1 ENHANCING BRADEN PRESSURE ULCER RISK ASSESSMENT IN ACUTELY ILL ADULT VETERANS By LINDA JOYCE COWAN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010
2 2010 Linda Joyce Cowan
3 To the military veterans who have faithfully served their country
4 ACKNOWLEDGMENTS I thank m y constant companion and Lord, Je sus Christ, for providi ng the strength and opportunity to complete this task while working fulltime. I could not have done it without Him. I gratefully acknowledge my di ssertation Chair, Joyce Stechmiller, for her immense support, steadfast leadership and patient mentoring. I will forever be influenced by her tremendous example of scholarly excellence a nd professionalism. I earnestly th ank my University of Florida committee members, Meredith Rowe, Rebecca Be yth, and Robert Cook for their patience, steadfast leadership, sage advi ce, and continuing support. I thank Courtney Lyder for his inspirational leadership in the field of pr essure ulcer prevention and treatment and for participating on my dissertation committee. I thankfully acknowledge my husband, Bill and youngest daughter, Debbie for their loving support, encouragement, and picking up the slack in household chores over the past 4 ye ars. I sincerely thank the rest of my immediate family: Paula and Pete Burt; Buzz, Brad and Terri Blazek; Ca rol Lawing; Katherine, Josh, April, Myra, Joanna, Shari, and Jamie Cowan; Donna and Doug Bullington; Sue Nettum, and Pat Norman for all of their prayers, encouragement, support, and forgiving my absences from many family functions and vacations while I have been in school. Seven of these family members are veterans, including my brother, Brad, who I than k for being a hero (and giving up so much for his country). I also appreciatively acknowledge my church family both at Grace Community Church and Faith Presbyterian Church for their steadfast prayers, words of encouragement and unwavering support while I complete d my studies. Lastly, I gratefu lly acknowledge my patients, co-workers and colleagues at th e North Florida/South Georgia Ve terans Health Administration, who are a constant source of encouragement a nd inspiration to me. I am truly blessed!
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ...........................................................................................................................7LIST OF ABBREVIATIONS ......................................................................................................... .8ABSTRACT ...................................................................................................................... .............101 INTRODUCTION .................................................................................................................. 12Statement of the Problem ...................................................................................................... ..12Background of the Problem ....................................................................................................13Identifying Risk ...............................................................................................................13Risk Assessment Tools ....................................................................................................14Purpose of the Study .......................................................................................................... .....17Significance of the Study ........................................................................................................19Clinical Importance of this Study .................................................................................... 20Limitations of the Study .................................................................................................. 212 REVIEW OF THE LITERATURE ........................................................................................22Etiology of Pressure Ulcers ................................................................................................... .22Tissue Tolerance ..............................................................................................................23External Forces of Pressure .............................................................................................24Current Definition and Stagi ng of Pressure Ulcers ................................................................25The Cost of Pressure Ulcers ...................................................................................................26Incidence and Prevalence of Pressure Ulcers .........................................................................27Guidelines and Mandates to Prevent Pressure Ulcers ............................................................ 29Pressure Ulcer Risk .................................................................................................................31Common Risk Factors .....................................................................................................32Age as a risk factor ...................................................................................................34Race or ethnicity as a pre ssure ulcer risk factor ....................................................... 34Smoking as a pressure ulcer risk factor .................................................................... 35Low BMI or serum albumin as a pressure ulcer risk factor .....................................36Inpatient length of stay as a pressure ulcer risk factor ............................................. 36Medical diagnoses as pressu re ulcer risk factors ..................................................... 37Pressure Ulcer Risk Assessment Tools ........................................................................... 38Current Understanding of Pressure Ulcer Risk Prediction ..............................................41Research Gaps .................................................................................................................413 RESEARCH METHODS .......................................................................................................44Design, Sampling and Setting ................................................................................................. 44Procedure ..................................................................................................................... ...........46
6 Data Collection ................................................................................................................47Data Analysis ...................................................................................................................51Assumptions ................................................................................................................... .55Additional Analyses ........................................................................................................... .....554 RESULTS ....................................................................................................................... ........57Characteristics of the Sample .................................................................................................57Comparison of Groups .......................................................................................................... ..59Pressure Ulcer Characteristics ................................................................................................63Regression Analysis ........................................................................................................... .....635 DISCUSSION, RECOMMENDATI ONS AND CONCLUSIONS .......................................74Recommendations for Practice ...............................................................................................77Recommendations for Future Research ..................................................................................79Limitations ................................................................................................................... ...........80Conclusions .............................................................................................................................82APPENDIX A BRADEN SCALE FOR PREDICTING PRESSURE SORE RISK ....................................... 83B PRESSURE ULCERS IN ACUTELY ILL VETERANS A PRELIMINARY M ODEL ....86C PRESSURE ULCERS IN ACUTELY I LL VETERANS A FINAL MODEL ................... 87D DIFFERENCES IN CAPTURING DIAGNOSES .................................................................88LIST OF REFERENCES ...............................................................................................................89BIOGRAPHICAL SKETCH .........................................................................................................97
7 LIST OF TABLES Table page 2-1 Examples of previous research id entifying pressure ulcer risk factors. .............................43 2-2 Abbreviated Braden scale subscales. ................................................................................. 43 4-1 Pressure ulcer st udy sample demographics. .......................................................................67 4-2 Total sample scale vari able descriptive statistics. ..............................................................68 4-3 Recorded medical diagnoses by chart review. ................................................................... 69 4-4 Comparison of mean differences between groups. ............................................................70 4-5 Differences between gro ups for predictor variables. ......................................................... 71 4-6 Pressure ulcer descriptions. ................................................................................................72 4-7 Logistic regression anal ysis results of models. .................................................................. 73 4-8 Relative risk of pressure ulcers by predictors in final model. ............................................ 73
8 LIST OF ABBREVIATIONS AHRQ Agency for Healthcare Research and Quality AMDA American Medical Directors Association ARNP Advanced Practice Registered Nurse CHF Congestive Heart Failure CMS Centers for Medicare and Medicaid Services CVA Cerebrovascular Accident CWS Certified Wound Specialist DM Diabetes Mellitus EPUAP European Advisory Panel FNP Family Nurse Practitioner GS Graduate student ICU Intensive Care Unit IHI Institute for Healthcare Improvement IRR Inter-rater Reliability JCAHO Joint Commission for Accredita tion of Healthcare Organizations LOS Length of Stay LPN Licensed Practical Nurse NA Nursing Assistant NGC National Guideline Clearinghouse NIH National Institutes of Health NPV Negative Predictive Value NPUAP National Pressure Ulcer Advisory Panel O.R. Operating Room OR Odds Ratio
9 PPV Positive Predictive Value PU Pressure Ulcer PURAS Pressure Ulcer Risk Assessment Scale PURS Pressure Ulcer Risk Screening RAS Risk Assessment Scale RN Registered Nurse RNAO Registered Nurses Association of Ontario SAWC Society for Advancement of Wound Care UK United Kingdom US United States UTI Urinary Tract Infection VA Veterans Administration WHS Wound Healing Society WOCN Wound, Ostomy, and Continence Nurses Society
10 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ENHANCING BRADEN PRESSURE ULCER RISK ASSESSMENT IN ACUTELY ILL ADULT VETERANS By Linda Joyce Cowan May 2010 Chair: Joyce Stechmiller Major: Nursing Sciences Preventing pressure ulcers is a significant health care challenge. Pressure ulcers increase inpatient length of stay, mortal ity, and complication rates. They reduce quality of life and pose a worldwide economic quandary. Many pressure ulcer risk factors have been identified over the past 20 years, yet current pressure ulcer risk assessment tools such as the Braden Scale for Predicting Pressure Sore Risk do not account for several of the most significant risk factors. The purpose of this retrospective study was to determine the pressure ulcer predictability of the Braden score alone, the Braden score + si gnificant medical factors, and significant medical factors alone. Medical factors inve stigated in this study include thos e reported in recent literature as significant pressure ulcer risk factors: diagnosis of gangrene, anemia, diabetes, malnutrition, osteomyelitis, pneumonia/pneumonitis, septicemia, candidiasis, bacterial skin infection, device/implant/graft complications, urinary tract infection, paralysis, senility, respiratory failure, acute renal failure, cerebrovascular accident, and congestive heart fa ilure; as well as age, race, hospital and intensive care unit length of stay days, surgery, operating room time in hours, smoking status, and a history of previous pressure ulcers. This study also examined differences between Braden scores and other associated risk factors in 213 acutely il l veterans with (n=100) and without (n=113) pressure ulcers in north Florida during January-July 2008.
11 A predictive model determined the Braden to tal score correctly classified 68% of total sample (65% veterans with PU, 70% veterans without PU). Adding fo ur significant medical factors correctly classified 78% total sample (74% with PU, 82% without PU). A predictive model with these four factors alone (presence of pneumonia/pneumonitis, candidiasis, severe nutritional compromise, and surgery) correctly classified 83% vetera ns with PU, and 72% veterans without PU (77% total sample correc tly classified). These findings suggest that identifying patients with severe nutritional co mpromise, the presence of pneumonia/pneumonitis, candidiasis, and surgery during hospi talization may be better able to identify veterans likely to develop pressure ulcers than current Braden risk assessment alone. More research is needed in this area to validate these findings in a larger sample, and provide dir ection for interventional studies, with an ultimate goal of redu cing incidence of pressure ulcers.
12 CHAPTER 1 INTRODUCTION Statement of the Problem Pressure ulcers are defined as, localized injury to the skin and/or underlying tissue usually over a bony prom inence, as a result of pr essure, or pressure in combination with shear and/or friction (NPUAP Press Release, 2007). Pr eventing pressure ulcers is one of the most significant challenges facing health care today (Armstrong, Ayello, Capitulo, Fowler, Krasner, Levine, Sibbald, & Smith, 2008). Pressure ulcers in crease inpatient length of stay, mortality, and complication rates (Maklebust, 2005; Reddy, Gill & Rochon, 2006) Pressure ulcers reduce quality of life and pose a considerable worldw ide economic quandary (Lyder, 2002; Maklebust, 2005; Spilsbury et al., 2007; Fogerty et al., 2008 ). The Joint Commission for Accreditation of Healthcare Organizations (JC AHO, 2008) reports approximatel y 2.5 million patients develop pressure ulcers each year in the United States (US), with a total estimated medical cost of up to $40,000 per ulcer, and approximately 60,000 deaths pe r year attributable to pressure ulcer complications. The United Kingdom (UK) estimates annual medical costs of pressure ulcers were 750 million pounds in 1998 (Banks, Graves, Bauer, & Ash, 2009). Scientific advances, superior medical technology, a nd innovative quality of care inte rventions have dramatically improved health care in the United States over the past 20 years (Vincent, Fink, Marini, Pinsky, Sibbald, et al., 2006). However, to date, pressure ulcer risk assessment and preventive measures have not significantly reduced overall national pressure ulcer prevalen ce or incidence during those same 20 years (Thomas, 2001; VanG ilder, MacFarlane & Meyer, 2008). Thomas (2001) poses an explanation for the unchanging incidence of pressure ulcers as a failure of known effective prevention treatment to be applied, or the failure of prevention strategies to be effective desp ite being applied (p. 298). Effective preventive measures may not
13 be applied if individuals are not appropriately identified as being at risk. Risk-screening tools are useless if they: 1) are not applicable to th e population being screened, 2) do not accurately account for significant risk factors, 3) are used inconsistently, or 4) are scored incorrectly (Thomas, 2001; Papanikolaou, Lyne, & Anthony, 2007). Furthermore, Thomas listed risk factors from epidemiological studies avai lable from 1989 to 1995, but admits that other significant risk factors may be unaccounted for in these studies (Thomas, 2001). Th is suggests that the data on which current preventive treatment measures are based may be out dated. Studies are needed to verify that modern day risk fact ors are accounted for, so that a ppropriate interventional studies may follow. Background of the Problem Pressure ulcers have also been known as bedsores for hundreds of years. The term decubitus ulcer has also been used historically to desc ribe pressure ulcers. Florence Nightingale in her fundamental book Notes on Nursing (1860 /reprinted 1976) br ought attention to prevention of pressure ulcers as primarily the responsibility of nursing when she said, If he (the patient) has a bedsore, it is generally the fault not of the disease, but of the nursing (p. 8). From the 1800s when Florence Nightingale charged nur ses with the responsib ility for preventing pressure ulcers, most of the medical community ha s been comfortable to leave this burden on the shoulders of nurses. More recently, however, the International Expert Wound Care Advisory Panel (Armstrong, Ayello, Capitulo, Fowler Krasner, Levine, Sibbald, & Smith, 2008) challenged not only nursing but also the entire hea lthcare industry with an immediate need to focus on the prevention of pressure ulcers. Identifying Risk The focus of pressure ulcer prevention has historically revolved around identifying risk factors and providing preventive interventions ai m ed at reducing those factors. An accurate
14 identification of what places a person at risk for developing a pressure ulcer is crucial in order to identify those at risk and initiate appropriate prevention interventions The National Pressure Ulcer Advisory Panel (NPUAP) lists Identifyi ng pressure ulcer risk factors and conducting valid/reliable risk assessments as one of the top 4 competencies for registered nurses in preventing pressure ulcers (NPUAP competency -based RN curriculum for preventing pressure ulcers, 2001). Lyder (2003) reports over 100 pressu re ulcer risk factors ha ve been identified in the literature. The Wound Hea ling Societys (WHS) Pressure Ulcer Prevention Guidelines (Stechmiller, Cowan, Whitney, Phillips, Aslam, et al., 2008) highlight the need for accurate identification of current risk factors in specific populations and a multidisciplinary approach to formulate prevention plans. Some common risk f actors listed in the WHS prevention guidelines as identified in previous studies include: imm obility, friction, shear, incontinence, moisture, age, altered level of consciousness, poor nutrition, poor perfusion, a nd certain skin conditions. The WHS guidelines stress the need for consistent use of pressure ulcer risk screening (PURS) tools (Stechmiller et al., 2008). Risk Assessment Tools Pressure ulcer risk screening tools such as the No rton Scale have been available since the 1960s (Norton, 1996). Pressure ulcer risk assessment tools presently utilized worldwide are the Norton Scale published in England in 1962, the Waterlow Scale published in England in 1984, and the Braden tool published in the US in 1987 (Papanik olaou, Lyne, & Anthony, 2007). The most widely used and tested of all risk assessment tools is the Braden Scale for Predicting Pressure Sore Risk developed by Barbara Braden and Nancy Bergstrom in the 1980s (Bergstrom, Braden, Laguzza, & Holman, 1987; Bryant, & Ni x, 2007). The Braden Scale is a well-tested instrument with six subscales used by health care providers to asse ss risk factors present that are associated with pressure ulcer development (Stotts & Gunningberg, 2007; Stotts & Wu, 2007).
