admissions aged 65 and older: Diagnosis by expert panel. Epidemiology of Dement ia in Nursing Homes Research Group. Gerontologist, 40 (6), 663 672. Manfredi, P. L., Breuer, B., Wallenstein, S., Stegmann, M., Bottomley, G., & Libow, L. (2003). Opioid treatment for agitation in patients with advanced dementia. International Journal of Ge riatric Psychiatry, 18 (8), 700 705. Margallo Lana, M., Swann, A., O'Brien, J., Fairbairn, A., Reichelt, K., Potkins, D., et al. (2001). Prevalence and pharmacological management of behavioural and psychological symptoms amongst dementia sufferers living in care environments. International Journal of Geriatric Psychiatry, 16 39 44. Matsuoka, K., Miyamoto, Y., Ito, H., & Kurita, H. (2003). Relationship between behavioral disturbances and characteristics of patients in special units for dementia. Psychiatry a nd Clinical Neurosciences, 57 (6), 569 574. McAuliffe, L., Nay, R., O'Donnell, M., & Fetherstonhaugh, D. (2009). Pain assessment in older people with dementia: Literature review. Journal of Advanced Nursing, 65 (1), 2 10. McConnell, E. S., Branch, L. G., Slo ane, R. J., & Pieper, C. F. (2003). Natural history of change in physical function among long stay nursing home residents. Nursing Research, 52 (2), 119 126. McConnell, E. S., Pieper, C. F., Sloane, R. J., & Branch, L. G. (2002). Effects of cognitive perfor mance on change in physical function in long stay nursing home residents. Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, 57 (12), 778 784.

PAGE 202

202 McCurren, C. (2002). Assessment for depression among nursing home elders: Evaluation of the MDS mood assessment Geriatric Nursing, 23 (2), 103 108. Meleis, A. I. (2007). Theoretical nursing: Development & progress (4th ed.). Philadelphia: Lippincott Williams & Wilkins. Menon, A. S., Gruber Baldini, A. L., Hebel, J. R., Kaup, B., Loreck, D., Zi mmerman, S. I., et al. (2001). Relationship between aggressive behaviors and depression among nursing home residents with dementia. International Journal of Geriatric Psychiatry, 16 (2), 139 146. Meyers, L. S., Gamst, G., & Guarino, A. J. (2006). Applied mu ltivariate research Thousand Oaks, California: Sage Publications. Mor, V. (2004). A comprehensive clinical assessment tool to inform policy and practice: Applications of the Minimum Data Set. Medical Care, 42 (4), 50 59. Mor, V., Berg, K., Angelelli, J., Gifford, D., Morris, J., & Moore, T. (2003). The quality of quality measurement in U.S. nursing homes. Gerontologist, 43 Spec No 2 37 46. Morris, J. N., Fries, B. E., Mehr, D. R., Hawes, C., Phillips, C., Mor, V., et al. (1994). MDS cognitive performance scale. Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, 49 (4), 174 182. Morris, J. N., Fries, B. E., & Morris, S. A. (1999). Scaling ADLs within the MDS. Journals of Gerontology. Series A, Biolo gical Sciences and Medical Sciences, 54 (11), 546 553. Morris, J. N., Hawes, C., Fries, B. E., Phillips, C. D., Mor, V., Katz, S., et al. (1990). Designing the national resident assessment instrument for nursing homes. Gerontologist, 30 (3), 293 307. Morriso n, R. S., & Siu, A. L. (2000). A comparison of pain and its treatment in advanced dementia and cognitively intact patients with hip fracture. Journal of Pain and Symptom Management, 19 (4), 240 248. National Center for Health Statistics. (2006). 2004 Nation al Nursing Home Survey. Retrieved February 22, 2011, from http://www.cdc.gov/nchs/data/nnhsd/nursinghomefacilities2006.pdf National Center for Health Statistics. (2009). Hea lth, United States, 2009 with special features on medical technology. Retrieved February 8, 2011, from http://www.cdc.gov/nchs/data/hus/hus09.pdf#105 Nelson, J. (1995). The influence of environmental factors in incidents of disruptive behavior. Journal of Gerontological Nursing, 21 (5), 19 24. Nemeth, K. A., Harrison, M. B., Graham, I. D., & Burke, S. (2004). Understanding venous leg ulcer pain: results of a longitudinal study. Ostomy Woun d Manage, 50 (1), 34 46.

