Effects of facility variation on the outcomes of stroke patients in the Veterans Administration's (VA) healthcare system

MISSING IMAGE

Material Information

Title:
Effects of facility variation on the outcomes of stroke patients in the Veterans Administration's (VA) healthcare system
Physical Description:
iv, 148 leaves : ill. ; 29 cm.
Language:
English
Creator:
Deane, Jared P
Publication Date:

Subjects

Subjects / Keywords:
Health Services Research thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Health Services Research -- UF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 2004.
Bibliography:
Includes bibliographical references.
Statement of Responsibility:
by Jared P. Deane.
General Note:
Printout.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 003103823
System ID:
AA00011795:00001


This item is only available as the following downloads:


Full Text










EFFECTS OF FACILITY VARIATION
ON THE OUTCOMES OF STROKE PATIENTS
IN THE VETERANS ADMINISTRATION'S (VA) HEALTHCARE SYSTEM













By

JARED P. DEANE


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

































Copyright 2004

by

Jared P. Deane
































This document is dedicated Thomas & Bemice Deane and James & Margaret Patillo (my
grandparents) for laying such a strong foundation on which I can stand tall and be proud.















ACKNOWLEDGMENTS

First and foremost, I would like to thank my Lord and Savior Jesus Christ for

blessing with the opportunity to pursue my Ph.D. He has provided me with the strength,

courage, and power to overcome what seemed to be an insurmountable task at times. He

will continue to receive all of my praise and honor.

I would like to thank the following people who have supported and endorsed me

throughout this learning and growing experience. I want to thank all the members of my

dissertation committee. Christopher Johnson, my chair and advisor, provided me with

the skills necessary to become a good health services organizational researcher. He

introduced me to the world of organization theory and showed me how interesting it can

be. He also let me share his office when there was no space in the trailer. Finally, he

provided funding which allowed me to finish my dissertation. I have admired Christy

Lemak's style of teaching since we both arrived at UF in 1998. I thank her for her

continued support of me. I appreciate those times that she made a special effort to meet

with me. Even though she was swamped with work, she never made me feel that I was

intruding on her time. Finally, I thank her for her expert review and comments on

anything I asked her to look over. I am forever indebted to you. I think my level of

writing has definitely increased since I began working with you. Keep drawing those

circles and boxes!!!!! I appreciate Dr. Tosi's assistance. Chris may have introduced me

to organization theory, but Dr. Tosi took my interest to the next level. I appreciate all the

informative readings that I received from his class. I think I will be using them for many









years after my time is done here. Until I met Pam Duncan, I thought Tarheels and

Wahoos were not supposed to coincide. She introduced me to the world of rehabilitation

and showed me how the work that an organizational researcher does can improve the

lives of people with disabilities. She truly opened my eyes to a whole new world.

I would like to thank the Department of Health Services Research, Policy, and

Management as well as the Dean's Office for providing me with the four years of

funding. I would not have been able to purse my dream of attaining my Ph.D. had it not

been for your generous support. I would also like to thank the faculty and staff in the

Department of Health Services Research, Policy, and Management at the University of

Florida. My professors have taught me many new things and opened my eyes to the

many different areas of health services research. I would like to thank the staff for being

there to address any problems that I would present to them. I also would like to thank my

fellow colleagues in the department for their support and encouragement. I feel as

though I have made some great friends during my time here. I would encourage them to

continue working hard because they will be in the same situation soon.

I would like to acknowledge the staff of the Rehabilitation Outcomes Research

Center (RORC) at the Malcolm Randall Veterans Administration Medical Center. They

have all been a very helpful resource in getting me to this point. In particular, I would

like to thank Elizabeth Cope, Ellen Esparolini, and Diane Cowper for their initial help

when I first started doing research at the RORC. I would also like to thank Kate

McKuhen for her assistance also. I would like to thank the Focus 1 group. In particular,

Drs. James Stansbury and Bruce Vogel have provided invaluable assistance regarding









methodology issues. Without the dedication of Dr. Dean Reker and Clifford Marshall, I

would not have the data necessary to complete the research for this dissertation.

I would like to thank Oscar Wood, Rich Greene, Darryl Nelson, Dave Kuti, Bryan

Patterson, Monde Qhobosheane, Carlos Taylor, and Earnest Hunter. You have stood by

me through good and bad times. For some you, our bond was established while I was in

graduate school. For others, it began back at UVA. Either way, our bond is strong now

and it will not ever be broken. In addition, I thank Nicale Whitehead-Nxumalo, Cheryl

Taylor, Stephanie Shepard, Rita Wharton, Shawn Agyepong, Nikkia Despertt, and

Rosalyn Hobson for their continued support and prayers through this long process.

Finally, I would like to thank Dr. Augustus A. Petticolas for his mentoring and words of

encouragement over the years. His advice has helped in ways he will never know. I

appreciate everything he has done for me.

Last but certainly not least, I thank my family for being supportive of me while I

was so far away from home. I do not remember one time when I reached out that they

were not there. I thank my mother for her prayers and encouragement and reminding me

that I could do this because God had it in his plan. She continues to inspire me everyday

to become a better Christian which makes me a better person. I thank my father for his

continued support and prayers. During my time here, he has continued to provide with

the proper insight that I needed to get this through the process of attaining this degree. I

thank him for sharing his wisdom. I want to thank Melvin for supporting me as if he I

was his real son. Although he was not obligated to do anything for me, he has done more

than enough. I thank my aunts (Joyce Cabell and Angela Thomas) and uncles (Robert

and Earnest Deane), and they respective spouses, for they support for me. I thank them









because every time I would see them, they seemed to know the right words to say to keep

me going. I thank my sister Tammy for being a good listener. She has had to listen to

me gripe more than anyone over the years and yet she never blew me off. I encourage

her to continue to seek all her personal goals in life. Once again, thanks everyone.
















TABLE OF CONTENTS
Page

ACKNOWLEDGMENTS ......................................................... iv

LIST OF TABLES................................................ xii

LIST O F FIG U R ES ..................................................................... ........................ xiii

ABSTRACT ....................................................... xiv

CHAPTER

1 IN TRO D U CTIO N .................................................................... ........................

A C culture in C change ................................................................. .............................
Stroke and Rehabilitation ......................................................... ........................ 4
R research Q uestions............................... ............................................................ 5
D issertation O overview ................................................... ........................................ 6

2 LITERATURE REVIEW........................................ ......................7

Measuring Stroke Outcomes........................... ....... .........................7
Problems with Stroke Assessment....................... ...... .........................11
Structure and Stroke Outcomes ......................................... ....................... 12
The Structure of Stroke Rehabilitation Care within the VA.....................................14
G general Structural Effects................................. .............................................. 16
Service Line Implementation....................................... ........................18
Rehabilitation Bedservice Units (RBUs) Structural Effects.....................................19
Limitation of Previous Research ................................... .........................21

3 THEORY AND HYPOTHESES.......................................................................22

Conceptual Model............... ..................................................................22
Open Systems Perspective...............................................................................23
Rational Systems Perspective......................................................................24
Natural Systems Perspective.............................................................................24
Structure-Process-Outcomes Paradigm ................................. ........................25
Contingency Theory ................... ............................25
Structure and Structural Forms...................... ... ................................27
Uncertainty and Context......................................................................32
The Concept of Fit.............................................................................34


viii









Issues w ith Fit................................................................. ......................... 34
Critique of the Contingency Theory......................... ................................36
Theoretical Convergence ...................................................... .......................... 39
H ypotheses........................................................................... .................................4 1
Context and Performance ...........................................................41
Structure and Performance .........................................................42
Context and Structure.................................................... ..........................45
Fit and Performance ................... ........................................46
Perform ance.................................... ................................ ....................... 46
Alternative Hypotheses ..........................................................46
Structural Characteristics..............................................47
Patient Characteristics ....................... ................................... 47
Summary of Hypotheses:..................... ....................................50

4 M E T H O D S ............................................................................ ...............................5 1

D ata Source and Sam ple........................................................ ......................... 51
Survey Population ................................... ............................ 52
Historical Background of the FSOD ....................... ............................52
Historical Background of the PTF.................................... .............................53
Background on the ARC, PSSG, and VSSC ...........................................54
Data Reliability and Validity........................... .... .........................55
M easurem ent................................................ ...................................................57
Stroke Patient Outcome...........................................................58
C o ntex t .......................................................................... ................................5 8
Structure ......................................................................... ...............................59
F it................................................................................. .................................. 6 1
Control V ariables...................................... ...............................................61
Structural ................................................. ......................................... 61
Patient........................ ....................... .............................. 62
Y ear ................................................................................ ...............................6 3
A analysis Strategy ............................................... .............................................63
D descriptive Statistics ............................................. ......................................... 67

5 DESCRIPTIVE STATISTICS AND RESULTS................... .............................68

Profile of Formal Rehabilitation Facilities ............................... ........................68
Organizational Context............................ ...... .........................70
Organizational Structure.......................... ........ ........................70
Variable conceptualization issues..............................................70
F in d in g s ............................................................................ .............................7 1
Patient Characteristics ............................................. .......................................74
Results of Hypothesis Testing ....................... ..................................76
Context-Performance Hypothesis......................... .... .....................76
Structure-Performance Hypotheses ................................. ......................77
Fit H ypothesis............................................. .............................................77
High/Low Performers Hypothesis...................... .... .............................79


ix









Control V ariables....................................... ..............................................80
Summary of Analytic Results.................. ...... .....................81

6 D ISC U SSIO N ....................................... ................................................................82

Key Points for Discussion ..................... ......................................82
Patient Population ............................................. .............................................. 83
Facility Setting Types........................................................... .........................84
Fit 85
High/Low Performing Facilities......................... ..... ......................87
Impact of VA Management ..........................................................88
Staffing R atios ............................................... ..........................................88
D disciplines .................................................. ............................................. 89
Size ..................... ..............................................................................................9 1
Careclass................ ................................................. ......................... 91
S getting ......................................................................... ...................................93
Contribution to the Theory ..................... ....................................... 94
L im itations......................................................................... ...................................96
D ata Lim stations ........................................ ............................................ 96
Variable Limitations........................ ....... ......................98
Recom m endations....................................... .............. .......................... 99
Recommendations for Future Research......................... ...............................99
Recommendation for Current Study.............................. ...................100
Conclusions.............................. .. ....................................101

APPENDIX

A FUNCTIONAL INDEPENDENCE MEASURE: ITEMS & LEVELS OF
FU N C T IO N ................................................................................ .....................103

B MEASURES USED IN THE ASSESSMENT OF STROKE OUTCOMES ...........107

C SUMMARY OF STUDIES FOCUSING ON THE STRUCTURE OF VA
HEALTHCARE FACILITIES ................................... .......................108

D SUMMARY OF CONTINGENCY THEORY IN THE HEALTHCARE FIELD
THA T UTILIZES FIT ........................................................... ....................... 15

E SUMMARY OF STUDY CONCEPTS AND VARIABLES................................119

F DESCRIPTIVE STATISTICS......................... ....... ........................120

G PAIRWISE CORRELATION MATRIX .................... ............................122

H REGRESSION FOR DIRECT AND CONTINGENT EFFECTS .........................132

LIST OF REFERENCES....................... ...............................134









BIOGRAPHICAL SKETCH........... .. ........................................................... 148















LIST OF TABLES


Table page

1 Number of Patients Listed by Facility Type Per Year .............................................69

2 Structural Characteristics by Setting Type......................... ............................72

3 Structural Characteristics by Year............................... .......................73

4 Patient Demographic Information According to Facility Type .............................75

5 Patient Demographic Information According Year of Admission.........................78

6 Regressions for High and Low Performing Facilities............................................79















LIST OF FIGURES


Figure page

1 Conceptual M odel ............................................ ........................................... 22

2 Interaction of Context and Structure on Performance......................................37

3 (a) Deviation from Context-Structure Relationship and (b) Expected Relationship
Between Absolute Values of Deviation Score and Performance...........................38















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

EFFECTS OF FACILITY VARIATION
ON THE OUTCOMES OF STROKE PATIENTS
IN THE VETERANS ADMINISTRATION'S (VA) HEALTHCARE SYSTEM

By

Jared P. Deane

August 2004

Chair: Christopher E. Johnson
Major Department: Health Services Administration

The purpose of the project was to determine if changes in the structure of VA

rehabilitation facilities fit the environment in which they operate to improve the

functional status of stroke patients. Functional Independence Measures (FIM) scores

for a sample of 9231 stroke patients who received formal rehabilitation in the VA's

rehabilitation bedservice units (RBUs) were obtained from the Functional Status

Outcomes Database (FSOD), which is housed at the VA's Austin Automation Center

(AAC). The environmental variable was created using an algorithm to calculate

FIM-Functional Related Group (FRG) score. Data on the structure of the facilities were

obtained from various VA databases. Finally, fit scores were determined by taking the

absolute value of the residuals predicted by the regression of each individual measure of

structure on context.








Contingency theory was used to determine if improved FIM scores are the

consequence of the fit between context and structure. To test this theory, an ordinary

least squares (OLS) regression was run, with the discharge motor FIM as the dependent

variable. Independent variables included measures of structure, context, and fit scores;

and control variables for structure and patient demographics. Lastly, a series of

regressions was run to determine if there was a difference in the structural characteristics

of facilities with high versus low FIM scores.

There was a fit between context and structure on 5 of the 9 structural variables.

These 5 fit combinations have a positive impact on FIM scores. Only size of facility had

both direct and contingent effects on FIM scores. In addition, high- and low-performing

facilities only differed structurally on the number of disciplines represented at each

facility.

We proved the theory that improved FIM scores are contingent on fit. At the same

time, very little structural difference was found in facilities with higher versus lower FIM

scores. Overall results are promising for VA management, showing that proper changes

in structure could improve outcomes for stroke patients.














CHAPTER 1
INTRODUCTION

Attitudes towards rehabilitation for stroke patients have changed. In the past, the

attitude of rehabilitation therapists was that there was not much that could be done in

order to assist the stroke patient. But recently, it has been demonstrated that rehabilitation

has the potential to create improvement in the physical functioning of stroke patients.

Within the Veterans Health Administration (VHA), a reorganization took place in 1995

to better serve veterans, including rehabilitative services. As part of the reorganization,

more discretion was given to the directors of newly created networks. The directors

would be able to reorganize the facilities better because they would have fewer facilities

for which they would be responsible, as compared to a national director. Part of the

reorganization would be to create a facility structure that fits the environment, or context,

in which rehabilitation care is provided. Given the importance of rehabilitation, the

structures that provide rehabilitation, and the environment in which the facility is located,

this dissertation examines the fit between the structure and context of VHA facilities that

provide rehabilitation services and its subsequent effects on stroke patient's outcomes.

A Culture in Change

The Veterans Health Administration, part of the Department of Veterans Affairs

(VA), is the largest integrated healthcare system in the United States (Kizer et al., 1997).

Although it provides a large amount of healthcare services to the nation's veterans, the

manner in which the VA was providing care in the past was not of the highest quality

(Jha et al., 2003). There were complaints regarding inadequate and inconsistent services









from these facilities (Kizer, 1998). Examples included long wait times, not treating

patients with respect, and long inpatients lengths of stay for services that were being

provided on an outpatient basis in the private sector.

In an effort to change the way healthcare services were provided as well as

change the image of the VA, a plan was initiated in 1995 regarding the structure of the

VA healthcare facilities. The idea would be to organize the current facilities into twenty-

two regional networks called Veterans Integrated Service Networks (VISNs). This new

organizational structure would resemble a structural form called an integrated delivery

network (IDN). IDNs originated in the private sector and were proven successful by

physicians, hospitals, and insurers who wanted to provide quality healthcare across the

continuum of care while working within fixed financial limits (Dowling, 1998).

Adopting an IDN organizational form would allow each VISN to provide continuity of

care that came about as a result of better integration of resources as well as "virtual"

delivery mechanisms such as outsourcing of services, partnerships, and strategic alliances

(Zajac & D'Aunno, 1997). With this change in organizational structure, the VA hoped to

create a new organization that was both efficient and productive, decentralized decision-

making, eliminated unnecessary bureaucracy, flattened administrative structures, and

streamlined communications.