15 The Braden Scale has been reported to have good inter-rater reliability with Cronbachs alpha between 0.83 to 0.99. Some studies report Braden specificity is betwee n 64 and 90% (with cut off risk scores of 18 or less) and sensitivity ranging from 83 to 100% (Ayello, 2007). Other studies suggest that nurses tend to over score rather th an underscore when using the Braden tool, underestimating pressure ulcer ri sk (Stotts & Gunningberg, 2007). At least one literature review of 31 studies regarding psychometric properties of the Braden Scale concluded that most of the interrater reliability measures (such as P earsons product-moment correlation and Cohens kappa) reported for the Braden tool were ina ppropriate measures (Kottner & Dassen, 2007). Laura Bolton (2007) extensively reviewed the cl inical evidence regarding the reliability and validity of the Braden, Norton, and Waterlow pr essure ulcer (PU) risk assessment tools vs. nurse clinical judgment across all settings (longterm care, re habilitation, acute care, ICUs, hospice, and pediatric care). She reports a meta-a nalysis of twenty studies (most of which were conducted in the 1990s or early 2000s) examining the Braden Pressure Ulcer Risk Assessment Scale (PURAS) across all settings results with an average sensitivity of 57% and specificity of 68% with a positive predictive value (PPV) of the at risk score of 23% and the negative predictive value (NPV) of the not at risk score of 91%. Five studies (conducted in the 1970s and 1980s) examining the Norton PURAS across a ll patient settings (home, hospital, and rehabilitation facilities) revealed an average se nsitivity of 47% and specificity of 62% with a PPV of the at risk score of 18% and the NPV of the not at risk score of 87%. Six studies examining the Waterlow PURAS across all settings (conducted in the 1980s and 1990s) reveal an average sensitivity of 82% and specificity of 27% with a PPV of the at risk score of 16% and the NPV of the not at risk score of 89%. Bolton (2007) compared this cumulative data to three studies that examined nur se clinical judgment across a ll settings, reporting an average
16 sensitivity of nurse clinical judgment of 51% and specificity of 60% with a PPV (the nurse accurately judging a patient to be at risk) of 33% and the NP V (the nurse accurately judging the patient to be not at risk) of 76%. Bolton (2007) concluded that the Braden tool is superior to the Norton and Waterlow scales in specificity and sensitivity, and is valid for predicting PU risk in multiple health care settings and count ries. However, Bolton al so concluded that, No evidence supports assessing pres sure ulcer risk on individual s in good clinical condition, as evidenced by low pre-operative and postoperative day 5 validity of the Braden Scale. (Bolton, 2007, p. 378). In addition, Pancorbo-Hildago, Garcia-Fernand ez, Lopez-Medina, & Alvarez-Nieto (2006) conducted a systematic review of 33 studies rega rding pressure ulcer risk assessment scales available for use today such as the Braden or No rton scales and found that the use of these scales has not changed the incidence of pr essure ulcers, but likewise concl uded that they ar e still better risk prediction tools than nurses clinical judgment (p. 108). Un fortunately, further limitations have been identified with the Br aden, Norton, and Waterlow pressure ulcer risk assessment tools. Literature suggests critical cut-off scores for eac h of the risk assessment scales (indicating at what point an individual is determined to be at risk) are disputable, and inconsistently applied (interpreted differently from se tting to setting). In addition, the risk factors which comprise the subscale categories of each tool (such as sensory, activity, moisture, nutrition, mobility, and friction in the Braden Scale) ha ve equal weights attributed to them, which may be statistically limiting. For example, certain Braden subscale fact ors (such as mobility) may be more important than other subscale factors for pr edicting risk (Berlowitz et al ., 2001), and one subscale only has a possible score of 1 to 3 (frict ion/shear) while the other subscales have a possible score of 1 to 4. Furthermore, it has been suggested that certai n Braden subscale definitions (such as patients
17 dietary intake or frequency of skin being mois t) are more difficult for nurses to determine or appropriately quantify than other Braden subscal e factors such as activity level (Papanikolaou, Lyne, & Anthony, 2007). Schoonhoven et al. (2006) cr iticized current pressure ulcer risk assessment tools by stating, Neither risk factors nor the weights attributed to them have been identified using adequate statis tical techniques (p. 65). This necessitates a re-evaluation of pressure ulcer risk screening tools for rele vancy and effectiveness to current populations. Papanikolaou, Lyne, & Anthony (2007) recommends, d ifferential weighted scoring techniques, advanced statistical methods, and large data sets be used to develop data driven and more robust risk assessment scales. (p. 285). Kottner & Dassen (2007) recommend that Braden tool interrater reliability be calculated and reporte d using intraclass correlation coefficients in combination with overall percen tage of agreement, instead of current reporting methods. In addition, patient acuity, medical technology, nursing hours at the be dside (Hall, Doran, & Pink, 2004; Kramer & Schmalenberg, 2005), nursing pr actice environments (Lake & Friese, 2006), and pressure ulcer risk factors identified in scientific rese arch (Fogerty et al., 2008) have changed in the past twenty years. Therefore, a modern statistical analysis is needed in a predictive model, demonstrating possible interact ions among currently iden tified risk factors and determining predictive contributi ons of each risk factor (inclu ding those comprising the Braden subscales) so that interventions may be directed at those risk factors that pose the strongest association with the development of pressure ul cers, particularly those that are modifiable. Purpose of the Study The purpose of this retrospective descriptive study was to determ ine the pressure ulcer predictability of the Braden score alone, the Br aden score + significant medical factors, and significant medical factors alone. Medical factors investigated in this study include those reported in recent literature as significant pressure ulcer risk factors: diagnosis of gangrene,
18 anemia, diabetes, malnutrition, osteomyelitis, pneumonia/pneumonitis, septicemia, candidiasis, bacterial skin infection, device /implant/graft complications, urin ary tract infection, paralysis, senility, respiratory failure, acute renal failure cerebrovascular accident, and congestive heart failure; as well as age, race, hospital and in tensive care unit length of stay days, surgery, operating room time in hours, smoking status, and a history of previous pressure ulcers. This study also examined differences between Braden sc ores and other associated risk factors in 213 acutely ill veterans with (n=100) and without (n =113) pressure ulcers in north Florida during January-July 2008. The aims of this research included: 1) to de termine the predictability of the Braden Scale total score on the development of pressure ulcers in an inpa tient acutely ill adult veteran population; 2) to determine if th e addition of other significant medical factors (diagnosis of gangrene, anemia, diabetes mellitus malnutrition, osteomyelitis, pneumonia/pneumonitis, septicemia, candidiasis, bacterial skin infection, complication of device or implant/graft, urinary tract infection, paralysis/CVA, senility, respirator y failure, acute renal failure, congestive heart failure, history of previous pressure ulcer, age, race, length of inpatient hospital and ICU stays, surgery, time in operating room, and smoking status ) to these Braden total scores enhance the models predictability of pressure ulcer develo pment in an inpatient acutely ill adult veteran population; and 3) to determine if selected me dical factors alone are significantly able to determine the development of pressure ulcers in an acute inpatient adult vetera n population. This retrospective descriptive study in an acu tely ill adult inpatient veteran population in north Florida from January-June 2008 determined the difference between Braden scores and other associated risk factors in veterans w ith and without pressure ulcers. Because this population is predominantly male, gender was not ex amined as an independent variable. Logistic
19 regression statistical analysis was utilized to determine how predictive Braden total scores were in a pressure ulcer predictive model (as well as examine each of the Braden sub-scores) with and without the inclusion of other medical factors. The results of this study enhance the current knowledge of pressure ulcer risk factors and the assessment tools used to screen for them and provide direction for future studies, with an ultimate goal of reducing incidence of pressure ulcers. Significance of the Study The m ost common pressure ulcer risk-screening tool in use today is the Braden Scale for Predicting Pressure Sore Risk (see Appendix A). Th is is the main pressure ulcer risk-screening tool utilized in Vetera ns Administration Health Care facilities nationwide. The Braden Scale was published in 1987 (Bergstrom, Braden, La guzza, & Holman, 1987; Bryant, & Nix, 2007). However, it has not changed significantly si nce first publication a nd does not account for important medical factors (such as age, race, length of stay, specific medical diagnoses, or smoking status) described in more recent literature as strongly a ssociated with the development of pressure ulcers (Ayello, 2007; VanGilder, Ma cFarlane, & Meyer, 2008 ; Fogarty et al., 2008). A systematic review of thirty-three studies i nvolving pressure ulcer ri sk assessment scales (Pancorbo-Hildago, Garcia-Fernandez, Lop ez-Medina, & Alvarez-Nieto, 2006) and a methodological review of risk assessment scales for pressure ulcers (Papanikolaou, Lyne, & Anthony, 2007) suggest there is great variability within the positive predictive value (PPV) and negative predictive value (NPV) of currently used risk assessment scales (RAS) between settings and health care providers. Furthermore, they point out the deficit of scient ific work investigating the subscale categories (risk factors which comp rise the scoring components of the scales) and appropriateness of cut-off (at risk) values of the scales, which vary among settings. Bergstrom,
20 Braden, Kemp, Champagne, & Ruby (1998) report Br aden scores of 18 or below should be used as the at risk cut-off score ( www.bradenscale .com ). In addition, there is limited data published in the literature descri bing current pressure ulcer risk factors among inpatient veterans in acute care settings. Further re search is needed to examine this population as well as investigate the signifi cance of other risk factors for pressure ulcer development described in scientific lite rature that are not acc ounted for by the common pressure ulcer risk assessment tools utilized t oday, such as specific hi gh risk medical diagnoses described by Fogerty et al. (2008). Clinical Importance of this Study An international expert panel published a consensus paper in 2008, which highlighted recen t changes in the U.S. Centers for Medica re and Medicaid Services (CMS) financial reimbursement amounts based on admission diagnosis codes for acute care and long-term care facilities (beginning in October 2008) that will no longer reimburse higher rates for patients that develop stage III or IV pressure ulcers after admission (Armstrong et al., 2008). This is thought to provide additional motivation to acute and lo ng-term care facilities to evaluate and improve their pressure ulcer prevention pr ograms. This discussion is signifi cant, as it stresses the urgency of a consensus among health care providers and particularly the wound care community in providing quality research as we ll as relevant, research-based toolkits, and accurate protocols that address: risk assessment and documentation, patient education, c linician training, and evidence-based effective intervention measures. Limitations of the plan mentioned in the CMS consensus paper include the lack of randomized control trials to know which interventions are most effective, and a lack of risk assessment t ools that are not only valid and reliable, but also up-to-date, accurate, easy to use, do not require intense training, and are applicable to current populations. In addition, Armstrong et al. (2008) point out that, Competence of the provider in
21 assessment is critical to do an accurate skin ( and risk ) assessment (p. 470). Skin and risk assessment is pivotal to pressure ulcer preven tion. From the days of Florence Nightingale until the present, nurses have been the primary provi der of accurate skin/risk assessments. However, under the new CMS ruling, this responsibility is going to also fall to the admitting physician/provider. Therefore, it is imperative that plans for eff ective pressure ulcer prevention incorporate a multidisciplinary approach involving a ll levels of care from the nursing assistant to the physician (Ho & Bogie, 2007; Howe, 2008; McInerney, 2008). Thomas (2001) suggests there may be few instances where pressure ulcers are unavoidable. However, most pressure ulcers ar e considered to be avoidable, therefore, preventable (Jalali & Rezaie, 2005; Bryant & Nix, 2007). The Cent ers for Medicare & Medicaid Services (CMS) reported 257,412 cases of preventable pressure ul cers (listed as secondary diagnosis) during the fiscal year 2007 (Armstrong et al., 2008). It is essential for the entire medical community to address this health i ssue in ways that will reduce these numbers. Limitations of the Study This study is lim ited to adult veterans ages 47 to over 85 within an acute hospitalization setting in the Southeastern Unite d States. The generalizability of this study is limited by the fact that this is a veteran population, is mostly male (97%), over half of th e subjects were over the age of 72, and the location is limited to veterans residing in the North Florida/South Georgia region.
22 CHAPTER 2 REVIEW OF THE LITERATURE Etiology of Pressure Ulcers Historically, pressure ulcers ha ve been described in the m edical literature since at least the 1500s when Fabricius Hildanus first documente d his understanding of the causes and clinical characteristics of bedsores. He highlighted the role of internal supernatural and external natural factors that interrupt th e supply of blood and nutrients to tissue as causes of bedsores. Mechanical pressure and incontinence were first identified as key factors in the development of pressure ulcers by French surgeon de la Motte in 1722 (Defloor, 1999). Research regarding the role of tissue ischemia in the formation of pressure ulcers and factors affecting the interruption of blood supply to human tissues have been published in the US since at least 1930 when Landis recorded capillary closing pressures (the level of external pressure required to occlude capillary blood and lymph circulation to /from tissues) of the average finger to be 32 mm Hg and it was suggested that any external pressu re higher than this could lead to tissue ischemia or necrosis (Kosiak, 1961; Defloor, 1999; Lyder, 2006). Defloor (1999) describes research done by Husain ( 1953), Kosiak (1959 & 1961), Lindan & Greenway (1965) Jonker (1978), Braden & Bryant (1990) Bennet & Lee (1985), Si deranko et al. (1992) and Sparks (1993) which built on the early work done by Landis (1930) to ultimately demonstrate that forces sufficient to exceed capill ary closing pressures is different according to an individuals diastolic blood pressure, weight, body build, age, body position or posture (sitting, supine, or side-lying), nut ritional status, and ti ssue perfusion. Furthermore, the internal effects of medications such as corticosteriods (decreased coll agen production & angiogenesis), psychological or physiological stre ss (increased cortisol levels and protein energy demands), diabetes (decreased sympathetic nervous sy stem function, capillary basement membrane
23 thickening, increased blood viscosity, impaired microcirculation), dehydration (decreased skin elasticity), and body temperature (i ncreased metabolic rate & ti ssue oxygen demands with fever) have also been described as influencing an individuals risk for pressure ulcer development by affecting tissue tolerance (Def loor, 1999; Lyder, 2006). Recently, work by Fogerty et al. (2008) has shown that in addition to advanced age, certain medical diagnoses, specifically related to infection and multisystem failure have been predictive of pressure ulcer formation in large populations. Tissue Tolerance The concep t of tissue tolerance attempts to explain external factors (such as moisture and the presence of friction and/or shearing forces ) and internal factors (such as age, nutrition, interstitial fluid levels, co llagen and elastin levels, low arteriolar pressure or poor perfusion/oxygenation) which results in unique physical characteristi cs of an individuals skin and underlying tissues that make them more or less susceptible to tissu e damage by sustained forces of pressure (Braden & Bergstrom, 1987). Th is idea of tissue tolera nce maintains that if tissue tolerance is low, shorter du ration of pressure will result in damage to tissues and likewise, if tissue tolerance is high, the individual is less susceptible and tissue damage would only occur with longer duration of sustained pressure applied (Braden & Bergstrom, 1987; Defloor, 1999). The main ideas supporting the te rm tissue tolerance was devel oped in the 1970s and 1980s based upon studies in animal models such as Dinsdale (1974) that demonstrated friction combined with pressure produced tissue damage faster than either friction or pr essure when applied independently. Tissue tolerance was listed as a major independent variable in Braden & Bergstoms 1987 Conceptual schema for the study of the etiology of pressure sores (Braden & Bergstrom, 1987; Bergstrom, Braden, Laguzza, & Holman, 1987). Defloor (1999) quoted a 1994 article by Meijer et al., in which they stated, individual susceptibility to pressure and shear
24 forces are as important as the actual external fo rces of pressure and shearing in the development of pressure ulcers. However, De floor was critical of any idea that tissue tolerance should be included in a causal model as an independent vari able rather than an intermediate variable stating, Tissue tolerance cannot cause pressure sores the existence of pr essure and/or shearing force is needed (Defloor, 1999, p. 207). External Forces of Pressure Defloor (1999) introduced a conceptual sche m e for pressure ulcer development that differentiated compressive force vs. shearing force as two independent in teractive variables and tissue tolerance for pressure vs. tissue tolerance for oxygen as two intermediate variables with pressure sores as the dependent variable. In this mode l, tissue mass, age, dehydration, protein and vitamin C deficiency, and st ress are factors contributing to tissue tolerance for pressure ; temperature, medication, protei n deficiency, smoking, blood pressu re, and presence of certain diseases (that affect oxygen supply, reactive hypere mia, and vascular occlusion) are factors contributing to tissue tolerance for oxygen Studies suggest that majo r alterations in the normal functioning of these human mechanisms or the cumu lative effects of minor changes in several of these mechanisms/factors combined with sustained external pressure (forces perpendicular to the skin) and/or the presence of exte rnal moisture (fecal or urinary in continence, excess perspiration) and/or friction and shear forces (slipping, sliding or rubbing forces parallel to the skin) will result in the development of pressure ulce rs (Defloor, 1999; Baranoski, 2006). Berlowitze & Brienza (2007) s uggest that deep tissue injury may be responsible for pressure ulcers as a result of stress (pressure) a nd strain (deformation) in soft tissue dependent on 4 contributing factors: ischemia caused by ca pillary occlusion, reperfusion injury, impaired lymphatic function, and prolonged mechanical deform ation of tissue cells a nd that superficial
25 (skin) injuries are not caused by pressure. Interestingly, they theorize skin injuries caused by friction/shear and moisture are mainly superf icial lesions and shoul d not be considered pressure ulcers (Berlowitze & Brienza, 2007, p.37) Baranoski (2006) desc ribes the differences between the two main theories of pressure ulcer etiology bein g that one is a top-to-bottom model: injury begins from skin destruction on the outside of the body at the surface level and progresses inward toward deeper tissues (such as Defloor theorized); and th e other theory is an inward to outward model: suggest ing that tissue damage begins inward (deep tissue) and moves outward (such as Berlowitz & Brienza suggested). Baranoski (2006) suggests that there is more current scientific data (such as ultrasound examination of deep ti ssue) to support the inward to outward model. If this deep tissue injury (Gefen, 2008) view is adopt ed among clinicians, it may affect pressure ulcer staging, pr essure ulcer risk assessment tools (where moisture/incontinence are presently considered significant risk factor s) and ultimately pre ssure ulcer prevention measures. Pressure ulcer research has identified severa l key physiological factor s responsible for the development of pressure ulcers in the human bo dy (tissue tolerance vs. external forces of pressure or friction/shear). However, there is a lack of research data that has determined the exact amount of time in which a pressure ulcer will develop in all persons. Gefen (2008) conducted a review of research findings from animal, human a nd in vitro studies and reported findings that indicate pressure ulcers may develop in high-risk i ndividuals in less than one hour of sustained pressure to vuln erable body tissue (Gefen, 2008). Current Definition and Staging of Pressure Ulcers The National Pressure Ulcer Advisory Pane l (NPUAP) defines a pressure ulcer as, localized injury to the skin and/or underlying tissue usually over a bony prom inence, as a result
26 of pressure, or pressure in combination with shear and/or friction (NPUAP Press Release, 2007). The degree of tissue damage is communicated through the use of a staging system from Stage I (less obvious tissue damage) to St age IV (damage may extend to bone), with Unstageable (depth of tissue damage is unde terminable) and Deep Tissue Injury (DTI) discoloration (depth of tissue damage likely deep er than physically obvious) recently added as descriptors (Black, Baharestani, Cuddigan, Dorner, Edsberg, & Langema, et al., 2007). The most common sites for pressure ulcer development ar e the sacrum, buttocks and heels ((Perneger, Heliot, Rae, Borst, & Gaspoz, 1998; Lyder, 2003; Vangilder, MacFarlane, & Meyer, 2008). The Cost of Pressure Ulcers Liter ature describes higher mortality rates in individuals with pressure ulcers (Redelings et al., 2005). Length of hospital stay has been report ed to be five times higher in patients with pressure ulcers versus those without pressure ulcers with similar admitting diagnoses (Graves, Birrell & Whitby, 2005). Pressure ulcers also increase medical complication rates such as sepsis, which occurs in as much as 30% of patients w ith pressure ulcers (Johns Hopkins: Lyder, 2000). In addition, sepsis has been re ported in almost 40% of all pressure ulcer related deaths (Whittington & Briones, 2004). As reported before, JCAHO attribut es approximately 60,000 deaths per year to pressure ulcer complica tions (JCAHO, 2008). Whittington & Briones (2004) estimate annual medical costs in the United States (US) associated with treating pressure ulcers exceed $5 billion dollars annually. Fogerty et al. (2008) estimates this cost to be higher at $10,845 per patient, exceeding a tota l $18.5 billion dollars annually. This cost does not include the cost of legal litigation in pressure ulcer cases, which is increasing. Reddy, Gill & Rochon (2006) report legal settlements fa vor long-term care residents in as much as 87% of cases against long term care facilities for failure to prevent pressure ulce rs (p.974). Furthermore, Jalali &
27 Rezaie (2005) suggest it may co st as little as $500 to preven t a pressure ulcer, indicating prevention is more cost-effective than treatment (Jalali & Rezaie, 2005, p.92). Incidence and Prevalence of Pressure Ulcers The incidence of pressure ulcers deals with the num ber of new pressure ulcers which develop in a particular populat ion over a specified period of time, while the prevalence of pressure ulcers deals with a snapshot picture of how many cases of pressu re ulcers exists in a population of interest at one partic ular reference point/date. It is a national concern that despite national pressure ulcer preventi on guidelines and directives, th e incidence and prevalence of pressure ulcers in the United St ates (US) has not significantly ch anged over the past 20 years. Whittington & Briones (2004) reported a pre ssure ulcer prevalence of 17% in 2,200 nationwide non-veteran acute care fa cilities in 1999. Using the same acute care facilities sample, they reported a prevalence rate of 14% in 2001 and a 16% prevalence rate in 2004. In addition, Whittington and Briones (2004) report ed a pressure ulcer incidence rate of 8% among the acute care facilities in 1999 and incide nce rates of 7% for both 2001 and 2004. They concluded that incidence rates of new pressure ulcers in acute care facilities varied widely among institutions but averaged around 7% over the 6 years of th e study from 1998 to 2004. In another article, Results of Nine Internationa l Pressure Ulcer Prevalence Surveys: 1989 to 2005 VanGilder, MacFarlane, & Meyer (2008) repor ted 148 acute care and long-ter m care facilities were first surveyed in 1989 and reported 9.2% overall preval ence of pressure ulcers in a sample of 34,987 patients. In 1995, 265 facilities were surveyed an d reported an overall pressure ulcer prevalence rate of 10.1% in a sample of 39,874 acute a nd long-term care patients. Between 1999 and 2005, pressure ulcer prevalence remained at 14.8% to 15.2% of the total sample reported (n=85,838) in 651 total acute and long-term care facilities. The overall pressure ulcer prevalence rate reported in 394 acute care facilities in 2003 was 15.4%, with a nosocomial prevalence rate of 6.9% in the
28 same facilities. The overall pressure ulcer prevalen ce rate reported in 533 acu te care facilities in 2005 was 14.6%, with nosocomial pressure ulce r prevalence of 7.3% in the same facilities (VanGilder, MacFarlane & Meyer (2008). The National Pressure Ulcer Advisory Panel (NPUAP) reported in 2001 that in cidence rates were as high as 17% for home care, up to 38% for acute care, and up to 23.9% for long-term care facilities. The NPUAP also reported 2001 prevalence rates as high as 29% for home care, up to 18% for acute care, and up to 28% for long term care facilities (NPUAP, 2001). These data i ndicate that US pressure ulcer prevalence rates have not improved at all since the 1980s (VanGilder, MacFarlane, & Meyer, 2008). Thomas (2001), Maklebust (2005), and Meau me & Faucher (2007) re-iterate that unchanging incidence/prevalence rates may be due to ineffective prevention strategies, problems with prevention measure implementation, unidentified risk, or lack of cons istent pressure ulcer prevention staff education measures. Furthermore, published data is limited on incidence and prevalence of pressure ulcers a nd risk factors among veteran popul ations except for veterans with spinal cord injuries or disorders (Garber & Rintala, 2003; Smith, Guihan, LaVela, & Garber, 2008) and veterans in long-term care facilities (Berlowitz, Bra ndeis, Anderson & Brand, 1997; Berlowitz et al., 1999; Brow n, 2003; Hickey et al., 2005). Garber & Rintala (2003) report findings of Yarkony and Heinemann (1995) that 32% of veterans with a spinal cord injury (SCI) developed at least one pressure ulcer within 20 years post-SCI a nd Carlson et al. (1992) reported 29% of veterans with SCI developed pressu re ulcers during acute care admissions. Smith, Guihan, LaVela, & Garber (2008) reported 36 % of 2,574 SCI respondents to a survey selfreported pressure ulcers during the previous ye ar (2002-2003). However, more research is needed among the acute care inpa tient (and outpatient) veteran popul ations. Acute care inpatient veterans may be at higher risk for pressure ul cers than their non-vetera n neighbors due to their
29 higher average age (38.9% of veterans > 65 year s old vs. 14% of non-vete rans > 65 years old) and prevalence of disability within veterans (26.8% of veterans have disability vs. 16.1% of nonveteran population) per US Census Bureau 2006 American Community Survey Data Set. Guidelines and Mandates to Prevent Pressure Ulcers The Veterans Health Administration ( VHA) Handbook policy 118.2 titled, Assessment and Prevention of Pressure Ulcers (2006) provides mandated comprehensive guidance for interdisciplinary approaches to assessment, reassessment, prevention and documentation of pressure ulcers relevant to all areas of VHA clinical practice (inc lusive of Acute Inpatient, Long Term Care Patients, and Outpatient populations ). National policies on pressure ulcer risk, including VHA policies, include a directive to pe rform accurate skin assessments and pressure ulcer risk assessment using the Braden or other widely accepted tool (NPUAP, 2001; VHA Handbook, 2006). The Agency for Health Care Po licy and Research (AHCPR) was established in 1989 as a result of United States (US) legi slation (Public Law 101-239 of the Omnibus Budget Reconciliation Act of 1989). The AHCPR was charge d to enhance the qual ity, appropriateness, and effectiveness of health care services and access to these se rvices within the US and its territories (AHCPR, 1992, publication No. 92-0047). One of the first Clinical Practice Guidelines published in 1992 by the AHCPR was Clinic al Practice Guide line #3 entitled, Pressure Ulcers in Adults: Prediction and Prevention (AHCPR publication No. 92 -0047). The Guideline was a result of a systematic review of 800 scientific manuscripts available from a literature search of that time. A panel of experts further analyzed a nd refined the guideline, as did peer-review and organizational reviewers. The focus of the 1992 gui deline is summarized in its 4 overall goals: (1) identifying at-risk individua ls who need prevention and the specific factors placing them at risk, (2) maintaining and improving tissue tolerance to pressure ulcers in order to prevent injury, (3) protecting against the adverse effects of external mechanical forces (pressure, friction, and
30 shear), and (4) reducing the incidence of pre ssure ulcers through edu cational programs. The guideline went on to describe specific nursing interventions aimed at accomplishing these goals (National Guideline Clearinghouse Archives, retr ieved 8/8/08). The AHCPR name was changed to Agency for Healthcare Research and Quality (AHRQ) but the intention of the guideline remained the same, to be evidence-based and revised and updated as needed to reflect new research findings. Unfortunately, this guide line was recently removed from the National Guideline Clearinghouse (NGC) web site because it no longer met the NGC Inclusion Criteria (for current date or content) and has yet to be replaced with a revised AHRQ guideline. However, more recent evidence-bas ed clinical practice guidelines have been posted on the NGC web site from other sources. These include: The Wound, Ostomy and Continence Nurses Society (WOCN) Guideline for the Prevention and Management of Pressure Ulcers published in 2003 and updated in 2004; The Registered Nurses Association of Ontario (RNAO) Guideline entitled Risk Assessment and Prevention of Pressure Ulcers, published in 2002 and revised in 2005; A Guideline entitled, Preventing Pressure Ulcers and Skin Tears by Ayello & Sibbald (2008) published in: Evidence-based Geriatric Nursing Pr otocols for Best Practice, 3rd Edition which was an update of a Guideline first published on the NGC website in 2003; and the American Medical Directors Association (AMDA) Guideline entitled, Pressure Ulcers in the Long-term Care Setting first published in 1996 and revised in 2008. Another evidence-based Guideline for the prevention of pressure ul cers is The Wound Hea ling Societys (WHS) Guidelines for the Prevention of Pressure Ulcers published by Stechmiller et al. in 2008. Most recently, the NPUAP (National Pressure Ulcer Advi sory Panel) in collaboration with the EPUAP (European Pressure Ulcer Advisory Panel) deve loped an international guideline for pressure ulcer prevention on th e NPUAP website at: www.npuap.org (2009). W hat all of these guidelines
31 have in common is the directive for health care pr oviders to appropriately identify individuals at risk of pressure ulcers, to use an accurate risk assessment tool (several guidelines specifically mention the Braden Scale), and to present a plan (intervention) to prevent pressure ulcers. In addition, the Institute for Healthcare Improvement (IHI) developed a national initiative called, 5 Million Lives Campaign It listed the prevention of pressu re ulcers by reliably using sciencebased guidelines for their prevention as one of the 12 necessary national interventions for healthcare to focus on beginning in 2006 (Dun can, 2007). Duncan (2007) proposed six key elements to pressure ulcer prevention which fall into either of two steps: ( 1) identify patients at risk, and (2) reliably implement prevention strate gies for all patients who are identified as being at risk. The predominant message is clear: an accurate identification of those individuals who are at risk of pressure ulcers is critical. Pressure Ulcer Risk The f irst step to prevent the development of pressure ulcers is to identify who is most likely to develop them (who are at greatest risk) with the ultimate goal to implement effective prevention measures. Risk factors are those factors or conditions that are noted to be most strongly associated with the outco me of interest. In order to pr ovide evidence-based preventive measures to prevent the development of pressure ulcers, an effective means of identifying those at highest risk is imperative. Current risk a ssessment tools may require further development, improved statistical evaluation, a nd possibly modification in orde r to remain applicable to present day populations (Defl oor & Grypdonck, 2005; Armstrong et al, 2008; Fogerty et al., 2008). In addition, studies such as Anthony, Re ynolds, & Russell (2000) demonstrating serum albumin enhanced the pressure ulcer predictabi lity of the Waterlow Risk Assessment Score, suggest that considering risk factors not acc ounted for on current risk assessment tools will enhance their pressure ulcer predictability.