PAGE 203

203 Norton, M. J., Allen, R. S., Snow, A. L., Hardin, J. M., & Burgio, L. D. (2010). Predictors of need driven behaviors in nursing home residents with dementia and associated certified nursing assistant burden. Aging & Mental Health, 1 4 (3), 303 309. Nsman, B., Bucht, G., Eriksson, S., & Sandman, P. O. (1993). Behavioural symptoms in the institutionalized elderly: Relationship to dementia. International Journal of Geriatric Psychiatry, 8 (10), 843 849. Olin, J. T., Fox, L. S., Pawluczyk, S., Taggart, N. A., & Schneider, L. S. (2001). A pilot randomized trial of carbamazepine for behavioral symptoms in treatment resistant outpatients with Alzheimer disease. American Journal of Geriatric Psychiatry, 9 (4), 400 405. Ott, B. R., Lapane, K. L., & Gambassi, G. (2000). Gender differences in the treatment of behavior problems in Alzheimer's disease. Neurology, 54 (2), 427 432. Ouslander, J. G. (1994). Maximizing the Minimum Data Set. Journal of the American Geriatr ics Society, 42 (11), 1212 1213. Palese, A., Menegazzo, E., Baulino, F., Pistrino, R., & Papparotto, C. (2009). The effectiveness of multistrategies on disruptive vocalization of people with dementia in institutions: A multicentered observational study. Jou rnal of Neuroscience Nursing, 41 (4), 191 200. nursing homes. Journal of the American Medical Directors Association, 10 (2), 98 106. Parmelee, P. A., Smith, B., & Katz, I. R. (1993). Pain complaints and cognitive status among elderly institution residents. Journal of the American Geriatrics Society, 41 (5), 517 522. Patel, V., & Hope R. A. (1992). A rating scale for aggressive behaviour in the elderly: The RAGE. Psychological Medicine, 22 (1), 211 221. Pautex, S., Herrmann, F. R., Michon, A., Giannakopoulos, P., & Gold, G. (2007). Psychometric properties of the Doloplus 2 observationa l pain assessment scale and comparison to self assessment in hospitalized elderly. Clinical Journal of Pain, 23 (9), 774 779. Perlman, C. M., & Hirdes, J. P. (2008). The aggressive behavior scale: A new scale to measure aggression based on the Minimum Data Set. Journal of the American Geriatrics Society, 56 (12), 2298 2303. Petek Ster, M., & Cedilnik Gorup, E. (2011). Psychotropic medication use a mong elderly nursing home residents in Slovenia: Cross sectional study. Croatian Medical Journal, 52 (1), 16 24.

PAGE 204

204 Pieper, M. J., Achterberg, W. P., Francke, A. L., van der Steen, J. T., Scherder, E. J., & Kovach, C. R. (2011). The implementation of the seria l trial intervention for pain and challenging behaviour in advanced dementia patients (STA OP!): A clustered randomized controlled trial. BMC Geriatrics, 11 (1), 12. Polit, D. F. (2010). Statistics and data analysis for nursing research (2nd ed.). New York: Prentice Hall. Polit, D. F., & Beck, C. T. (2004). Nursing research: Principles and methods (7th ed.). Philadelphia: Lippincott. Pollock, B. G., Mulsant, B. H., Rosen, J., Sweet, R. A., Mazumdar, S., Bharucha, A., et al. (2002). Comparison of citalopram, perphenazine, and placebo for the acute treatment of psychosis and behavioral disturbances in hospitalized, demented patients. The American Journal of Psychiatry, 159 (3), 460 465. Porsteinsson, A. P., Tariot, P. N., Erb, R., Cox, C., Smith, E., Jakimovich, L., et al. (2001). Placebo controlled study of divalproex sodium for agitation in dementia. Am J Geriatr Psychiatry, 9 (1), 58 66. Rahim Williams, F. B., Riley, J. L., 3rd, Herrera, D., Campbell, C. M., Hastie, B. A., & Fillingim, R. B. (2007). Ethnic iden tity predicts experimental pain sensitivity in African Americans and Hispanics. Pain, 129 (1 2), 177 184. Rahman, A. N., & Applebaum, R. A. (in press). The nursing home Minimum Data Set assessment instrument: Manifest functions and unintended consequences P ast, present, and future. Gerontologist Rantz, M. J., Petroski, G. F., Madsen, R. W., Mehr, D. R., Popejoy, L., Hicks, L. L., et al. (2000). Setting thresholds for quality indicators derived from MDS data for nursing home quality improvement reports: an u pdate. Joint Commission Journal on Quality and Patient Safety, 26 (2), 101 110. Remington, R., Abdallah, L., Melillo, K. D., & Flanagan, J. (2006). Managing problem behaviors associated with dementia. Rehabilitation Nursing, 31 (5), 186 192. Research Data As sistance Center [ResDAC]. (2010). Requesting CMS's Identifiable Data Files (RIFs) New use requests. Retrieved February 12, 2011, from http://www.resdac.org/medicare/requesting_da ta_NewUse.asp Reynolds, K. S., Hanson, L. C., DeVellis, R. F., Henderson, M., & Steinhauser, K. E. (2008). Disparities in pain management between cognitively intact and cognitively impaired nursing home residents. Journal of Pain and Symptom Management, 35 (4), 388 396. Richards, K. C., & Beck, C. K. (2004). Progressively lowered stress threshold model: Understanding behavioral symptoms of dementia. American Geriatrics Society, 52 (10), 1774 1775.