Although this reorganization had the potential to provide a great benefit to the

VA, there were some issues to overcome. A change would mean a departure from the

"business as usual" manner in which healthcare had been provided by the VA for

decades. According to Kizer (1998), before the change to the VISN structure in 1995, the

VA was an "older, monolithic, military-style top-down organization" (p.1). In order for









the VISN structure to work, a "180-degree shift in management philosophy and

execution, plus an intense application of integrated management network systems" (p.1)

had to be used. Therefore, a complete change in the culture of the organization had to

take place.

By itself, change in culture is difficult, but when one is trying to change a culture

that is, according to Vestal et al. (1997), "born of military and government lineage and

aged for generations" (p.340), the change is that much more difficult. In their very

general analysis of the culture within the VA, Vestal et al. (1997) found that pre-1995,

the culture was very functional and did not emphasize unification, adaptiveness, and

flexibility. Implementing a VISN structure would mean a change to a culture that is

opposite to that of the traditional VA culture. This type of change would take significant

time. It has been suggested that effective change would take five to ten years (Chapin,

1994). This would seem to be the time frame that current VA facilities that provide

rehabilitation care are following. In a study of organization culture in VA rehabilitation

teams, Strasser et al. (2002) found that team members perceived working in a culture that

is still impersonal and not very dynamic following three years post-VISN

implementation. The perceptions of the managers who manage these teams were the

exact opposite. They perceived the current structure as one that provides a culture that is

more personable and dynamic. One could infer from the results that the "vision for

change" is working its way down to the clinicians, with the administrators already

buying-in to the idea.

With regards to this study, this change in culture to a more patient-focused type of

care meant that the decisions regarding the health of the patient would be made as close









to the patient-level as possible, with the national headquarters providing support rather

than being the primary overseer (Kizer, 1995). With the VA adopting an organizational

structure that is characterized by the ability to be flexible and adaptive, there will be

multiple ways in which facilities organize to provide rehabilitation care. These changes

in the facility's structure could be a catalyst for improvements in outcomes of the stroke

patients who receive their care from the VA. This study examines whether the structural

changes, indeed impact outcomes of stroke patients.

Stroke and Rehabilitation

Stroke is a disease that is affecting our country in large numbers. It is the third

leading cause of death in the United States, trailing only cancer and heart disease

(American Heart Association, 2003). In addition to being one of the leading causes of

death in this country, it is the leader in terms of adult onset disability. For example, in

1999, over 1.1 million Americans reported difficulty with functional limitations,

activities of daily living (ADLs), and disabilities as a result of suffering from a stroke

(Centers for Disease Control, 2001).

Having a stroke can have differential effects on individuals. Someone who has no

lingering side effects of the stroke would be classified as a mild stoke case. A moderate

stroke classification would be more severe than a mild stroke and would have

neurological deficit that affected continence, mobility, and ability to look after

themselves (Kalra et al., 2000). The final classification of stroke severity would be

severe. These cases would include patients who tend to have swallowing problems that

should not amenable to modifications in the patient's dietary regimen and also require

heavy nursing assistance.









Despite advances in acute stroke care, there is still a lack of interventions

available that can either minimize or reverse the effects of stroke (Katzan et al., 2000;

Mohr, 2000; Albers et al., 2000). Therefore, stroke patients have to rely heavily on

rehabilitation care (Duncan et al., 2002). According to Stineman (2001), rehabilitation

"is a set of philosophies, treatments, and therapies that, when combined with natural

recovery, are intended to enhance patient's potential for participating in meaningful life

experiences" (p.148). In this and other Western countries, rehabilitation is provided in

either an inpatient or outpatient setting (Langhorne & Duncan, 2001). Within the VA,

rehabilitation is provided in both settings with the inpatient setting being further

subdivided into acute and subacute bedsections. Acute bedsections provide services in

addition to rehabilitation whereas in a subacute bedsections, the VA has designated these

units strictly for rehabilitation. This study will examine whether being in either an acute

or subacute rehabilitation bedsection has an effect on a stroke patient's outcomes.

Research Questions

This dissertation is based in a contingency theory framework that assumes the fit

between a VA facility's structure and context has an impact on stroke patients who

receive rehabilitation. Further, implementation of the network, or VISN structure, in the

VA shifted discretion for reorganization of the facilities from the VA's national

headquarters to the directors of the individual VISNs. This meant that each organization

could be organized to better fit its environment. This dissertation presents a conceptual

model that suggests how and why organizational fit is important for patient outcomes.

These issues are examined among a national sample of facilities within the VA that

provide rehabilitation care. Research questions include the following:









1. Does the environment (context) in which rehabilitation care is provided have
an effect on stroke patient outcomes?

2. Does the structure of the rehabilitation facility effect stroke patient outcomes?

3. Does the structure of the facility effect the context in which rehabilitation is
provided?

4. Does the fit between a rehabilitation facility's structure and their context
improve stroke patient outcomes?

5. Is there a structural difference between facilities with higher patient outcomes
and facilities with lower patient outcomes?

Dissertation Overview

This dissertation is organized into six chapters. Chapter 1 introduces the reader to

the areas of stroke, stroke rehabilitation, and reorganization within the VA. Chapter 2

provides a comprehensive review of two bodies of research: (1) studies that identify key

assessors of stroke patient outcomes and (2) studies of the effects of structure on

rehabilitation. A conceptual model and research hypotheses regarding how and why

organizational structure affects patient outcomes are presented in Chapter 3, along with a

discussion and synthesis of the two theoretical perspectives in which they are based

(structure-process-outcome paradigm and contingency theory). Chapter 4 details the

methods used in the research, including information regarding the population, sample,

measures, and analyses. A descriptive profile of stroke rehabilitation units and a

summary of the analytic results from the dissertation are presented in Chapter 5. Finally,

in Chapter 6, the results from this study are discussed, with attention to theoretical

implications, management and policy implications, study limitations, and suggestions for

future research.














CHAPTER 2
LITERATURE REVIEW

The central aim of this dissertation is to examine the relationship between the

facilities that provide rehabilitation services within the Veterans Administration (VA) and

stroke patient outcomes. This chapter reviews and summarizes previous research on (1)

issues associated with assessing stroke; (2) the impact of structure on stroke

rehabilitation; and (3) the effects of organizational change on patient outcomes within the

VA. First, I summarize the body of research that describes the various instruments that

are used to assess stroke. Next, I introduce the important structural characteristics of

facilities that provide rehabilitation services. This will be followed by a summary of the

effects of organizational change within the VA on the manner in which services are

provided to veterans and the subsequent effects of these changes on patient outcomes.

Finally, I discuss the limitations of this body of research in light of the current study.

Measuring Stroke Outcomes

Currently, there are many different instruments available to clinicians that can

assess stroke and improvement in stroke patients after rehabilitation. Each instrument is

unique and has its own idiosyncrasies, yet they all measure some form of functional

status. These different instruments can be grouped into one of four categories (Duncan et

al., 2000). The first category includes instruments that measure impairment. Impairment

is defined by a loss or abnormality of psychological, physiological, or anatomical

structure or function (World Health Organization, 1980). Included in this classification

are scales that measure deficits ranging from various neurological symptoms to cognitive









deficit and even depression. The impairment instrument that is the most often used is the

NIH stroke scale (NIHSS). This scale uses the summation of the individual elements of a

neurological examination to come up with an overall score (Brott et al., 1989; Goldstein

1989).

The second group of assessors measures level of activity. Measures of the level

of activity are generally referred to as activities of daily living (ADLs). The various

scales within this group measure performance in occupational functions that are

necessary for independent living. ADLs are most useful for assessing the effects of

rehabilitation (Reding & Mcdowell, 1987). Instruments included in this group include

the Barthel Index (BI) and the Functional Independence MeasureTM (FIM) motor score.

The BI is a widely used ADL scale that is comprised often items ranging from feeding to

bladder control to stair climbing (Mahoney & Barthel, 1965). Varying weights are

assigned to the ten items. The scores range from 0 (completely dependent) to 100

(independent in basic ADLs). Higher scores represent a higher degree of independence.

The FIM motor scale is 1 or 2 scales of the FIM; the other being cognitive. It

assesses self-care, sphincter control, mobility, and locomotion. Unlike the BI, each of the

thirteen categories is equally ranked using a seven-point scale. According to Granger and

Hamilton (1996), each point on the scale represents a "gradation of independence and

reflects] the amount of assistance a patient requires" (p.l17). This instrument, as well as

the BI, is recommended by the Agency for Health Care Policy and Research (AHCPR)

for measuring poststroke disability (van de Putten et al., 1999; Kidd et al., 1995).

Appendix A provides specific definitions for each of the physical and cognitive FIM

items.









Similar to the BI, lower scores on the motor FIM indicate more dependence. For

the FIM, there are two types of dependence that are related to the degree of assistance

that is required in order to perform the task on the measure. Modified dependence is

defined as an assistant providing less than half the effort required to complete the task,

while complete dependence is when the assistant provides more than half of the effort.

Another instrument that falls into this category would be the modified Rankin

scale (MRS). The scale is a global disability scale that measures independence rather

than performance on specific tasks, like the Barthel Index. By measuring independence,

the user of the Rankin gets a sense of the mental, as well as physical, adaptation to

neurologic deficits (van Swieten et al., 1988). Therefore, the Rankin would be a better

evaluator of whether or not patients can look after themselves in daily life when

compared to ADL scores.

Quality of life and health status measures lead the third group of assessors. One

instrument that falls into this category is the Short-form 36 (SF-36). It measures eight

health concepts ranging from physical functioning to energy/fatigue level. This survey

fills the gaps between lengthy surveys used in research projects and single-item health

measures used in national surveys and clinical investigations (Ware & Sherbourne,

1992).

Another measure that would be classified in the quality of life and health status

measures is the Stroke Impact Scale (SIS). As the name implies, it is a scale that is

specifically designed to comprehensively assess stroke specific outcomes (Lai et al.,

2003). It assesses eight domains including strength, hand function, ADL and

instrumental ADL (IADL), mobility, emotion, memory, communication, and









participation (Duncan, Bode et al., 2003). This instrument was developed to extend the

range of function that was previously measured by the BI and SF-36. A shorter version

of the instrument (SIS-16) was developed to assess stroke approximately 1 to 3 months

poststroke. The SIS-16 has shown to be sensitive to functional differences across all

levels of the stroke when compared to the BI (Duncan, Lai et al., 2003).

The third scale in this category is the Stroke Specific-Quality of Life (SS-QOL)

measure. This measure has 12 domains of which some are similar to the SIS, i.e.,

mobility and thinking. Just like the SIS, the SS-QOL has a broader coverage of functions

that the SF-36. Examples of items included in the SS-QOL but not the SF-36 include

language, hand function, cognition, and vision (Williams, Weinberger, Clark et al.,

1999). As proof of this broader coverage, when used to assess mild to moderate ischemic

stroke patients, the SS-QOL score was associated significantly with patient ratings of

overall health related quality of life at 1 month poststroke, while there was no association

with the SF-36 (Williams, Weinberger, Harris et al., 1999).

The final category includes measures that are self-developed. There are but a few

of these studies available, yet none are validated for stroke use (Duncan et al., 2000).

Appendix B details the various measures used in the assessment of stroke outcomes as

well as information regarding the testing of validity and reliability for each instrument.

As mentioned previously, there are many instruments to choose from that can

adequately assess stroke and improvement in stroke patients after rehabilitation. For this

study, I am interested in using a measure that would fall into the levels of activity

category as well as measure progress made while in rehabilitation. The motor FIM

satisfies both desires. This instrument was also chosen because it is a comprehensive









assessment tool, although it is simply a basic indicator that collects a minimum amount of

information that is necessary for disability assessment (UDS Data Management Service,

1990). Finally, the motor FIM was chosen as the outcome measure because its scores can

be translated from an ADL measure to a categorical disability measure (MRS), although

there is a ceiling effect and the translation is not complete (Kwon, 2004).

Problems with Stroke Assessment

There are many issues that arise when one is accurately trying to assess stroke

patients. The three issues that are the most important for this study include consistency

problems in the selection of outcome measures, a lack of consistency in the timing of the

assessment of stroke, and the selection of arbitrary points to determine favorable or

unfavorable outcomes. With regard to the first issue, there are consistency issues with

selecting appropriate outcome measures because primary measures are being chosen that

are measuring outcomes other than ADLs. Duncan et al. (2000) suggested in their review

of appropriate outcome measures for stroke that the "primary outcome measure should be

at the level of activities capturing not only basic ADLs, but also including IADLs and

advanced mobility" (p.1435). This study addresses this issue by using an outcome

measure (the FIM) that falls into the ADL category. Even though it is suggested that

impairment not be the primary outcome measure, it still has a significant role as an

outcome measure. In the same review, Duncan et al. (2000) suggested the role for

impairment outcome measures would be as an assessor of drug efficacy or therapeutic

interventions in terms of neurological recovery.

The second issue that plagues stroke assessment is timing. When a clinician or

researcher is selecting the appropriate time to assess stroke, they have to select a time that

allows for complete recovery from the stroke. A time should be selected where









spontaneous recovery has plateaued. Duncan et al. (2000) suggest that this plateau

occurs around the five or six month period. This study does not address this issue

because the data only assess motor and cognitive function prior to discharge, not the

recommended five or six-month period.

The final issue that makes stroke assessment difficult is the selection of arbitrary

cutoff points that determine favorable versus unfavorable outcomes. These cutoff points

make it difficult to compare results across trials. To address this issue in the present

study, the previously defined scales for the FIM, which determine levels of

independence, will be used to determine favorable or unfavorable outcomes. A favorable

outcome would be an increase in the level of independence.

Structure and Stroke Outcomes

Previous research on the effects of structure on stroke patients' outcomes has been

concentrated into one of three areas: (1) payment structures, (2) various organizational

configurations, and (3) stroke unit effects on outcomes. Regarding payment structures,

Yip et al. (2002) examined the effects that the implementation of the Prospect Payment

System (PPS) has on the patient case mix and the utilization of rehabilitation services for

all diseases, including stroke, in skilled nursing facilities (SNF). The study found that

there were significant differences between fee-for-service (FFS) and managed care SNF

patients. Post-PPS implementation, FFS patients had shorter lengths of stay (LOS) in the

hospital and higher role emotional scores when compared to managed care patients. On

the other hand, managed care patients had lower physical functioning scores than FFS

patients. In terms of rehabilitation utilization, Yip et al. (2002) found that post-PPS

implementation, in general, both types of patients had shorter stays for rehabilitation. In









addition, the number of minutes spent with the physical therapist decreased by almost 20

minutes per day.

In a similar study of elderly patients, Kramer et al. (2000) looked at the treatment

and outcomes for stroke patients who were enrolled in either a Medicare FFS plan or a

Medicare Health Maintenance Organization (HMO). After evaluation of these patients

twelve months post-stroke, it was concluded that short-term functional outcomes and

eventual community residence rates were lower for the Medicare HMO enrollees. These

results are consistent with the lower amounts of rehabilitation (nursing home versus

rehabilitation hospital) along with access to less specialty care, which is common to all

HMO enrollees.

The second area of previous research focused on the effects of different types of

facilities on outcomes. Kramer and his colleagues (1997) compared outcomes for stroke

patients. The three different types of facilities that were compared were rehabilitation

hospitals, subacute nursing homes, and traditional nursing homes. The results of the

study showed that patients who were admitted to the rehabilitation hospital were more

likely to return to the community and recover their activities of daily living (ADLs). In

addition, subacute nursing home patients were more likely than traditional nursing home

patients to return to the community.