32 Common Risk Factors W ithin the past 20 years, major risk factor s identified (see Table 2-1 for examples) for pressure ulcer development include increased age, impaired mobility, decreased physical activity, poor nutrition, urinary and/or fecal inco ntinence, sensory impairment, friction, shear, moisture, low BMI, altered level of consciousne ss, poor perfusion and certain skin conditions (Allman, 1997; Ayello & Lyder, 2001; Lyder, 2006; Reddy, Gill, & Rochon, 2006; Stechmiller et al., 2008). Several studies have identified additional risk factors that include smoking status, diabetes mellitus (DM), coronary artery disease (CAD), renal failure, intensive care unit (ICU) stay greater than 3 days, ventilator depende ncy, pneumonia/pneumonitis, fever/sepsis, obesity, female gender, and peripheral vascular disease (PVD) (Berlowitz et al ., 2001; Lyder, 2006; de Souza, & Santos, 2007). Several studies suggest ethni city or race may be a significant risk factor, with people of darker skin tones having up to 5 times higher risk than their lighter skinned neighbors (Fogerty et al. (2008). Unfortunately, conflicting data from smaller studies do not show significant differences between age or race or nutritional valu es among those with pressure ulcers vs. those without pressure ulcers (Smith, Guihan, LaVe la, & Garber, 2008). Significant limitations of most pressure ulcer predictive studies are: small sample sizes, convenience sampling, potential bias due to under-reporting of pr essure ulcers or inappropriately considering skin tears or other non-pressure related skin c onditions as pressure ulcers, confounding variables not examined, and lack of scientific rigor such as is seen in randomized controlled trials (VanGilder, MacFarlane, & Meyer, 2008). Fogerty et al. (2008) conducted a very larg e case-control study re viewing admission and discharge data from over six million subjects (N ationwide Inpatient Sample) to identify risk factors and demographic differences between those who developed pressure ulcers and those that did not. Some may describe their study as a ne sted case-control (Gordi s, 2004) because they
33 identified a cohort (inpatients in the NIS dataset), followed them from their hospital admission until hospital discharge (during 2003), and separa ted them into 2 groups: those who developed pressure ulcers (cases) and t hose that did not (controls). There were 94,758 incident pressure ulcers documented among a final discharge sample of 6,610,787 persons. Utilizing multivariate logistic regression analysis on 45 common diagnoses identified in persons with pressure ulcers, they reported odds ratios (estimat e of relative risk) for the most si gnificant risk factors associated with developing pressure ulcers. Analysis was also conducted stra tifying the sample by age, race and gender. Age over 75 years was th e strongest pressure ulcer risk factor identified with an Odds Ratio (OR) of 12.63. Other strong risk f actors identified by Fogert y et al. (2008) included over 28 medical diagnoses with an Odds Ratio over 2.0, indicating two or more times the average risk for pressure ulcer s among individuals w ith these diagnoses. Age 59 to 75 years was a strong risk factor (OR 5.99, no Confidence Interv al reported), and Afri can American race (OR 5.71, 95% CI 5.35-6.10). Fogerty et al also reported a statistically significant interaction between race and age, such that as African Am ericans age, their risk of developing pressure ulcers increases faster than the risk Caucasians experience as they age, indicating noteworthy racial disparities. Other signifi cant findings identified in thei r study highlight some of the strongest risk factors are non-m odifiable (age, paralysis, race) while others are potentially modifiable (infection, nutritional deficiencies). Therefore, explor ation is needed to determine when interventions are most eff ective in those persons with non-m odifiable risk factors (such as age > 75), or if perhaps interv entions should be initiated in all persons over 75 years old or all persons identified with particular diagnoses. Investigations are also needed that examine the most effective preventive interventions to reduc e or eliminate the identified modifiable risk factors (infection and nutritional deficiencies) and ways to accurately identify them in patients.
34 Age as a risk factor One particular populatio n of interest is our elderly population because pressure ulcers disproportionately affect the ag ing population. Many pressure ulcer studies indicate a strong positive association between older age and the deve lopment of pressure ulcers (Bergstrom et al., 1998; Fisher, Wells, & Harrison, 2004; Whittington, K., & Briones, R., 2004; Maklebust, 2005; Schoonhoven et al., 2006; Scott et al., 2006; Lyder, 2006; Fogerty et al., 2008; Stechmiller et al., 2008). Almost 80% of all pressure ulcer relate d deaths occur in pe rsons over 75 years old (Whittington & Briones, 2004). The US Census Bureau in a press release March 13, 2001 reported there were 3.1 million people over 65 y ears old in the US in 1900, 34.6 million in 1999, with numbers projected to reach 54 million persons by the year 2020 and 87 million by 2040. Pressure ulcer development is a critical problem that is growi ng. Even though data exists to suggest that not all pressu re ulcers in the elderly are avoidabl e and some pressure ulcers develop as a result of multiple systems failure and endof-life physical decline (Thomas, 2001; Padula, Osborne, & Williams, 2008), the data also supports that most pressure ulcers are preventable. Therefore, unless effective interventions are id entified and implemented, this national problem threatens to become epidemic. Id entification of the strongest a nd most accurate pressure ulcer risk factors is imperative in the present day popula tion in order to select patients for appropriate prevention interventions. It is also important to identify if adva nced age alone is a strong enough risk factor to implement preventive interventi ons for all individuals over a certain age group and/or to evaluate whether current preventive meas ures actually reduce incident pressure ulcers in this vulnerable population. Race or ethnicity as a pressure ulcer risk factor Lyder (2006) cites the la ck of sufficient non-white populations in the m ajority of pressure ulcer studies as a problem resulting in conflicting data regarding race or ethnicity as a risk factor
35 for pressure ulcers. However, more recently, th e large study conducted by Fogerty et al. (2008) suggests there are racial disparit ies among individuals with pressure ulcers and ethnicity or race may be a significant pressure ulcer risk factor, with people of darker sk in tones having up to 5 times higher risk than their light er-skinned neighbors. Fogerty et al. reported African American race with an Odds Ration of 5.71 (95% CI 5.35-6.10) in addition to a statistically significant interaction between race and age, such that as African Americans age, their risk of developing pressure ulcers increases faster than Caucasians risk as they ag e. Maklebust (2005) describes at least one explanation for higher pressure ulce r rates among darker-skinn ed individuals than among lighter-skinned persons. She theorized that dark skin does not exhibit noticeable early hyperemic pigment changes (blanchable erythema) w ith superficial pressure that is classically identified as a hallmark to impending pressu re damage at a point when interventions (turning/relieving the pressure off of the effected body part) may reverse these pressure effects. In darker skinned persons, the tissue damage related to unrelieved pressure may not be noticeable until the damage is much deeper and not reversible (Maklebust, 2005, p.369 Smoking as a pressure ulcer risk factor Sm oking has been identified as a strong risk f actor for pressure ulcer development in the scientific literature (Lyder, 2002), but too often is not examined as a risk factor in pressure ulcer studies (Fisher, Wells, & Harrison, 2004; Meaume & Faucher, 2007; Padula, Osburn, & Williams, 2008). The most plausible reason fo r smoking being related to pressure ulcer development is the effects of smoking on tissue o xygen perfusion due to the actions of nicotine on the microvasculature of the skin and increa sed blood viscosity in smokers (Tur, Yosipovitch & Oren-Vulfs, 1992; Monfrecola, Riccio, Savarese Posteraro, & Procaccini, 1998). Other ways smoking impairs tissue tolerance and wound healing ha s been explored in se veral studies such as done by Srensen, Zillmer, gren, Ladelund, Karl smark, and Gottrup (2009) that demonstrated
36 higher transepidermal water loss in smokers vs. non smokers (p<0.01) and epidermal blister levels of matrix metalloproteinase 8 (MMP-8 ) twice as high in smokers as in non smokers (p<0.01). However, conflicting data exists sugge sting that while smokers experience impaired wound healing (Padubidri, Yetman, Browne, Lu cas, Papay, Larive, & Zins, 2001), nicotine may stimulate angiogenesis in the repair of ischemic tissue (Martin, Mousa, S.S., Shaker, & Mousa, S.A., 2009). Low BMI or serum albumin as a pressure ulcer ris k factor Scientific findings are contr oversial regarding Body Mass Inde x (BMI) scores as a risk factor for pressure ulcers. Several studies s uggest that low BMI is associated with higher prevalence of pressure ulcers, and that highe r BMI scores may have a protective affect (Berlowitz et al., 2001, Lyder 2006). Mechanisms be hind this association ar e likely due to low BMI as an indicator of inadequate nutrition an d/or depletion of lean muscle mass (Lyder, 2006) and the resulting decreased subcutaneous fat (body padding) and higher peak pressures noted over bony prominences of cachectic persons (Defloor 1999). Some of these findings suggest that while high BMI may afford a protective effect (l ower peak pressures due to diffuse area of pressure with greater surface area), a very hi gh BMI (morbidly obese with BMI over 28) may be a potential risk due to greater surface areas of increased pressure, immobility and friction/shear forces. In addition, low serum albumin (or prealbumin) levels, while not the most accurate overall nutritional marker, can indicate a nutritiona l compromise and it has been suggested that low levels of serum albumin may be useful in predicting pressure ulcers (Anthony, Reynolds, & Russell, 2000). Inpatient length of stay as a pressure ulcer risk factor Length of stay (num ber of days) has been iden tified in several studies as an independent positive predictor of pressure ulcers, whereby the greater number of days spent in the hospital,
37 the greater risk of acquiring a pressure ulcer. At the very least it is a confounding variable (Theaker, C., Mannan, M., Ives, N., & Soni N, 2000; Fisher, Wells, & Harrison, 2004). Likewise, hours spent in a operati ng room (OR) and number of days in an intensive care unit (ICU) have also been shown in several studies to be strongly associated with pressure ulcer development, such that the greate r the hours spent in the OR/the greater number of days in ICU, the greater risk of pressure ulcer deve lopment (Fisher, Wells, & Harrison, 2004). Medical diagnoses as pressure ulcer risk factors Fogerty et al. (2008) demons trate a strong association be tween m ore than 17 medical diagnoses and the development of pressure ulcers within a very large nationwide sample of over 6 million subjects admitted to acute care facili ties in the US, with 94, 758 persons from the sample developing pressure ulcers. These dia gnoses include (listed in descending order): diagnosis of gangrene (OR 10.94, 95% CI 10.43-11.48), paralysis (OR 10.30, 95% CI 9.9610.96), septicemia (OR 9.78, 95% CI 9.3310.26) osteomyelitis (OR 9.38, 95% CI 8.81-9.99), nutritional deficiencies (OR9.18, 95% CI 8.81-9.99), pneumonitis (OR 8.70, 95% CI 8.33-9.09), urinary tract infection (OR 7.17, 95% CI 6.96-7.38), bacterial infec tion/bacterial skin infection (OR 5.71/3.24, 95% CI 5.49-5.93/3.12-3.38), senility (OR 4.84, 95% CI 4.62-5.07), candidiasis (OR 4.63, 95% CI 4.41-4.86), respiratory fail ure (OR 4.47, 95% CI 4.21-4.76), acute renal failure (OR 4.16, 95% CI 4.00-4.33), cerebrovascular accident or failure (OR 4.04, 95% CI 3.834.27), diabetes mellitus with complications (OR 2.63, 95% CI 2.54-2.73), congestive heart failure (OR 2.63, 95% CI 2.55-2.73), anemia (O R 2.62, 95% CI 2.52-2.73), and complication of device, implant or graft (OR 2.48, 95% CI 2.35-2.61). The purpose of identifying a risk factor is to be able to intervene with preventive measures. Identifying the strongest pressure ulcer risk factors is the first st ep toward being able to provide evidence-based interventions and thereby lower th e likelihood of someone developing a pressure
38 ulcer. Pressure ulcer risk assessment tools provi de a tangible way to quantify potential risk so that interventions may be reserved for those at highest risk and avoid unnecessary interventions with higher financial expenditu res on those who do not need them (Defloor & Grypdonck, 2005). Research is needed that enhances or improves cu rrent risk assessment tools in such a way as to identify those individuals at high est risk. Research is also needed to determine the most effective evidence-based preventive interventions and evaluate preventive measures currently utilized, with an ultimate goal of reducing the incidence rates of pressure ul cers in high-risk populations. Pressure Ulcer Risk Assessment Tools Common pressure u lcer risk assessment tools in use today are largely based on earlier tools such as the Norton Scale published in 1962. Dor een Norton (along with Rhoda McLaren and Dr. Norman Exton-Smith) developed the Norton Scal e in Great Britain during the 1950s (Norton, 1996). It is the first of all of the pressure ulce r risk assessment scales; indeed it is one of the earliest risk assessment scales of any kind. At first, Doreen No rton and her colleagues devised a data collection tool with columns to describe al l factors noted in every patient that might be relevant to pressure ulcer development (p.39) such as a patients weight, build, appetite, medications, preventive measures (14 different skin care products), treatment measures, site and condition of skin, and skin changes. They develope d a rating scale (at a time when rating scales were uncommon) with 5 elements that had weighted descending values for each element from 4 to 1. The elements or factors in their tool were listed as column headings for general physical condition (Norton later said she intended this head ing to include overall nutritional state), mental condition, mobility, activity, a nd incontinence (Norton, 1996). Norton reports the tool was scored for a normal or good function in each factor and for very poor or bad function, with a total possible high scor e of 20 (patient in good overall condition) and low score of 5 (patient in poor overall condition). Norton explains, A descending scale was selected because it
39 correlated with a decline in the patients condit ion (p.39). The Norton conceptual model is a simple model based on their observed factors of general physical condition, mental condition, mobility, activity, and incontinence quantified with a Likert-type scale for each of these factors totaled as one independent variable and pressure ulcer RISK as the dependent or outcome variable. This model proposed that lower total scores have a strong a ssociation with higher pressure ulcer risk (Norton, 1996). Building on early pressure ulcer research done by Norton and others, the Braden Scale for Predicting Pressure Sore Risk (see Appendix A for the full tool and Table 2-2 for an abbreviated description of the subscales) was first publis hed in 1987 (Bergstrom, Demuth, & Braden, 1987; Bergstrom, Braden, Laguzza, & Holman, 1987), and is probably the most widely used pressure ulcer risk assessment tool available today. The th eoretical framework is based on a physiological model depicting factors that contribute to the de velopment of pressure ulcers. It includes factors affecting intensity and duration of pressure (decreased mobility, decreased activity, and decreased sensory perception), which combine w ith intrinsic factors (a ge, nutrition, vascular perfusion) and extrinsic factor s (increased moisture, increased friction, and increased sheer forces) that affect tissue tolera nce (Pieper, 2007). The Braden S cale is publicized as the most extensively tested and studied of the assessmen t tools. The Braden Scale has a potential score ranging from 6 to 23 derived from total scores of its six subscales (sensory/perception, mobility, activity level, moisture/incontin ence, nutrition, and friction/shear). Lower scores on the Braden Scale indicate greater risk for pressure ulcer deve lopment. Very high risk = 9 or below; High risk = 10-12; Moderate risk= 13-14; and Mild risk = 15-18 (see Table 2-2). There is literature by Braden to suggest that if a person has other majo r risk factors present (advanced age, fever, poor nutrition, or hemodynamic instability) their score should be advanced to the next highest level of
40 risk, yet observational studies s uggest nurses do not routinely do th is. Research suggests that nurses frequently underestimate the level of pressu re ulcer risk (Bergstrom et al., 1987; Braden & Bergstrom, 1994; Ayello & Braden, 2002; Stotts & Gunningberg, 2007). Jalali and Rezaie (2005) report sens itivity of risk assessment tools to be the percent of individuals who developed a pressu re ulcer who were assessed (by the tool) to be at risk and specificity to be the percent of individuals who do not develop a pressure ulcer who were assessed (by the tool) not to be at risk (p.94). They report sensitivity and specificity for the Norton scale to be 49% and 100%, the Braden scale was 53% and 100% respectively, which differs slightly from other reports of Braden sensitivity and specificity of 57% and 68%, respectively (Pancobo-Hildalgo et al., 2006; Bolton, 2007). Defloor (1999) criticized Braden & Bergst roms conceptual model because it did not include factors identified in other studies as strongly associated w ith pressure ulcer development, such as specific diseases, dehydration, prot ein deficiency, body build, position, etc. He described his own conceptual scheme of pressu re sore formation, utilizing known risk factors and pathophysiology and expanding on the factors listed in the Braden & Bergstrom model. Defloor also noted that more rese arch is needed especially in re gards to factors such as smoking and low serum protein levels, as well as the influence of preventive measures. Pressure ulcer risk assessment tools provide a tangible way to quantify potential risk so that interventions may be reserved for those at highest risk and avoid unnecessary interventions and higher financial expenditures on those not at risk (Defloor & Gr ypdonck, 2005). Research is needed that enhances or improves current risk assessment tools in such a way as to identify those individuals at highest risk as well as determine if preventive measures currently utilized have any significant affect on the incidence of pre ssure ulcers in high-risk populations.