PAGE 205

205 Richter, T., Mann, E., Meyer, G., Haastert, B., & Kopke, S. (i n press). Prevalence of psychotropic medication use among German and Austrian nursing home residents: A comparison of 3 cohorts. Journal of the American Medical Directors Association Riley, J. L., 3rd, Wade, J. B., Myers, C. D., Sheffield, D., Papas, R. K ., & Price, D. D. (2002). Racial/ethnic differences in the experience of chronic pain. Pain, 100 (3), 291 298. Roberts, C. (1999). The management of wandering in older people with dementia. Journal of Clinical Nursing, 8 (3), 322 323. Robinson, L., Hutchings D., Corner, L., Finch, T., Hughes, J., Brittain, K., et al. (2007). Balancing rights and risks: Conflicting perspectives in the management of wandering in dementia. Health, Risk, & Society, 9 (4), 389 406. Rodda, J., Morgan, S., & Walker, Z. (2009). Are c holinesterase inhibitors effective in the management of the behavioral and psychological symptoms of dementia in Alzheimer's disease? A systematic review of randomized, placebo controlled trials of donepezil, rivastigmine and galantamine. International Psy chogeriatrics, 21 (5), 813 824. Rojas Fernandez, C. H., Eng, M., & Allie, N. D. (2003). Pharmacologic management by clinical pharmacists of behavioral and psychological symptoms of dementia in nursing home residents: results from a pilot study. Pharmacother apy, 23 (2), 217 221. Ryden, M. B. (1988). Aggressive behavior in persons with dementia who live in the community. Alzheimer Disease and Associated Disorders, 2 (4), 342 355. Scherder, E. J., Herr, K., Pickering, G., Gibson, S., Benedetti, F., & Lautenbacher, S. (2009). Pain in dementia. Pain, 145 (3), 276 278. Scherder, E. J., Oosterman, J., Swaab, D., Herr, K., Ooms, M., Ribbe, M., et al. (2005). Recent developments in pain in dementia. BMJ, 330 (7489), 461 464. Scherder, E. J., Slaets, J., Deije n, J. B., Gorter, Y., Ooms, M. E., Ribbe, M., et al. (2003). Pain assessment in patients with possible vascular dementia. Psychiatry, 66 (2), 133 145. Schmidt, R., Bach, M., Dal Bianco, P., Holzer, P., Pluta Fuerst, A., Assem Hilger, E., et al. (2010). Deme ntia and pain [Abstract]. Neuropsychiatrie, 24 (1), 1 13. Schneider, L. S., Dagerman, K. S., & Insel, P. (2005). Risk of death with atypical antipsychotic drug treatment for dementia: Meta analysis of randomized placebo controlled trials. JAMA, 294 (15), 193 4 1943. Schnelle, J. F., Wood, S., Schnelle, E. R., & Simmons, S. F. (2001). Measurement sensitivity and the Minimum Data Set depression quality indicator. Gerontologist, 41 (3), 401 405. Schonfeld, L., King Kallimanis, B., Brown, L. M., Davis, D. M., Kearn s, W. D., Molinari, V. A., et al. (2007). Wanderers with cognitive impairment in Department of Veterans Affairs nursing home care units. Journal of the American Geriatrics Society, 55 (5), 692 699.