Finally, the research on patients in stroke units provided expected results. Several

studies found that stroke patients who were treated in stroke units had improved

outcomes, reduced LOS, and did not have increased therapy time (Kalra et al., 1993). In

another study, Kalra and Eade (1995) found that stoke units, when compared to general

wards, improved the outcomes of severe stroke patients as well. Based on the results, the









researchers posited that improvement in outcomes may be attributed to multidisciplinary

teams that have expertise in treating stroke patients and an environment that is more

dynamic which leads to more innovative management strategies. These results were

further confirmed in a more recent study done by Evans et al. (2001), who also looked at

management styles for stroke units and general wards. They found that the "specialists

aspects of stroke management cannot be replicated in generic settings, even when

ongoing specialist stroke-team care is provided" (p.1590).

Finally, Evans et al. (2002) looked at the effects of stroke units from a clinical

perspective. Their research explored unit effectiveness based on the type of stroke

subtype. The results showed that patients with large-vessel infarcts have improved

outcomes while those with lacunar syndromes did not improve when admitted to a stroke

unit.

The Structure of Stroke Rehabilitation Care within the VA

The purpose of examining the structural effects of an organization when assessing

performance is that the structural indicators assess the capacity that the organization

possesses for effective performance (Scott, 1998). According to Scott (1998), these

indicators include "all measures based on organizational features or participant

characteristics presumed to have an impact on organizational effectiveness" (p.358). For

healthcare facilities, these measures include the adequacy of facilities and equipment as

well as the qualifications of the medical staff in terms of training and certification.

Specifically for this study, the major organizational characteristics of stroke

rehabilitation care can have significant effects on patient outcomes (Reker et al., 2000).

One of the major characteristics is the presence of a multidisciplinary team. These teams

bring various skills to the table when treating stroke patients. As previously mentioned, a









portion of the stroke patients who receive rehabilitation care within the VA's healthcare

system, care is provided in rehabilitation bedservice units (RBUs). RBUs are considered

the upper tier of care for stroke patients. It can be identified as a location with a formal

rehabilitation bed unit that is either classified as acute or subacute. This classification

infers the unit is organized for specific impairments, with stroke being one of those

impairments. It is believed that the teams that service these RBUs are multidisciplinary

and that serve only those bed units. The dedicated staff allows for more organized care

because they are organized into teams (Reker et al., 2000). These organized

multidisciplinary teams can be used to benefit postacute stroke patients (Langhome &

Duncan, 2001).

Regarding structure-outcomes research within the VA, there has been a relatively

small stream of research on the effects of the VA's structure on the outcomes of stroke

patients who require rehabilitation with a majority of the research having been done more

recently. In an effort to better organize research that was done regarding VA structure

and outcomes, Hoenig et al. (1999) proposed that the structure-process-outcomes model

be considered. This model posits that structure effects outcomes in an indirect manner

(Donabedian, 1980). The conclusion from this theoretical study was that utilizing this

model for rehabilitation research would promote better quality in research regarding

stroke rehabilitation.

Following that study, Hoenig and colleagues (2000) developed a taxonomy for the

classification of stroke rehabilitation services. The expert panel determined that

structural characteristics such as personnel, physical facilities, coordination of care, and









hospital characteristics had face and/or construct validity. This study provided an initial

taxonomy for assessing stroke rehabilitation units.

For the few studies that have looked at the general structure of the VA's facilities

and their effects on stroke patient's outcomes, the results have varied (Duncan et al.,

2002; Hoenig et al., 2000, 2002; Stineman & Asch, 2001; Hoenig et al., 2000; Camberg

et al., 1997). This stream of research has tended to fall in one of two categories: (1) the

effects of VA structure, in general, on the stroke patient's outcomes or (2) more

specifically, the effects of RBU structure on the stroke patient's outcomes. Each section

will be reviewed individually. A summary of these studies can be located in Appendix C.

General Structural Effects

Hoenig et al. (2002) empirically tested the structure-process-outcomes model for

stroke rehabilitation care. This was accomplished by using the previously proposed

taxonomy that was developed for assessing stroke rehabilitation units (Hoenig et al.,

1999). The results of this study proved the structure-process-outcome thesis that

structure affects patient's outcomes in an indirect manner via processes of care.

Continuing with the stream of research that was focused on structure and its

effects on outcomes, Hoenig et al. (2001) examined structural effects on stroke patients

for 182 VA facilities. As was hypothesized, the study proved that there was variation in

the structures that provide rehabilitation care for stroke patients. According to the

researchers, the variation in the structure plays a direct role in the patterns of care

provided and an indirect role in the outcomes.

In order to understand the potential impact of the setting on stroke patient's

outcomes, Duncan et al. (2002) looked at the effects of the setting on guideline

adherence. Guideline adherence is an important piece of the rehabilitation puzzle









because greater levels of adherence to these guidelines may improve patient outcomes.

Unfortunately, this study concluded that at six months, compliance with acute

rehabilitative care was unrelated to patient outcomes. It is believed that this relationship

happens because there is a confounding effect that occurs between processes of care and

the severity of the residual disabilities (Duncan et al., 2002). Other analyses showed that

there was a difference in compliance levels based on the setting. Rehabilitation

compliance in the nursing home setting was substantially worse than in the inpatient

setting.

Not all of the research on settings has shown significant effects. In a similar study

looking at postacute stroke guideline compliance, Reker et al. (2002) looked at how

adherence played a role in patient satisfaction. As part of the study, characteristics of the

structure in which rehabilitation is provided was included. The results of the study found

that structure did not have a significant effect on various measures of a patient's

satisfaction.

The setting in which rehabilitation takes place not only has an effect on stroke

patients while they are in rehabilitation, but it also plays a role in stroke patients

continued improvement depending on where they are discharged. Camberg et al. (1997)

looked at how stroke patients, among other types of patients with neurological

impairments, performed after being discharged from a VA facility. The results of the

study showed that for the stroke patients, at one month post-stroke, a larger percentage of

patients who were admitted to VA nursing homes were readmitted to VA hospitals as

compared to stroke patients who were discharged to the home or a community nursing









home. At the sixth month post-stroke time interval, the differences in readmission

patterns were small.

Service Line Implementation

As mentioned previously in the Introduction section, the VA decided to

reorganize in an effort to improve the healthcare services that it was providing to the

veterans. By implementing the VISN structure, the VA was essentially adopting an

organizational form that was similar to the IDN that had proved to be effective in the

private sector. The change to a structure that resembled an IDN was chosen because

having the VA hospitals operating as individual entities was not producing the outcomes

that the VA leadership desired (Kizer & Pane, 1997). Although this change was a move

in the right direction, there were two organizational challenges that the VA could

possibly face with this implementation. The first challenge is delivering service that is

consistent in both content and quality for patients who would receive care from multiple

providers, in multiple settings, and possibly in multiple geographic regions. The second

challenge is providing such care in a cost-effective manner (Charns et al., 2001).

One way to combat these challenges is to implement clinical service lines,

sometimes referred to as simply service lines. A service line is defined as "a family of

organizational arrangements based on a hospital's outputs, rather than on its inputs. They

can be defined as either (1) procedures or interventions, such as surgery; (2) management

of diseases, such as comprehensive care for cancer; or (3) management of care for and/or

maintaining health of identifiable segments of the population (Chains et al., 2001).

Based on these classifications, it seemed that the VA had classifications for both the

management of diseases and care for identifiable populations. The two most popular

service lines were for mental health and primary care during the initial years after the









mandate for reorganization was handed down (1997-1999). Charns et al. (2001)

performed an analysis of the facilities that had these service lines to see the effect that

they had on outcomes between fiscal years (FY) 1997 and 1998. These results of the

study concluded that the implementation of either service line had a significant

improvement in the outcomes when compared to facilities that did not have service lines.

However, one explanation for the results was that the effects may be that easily

detectable in such a short period of time.

Rehabilitation Bedservice Units (RBUs) Structural Effects

Although the VA did not have service lines specifically for stroke, there were

organizational changes. The VA implemented the rehabilitation bedservice unit (RBU)

to better serve it stroke patients and at the same time answer the call sent forth by then

Undersecretary Kizer for reorganization. As mentioned earlier, the rehabilitation

bedservice unit (RBU) is the VA's answer to having an organized structure for

rehabilitation that has multiple disciplines represented when providing rehabilitation care

for stroke patients. There were four studies that were done that looked at the RBUs

effects on stroke outcomes (Reker et al., 2000, 1998; O'Donnell & Hamilton, 1997; Chen

et al., 2002;).

In their study of RBUs, Reker et al. (2000) wanted to determine if the structure of

the RBUs had an effect on patient outcomes. The results of their study showed that

between 6 and 13 percent of the variation in the outcomes of the stroke patients could be

attributed to the rehabilitation unit. Also of interest was the fact that the lower the

physician workload ratio, the higher the patient functional gain. This is interesting

because, based on the way the ratio was constructed, the lower ratio usually indicated that

the physician would be working more. Intuitively, one would believe that the opposite









would be true. The rationalization for this relationship is similar to that for surgeons;

higher volumes of procedures net better outcomes (Showstack et al., 1987; Hannan et al.,

1995; Hannan et al., 1998). In addition to the structure's effect on the patient's

outcomes, the study revealed that a more diversified staff was associated with an

increased length of stay. The increased length of stay was associated more with

coordinating issues between the staff than an issue of the patient's health status.

Two other studies looked at issues regarding the RBUs and referral sources to see

its impact on outcomes. The results were mixed. In the first study, O'Donnell and

Hamilton (1997) found that the referral source, that is, where the patient was before they

came to the RBU, was a significant predictor of stroke rehabilitation outcomes. It was

also a factor in non-structural characteristics such as admission function, length of stay

and home discharge. These results were similar to those found in similar types of

research done in non-VA settings (Hosek et al., 1986; Heinemann et al., 1987; Stineman

& Williams, 1990). In the second study, Reker et al. (1998) also looked at referral source

to RBUs. Their results were quiet different. The referral source was not a significant

predictor of the stroke patient's outcomes. However, similar the first study, the referral

source was a predictor of the length of stay.

Finally, Chen et al. (2002) looked at RBUs and their effects on functional gains

and treatment intensity. The results of the study showed that none of the structural

characteristics had a significant effect on treatment intensity. The study also

demonstrated that there were facility differences in lengths of stay, provision of

rehabilitation therapies, and functional gains.









Limitation of Previous Research

The studies discussed here represent a strong foundation from which to research

the rehabilitation facilities and their subsequent effect on the outcomes of stroke patients.

However, several gaps appear from this stream of research. First, there are issues with

selecting an appropriate outcome measure by which to assess the stroke patient. It is

believed that selection of the FIM as the outcome measure will address the issues raised

in previous studies. The primary reason for the selection of the FIM was that it is the

industry standard for rehabilitation (vawwl .va.gov/health/rehab/FSOD.htm). Secondly,

the FIM was selected because it was constructed specifically to measure functional status,

unlike many of the other measures used.

In addition to selecting an appropriate measure to assess stroke, many of the

studies only looked at data for a time frame of one to two years. This limited time frame

does not allow for the temporal effects to be examined. Secondly, a majority of the data

that was used for the previous studies were collected either before or immediately after

the 1995 mandate for VA facilities to reorganize. Therefore, these analyses may not fully

account for the change in structure and its effects on outcomes.

The final gap regarding this research centers around previous organizational

research. In the past, the stream of research that explored structural effects was not

grounded in management theory. One aim of this study is to fill the gaps in the previous

research by performing research that is grounded in the management theory. This can

help in explaining some of the decisions that are made in reference to the restructuring of

the VA's facilities and its subsequent effect on patient outcomes.
















CHAPTER 3
THEORY AND HYPOTHESES

In this chapter, I will draw upon contingency theory and the structure, process,

outcome (SPO) perspective to develop a conceptual model of the relationship between

rehabilitation bedservice units (RBUs) and the outcomes of stroke patients. The

discussion and support of this model includes a review of the relevant theoretical

literature and statement of research hypotheses.

Conceptual Model

The purpose of this study is to examine the relationships between VA facilities

that provide rehabilitation care and stroke patient outcomes. The proposed conceptual

model is presented in Figure 1.


CONTEXTUAL
Task Uncertainty
FIM-FRG

STRUCTURAL
Discharge
Specialization Functional
Staffing ratios
No of disciplines Independence
No of total beds Measure
Standardization CONTROL (FIM)
Type of RBU bed Structure Score
PT Tenure
Complexity Age of Facility
Type of facility Patient
Marital Status
Age
Year



Figure 1. Conceptual Model









The model, which draws from several organizational perspectives, asserts that the

Functional Independence Measure (FIM) scores of stroke patients who receive

rehabilitation in VA-sponsored RBUs are a function of the severity of the patient's

impairment and case mix. In addition, the FIM scores are a function of the availability of

various care providers, how the degree of standardization of protocols used for

administering rehabilitation services, and the types of other services that are being

offered by the facility in which the RBUs are located. Finally, FIM scores are a function

of the fit between the contextual and structural characteristics of the facility in which

rehabilitation care is provided.

This model of facility variation and patient outcomes draws upon several

perspectives within organizational theory: the structure-process-outcomes (SPO)

paradigm, the open, rational, and natural systems perspectives, as well as contingency

theory. Each of these theories is briefly reviewed and linked in the following sections.

Open Systems Perspective

Just as the name implies, the open systems perspective states that organizations

are not fixed, closed systems. Instead, Hall (1972) states that these organizations "affect

their environment and are affected by them" (p.4). An open systems view of

organizations suggests that organizational environments shape internal structures and

boundaries (Scott, 1998). In other words, organizations rely on an exchange of inputs

and outputs with their environment (Katz & Kahn, 1978; Pfeffer & Salancik, 1978). The

Veterans Health Administration (VHA), a government healthcare system, has influences

from the external environment, including the private sector and the overall federal

government. For example, the VA decided in 1995 to adopt an organizational form that

resembles those that were proven successful in the private sector (i.e., integrated delivery









networks (IDNs)). The open systems perspective allows for analysis of how this external

force has affected outcomes in the VA's healthcare system.

Rational Systems Perspective

The rational systems perspective looks at organizations as being created for

specific purposes. According to March and Simon (1958), organizations have very

specific structures. These structures are constructed to seek specific goals Etzioni (1964).

The rational systems perspective would best describe the structure of the healthcare for

the VA prior to the reorganization mandate in 1995. Before the mandate, the main focus

of the healthcare system was the individual facility. The specific goal at that time was to

fund large medical centers to provide care. The result was that these large entities began

to compete against each other (Kizer et al., 1997).

Natural Systems Perspective

The natural systems perspective suggests that organizations not only strive to

reach specific goals, but according to Gouldner (1959), an organization seeks "to survive

and to maintain its equilibrium...even after its explicitly held goals have been

successfully attained" (p.405). The VA felt it could reach its goals of improved patient

care and at the same time survive threats from both internal and external forces by

creating the VISN structure. It was felt that by adding this new layer of management to

the current structure of the VA, a successful shift could be made from a focus on facilities

to the populations that these facilities serve (Kizer et al., 1997). This new structure

would also negate the competition between facilities by creating local and regional

networks that emphasized a continuum of care.









Structure-Process-Outcomes Paradigm

The structure-process-outcomes (SPO) paradigm was conceptualized by

Donabedian and is often used to assess the quality of medical care that is provided to

patients (Donabedian, 1980). Structural measures are defined as the "human, physical,

and financial resources that are needed to provide medical care" (p. 81). These

characteristics are fairly stable. Process measures are the procedures that are performed

by practitioners (Gustafson & Hundt, 1995), while outcomes are changes in the health

status of the patient as a result of the consumption of the medical care. There are two

types of outcomes: technical and interpersonal. Technical outcomes include the physical

and functional aspects of care, while interpersonal outcomes deal more with the patient's

satisfaction with the care received (Donabedian, 1980). This study only focuses on

technical outcomes of rehabilitation services.

This paradigm suggests that structure affects processes directly and outcomes

indirectly via processes. VA researchers have used this paradigm to explain

rehabilitation outcomes in previous research (refer to Table 3). This study will follow the

same model. The characteristics of the facility (structure) where rehabilitation services

are provided are believed to have an effect on the FIM scores of the stroke patients at

discharge (outcomes).