41 Current Understanding of Pressure Ulcer Ris k Prediction With regard to pressure ulcer risk predic tion, using data collected from research studies on populations twenty years ago poses a problem for application to the current population. Most of the updated pressure ulcer pr evention guidelines available on the National Guideline website ( www.guidelines.gov ) are still based on those risk factors iden tified over twenty years ago and these may not carry the same relevance t oday (RNAO, 2005; AMDA, 2008). Vincent et al. (2006) describes medical technology and clinical procedure advances as well as process of care (organizational/policy) changes within the emer gency medicine and intensive care unit (ICU) arenas over the past 25 years. In addition, mo re patients are having procedures done on an outpatient basis so fewer patients with minor conditions are being admitted to the hospital (Edelman, Weiss, Ashton, & Wray, 1995; CDC, Ambulatory Surgery in the US: 1995). These changes are likely to alter acuity levels, numbers of patient transfers within facilities, and length of stays for patients being admitted to hospitals. Essentially, these factors are apt to change the face of the inpatient population and impact character istics of those at risk of a pressure ulcer. Research Gaps As stated previously, the risk assessm ent t ools commonly used in acute care in the United States are the Braden and Nort on scales. Doreen Norton (1996) conducted her research involving 600 patients (average age 79) in th e geriatric firm of a London hos pital (over a 2-year period of time) in the 1950s. Her scale is based on observed factors in that populatio n. Predictive statistical analysis of each factor was not done. Barbara Braden (Bergstrom, Braden, Laguzza, & Holman, 1987) built on what Norton had done and modified her scale to fit observed factors of the 1980s. She also suggested the at risk cut off score s hould be 16. Only predicti ve (criterion-related) validity was reported as sensitivity and specific ity for each possible total score of the Braden Scale (9 to 23), indicating a maximum 100% sensitivity and 64 to 90% specificity when using
42 total score of 16 or less as the critical at risk score for developing a pressure ulcer. Bergstrom, Braden, Laguzza, & Holman (1987) reported an in ter-rater reliability of the Braden Scale between RN, GS, LPN, and NA staff on each un it (that tested the scale) using Pearson Correlations (86 RN/GS pairs ha d the highest IRR: r= .99, p< .001, 88% agreement; and NA had the lowest IRR: r=.84, p<.001, 12 to 46% agreement) Predictive statistica l analysis, amount of contribution or weights of each s ubscale factor within the tool was not reported. Interestingly, a more recent study suggests that, even after specif ic in-depth training on how to use the Braden Scale, nurses produced reliable Braden Scores only 65% of the time af ter training (Magnan & Maklebust, 2008). While the Norton and Braden risk assessment t ools were derived from factors identified in predictive models of pressure ulcer development in the 1960s and 1980s (and mostly in rehabilitation or long term care settings), there have been no new widely accepted tools based on current pressure ulcer predictive studies within the past 20 years. This is a significant gap in the research. In order to suggest i nnovative pressure ulcer prevention interventi ons, one must start with identifying (or verifying) th e strongest predictors of pressu re ulcer development in presentday populations and settings. A true reflection of current risk factors for specific populations, evidence based risk assessment tools, and ways to improve the accurate use of risk assessment tools are needed (Armstrong et al., 2008). Fogerty et al. (2008) has provid ed us with extremely valuable pressure ulcer predictor information in current day inpatient non-federal populations. This information needs to be examined further an d research is needed to apply that information to the veteran population and investigate these and other factors iden tified in scientific literature associated with pressure ulcer development to determine if these factors should be accounted for
43 in addition to or in place of those factors indicated on current risk assessment tools being used in veteran facilities. Table 2-1. Examples of prev ious research identifying pr essure ulcer risk factors. Researchers Year Population Sample Size Method Strongest Risk Factors Identified Schoonhoven, L., Grobbee, D., Donders, A., et al. 2006 Adult pts in 2 acute care hospitals in Netherlands 1,229 (121 developed stage II to IV pressure ulcers) Prospective Cohort Age, weight at admission, abnormal appearance of skin, friction & shear, planned surgery in coming week Fisher, A., Wells, G., Harrison, M 2004 1993 to 1996 acute care hospitals 1,992 derivation sample, 581 validation sample Prevalence Age, male gender, sensory perception, moisture, mobility, nutrition, friction/shear, Young, J., Nikoletti, S., McCaul, K. Et al. 2002 1998 to 2000 in Western Australia 1,394 3 Cross sectional prevalence studies Age, Braden Score Theaker, C., Mannan, M., Ives, N., & Soni, N. 2000 Adult ICU patients in UK facility 286 (77 developed stage I to IV pressure ulcers) Prospective Norepinephrine infusion, APACHE II score, fecal incontinence, anemia, length of stay Table 2-2. Abbreviated Br aden scale subscales. 1 2 3 4 Sensory Perception Completely limited Very limited Slightly limited No impairment Moisture Constantly moist Very Moist Occasionally moist Rarely moist Activity Bedfast Chairfast Walks occasionally Walks frequently Mobility Completely immobile Very limited Slightly limited No limitation Nutrition Very poor Probably inadequate Adequate Excellent Friction and Shear Definitely a problem Potential problem No apparent problem -blank -Total Score: 9 or less = very high risk, 10-12 = high risk, 13-14= moderate risk, 15-18=at risk, over 18=not at risk (references: Bergstrom, Demuth, & Braden, 1987; Bergstrom, Braden, Kemp, Champagne, & Ruby, 1998; Stotts, 2007) See Appendix A for full Braden Scale for Predicting Pressure Ulcer Risk
44 CHAPTER 3 RESEARCH METHODS Design, Sampling and Setting This retrosp ective descriptive study in an acu tely ill adult inpatient veteran population in north Florida from January-June 2008 determined the differences between Braden scores and other associated risk factors in veterans w ith and without pressure ulcers. This study was designed to determine the pressure ulcer predictability of the Braden score alone, the Braden score + significant medical factors, and signifi cant medical factors al one. A predictive model using logistic regression was considered in the methodology of this study. Medical factors investigated in this study include those reported in recent literature as significant pressure ulcer risk factors: diagnosis of gangrene, anemia, diabetes, malnutrition, osteomyelitis, pneumonia/pneumonitis, septicemia, candidiasis, b acterial skin infection, device/implant/graft complications, urinary tract infection, paralysis, senility, respiratory failu re, acute renal failure, cerebrovascular accident, and conge stive heart failure; as well as ag e, race, hospital and intensive care unit length of stay days, surgery, opera ting room time in hours, smoking status, and a history of previous pressure ulcers. Because this population is predominantly male, gender was not examined as an independent variable. Logistic regression statistical analysis was utilized to determine how effective Braden total scores were in a pressure ulcer predictive model (as well as examine each of the Braden sub-scores) with and without the inclusion of other medical factors. A separate predictive model was examined using only the mo st robust medical fact ors associated with pressure ulcer incidence in this sample (without the Braden score) in an attempt to develop the most parsimonious predictive model. The results of this study enhances th e current knowledge of
45 pressure ulcer risk factors and risk screening/ assessment tools and provides direction for future studies, with an ultimate goal of re ducing incidence of pressure ulcers. The research was conducted entirely at the North Florida / Sout h Georgia Veterans Health Administration acute inpatient faciliti es. Purposive sampling was used. All patient records from the North Florida / South Geor gia Veterans Administration from January 2008 through June 2008 were reviewed to obtain data listed on the data collection sheet (see Appendix B) until all veterans who developed pressure ulcers during their hospitalization were identified and at least the same number of veterans hospi talized in the same f acility but who did not develop pressure ulcers during their hospitalization were documen ted. From general estimates of 10 subjects per variable within the regression analysis and an estimat ed 10 or less final variables, the sample size of 210 was determined sufficient. Furthermore, power analysis was conducted by a statistician prior to the conclusi on of data collection to assure ad equate sample size to reach .80 (80%) power. If the risk factors were present in at least 20% of subjects, the study should have adequate power to detect risk factors with odds ra tios of 3.5 or greater. If the risk factors were present in at least 50% of the subjects, the study should have ad equate power to detect risk factors with odds ratios of 2.5 or greater (with a sample of at least 100 cases and at least 100 controls). Over 500 electronic pa tient records were reviewed to obtain the minimum number of 100 subjects with incident pressu re ulcers and at least the same number of subjec ts hospitalized during the same time without incident pressure ulcers but who had enough re corded data in their chart to complete the study da ta collection sheets (s ee Appendix B), resul ting in a total study sample of 213 adult veterans admitted to the North Florida/South Georgia Veterans Administration acute care facility between January 2008 and through the end of June 2008. Veterans Administration Nursing Outcomes Da tabase (VANOD) records were examined to
46 obtain the records of all those ve terans who developed pressure ulcers while hospitalized during the selected dates. This resulted in 100 cas es with pressure ulcers. VANOD database and admissions records were then examined to obt ain records of patients hospitalized during the same time frame but who did not develop pressure ulcers (10 or more subjects per month) and whose Braden Scores were less than 19. These subjects were randomly selected by going down an alphabetical list of admissions for each mont h during the same time period. This resulted in 113 subjects without pressure ulcers. Exclusion criteria for the study were: non incident pressure ulcers and missing more than 6 required fields fr om the data collection sheet. Over 240 charts were excluded due to missing 6 or more key com ponents of the data collection tool, as were over 20 patients that were listed by reports as having developed pressure ulcers while hospitalized but further examination of admission documentation actua lly recorded the pressure ulcer as present on admission. This resulted in more than 287 records being excluded. No subject was included more than once. All data collect ed on those patients with pressu re ulcers was recorded only if documented prior to pressure ulcer development. Procedure Internal Review Board (IRB) approval for this retrospective data analysis as an exem pt study (no personally identifying information to be recorded) was obtained from both the University of Florida (UF) and the Veterans Ad ministration (VA). As an exempt retrospective analysis, no informed consent was necessar y. Data was collected using the Veterans Administration (VA) Computeri zed Patient Record System (CPRS) accessed at the VA in Gainesville, Florida. Medical data already collected and stored w ithin the patient record system was reviewed, starting with those patients adm itted to the North Florida/South Georgia VA acute care facility anytime during Ja nuary 2008 to June 2008 identifi ed by Veterans Administration Nursing Outcomes Database (VANOD) records, pr evalence surveys and tracking logs as having
47 developed a pressure ulcer during their admi ssion. Next, random patient admissions were reviewed from the same time period January to June 2008 excluding those patients who were identified as having developed pressure ulcers, or having data previous ly collected, or missing more than 6 required fields on the data collecti on sheet. The patient records were examined and only the non-identifying data listed on the data collection sh eet (Appendix B) was recorded. Data from the data collection sheet wa s entered onto a spreadsheet stored on a secure file on the VA intranet and a VA security-approved jump drive (for transport during data analysis). The completed data collection sheets, as well as the data spreadsheet and jump drive, were stored in a locked location at the North Flor ida South Georgia VA facility. Data Collection Data was co llected using a Data Collection Tool (see Appendix B). Data from the patient data records of the VA was queried using the top 17 diagnoses identified in the Fogerty et al. study (2008). These diagnoses were reported as bina ry coded categorical variables with no (not present) coded as 0 and yes (diagnosis was present) coded as 1. The diagnoses were reported as present during the hospitalization if these di agnoses were identified using International Classification of Diseases ve rsion 9 (ICD-9) as a discharg e diagnosis. Typically up to 10 discharge diagnoses were recorded on the discha rge summary of patients within this study, with 4 discharge diagnoses recorded in this study be ing the least number a nd 11 discharge diagnoses recorded as the most for one subject. Since th e average number of discha rge diagnosis codes was between 4 and 10, there was concern that many active and pertinent di agnoses would not be accurately captured. To address this concern, the patients medical record for the hospitalization under review was examined and the diagnoses or medical factors examined in this study were recorded as present in the subject if ther e was any evidence provided in medical provider assessment documentation, specialist consults with diagnosis confir mation, active disease
48 problem lists, or laboratory and diagnostic studies (such as radiological images interpreted by a radiologist) as well as by discha rge ICD9 codes. One individual, the primary investigator, who is an advanced registered nurse practitioner and bo ard certified in family practice as well as a certified wound specialist, accomplished all data collection and medical record reviews. Reported diagnoses included: gangrene ( 785.4 and related codes); anemia (280.0 and related codes), diabetes mellitus with comp lications (250.1-250.9); malnutrition (260, 261, 262, 263.0-263.9, 995.85); osteomyelitis (730.0-730.9); pne umonia or pneumonitis (480.0, 486, and related codes); septicemia (038.1-038.9, 998.59 a nd related codes); candidiasis codes (111.8, 112.0, 112.2, 112.84, 112.89); bacterial skin infec tion or cellulitis (682.6, 682.7, 686.9, 998.51 and related codes); complication of devi ce or implant/graft (996.0-996.89, 429.4-429.9); urinary tract infection (098.0, 098.2, 131.00, 559.0, 597.80); paralysis (045.0, 300.11, 332.0 and related codes); senility (259.8, 290.10 and related c odes); respiratory fa ilure (348.8, 518.81-581.84), acute renal failure (403.91, 404.02 and related co des), cerebrovascular (437.8), and congestive heart failure (428.0, 428.1, 428.9). There were no mycosis (031.9) codes found. Since some diagnoses are commonly similar or conceptually related, certain dia gnoses were examined separately and then combined counting overla pping cases only once if they had both diagnoses (CVA + paralysis; pneumonia/pneumonitis + respiratory failure). In addition, hemoglobin and hematocrit were also recorded but are considered in the diagnosis of anemia, so only diagnosis of anemia was used in analysis. Similarly, low body mass index (BMI) and low serum albumin or pre-albumin levels were recorded, but since these indices are included in the registered dieticians nutritional assessment, it was decided to select the registered dietician nutritional assessment of severe nutritional compromise as a proxy for malnutrition and not use the lab
49 indices in the final predictive m odels, so that cases would not be counted more than once for factors that were similar conceptually. A diagnosis or documented histor y of previous pressure ulce r prior to current admission (707.0.9) was also queried and in cluded as a coded categorical independent variable (no=0 / yes=1). Age, race, and length of stay for hospital admission, ICU or operating room time (if applicable) were also recorded. Age was reported as a continuous scale variable in number of years (up to maximum age of 85). All ages over 85 were recorded only as or older to further protect the identity of these in dividuals. Gender was reported as a total sample frequency in demographics but was not considered as a variable in the statistical anal ysis due to the >85% male veteran population within the VA system. Race/ethnicity was reported as a categorical variable (African American / Caucasian / Othe r). Length of hospital stay and ICU stay (if applicable) was reported as a c ontinuous variable (number of days), while time in operating room (if applicable) was repor ted as number of hours, r ounded to one decimal place. The outcome or dependent variab le was reported as a dichotom ous coded variable (did not develop a pressure ulcer was recorded as 0, and did develop a pressure ul cer was coded as a 1). The incidence of pressure ulcers was identified by any diagnosis of pressure ulcer (including suspected deep tissue injury discoloration or DTI and unstageable pressure ulcers) during the current acute inpatient admission that was substantiated by detail ed skin and wound assessments by a licensed nurse or physician and/or wound specialist consult asse ssment. All subjects identified as having developed a pressure ulcer were further reported as frequencies by stage and location of pressure ulcer. Patient records were crosschecked with prevalence data collected quarterly by the VA system.