PAGE 206

206 Schreiner, A. S. (2001). Aggressive behaviors among demente d nursing home residents. International Journal of Geriatric Psychiatry, 16 (2), 209 215. Schreiner, A. S., Yamamoto, E., & Shiotani, H. (2000). Agitated behavior in elderly nursing home residents with dementia in Japan. Journals of Gerontology. Series B, P sychological Sciences and Social Sciences, 55 (3), 180 186. Schuler, M., Njoo, N., Hestermann, M., Oster, P., & Hauer, K. (2004). Acute and chronic pain in geriatrics: Clinical characteristics of pain and the influence of cognition. Pain Medicine, 5 (3), 253 262. Sengupta, M., Bercovitz, A., & Harris Kojetin, L. D. (2010). Prevalence and management of pain, by race and dementia among nursing home residents: United States, 2004. NCHS Data Brief (30), 1 8. Sgadari, A., Morris, J. N., Fries, B. E., Ljunggren, G., Jonsson, P. V., DuPaquier, J. N., et al. (1997). Efforts to establish the reliability of the resident assessment instrument. Age and Ageing, 26 Suppl 2 27 30. Shaping Long Term Care in America Project. (2007). Long term care: Facts on care in the US. R etrieved March 8, 2011, from http://ltcfocus.org/StateTable.aspx Shega, J. W., Emanuel, L., Vargish, L., Levine, S. K., Bursch, H., Herr, K., et al. (2007). Pain in persons with dementia: Complex, common, and challenging. The Journal of Pain, 8 (5), 373 378. Shega, J. W., Ersek, M., Herr, K., Paice, J. A., Rockwood, K., Weiner, D. K., et al. (2010). The multidimensional experience of noncancer pain: Does cognitive status matter? Pain Medicine, 11 (11), 1680 1687. Shega, J. W., Hougham, G. W., Stocking, C. B., Cox Hayley, D., & Sachs, G. A. (2005). Factors associated with self and caregiver report of pain among community dwelling persons with dementia. Journal of Palliative Medicine, 8 (3), 567 575. Sherratt, K., Thornton, A., & Hatton, C. (2004). Emotional and behavioural responses to music in people with dementia: An observational study. Aging & Mental Health, 8 (3), 233 241. Shin, J. H., & Scherer, Y. (2009). Advantages and disadvantages of using MDS data in nursing research. Journal of Gerontological Nursing, 35 (1), 7 17. Siegler, E. L., Capezuti, E., Maislin, G., Baumgarten, M., Evans, L., & Strumpf, N. (1997). Effects of a restraint reduction intervention and OBRA '87 regulations on psychoactive drug use in nursing homes. Journal of the American Geriatrics Society, 45 (7), 791 796. Sink, K. M., Covinsky, K. E., Newcomer, R., & Yaffe, K. (2004). Ethnic differences in the prevalence and pattern of dementia related behaviors. Journal of the American Geriatrics S ociety, 52 (8), 1277 1283.