Contingency Theory

Contingency theory (Van de Ven & Drazin, 1985; Donaldson, 2000) is helpful in

assessing whether the organizational form that was chosen as a result of the

implementation of the VISN structure was appropriate for the improvement of patient

outcomes. The theory states that a fit between an organization's context and structure

effects performance. A match between the context and structure, indicates better









outcomes for the patient. In addition to the fit, contingency theory provides a perspective

for understanding how patient outcomes can be related to the organization of the facility

in which care is provided. It could also be helpful in explaining how individual

healthcare facilities within the VA system have adopted their internal structure to match

that external environment in which the facility is located (Astley & Van de Ven, 1983).

There are two preliminary assumptions that underlie contingency theory: (1) there

is no best way to organize and (2) any way of organizing is not equally effective

(Galbraith, 1973). These two assumptions challenge the previous work of organizational

researchers. The first assumption challenged the work of administrative theorists who

tried to develop principles that were applicable to all organizations in all times and places

(Scott, 1998). These efforts fail to recognize that there are a variety of tasks that are

performed in organizational settings. By attempting to develop principles that are

applicable to all organizations in all times and places, these administrative theorists have

overlooked the variance that is explicit in these different tasks.

The second assumption challenged the work of economists in their development

of the theory of the firm (Scott, 1998). Economists believed that organizational structure

does not have any effect on organization performance. However, this belief began to

change after Williamson's research showed that organizational structure does have a role

to play in performance (Williamson, 1985). The role entails how close the decision

maker is relative to the subordinates, in terms of location on the organizational hierarchy,

that the decision is made. A close proximity between the two means the decision maker

should be better informed because the is "closer" to activities of the subordinates and

thus, possesses more information. Therefore, a more informed decision could result in an









improvement in performance. For this study, the implementation of the VISN structure

should improve patient outcomes because he decision maker is now "closer" to the

providers.

A third assumption was posited based on Lawrence and Lorsch's (1967) argument

that the unit in the facility may face different external demands than the facility as a

whole. Scott (1998) built on this idea and posited the assumption that the best way to

organize is contingent upon the environment in which the organization relates. He felt

that this assumption best represented the beliefs of contingency theorists. For this

particular study, the unit that would face these varying external demands are the

individual RBUs.

Structure and Structural Forms

As stated previously, for contingency theory, structure is one of the determinants

of performance (Burs & Stalker, 1961; Woodward, 1965; Lawrence & Lorsch, 1967).

There are two possible structural forms that an organization can take in response to its

environment based on the type of task that has to be performed. The first of these

structural forms is mechanistic. Mechanistic forms are more effective for environments

that are relatively simple and stable, have tasks that are performed in a routine fashion,

and have a high percentage of its workforce classified as nonprofessional (Burs &

Stalker, 1961; Lawrence & Lorsch, 1967; Perrow, 1967; Thompson, 1967). Quoting

Burs and Stalker (1975), mechanistic systems are also characterized by:

* the specialized differentiation of functional tasks into which the problems and tasks
facing the concern as a whole are broken down;

* the abstract nature of each individual task, which is pursued with techniques and
purposes more or less distinct from those of the concern as a whole; i.e., the









functionaries tend to pursue the technical improvement of means, rather than the
accomplishment of the ends of the concern;

* the reconciliation, for each level in the hierarchy, of these distinct performances by
the immediate superiors, who are also, in turn, responsible for seeing that each is
relevant in his own special part of the task;

* the precise definition of rights and obligations and technical methods attached to each
functional role;

* the translation of rights and obligations and methods into the responsibilities of a
functional position;

* hierarchic structure of control, authority and communication;

* a reinforcement of the hierarchic structure by the location of knowledge of actualities
exclusively at the top of the hierarchy, where the final reconciliation of distinct tasks
and assessment of relevance is made


* a tendency for interaction between members of the concern to be vertical, i.e.,
between superior and subordinate

* a tendency for operations and working behavior to be governed by the instructions
and decisions issued by superiors;

* insistence on loyalty to the concern and obedience to superiors as a condition of
membership

* a greater importance and prestige attaching to internal (local) than to general
(cosmopolitan) knowledge, experience, and skill (p.41).

The second of these structural forms is organic. This organizational form is likely

to be more effective when the environment is complex and dynamic, tasks are performed

in a nonroutine manner, and a high percentage of the workforce is professional (Burns &

Stalker, 1961; Lawrence & Lorsch, 1967; Perrow, 1967; Thompson, 1967). This would

tend to describe the environment for stroke rehabilitation. Quoting from Burs and

Stalker (1975) once more, organic management systems are also characterized by:









* the contributive nature of special knowledge and experience to the common task of
the concern;


* the "realistic" nature of the individual task, which is seen as set by the total situation
of the concern;


* the adjustment and continual re-definition of individual tasks through the interaction
with others;


* the shredding of "responsibility" as a limited field of rights, obligations and methods
(problems may not be posted upwards, downwards, or sideways as being someone
else's responsibility);


* the spread of commitment to concern beyond any technical definition;


* a network structure of control, authority, and communication. The sanctions which
apply to the individual's conduct in his working role derive more from presumed
community of interest with the rest of the working organization in the survival and
growth of the firm, and less from a contractual relationship between himself and a
non-personal corporation, represented for him by an immediate superior;


* omniscience no longer imputed to the head of the concern; knowledge about the
technical or commercial nature of the her and now task may be located anywhere in
the network; this location becoming the ad hoc center of control authority and
communication;


* a lateral rather than a vertical direction of communication through the organization,
communication between people of different rank, also, resembling consultants rather
than command;


* a content of communication which consists of information and advice rather than
instructions and decisions;

* commitment to the concern's tasks and to the "technological ethos" of material
progress and expansion is more highly valued than loyalty and obedience;

* importance and prestige attach to affiliations and expertise valid in the industrial and
technical and commercial milieux external to the firm (p.43).









As mentioned earlier in this chapter, prior to reorganization in 1995, the structure

of the VHA could be considered mechanistic because the focus was on the funding of

care for facilities via large medical centers. Initially, it may have been simple to have this

organizational form because the VA's only mission at its inception was to provide care to

veterans with service-connected disabilities (Kizer et al., 2000). Presently, the VA has

five missions that makes providing healthcare much more complicated. That is why a

change to the VISN structure, and thus a change to a more organic form, seems to be the

best choice. According to Kizer et al. (1997), allowing individual VISN management to

choose how to structure the organization for their respective regions would make it easier

for the VA to move from focusing on facilities to populations by creating a system of

regional networks whose main goal was to provide a continuum of care.

There have been a variety of ways in which structure has been conceptualized and

many have shown to have an effect on performance. In their book that studied the

performance of hospitals, Flood and Scott (1987) defined structure in terms of four

dimensions: differentiation, coordination (or integration), power, and staff qualifications.

Differentiation and coordination are two variables that are fundamental to contingency

theory (Lawrence & Lorsch, 1967) and should probably be used in any study that

involves measuring structure. Both are important for organizational performance.

Differentiation is concerned with how the specialized areas of the organization are

grouped to perform a needed task. In general, most organizations are organized by

function. These functions represent the various professions that are needed in order for

the organization to perform. They also interact mostly with others in their department

(Charns & Tewksbury, 1993). Having organizations grouped by specialty has the









advantage of making it easier for the profession represented to maintain professional

standards. Also, it is easier for the department to maintain expert knowledge.

Integration involves the coordination of the activities that are performed in an

effort to have optimal performance. In general, when you have organizations that are

organized along functional lines, the level of coordination is very important. It is

important because if the level of interdependence is low, then the quality and efficiency

of the work that is being performed may be sacrificed (Charns & Tewksbury, 1993).

Traditionally, in health service organizations, there has been more emphasis on

differentiation by function than the integration of the functions. However, more recently

things are beginning to change more towards greater coordination of the functions.

Aside from differentiation and integration, there are other structural variables that

have also shown to have an effect on performance. They are size and heterogeneity

(Sicotte & BWland, 2001); staff diversity and capacity (Fennell et al., 2000); job

assignment, hierarchy, and closeness of supervision (Rohrer et. al, 1993); participation

and formalization (Alexander & Randolph, 1985); and many others.

Regarding the structure of the VA hospitals, many of the same variables are used

to conceptualize structure (see Appendix C). Most of the classifications in previous

studies are variables that Zinn and Mor (1998) refer to as immutable classifications. This

classification is associated more with the structure-process-outcomes paradigm

(Donabedian, 1980). This study will depart from the previous studies when

conceptualizing structure and will focus more on the strategic dimensions of structure

that are associated more directly with the contingency theory (Zinn & Mor, 1998).









Uncertainty and Context

Since the organic form of an organization results from an unstable environment,

the issue of environmental uncertainty plays an important role. There are two types of

uncertainty: task and environmental. Task uncertainty is defined as the difference

between the amount of information required to perform a task and the amount of

information that the organization possesses. Therefore, the greater the task uncertainty,

the greater the amount of information is that required in order for decision-makers to

execute a task at a given level of performance. It is this uncertainty that prevents all tasks

from being routine (Galbraith, 1973) and all organizations from adopting a mechanistic

form. This holds true for rehabilitation care. Since each stroke patient differs in their

responses to similar therapies, there is a need for more information regarding the patient

before a course of therapy can be prescribed versus if all patients responded to all

therapies the same way.

Environmental uncertainty is created when there are scarce resources and a lack

of information regarding environmental fluctuations, availability of exchange partners,

and available rates of exchange in an interoganizational field (Cook, 1977). Bums and

Stalker (1961), as well as Lawrence and Lorsch (1967), were among the first researchers

to mention that this uncertainty could play a role in the types of forms that were chosen

by organizations. They found that when greater uncertainty was present, a more loose

structure was more effective. The "loose" structure is believed to be the organic form.

To address the issue of uncertainty in the environment, it is suggested that the

organization consider creating slack. Slack is defined as an unused resource. It reduces

the information processing demands by reducing the interdependence between two or

more groups. However, an organization must be cautious when using slack resources









because they have an associated cost (Galbraith, 1973). If the slack resources are too

excessive, there is an increase in cost to the organization, which likely will be shifted to

the consumer.

Uncertainty is part of the conceptualization of the context construct that is key to

testing contingency theory. However, context may be conceptualized in a number of

different ways. Galbraith (1973), as well as Lawrence and Lorsch (1967), define the

environment in terms of the task. They mention how the environment is characterized by

the amount of complexity and uncertainty it poses for the organization (Scott, 1998).

Both dimensions of the environment are defined similarly in the contingency theory

literature. For example, Drazin and Van de Ven (1985) define task uncertainty as the

amount of task variability and task difficulty. Alternatively, Sicotte and B61and (2001)

define task complexity in the same manner; the interaction between task variability and

task difficulty. This study will only refer to task uncertainty and define it as the

interaction between task difficulty and variability. It should be noted that in this study, as

well as other contingency theory studies, the environment is sometimes encompassed by

the larger construct referred to as context. This is a broader construct that also includes

intraorganizational variables, such as size of the facility (Donaldson, 2000).

In classical contingency theory, task difficulty is conceptualized as the

analyzability or predictability of the work encountered (Van de Ven, 1976). Also in its

classical form, task variability is conceptualized as the number of exceptions encountered

while performing the task (Van de Ven, 1976). When contingency theory is used for

medical studies, other medically related conceptualizations for both task difficulty and

variability have been used. For example, Sicotte and B6land (2001) posit that task









difficulty could be defined as illness severity because severity "governs the consulting

and prescription processes" (p.168), while task variability is defined as patient casemix.

This alternate definition is justified on the premise that the proportion of infrequent types

of patients causes uncertainty for the provider due to the infrequency in which the

provider has to treat this type of patient.

The Concept of Fit

Fit is the central concept of contingency theory. Donaldson (2000) explains that

fit is important because "it explains variations in organizational performance,

organizational change, and associations between context and structures" (p.185). The

associations between context and structures are what constitute fit; the higher the

association between the two constructs, the better the fit. Organizations that have a better

fit exhibit higher organizational performance. But just as a fit between structures and

context may lead to higher performance, a misfit can have a negative effect on

performance (Galbraith, 1973). If an organization has a misfit between its context and

structure, a change in the organizational structure could facilitate an improvement in

performance. The context could also be changed for a better fit, but it is more difficult to

change context than change the structure. Specifically for this study, a better fit between

the contextual and structural variables should result in better outcomes for the stroke

patients receiving rehabilitation care in the VA.

Issues with Fit

One issue with testing fit involves determining the best empirical measures of

performance. Traditionally, in the manufacturing and retail sales industries, performance

is measured by return-on-investment or profit margin (Young et al., 2001). These are

plausible options for the healthcare industry if financial performance is the primary focus.









Aside from financial measures, clinical measures such as a patient's assessment score

(i.e., FIM score) or the change in the health of a group or population (i.e., the community

used fewer services) could be used. These measures help researchers identify how fit

effects the patient. Using a financial measure does not properly assess the impact of fit

on the patient. This study uses the patient's discharge FIM score as the empirical

measure of performance.

Another issue involves the level at which the fit of the organization should be

assessed. Quoting Lawrence and Lorsch (1967), they suggest that when assessed, fit

could be analyzed at two levels:

* the structural features of each organizational subunit should be suited to the specific
environment to which it relates; and

* the differentiation and mode of integration characterizing the larger organization
should be suited to the overall environment within which the organization must
operate (p. 113).

By assessing fit at both the unit and facility levels, it is possible that more than

one indicator would be needed. For example, with regards to the healthcare industry, the

unit's measure of fit could be the change in a patient's functional status as a result of a

change in the internal structure of an organization. Alternatively, the measure of fit for

the organization could be some financial indicator to see if the reorganization resulted in

an increase in profit. This study will focus solely on the fit-performance relationship on

the subunit level due to the scope of this study.

There are a number of studies in the healthcare literature that were interested in

examining how fit affected performance (see Appendix D). Each of these studies

examined fit at the subunit level. Schoonhoven (1981) looked at fit for acute care









hospital operating room suites while Argote (1982) used emergency units as the subunit

setting. Finally, Alexander and Randolph (1985) wanted to test fit and its effect on

performance in the nursing units. This study continues this trend by performing fit

analyses at the RBU level. However, unique to this study will be the fact that the

independent variables are a mixture of both quantitative and qualitative measures. In the

previous studies, a majority of the variables specified were qualitative.

Critique of the Contingency Theory

Although it is a very popular theory, contingency theory does have it opponents.

Schoonhoven (1981) criticizes the theory on a number of levels. Her study reported five

reasons why previous attempts at using contingency theory have provided mixed support.

They are:

a lack of clarity regarding the development of the set of interrelated propositions

the notion that empirical interactions are being predicted is often times blurred

theoretical statements fail to provide clues about the specific form of the interaction
intended

* the operational and computational procedures used imposes assumptions on an
already imprecise conceptual framework

* the assumptions of symmetrical effects are hidden in the theory's language.

Except for the first critique, which deals with the legitimacy of contingency theory as

a whole, the other critiques posited by Schoonhoven (1981) center around the

conceptualization of the fit term. When it is conceptualized as an interaction of context

and structure, fit has a number of limitations. The first is that the intercorrelations among

context and structure variables could lead to nonorthogonal factor designs. When this

situation arises, there are difficulties in decomposing and assessing the differences










between the interaction and intercorrelation effects on a dependent variable in such

situations (Green, 1977). Secondly, there could be a restricted range of structural

variation existing within each level of context. Thirdly, if the explanatory variables are

either polychotomized or dichotomized, there could be a loss of information that reduces

the ability of the model to detect interactions (Miller, 1981; Pierce et al., 1979). Finally,

significant interaction terms may result solely as a function of the scale of measurement

of the dependent variable. Monotonic or logarithmic transformations of the dependent

variable may reduce the effect of the interaction to insignificant levels (Green, 1977).