50 Non-identifiable data was recorded directly from the computerized patient data system onto a data collection tool (see Appendix A) a nd then onto an Excel spreadsheet, transformed onto an SPSS data set. If individual records were missing more than six pieces of required information on the data collection sheet (except for laboratory values), the record was not used in the sample. A total of 108 veterans in the total sample di d not have sufficient data recorded in their medical record to determine if they had any histor y of previous pressure ulcers. Since there were 108 subjects out of 213 with missing data on history of previous pressure ulcers, this variable was not included in the regression analysis. A total of 71 veterans reported recent weight loss just prior to hospitalizat ion but 95 veterans did not have this data recorded in their medical record nor had recent weights recorded to determin e this information, so this variable was also not included in final analysis due to the larg e number of missing cases. Similarly, serum prealbumin laboratory values were only recorded in 58 total subjects within the sample, so this variable was not included in final data analysis. The National Institutes of H ealth (NIH) report that body ma ss index (BMI) is a number calculated by dividing weight in pounds by he ight in inches squared and multiplying by a conversion factor of 703. This number is used as an indicator of body fat to screen people for health risk weight categories ( http://www.nhlbisupport.com/bmi/ ). BMI is not an adequate m easure of total nutrition. However, since a nutritional assessment performed by a registered dietician (RD) at the VA includes classifying th e patients nutritional compromise (none, mild, moderate, or severe) based on a combination of an thropometric, biological, clinical, and dietary history data that include s BMI, nutritional history, unintentional weight loss as a percent of usual body weight, percent of ideal body weight, diet serum albumin, and total lymphocyte count
51 (Lowery, Hiller, Davis, & Shore, 1998), a deci sion was made to run this variable in the predictive models alone and run another separa te LR model with th e BMI factor. The RD nutrition assessment of severe malnutrition wa s selected as a proxy for a diagnosis of malnutrition, since it was determined to be a more accurate appraisal of nutrition than ICD9 code diagnoses. The diagnosis of urinary tract infec tion (599.0) occurred 5 times concurrently with urogenital candida infection, so the ICD-9 dia gnosis code of 112.2 (urogenital candidiasis) was captured under the category of candidiasis but if the patient had a separate diagnosis code of 599.0, the case was also left in the UTI category. Data Analysis Descriptive statistics (means, stand ard devi ations and ranges for age, hospital and ICU length of stay, total Braden scores, albumin, hemoglobin, hema tocrit, total body mass index or BMI, and total hours in operating room) were re ported for the total sample as well as for each outcome group (those that did vs. did not develop pressure ulcers). Frequencies were reported for race, smoking status, recent wei ght loss, BMI category, surgery during hospitalization, patient history of previous pressure ulcer, nutrition category as determined by a registered dietician (RD), and each predetermined high-risk diagnosis as identified by Fogerty et al. (2008). These include: gangrene, anemia, diabetes with co mplication, osteomyelitis, pneumonia/pneumonitis, sepsis/septicemia, bacterial skin infection, compli cations of device or impl ant/graft, urinary tract infection, malnutrition, paralysis, senility, respiratory failure, acu te renal failure, cerebrovascular accident, and congestive heart failure. Candidiasis (a ny site candida infection) is recorded instead of mycosis (ICD-9 031.9) because there were no 031.9 ICD-9 codes recorded in any of the charts. The frequencies of all of these diagnoses were reported for the total sample as well as each of the two outcome groups. Bivariate an alysis was conducted to examine significant differences between with pressure ulcer and without pressure ulcer groups. Differences between
52 groups with regard to scale variables were examin ed using independent samples t-tests statistics. Differences between groups with regard to ca tegorical variables were examined with Chisquared and Mann-Whitney U statistics. Variables we re entered into logis tic regression models only if significant bivariate differences were note d. Independent samples t-tests were used to look for differences between group means for ag e, Hgb/Hct, BMI, prealbumin, albumin, length of hospital and ICU stays, hours in the operating r oom (if applicable), total Braden Scores, and Braden Subscores. Non parametric statistics su ch as Chi-square and Mann-Whitney U statistics were calculated to explore differences between the two outcome groups for race, smoking status, patient history of previous pressure ulcer, surg ery, severe nutritional compromise, and each predetermined medical diagnosis. The data was analyzed comparing differences between those individuals that actually did de velop a pressure ulcer and those that did not develop a pressure ulcer within the sample group of adult acute inpatient veterans. Multivariate analysis in this study included ex amination of the standardized residuals and DfBetas for outliers and influential cases for scale independent variab les (age, total Braden Scores, length of stay in days, length of ICU stay in days, a nd hours in the opera ting room). Any case with a standardized residual (ZRE) >3 was judge d to be an outlier, but this did not represent more than 5% of cases with ZRE scores over 2.0, so was not determined to be potentially threatening regarding possible bias due to outliers (Field, 2005). Any case identified as an outlier was examined for possible data entry error or mi scalculation. If the case was determined to be valid, it was left in the analysis. As there were no DfBetas >1 reported in the analysis, no cases were considered influential cases for this study (Field, 2005). Logistic re gression (LR) analysis was used to determine how effective a pressure ulcer predictive model was using Braden total scores with and without the incl usion of other medical factors identified in the scientific
53 literature. Whether a patient did or did not develop a pressure ul cer was the dichotomous (binary) categorical dependent variable. Independent variables (Braden to tal scores and sub-scores, and specific medical factors) that were found to have a significant association with pressure ulcers were loaded in the logistic re gression (LR) model in a stepwise and forward LR approach as determined by previous literature and the SPSS software. A registered dietician (RD) nutritional assessment at the VA includes classifying the patients nutritional compromise (none, mild, mo derate, or severe) based on a combination of anthropometric, biological, clinical, and dietary history data such as: BMI, nutritional history, unintentional weight loss as a percent of usual body weight, per cent of ideal body weight, diet, serum albumin, and total lymphocyte count (Lower y, Hiller, Davis, & Shore, 1998). An a priori decision was made to run the RD nutritional asse ssment of severe nutritional compromise as a proxy variable for malnutrition in the predictive models by itse lf (without other nutritional indicators such as BMI or albumin) and run anot her LR model with BMI if necessary. Any factor that was not found to be a signifi cant predictor within the LR wa s deleted from the model and the model re-run in order to determine the most pars imonious model. In the event that two factors were similar in concept and may have subjec ts counted for both fact ors (such as CVA and paralysis or pneumonia/pneumonitis and acute respirat ory failure) the factors were run separately in different models and if bot h were significant predictors, we re run in another model as a combined new factor (with overlapping cases co unted only once). If only one of the variables similar in concept was found to be a significant predic tor, it was selected as the variable to run in the final model. For example, CVA/paralysis was combined to form a new variable, while pneumonia/pneumonitis was run alone and acute respiratory failure was eliminated from final analysis as it was not a significant predictor. F actors where >20% of the sample did not have the
54 recorded data were not included in the LR analysis (his tory of previous pre ssure ulcers, history of recent weight loss, an d pre-albumin levels). The first step of logistic regr ession (LR) was to run a predictive model with total Braden scores entered in step one as a predictor alone. Next in a sepa rate LR, the Braden sub-scores (sensory perception, moisture, activity, mobilit y, nutrition, friction) were run as independent variables (predictors) entered all together in fo rward LR method in step one. Thirdly, in a separate LR model, the Braden total score was en tered in step one (force d entry), with nutrition category of severe nutritional compromise by RD en tered in step two. Fourth, Braden total scores were entered in the first step of a LR model (see Table 4-5) and th en all other significant factors (surgery, BMI, candidiasis, hospital los, CVA/paralysis, sepsis, UTI, pneumonia/pneumonitis, and senility/dementia) were entered in the second step of LR in a forward LR method by SPSS. In models 5 through 7 the most parsimonious model (based on the least amount of significant variables that could most accurately predict the subjects who developed pressure ulcers within the sample) was run first with and then without entering the Braden total scores in step one. Binary logistic regression analysis was c onducted using SPSS version 17.0. Binary logistic regression was utilized in SPSS fo r statistical analys is of the data to examine significant predictors of pressure ulcers within the identi fied high risk population. Data analysis was run again by a statistician using the same data set with SAS software program to assure similar outcomes. Classification tables and goodness of fit statistics were examined for each final predictive model (with total Braden scores, Br aden sub-scores, both w ith and without other significant medical factors, and those medical factors alone). Significant differences in the models were explored to determ ine if the identified di agnosis codes or other variables enhanced the predictive ability of the Braden Score.
55 Assumptions Assum ptions to be met for logistic regr ession (non-parametric) analysis include representative sample (as random as possible) independence of scores, no empty cells, and muticollinearity. Representativeness of the purposive sample was met due to the sample size (n > 200) and relatively random selection of subjects. Subjects were selected to include all patients that developed pressure ulcer s while inpatient from Janua ry 2008 to June 2008, and then randomly from each of the same months among patie nts that did not develop pressure ulcers but had Braden scores 18 or less, until over 200 to tal sample size was achieved. Independence (cases must be independent of each other) was me t because there were no repeated measures and all data collected was from individual participants. In cases wher e conceptually similar variables were counted (such as pneumonia/pneumonitis and respiratory failure), the two variables were examined separately in the models, as well as co mbined in a new variable (CVA/paralysis) with the overlapping cases only counted once. Empty cel ls must be avoided in bivariate crosstabs analysis. In the event of a larg e number of missing cases, such as where history of weight loss or history of previous pressure ulcer was not r ecorded in many charts, the variables were not included in final analysis. The assumption of normality does not apply to logistic regression (LR), so no distribution histograms or Sh apiro-Wilk statistics were reported. Additional Analyses W ithin the first month of data collection, it was noted that the ICD9 discharge diagnosis codes did not always reflect all of the actual pertinent diagnoses present during hospitalization, so the investigator went back to the IRB with a request to collect additional data to include chart review of active problem lists, sp ecialist consults, labs provider progress not es and assessments which would identify if the patient had the diag nosis of interest present during hospitalization prior to any development of pressure ulcer. A ppendix E reports specific medical diagnoses by
56 ICD9 code recorded in the veterans medical re cord upon discharge for the total sample versus how many actual diagnoses were present in the medical record during th at hospitalization as evidenced by documentation in the provider notes, active problem list or la bs during that hospital stay (as well as discharge dia gnosis code list). The frequency of diagnosis is reported for both groups within the sample (without pressure ulcers vs. with pressure ulcers).
57 CHAPTER 4 RESULTS Characteristics of the Sample The total sample included 206 m ale (97% of sample) and 7 female subjects (3% of total sample). Gender was not included in the statistical analysis becaus e the majority of veterans are male, but is reported in the sample demographi cs. Fifty-nine veterans in the sample were smokers and 1 veteran did not have smoking status recorded. Twenty nine (13.6%) veterans in the total sample were African Americans, 171 (8 0.3%) were Caucasians, and 10 veterans (4.7%) had their race listed as other in the medical re cord. Three veterans within the sample did not have any race data recorded. All subjects th at did not have information documented on the specific factor of interest listed in Table 4-1 are listed as missing cases from sample in the table. Age of subjects in the total sample ranged from 47 years of age to >85 years of age, with 50% of the total sample over the age of 72. All subj ects over 85 years of age were listed as or older in order to protect potential ly identifiable data and maintain the exempt status of the IRB approval for the study. Age and othe r continuous numerical (or scale) variables are described for the entire sample in Table 4-2. The mean age of the total sample was 71years of age (SD 10.6, range 47 to > 85 years old). Hospital length of st ay in days ranged from 2 to 110 days with an average for the total sample of 11.72 days. Inte nsive care unit (ICU) lengt h of stay in days averaged 8 days for the entire sample. Opera ting room (OR) time in hours averaged 4.7 hours with a range of 0.5 12 hours for the total samp le. Hemoglobin (Hgb) levels averaged 9.94 g/dL for the total sample. Hematocrit (Hct) levels av erage 30.5% for the total sample. Twenty seven (12.7%) veterans had a diagnosis of cerebrovascular accident (CVA). Sixteen (7.5%) veterans had a diagnosis of paralysis. Seven veterans (3.3%) had a diagnosis of osteomyelitis. Thirty five
58 (16.4%) veterans had a diagnosis of senility or dementia. Forty (18.8%) veterans had a diagnosis of sepsis or septicemia. Eighty six (40.4%) veterans had a diagnos is of urinary tract infection. Seven (3.3%) veterans had a diagnosis of gangren e. One hundred and ninety eight (93%) of the veterans in the total sample had some dia gnosis or lab report duri ng their hospitalization diagnostic of anemia. Ninety one (42.7%) of th e veterans in the sample had a diagnosis of diabetes mellitus with complication. Fifty (23.5%) veterans had a diagnosis of pneumonia or pneumonitis. Thirty eight (17.8%) veterans had a diagnosis of candidiasis or candida fungal infection (skin, urogenital, esophag itis, or thrush). Twenty seve n (12.7%) veterans had diagnosis of bacterial skin infect ion or cellulitis. Thirty four (16%) veterans had diagnosis of graft or device complication. Thirty one (14.6%) veterans had diagnosis of acute respiratory failure. Seventy three (34.3%) veterans had diagnosis of acute renal failure. Fift y five (25.8%) veterans had a diagnosis of congestive heart failure (CHF) during the hospitalization. Seventy nine (37.1%) of veterans within the sample had a body mass index (BMI) over 28. One hundred and eight (50.7%) of veterans within the sample had a BMI be tween 19 and 28. Twenty four (11.3) veterans had a BMI below 18. Two hundred and two veterans within the sa mple had a nutritional assessment by a registered dietician (RD). Thirty nine (18.3%) veterans had a diagnosis by RD of mild nutritional compromise 124 (58.2%) had a diagnosis by RD of moderate nutritional compromise, and 39 (18.3%) had a diagnosis by RD of severe nutritional compromise. Three veterans in the sample had no nutritional comp romise as determined by RD evaluation. The study sample includes a total number of 36 (17%) vetera ns that had surgery during their hospitalization and 177 vetera ns that did not have surger y (see Table 4-1). Seventy six veterans had no history of previ ous pressure ulcers. Twenty nine veterans did have a previous history of pressure ulcers. However, a total of 108 veterans in the sample did not have this
59 information recorded. A total of 71 veterans reported recent weight loss just prior to hospitalization but 95 veterans did not have this data recorded in their medical record nor had recent weights recorded to determine this info rmation. Serum albumin levels averaged 2.97 g/dL for the total sample. Serum pre-albumin laborat ory values averaged 9.67 mg/dL for the total sample, but pre-albumin levels we re only recorded in 58 total subj ects. Please see Table 4-3 for frequencies of medical diagnoses by medical chart review (inc luding ICD 9 codes). Comparison of Groups Com parisons of the veterans who did develop pressure ulcers during hospitalization versus those veterans who did not develop pressure ul cers during their acute hospitalization are made with the following tables and figures. Table 4-4 describes the means and standard deviation of other numerical, continuous (sca le) variables for both groups (with /without PU) as well as twotailed t-test results for significant differences between means of several of the scale variables (age, hospital length of stay in days, ICU length of stay in days, operating room time in hours, serum albumin, serum prealbumin, hemoglobin in g/dL, hematocrit %, and Body Mass Index). Table 4-4 examines the t-test for significant differences between average total Braden Risk Assessment scores and each sepa rate Braden Scale sub-score of : sensory perception, moisture, activity, mobility, nutrition, and friction/shear. Average age of veterans with PU was 71.5 year s. Average age of vete rans without PU was 70.5 years (see Table 4-4). The difference in mean ages of veterans with in the two groups was 1.5 years, which was not statisti cally significant (p = 0.522). Hospita l length of stay in days ranged from 2 to 110 days with an average for the veterans without PU of 8.5 days, and those with PU averaged 15.4 days. The mean difference was 6.92 days, which was statistically significant (p = 0.000). Intensive car e unit (ICU) length of stay in days averaged 5.72 days for the veterans without PU and 8.87 days for vetera ns with PU. The mean difference is 3.15 days,
60 which was statistically signifi cant (p = 0.039). Operating room (OR) time in hours averaged 3.4 hours for the veterans without PU and 5.15 hours for veterans with PU. The mean difference is 1.35 hours, which was not statistically signi ficant (p = 0.394). Hemoglobin (Hgb) levels averaged 10.43 g/dL for the veterans without PU and 9.39 g/dL for the veterans with PU; the mean difference (1.02 g/dL) is statistically sign ificant (p=0.000). Hema tocrit (Hct) levels averaged 32.0% for veterans without PU a nd 28.8% for the veterans with PU; the mean difference (3.2%) is statistically significant (p=0.000). Serum al bumin levels averaged 3.26 g/dL for the veterans without PU, and 2.66 g/dL for thos e with PU; the mean difference (0.6 g/dL) is statistically significant (p=0.000) Pre-albumin levels averaged 12.21 mg/dL for the veterans without PU, and 8.44 mg/dL for those with PU; the mean difference (3.77 mg /dL) is statistically significant (p=0.021). Body mass index (BMI) averag ed 27.7 for the veterans without PU, and 25.5 for those with PU; the mean difference (2.2) is statistically si gnificant (p=0.029). Total Braden scores averaged 14.6 for the veterans wi thout PU, and 13 for those with PU; the mean difference (1.6) is statistically significant (p=0.00 0). Braden sensory perception sub-score averaged 2.86 for the veterans without PU, and 2.65 for those with PU; the mean difference (0.21) is not statistically signi ficant (p=0.062). Braden moisture sub-score averaged 3.18 for the veterans without PU, and 3.10 for those with PU; the mean difference (0.08) is not statistically significant (p=0.395). Braden activ ity sub-score averaged 2.02 for the veterans without PU, and 1.31 for those with PU; the mean difference (0.71) is statistically significant (p=0.000). Braden mobility sub-score averaged 2.43 for the veterans without PU, and 2.09 for those with PU; the mean difference (0.34) is statis tically significant (p=0.000). Braden nutrition sub-score averaged 2.12 for the veterans without PU, and 2.02 for those with PU; the mean difference (0.10) is not statistically significant (p=0.292) Braden friction sub-score av eraged 2.12 for the veterans
61 without PU, and 1.79 for those with PU; the mean difference (0.33) is statistically significant (p=0.000). Table 4-5 examines the differences between groups (those without pressure ulcers vs. those with pressure ulcers) in the sample re garding the frequency of specific categorical demographic and medical factors. Comparison between groups was examined using Chi-squared statistic and the level of significance of this test is reported for each factor. Racial distribution of the sample includes 29 African American vetera ns (14 without PU, 15 with PU), 171 Caucasian veterans (87 without PU, 84 with PU), 10 listed as other ethnicity (9 without PU and 1 with PU), and 3 veterans who did not have any race recorded (no pressure ulcers were present in these three veterans). The difference in racial distri bution as a whole among those veterans that did develop pressure ulcers and thos e that did not was statistically significant (p=0.033). A total of 36 veterans had surgery during their hospitaliz ation (12 without PU, 24 with PU), and 177 veterans that did not have surgery (101 without PU, 76 with PU); the group difference is statistically significant (p=0.009). A total of 59 veterans within the sample were current smokers (35 without PU, 24 with PU), and 153 veterans reported they were not current smokers (77 without PU, 76 with PU); the group difference is not statistically significant (p=0.321). A total of 202 veterans (95% of the total sample) had a nutritional evaluation done by a Registered Dietician (RD). Eleven veterans (5%) did not ha ve a nutritional evaluation by a RD during their hospitalization. Of those subjects who had a nutritional evaluation, 3 were determined to have no nutritional compromise (none of th e 3 developed pressure ulcers); 39 (19%) were determined to have mild nutritional compromise (30 without PU, 9 with PU); 124 (61%) were determined to have moderate nutritional compromise (64 with out PU, 60 with PU); and 39 subjects were determined to have severe nutritional compromi se (8 without PU and 31 with PU); the group