PAGE 207

207 Sink, K. M., Holden, K. F., & Yaffe, K. (2005). Pharmacological treatment of neuropsychiatric symptoms of dementia: A review of the evidence. JAMA, 293 (5), 596 608. Sival, R. C., Haffmans, P. M., Jansen, P. A., Duursma, S. A., & Eikelenboom, P. (2002). Sodium valproate in the treatment of aggressive behavior in patients with dementia: A randomized placebo controlled clinical trial. International Journal of Geriatric Psychiatry, 17 (6), 579 585. Sloane, P. D., Davidson, S., Knight, N., Tangen, C., & Mitchell, C. M. (1999). Severe disruptive vocalizers. Journal of the American Geriatrics Society, 47 (4), 439 445. Sloane, P. D., Miller, L. L., Mitchell, C. M., Rader, J., Swafford, K., & Hiatt, S. O. (2007). Provision of morning care to nursing home residents with dementia: Opportunity for improvement? American Journal of Alzheimer's Disease and Other Dementias, 22 (5), 369 377. Sloane, P. D., Mitchell, C. M., Preisser, J. S., Phillips, C., Commander, C., & Burker, E. (1998). Environmenta l correlates of resident agitation in Alzheimer's disease special care units. Journal of the American Geriatrics Society, 46 (7), 862 869. Smalbrugge, M., Jongenelis, L. K., Pot, A. M., Beekman, A. T., & Eefsting, J. A. (2007). Pain among nursing home patie nts in the Netherlands: Prevalence, course, clinical correlates, recognition and analgesic treatment. BMC Geriatrics, 7 3. Smith, M., Gerdner, L. A., Hall, G. R., & Buckwalter, K. C. (2004). History, development, and future of the progressively lowered st ress threshold: A conceptual model for dementia care. Journal of the American Geriatrics Society, 52 (10), 1755 1760. Snow, A. L., Chandler, J. F., Kunik, M. E., Davila, J. A., Balasubramanyam, V., Steele, A. B., et al. (2009). Self reported pain in persons with dementia predicts subsequent decreased psychosocial functioning. American Journal of Geriatric Psychiatry, 17 (10), 873 880. Snowden, M., McCormick, W., Russo, J., Srebnik, D., Comtois, K., Bowen, J., et al. (1999). Validity and responsiveness of the Minimum Data Set. Journal of the American Geriatrics Society, 47 (8), 1000 1004. Snowden, M., Sato, K., & Roy Byrne, P. (2003). Assessment and treatment of nursing home residents with depression or behavioral symptoms associated with dementia: A review of the literature. Journal of the American Geriatrics Society, 51 (9), 1305 1317. Song, J. A., & Algase, D. ( 2008). Premorbid characteristics and wandering behavior in persons with dementia. Archives of Psychiatric Nursing, 22 (6), 318 327. Song, J. A., Lim, Y. M., & Hong, G. R. (2008). Wandering behaviour of persons with dementia in Korea: Investigation of relate d factors. Aging & Mental Health, 12 (3), 366 373.

PAGE 208

208 Steinberg, M., Sheppard, J. M., Tschanz, J. T., Norton, M. C., Steffens, D. C., Breitner, J. C., et al. (2003). The incidence of mental and behavioral disturbances in dementia: The cache county study. The J ournal of Neuropsychiatry and Clinical Neurosciences, 15 (3), 340 345. Stevenson, K. M., Brown, R. L., Dahl, J. L., Ward, S. E., & Brown, M. S. (2006). The discomfort behavior scale: A measure of discomfort in the cognitively impaired based on the Minimum D ata Set 2.0. Research in Nursing and Health, 29 (6), 576 587. Stineman, M. G., & Maislin, G. (2000). Clinical, epidemiological, and policy implications of Minimum Data Set validity. Journal of the American Geriatrics Society, 48 (12), 1734 1736. Stranjalis, G., Kalamatianos, T., Stavrinou, L. C., Tsamandouraki, K., & Alamanos, Y. (2011). Neck pain in a sample of Greek urban population (15 65 years): Analysis according to personal and socioeconomic characteristics. Spine Suh, G. H. (2004). Agitated behaviours among the institutionalized elderly with dementia: Validation of the Korean version of the Cohen Mansfield Agitation Inventory. International Journal of Geriatric Psychiatry, 19 (4), 378 385. Suh, G. H., Yeon, B. K., Shah, A., & Lee, J. Y. (2005). Mortalit y in Alzheimer's disease: A comparative prospective Korean study in the community and nursing homes. International Journal of Geriatric Psychiatry, 20 (1), 26 34. Svansdottir, H. B., & Snaedal, J. (2006). Music therapy in moderate and severe dementia of Alz heimer's type: A case control study. International Psychogeriatrics, 18 (4), 613 621. Szor, J. K., & Bourguignon, C. (1999). Description of pressure ulcer pain at rest and at dressing change [Abstract]. Journal of Wound Ostomy & Continence Nursing, 26 (3), 1 15 120. Taft, L. B. (1989). Conceptual analysis of agitation in the confused elderly. Archives of Psychiatric Nursing, 3 (2), 102 107. Tait, R. C., & Chibnall, J. T. (2008). Under treatment of pain in dementia: Assessment is key. Journal of the American Med ical Directors Association, 9 (6), 372 374. Takanori, S. (2004). An overview of human interactive robots for psychological enrichment. Proceedings of the IEEE, 92 (11), 1749 1758. Talerico, K. A., Evans, L. K., & Strumpf, N. E. (2002). Mental health correlat es of aggression in nursing home residents with dementia. Gerontologist, 42 (2), 169 177. Teno, J. M., Kabumoto, G., Wetle, T., Roy, J., & Mor, V. (2004). Daily pain that was excruciating at some time in the previous week: Prevalence, characteristics, and outcomes in nursing home residents. Journal of the American Geriatrics Society, 52 (5), 762 767.