As a solution to this issue, fit scores for this study were constructed using the

residual of the regression of structure and context (Hage, 1980). Figure 2 depicts a

typical interaction hypothesis of contextual heterogeneity and structural complexity on

organizational performance. The positively sloped line indicates combinations of context






Hg
Performance









X Low
SPerformance
1Zi
Homogeneous Heterogeneous
ORGANIZATIONAL ENVIRONMENT


Figure 2. Interaction of Context and Structure on Performance










and structure that leads to high performance, while the negatively sloped line indicates

combinations of context and structure which lead to lower performance.

Figure 3a shows the deviation from the high performance line. The distance a

facility falls from the predicted line equals the amount of decrease in performance.

Figure 3b plots the absolute value of the residual fit score against performance. As

expected, facility A would have lower performance than facility B because it is farther

away from that predicted line in Figure 3a. If the slopes on the graphs were switched,

i.e., the structure versus context graph had a negative slope while the performance versus

deviation score was positive, this would indicate a misfit. This study will use the same





a -a (a)





0 L Organization B
SIOrganization A
--- Amount of deviation

SLow ORGANIZATIONAL CONTEXT High


SOrganizationA (b)
0



SOrganization B



Low PERFORMANCE High


Figure 3. (a) Deviation from Context-Structure Relationship and (b) Expected
Relationship Between Absolute Values of Deviation Score and Performance.









techniques as described above but more emphasis is on whether the slope of performance

(FIM score) versus residual is negative or positive, indicating a good or bad fit,

respectively, between each structure/context combination.

Another issued raised against the use of contingency theory was the

conceptualization of the variables. Initial conceptualizations of the variables were

thought to be too abstract and were not considered decision variables (Argyris, 1972).

For example, constructs such as formalization, centralization, and specialization are

measured as scales and did not change that much. The issue is addressed in this study by

not having all conceptualizations of the independent variables as scalar.

Theoretical Convergence

The paradigms discussed above offer many different ways to view how

organizations affect performance. These paradigms can be merged with contingency

theory to provide a more robust model, which better explains the organization-

performance relationship. To help understand contingency theory better, Lawrence and

Lorsch (1967) provided a general framework that encompasses all three perspectives

from which organizations are viewed (i.e., rational, open, and natural). They offer that

contingency theory is viewed similar to the open systems perspective which Scott (1998)

defines as "systems of interdependent activities linking shifting coalitions of participants;

[and] the systems are embedded in...the environments in which they operate" (p.89).

Lawrence and Lorsch (1967) also posit that the rational and natural perspectives are used

to identify the organizations that adapted to their environments by using a mechanistic or

organic form. Just as the two organizational forms are at opposite ends of the continuum,

so are the perspectives. The rational perspective, which is characterized by organizations

that pursue specific goals and have formalized structures, would represent the








mechanistic form. The organic form would be represented by the natural perspective,

which sees organizations as pursuing multiple interests with an informal structure (Scott,

1998). Unlike Etzioni's (1964) model, which views rational and natural systems as

different aspects of the same organization, Lawrence and Lorsch's (1967) view sees the

two perspectives on the ends of the same continuum as different kinds of organizations.

They choose this view because they believe that the open system perspective is a more

comprehensive framework, while the natural and rational perspectives are partial views

of organizations (Scott, 1998).

The structure, process, and outcomes (SPO) paradigm and contingency theory

offer complimentary perspectives on organizational structure (Zinn & Mor, 1998). Both

are interested in how structure affects outcomes or performance. Contingency theory is

concerned with how structure and context together affect performance while the SPO

models are concerned with the effects of an organization's structure on performance,

although these effects are indirect. When studying quality of care, performance can be

considered the equivalent of the outcomes of the patient since it is the change in patient's

functional status that is the dependent variable of interest with the SPO paradigm. Since

most of the studies done using contingency theory in the healthcare sector regarded the

performance of the organization, merging the SPO paradigm with contingency theory

will provide a basis for allowing the focus of the dependent variable to shift from the

performance of the organization to the performance of the individual. This is what

Rohrer et al. (1993) exhibited in their study when they used contingency theory to test the

structural and contextual effects on patient's functional status.









The SPO model and contingency theory differ in the manner in which the

organization is conceptualized. In the SPO model, focus is on the organizational capacity

and capability (Zinn & Mor, 1998). Examples would include the human, physical, and

financial resources that are needed to provide care (Donabedian, 1980). Regarding

contingency theory, there is more focus on the structure in terms of the formalization,

specialization, standardization, complexity, and centralization. Healthcare examples for

this type of conceptualization would include the use of clinical protocols, the use of

interdisciplinary teams, and locus of control (Zinn & Mor, 1998). Even though there are

differences, the theoretical convergence of contingency theory and the SPO paradigm

offer an improved theory in which to study the effect of fit between context and structure

on stroke patient's FIM scores.

Hypotheses

This study builds upon the previous integrative perspectives for contingency

theory and the SPO paradigm and suggests organizational and contextual factors that may

have an effect on patient outcomes. First, the study will determine which variables have

a significant effect on performance. Then fit will be analyzed to see if it has either a

positive or negative effect on performance. Hypotheses for each of the relationships are

discussed in the following paragraphs.

Context and Performance

As previously described in detail, contingency theory posits that contextual

factors are one of the two constructs that have an effect on an organization's

performance. Context includes the environment in which the organization is located as

well as the nature of the work that is performed. There have been only a few studies in

the healthcare literature that exclusively examined the effect of context on performance









(Sicotte & BWland, 2001; Rohrer et al., 1993; Leatt & Schneck, 1983). Of these studies,

only the Sicotte and Brland (2001) and Leatt and Schneck (1983) studies actually

conceptualized context in terms of task uncertainty. This study will conceptualize

context in a similar fashion. The variable for task uncertainty will be conceptualized

such that higher levels of uncertainty are indicated by lower scores. It will also follow

the same conceptualizations as Rohrer et al. (1993) when they had the dependent variable

defined as a clinical outcome. Therefore, it is hypothesized that as the task uncertainty

increases, the stroke patient's FIM score will decrease because more difficult and

variable tasks are performed on patients with the worse conditions.


Hypothesis 1: As the task uncertainty increases, the patient's FIM score will increase.


Structure and Performance

Contingency theorists describe the elements of organization structure in terms of

specialization, standardization, and complexity. Specialization is defined as the degree to

which the tasks are subdivided into jobs for the different disciplines represented (Zinn &

Mor, 1998). This study does not focus on the actual task subdivisions, but instead seeks

to capture the number of staff for each discipline available per patient load to determine if

the staff can perform its designated duties and improve patient care (Rohrer et al., 1993).

As staffing increases, the more expertise is available to be potentially utilized by the

patient. This is believed to improve FIM scores. Since staffing is conceptualized as a

ratio, an increase in staffing indicates a decrease in the staffing ratio.


Hypothesis 2: The higher the staffing ratio for the various disciplines, the lower the

patient's FIM score.









Along the same lines, there are various disciplines that are necessary to provide

the type of multidisciplinary care for stroke rehabilitation that is needed to improve

outcomes (Langhome & Duncan, 2000). The more diverse the rehabilitation team, the

more improved the patients outcomes should be. This is true because representation from

different disciplines can bring about new ideas of how to treat the more difficult patients.


Hypothesis 3: The higher the number of disciplines represented, the higher the patient's

FIM score.


Finally, in the classification of specialization, is the issue of size of the facility in

which the RBU is located. According to contingency theory, size is negatively related to

performance because as the size of the organization increases and the structure remains

constant, the facility moves out of the fit range into a misfit range. This would have a

negative impact on performance (Donaldson, 2000). But this hypothesis does not hold

true for RBUs. They have tended to be associated with larger inpatient facilities that are

generally located in large urban settings (Reker et al., 2000). Therefore, larger facilities

will possess the specialized units and have better patient outcomes.


Hypothesis 4: As the size of the facility increases, the patient's FIM score increases.


The next classification of structure emphasizes the standardization of the work

that is performed and how it relates to patient outcomes. Standardization refers to the

extent to which similar work is performed in a uniform manner (Zinn & Mor, 1998).

Regarding this study, although the care that is provided in the RBUs are provided by

multidisciplined teams (Hoenig et al., 2002; Langhorne & Duncan, 2000), these teams are









not concentrated solely on rehabilitation in an acute RBU. Therefore, the rehabilitation

care that is provided in a subacute RBU should be better than the care provided in an

acute RBU. By treating rehabilitation and non-rehabilitation patients, the providers in the

acute RBUs may be less likely to adhere to specific rehabilitation protocols because they

have to adhere to protocols for other medical conditions. Attempting to remembering the

details of multiple protocols could prove to be difficult.


Hypothesis 5: If the patient receives rehabilitation care from an acute RBU, their FIM

scores will be lower when compared to receiving care in a subacute RBU.


Finally, the complexity of the organization will be examined. Complexity

involves the number of discrete units and their arrangement in the organization (Zinn &

Mor, 1998). As the facility offers more services, it becomes more complex (Hoenig et

al., 2000). For example, being in an extended care facility would be considered the least

complex facility, being in an inpatient facility would be the second least complex, and

being in both an inpatient and extended care facility (denoted as BOTH in this study)

would be the most complex. Complexity is thought to expose rehabilitation patients to

more disciplines and thus more expertise on how to improve the patient's functional

status.


Hypothesis 6: When compared to being in an extended care facility, a patient in an

inpatient facility will have higher FIM scores.









Hypothesis 7: When compared to being in an extended care facility, a patient in both an

inpatient and extended care facility for the same episode of care, will have higher FIM

scores.


Context and Structure

There has been much debate regarding the relationship between the context in

which an organization is operating and the structure of the organization. Previous studies

that examine this relationship have had mixed results. One explanation can be attributed

to the fact that there are many definitions of context.

In terms of context being conceptualized in the traditional terms of contingency

theory, Woodward (1965) found a link between context and structure, but the link was

not proven empirically. There was subsequent research performed that indeed backed

Woodward's conclusion. Gerwin (1979) found that greater task uncertainty lead to less

structured roles. Additionally, Miller et al. (1991) utilized a meta-analysis to find a

robust relationship between the context and structure. In the previous studies that

unsuccessfully tried to replicate Woodward's (1965), most used a conceptualization of

context that was different from Woodward's. These previous studies show why it is

important to test the relationship between context and structure in this study. For this

study, to test the relationship between context and structure, a fit score will be generated

by utilizing the residual of the regression of structure on context (Hage, 1980). The fit

score of each context-structure combination will tell how well the two measures are

aligned in an effort to facilitate high performance.









Hypothesis 8: As a result of organizational changed in the facilities, the context and

structure will now have a better fit.


Fit and Performance

As stated previously, the fit between an organization's structure and its context

has an effect on performance (Drazin & Van de Ven, 1985; Fry & Slocum, 1984; Dewar

& Werbel, 1979). This study suggests fit will have a positive impact on patient's

outcomes, while misfit will have a negative impact.


Hypothesis 9: The better the fit between context and structure, the better the patient's

FIM score.


Performance

For this study, performance is important because it determines how much the

stroke patients have improved their functional status from admission to discharge.

According to contingency theory, fit effects performance. In this study, performance is

defined as the discharge motor FIM score of the patient. There are facilities where

performance is higher than others. The structural characteristics of the high performing

facilities are thought to be different than the low performing facilities.


Hypothesis 10: When compared to low performing facilities, high performing facilities

will differ in terms of their structural characteristics.


Alternative Hypotheses

Variation in the FIM scores of stroke patients who receive rehabilitation services

in the VA's healthcare network may be associated with factors other than the structural









and contextual variables named in the above hypotheses. There are many ways in which

to conceptualize these two constructs, so the conceptualizations that were chosen were

those that were deemed the best representation of the environment in which rehabilitation

services for stroke patients is provided. In this dissertation, I include several control

variables that measure organizational and patient characteristics that offer alternative

explanations to the already presented hypotheses.

Structural Characteristics

All of the various disciplines that provide services do not have the same level of

competence. This can be attributed to the fact that providers have various levels of

experiences when it comes to providing rehabilitation (Morrison et al., 1992). Some

providers have very little experience while others have been providing services for many

years. The experience gained over time assists the providers in being able to handle more

severe stroke patients (Kostiuk et al., 1989).

Another structural issue is the age of the facility. Facility age is considered to be

important because it is considered an antecedent of organizational change (Lee &

Pennings, 2002; Kraatz & Moore, 2002). The older the organization, the more inertia it

possesses. This inertia may have a negative effect on the adoption of new structural

forms (Hannan & Freeman, 1984). With regards to this study, the older the facility in

which rehabilitation is provided, the more resistance there would be for the

implementation of the new VISN structure which may affect the performance of the

stroke patient's rehabilitation.

Patient Characteristics

Since this study has a dependent variable that is at the individual level and

independent variables at the facility level, there have to be control variables that take into









account the variation in the patient population sampled for this study. These control

variables for this study include the patient's age, race, and marital status.

Age is included in this model as a control variable because it is the most

important risk factor for developing a stroke as well as having an effect on stroke

recovery (Kulger et al., 2003). Paolucci et al. (2003) found that even when severity of

stroke was the same, the older the stroke patient was, the greater the disability in

activities of daily living (ADL) and mobility, as well as minor results of rehabilitation

treatment. In addition, Ergeletzis et al. (2002) noted that stroke patients above 80 years

of age were less likely to return home from a stroke when compared to stroke patients

under 80 years of age.

Race was included as a control variable because it captures the potential for

disparity that occurs in providing healthcare. African-Americans tend to suffer from

stroke more in terms of having greater incidence and mortality (Otten et at, 1990). They

not only suffer a higher rate of mortality, but survival tends to result in more severe

residual motor impairment (Homer et al., 2003; Homer et al., 1991) as well as worse

functional status after the stroke (Homer et al., 1991; Sacco et al.; 1991) when compared

to whites. For those stroke patients who are admitted to nursing homes, there are still

racial disparities. According to Christian et al. (2003), Asain/Pacific Islanders, Blacks,

and Hispanics eligible for anticoagulant therapy were given the drugs less often than non-

Hispanic whites.

The social support that a stroke patient receives after a stroke can have an effect

on the outcomes of the patients' rehabilitation. This study will focus on the social






49


support measure of marital status. Marital status has previously shown to have an

important role in explaining differences in subjective functioning (Koukouli et al., 2002)..

Finally, a control variable was added to control for the multiples years in which

the study takes place. This variable is important because after the implementation of the

VISN structure, it is believed that patient's FIM scores should improve over time. By

including this dichotomous control variable in the model, the improvement in patient's

FIM scores can be compared to see if the improvement does indeed occur.









Summary of Hypotheses:

Research Question 1: Does the context in which rehabilitation is provided
have an effect on stroke patient outcomes?

Hypothesis 1: As the task uncertainty score of the stroke patient increases, the
patient's FIM score will increase.

Research Question 2: Does the structure of the rehabilitation facility effect
the stroke patient outcomes?

Hypothesis 2: The higher the staffing ratio for the various disciplines, the lower
the patient's FIM score.

Hypothesis 3: The higher the number of disciplines represented, the lower the
patient's FIM score.

Hypothesis 4: As the size of the facility increases, the patient's FIM score should
increase.

Hypothesis 5: When compared to receiving care in a subacute RBU, a patient
who receives rehabilitation care from an acute RBU will have a higher FIM score.

Hypothesis 6: When compared to being in an extended care facility, a patient in
an inpatient facility will have a better FIM score.

Hypothesis 7: When compared to being in an extended care facility, a patient in
both an inpatient and extended care facility, will have a better FIM score.

Research Question 3: Does the structure of the facility effect the context in
which rehabilitation is provided?

Hypothesis 8: The smaller the fit score, the more effect structure has on context.

Research Question 4: Does the fit between a rehabilitation facility's structure
and their context improve stroke patient outcomes?

Hypothesis 9: The better the fit between context and structure, the better the
patient's FIM score.

Research Question 5: Is there a structural difference between facilities with
higher patient outcomes and facilities with lower patient outcomes?