62 difference is statistically si gnificant (p=0.000). Body Mass Index (BMI) was examined as both a scale numerical variable and as a categorical variable with 2 extremes and one midlevel BMI range labeled Normal or Average for the purpose of this study: Low BMI <19 (underweight), Normal BMI 19-28, and High BMI > 28 (obese). Only 2 veterans did not have any BMI calculated or recent weight documented to calc ulate a BMI from. There were 24 veterans with Low BMI <19 (8 without PU, 16 with PU), 108 veterans with Normal BMI 19-28 (53 without PU, 55 with PU), 79 veterans with High BMI > 28 (50 without PU, 29 with PU); the group difference is statistically significant (p= 0.023) Recent weight loss was recorded on only 71 veterans (47 veterans had stable weights record ed for the 3 months prior to admission) and 95 veterans had this data missing in their chart. Of the 71 veterans w ith recent weight loss documented, 25 did not develop a PU during hospitalization and 46 did develop a PU during their hospitalization; the group difference is sta tistically significant (p= 0.000). The total sample (213 veterans) only had 29 (14%) wi th previous pressure ulcers documented in their chart. Six (2.6%) of these 29 veterans did not develop pressure ulcers, while 23 did develop PU (79%); the group difference is statistically significant (p= 0.000). Please see Table 4-5 for each diagnosis (anemia, cerebrovascular accident or CVA/pa ralysis, congestive heart failure or CHF, osteomyelitis, gangrene, sepsis, diabetes mellitu s or DM with complications, bacterial skin infection, urinary tract infection or UTI, pneum onia/pneumonitis, senility/dementia, candidiasis, device or graft complication, acute respirator y failure, and acute renal failure) and the corresponding frequencies and group differences of these diagnoses in the total sample. Those diagnoses that demonstrated statistically significant differences between groups were: CVA/paralysis (p = 0.001), sepsis (p = 0.029), pneumonia/pneumonitis (p = 0.000), acute respiratory failure (p = 0.001), senility/dementia (p = 0.039), candidiasis (p = 0.000), and UTI (p
63 = 0.000). The diagnosis of urinary tract infectio n (599.0) occurred 5 times concurrently with urogenital candida infection, so the ICD-9 dia gnosis code of 112.2 (urogenital candidiasis) was captured under the category of candidiasis but if the patient had a separate diagnosis code of 599.0, the case was also left in the UTI category. Out of curiosity, these 5 cases were removed from both variables and each run separately only c ounting the 5 cases once, but it did not affect the data analysis, so all 5 cases we re left in under both variables. Pressure Ulcer Characteristics The veterans with pressure ulcers w ere typically identified with stage II pressure ulcers or worse. Only 5 veterans with pressure ulcers we re reported as having developed stage I pressure ulcers, 74 veterans developed stage II pressure ulcers, 5 veterans developed stage III, no veterans were identified as having developed stage IV pr essure ulcers, 13 vetera ns developed suspected Deep Tissue Injury, and 3 veterans developed un stageable pressure ulcers. The most frequent anatomical location of the pressu re ulcers in the group of veterans in the sample who developed pressure ulcers during their hos pitalization was the sacrum/coccyx area (n=52), and buttocks (n=25), followed by heel (n=8), hip (n=1), other sites (n=9), and multiple sites of pressure ulcers (n=5). Please see Table 4-6 for pressure ulcer stages and locations. Regression Analysis Logistic regression was first r un en tering the total Braden score as a predictor variable in step one and developed pressure ulcer as the dependent variable. Goodness of Fit statistics (-2 log likelihood) were examined for every model and classification tables were examined for how accurately the model was able to predict cases. Please see Table 4-7 for Logistic regression analysis results. This first Braden only model (Model 1) accurately classified 68% of the total sample (70% accuracy in the no PU group and 65% accuracy in the PU group). Secondly, the significant subscales of the Braden were entered in step one of a separate LR model (activity,
64 mobility, and friction entered in forward LR method). The SPSS statistical program eliminated mobility as a predictor variable from the m odel (Model 2), as it did not reach appropriate significance. The Braden activity and friction subscale scores in a predictive model by themselves could accurately classify 72% of to tal sample (65.5% without PU were correctly classified and 80% with PU were correctly classi fied). Thirdly, total Braden scores was entered in step one (forced entry) and RD nutritional ev aluation (nutritional comp romise by categories of none, mild, moderate, severe) was entered in step two of a separate LR model (Model 3). This was done in a separate model so th at other indices (if si gnificant) which are al ready considered in the RD nutritional evaluation (such as BMI) coul d be entered in a separate model and the strongest predictor selected for final regres sion models. The total Br aden + RD nutritional compromise categories in a model (Model 3) could accurately predict 73% of total sample (74% no PU/ 70% yes PU groups). Next a separate LR model (Model 4) was run with total Braden scores entered in step 1 (f orced entry) and all other si gnificant variables (surgery, pneumonia/pneumonitis, CVA/Paralysis, UTI, ICU los, hospital los, Hct, BMI category, sepsis/septicemia, senility/dementia, albumin, ca ndidiasis) entered in step 2 in a forward LR method in SPSS. In this model (Model 4), SPSS eliminated BMI, ICU los, Hct, albumin, sepsis/septicemia, senility/dementia, and hospital los as significant contributors to the predictive model. Only pneumonia/pneumonitis, candidiasis, su rgery, UTI, and CVA/paralysis were left in the model. In the fifth model, Braden scores as a predictor variable was en tered in step 1 of the LR (forced entry) and pneumonia/pneumonitis, candidiasis, surgery, severe nutritional compromise by RD eval, CVA/paralysis, and UTI we re entered in step 2 in a forward LR method in SPSS This model (Model 5) could accurately predict 80.8% of tota l sample (85% no PU/ 76% of yes PU group). In an attempt to select the most parsimonious model with the least
65 number of variables that could pr edict the most cases, the sixth model was run with total Braden scores entered in step 1 (forced entry) and only the strongest of the previous models predictors entered in a forward LR method in step 2 ( pneumonia/pneumonitis, candidiasis, surgery, severe nutritional compromise by RD evaluation). This model could accurately predict 78.4% of total sample (82.3% no PU and 74% of the yes PU gr oup). The final model was run without the total Braden score since the Braden total score in se veral models did not achie ve adequate level of significance in step 2. Model 7 was run with only pneumonia/pneumonitis, candidiasis, surgery and severe nutritional compromise by RD eval uation (a proxy for mal nutrition). These four predictor variables entered into step 1 in forward LR fashion were able to correctly classify 77% of the total sample, with 71.7% of the no pressure ulcer group correctly classified, and more importantly, 83% of the with pressure ulcer gr oup correctly classified (see Table 4-7). The relative risk of incident pressure ulcers for veterans in the sample with low Braden total scores is estimated with an odds ratio of 0.784 (95% confidence interval 0.675 0.910, p=0.001), such that for every point change lower in the Braden score (from 18), the veteran in the sample was 1.3 times more likely to develop a pressure ulcer. The rela tive risk of incident pressure ulcers for veterans in the sample with a diagnosis of pneumonia or pneumonitis is estimated with an odds ratio of 7.9 (95% CI 3.4 18.61, p <0.001), indicating a veteran in the sample with pneumonia/pneumonitis was almost 8 times more likely to have a pressure ulcer than someone without either of these diagnoses. Th e relative risk of incident pressure ulcers for veterans in the sample with a candidiasis diagnosis (skin, esophageal, urinary, or blood candidiasis) is estimated with an odds ratio of 9.5 (95% CI 3.4 26.4, p 0.000), indicating a veteran in the sample with candidiasis was 9 times more likely to develop pressure ulcers than those that did not have this diagnosis. The relative risk of incident pressure ulcers for veterans in
66 the sample with a severe nutritional compromise is estimated with an odds ratio of 4.98 (CI 1.912.9, p 0.001), such that veterans who were evalua ted by a registered dietician and assessed to have severe nutritional compromise were almost 5 times more likely to have a pressure ulcer develop than those veterans who had no, mild or moderate nutritional compromise. Finally, the occurrence of surgery during hospitalization al so added to the risk of pressure ulcer development. The relative risk of incident pressu re ulcers for veterans in the sample who had surgery during their hospitaliza tion is estimated with an odds ratio of 5.8 (CI 2.5 13.8, p 0.000), suggesting that surgery durin g hospitalization for the veterans in our sample increased their risk of pressure ulcer almost 6 times that of the veterans that di d not have surgery (see Table 4-8).
67 Table 4-1. Pressure ulcer study sample demographics. Total sample n= % SampleMissing cases from sample Male 206 97% Female 73 %Race: -3 African American 29 14% Caucasian 171 80% Other race 10 5% Surgery 36 17% 0 Smokers 59 28% 1 BMI by category: -2 Low BMI <19 24 11% Norm 19 to 28 108 51% High BMI >28 79 37% Recent wt loss 71 33%95/44% Previous PU 29 14%108/51% RD Nutrition Assessment: 202 95%11/5% Mild compromise 39 18% Moderate compromise 124 58% Severe compromise 39 18% Anemia 198 93% 0 CVA/paralysis 34 16% 0 CHF diagnosis 55 26% 0 Osteomyelitis diagnosis 73 %0 Gangrene 73 %0 Sepsis 40 19% 0 DM w/complications 91 43% 0 Bacterial skin infection 27 13% 0 Urinary tract infection 91 43% 0 Pneumonia/pneumonitis 50 23% 0 Senility/Dementia 35 16% 0 Candidiasis 38 18% 0 Device/graft complication 34 16% 0 Acute Resp failure 31 15% 0 Acute Renal failure 73 34% 0
68 Table 4.2. Total sample scale va riable descriptive statistics. Variable N Mean Median Std. Deviation Minimum Max Age 213 70.97 72 10.644 47 85 Total Braden score 213 13.86 14 2.482 8 19 Hosp los 213 11.72 8 10.924 2 110 ICU los 82 7.84 5 8.1 1 44 OR time 35 4.72 4.5 2.525 .5 12.0 Hemoglobin 213 9.9404 10 1.88824 5.90 14.60 Hematocrit 212 30.4981 30.75 5.65602 18.60 45.20 Albumin 204 2.9735 3 .76820 1.10 4.70 Prealbumin 58 9.6724 9 5.91278 3.00 26.00 BMI 211 26.6256 25.5 7.21859 13.70 59.50
69 Table 4-3. Recorded medical diagnoses by chart review. Diagnosis ICD9 Codes Total # subjects with diagnosis (including ICD9, labs, active problem list, and provider notes) Acute Renal Failure 584.9, 586.0 403.91, 404.02 73 Acute Resp. Failure 518.8131 Anemia 280.0, 280.9, 285.21, 285.22, 285.29, 285.9 198 CHF 428.0, 428.1, 428.9 55 CVA 437.8, 438.20, 438.89 27 Device or Graft complications E878, 996.0-996.89, 999.31, 429.4429.9 34 DM with complications 250.00, 250.01, 250.13, 250.30-250.80 91 Gangrene 785.4 7 Malnutrition (moderate or severe by RD) 262, 273.8, 269.9, 262, 263.9 170 mod + severe 39 severe only Candidiasis 111.8, 112.0, 112.2, 112.84, 112.89 38 Osteomyelitis 730.0-730-9, 730.17, 730.27 7 Paralysis 332.0, 344.01, 344.1, 342.91, 342.90, 358.00 16 Pneumonia or Pneumonitis 480.0, 481, 482.0, 482.9, 486 50 Senility or Dementia 290.40, 294.10, 331.0, 331.82, 780.09, 780.97 35 Sepsis 038.10, 038.42, 038.9, 785.52, 995.91, 995.92 40 Bacterial skin infection 681.00, 681.1, 682, 682.2-682.7, 686.9, 998.51 27 UTI 559.0, 597.80, 098.0, 098.2 91
70 Table 4.4. Comparison of mean differences between groups. Variable Mean No PU Mean Yes PU Mean Diff T Sig. (2 tail) p Age 70.5 71.5 1.5 -.642 .522 Hosp los (days) 8.48 15.4 6.92 -4.67 .000 ICU los (days) 5.72 8.87 3.15 -2.1 .039 OR time (hrs) 3.80 5.15 1.35 -.890 .394 Hemoglobin 10.42 9.4 1.02 4.13 .000 Hematocrit 32 28.8 3.2 4.26 .000 Albumin 3.3 2.7 0.6 6.08 .000 Prealbumin 12.2 8.43 3.77 2.09 .021 BMI 27.7 25.5 2.2 2.2 .029 Total Braden 14.6 13 1.6 5.148 .000 Braden sensory 2.86 2.65 0.21 1.874 .062 Braden moisture 3.18 3.10 0.08 .852 .395 Braden activity 2.02 1.31 0.71 6.635 .000 Braden mobility 2.43 2.09 0.34 4.016 .000 Braden nutrition 2.12 2.02 0.10 1.057 .292 Braden friction 2.12 1.79 0.33 4.320 .000
71 Table 4-5. Differences between gro ups for predictor variables. Total sample=213 No PU = 113 Yes PU = 100 Total sample n=x / % sample NO PU n/% this group YES PU n/%this group ChiSquare Statistic Sig. p Race: African American 29/14%14/12%15/15%Caucasian 171/80%87/77%84/84%Other race 10/5%9/8%1/1%8.726 .033 Yes Surgery 36/17%12/11%24/24%6.763 .009 No Surgery 177/83%111/98%76/76%Yes severe nutritional compromise 39/18%8/7%31/31%20.295 .000 No severe nutritional compromise 174/82%105/93%69/69%BMI by category: 9.528 .023 Low BMI <19 24/11%8/7%16/16%Norm 19 to 28 108/58%53/46%55/55%High BMI >28 79/37%50/44%29/29%Yes CVA/paralysis 34/16%9/8%25/25%11.48 .001 No CVA/paralysis 179/84%104/92%75/75%Yes Sepsis 40/14%15/13%25/25%4.782 .029 No Sepsis 173/81%98/86%75/75%Yes UTI 91/43%33/29%58/58%12.479 .000 No UTI 122/57%80/71%42/42%Yes Pneum/pneumonitis 50/23%10/9%40/40%28.657 .000 No pneumonia/pneumonitis 163/77%103/91%60/60%Yes Senility/Dementia 35/16%13/12%22/22%4.256 .039 No Senility/Dementia 178/84%100/88%78/78%Yes Candidiasis 38/18%6/5%32/32%25.784 .000 No Candidiasis 175/82%107/95%68/68%Yes respiratory failure 31/15%8/7%23/23%10.813 .001 No acute resp failure 182/85%105/93%77/77%Yes acute renal failure 73/34%35/31%38/38%1.163 .281 No acute renal failure 140/66%78/69%62/62%
72 Table 4.6. Pressure ulcer descriptions. n= x (% of pressure ulcer group) Number of subjects with PU 100 Number of Stage I PU 5 (5%) Number of Stage II PU 74 (74%) Number of Stage III PU 5 (5%) Number of Stage IV PU 0 Number of DTI PU 13 (13%) Number of Unstageable PU 3 (3%) Location of pressure ulcers: Buttocks 25 (25%) Heel 8 (8%) Hip 1 (1%) Sacrum/coccyx 52 (52%) Multiple locations 5 (5%) Other location (ankle, knee, foot, etc) 9 (9%)
73 Table 4-7. Logistic regression an alysis results of models. Model and Step Variable Name % Total Cases Correctly Classified % No PU Correctly Classified % Yes PU Correctly Classified Model -2 log Liklihood Model Chi-Square Statistic Sig. p 1/1 Total Braden 67.6%69.9%65%269.635 24.852.000 2/1 Only 2 Braden Subscales: 72.3% 65.5% 80% 243.005 51.481 .000 Activity Friction 3/1 Total Braden + 72.3%74.3%70%240.021 54.466.000 3/2 RD Nutrition Eval 4/1 Total Braden 78.9%81.4%76.3%174.761 100.986.000 4/2 + Pneumonia + Candidiasis + Surgery + UTI + CVA/Paralysis 5/1 Total Braden 80.8%85%76%191.890 102.597.000 5/2 +Pneumonia +Candidiasis +Surgery +Severe RD Maln + CVA/Paralysis +UTI 6/1 Total Braden 78.4%82.3%74%201.456 93.031.000 6/2 +Pneumonia +Candidiasis +Surgery +Severe RD Maln 7/1 +Pneumonia 77.0%71.7%83%212.239 82.248.000 +Candidiasis +Surgery +Severe RD Maln Table 4-8. Relative risk of pressure ulcers by predictors in final model. Odds Ratios 95% CISig. p Braden total scores 0.7840.675-0.910 0.001 Pneumonia/ pneumonitis 7.93.4-18.61 0.000 Candidiasis 9.5 3.4-26.4 0.000 Surgery 5.8 2.5-13.8 0.000 Severe Maln by RD 4.98 1.9-12.9 0.001
74 CHAPTER 5 DISCUSSION, RECOMMENDATIONS AND CONCL USIONS The aims of this retrospective descriptive st udy include: 1) to determine the predictability of the Braden Scale total score on the development of pressure ulcers in an inpatient acutely ill adult veteran population, 2) to determine if th e addition of other significant medical factors (diagnosis of gangrene, anemia, diabetes mellitus malnutrition, osteomyelitis, pneumonia/pneumonitis, septicemia, candidiasis, bact erial skin infection, complication of device or implant/graft, urinary tract infection, paralysi s/CVA, senility, respiratory failure, acute renal failure, congestive heart failure, history of previous pressure ulcer, age, race, length of inpatient hospital and ICU stays, time in operating room, a nd smoking status) to these Braden total scores enhance the models predictability of pressure ul cer development in an inpatient acutely ill adult veteran population, and 3) to determ ine if selected medical factors alone are significantly able to determine the development of pressure ulcers in an acutely ill inpatient adult veteran population. The overall conclusion from this analysis was that a logistic regression model of pressure ulcer development in acutely ill veterans indicated 5 predictors able to determine a statistically significant risk of pressure ulcer development. Sp ecifically, the logistical analysis indicated that high risk Braden total scores (mean =13), the presence of a diagnosis of pneumonia/pneumonitis, candidiasis, a severe nutritional compromise as determined by a registered dietician (RD), or surgery during hospitalization can be predictive of the development of pressure ulcers in acutely ill veterans. The analysis included a predictive m odel using binary Logistic Regression, which determined that the Braden total score alone was correctly able to classi fy 68% of the sample (70% subjects which did not develo p pressure ulcers were correctly classified and 65% cases that did develop pressure ulcers were correctly classified in this B raden score only model). Please see Table 4-7. Adding the presence of pneumonia/pneumonitis, candidiasis, severe nutritional
75 compromise and surgery during hospitalization was able to correctly classify an additional 10% of the sample (78% of total sample correctly classified: 82% without PU, 74% with PU). The final five-factor model summary -2 Log likelihood statistic indicated a g ood fit of the model to the data (Table 4-7). This analysis supports the notion that accounting for at least 4 additional medical factors to the Braden total scores will enhance the models predictability of pressure ulcer development in an inpatient acutely ill adult veteran populat ion. The diagnosis of pneumonia/pneumonitis or candidiasis was the highest influential medical factor next to total Braden scores for the model. The odds ratios of the four strongest pressure ulcer risk predictors: pneumonia or pneumonitis odds ratio of 6.9 (95% CI 2.9 16.75, p <.001), candidiasis odds ratio 9.0 (95% CI 3.2-25.7, p .000), severe nutritional compromise odds ratio of 4.98 (CI 1.9-12.9, p .001), and surgery during hospitaliza tion odds ratio of 5.8 (CI 2.5 13.8) were consistent with Fogerty et al. (2008) that reported OR of 9.18 (95%CI 8.81-9.99) for a malnutrition diagnosis, OR of 3.47 and 8.70 (95%CI 3.33-3.61 and 8.33-9.09) for diagnosis of pneumonia or pnuemonitis, respectively; and an OR 4.63 (95%CI 4.41-4.86) for a diagnosis of mycosis. Fogerty et al. did not list their ICD9 codes for mycosis, but the ICD9 code book lists diagnosis codes of 111.9, 117.9, 112.0, 117.9, 112.1, 111.8, and 111.9, most of which are typically classified as candidiasis. So, for the purpose of this study candidiasis (ICD9 codes 111.8, 112.0, 112.2, 112.84 and 112.89) was used as the mycosis predictor variable, and the odds ratio of 9. 0 is twice that of the Fogerty et al. study. Perhaps this is because more true diagnosis of candidiasis were identified by chart review, or perhaps because Fogerty et al. (2008) only used one or two of the related discharge diagnosis codes. The findings of this study are also consistent with those other studies that suggest that the occurrence of surgery is a significant risk fa ctor for pressure ulcer s (Aronovitch, 2007; Uzun & Tan, 2007). There is some evidence that several factors may have a combined effect (evidence
76 by slight changes in odds ratios for some factors with the addition of other factors in the model). For example, diagnosis of pneumonia/pneumonitis odds ratios fluctuated a bit from OR 6.9 to 8.5 with the addition of other factors. Other interesting findings of this study include the Br aden sub-scores which had demonstrated a stronger inverse relationship to pressure ulcer devel opment in correlational examination than the Braden total score. Activity, mobility, and friction sub-scores of the Braden Scale were strongly associated with pr essure ulcers independent of the Braden Total score. The activity sub-score of the Braden demonstrated a stronger association with development of pressure ulcers than the total Braden score (Pearson Correlation -.415, p<.001 versus -.334, p<.001 for the Braden total score), indicating as activity su b-scores decreased, the risk of pressure ulcers in creased. Friction sub-score (Pea rson Correlation -.285, p<.001), and mobility sub-score (Pearson Correlation -.266, p<.001) also demonstrated significant inverse associations with pressure ulcer development but not as strong as the total Braden score or the activity sub-score, and they did not retain their significance in logistic regression models. The LR model that was run with the Braden subs cores of activity and friction was the second strongest model at correctly clas sifying veterans in the sample who did develop pressure ulcers (80% accuracy with PU; 72.3% total sample accuracy; 65.5% accuracy cl assifying the without PU group). This supports findings in another study by Kottner, Halfens, & Dassen (2009) who suggest the Braden subscale items of moisture, sensory percep tion and nutrition contain the largest amount of measurement error (p. 1307). This may be due to the ambiguous and vague descriptions and instructions for use of the Braden tool for pressure ulcer risk measurement (see Appendix A), or may indicate the need for further st aff instruction on the prope r use of the tool.