PAGE 209

209 Testad, I., Aasland, A. M., & Aarsland, D. (2007). Prevalence and correlates of disruptive behavior in patients in Norwegian nursing homes. International Journal of Geriatric Psychiatry, 22 (9), 916 921. Tjia, J., Rothman, M. R., Kiely, D. K., Shaffer, M. L., Holmes, H. M., Sachs, G. A., et al. (20 10). Daily medication use in nursing home residents with advanced dementia. Journal of the American Geriatrics Society, 58 (5), 880 888. Touhy, T. A. (2004). Dementia, personhood, and nursing: Learning from a nursing situation. Nursing Science Quarterly, 17 (1), 43 49. Tripathi, M., & Vibha, D. (2010). An approach to and the rationale for the pharmacological management of behavioral and psychological symptoms of dementia. Annals of Indian Academy of Neurology, 13 (Suppl. 2), 94 98. Vance, D. E., Burgio, L. D., Roth, D. L., Stevens, A. B., Fairchild, J. K., & Yurick, A. (2003). Predictors of agitation in nursing home residents. Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 58 (2), P129 137. Villanueva, M. R., Smith, T. L., Erickso n, J. S., Lee, A. C., & Singer, C. M. (2003). Pain Assessment for the Dementing Elderly (PADE): Reliability and validity of a new measure. Journal of the American Medical Directors Association, 4 (1), 1 8. Volicer, L. (2007). Epidemiology of wandering. In A L. Nelson & D. L. Algase (Eds.), Evidence based protocols for managing wandering behaviors (pp. 53 72). New York: Springer Publishing Company. Volicer, L., Bass, E. A., & Luther, S. L. (2007). Agitation and resistiveness to care are two separate behavior al syndromes of dementia. Journal of the American Medical Directors Association, 8 (8), 527 532. Volicer, L., Van der Steen, J. T., & Frijters, D. H. (2009). Modifiable factors related to abusive behaviors in nursing home residents with dementia. Journal of the American Medical Directors Association, 10 (9), 617 622. Voyer, P., Verreault, R., Azizah, G. M., Desrosiers, J., Champoux, N., & Bedard, A. (2005). Prevalence of physical and verbal aggressive behaviours and associated factors among older adults in lo ng term care facilities. BMC Geriatrics, 5 13. Wang, W. W., & Moyle, W. (2005). Physical restraint use on people with dementia: A review of the literature. Australian Journal of Advanced Nursing, 22 (4), 46 52. Wang, Y., Byers, K. L., & Velozo, C. A. (2008 ). Rasch analysis of Minimum Data Set mandated in skilled nursing facilities. Journal of Rehabilitation Research & Development, 45 (9), 1385 1400.

PAGE 210

210 Warden, V., Hurley, A. C., & Volicer, L. (2003). Development and psychometric evaluation of the pain assessmen t in advanced dementia (PAINAD). Journal of the American Medical Directors Association, 4 (1), 9 15. Weiner, D., Peterson, B., & Keefe, F. (1999). Chronic pain associated behaviors in the nursing home: Resident versus caregiver perceptions. Pain, 80 (3), 577 588. Weiner, M. F., & Teri, L. (2003). Psychological and behavioral management. In M. F. Weiner & A. M. Lipton (Eds.), The dementias: Diagnosis, treatment and research (3rd ed., pp. 181 218). Washington, DC: American Psychiatric Publishing. Wetzels, R. B., Zuidema, S. U., de Jonghe, J. F., Verhey, F. R., & Koopmans, R. T. (2010). Determinants of quality of life in nursing home residents with dementia [Abstract]. Dementia and Geriatric Cognitive Disorders, 29 (3), 189 197. Whall, A., Gillis, G., Yankou, D ., Booth, D., & Beel Bates, C. (1992). Disruptive behavior in elderly nursing home residents: A survey of nursing staff. Journal of Gerontological Nursing, 18 (10), 13 17. Williams, B. C., Li, Y., Fries, B. E., & Warren, R. L. (1997). Predicting patient sco res between the functional independence measure and the minimum data set: Development and performance of a FIM MDS "crosswalk". Archives of Physical Medicine and Rehabilitation, 78 (1), 48 54. Won, A., Lapane, K., Gambassi, G., Bernabei, R., Mor, V., & Lips itz, L. A. (1999). Correlates and management of nonmalignant pain in the nursing home. SAGE Study Group. Systematic Assessment of Geriatric drug use via Epidemiology. Journal of the American Geriatrics Society, 47 (8), 936 942. Won, A., Lapane, K. L., Vallo w, S., Schein, J., Morris, J. N., & Lipsitz, L. A. (2006). Long term effects of analgesics in a population of elderly nursing home residents with persistent nonmalignant pain. Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, 61 ( 2), 165 169. Wood, S., Cummings, J. L., Hsu, M. A., Barclay, T., Wheatley, M. V., Yarema, K. T., et al. (2000). The use of the neuropsychiatric inventory in nursing home residents: Characterization and measurement. American Journal of Geriatric Psychiatry, 8 (1), 75 83. Wu, H. Z., Low, L. F., Xiao, S., & Brodaty, H. (2009). A pilot study of differences in behavioral and psychological symptoms of dementia in nursing home residents in Sydney and Shanghai. International Psychogeriatrics, 21 (3), 476 484. Wu, N., Mor, V., & Roy, J. (2009). Resident, nursing home, and state factors affecting the reliability of Minimum Data Set quality measures. American Journal of Medical Quality, 24 (3), 229 240. Youngblut, J. M., & Casper, G. R. (1993). Single item indicato rs in nursing research. Research in Nursing & Health, 16 (6), 459 465.