Hypothesis 10: When compared to low performing facilities, high performing
facilities will differ in terms of their structural characteristics.














CHAPTER 4
METHODS

This chapter presents an overview of the research methods used in this

dissertation. It includes sections that describe (1) the data source and sample; (2)

measures; and (3) the analysis strategy employed to evaluate the associations regarding

the fit between the context and structure of VA facilities and their effects stroke patient's

outcomes.

Data Source and Sample

This study uses two groups of data: patient and structural. Patient data come from

two sources. The first is the Functional Status Outcomes Database (FSOD). The FSOD

is a database that was developed by the VA, in conjunction with the Uniform Data

System for Medical Rehabilitation (UDSMR), to track the functional status of patients

who received rehabilitation care in either the VA's acute or subacute inpatient

rehabilitation units. In addition to the FSOD, the Patient Treatment File (PTF) also

provided information on the patient. It is the VA's inpatient file on all patients who

receive healthcare in the VA. These two databases are the main sources of data for this

study. Structural data primarily comes from the VA's Allocation Resource Center (ARC)

and Personnel and Accounting Integrated Data (PAID) database as well as two VA web-

based applications: the Planning Systems Support Group (PSSG) and the VISN Support

Service Center (VSSN). In addition, the PTF and FSOD provide structural data. This

study specifically uses data for fiscal years 1996-2002. The following sections describe









the data collection protocol procedures used in the FSOD and PTF, and discuss them with

respect to the current study

Survey Population

For this study, a patient in the FSOD is defined as a veteran who has had a stroke

and either been admitted or transferred to a RBU to receive inpatient rehabilitation.

Inclusion for this study involves the patient having an admission date, discharge date, and

a date of birth that is not equal to the admission or discharge date. Also, for each patient

record, there has to be a score entered for all of the subscales of the FIM. Finally, it is

possible for a patient to have more than one episode of care in this dataset.

Historical Background of the FSOD

The main source for data regarding rehabilitation studies is the Uniform Data

System for Medical Rehabilitation (UDSMR). It was developed as a means to quantify

disability and rehabilitation outcome. This database is the largest national registry of

standardized information on medical rehabilitation patients in the United States and

contains information on more than 4 million patients. Currently, more than 850

rehabilitation facilities and hospitals, both VA and non-VA, that use the UDSMR data

service. This represents approximately 70% of all comprehensive medical rehabilitation

facilities in the United States. The information in the database not only includes the FIM

measures, it also includes information regarding the patient population demographics,

facility characteristics, and source of payment (www.fimcontinuum.com).

Starting in 1989, VA rehabilitation units began to report data to the VA-UDSMR.

By 1994, 54 of the 62 units were enrolled (Bates & Stineman, 2000). In 1997, through a

cooperative agreement between UDSMR and the VA, an outcomes database was

developed for patients receiving care in its acute and subacute inpatient rehabilitation bed









units. This database was called the Functional Status Outcomes Database (FSOD).

With the creation of this database, VA clinicians now had the ability to track patient

outcomes from onset of a stroke in the inpatient setting through the continuum of care

until discharge (Cowper et al, 2004). For this study, only data from patients who

received rehabilitation in a formal rehabilitation bedunit will be used for analysis.

Historical Background of the PTF

The Patient Treatment File (PTF) is a set of national databases that contain

inpatient care data for the VA (VIREC, 2000). There are eight files which comprise the

PTF: the PTF Main, PTF Bedsection, PTF Procedure, PTF Surgery, PTF Extended Care

Main, PTF Extended Care Bedsection, PTF Extended Care Procedure, and PTF Extended

Care Surgery files. This study will only used data from the PTF Main and PTF Extended

Care Main files (Cowper et al., 1999). The information in the PTF Main file dates back

to fiscal year 1970. The other files were added in the mid-1980s.

The PTF Main file contains information about the patient's entire inpatient stay.

The data available ranges from a patient's demographic information (e.g., gender, race,

marital status, date of birth, etc.) to summary information about the inpatient episode

(e.g., date of admission, diagnosis, location where care was provided, length of stay, etc.)

(Cowper et al., 2002).

The PTF Extended Care file contains the same demographic and episode of care

information as the PTF Main but represents care provided in a VA nursing home care

unit (NHCU) or domiciliary. For those patients who received care in both an inpatient

and extended care facility in the same episode, it is possible to link information on a

patient who has received care in both settings. This study contains a number of these

cases.









Background on the ARC, PSSG, and VSSC

Allocation Resource Center (ARC). The Allocation Resource Center (ARC) is

a clinically-focused health systems information and management group that assists VA

policy and operations management functions. Services provided include developing,

maintaining, and using decision support patient-specific workload and expenditure

databases. A major responsibility of the ARC is the development, implementation, and

maintenance of the management information system that supports the VHA's budget

process. It also supports VA Headquarters and field offices through the simulation and

evaluation of alternative budget scenarios, management policies, and decisions. It has

been in operation since 1982 (vaww.arc.med.va.gov). For this study, the ARC was used

to obtain unit-level staffing information that was specific to the rehabilitation services

offered by the VA.

Planning Systems Support Group (PSSG). The VA Planning Systems Support

Group (PSSG) is a field-based group of the VHA Office of the Assistant Deputy Under

Secretary for Health with the primary mission of supporting national policy objectives

and strategic goals. Since 1983, this group has been involved in the development,

maintenance, distribution and support of PC-based projection models. It also has

pioneered the use of Geographic Information Systems (GIS) as a tool in VHA. Definition

of service area boundaries, market penetration, utilization analysis, and demographic

analysis are a few of the services that this project is called upon to perform in achieving

the goals mandated by VHA Headquarters. The primary goals established are:

enrollment, legislative initiatives, VHA priorities, Medicare and strategic planning. PSSG

has been involved with such national issues as development of network boundaries,

Medicare subvention, veteran enrollment, tracking current VHA service sites, and









determining geographic access to VHA services (vaww.pssg.med.va.gov). The PSSG

was used to determine characteristics of the facility in which rehabilitation services were

provided.

VISN Support Service Center (VSSC). Another on-line database that was used

to obtain facility characteristics was the VISN Support Service Center (VSSC). The

VSSC is a healthcare information and technical support organization serving both the

needs of the Deputy Under Secretary for Health for Operations and Management and the

VISN networks. It supports field operations in the areas of Information Management;

Capital Programs; Capital Asset Realignment for Enhanced Services (CARES); National

Veterans Service and Advocacy Program (NVSAP); and, Planning and Data Analysis

(klfmenu.med.va.gov).

Personnel and Accounting Integrated Data (PAID) database. The PAID

database is the major accounting system for personnel in the VHA. It contains

information regarding employee cost center, job title, salary, and education level (Hoenig

et al., 2001). The database was used in this study to acquire information regarding the

number of disciplines associated with rehabilitation that were available at each facility as

well as facility-level staffing information on specific disciplines.

Data Reliability and Validity

There were no studies found that have directly tested the reliability of the FSOD.

However, there are studies that have tested the reliability and validity of the FIM and

these studies can serve as a proxy for a lack of FSOD reliability studies because the

FSOD is based on the FIM (http://vawwl.va.gov/health/rehab/FSOD.htm). Regarding

internal consistency, studies by Granger and Hamilton (1996) and Dodds et al. (1993)

found that the Cronbach's alpha was high for both the admission and discharge FIM









measures. Both studies reported admission and discharge FIM alphas of 0.93 and 0.95,

respectively.

In addition to the internal consistency of the FIM being tested, there were also

studies that tested the inter-rater reliability. In a study that had patients from twenty-five

facilities who were assessed by physicians, nurses, and therapists, the mean kappa index

of agreement between ratings for each item in the instrument was .71 (Hamilton et al.,

1991). In a more recent study that looked at patients who were assessed by nurses, the

mean kappa coefficient of assessment was .65 (Haas et al., 2002). Overall, both studies

showed that the FIM has moderate to high inter-rater reliability.

Regarding validity, asking clinicians to judge the FIM's scope and ease of

administration tested the content validity. This assessment by the clinicians led to the

addition of the social adjustment and cognition items to the instrument. Also resulting

from the clinician's assessment was the expansion of the answer categories to include

modified and complete dependence (Keith et al., 1987).

For construct validity, admission and discharge FIM scores decreased with

increasing age. In addition, the FIM subscales discriminated differences between stroke

types. There was a small but significant difference in admission (p < .005) and discharge

(p < .05) FIM for stroke victims with right and left sided body involvement. Statistically

significant differences only were found in the Communication subscale (Dodds et al.,

1993).

The reliability of the PTF was tested by comparing patient level information in

the PTF with patient medical records (Kashner 1998). Higher agreement was found

between patient demographic characteristics (average K =.92) and inpatient principal









diagnoses (average K =.75). Only moderate agreement was found between inpatient

secondary diagnoses (average K =.62), inpatient bedsection location (average K =.53), and

outpatient clinic stops (average K =.46).

It should be noted that when a principal diagnosis of rehabilitation was

investigated for agreement, the kappa statistics was one of the lowest scores calculated (K

=.54). Although the kappa statistic had one of the lowest ratings, it would still be

considered to have moderate agreement. Also, the study will only used data that is

associated with the two highest inter-rater reliability scores: patient demographic

information and inpatient principal diagnoses. Therefore, the PTF is a reliable source of

data for this study.

A review of the literature shows that there currently are no studies that establish

the validity of the data contained in the PTF. In absence of these studies, Kashner (1998)

recommends that use of care data be constructed from a number of sources. This study

will be following this suggestion not only because of the recommendation of Kashner

(1998), but also because there are structural data that are needed that both the PTF and

FSOD do not provide.

Measurement

The following sections describe information about how each concept in this study

is operationalized and measured. A summary of study concepts, measures, and data

sources are provided in Appendix E. Each measure used in the study is described in the

following five sections: (1) stroke patient outcome (dependent variable); (2) context; (3)

structure; (4) fit; and (5) alternative explanations.









Stroke Patient Outcome

The objective of this study is to understand the association regarding the fit

between the context and structure of VA facilities and its association with stroke patient's

outcomes. Based on the body of research regarding the appropriate outcome measures

for stroke summarized in Chapter 2, the outcome measure that was selected for this study

is the discharge motor Functional Independence Measure (FIM).

As mentioned previously, the FIM is a good outcome measure for this study for a

number of reasons. First, the measure is the industry standard for rehabilitation. It is an

instrument that is used around the world to assess functional gains in patients who

required rehabilitation (www.fimcontinuum.com). Secondly, it measures functional

status at both admission and discharge from rehabilitation. The admission FIM can be

compared to the discharge FIM score to see exactly how much improvement has been

made in terms of level of independence (Linacre et al., 1993). Previous research has

shown that the admission FIM score is one of the better predictors of discharge FIM

score (Inouye, 2000). Finally, the motor score of the FIM can be translated from an ADL

measure to a categorical disability measure (Kwon, 2004). Therefore, for this

dissertation, the dependent variable will be operationalized as the discharge motor FIM

score of the patient.

Context

Context refers to amount of uncertainty in the tasks that are performed on the

stroke patient to improve their functional status. The tasks that are more difficult and

variable have a negative impact on outcomes (Sicotte & BWland, 2001; Rohrer et al.,

1993). For rehabilitation, the FIM-Functional Related Group (FRG) is the measure that

captures both the degree of difficulty and variability in the tasks needed in the treatment









of the various types of stroke patients (Bates & Stineman, 2000). Thus, task uncertainty

is operationalized as the FIM-FRG score for each patient. This score is based on

admission motor and cognitive FIM scores as well as the age of the patient. They are

coded so that the higher the score, the less severe the stroke (Stineman et al., 1997).

Structure

Three elements of the contingency theory will be used to conceptualize structure:

specialization, standardization, and complexity. The measures that are used to

operationalize each element are discussed below.

Specialization regards how the tasks are subdivided for the various disciplines

that provide services (Zinn & Mor, 1998). The manner in which the tasks are subdivided

is important because in rehabilitation, multidisciplinary teams are shown to produce

better outcomes in stroke patients (Langhorne & Duncan, 2000). If these teams are not

staffed with the appropriate levels of the clinical staff, the patient may not realize their

maximum functional gain.

In general, staffing is conceptualized in terms of the ratio of the number of full-

time equivalent employees (FTEEs) to some measure of scale and quantity utilization

(Snail & Robinson, 1998). Examples of such scale and quantity utilization variables

include total number of beds and average daily census, among others. For this

dissertation, the staffing ratio is conceptualized as the ratio of the number of FTEEs for

each rehabilitation unit to the volume of stroke patients for that facility. Staffing ratios

will be calculated for doctors, physical and occupational therapists (PTs and OTs,

respectively), licensed practical nurses (LPNs), speech language pathologist, nurse

practitioners, and registered nurses (RNs). The details on how the stroke volume is

calculated are offered later in this chapter.









Since each facility has the discretion of organizing its facility in the manner

deemed sufficient for survival in the local market, there is variation in the how the

multidisciplinary teams are organized. Recent research has shown that the more diverse

the rehabilitation team, the more improved the outcome for the stroke patient should be

(Langhome & Duncan, 2000). Therefore, the number of disciplines will be

operationalized by summing each discipline represented for each individual facility.

Finally, in the classification of specialization, the size of the facility is an

indication of the types of resources that are available. In addition to resources, larger

facilities have tended to be associated with larger inpatient rehabilitation facilities (Reker

et al., 2000). In this study, size is determined by the total number of beds. This measure

is computed by calculating the average number of operating beds in the facility for the

fiscal year.

The next concept of structure is concerned with the standardization of the

processes that are performed on the rehabilitation patients. Standardization refers to the

extent to which similar work is performed in a uniform manner (Zinn & Mor, 1998).

Regarding this study, although the care that is provided in the RBUs is provided by

multidisciplined teams, the teams located in an acute RBU are not dedicated solely to

rehabilitation. To capture the difference between receiving care in an acute versus

subactue RBU, a dichotomous variable will be created to compare being discharged from

a subacute RBU with being discharged from an acute RBU.

The last concept of structure to be studied is complexity. Previous research in this

area shows that as facilities offer more services, they are considered more complex

(Hoenig et al., 2000). Therefore, complexity will be operationalized as the type of









facility in which care is provided. Dichotomous variables will be created to represent if

rehabilitation care was provided in an inpatient setting, an extended care setting, or both.

Extended care is offered in domiciliaries, VA nursing homes, or community nursing

homes (VA Information Resource Center, 2000). As far as classification of least to most

complex facility is concerned, being in an extended care facility would be considered the

least complex facility, being in an inpatient facility would be the second least complex,

and being in both an inpatient and extended care (BOTH) facility would be the most

complex.

Fit

In order to evaluate the hypothesis associated with fit (Hypothesis 13), residual

terms were created to represent the interaction between the structure and context. The

residuals equals the fit variable. Further detail on the conceptualization of the fit variable

will be given in the Analysis Strategy section of this chapter.

Control Variables

To account for possible alternative explanations for variation in stroke patient's

outcomes, the following control variables are included in the analyses. They are

categorized into three groups: structural, patient, and year.

Structural

Human capital. As rehabilitation clinicians gain experience with stroke patients,

they become more knowledge regarding treatment for patients with varying degrees of

stoke. As the clinicians become more knowledgeable, the outcomes of the stroke patients

should improve. In rehabilitation, it is difficult to capture the knowledge gained because

many of the disciplines that compose multidisciplinary teams also provide services

outside of the rehabilitation realm. The physical therapist is the only discipline that









devotes 100% of their time to rehabilitation. Current theoretical models of physical

therapists expertise reveal that this expertise comes the longer they work in rehabilitation

(Jensen et al., 2000; Jensen et al., 1992; Jensen et al., 1990). Therefore, in this study,

human capital is operationalized as the average tenure of the physical therapists for each

facility.