77 Recommendations for Practice Several striking im plications are suggested from the analysis of data from this study. Most strikingly, the total Braden Score, while still able to correctly classify 68% of the cases in the study, was only able to correctly classify 65% of t hose with pressure ulcers. This implies that the Braden score alone may not be the strongest pred iction model of pressure ulcer risk. Since other sub-scores of the Braden were not significantly associated with the development of pressure ulcers (sensory perception sub-score, nutrition sub-score, and moisture sub-score) in many analyses, it may suggest that there is a deficit in the appropriate scoring of these factors and a need for further education in the best way to quantify risk in th ese categories. These sub-scores, if inappropriately scored may reduce the overall predictive ability of the total Braden score. Activity, friction, and mobility sub-scores of the Braden Scale were all significantly associated with pressure ulcer development on Pearson Corr elations, but not all of these were able to maintain that significance duri ng logistic regression analysis. The activity sub-score of the Braden Scale was the most strongly associated factor with pressure ulcer development in logistic regression models (OR.348, CI .212-.570, p .000), indicat ing a 2.9 times greater risk of pressure ulcer development for every point below 4 on th e activity scale. In fact, it was the overall strongest predictor of pressure ul cers regardless of othe r factors in the models This indicates that nurses who use the Braden Scale may be better able to quickly determine a patients activity level as scored by the Braden Scale because th e scoring mechanism is far more concrete and delineated for Activity (bed bound, chair bound, walks occasionally, walks frequently) than some of the other Braden sub-scores, where sc oring directions are more vague. Moisture, for instance may be difficult to note if the patient is constantly moist versus very moist or occasionally moist versus rarely moist. Pers piration, fever, wound exudates, urinary or fecal incontinence may produce varying de grees of skin moisture that may be difficult to fit to one
78 category. These findings are consistent with Berlowitz et al. (2001) who found mobility (perhaps more a measure of activity as measured by activi ties of daily living in their study) and urinary incontinence to be somewhat difficult to meas ure or quantify but both were still strongly associated with the development of pressure ulcers. These findings imply that better, more consistent education is needed for nurses w ho are conducting pressure ulcer risk assessments (Lyder & Ayello, 2007) and that simpler, more c oncrete and direct methods of risk assessment would be appropriate, with less chance of variance of interpretation. The presence of severe nutritional compromise has been implicated as a pressure ulcer risk factor in numerous studies (Banks, Graves, Bauer, & Ash, 2009; Fogerty et al., 2008; Fisher, Wells, & Harrison, 2004). The nutrition sub-score of the Braden Scale was more weakly correlated to the registered diet icians assessment of nutritional compromise (r=.220) than it was to serum albumin (r=.326), indicatin g that nurses may be relying on inadequate factors to assess nutritional status. Langkamp-Henken, Hudgens, St echmiller & Herrlinger-Garcia (2005) suggest that mini-nutritional assessment (MNA) and screen ing scores are far more accurate than other measures of nutritional status co mmonly used by nurses (such as amount of intake queried using the Braden tool) or laboratory i ndices (such as serum albumin) al one. In this research study, a registered dietician did a full nutritional assessm ent (more extensive than the MNA) of 95% of subjects. The north Florida/south Georgia VA has re gistered dieticians who typically complete a full nutritional assessment with in 48 hours of admission. The resulting classification of a patients nutritional status (no compromise, mild compromise, m oderate compromise, and severe compromise) takes into account anthropometric (BMI, etc), biological (serum albumin, total lymphocyte count, etc), clinical, an d dietary history data and is readily available for nurses to view in the medical record. This was deemed the best assessment of overa ll nutritional status and
79 the classification of severe nutritional compromise was the most strongly associate with pressure ulcer development than any othe r nutritional factor collected in this study. Since a severe nutritional compromise was strongly associated with pressure ulce r development, it is suggested that MNA or RD nutritional assessments be utilized for determining a more accurate nutritional risk of pressure ulcer development rather th an nutrition subscales of the Braden tool. Recommendations for Future Research Lyder (2003) reported that over 100 pressure ulcer risk factor s have been reported in the literature. T his study examined only several of what was determined to be the strongest predictors of pressure ulcers identified in the liter ature. More research is needed to validate the findings of this study in a la rger population (examining medi cal risk factors of severe malnutrition, pneumonia/pneumonitis, candidiasis, surgery and the development of pressure ulcers). In addition, more research is needed to examine factors that were significantly associated with pressure ulcer risk in this study and consis tent with other research studies (such as UTI, history of previous pressure ulcers, sepsis, race, anemia), but were not able to maintain their significance during LR in this study (possibly due to sample size). Similarly, age and smoking status have been identified as pres sure ulcer risk factor s in the literature but were not found to be significant predictors in this st udy. Reasons for this need to be explored further but may include a somewhat narrow age range of the veteran population and the large number of smokers in younger persons. More research is also needed in developing a risk assessment tool that is not vague in its components and does not allow room for variance in user interpretation. A dynamic medical risk factor determinant tool or check of f list where factors are queried with yes or no only questions may be better suited for ti mely pressure ulcer risk assessment and reassessment. In addition, it would be beneficial if the tool descri bed appropriate interventions for each yes factor. For example, if the tool queried, Does the patient experience urinary
80 incontinence? and the answer was yes, the tool would direct the use of a barrier cream. This hypothetical tool needs to be dynamic in so far as it needs to be modifiable and reassessed on a regular basis (perhaps every 3 to 5 years) with the top 4 to 5 medical factors present in the literature replacing factors queried in the tool (if necessary). Questions in the tool with yes/no answers should be based on these most significant risk factors identified in the literature. Questions could be simply phrased such as, H as the person been identified with candidiasis, pneumonia/pneumonitis, or surgery during this ad mission? or, Does a MNA or RD nutritional assessment show severe nutritional compromis e? or, Is the person ab le to lift all four extremities off of the bed unaided and hold it up for a count of ten? or, Is the person bedbound or chair bound except for transfers? This tool should be developed and tested for validity and reliability (perhaps a Cowan pressure ul cer risk assessment questionnaire?). The next logical research step suggested by the results of this study is to perform a larger retrospective review with at leas t 500 subjects in a different vete ran population collecting data on the 4 most significant medical factors as well as those that were not able to sustain statistical significance through all steps of the LR or had to o many missing cells (UTI, history of previous pressure ulcer, recent history of weight loss, CVA/paralysis) or had t oo few subjects in each category (race, sepsis, gangrene, etc). If the final four factor only pressure ulcer predictive model (pneumonia/pneumonitis, candidiasis, surgery, severe nutritional compromise) can be validated through other larger studies, it coul d represent significan t clinical improvement in pressure ulcer risk assessment (quicker, more concrete, less ro om for provider interpre tation, increased interrater reliability). Limitations One crucial discovery during early data colle c tion for this study was that most actual medical diagnoses are not adequately captured in the discharge diagnoses ICD9 list for billing
81 and recording purposes in the patients medical record (see Table 4-6). Typically only 4 to 10 discharge diagnoses were recorded on the discharge summary. Often, important diagnoses that were present during the hospitalization were not recorded on discharge summary. This necessitated a revised data collect ion data tool early in the data collection to capture medical diagnoses by provider notes/specialist assessment, and diagnostic results/r adiologist report. The revised data collection tool was resubmitted to the IRB and approved and used for all data collection (see appendix B). This ha s enormous implications for all data analysis where medical diagnoses are reviewed and obtained from either admission or discharge data or coded databases. There is a very real threat to the validity of findings if the co rrect diagnoses are not captured and/or under reported. Fogerty et al. (2008) had an enormous sample (six million persons) in their predictive model regarding medical diagnos es and pressure ulcer development. However, they relied solely on ICD9 diagnosis codes list ed in the patients discharge medical record (National Inpatient Sample). If many actual di agnoses that patients experience during acute inpatient hospitalizations are not captured or co ded corrected, this poses a significant limitation to the interpretation of their data analysis. Another potential limitation of this study is the sample size of 213. The statistical power analysis done prior to data collection determined that the sample was adequate to achieve 80% power if the risk factor was present in at least 20 % of the population and the odds ratio was at least 3.5. This may have limited the predictive ability of the factors in the model that potentially had an odds ratio of less than 3.5. A larger sample could potentially demonstrate more factors in this study that would significantly add to the predictive model. In addition, the missing data in so many subjects with regard to hist ory of previous pressure ulcer and history of recent weight loss
82 eliminated them from the analysis, and they ma y be important factors to consider in future studies. Conclusions In conclusion, this study does dem onstrate suppo rt for the idea suggested by Fogerty et al. (2008) that medical factors such as surgery and malnutrition, and diagnoses such as pneumonia/pneumonitis and candidiasis present in the patient during hospitalization can enhance (if not surpass) the Braden Scale in a pressure ulcer predictive model (see Table 4-8). Findings from this study suggest that identifying patients with severe nutritional co mpromise, the presence of pneumonia/pneumonitis and/or candidiasis, and the event of surgery during hospitalization may be better able to identify veterans at high risk of pressure ulcers than current Braden risk assessment scores alone. In all LR models run wi th the Braden scale and the addition of other significant factors within this study, the predicti ve accuracy of the Braden scale was improved (improved 4 to 7 % for total model accuracy, improved up to 13% for accuracy in predicting the NO PU group, and improved accuracy up to 18% in pr edicting PU in the YES PU group) with the addition of other medical factors such as diagnosis of pneumonia/pneumonitis. More research is needed to validate these findi ng and to explore relevancy of cu rrent risk assessment techniques as well as provide direction for interventional stud ies (for example: which are the most effective pressure ulcer prevention interventions?), with an ultimate goal of reducing incidence of pressure ulcers.
83 APPENDIX A BRADEN SCALE FOR PREDICTI NG PRESS URE SORE RISK
85 APPENDIX B COWAN DATA COLLECTION TOOL Please DO NOT include any private information on the patient, such as name or identification number. Age in years___________(all ages over 85 will just be recorded as "85 or older"). Developed Pressure ulcer during admission? (circle one) : DID / or DID NOT Worst Stage: Stage I Stage II Stag e III Stage IV DTI Unstageable Location: sacrum-coccyx / heel / buttocks / hip / ischium / other Gender (circle) Male / Female / not listed Ethni city (circle) African American / Caucasian / Other Total Lowest Braden Score during admission (prior to pressure ulcers): __________ Subscale scores: Sensory 1 / 2 / 3 / 4 Moisture: 1 / 2 / 3 / 4 Wound Consult done? Y / N Activity: 1 / 2 / 3 / 4 Mobility: 1 / 2 / 3 / 4 Specialty mattress ordered? Y/N Nutrition: 1 / 2 / 3 / 4 Friction: 1 / 2 / 3 Nutrition supplem ordered? Y/N Length of inpatient admission in days: _________ Length of ICU stay in days: __________ Have surgery during current admission? Y / N Time in operating room in hours: __________ Smoking (circle one) Not Current Smoker / Currently Smokes / Not listed Presence of following diagnosis codes during admission or at discharge? Gangrene (785.4): YES / NO / APL (active problem list) / PN (provider notes) / LB (labs/radiology) Anemia (280.0 and related codes) YES / NO / APL / PN / LB Actual lowest Hgb during admission _____ Actual lowest HCT during admission ______ Diabetes mellitus w/ complications (250.1-250.9) YES / NO / APL / PN / LB Malnutrition (260, 261, 262, 263.0-263.9, 995.85); YES / NO / APL / PN / LB BMI: <19 /19-28/ >28 Lowest albumin _____ Lowest Pre-albumin _____ Nutrition Consult: Y / N Compromised?: None / Mild / Mod / Severe Osteomyelitis (730.0-730.9); YES / NO / APL / PN / LB Pneumonia/pneumonitis (112.4, 480.0, 486, and relate d codes); YES / NO / APL / PN / LB Septicemia (038.1, 998.59 and related codes); YES / NO / APL / PN / LB Candidiasis (111.8, 112.2 and related codes) YES / NO / APL / PN / LB Bacterial Bacterial skin infection (686.9, 998.51 and related codes); YES / NO / APL / PN / LB Complication of device or implant/graft (996.0-996.89, 429.4-429.9); YES / NO / APL / PN / LB Urinary tract infection (098.0, 098.2, 131.00, 559.0, 597.80); YES / NO / APL / PN / LB Paralysis (045.0, 300.11, 332.0 and related code s); YES / NO / APL / PN / LB Senility (259.8, 290.10 and related codes); YES / NO / APL / PN / LB Respiratory failure (348.8, 518.81-581.84), YES / NO / APL / PN / LB Acute renal failure (403.91, 404.02 and related codes), YES / NO / APL / PN / LB Cerebrovascular accident (437.8), YES / NO / APL / PN / LB Congestive heart failure (428.0, 428.1, 428.9) YES / NO / APL / PN / LB History of previous pressure ulcer prior to current admission (707.0.9) YES / NO / Unknown Actual diagnosis codes listed at discharge: _______________________________________________
86 APPENDIX C PRESSURE ULCERS IN ACUTELY ILL VETERANS A PRELIMINARY MODEL Total Lowest Braden Score (6-23) Time in O.R. in hours Length of ICU stay in days Diagnosis Present (Y / N): Gangrene Paralysis Septicemia Osteomyelitis Malnutrition (severe) Pneumonitis / pneumonitis Urinary tract infection Bacterial infection / Bacterial skin infection Senility/Dementia Candidiasis Acute Respiratory failure Acute Renal Failure Cerebrovascular accident Diabetes mellitus w/ complications Congestive Heart Failure Anemia Device or Graft complication History of previous pressure ulcer (Y / N) Subscales: Sensory 0-4 Activity 0-4 Moisture 0-4 Mobility 0-4 Nutrition 0-4 Friction 0-3 DID NOT develop pressure ulcer during admission vs. DID develop pressure ulcer during admission Stage I Stage II Stage III Stage IV DTI Unstageable Ethnicity: A/C/O Current Smoking Status: Yes / No / Not listed Age in years Length of hospital stay in days
87 APPENDIX D PRESSURE ULCERS IN ACUTELY ILL VETERANS A FINAL MODEL Severe nutritional compromise DID NOT develop pressure ulcer during admission vs. DID develop pressure ulcer during admission Pneumonia/pneumonitis Candidiasis Event of Surgery during hospitalization Total Braden Scores Activity & Friction
88 APPENDIX E DIFFERENCES IN CAPTURING DIAGNOSES Diagnosis ICD9 Codes Number of subjects with this ICD9 code listed at discharge Number of subjects with diagnosis evident during hospitalization from chart review (including ICD9 codes, labs, active problem list, and provider notes) Acute Renal Failure 584.9, 586.0 403.91, 404.02 No PU: 28 Yes PU: 30 No PU: 35 Yes PU: 38 Acute Resp. Failure 518.81 No PU: 7 Yes PU: 20 No PU: 8 Yes PU: 23 Anemia 280.0, 280.9, 285.21, 285.22, 285.29, 285.9 No PU: 28 Yes PU: 20 No PU: 102 Yes PU: 96 CHF 428.0, 428.1, 428.9 No PU: 22 Yes PU: 19 No PU: 28 Yes PU: 27 CVA 437.8, 438.20, 438.89 No PU: 1 Yes PU: 2 No PU: 6 Yes PU: 21 Device/Graft complications E878, 996.0-996.89, 999.31, 429.4-429.9 No PU: 14 Yes PU: 14 No PU: 16 Yes PU: 18 DM with complications 250.00, 250.01, 250.13, 250.30250.80 No PU: 14 Yes PU: 8 No PU: 50 Yes PU: 41 Gangrene 785.4 No PU: 3 Yes PU: 0 No PU: 4 Yes PU: 3 Malnutrition (Severe) 262, 273.8, 269.9, 262, 263.9 No PU: 2 Yes PU: 2 No PU: 8 Yes PU: 31 Candidiasis 031.9, 111.8, 112.0, 112.2, 112.84, 112.89 No PU: 4 Yes PU: 9 No PU: 6 Yes PU: 32 Osteomyelitis 730.0-730-9, 730.17, 730.27 No PU: 3 Yes PU: 3 No PU: 3 Yes PU: 4 Paralysis 332.0, 344.01, 344.1, 342.91, 342.90, 358.00 No PU: 2 Yes PU: 5 No PU: 4 Yes PU: 12 Pneumonia or Pneumonitis 480.0, 481, 482.0, 482.9, 486 No PU: 7 Yes PU: 32 No PU: 10 Yes PU: 40 Senility or Dementia 290.40, 294.10, 331.0, 331.82, 780.09, 780.97 No PU: 9 Yes PU: 11 No PU: 13 Yes PU: 22 Sepsis 038.10, 038.42, 038.9, 785.52, 995.91, 995.92 No PU: 11 Yes PU: 12 No PU: 15 Yes PU: 25 Bacterial skin infection 681.00, 681.1, 682, 682.2-682.7, 686.9, 998.51 No PU: 11 Yes PU: 10 No PU: 13 Yes PU: 14 UTI 559.0, 597.80, 098.0, 098.2, 112.2 No PU: 25 Yes PU: 37 No PU: 33 Yes PU: 58
89 LIST OF REFERENCES Agency for Health Care Policy and Research (1992). Pressure Ulcers in Adults: Prediction and Prevention. Clinical Practice Guideline Num ber 3. AHCPR pub. No. 92-0047. Retrieved August 8, 2008 from: www.ncbi.nih.gov Allm an R. (1997). Pressure ulcer prevalence, in cidence, risk factors, and impact. Clinical Geriatric Medicine, 13, 421-436. American Medical Directors Asso ciation. (2008). Pressure ulcers in the long-term care setting. Retrieved August 8, 2008 from: www.guidelines.gov Anthony, D., Reynolds, T., & Russell, L. (2000). An investigation into the use of serum album in in pressure sore prediction. Journal of Advanced Nursing, 32 (2), 359-365. Armstrong, D., Ayello, E., Capitulo, K., Fowler, E., Krasner, D., Levine J., Sibbald, R.G., & Smith, A. (2008). Opportunities to Improve Pressure Ulcer Prevention and Treatment: Implications of the CMS Acute Care Present on Admission (POA) Indicators/HospitalAcquired Conditions (HAC) Ruling. International Expert Wound Care Advisory Panel. Roundtable discussion held Fe bruary 2, 2008 in Chicago, IL. Aronovitch, S. (2007). Intraoperative ly acquired pressure ulcers: ar e there common risk factors? Ostomy Wound Management, 53 (2), 57-69. Ayello, E. (2007). Predicting pressure ulcer risk. American Journal of Nursing, 107(11), 45-47. Ayello, E., Baranoski, S., & Salati, D. (2005) A survey of nurses wound care knowledge. Advances in Skin & Wound Care, 18(5), 268-278. Ayello, E., & Braden, B. (2002). How and w hy to do pressure ulcer risk assessment. Advanced Wound Care, 15 (3), 125-31. Ayello, E. & Lyder, C. (2001). Pressure ulcers in persons of color, race, and ethnicity. In J. Cuddinton (Ed.), Pressure Ulcers in America: Prevale nce, Incidence, and Implications for the Future. Washington: National Pressure Ulcer Advisory Panel Ayello, E. & Lyder, C. (2007). Protecting patients from harm: preventing pressure ulcers in hospital patients. Nursing, (37) 10, 36-40. Ayello, E. & Sibbald, G. (2008). Preventing pressure ulcers and skin tears. Evidence Based Geriatric Nursing Protocols for Best Practice, Hartford Institute for Geriatric Nursing. Retrieved January 1, 2009 from: http://www.guidelines.gov/sear ch/searchresults.aspx? Type=3 &txtSearch=pressure+ulcers &num=20
90 Banks, M., Bauer, J., Graves, N. & Ash, S. (2009) Malnutrition and pressure ulcer risk in adults in Australian health care facilities. Clinical and Sports Nutrition. Retrieved March 12, 2009 from: http://www.sciencedirect.com/scienc e? _ob=ArticleURL&_udi=B6TB0-4XY4SHN8&_user=10&_coverDate=12%2F16%2F2009& _rdoc=1&_fmt=high&_orig=search&_s ort=d&_docanchor=&view=c&_acct=C 000050221&_version=1&_urlVersion=0&_useri d=10&md5=f00faf881f195e6a61e80de873788700 Baranoski, S. (2006). Raising awareness of pressure ulcer prev ention and treatment. Advances in Skin & Wound Care, 19 (7), 398-404. Bergstrom, N., Braden, B., Kemp, M., Champa gne, M., & Ruby, E. (1998). Predicting pressure ulcer risk: A multisite study of the predic tive validity of the Braden Scale. Nursing Research, 47 (5). 261-269. Bergstrom, N., Braden, B., Laguzza, A., & Holman, V. (1987). The Braden Scale for predicting pressure sore risk. Nursing Research, 36 (4), 205-210. Bergstrom, N., Demuth, P., & Braden, B. (1987). A clinical trial of the Braden Scale for Predicting Pressure Sore Risk. Nursing Clinics of North America, 22 (2), 417-428. Berlowitz, D., Anderson, J., Brandeis, G., Lehner, L, Brand, H., Ash, A., & Moskowitz, M. (1999). Pressure ulcer development in th e VA: Characteristics of nursing homes providing best care. American Journal of Medical Quality, 14 (1), 39-44. Berlowitz, D., Brandeis, G., Anderson, J., & Bra nd, H. (1997). Predictors of pressure ulcers healing among long-te rm care residents. Journal of the American Geriatrics Society, 45(1), 30-34. Berlowitz, D., Brandeis, G., Morris, J., Ash, A., Anderson, J., Kader, B., & Moskowitz, M. (2001). Deriving a risk-adjustment model fo r pressure ulcer development using the minimum data set. American Geriatrics Society, 49 (7), 866-871. Berlowitz, D., & Brienza, D. (2007). Are all pre ssure ulcers the result of deep tissue injury? A review of the literature. Ostomy Wound Management, 53 (10), 34-38. Black, J., Baharestani, M., Cuddigan, J., Dorner B., Edsberg, L., & Langemo, D. et al. (2007). National Pressure Ulcer Advi sory Panels updated pressure ulcer staging system. Dermatological Nursing, 19(4), 343-349. Bolton, L. (2007). Which pressure ulcer risk assessm ent scales are valid for use in the clinical setting? Wound Ostomy Continence Nurse, 34(4), 368-381. Braden, B., & Bergstrom, N. (1987). A conceptual schema for for the study of the etiology of pressure sores. Rehabilitation Nursing, 12 (1), 8-12. Braden, B., & Bergstrom, N. (1994). Predictive va lidity of the Braden Scale for pressure sore risk in a nursing home population. Research Nursing Health, 17 459-70.