PAGE 211

211 Zanocchi, M., Maero, B., Nicola, E., Martinelli, E., Luppino, A., Gonella, M., et al. (2008). Chronic pain in a sample of nursing home residents: prevalence, characteristics, influence o n quality of life (QoL). Archives of Gerontology and Geriatrics, 47 (1), 121 128. Zeisel, J., Silverstein, N. M., Hyde, J., Levkoff, S., Lawton, M. P., & Holmes, W. (2003). Environmental correlates to behavioral health outcomes in Alzheimer's special care u nits. Gerontologist, 43 (5), 697 711. Zhang, N. J., Paek, S. C., & Wan, T. T. H. (2009). Reliability estimates of clinical measures between Minimum Data Set and online survey certification and reporting data of US nursing homes. Medical Care, 47 (4), 492 495 Zieber, C. G., Hagen, B., Armstrong Esther, C., & Aho, M. (2005). Pain and agitation in long term care residents with dementia: Use of the Pittsburgh Agitation Scale [Abstract]. International Journal of Palliative Nursing, 11 (2), 71 78. Zimmerman, S., Sl oane, P. D., Williams, C. S., Dobbs, D., Ellajosyula, R., Braaten, A., et al. (2007). Residential care/assisted living staff may detect undiagnosed dementia using the Minimum Data Set cognition scale. Journal of the American Geriatrics Society, 55 (9), 1349 1355. Zuidema, S. U., Derksen, E., Verhey, F. R. J., & Koopmans, R. T. C. M. (2007). Prevalence of neuropsychiatric symptoms in a large sample of Dutch nursing home patients with dementia. International Journal of Geriatric Psychiatry, 22 (7), 632 638. Zwa khalen, S. M., Koopmans, R. T., Geels, P. J., Berger, M. P., & Hamers, J. P. (2009). The prevalence of pain in nursing home residents with dementia measured using an observational pain scale. European Journal of Pain, 13 (1), 89 93.

PAGE 212

212 BIOGRAPHICAL SKETC H Hyochol Ahn received his Bachelor of Engineering degree in Electr ical Engineering from the University of Seoul in South Korea and earned Master of Science degree in Electrical and Computer E ngineering from the University of Florida in 2004. While enroll ed in the Ph.D. program in Electrical and Computer Engineering, he developed a strong desire to learn more about human physiology, pathophysiology, disease management and caregiving. Consequently, he changed his major to nursing in order to combine his eng ineering knowledge with nursing science. He graduated from the University of Florida w ith his Bachelor of Science in N ursing in 2007, and practiced as a registered nurse at the Shands H ospital at the University of Florida. He went on to earn his Master of Science in N ursing in 2009 and his certification as an Adult Nurse Practitioner from American Nurses Credentialing Center in 2010 He received his Nurse Educator Certification from University of Florida College of Nursing in 2011. During his Ph.D. program in Nursing, he has taught several classes to undergraduate and graduate nursing students earned a research grant from Sigma Theta Tau Alpha Theta Chapter, participated in three NIH projects, and published several articles. His minor course of study for his doctorate is clinical and translational science.