Facility age. As the facility increases in age, it becomes more resistant to

organizational change (Lee & Pennings, 2002; Kraatz & Moore, 2002). This would

indicate that older facilities in the VA would not be as quick to reorganize as younger

facilities following the 1995 mandate for reorganization. Hence, the age of the facility is

measured as the number of years the facility which renders rehabilitation services has

been in operation.

Patient

There are three control variables that control for the variation in the patient

population that receives rehabilitation for stroke in the VA's healthcare system. Patient's

age is operationalized as the category to which each patient belongs. The categories are

grouped in to eight groups, with each group having a nine year range with the exception

of the first group which ranges from 0-24 years of age. This measure categorizes patients

into eight age categories according to their chronological age. The race category groups

patients into one of six categories (Hispanic-Black, Hispanic-White, American Indian,

Black, White, and Asian). The racial categories will be taken directly from the PTF and

none of the groups will be grouped.

Finally, marital status is a categorical variable that assesses the marital status of

the patient when they receive rehabilitation care. The categories include single, married,

widowed, separated, and divorced. The patient is classified into one of six categories.









For all three categorical patient-control variables, they were converted to dichotomous

variables and a reference variable was chosen. This allows for comparison between

groups of the different categories on the discharge motor FIM.

Year

There is a control variable for the year in which the patient receives rehabilitation

care. Similar to the patient control variables, this variable will be converted from a

categorical variable to a dichotomous one with the first year as the reference variable.

Analysis Strategy

This section describes the analysis strategy used to test the hypotheses put forth in

this dissertation. The analysis strategy includes the following steps: (1) selecting true

stroke cases; (2) calculating the fit variables; (3) the regression technique for a

continuous performance measure; (4) the regression technique for dichotomous

dependent variable; and (5) determining variation in structural characteristics. Each of

these steps in the analysis strategy is described below.

Step 1 Selecting True Stroke Cases

In this dissertation, I am interested in look at how changes in the structure effects

stroke patients. When analyzing the structure of an organization, staffing levels of

various disciplines is important. The data sources that were described previously do not

provide staffing ratios at the subunit level. Therefore, these ratios have to be constructed.

As mentioned in the measurement section, for this study, staffing ratios will be

conceptualized as the ratio of the number of FTEEs to volume of stroke patients. The

numerator was provided from the VA's database, but the denominator had to be

calculated, due to a lack in the accuracy of the stroke volume. The accuracy issue can be

traced to the lack of compliance by the rehabilitation facilities in submitting FIM scores









to the FSOD for the early years of the study (Reker et al., 2004). To compensate for this

inaccuracy, a specificity algorithm will be used to identify the most true stroke patients

with a new stroke. This algorithm selected patients using the International Classification

of Disease, 9th Revision, Clinical Modification (ICD-9-CM, or ICD-9) codes (American

Medical Association, 1995). Patients are selected as followed:

* If admission or discharge primary diagnosis is 430.xx, 431 .xx, 432.xx, or 436.xx, or

* Admission or discharge primary diagnosis is V57.xx (Rehabilitation) and any

secondary diagnosis is 342.xx (Hemiparesis), 430.xx, 431 .xx, 433.xx, 434.xx, 435.,,,

436.xx, 437.xx, or 438.xx, or

* Admission or discharge primary diagnosis is 433.xx or 435.xx and any secondary

diagnosis code is 342.xx, 430.xx, 431 .xx, 432.xx, 434.xx, or 436.xx (Reker et al.,

2001).

Once the specificity algorithm is run and the "most true" strokes are determined, the

appropriate number of FTEEs will be divided by the stroke volume to calculate the

staffing ratio.

Step 2 Calculating the Appropriate Fit Variable

As mentioned in the measurement section, residual terms have to be created to

conceptualize fit variables for this study. Following the recommendations of Dewar and

Werbel (1984), Hage (1980), and Fry and Slocum (1979), the following analysis

approach is taken with respect to the construction of fit variables. First, the individual

structure variables are regressed onto the contextual variable. From these results, the

studentized residuals are generated by using the expression:









ri = ei/ (s(,fl-h,)


Studentized residuals are used instead of standardized residuals because they

recognize that the error associated with predicting values far from the mean is larger than

the error associated with predicting values closer to the mean (Chatterjee & Hadi, 1986).

The test can be interpreted as the t-statistic for testing the significance of a dummy

variable equal to 1 in an observation in question and 0 otherwise (Belsley, Kuh, &

Welsch, 1980).

Once the residuals are generated, the absolute value of the residuals will be taken

because there is no distinction between residuals that fall above or below the prediction

line. For this study, it is not important whether the residuals fall above or below the

prediction, or fit, line, but rather how far is the distance from fit line. Once the fit scores

are entered into the overall model equation, the standardized coefficients will be

generated in order to compare the effect sizes of the fit variables in comparison to the

other variables in the model.

Step 3 -Regression Technique for a Continuous Performance Measure

For this dissertation, there is one dependent variable (discharge motor FIM) that

reflects the outcome of patients who receive rehabilitation in the VA's healthcare system.

Discharge motor FIM is a continuous variable and therefore, the equations predicting this

dependent variable will use ordinary least squared (OLS) regression.

In a normal regression where the model is specified as


Yi = Xi, + E,









(x,,ti) are independently and identically distributed (i.i.d.) with variance o2. With these

assumptions, the regression produces consistent parameter estimates and standardized

errors that can be used for valid statistical inference about the coefficient (Kennedy,

1998). In this study, i.i.d. is not assumed. Therefore, corrections in the model have to be

made. To correct for the relaxed assumption of (Xi,E) being identically distributed, a

robust regression will be utilized. To address the relaxed assumption of independence,

the "cluster( )" option in STATA will be used (STATA, 2001). This study will cluster on

individual VA station. This option produces the appropriate standard errors even if the

observations are correlated. The "cluster ()" option also requires only the observations

be independent across the cluster.

Step 4 Determining Variation in Structural Characteristics

In an effort to detect differences in the characteristics of facilities that have

patients who have higher discharge FIM scores when compared to facilities with lower

scores, a series of regressions will be run. The first step will be to aggregate all the

patient-level data to the facility-level by year. In order to accomplish this, a number of

steps will be performed based on the type of variable. For the workload ratios and other

continuous variables (i.e., number of total beds and number of disciplines represented),

the average for each facility will be calculated. If the variable is dichotomous, a ratio was

created. This ratio represents the number of positive responses, i.e., those that have a 1

versus a 0, to the total number of patients in a particular facility that received care.

The dependent variable is then calculated at the facility level. Similar to the

continuous independent variables, the dependent variable is averaged for each facility for

each year. The next step will be to classify high versus low performing facilities. This is








done by dividing the facilities into quartiles according to the FIM scores, with higher

scores being grouped in the fourth quartile and lower scores grouped in the first quartile.

Next, four dichotomous variables will be created which indicate if a facility is listed in a

particular quartile category.

Finally, to test for differences among the facility characteristics, a series of

regressions will be run with each structural variable as the dependent variable and the

quartile variables as the independent variables, with lowest quartile serving as the

reference. If the structural variable is continuous, an OLS will be run according to Step 3

described above. If the dependent variable is a dichotomous variable, a logistic

regression (logit) will be run. In a logit, the dependent variable is binary in nature and

only has two possible outcomes. The model is represented by the expression:

logit (HI) = a + X = log (H / 1 H)

where H / 1 is the odds of success or failure and log (II / 1 I) is the logistic

transformation, or logit (Agresti & Finlay, 1997). One feature of the logit model is that

the stochastic ingredient is not represented by the error term. In this model, the stochastic

element is inherent in the modeling itself (Kennedy, 1998). Since the data is aggregated

at the facility level, only a robust regression will be run. The facility average FIM scores

for each year are viewed as individual data points.

Descriptive Statistics

Descriptive statistics and pairwise correlations are provided in Appendices F and

G, respectively.














CHAPTER 5
DESCRIPTIVE STATISTICS AND RESULTS

This chapter provides a descriptive profile of the facilities which provide formal

rehabilitation care to stroke patients within the VHA. The profile will examine facility

characteristics of all the facilities and the patients in these facilities. Descriptive statistics

for each variable are presented according to (1) the facility setting in which the RBU is

located, and (2) temporally. Following the descriptive profile, this chapter will present

the results of the analysis described in Chapter 4. Results from hypotheses test are the

presented for the effects of context, structure, and fit on FIM scores as well as structural

differences between high and low performing facilities.

Profile of Formal Rehabilitation Facilities

Of the more than 1100 facilities that are operated by the VA, there are 62

inpatient and 29 extended care facilities that either currently provide formal rehabilitation

in RBUs or have provided rehabilitation care in the past. This study contains a sample of

9231 patients who received formal rehabilitation care from fiscal years 1996-2002. A

majority of the veterans in this population received rehabilitation care in an inpatient

facility. Table 1 breaks down the patients by year and type of facility where

rehabilitation care was given.

In the following sections, the RBUs are examined according to facility type and

temporally. They are then compared to each other on organizational context and

structure.












































































0 c









Organizational Context

On average, patients who received rehabilitation care had an average FRG score

of 4.4. According to the FRG algorithm developed by Bates and Stineman (2000), this

FIM score indicates that the average patient in this study would be classified as having

had a moderate stroke. When the settings are compared, there is no difference in average

FIM. When looking at FRG classifications over time, average scores fell slightly over

the period of this study, going from a high of 4.6 in 1996 to a low of 4.3 in 2001, and

then increasing minimally to 4.4 in 2002.

Organizational Structure

Variable conceptualization issues

There was an issue of how to compensate for incomplete data due to a lack of

reporting for a number of disciplines. This section discusses the problem and how each

was resolved.

The first issue deals with the staffing ratio conceptualizations for physical and

occupational therapists (PT and OT, respectively). The data were sparsely reported for

these two disciplines individually, so it was decided to combine the two disciplines and

name the variable therapists. This was accomplished by summing the number of full-

time equivalent employees (FTEEs) for each facility and then dividing by the number of

strokes for the facility. After the new variable was constructed, there were still many

cases where FTEEs were not reported. To solve this problem, facility-level FTEE data

was used. The values for the PTs were the same as the unit-level data for the

rehabilitation unit. This was expected since the PTs are committed 100% to

rehabilitation. In addition, the reporting of FTEE data for OTs at the facility level was

more complete with fewer missing values when compared to the unit level. Therefore,









conceptualizing the therapists variable at the facility, accounts for some of the missing

OTs at the unit level.

The same issue arose for both the speech language pathologist and nurse

practitioners. Both variables were sparsely reported for the unit-level. At the facility

level, FTEEs were still sparsely reported for speech language pathologists and not

reported at all for nurse practitioners. To address this issue, a dichotomous variable was

created to account for the presence of either or both disciplines (referred to as SLPNP).

This variable was also conceptualized at the facility level.

The data for registered nurses (RNs) was the most incomplete of any of the

disciplines with over two-thirds of the FTEEs for the rehabilitation unit missing. It was

assumed that these were missing values because it is difficult to image such a lack of RN

availability for rehabilitation. Similar to the therapists' resolution, facility-level data was

used to calculate staffing ratios instead of the unit-level.

The final disciplines with missing data issues were licensed practical and

vocational nurses (LPNs and LVNs, respectively). The VA reports these two disciplines

as one (referred to as LPNs). However, there was no data available for this variable at

the facility level. Since LPNs may not play as prominent a role as RNs in rehabilitation,

it was decided that if no data was reported for LPNs, the variable was coded as a 0 to

indicate that no LPN was reporting to that facility.

Findings

When comparing setting differences for structural characteristics in terms of

staffing ratios, there is a consistent trend. The staffing ratios averages for all disciplines

are similar for patients who receive rehabilitation in the inpatient and BOTH settings

while the staffing averages for the extended care setting are higher. This relationship









would be expected since the denominator in the workload ratio, number of true stroke

cases, is smaller for the extended care facility when compared to the inpatient and BOTH

facilities. A decrease in the denominator indicates that the number of FTEEs is similar

for inpatient and extended care facilities (Table 2).

Table 2. Structural Characteristics by Setting Type

Extended
Main Care BOTH
Variable Mean SD Mean SD Mean SD
Discharge FIM 66.66 19.60 65.52 19.96 67.54 19.47
Functional Related Group (FRG) 4.41 2.46 4.47 2.48 4.35 2.42
Physician 0.07 0.12 0.11 0.20 0.06 0.09
Therapists 0.33 0.34 0.66 1.47 0.34 0.68
Presence of SLPNP (%) 27.00 0.44 0.13 0.34 0.16 0.37
Licensed Practical Nurse (LPN) 0.01 0.05 0.01 0.07 0.02 0.07
Registered Nurse (RN) 7.39 10.01 14.96 22.47 9.01 11.92
No. of Disciplines (Disciplines) 3.61 0.98 3.52 0.95 3.65 1.08
Av. Total Operating Beds (Size) 337.79 172.49 206.31 143.40 206.62 128.78
% of Acute Beds 89.00 0.31 0.24 0.43 0.28 0.45


In terms of the physical characteristics of the organization, the size of the BOTH

facilities are smaller in comparison to inpatient facilities. The smaller size of the

facilities would also explain why the presence of a SLPNP is smaller when compared to

the inpatient setting alone. In addition, smaller facilities tend to have a larger percentage

of subacute beds.

When looking at the temporal trends of the data, there are a few variables that

have been pretty steady while others have changed over time. Regarding the steady-state

variables, staffing ratios for physicians, therapists, and the number of disciplines

represented have not changed in the time frame of this study. The RN staffing ratio was

steady except for a peak of in 2001 (Table 3).









0 (0 D -T co 0) 1- ^-
0 CO -I 0 r 4
c) 0 N co c n M .o 0o 0 1 c 0

c (a co to 06 cO N oo 0 r- a)
d co 0 6 o o o o o
eo~ c C) (a N- N-


Nl iD oido d 6 r- o o d d d
o 00 C 6 0 6;


cO CO ( 0 o 0 C
c (N M C ) ( .- 0
S dd o d a c ^r d d d d
0 4 ( C) o 0

oc (





I ) o (D 0 (a C) (a -T
S dd o d N d d c
Ci O4 C ) C) N- () C) C) C)
o 0 CD o CD o


c N




c, idd d co N 0d d g
m C i CO' (D Ol T- 0 0 CM













Nlu Oj C4dd d d r d ,o d d d d
7- o



c" c o M M
Q 6idd d N- (a N d Cd)d d
0) (N
0 ))




Co C) M (t 0) N OD C-
(D (

S, S6c 0 0 Co

C 6 C ) N ( Cv ) C o c)
u dd d X N c c d T d d
(0 ma

C) (a) OD
>c) L o (aN- I) 0) ODC(a(a 0) (N ( (

oO ( L6 6) C) C) C)


(N



N~)(N
co LO ( C) C 0 d C






iC 6 C) xa C 5 C) C) C)0





,0a,,C .-ImI C Co ".a: aF
Sa = 16
IL (L f i (a U) (. Z w z 0z Q (a a, 1 CL a 0 (









Other structural variables fluctuated during the time frame of the study. The

staffing ratio for LPNs differed by a factor of 10 from 1996 to 1998 and then remained

steady. Regarding SLPNP variable, the presence of a speech language pathologists or

nurse practitioner was very sporadic in the first four years of the study, but between 1999

and 2000, there was a twelve percent increase to 31%. This trend has continued to climb

to a high of 35% in 2002. As indicated in Table 3, the RN staffing ratio decreased at the

same time the SLPNP variable increased.

Finally, the physical characteristics of the facilities have shown different trends.

The size of the facilities where the RBUs are located are decreasing. In 1996, a stroke

patient received rehabilitation care in a facility that averaged 530 beds. That number has

dropped every year to a low of approximately 236 beds in 2002. A similar decrease is

seen in the percentage of acute RBUs.

Patient Characteristics

The rehabilitation services that are provided by the VA are received by a diverse

set of veterans (Table 4). In terms of race, rehabilitation administered either in an

inpatient or BOTH setting is given to the same proportions of the population with whites

making up the largest percentage of the population (62%) and Blacks being the second

largest percentage (24%). The proportions are a little different for the extended care

setting. In this setting, whites are admitted in larger proportions (67%) when compared

to the inpatient or BOTH setting while the proportion of Blacks decreases by almost 5%

from 24% to 19%. In terms of marital status, the proportions are equal regardless of

setting with almost half of the patients being married and another quarter being divorced.