91 Braden Scale for Predicting Pressure Sore Risk. 1987. Retrieved January 29, 2010 from: http://www.bradenscale.com Brown, G. (2003). Long-term outcomes of full thic kness pressure ulcers: Healing and mortality. Ostomy Wound Management, 49 (10), 42-50. Brown, S. (2004). The Braden Scale: A review of the research evidence. Orthopedic Nursing, 23(1), 30-38. Bryant, R. & Nix, D. (2007). Acute & Chronic Wounds: Curre nt Management Concepts (3rd ed.). St. Louis: Mosby. Centers for Disease Control and Prevention (1997). Ambulatory Su rgery in the United States, 1995. Retrieved August 11, 2008 from: http://www.cdc.gov/nchs/data/ad/ad296.pdf Defloor, T. (1999). The risk of pressure sores: a c onceptual schem e. Journal of Clinical Nursing, 8, 206-216. Defloor, T., & Grypdonck, M. (2005). Pressure ulcers: Validation of two risk assessment scales. Journal of Clinical Nursing, 14, 373-382. Dinsdale, S. (1974). Decubitus ulcers: role of pressure and friction in causation. Archives of Physical Medicine and Rehabilitation, 55, 147-152. Duncan, K. (2007). Preventing pressure ulcers: The goal is zero. The Joint Commision Journal on Quality and Patient Safety, 33 (10), 605-610. de Souza, D., & Santos, V. (2007). Risk f actors for pressure ulcer development in institutionalized elderly. Rev Lat Am Enfermagem, 15(5), 958-964. Edelman, D., Weiss, T., Ashton, C., & Wray, N. (1995). Outpatient Durgery Utilization in Veterans Affairs Hospitals 1981-1989. MEDICAL CARE, 33(3). 246-255. European Pressure Ulcer Advisory Panel a nd National Pressure Ulcer Advisory Panel. Prevention and treatment of pressure ulcers: quick reference guide. Washington DC: National Pressure Ulce r Advisory Panel; 2009. Field, A. (2005). Discovering Statistics Using SPSS (Second Ed.). Thousand Oaks, Sage Publications. Fisher, A., Wells, G., & Harrison, M. (2004). Factors associated with pressure ulcers in acute care hospitals. Advances in Skin & Wound Care, 17 (2), 80-90. Fogerty, M., Abumrad, N., Nanney, L., Arbogast, P ., Poulose, B., & Barbul, A. (2008). Risk factors for pressure ulcers in acute care hospitals. Wound Repair and Regeneration, 16, 11-18.
92 Garber, S. & Rintala, D. (2003) Pressure ulcers in veterans with spinal cord injury: A retrospective study. Journal of Rehabilitative Research and Development, 40 (5), 433442. Gefen, A. (2008). How much time does it take to ge t a pressure ulcer? Integrated evidence from human, animal, and In Vitro studies. Ostomy Wound Management, 54 (10), 26-35. Gordis, L. (2004). Epidemiology (3rd ed. pp. 172) Philadelphia: Elsevier Inc. Graham, N. (1995). Quality in Health Care: Theory, Application, and Evolution. (pp 190-195). Gaithersburg: Aspen Publishers, Inc. Graves, N., Birrell, F., & Whitby, M. (2005). Effect of pressure ulcers on length of hospital stay. Infection Control Hospital Epidemiology, 26, 293-297 Hickey, E., Young, G., Parker, V., Czarnowski, E., Saliba, D., & Berlowitz, D. (2005). The effects of changes in nursing home staffing on pressure ulcer rates. Journal of American Medical DirectorsAssociation, 6, 50-53. Hall, M., Doran, D., & Pink, G. (2004). Nurse sta ffing models, nursing hours and patient safety outcomes. Journal of Nursing Administration 34 (1), 41-45. Ho, C. & Bogie, K. (2007). Integrat ing wound care into cl inical practice. Ostomy Wound Management, 53(10), 18-21. Howe, L. (2008). Education and empowerment of the nursing assistan t: Validating their important role in skin care and pressure ulcer prevention, and demonstrating productivity enhancement and cost savings. Advances in Skin and Wound Care, 21, 275-281. International Expert Wound Care Advisory Panel (February, 2008). Roundtable discussion: Opportunities to Improve Pressure Ulcer Prevention and Treatment: Implications of the CMS Acute Care Present on Admission (POA) Indicators/Hospital-Acquired Conditions (HAC) Ruling. Joint Commission for Accr editation of Hea lthcare Organizations (JCAHO). Joint Commission Perspectives on Patient Safety (2008): Strategies for pr eventing pressure ulcers. Retrieved January 14, 2010 from: http://www.jcrinc.com/Pressu re-Ulcers-stage-III-IVdecubitis-u lcers/ Jalali, R. & Rezaie, M. (2005). Predicting pressure ulcer risk: Comparing the predictive validity of 4 scales. Advances in Skin & Wound Care, 18 (2), 92-97. John Hopkins Geriatric Summer Scholars Program Required R eading Section 8, Chapter 30: Pressure Ulcers. Lyder, C. (2000) Retrieved November 11, 2008 from: http://www.hopkinsmedicine.org/geriatrics/e ducation/MedStudents/GeriatricsSumm erSc holarsProgram/Section8.pdf Kirsner, R. & Marcus, A. (2008). Spotlighting the top-10 research findings in wound care. Skin & Aging, 26-27.
93 Kottner, J., & Dassen, T. (2007). Interpreting interra ter reliability coefficients of the Braden scale: A discussion paper. International Journal of Nursing Studies, 45, 1238-1246. Kottner, J., Halfens, R., & Dassen, T. (2009). An in terrater reliabili ty study of the assessment of pressure ulcer risk using the Braden scale and the classification of pressure ulcers in the home care setting. International Journal of Nursing Studies, 46, 1307-1312. Kosiak, M. (1961). Etiol ogy of decubitus ulcers. Archives of Physical Medicine and Rehabilitation, 42, 19-29. Lake, E., & Friese, C. (2006). Variations in nursing practice environments : relation to staffing and hospital characteristics. Nursing Research, 55 (1), 1-9. Langkamp-Henken, B., Hudgens, J., Stechmiller, J., & Herrlinger-Garcia, K. (2005). Mini nutritional assessment and screening scores ar e associated with nutritional indicators in elderly people with pressure ulcers. Jpurnal of the American Dietetic Association, 105(10), 1590-1596. Lowery, J., Hiller, L., Davis, J, & Shore, C. ( 1998). Comparison of professional judgment versus an algorithm for nutritional status classification. Medical Care, 36 (11), 1578-1588. Lyder, C. (2002). Pressure ulcer prevention and management. Annual Review Nursing Research, 20, 35-61. Lyder, C. (2003). Pressure ulcer prevention and management. JAMA, 289 (2), 223-226. Lyder, C. (2006). Assessing risk and preventing pressu re ulcers in pati ents with cancer. Seminars in Oncology Nursing, 2 (3), 178-184. Magnan, M., & Maklebust, J. (2008 ). The effect of web-based Braden Scale training on the reliability and precision of Braden Scale pressure ulcer risk assessments. Journal of Wound Ostomy and Continence Nursing, 35(2), 199-208. Maklebust, J. (2005). Pressure ulcers: The great insult. Nursing Clinics of North America, 40 365-389. Martin, J.W., Mousa, S.S., Shaker, O., & Mousa, S.A. (2009). The multiple faces of nicotine and its implications in tissue and wound repair. Experimental Dermatology, 18 (6), 497-505. McInerney, J. (2008). Reducing hospital-aquired pressure ulcer prevalence through a focused prevention program. Advances in Skin & Wound Care, 21 (2), 75-78. Meaume, S., & Faucher, N. (2007). Heel pre ssure ulcers on the increase? Epidemiological change or ineffective prevention strategies? Journal of Tissue Viability, 17, 30-33. Monfrecola, g., Riccio, G., Savarese, C., Posteraro, G., & Procaccini, E.M. (1998). The acute effects of smoking on cutaneous microcircula tion blood flow in habitual smokers and nonsmokers. Dermatology, 197(2), 115-118.
94 National Pressure Ulcer Adviso ry Panel (2001). Pressure Ulce r Prevention: RN Competencybased Curriculum. Retrieved March 28, 2005 from: www.npuap.org National P ressure Ulcer Advisory Panel (NPUAP). Press Release: Pressure Ulcer Stages Revised by NPUAP. Washington, DC: February, 2007 Nightingale, F. (1860/1976). Notes on Nursing: What it is, and What it is Not New York: Buccaneer Books, Inc. Norton, D. (1996). Calculating the risk : Reflections on the Norton Scale. Advances in Wound Care, 9(6), 38-43. Padubidri, A.N., Yetman, R., Browne, E., Lucas, A., Papay, F., Larive, B., & Zins, J. (2001). Complications of postmastectomy breast rec onstructions in smoke rs, ex-smokers, and nonsmokers. Plastic Reconstructive Surgery, 107 (2), 342-351. Padula, C.A., Osborne, E., & Williams, J., (2008). Prevention and early de tection of pressure ulcers in hospitalized patients. Journal of Wound Ostomy and Continence Nursing, 35(1), 65-78. Papanikolaou, P., Lyne, P., & Anthony, D. (2007). Risk assessment scales for pressure ulcers: A methodological review. International Journal of Nursing Studies, 44, 285-296. Pancorbo-Hildago, P., Garcia-Fernandez, F., Lop ez-Medina, I., & AlvarezNieto, C. (2006). Risk assessment scales for pressure ulcer prevention: a systematic review. Journal of Advanced Nursing, 54(1), 94-110. Perneger, T., Heliot, C., Rae, A., Borst, F., & Gaspoz, J. (1998). Hospital-acquired pressure ulcers. Archives of Internal Medicine, 158, 1940-1945. Pieper, B. (2007). Mechanical forces: Pressure, shear, and friction. In R.A. Bryant & D.P. Nix (Eds.), Acute & Chronic Wounds: Current Management Concepts (3rd ed.,pp.205-230). St. Louis: Mosby. Reddy, M., Gill, S., & Rochon, P. (2006). Preventing pressure ulcers: A systematic review. JAMA, 296 (8), 974-984. Redelings et al (2005). Pressure ul cers: More lethal than we thought? Advances in Skin and Wound Care, 18 (7). 367-372. Registered Nurses Association of Ontario (2005). Risk assessment & prevention of pressure ulcers. Retrieved August 8, 2008 from: www.guidelines.gov Schoonhoven, L., Grobbee, D., Donders, A., Algra, A., Grypdonck, M., & Bouse ma, M. et al. (2006). Prediction of pressure ulcer developmen t in hospitalized patients: a tool for risk assessment. Quality and Safety in Health Care,15(1), 65-70.
95 Scott, R., Gibran, N., Engrav, L., Mack, C., & Rivera, F. (2006). Incidence and characteristics of hospitalized patients with pressure ul cers: Sate of Washington, 1987 to 2000. Journal of Plastic and Reconstructive Surgery 630-634. Smith, B., Guihan, M., LaVela, S., & Garber, S. (2008). Factors predicting pressure ulcers in veterans with spinal cord injuries. American Journal of Physical Medicine & Rehabilitation, 87 (9), 750-757. Srensen, L.T., Zillmer, R., gren, M., Ladelun d, S., Karlsmark, T., & Gottrup, F. (2009). Effect of smoking, abstention, and nicotine patch on ep idermal healing and collagenase in skin transudate. Wound Repair and Regeneration, 17 (3), 347-353. Spilsbury, K., Nelson, A., Cullum, N., Iglesias, C. Nixon, J. & Mason, S. (2007). Pressure ulcers and their treatment and effects on quality of life: hospital inpatient perspectives. JAN:Original Research, 494-505. Stechmiller, J., Cowan, L., Whitney, J., Phillips, L., Aslam, R., Barbul, A., Gottrup, F., Gould, L., Robson, M., Rodeheaver, G., Thomas, D., & Stotts, N. (2008). Guidelines for prevention of pressure ulcers. Wound Repair and Regeneration, 16 151-168. Stotts, N., & Gunningberg, L. (2007). Predicting pressure ulcers risk. American Journal of Nursing, 107(11), 40-48. Stotts, N. & Wu, H. (2007). Hospital Recovery is facilitated by prevention of pressure ulcers in older adults. Critical Care Nursing Clinics of North America, 19, 269-275. Theaker, C., Mannan, M., Ives, N., & Soni, N. (2000). Risk Factors for pressure sores in the critically ill. Anesthesia, 55 (3), 221-224. Thomas, D. (2001). Are all pr essure ulcers avoidable? Journal of American Medical Directors Association, 2, 297-301. US Census Bureau. Data from the 2006 American Community Survey. Retrieved November 11, 2008 from: http://factfinder.census.gov/serv let/S TTable?_bm=y&-geo_id=01000US&qr_name=ACS_2006_EST_G00_S2101& -ds_name=ACS_2006_EST_G00_ Uzun, O. & Tan, M. (2007). A prospective, descriptive pressure ulcer risk factor and prevalence study at a university hospital in Turkey. Ostomy Wound Management, 53 (2), 44-56. VanGilder, C., MacFarlane, G., & Meyer, S. (2008). Results of nine international pressure ulcers prevalence surveys: 1989 to 2005. Ostomy Wound Management, 54 (2), 40-54. van Rijswijk, L. (2008). Has the pendulum reached the sun? Ostomy Wound Management, 54(10), 6. Vincent, J., Fink, M., Marini, J ., Pinsky, M., Sibbald, W., Singer, M., Suter, P., Cook, D. Pepe, P, & Evans, T. (2006). Intensive care and em ergency medicine: Progress over the past 25 years. Chest, 129, 1061-1067.
96 Weiner, M. G., Lyman, J., Murphy, S., & Weiner, M. (2007). Electronic health records: Highquality electronic data for higher-quality clinical research. Informatics in Primary Care, 15, 121-127. Wound Ostomy & Continence Nurses Society (2004). Guideline for the prevention and management of pressure ulcers. Retrieved August 8, 2008 from: www.guidelines.gov Whittington, K. & Briones, R. (2004). Nationa l prevalence and Incidence study: 6-yesr sequential acute care data. Advances in Skin & Wound Care, 17 (9), 490-494.
97 BIOGRAPHICAL SKETCH Linda Cowan was born in Florida. E xcept for a fe w years living in Malaysia as a child with her parents, she spent a great deal of her life in Miami, Florida. She attended Jackson Memorial Hospital School of Nursing as well as Miami-Dade Community College in the late 1970s and early 1980s. She was employed as a registered nurse for over 20 years before going back to graduate school in 2001. Linda completed her Ma sters degree in Nursing in 2004 and began the doctoral program at the University of Florida in 2005. She majored in Nursing Sciences with a minor in Epidemiology, and she completed a Public Health Certificate in April, 2008 through the graduate program at the UF College of Public Health & Health Professions. She is currently employed full time at the North Florida/South Georgia Veterans Health Administration (VA) as a Wound and Ostomy Consultant and Certified W ound Specialist (CWS). She is also licensed by the state of Florida as an Advanced Registered Nurse Practitioner with a board certification in family practice. She currently resides in the North Florida area with her husband and children (the youngest of which also attends college). She is a member of the Southern Nursing Research Society (SNRS), the American Public Hea lth Association (APHA), Sigma Theta Tau, Gerontological Society of Amer ica (GSA), the Wound, Ostomy and Continence Nurses Society (WOCN), and the Wound Heali ng Society (WHS), where she serves on the Education Committee. She has participated in several res earch studies involving wound care over the past 7 years. She graduates from the University of Flor ida with her Doctor of Philosophy in Nursing in May, 2010. Her career goals include continuing re search, teaching, scientific publication, and assisting to establish a wound and ostomy profe ssional multidisciplinary training center at the VA.