Table 4. Patient Demographic Information According to Facility Type

Extended
Race Inpatient Care BOTH
Hispanic White 668(8) 18(6) 46(7)
Hispanic Black 60 (.6 0(0) 0(0)
American Indian 36 (.4) 0(0) 1 (.1)
Black 2001(24) 62(19) 166(23.9)
Asian 43(1) 0 (0) 4(1)
White 5114 (62) 214(67) 432 (62)
Unknown 295 (4) 26 (8) 45 (6)
Total 8217 320 694

Extended
Marital Status Inpatient Care BOTH
Missing 51 (1) 7(2) 12(2)
Single 844 (10) 33(10) 85(12)
Married 4030(49) 151(47) 309(45)
Widowed 1004 (12) 39(12) 91 (13)
Separated 358 (4) 14 (4) 29 (4)
Divorced 1924 (23) 76 (24) 168 (24)
Total 8217 320 694

Extended
Age Inpatient Care BOTH
0-24 11 (.1) 0(0) 0(0)
25-34 30 (.4) 0(0) 0 (0)
35-44 190(2) 10(3) 15(2)
45-54 1157(13.6) 52(16) 81(12)
55-64 1992(24) 76(24) 144(21)
65-74 2717(33) 88(28) 208 (30)
75-84 1932 (24) 87 (27) 220 (32)
85+ 188 (1.9) 7(2) 26(4)
Total 8217 320 694

Note: Percentage of facility total in parentheses ()


Finally, in terms of the age of the stroke patients, over three-quarters of the

patients in this study fall into the three age categories which encompasses the ages of 55-

84. In addition, the BOTH rehabilitation setting had a noticeably larger percentage of

patients 75 and older when compare to the other settings (36% vs. 26% and 29%).









When looking at patient characteristics temporally, the percentage of the

population for Hispanic-Whites has increased while the percentage of Whites has

decreased, while all other racial and ethnic groups remained the same (Table 5). This

change in the population of the VA is a reflection of the increase in the Hispanic

population in the general population. A change has also taken place in the age of the

stroke patients in this study. Between 2001 and 2002, there is a 5% increase in the age

group 55-64, while at the same time there is a gradual decline in percentage of stroke

patients who are between 65-74 years old.

Results of Hypothesis Testing

The results are presented in a manner that follows the analysis strategy presented

in Chapter 4. As mentioned in the Analysis Strategy section for the appropriate

regression technique, in order to test the hypotheses for direct and contingent effects, an

OLS will be run with the structure, context, fit, and control variables representing the

independent variables and discharge motor FIM score representing the dependent

variable. Next, a series of regressions will be run to determine structural differences in

facilities with higher FIM scores when compared to facilities with lower scores.

Context-Performance Hypothesis

There is one hypothesis regarding the effects of context on performance. As

shown in Appendix H, there is support for Hypothesis 1, which predicts that an increase

in task uncertainty will be positively associated with discharge motor FIM scores

(coefficient=4.87, p=0.00).









Structure-Performance Hypotheses

There are six hypotheses regarding the effects of structure on a patient's

outcomes: three represents dimensions of specialization, one of standardization, and two

of complexity. Each is discussed below.

Specialization. There is partial support for Hypothesis 2 which posits that lower

staffing ratios will be positively associated with improved FIM scores. Of all the

disciplines represented, only lower physician and RN staffing ratios have the predicted

effect on stroke patient's FIM scores, yet they were non-significant.

There is no support for Hypothesis 3 which predicts that the number of different

disciplines represented in providing rehabilitation care will be positively associated with

FIM scores. There was also no support for relationship Hypothesis 4 which forecasts an

increase in the size of the facility will be positively associated with FIM scores.

Standardization. There is no support for Hypothesis 5 which hypothesizes that

when compared to rehabilitation care received in a subacute RBU, receiving care in an

acute RBU will be associated with lower FIM scores.

Complexity. Support is provided for Hypothesis 6 which states that when

compared to being admitted to an extended care facility, being admitted to an inpatient

facility is positively associated with FIM scores (coefficient=10.98, p=0.00). Also,

support is given to Hypothesis 7 which states that when compared to being admitted to an

extended care facility, being admitted to a BOTH facility is positively associated with

FIM scores (coefficient=l 1.69, p=0.00).

Fit Hypothesis

Partial support for Hypothesis 9 was found. This hypothesis asserts that

the fit between context and structure has a positive effect on FIM scores.



























Sr-
., e.7 N

(N,-.- L (aNC
0 r ^- l'- N O-
000 (D (0
(.- N rN








0 ( N a'
(N Ne












C,
N






o~aa,

oc 0
1m 0
0,ZdS1P 8 r5,K aa :





S.i- ? 0






g *-: a -

J. N
' g N






(DN (a N- (a 0'-
(N r-





77~o~ C

(i (0 (7 c 0 c -

S~~~ BE a^ ^*


N '- b
-, hO C M -C


(.7.1 (a ( 04 (
N ( ( (a (a'-




-N
a. -1 (N J ;1J"





0 C 0

(N (a (co
Cc m
'- (a '- (.

0 NN
S ~ c co5 ?
m


J '(, .7.







z 0S 2 cc-










c C,
0N (.
77 ..7 7 .

-.a'- N^


r^ in r~ zy (MN




c
(a


n) ci m) o 5 -' s
a? s 'a~
S51 .7
0 '-(aN (.7'-lil6I


(N.- (.7t( (a- ( ('e


(6 C.
tN T-c)f O ^ \






N -' .7 C u"- .- l
00







c i









a,








00 0 U)C0
(CO a


















04N
0) CO C c C ', O
iu m S1


















'^T fv o04 00C
7'' ~ 'i'iD H; -






(a (a (a(a( (a













<0 (N (7 o ( i( NI
N


(N
(a'
m




IT,
5.:7


-. 0,










cp
5










C,









ca
CT '"'

















N (.

N
(.7-




.7- I7








0
Sc,



t'


0
Sa
~ ^ cai
C


7)
S 1 )








79



The fit for FRG-MD, FRG-SLPNP, FRG-RN, FRG-size, and the FRG-BOTH setting


were all negative and significant which indicates a good fit. The FRG-therapists, FRG-


LPN, FRG-discipline, FRG-match, were all significant and positive, which indicates a


bad fit. According to the standardized beta coefficients in Table E, the fit scores for


FRG-MD and FRG-RN had to largest effect on FIM scores.


High/Low Performers Hypothesis


Finally, this study shows mixed support for Hypothesis 10 which posits that


facilities with stroke patients that have higher FIM scores are significantly different than


facilities with lower FIM scores (Table 6).


Table 6. Regressions for High and Low Performing Facilities


Disciplines Ratio


Number of clusters (station) = 65

Robust
Disciplines Coef. Std. Err. t
------------- - -
Percentile2 .3134796 .1454099 2.16
Percentile3 .2737844 .1174411 2.33
Percentile4 .3479624 .1996628 1.74
cons 3.272727 .1023647 31.97


SLPNP

Log likelihood


SLPNP P
-------------+----
Percentile2
Percentile3 .
Percentile4 I
cons -1

Inpatient Facility

Log likelihood =


Inpatient |
---- --------+____.
Percentile2 |
Percentile3 |
Percentile4
cons 2


-192.95099


Coef.

5958188
3742126
5389965
.504077



-111.027


Coef.

1.69538
0924571
5640326
.054124


Robust
Std. Err.

.3296515
.3841519
.4804042
.3294497






Robust
Std. Err.

.7605521
.5088266
.5511702
.4042707


F( 3, 64) = 2.59


Prob > F
R-squared
Root MSE


P>Itj [95% Conf.
.....................
0.035 .0229899
0.023 .0391688
0.086 -.0509099
0.000 3.06823

Wald chi2(3)
Prob > chi2
Pseudo R2
.....................

P> z| [95% Conf.

0.071 -.0502862
0.330 -.3787113
0.262 -.4025785
0.000 -2.149787

Wald chi2(3)
Prob > chi2 =
Pseudo R2


P>Izl

0.026
0.856
0.306
0.000


[95% Conf.

.2047257
-.9048246
-1.644306
1.261768


= 0.0607
= 0.0213
=.94061


Interval]

.6039693
.5083999
.7468346
3.477224

3.55
0.3149
0.0084


Interval]

1.241924
1.127136
1.480572
-.8583679

8.03
0.0455
0.0582


Interval]

3.186035
1.089739
.5162412
2.84648









When compared to low performing facilities, the other three performance groups

are all significantly different for the number of disciplines represented. The effects are

both positive and significant (coefficient=.313, p=.035; coefficient=.348, p=.086; and

coefficient=.274 and p=.023, respectively). There is also a positive and significant

difference between the lowest and second lowest performing facilities for the presence of

a SLPNP (coefficient=.596, p=.071). The same holds true for receiving care in an

inpatient rehabilitation setting (coefficient=1.70, p=.026).

Control Variables

As described previously, control variables were entered into the regression model

to control for factors other than those that have been tested in the above hypotheses.

Each control variable with a significant (p<.10) effect is described below.

According to Table E, all else equal, the more tenure the physical therapists are,

the lower the FIM scores (coefficient=-.107, p=.015). By contrast, older facilities have a

positive effect on FIM score (coefficient=.038, p=.000).

For the patient control variables, when compare to Black stroke patients,

Hispanic-Black (coefficient=3.98, p=.016) and Hispanic-White (coefficient=2.40,

p=.000) patients tended to have higher FIM scores. Regarding marital status, married

stroke patients (coefficient=-.904, p=.077) tended to have higher FIM scores that single

patients.

Finally, for age, when compared to the younger age group of stroke patients (0-

24), patient in all the older age groups had lower FIM scores. The FIM scores are lower

as the patient gets older. Table E lists the coefficients and p-values for each age category.






81


Summary of Analytic Results

The analyses performed in this investigation helps to shed light on the effects of

context and structure on performance. The results of the analyses suggest that context

and structure, individually and collectively, have an effect on the FIM scores of patients

who receive care in VA rehabilitation facilities. The results also prove that there is

minimal structural difference between facilities who have patients with high FIM scores

and those with low FIM scores. Finally, just as contingency theory suggests, the fit

between context and structure does play a role in a patient's discharge FIM score. This

relationship was suggested by the conceptual model in this study. The implications of

these findings will be discussed in Chapter 6.














CHAPTER 6
DISCUSSION

This chapter includes a discussion of the key findings from this dissertation, with

particular emphasis on their implications for stroke rehabilitation, management decision-

making within the VA, and organizational theory. First, study findings are interpreted

and discussed in terms of five key discussion points. Next, implications for each finding

are discussed. This is followed by suggestions for contingency theory. Finally,

limitations of the current research are presented and suggestions are offered for future

research.

Key Points for Discussion

The overall premise of this study is that the appropriate context and structure can

have a positive impact on stroke patients who receive rehabilitation care in the VA. I

suggest this influence is a function of the fit between the context and structure. Study

findings provide support for the proposed model.

The findings from this dissertation are summarized in this chapter in the following

five areas:

1. Heterogeneous patients receive rehabilitation care from the VA. VA stroke
patients exhibit trends that are consistent with the general population.

2. Within the VA, the facilities (i.e., inpatient, extended care, inpatient and
extended care (BOTH)) in which rehabilitation is provided are similar in
terms of the populations in which they serve.

3. The fit between context and structure is an important determinant ofFIM
scores.









4. Structural differentiation is observed only on one structural characteristic
when comparing low and high performing facilities in terms of FIM scores.

5. VA management plays a significant role in improving the quality of care that
is provided to the veterans who receive rehabilitation at the VA's facilities.

Each of the above points will be discussed in greater detail below.

Patient Population

On average, patients who were included in this study tended to have moderate

strokes more often than severe or mild strokes. A majority of the patients being classified

as having a moderate stroke indicates that the providers, who recommend the patient for

rehabilitation, and the subsequent submission of FIM scores to the FSOD, are following

VA guidelines. These guidelines state that in order to be admitted into the FSOD, the

patient has to have the potential to benefit from rehabilitation services.

In terms of race, the years that this study covers saw an increase in the population

of the Hispanic-Whites among VA stroke patients. This increase in the stroke population

is similar for the general veteran population (vaww.va.gov/vetdata/SurveyResults). The

same increase is also occurring nationally (www.census.gov). As the Hispanic

population continues to grow in the VA, the population of non-Hispanic Whites has

decreased over the same period of time.

Another category where change is occurring is in the age distribution of VA

stroke patients. From fiscal years 2001 to 2002, the percentage of the patients in the 55-

64 age category increased by 5%. The five percent increase represents the largest one

year increase in any of the age categories over the timeframe of this study. Once again,

this change in the age demographic follows a trend that is occurring in both the veteran

population (vaww.va.gov/vetdata/SurveyResults) and the general population

(www.census.gov). This category is starting to see an increase because of the "baby-









boomers." As the "boomers" start to age, they may require more medical services. In

addition, as the percentage of the population as a whole sees a decrease in private health

insurance, the "boomers" who are veterans, may be utilizing the VA more because of

affordability issues with private insurance.

Facility Setting Types

There are three types of facilities in which rehabilitation services are offered:

inpatient, extended care, and both inpatient and extended care (BOTH). For the inpatient

setting, there was a peak in inpatient admissions in 1999, but starting in 2000 and

continuing until 2002, there was a decline in the number of stroke patients who were

admitted to the inpatient setting. This was attributed to the fact that in 1996, Congress

enacted legislation that expanded the eligibility for the complete continuum of care. This

expansion included outpatient care and prescription drugs (vawwl.va.gov/cares).

Expansion of these programs allowed for more patients to be treated in an outpatient

setting. The legislation may also explain why over the same period of time inpatient

admissions were declining, patients in the extended care and BOTH settings increased,

with the BOTH setting having a larger increase (Table 1).

When looking at race, the distribution of stroke patients is similar for the inpatient

and BOTH settings. However, there is a difference in the racial composition for the

extended care setting when compared to the other two. With all other races equal for the

inpatient and BOTH settings, the percentage of Blacks is 5% lower than the other settings

while the percentage of Whites is 5% higher. An explanation for the decrease in Black

veterans could be a result of Black families choosing not to admit their loved ones into an

extended care facility but instead assuming the role of caregiver. One study showed that









Black caregivers, when compared to White caregivers, provide a higher intensity of care

while less likely to report difficulty with providing care (Navaie-Waliser, 2001).

In addition, over three-quarters of the patients admitted to each of the setting

types are between the ages of 55-84. One reason for such a small percentage of those

younger than 55 is that these patients probably have private health insurance. Along the

same line, the category which includes the 55 year old patients is the category that has the

first group of retirement eligible veterans included in it.

When looking at the various ages categories in the different setting, patients in the

75-84 category were in the BOTH setting at a higher rate than any other setting. This

could be attributed to the fact that older patients take longer to recover from stroke than

younger patients (Paolucci et al., 2003). The patient may also need more rehabilitation to

regain their functional status and are transferred to an extended care facility to get this

extra care.

Fit

The central thesis of contingency theory is that the fit between context and

structure will have a positive effect on performance. If there is a lack of fit, or misfit,

between context and structure, the effect on performance will be negative. Study

findings indicate that there is indeed a fit between context and structure and the fit effects

performance in a positive manner.

Explanation for fit. In terms of the three constructs of structure and their fit with

context, this study provides interesting results. Of the seven variables in the construct

specialization, four were would be considered good fits. One interesting point is that for

the staffing ratios, those that were non-significant when their effect on FIM scores were

tested directly, were significant when combined with the FRG variable to calculate the fit




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EFMGCHF7F_9OO0IY INGEST_TIME 2012-08-30T13:48:13Z PACKAGE AA00011795_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES