DIAGNOSIS & DISRUPTION:
UNDERSTANDING GUIDELINE-CONSISTENT DEMENTIA CARE AND
PATTERNS OF TRANSITIONS EXPERIENCED BY INDIVIDUALS WITH
DEMENTIA
by
Saskia Nikali Sivananthan
BSc, The University of British Columbia, Canada, 2004
MSc, McGill University, Canada, 2008
A THESIS SUBMITTED IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
in
The Faculty of Graduate & Postdoctoral Studies
(Population & Public Health)
THE UNIVERSITY OF BRITISH COLUMBIA
(Vancouver)
February 2015
©Saskia Nikali Sivananthan, 2015
ii
ABSTRACT
Introduction
Early detection and management, as outlined in dementia guidelines, can improve
outcomes and potentially reduce the number of transitions experienced between care-
facilities. The objectives of this thesis were to: conduct a critical assessment of the
literature on actual versus guideline-consistent dementia care; use population-level
data to examine variations in receipt of guideline-consistent dementia care; understand
longitudinal patterns of transitions and the factors that contribute to transitions; and
assess if receipt of guideline-consistent dementia care and/or high quality primary care
are associated with the number of transitions experienced.
Method
Population-level administrative data in British Columbia were used to identify cohorts of
individuals newly diagnosed with dementia in 2001/02 and 2009/10. Dementia
guidelines were used to characterize clinical care individuals could receive. The
proportion of patients diagnosed with dementia in 2009/10 who received guideline-
consistent dementia care and the probability that patient characteristics affected the
likelihood of receiving each process was assessed. Those diagnosed with dementia in
2001/02 were followed forward 10-years after the first point of diagnosis and points of
care when transitions were highest as well as factors that contribute to those transitions
were assessed. The association of receipt of guideline-consistent dementia care and/or
high quality primary care with the number of transitions experienced in the year of
diagnosis was assessed using negative binomial regression.
iii
Results
Older patients were less likely to receive guideline-consistent dementia care. A quarter
of the cohort received an antipsychotic or non-recommended benzodiazepine.
Individuals living within rural health-authorities or of low income were more likely to
receive antipsychotic treatment. Transitions were highest in the year of diagnosis and at
end-of-life, driven by acute hospitalizations. Higher morbidity, living within rural
health-authorities, and having behavioural symptoms associated with dementia were
strongly associated with the number of transitions experienced. Lastly, receipt of
guideline-consistent dementia care and high quality primary care, independently, were
associated with fewer transitions across care settings.
Conclusion
Patterns of inequality by age and income may signal barriers to guideline-consistent
dementia care. The spike in transitions in the year following diagnosis highlights a
distressing period for individuals with dementia, but suggests a useful target for
interventions.
iv
PREFACE
This statement is to certify that the work in this thesis was conceived, designed,
analyzed and written by Saskia Sivananthan. All research in this dissertation was
approved by the University of British Columbia’s Behavioural Research Ethics Board:
UBC BREB Number H11-02814.
Saskia Sivananthan was entirely responsible for the work in Chapters 1, 2, 4, 6, 7 and 8.
Chapter 3 is based on work conducted by Saskia Sivananthan, Joseph Puyat and
Dr. Kimberlyn McGrail. Saskia Sivananthan was responsible for the conception, design,
development and management of the study, conducting the analysis and interpretation
of the data, writing and revising the manuscript. Joseph Puyat contributed to the
collection and analysis, Dr. Kimberlyn McGrail contributed to the analysis, interpretation
and edited the manuscript.
Chapter 5 is based on work conducted by Saskia Sivananthan, Ruth Lavergne and
Dr. Kimberlyn McGrail. Saskia Sivananthan was responsible for the conception, design,
development and management of the study, conducting the analysis and interpretation
of the data, writing and revising the manuscript. Ruth Lavergne and Dr. Kimberlyn
McGrail contributed to the analysis, interpretation and edited the manuscript.
A version of Chapter 3 has been published:
Sivananthan SN, Puyat JH, McGrail KM. Variations in self-reported practice among
physicians providing clinical care to individuals with dementia: a systematic review.
Journal of American Geriatric Society 2013; 61(8):1277-85.
v
A version of Chapter 5 has been accepted:
Sivananthan SN, Lavergne MR, McGrail KM. Caring for dementia: A population-based
study examining variations in guideline-consistent medical care. Alzheimer’s &
Dementia
A version of Chapter 6 is under review:
Sivananthan SN, McGrail KM. Diagnosis & Disruption: Identifying points of care when
transitions are highest for individuals with dementia
A version of Chapter 8 has been accepted:
Sivananthan SN. Short End of the Stick? Health Care Inequity in a Vulnerable
Population of Individuals with Dementia. Canadian Geriatrics Society Journal of CME
2014.
vi
TABLE OF CONTENTS
ABSTRACT ........................................................................................................................................................ ii
PREFACE .......................................................................................................................................................... iv
TABLE OF CONTENTS ................................................................................................................................. vi
LIST OF TABLES ............................................................................................................................................... x
LIST OF FIGURES .......................................................................................................................................... xii
LIST OF ABBREVIATIONS ......................................................................................................................... xiv
ACKNOWLEDGEMENTS ............................................................................................................................ xv
DEDICATION ............................................................................................................................................... xvii
CHAPTER 1- INTRODUCTION ................................................................................................................... 1
1.1 Introduction to the Area of Inquiry ............................................................................................. 1
1.2 Research Objectives and Hypotheses ........................................................................................ 3
1.3 Thesis Roadmap ............................................................................................................................... 10
CHAPTER 2- BACKGROUND & LITERATURE REVIEW .................................................................... 13
2.1 What is Dementia? .......................................................................................................................... 13
2.2 The Social and Financial Impact of Dementia ....................................................................... 17
2.3 Care for Dementia in the Primary Care Setting .................................................................... 19
2.4 Overview of Services Outside Primary Care ........................................................................... 22
2.4.1 Home and Community Care in British Columbia ......................................................... 24
2.4.2 Long-Term Care in British Columbia ................................................................................. 26
2.5 American and Canadian Guidelines for Dementia Care .................................................... 28
2.5.1 Dementia Care Guidelines in British Columbia ............................................................. 33
2.6 Variations in Dementia Care ........................................................................................................ 36
2.7 Conceptual Framework .................................................................................................................. 38
2.8 Primary Outcome Measure: Transitions and Consequences ............................................ 43
2.8.1 Transitions for Seniors ............................................................................................................ 45
2.8.2 Transitions for Seniors with Dementia ............................................................................. 45
CHAPTER 3 – A SYSTEMATIC REVIEW OF VARIATIONS IN DEMENTIA CARE PHYSICIAN
PRACTICE PATTERNS ................................................................................................................................. 48
3.1 Introduction ....................................................................................................................................... 48
3.2 Systematic Review Methods ........................................................................................................ 49
3.2.1 Search Strategy ......................................................................................................................... 49
3.2.2 Data Collection and Data Extraction ................................................................................. 51
3.2.3 Data Analysis ............................................................................................................................. 54
vii
3.3 Results .................................................................................................................................................. 54
3.3.1 Variations in Dementia Care Processes ............................................................................ 60
3.4 Discussion ........................................................................................................................................... 63
3.4.1 Limitations .................................................................................................................................. 66
3.5 Conclusion .......................................................................................................................................... 67
CHAPTER 4 – METHODOLOGY ............................................................................................................... 69
4.1 The Health Services Research Lens ........................................................................................... 69
4.1.1 Use of Administrative Databases ....................................................................................... 70
4.2 Administrative Data Sources ....................................................................................................... 71
4.2.1 The Consolidation File ............................................................................................................ 73
4.2.2 The Medical Services Plan File ............................................................................................. 74
4.2.3 The Home and Community Care File ................................................................................ 76
4.2.4 The Hospital Separations File .............................................................................................. 77
4.2.5 The Vital Statistics Deaths File ............................................................................................. 77
4.2.6 PharmaNet .................................................................................................................................. 78
4.2.7 The College of Physicians & Surgeons of BC File ........................................................ 79
4.3 Study Design ..................................................................................................................................... 79
4.3.1 Identification of Study Cohort ............................................................................................. 80
4.4 Construction of Longitudinal Files ............................................................................................ 83
4.4.1 Analytic and Service Variables............................................................................................. 84
4.4.1.1 Health Status ...................................................................................................................... 88
4.4.1.2 Physician Attribution Process ....................................................................................... 89
4.4.2 Outcome Variables .................................................................................................................. 96
4.4.2.1 Long-Term Care Facility Transition ............................................................................ 97
4.4.3 Analytic Models......................................................................................................................... 99
CHAPTER 5 – CARING FOR DEMENTIA: A POPULATION-BASED STUDY EXAMINING
VARIATIONS IN GUIDELINE-CONSISTENT CARE .......................................................................... 103
5.1 Introduction .................................................................................................................................... 103
5.2 Methods ........................................................................................................................................... 104
5.2.1 Data Sources ........................................................................................................................... 104
5.2.2 Study Population ................................................................................................................... 105
5.2.3 Recommended Dementia Care Guidelines .................................................................. 105
5.2.4 Explanatory Variables .......................................................................................................... 108
5.2.5 Statistical Analysis ................................................................................................................. 108
5.3 Results ............................................................................................................................................... 109
viii
5.3.1 Final Study Participants....................................................................................................... 111
5.3.2 Variations in Dementia Care Processes ......................................................................... 114
5.4 Discussion ........................................................................................................................................ 120
5.4.1 Limitations ............................................................................................................................... 124
5.5 Conclusion ....................................................................................................................................... 125
CHAPTER 6 – IDENTIFYING POINTS OF CARE WHEN TRANSITIONS ARE HIGHEST AND
THE FACTORS THAT CONTRIBUTE TO THEM ................................................................................ 127
6.1 Introduction .................................................................................................................................... 127
6.2 Methods ........................................................................................................................................... 129
6.2.1 Data Sources ........................................................................................................................... 129
6.2.2 Study Population ................................................................................................................... 130
6.2.3 Primary Outcome – Transitions ........................................................................................ 130
6.2.4 Explanatory Variables .......................................................................................................... 131
6.2.5 Statistical Analysis ................................................................................................................. 132
6.3 Results ............................................................................................................................................... 134
6.3.1 Baseline Characteristics ....................................................................................................... 134
6.3.2 Aggregate Healthcare Use and Transitions ................................................................. 137
6.3.3 Transitions during Year of Diagnosis ............................................................................. 140
6.3.4 Transitions at End-Of-Life .................................................................................................. 144
6.3.5 Factors Associated with Transitions ............................................................................... 146
6.4 Discussion ........................................................................................................................................ 147
6.4.1 Limitations ............................................................................................................................... 154
6.5 Conclusion ....................................................................................................................................... 155
CHAPTER 7 – GUIDELINE-CONSISTENT DEMENTIA CARE AND HIGH QUALITY PRIMARY
CARE – IS THERE AN ASSOCIATION WITH TRANSITIONS? ...................................................... 156
7.1 Introduction .................................................................................................................................... 156
7.2 Methods ........................................................................................................................................... 157
7.2.1 Data Sources and Study Population............................................................................... 157
7.2.2 Primary Outcome - Transitions ........................................................................................ 158
7.2.3 Measures of Guideline-Recommended Dementia Care & High Quality Primary
Care ....................................................................................................................................................... 158
7.2.4 Controlled Explanatory Variables .................................................................................... 161
7.2.5 Statistical Analysis ................................................................................................................. 162
7.3 Results ............................................................................................................................................... 163
7.3.1 Descriptive Analyses ............................................................................................................ 163
ix
7.3.2 Receipt of Guideline-Consistent Dementia Care and High Quality Primary Care
Process ................................................................................................................................................. 164
7.3.3 Regression Analysis .............................................................................................................. 167
7.4 Discussion ........................................................................................................................................ 171
7.4.1 Limitations ............................................................................................................................... 175
7.5 Conclusion ....................................................................................................................................... 177
CHAPTER 8- CONCLUSION .................................................................................................................. 178
8.1 Summary and Key Contributions ............................................................................................ 178
8.2 Research Findings ......................................................................................................................... 181
8.2.1 Variations in Guideline-Consistent Dementia Care – Lesson Learned from a
Systematic Review ........................................................................................................................... 181
8.2.2 Persistent Variation – Getting Past Self-Report using Administrative Data .... 183
8.2.3 Transitions – The Year of Diagnosis and Disruption ................................................ 184
8.2.4 The Effectiveness of Guideline-Consistent Dementia Care ................................... 186
8.3 Strengths and Limitations.......................................................................................................... 188
8.4 Future Directions .......................................................................................................................... 190
8.5 Areas for Further Inquiry ............................................................................................................ 194
8.6 Conclusion ....................................................................................................................................... 196
BIBLIOGRAPHY .......................................................................................................................................... 198
APPENDIX A, CHAPTER 3 ...................................................................................................................... 218
A.1 Objective of Study Protocol ..................................................................................................... 218
A.2 Method ............................................................................................................................................. 218
A.2.1 Study Design .......................................................................................................................... 218
A.2.2 Population/Comparators ................................................................................................... 218
A.2.3 Interventions........................................................................................................................... 218
A.2.4 Outcomes ................................................................................................................................ 219
A.2.5 Inclusion and Exclusion Criteria....................................................................................... 219
A.3 Review Process .............................................................................................................................. 219
A.3.1 Search Method ....................................................................................................................... 219
A.3.2 Data Collection and Extraction Procedure .................................................................. 220
A.3.3 Critical Appraisal ................................................................................................................... 220
APPENDIX B, CHAPTER 4 ....................................................................................................................... 221
APPENDIX C, CHAPTER 5 ...................................................................................................................... 222
APPENDIX D, CHAPTER 6 ...................................................................................................................... 229
x
LIST OF TABLES
Table 1.1 – Overview of thesis objectives and corresponding analytical chapters ............... 3
Table 2.1 - International classification of disease codes related to dementia, version 9 ......
..................................................................................................................................... .........................................17
Table 2.2 - International classification of disease codes related to dementia, version 10
........................................................................................................................................................................... 17
Table 3.1 - Characteristics of studies evaluating physician practice patterns in dementia
care ................................................................................................................................................................... 57
Table 3.2 - Variations in physicians practice patterns measured by proportions for each
dementia care process .............................................................................................................................. 62
Table 4.1 - Key variables and data sources for analysis ................................................................ 72
Table 4.2 – Detail of variables measuring characteristics in conceptual framework .......... 84
Table 5.1 – Proportion of seniors 69+ years of age residing in a LTC facility receiving
guideline-consistent care ...................................................................................................................... 110
Table 5.2 – Baseline characteristics of community-dwelling individuals newly diagnosed
with dementia in 2009/10 ..................................................................................................................... 112
Table 6.1 – Baseline characteristics of individuals newly diagnosed with dementia in
2001/02 ........................................................................................................................................................ 135
Table 6.2 – Transitions and healthcare use during the entire study period ....................... 138
Table 6.3 – Cohort characteristics stratified by rate of transitions ........................................ 139
Table 6.4 – Multivariate regression analysis examining patient and practice factors
associated with transitions experienced .......................................................................................... 147
Table 7.1 – Description of primary care dimensions measured .............................................. 161
Table 7.2 – Summary of sub-analyses and sub-cohorts ............................................................ 163
Table 7.3 –Dementia care and primary care use in year of diagnosis .................................. 166
Table 7.4 – Odds of receiving guideline-consistent care based on receiving a diagnosis
in the community (relative to receiving an initial diagnosis in hospital) ............................ 167
Table 7.5 – Multivariate regression analysis examining association of dementia care and
primary care with transitions experienced in the year of diagnosis ..................................... 170
Table B.1 - Cross-walk mapping a sample of drug identification numbers to
corresponding anatomical therapeutic chemical groups for medications of interest ... 221
xi
Table C.1 – Baseline characteristics of individuals newly diagnosed with dementia in
2009/10 who died during study period ........................................................................................... 222
Table C.2A – Percentage of individuals who died during study period who received
laboratory testing and imaging processes outlined in guidelines stratified by patient
characteristics ............................................................................................................................................ 224
Table C.2B – Percentage of individuals who died during study period who received
prescriptions and dementia care management processes outlined in guidelines stratified
by patient characteristics ...................................................................................................................... 225
Table C.3A - Proportional odds ratio for factors associated with receiving guideline-
consistent care from multivariate modelling ................................................................................. 226
Table C.3B – Proportional odds ratio for factors associated with receiving guideline-
consistent prescriptions and dementia care management processes, from multivariate
modelling .................................................................................................................................................... 227
Table D.1 - Longitudinal health care use ........................................................................................ 229
Table D.2 – Principal diagnoses for patients newly diagnosed with dementia in hospital
only ................................................................................................................................................................ 230
xii
LIST OF FIGURES
Figure 2.1 – Assessment and publicly-funded service options in BC ...................................... 23
Figure 2.2 – Conceptual framework guiding the thesis ................................................................ 40
Figure 3.1 – Flowchart describing the approach used to identify all eligible studies ....... 56
Figure 3.2 – Geographical locations of eligible studies from the USA .................................... 59
Figure 4.1 – Historical cohort study design ....................................................................................... 81
Figure 4.2 – Example of transitions that can be experienced over study period ................ 97
Figure 4.3 – Pathways to publicly funded home and community care support .................. 98
Figure 5.1 – Flow of study participants ............................................................................................ 110
Figure 5.2 – Location of diagnosis of dementia (Cohort 2) ...................................................... 113
Figure 5.3A – Percentage of seniors receiving guideline-consistent lab testing and
imaging ........................................................................................................................................................ 115
Figure 5.3B – Percentage of seniors receiving guideline-consistent prescription and
dementia care management processes ........................................................................................... 116
Figure 5.4A – Adjusted odds ratio of receiving calcium, B12, laboratory tests and
computed tomography scans.............................................................................................................. 117
Figure 5.4B – Adjusted odds ratio of receiving acetyl-cholinesterase inhibitors,
antipsychotics and non-recommended benzodiazepines ........................................................ 118
Figure 5.4C – Adjusted odds ratio of receiving a physical exam, counselling or specialist
referral .......................................................................................................................................................... 118
Figure 6.1 – Location of diagnosis of dementia (Cohort 1) ...................................................... 136
Figure 6.2 – Survival curve of cohort ................................................................................................ 137
Figure 6.3 Transitions rate in each year for those with <1.0, 1.0-1.9, 2.0-3.9 and 4.0+
average transitions per year for 2001/02-2003/04, by year of death (upto 2003/04) .... 139
Figure 6.4A – Transition rates over time for those diagnosed in 2001/02, by year of
death ............................................................................................................................................................. 140
Figure 6.4B – Hospitalization rates over time for those diagnosed in 2001/02, by year of
death ............................................................................................................................................................. 141
Figure 6.4C – Other transitions over time for those diagnosed in 2001/02, by year of
death ............................................................................................................................................................. 141
xiii
Figure 6.5 – Number and proportion of those eligible moving from community to LTC,
by years since diagnosis ........................................................................................................................ 143
Figure 6.6 – Frequency of different transition types experienced by newly diagnosed
dementia patients .................................................................................................................................... 144
Figure 6.7A – Hospitalization rate over time for those diagnosed in 2001/02, by LTC
status and year of death ........................................................................................................................ 145
Figure 6.7B – Hospitalization rate over time for those diagnosed in 2001/02, by LTC
status and number of major ADGs .................................................................................................... 145
xiv
LIST OF ABBREVIATIONS
AChI Acetyl-cholinesterase inhibitor
ACG Adjusted Clinical Groups
ADG Aggregated Diagnostic Groups
AD Alzheimer’s Disease
ADLs Activities of Daily Living
ATC Anatomical Therapeutic Chemical
BC British Columbia
CT Computed Tomography
DAD Discharge Abstract Database
DINs Drug Identification Numbers
HA Health Authority
HCC Home and Community Care
HSR Health Service Research
ICD International Classification of Disease &
Related Health Problems
LTC Long-term care
MRI Magnetic Resonance Imaging
MSP Medical Services Plan
NINCDS-
ADRDA
The National Institute on Aging and the
Alzheimer’s Association Workgroup
PCPs Primary care physicians
RUBs Resource Utilization Bands
SES Socioeconomic status
TSH Thyroid Stimulating Hormone
xv
ACKNOWLEDGEMENTS
I owe a big debt of gratitude first to my supervisor and mentor, Dr. Kimberlyn
McGrail. Kim supported, mentored and nurtured me through my uphill battle to induct
myself into a new field, make connections across the country and develop a meaningful
thesis that I am proud of. This body of work would not be what it is without her sage
advice, enthusiasm for results big and small and willingness to be available whenever I
needed her. Thank you Kim for bringing me into the CHSPR family and for being such a
wonderful supervisor and friend, what more could a graduate student possible ask for?
A big thank you to my secondary supervisor, Dr. Morris Barer, for such thoughtful
and meticulous feedback on all of my work. Given everything on your plate, I
appreciate the time invested in reviewing my work with a fine toothcomb and for asking
the difficult questions. Your running commentary made revisions both valuable and
entertaining. As well, I want to thank my committee members, Dr. Margaret McGregor
and Dr. Jason Sutherland for their encouragement and for sharing their insights so
generously.
To Ruth Lavergne, a special thank you not only for your contributions to this
thesis, but also for the pep talks, skype calls, drinks, comfort food and countless other
generous ways in which you made this PhD such a great experience. It won’t be
forgotten. To all my colleagues at CHSPR, particularly Dawn Mooney (I owe you a drink!
Your ability to dawnify never ceases to amaze me), Sandra Peterson (for your ninja SAS
skills and patient responses as I taught myself to code), Rachael McKendry and Lindsay
Hedden, your support helped me build this thesis, thank you.
xvi
I want to acknowledge my two mentors, Dr. Lynn Beattie and Dr. Larry Chambers,
both of whom gave of their time freely, listening to my thesis as it grew and morphed,
and always providing sounding boards I could trust. I appreciate your advice and
friendship. I also want to thank Responsive Care Management, Bill Dillane and Chris
Dalgish who made it possible for me to ground the framework of my thesis by truly
experiencing life as a senior at a long-term care facility. A special thank you to Karie
Warner (Mill Creek Care Centre), Cathy Fiore (O’Neill Centre), the warm staff and
residents at both Mill Creek Care Centre and O’Neill Centre who answered all my
questions, treated me as a long-term care resident no-questions-asked while making
my stay as comfortable and informative as they could.
Lastly, thank you to mi familia. Amma, Appachchi and Angelo, you have always
supported me, my decisions and harebrained adventures. Thank you for being my
backbone, strongest champions and defenders. You made the sacrifices to give me the
opportunities to be who I am today. I can reach higher in the assurance that you will
always be there for me. To Travis, my very own personal cheerleader, thank you for
believing in me, nudging me forward when I was hesitant and for never ever allowing
me to compromise my dreams. Your unshakable faith in me and dedicated partnership
is the foundation of my work.
xvii
DEDICATION
This thesis is dedicated to my two favourite people who taught me to value
independence and joie de vivre at any stage of life: my grandmother Sybil Maud Mary
Dehideniya and Vincent.
Grandma, you are an inspiration, I am so proud to have a role model like you.
Vince, I still miss you. I know you’re enjoying a beer and a good laugh somewhere.
1
CHAPTER 1- INTRODUCTION
1.1 Introduction to the Area of Inquiry
Dementia is one of the most significant causes of disability among seniors,
affecting 35.6 million people worldwide in 2010 and projected to double every 20
years
1
. In Canada, an estimated 747,000 individuals were living with dementia in 2011
(14.9% of Canadian 65 years or older)
2
. In addition to memory loss, as the disease
progresses, impairment manifests itself in other symptoms that include language
disorders, difficulty with physically executing learned movement, inability to recognize
objects, confusion and behavioral disturbances
3
. The degenerative nature of the
disease means that as it progresses, individuals require increased care, social support
and assistance with activities of daily living (ADLs)
4,5
. It is a chronic disease with often
devastating and long-term financial and social implications for the patient, caregivers
and families.
The key contact in establishing a diagnosis, managing the dementia and
potentially connecting patients with community resources is the primary care physician.
However, many primary care physicians have limited training in geriatric care generally
and dementia care specifically. Best practice dementia care guidelines can therefore
play an essential role in helping primary care physicians confirm a diagnosis of
dementia and then make decisions around managing their patients’ care, particularly
during the early stages after diagnosis. Guidelines on dementia care have been
available, well established and regularly updated for almost three decades
6,7
. Despite
evidence that adherence to guidelines on dementia care can improve quality and
outcomes for patients, reports of poor detection and inadequate management persist
8,9
.
2
Due to the chronic nature of the disease, physicians need to be able to determine
a diagnosis, and develop a management plan as early as possible. Equally important is
the development of a longer-term care and life management plan for a condition that
will worsen in severity, has no known effective treatment, and significantly affects
quality of life and outcomes for patients and those around them. Within that longer
term plan, there will be an important series of care-change points, or transitions.
A transition is defined as a situation in which “an individual physically moves
from one place to another and stays there for at least one night”
10
. The healthcare
system is fractured, often forcing patients to transition between different facilities to
receive care when experiencing highly stressful events. A high number of transitions
represents a serious challenge to dementia patients and their caregivers because of the
importance of stability and familiarity to those suffering from dementia
11,12
. A greater
number of transitions between care settings also poses a serious challenge to the
continuity of care and the safety of the patient and is associated with medication errors,
preventable hospital readmissions and increased mortality risk
13–17
. Some transitions
are inevitable, but failure to develop a condition-sensitive care plan can lead to more
transitions than are necessary, with effects on the health status and quality of life of the
affected patients.
The chronic nature of dementia requires care that provides accessibility, care
coordination and continuity in order to meet the ongoing biopsychosocial needs of
people with a primary diagnosis of dementia
18,19
. These are all attributes of high
performing primary care. High quality primary care has been shown to be associated
with fewer of the safety and quality deficiencies identified during transitions
20–22
.
3
However few studies have assessed the experience of primary care in elderly
populations and particularly not in vulnerable elderly populations such as those with
dementia
23
. As well, primary care physicians are in the best position to provide
guideline-consistent dementia which has been shown to be associated with better
outcomes for patients
24–26
, though no studies have assessed the association between
receipt of guideline-consistent dementia care and transitions.
1.2 Research Objectives and Hypotheses
This thesis is designed to add needed research evidence to the area of primary
care for people with dementia. The specific objectives and hypotheses of my thesis and
the corresponding analytic chapters where they are found are outlined in Table 1.1 and
discussed in more detail below.
Table 1.1 – Overview of thesis objectives and corresponding analytical chapters
Objective Chapter
Conduct a critical assessment examining the existing research
literature on actual physician practice patterns associated with the
care of people with dementia and to what extent those practice
patterns are consistent with published guidelines
3
Examine population-based variations in receipt of guideline-
consistent dementia care and patient factors that are associated
with that care
5
Examine transitions experienced longitudinally to establish
patterns that can help identify points of care when transitions are
highest and the factors that contribute to those transitions
6
Assess the association between receipt of guideline-consistent
dementia care and/or high quality primary care and the number of
transitions experienced
7
4
Research Objective 1 – A Critical Assessment of Available Research on the extent
to which Dementia Care Practice Patterns are Consistent with Guidelines
Primary care physicians are gatekeepers who are often the first point of contact for
people with dementia and are therefore in the unique position to make the diagnosis of
dementia and to set in motion decisions about clinical care
7,8,27
, but limited research is
available on physician practice patterns in dementia care. Available studies indicate
wide variations in methodologies and responses regarding dementia care provided by
physicians
28,29
. Best practice guidelines for dementia care have been established in the
medical community for several decades. The use of best practice guidelines has been
shown to reduce variation and maintain, or even improve quality of care
30
. Therefore,
my first objective is to determine the current state of knowledge on physician practice
patterns with regard to dementia care using best practice guidelines and to determine
to what extent actual practice, as reported in the literature, is consistent with guideline
recommended care. I assess seven dementia care processes recommended by
guidelines: formal memory testing; imaging; laboratory-testing; interventions;
counselling; community service; and specialist referrals. Results are presented in
Chapter 2.
Research Question 1: Based on the current literature, to what extent is actual practice
consistent with guideline-recommended care?
Hypothesis:
i) There is wide variation in physician practice patterns with regard to provision of
dementia care, particularly discretionary dementia care processes.
5
Research Objective 2 – Population-based Variation in Receipt of Guideline-
Consistent Dementia Care for Individuals with Dementia
The systematic review undertaken to meet research objective 1 showed wide
variation in the provision of guideline-recommended care by physicians. Despite this,
the available studies had significant limitations in study design, sampling and sample
size, limiting interpretation of the extent of variation and its effect on outcomes of care
for those with dementia
31
. A recent publication from the Canadian Consensus
Conference on the Diagnosis and Treatment of Dementia noted “perhaps more
important than any of these knowledge gaps are the large gaps between what we know
about diagnosis and what is practiced in many family physicians’ offices”
32
. Using
administrative data to mitigate many of the limitations identified in past research, I
examine the receipt of clinical services by seniors newly diagnosed with dementia in
British Columbia (BC), and compare patterns of care to those recommended by
dementia care guidelines. By conducting this analysis at a patient level, I am able to
examine potentially modifiable patient characteristics associated with any variation and
differentiate warranted and unwarranted variation. I assess six dementia care processes
measurable in administrative data: imaging; laboratory-testing; physical examination;
pharmaceutical interventions; counselling; and specialist referrals, all within the first year
of diagnosis. The specific research questions are as follows:
Research Question 2a: What proportion of individuals newly diagnosed with dementia
in BC receives each of the guideline-consistent dementia care processes?
6
Research Question 2b: Is there variation in receipt of guideline-consistent care, and if
so, what are the patient characteristics most strongly associated with this variation?
Hypothesis:
i) A high proportion of individuals will receive guideline-consistent laboratory and
imaging processes of dementia care as these are more regularly administrated, while a
lower proportion of individuals will receive all other, more discretionary processes of
care.
ii) There will be warranted variation in receipt of guideline-consistent care that can be
explained by geographic location and health status.
iii) There will be unwarranted variation by other patient characteristics, indicating
potential inequity and lack of quality of care.
Research Objective 3 – Patterns of Longitudinal Transitions Experienced by
Individuals with Dementia
The prevalence of moves between care settings (home, hospital, LTC-facilities), or
transitions, is high in the elderly
33,34
. A high number of transitions between care
settings poses a serious challenge to continuity of care and the safety of the patient as
evidenced by medication errors, quality deficiencies, preventable hospital readmissions
and increased risk of mortality
13–17
.
For extremely vulnerable individuals with dementia, safety challenges are
compounded due to their disorientation
12,35
. Guidelines for good dementia care stress
7
the need for continuity and familiar environments which cannot be consistent with a
high number of transitions
10,11,26,36
.
A few studies have begun to examine transitions, describing the dynamic movement
of seniors with dementia
12,37
. However, these studies are often over short periods of
time with small cohorts and none examine the factors that contribute to these
transitions. The objective of this study was to examine transitions experienced
longitudinally to establish patterns that can be used to identify points of care when
transitions are highest and the factors that contribute to those transitions to help assess
the relationship between receipt of care and outcomes.
Research Question 3a: What is the rate of transitions experienced over the study
period and what are the points of care during which transitions are highest?
Research Question 3b: Are there individuals who consistently experienced a high
number of transitions and if so what are their characteristics?
Research Question 3c: What are the types of transitions experienced and what factors
influence these transitions?
Research Question 3d: What factors influence the number of transitions during the
peak points of care and overall?
Hypothesis:
i) The number of transitions experienced will be higher at the point of diagnosis and at
the end-of-life compared to the time period after diagnosis and before their end-of-life.
8
ii) The individuals who consistently experience a high number of transitions will have a
high number of co-morbidities in addition to dementia and behavioural symptoms of
dementia compared to those with a low number of transitions.
iii) Hospitalizations will drive transitions, but a large proportion of the cohort will also
move to LTC. Age, morbidity and behavioural symptoms will be the factors associated
with hospitalizations or a move to LTC. The lack of a caregiver will be significantly
associated with a move to LTC but not hospitalizations.
iv) The factors associated with a higher number of transitions will be age, gender (male),
morbidity, behavioural symptoms of dementia and the lack of a caregiver both during
the peak points and overall.
Research Objective 4 – Association of Guideline-Consistent Dementia and High
Quality Primary Care Received with Transitions Experienced by Individuals with
Dementia
The higher the number of transitions experienced, the higher the risk of morbidity
and mortality due to medication errors, poor communication, adverse drug interactions
and avoidable re-hospitalizations
33
. The clinical trajectories of individuals with
dementia have been well-documented and the care needs of these patients should be
anticipated by primary care doctors which would minimize the number of transitions
experienced. The chronic nature of dementia requires longitudinal care that provides
accessibility, care coordination and continuity in order to meet the ongoing
biopsychosocial needs of people with a primary diagnosis of dementia, which are also
9
the facets of high performing primary care (Boustani, Sachs, & Callahan, 2007; Boustani,
Schubert, et al., 2007; Chodosh et al, 2006).
There is good evidence for the benefits of accessibility, continuity of care and
coordination, all of which are key features of high quality primary care. Family
physicians are best positioned to ensure effective care coordination which has been
shown to be associated with better outcomes
38–40
. However, many of these studies
have been conducted in adults and young children but not in elderly populations and
particularly not in vulnerable elderly populations such as those with dementia
23
. As
well, guideline-consistent dementia care is thought to be associated with better
outcomes, though barriers to implementing these guidelines indicate not enough
studies demonstrating this association
41
.
My final objective therefore is to examine the association between receipt of
guideline-consistent dementia care and/or high quality primary care and the number of
transitions experienced by individuals with dementia.
Research Question 4a: Is receipt of guideline-consistent dementia care in the year of
diagnosis associated with the number of transitions the patient experiences in that
year?
Research Question 4b: Is receipt of high quality primary care associated with the
number of transitions the patient experiences in the year in which dementia is first
diagnosed?
10
Hypothesis:
i) Holding all else constant, receipt of dementia care that is consistent with guidelines,
and appropriate high quality primary care that encompasses continuity of care and
coordination will contribute to fewer transitions.
ii) Receipt of guideline-consistent dementia care will be associated with receiving high
quality primary care as measured by continuity of care.
1.3 Thesis Roadmap
This thesis includes 8 chapters in total. Chapter 1 (which you are currently
reading) serves as a preamble, providing an executive overview of the thesis and the
main objectives examined.
In Chapter 2, I provide a literature review relevant to my work. I define dementia
and describe its impact on afflicted patients, those immediately surrounding them, and
the societies and communities in which they are embedded. I then explore the care that
patients with dementia receive within and outside the primary care context, particularly
examining the role of guidelines in defining what is considered good dementia care. I
provide an overview of variations in dementia care (which is explored in more detail as
part of a systematic review in chapter 3) prior to describing the conceptual framework
designed to guide and inform the analytic work of this thesis. Following this, I describe
the current literature on transitions and their impact on patients.
Chapter 3 is a systematic review of the relevant literature on measures of
dementia care processes and physician practice patterns in the context of providing
11
guideline-consistent dementia care. The review highlights significant differences in
dementia care practice patterns indicative of inconsistent care received by patients with
dementia, but also the limitations in the study methods employed, clarifying the need
to examine, at a population-level, the differences in dementia care received by patients
and the patient characteristics associated with those differences.
Chapter 4 provides a description of my research design, data sources and
methods of analysis. It contains a brief history and description of the use of
administrative data in research and outlines in detail the linked administrative datasets
utilized, variable definitions and operationalized.
Chapter 5 describes my attempt to fill the gap in knowledge identified through
the systematic review in chapter 2. It is a population-level analysis examining the
receipt of guideline-consistent dementia care in BC during the first year of diagnosis. I
describe the proportions of individuals who receive guideline-consistent dementia care,
variations in care experienced and the patient characteristics that influence them.
Chapter 6 descriptively assesses transitions experienced by patients with
dementia from the year of diagnosis to a decade later or end-of-life. I explore patterns
and types of transitions, identify points of care when transitions are highest as well as
the demographic factors that influence these transitions.
Chapter 7, the final empirical chapter of my thesis contains a study that
examines if there is an association between the nature of guideline-consistent dementia
care patients receive at initial diagnosis, high quality primary care and the number of
transitions experienced.
12
Chapter 8 is the concluding chapter to my thesis which summarizes my results,
situates those results as contributions to the field, and makes recommendations for
policy, practice and future research.
13
CHAPTER 2- BACKGROUND & LITERATURE REVIEW
2.1 What is Dementia?
Dementia is an umbrella term that applies to a class of disorders characterized by
memory loss and affecting higher brain function as the brain deteriorates
3
. It can be
chronic or progressive in older adults, reversible (sometimes as a symptom of an illness
which when treated, reverses) or irreversible. The irreversible dementias include
vascular dementia, frontotemporal dementia, Lewy body dementia, Creutzfeldt-Jakob
disease and the most common, Alzheimer’s Disease (AD). Worldwide, 24.3 million
people were estimated to be living with dementia in 2001 with that number doubling
every twenty years
1
. The highest prevalence of dementia is in China, and the majority
(60.1%) of all people with dementia live in developing countries
1
.
In 1907 Dr. Alois Alzheimer presented the clinical and neuropathological features
of “[a]n unusual illness of the cerebral cortex” seen in a 55-year old woman he had
observed for five years in a Frankfurt asylum
42
. The disease that bears his name is now
the most common form of dementia (accounting for 50-60% of all cases), with an
exponential increase in incidence in those aged 65 and over in developed countries
3
.
Alzheimer’s Disease can be either sporadic or familial. Familial AD is an autosomal
dominant disorder with the age of onset generally before age 65, however this is a rare
form with a low prevalence of 0.1%
43
. In the sporadic form, which is the most common,
the apolipoprotein E allele 4 has been shown to contribute most of the genetic risk. The
key clinical symptom is progressive memory loss, and two distinct neuro-pathological
features: extracellular plaque deposits and neurofibrillary tangles. Over one hundred
14
years after its identification, the cause(s) of AD are still debated and a definitive
diagnosis of AD can still only be made post-mortem.
Vascular dementia is the second most common form of dementia in the elderly
and is usually a result of a series of small, unnoticed strokes, often co-existing with AD.
It is characterized by either acute onset (for a large stroke) or a step-wise progression of
cognitive impairment
44
. Often, though, patients present as mixed vascular and AD
dementia. Other dementias include frontotemporal dementia (including Pick’s Disease),
Lewy body dementia, Creutzfeldt-Jacob Disease and dementia that occurs with chronic
neurodegenerative diseases such as Huntington’s disease, Parkinson’s disease and
Multiple Sclerosis which have the same gradual progression of cognitive decline as AD
2,45
.
Early-stage dementia for the two most common forms presents as deteriorating
memory and a decline in verbal fluency
46
. Other common symptoms include aphasia
(language disorder), apraxia (inability to articulate thoughts or physically execute
learned movement), agnosia (inability to recognize objects), deterioration of higher
cortical function (confusion, disorientation) and behavioural disturbances (depression,
agitation, delusion). Dementia is diagnosed when these symptoms progress to the
point where they affect and individual’s ability to care for themselves
3
.
A definitive physiologic-based diagnosis of dementia can still only be made
post-mortem by neuropathology. However, an accurate functional diagnosis can be
made through a combination of clinical evaluation, cognitive screening, laboratory
testing and structural imaging
5,32
. Canadian consensus guidelines outline three
conceptual components to the diagnostic process: 1) the clinical diagnosis, 2) searching
15
for the potential cause and 3) identifying treatable comorbid conditions or factors that
may by contributing to the dementia
32
.
The clinical diagnosis is based on the patient’s history, collateral history from a
family member or caregiver (without the patient present), a physical examination and a
cognitive assessment. The patient’s history would indicate the cadence of the illness
and other potential risk factors or causes which can be confirmed by the collateral
history from the family member. For example, a gradual onset is often indicative of AD
while a step-wise progression is more consistent with vascular dementia. Alcohol abuse
may be a cause for dementia and a family history of dementia is considered a risk
factor. The physical examination would be conducted to search for specific signs such
as a possible stroke. Finally, cognitive function can be tested using any number of
validated cognitive tests, all of which provide an indication of the severity of memory
and cognitive loss.
Once the presence of dementia is established, the physician would search for a
specific cause, first by ruling out other medical and potentially reversible reasons for
dementia using further clinical evaluation, laboratory testing and imaging. For example,
renal failure, a brain tumor or subdural hemorrhages can present as memory loss but
are treatable and easily identified via laboratory testing and imaging. As well,
individuals with dementia are at risk for delirium and depression, both of which may be
further explored by the physician. A combination of these techniques can allow for an
accurate clinical diagnosis of dementia, however the challenges of making this
diagnosis should not be underestimated given the range of symptoms and confounding
factors present in patients.
16
A diagnosis of dementia is recorded in the patient’s record, but also in diagnostic
codes that physicians use to bill for their services. Those diagnostic codes are made up
of a United Nations’ sponsored standard classification system used by the World Health
Organization and adopted internationally to classify diseases as well as more nuanced
symptoms, signs, abnormal findings and external causes of injury known as The
International Classification of Diseases and Related Health Problems (ICD)
47
. The system
provides a means of mapping health conditions under general categories with
variations of the condition assigned by codes that can (in its most recent version) be up
to six characters long. It enables a comparability in the classification, storage and
retrieval of health data and statistics across jurisdictions and over time, and is used for
compiling morbidity data, mortality data, as part of reimbursement systems, and as an
input to resource allocation decisions
48
.
The ICD is periodically revised. Administrative data in BC include both the 9
th
(ICD-9) and 10
th
(ICD-10) revisions of the system
49
. The 10
th
revision of the ICD codes
involved a functional change in structure and was not a simple updating. This resulted
in more specificity of information conveyed through the codes, an increase in character
length to allow the ability to add more codes and fuller code titles to better reflect
medical advances in knowledge
50
. There are several codes that capture dementia in
both the ICD 9 and 10 revisions (Table 2.1 and 2.2).
17
Table 2.1 - International classification of disease codes related to dementia, version 9
ICD VERSION CODE DESCRIPTOR
ICD-9 290.0 Senile dementia uncomplicated
ICD-9 290.2 Senile dementia with delusions or depressive features
ICD-9 290.3 Senile dementia with delirium
ICD-9 290.4 Arteriosclerotic dementia
ICD-9 294.1 Dementia in conditions classified elsewhere
ICD-9 331.0 Alzheimer disease
ICD-9 331.1, 331.11,
331.19, 331.2,
331.82, 331.89,
331.9
Other cerebral degenerations
ICD-9 797 Senility without mention of psychosis
Table 2.2 - International classification of disease codes related to dementia, version 10
ICD VERSION CODE DESCRIPTOR
ICD-10 F00 Dementia in Alzheimer’s Disease
ICD-10 F01 Vascular Dementia
ICD-10 F02 Dementia in other diseases
ICD-10 F03 Unspecified dementia
ICD-10 G30 Alzheimer’s Disease
2.2 The Social and Financial Impact of Dementia
The first of the largest recent Canadian birth cohort – the baby boomers - turned 65 in
2011, and by the year 2015 Canada will have more individuals aged 65 and older than
individuals under 15 years of age
51
. Dementia is one of the most significant causes of
disability among Canadians aged 65 and older. It is projected that by 2038 nearly 1.12
million Canadians (2.8% of the Canadian population) will be living with dementia
2
. The
nature of dementia means that as it progresses, individuals require increased help with
day-to-day ADLs and social support. This additional responsibility usually falls to an
informal caregiver who is often a spouse or adult child. Caregivers of patients with
dementia dedicate on average 90 hours per week providing care and managing
treatment while also bearing financial costs that are both direct (such as home support
18
or therapy) and indirect (such as lost income) in order to spend time assisting patients
52
. The social consequences of dementia therefore affect both the patient and the
caregiver(s).
Unlike other chronic diseases, dementia affects cognitive function as opposed to
physical function. Because of this, less is known about the inner needs and experiences
of individuals with dementia as they become unable to express themselves verbally.
The literature reviewed in the next chapter reveals that patients with early stage
dementia describe a loss of control and self-identity, as well as a range of emotions
including anger, fear, shame, frustration and stigmatization
53
. Many attempt to
continue their lives as best they can after the diagnosis, but feel isolated socially and by
their healthcare professionals. They expect suitable individual care from health
professionals who have the knowledge to meet their needs at the different stages of the
disease, but express frustration at their needs being overlooked or misunderstood, and
at no longer being treated as accountable adults
53
. Social integration, which is
emphasized in the care of younger adults with cognitive impairment, is often replaced
with an emphasis on safety and efficiency when it comes to seniors, with the implication
of reduced freedom or autonomy
54
. The 2003 World Health Report Global Burden of
Disease estimates that dementia accounts for 11% of total years spent with a disability
in individuals over the age of 60. This is more than all cancers, stroke or cardiovascular
disease combined
1
.
Informal caregivers of individuals with dementia describe feelings of isolation,
anger, limited social support and feeling overwhelmed in addition to physical and
psychological distress
55
. Caregivers of individuals with dementia have 46% more
19
physician visits, 71% more prescribed medications, 63% higher mortality risk, poorer
self-rated health and a higher incidence of depression and anxiety compared to
matched non-caregivers
55
. Unsurprisingly, the leading reason for institutionalization
of seniors with dementia is caregiver distress
56
.
The annual cost of healthcare for a person with dementia is estimated to be
approximately three times that of a comparable person without dementia (accounting
for direct medical and non-medical costs)
57–60
. Costs additionally increase dramatically
with increased disease severity
61
. Half of all cost is associated with hospitalizations
(including for potentially preventable ambulatory-care sensitive conditions)
59,62
. While
there is some variation in cost estimates associated with study design (the costs
emerging from the more common cross-sectional designs are higher than those from
longitudinal designs
63
, the cumulative economic costs of dementia are incontestably
substantial -- projections for Canada as high as $872 billion by 2038 have been
reported
2
. In short, dementia represents a clear and present danger for all unprepared
health care systems.
2.3 Care for Dementia in the Primary Care Setting
The care of patients with dementia in the primary care setting is complex. Dementias
often go un- or under-diagnosed in primary care, with some estimating as many as
two-thirds of patients with dementia not being identified
8,9,64
. Most studies indicate
that patients with dementia have a higher burden of co-morbidity than comparable
controls, though there is still some debate around this in the literature
62,65
. Patients
with dementia who also have co-morbidities often require a higher level of care
20
because of difficulties with communication and management
4
. This is complicated by
multiple prescriptions to manage these co-morbidities in addition to psychotropic
drugs for behavioural and psychological symptoms of dementia. Drug interactions can
result in higher use of other parts of the healthcare system
66
. The interaction of
cognitive, behavioural and functional symptoms of dementia leads to a decreased
quality of life for the patient, and often, for (particularly informal) caregivers as well
55
.
In BC, as in many other jurisdictions, primary care doctors are the first point of
health care system contact for patients or family members who suspect a cognitive
problem. Primary care physicians therefore are in the unique position to make the
diagnosis of dementia and to set in motion decisions about clinical care, guided by
well-established dementia care guidelines (discussed in more detail in section 2.5)
8,27,67
.
The primary care physician needs to first provide an accurate, and preferably early,
diagnosis. They then, more importantly, need to provide long-term management of
dementia in the context of additional co-morbidities
7
. Current dementia care also
recognizes the need for counselling and frequent monitoring of the health status of
both the patient and the caregiver (regardless of whether the caregiver is a patient of
the physician) as informal caregivers play such an essential role in longitudinal
dementia care
8
. The initial diagnosis, counselling and management of symptoms fall
within what is considered ‘dementia care’ as outlined by guidelines (see section 2.5) and
ideally occur within the first year of diagnosis. However, as previously noted, dementia
is a chronic, progressive disease and individuals with dementia often have a high
burden of co-morbidity requiring physicians to provide longitudinal primary care to
manage the dementia in the context of those co-morbidities.
21
High quality primary care contributes to better outcomes for patients including
better management of chronic conditions, fewer preventable hospitalizations, shorter
lengths of stay, reduced use of emergency rooms and reduced risk of mortality
20
. It has
been defined as having four main features: first point of access for each new need;
longitudinal person-focused (not disease-focused) care; coordination of care provided
in other settings or by other practitioners; and comprehensive care that addresses most
health needs
22
. While less is known about the association of receipt of high quality
primary care with better outcomes of care specifically for people with dementia, it is
increasingly clear that in order to meet the ongoing biopsychosocial needs of people
with dementia, particularly in the context of co-morbidities, all the facets of high quality
primary care as defined by Starfield will need to be addressed
8,18,19
.
The complex range of cognitive, social and functional problems involved with
dementia often makes it difficult for primary care physicians to manage on their own.
The participation of physicians in multidisciplinary professional groups is
recommended, often involving a social worker, geriatric psychiatrist and nurse case
manager to ensure effective care coordination and to provide support to the patient
and caregiver
68
. Physicians are also an important source of referral to home and
community services. While, it would be ideal for patients to learn of these support
services from physicians during their visit, awareness and outreach have increased so
that referrals can now come from nurses, social workers, family members or the
dementia patients themselves
69
.
All in all, primary care physicians are a pivotal hub. The literature however, also
indicates several barriers to practice. Busy primary care physicians have expressed
22
concern about the lack of resources, expertise and time to provide the level of dementia
care required resulting in delayed recognition of dementia and adverse outcomes for
the patient and their caregivers
70
. The complexity and challenges associated with
behavioural symptoms that accompany dementia are particularly difficult challenges for
physicians
71
. Finally, despite a high degree of accuracy when dementia diagnostic
guidelines are applied, physicians often do not change their practice behaviour to align
with widely available guidelines for dementia care
26,28
.
2.4 Overview of Services Outside Primary Care
There are a broad range of health and social services available for all seniors that
dementia patients and their families may access beyond the standard set of services
offered by physicians and hospitals. Patients with dementia generally need both clinical
and social services to support them in their longitudinal care. While the primary care
physician is often the first contact at the diagnosis of Alzheimer’s disease and other
dementias, dementia-specific home and community support services are available and
have grown over the years.
In BC, access to other publicly-funded services is mediated through an
assessment process
72
. As noted, referrals for assessment can come from a variety of
sources. Once a referral is made, there is a formal process by which the individual’s
physical, mental and functional state is evaluated, along with an assessment of the
adequacy of informal supports. All of this information is used to determine the amount,
intensity and duration of care individuals are eligible to receive through the public
health care system
73–75
. Publicly-funded services may still have a provider pay
23
component, which is based on one’s ability to pay. Individuals of course also have the
option either of adding to these publicly funded services with additional care, or to
forego public services entirely
76
. Patients may have several ‘episodes of care’ where
they move in and out of the system as needed after their initial assessment, but many
older adults become long term users as they become more medically frail
72
.
The publicly-funded services potentially available in BC fall into three broad
categories: home services, community services and long-term care (LTC) services
77
(Figure 2.1).
Figure 2.1 – Assessment and publicly-funded service options in BC
Home services include home nursing, occupational therapy, physical therapy,
home support for ADLs and palliative care
72
. Community services range from respite
care (provided at home or in the community or on a short-term basis at a facility), to
adult day centres, specialized education and peer support programs run by health
organizations and local chapters of the Alzheimer’s Society. Use and review of these
services are still sparse, though several randomized control trials have demonstrated
24
their efficacy in specific settings
78,79
. If adults with dementia are no longer able to live
safely in the community or require a higher level of care than is possible with home
services, assisted living/supportive housing and LTC/residential care are available.
2.4.1 Home and Community Care in British Columbia
Remaining in the community for as long as is feasible is considered the best
possible option both by people with dementia and their caregivers
73,80,81
. The majority
of older adults prefer to continue living independently in their own homes, though a
diagnosis of dementia can eventually eliminate this possibility due to safety concerns.
Home care as well as other community-based services provide a means for older adults
with dementia (and their caregivers) to extend their length of stay in the community by
managing their health conditions at home and assisting with ADLs such as help with
bathing and preparing meals
82
.
Like many provinces in Canada, and consistent with the wishes of individuals and
their families, having seniors remain in the community for as long as possible has been
a priority for BC
83
. Recent analyses, however, have demonstrated that despite this,
access to and use of home and community care (HCC) services have decreased over
time for a variety of reasons including policy changes and shifts in the demographic of
seniors who might need home care
72,84–87
.
The literature surrounding the cost of home care has also seen a shift. While
home care was previously unequivocally considered more cost-effective for older adults,
current studies indicate that this is only the case under specific circumstances, as
informal care giving costs are often not adequately accounted for (e.g. loss of caregiver
25
income, caregiver stress and healthcare utilization)
72,88–90
. Regardless, remaining in the
community is overwhelmingly what most Canadian seniors (93% of those surveyed)
prefer
51,91
.
For individuals with dementia, remaining in the community as long as possible
has a number of benefits. Transitioning to a new environment such as a LTC facility is
stressful for older adults because of the nature of the change and the association of loss
11,80
. Home is associated with continuity, autonomy, competency, privacy and control
92–
95
. This is doubly so for individuals with dementia who feel that the home environment
provides a sense of personhood and normalcy
93,94,96
. In the face of cognitive and
functional disconnection and losses faced through dementia, a familiar environment
and routine is useful in the management of dementia care and can be provided with the
assistance of home and community care services
97
.
For many informal caregivers, being able to provide care to their loved ones at
home allows them to ensure that these individuals are receiving individualized care that
is appropriate and meets their needs and standards
80
. Given that individuals with
dementia often have multiple co-morbidities, primary care physicians who provide early
intervention with appropriate monitoring and management of the multiple conditions
may be able to help their patients remain in the community longer
26
. Additionally,
monitoring of caregiver well-being and assistance in mobilizing social support by
primary care physicians can help pre-empt and relieve some of the caregiver stress that
is often the trigger for institutionalization
12,55
.
26
2.4.2 Long-Term Care in British Columbia
Home-based services can help people with dementia age in place, but eventually
it often becomes very difficult for informal caregivers to manage care for people with
dementia at home, particularly during the later stages of the disease or when
behavioural symptoms become unmanageable. At that point a move to alternative
forms of housing that provide additional supports becomes necessary
98
.
In BC the housing menu is somewhat more limited for those with dementia
relative to seniors without dementia (Figure 2.1). Assisted living, which is described as a
residence that “offers housing, hospitality services and personal assistance to adults
who can live independently but require regular help with daily activities”
99
is a form of
care between home care and LTC, but is considered unsuitable for people with
moderate to more advanced dementia since eligibility criteria in BC (as with most other
provinces and states) requires that individuals be able to direct their own care
11,99
.
Therefore, individuals in Assisted Living whose dementia makes this impossible, face a
second transition to LTC --often the last, and only option
100,101
. But even at this level of
care, most LTC facilities as they are currently resourced, are inadequately equipped for
dementia patients. Dementia patients are often sequestered in separate or special
secured wings with different resourcing within the facility because they require more
support than the average LTC resident for any given level of other disabilities or
limitations.
While individuals with dementia undeniably require additional support, current
literature suggests that more home-like environments are better-suited for individuals
with dementia (as well as for all older adults living in LTC facilities). This means not just
27
‘home-like attributes’ implemented in the form of decor and the naming of facilities,
but social relational care practices as well that are meaningful to the persons with
dementia, their families and staff
81
. This idea is being developed in some LTC facilities
in BC based on several frameworks
102,103
but has been extended further by others. The
Dementia Village in the Netherlands and Beatitudes Campus in Arizona, USA are two
such examples. They feature open environments that encourage patients with dementia
to be autonomous, focuses on individuality and relationships while maintaining a
home-like environment, models that BC is using as templates
104,105
. Residents of LTC
facilities surveyed regarding their preferences have indicated support for this sort of
social integration as opposed to an over-emphasis on safety and efficiency
54
.
Unfortunately, many of the facilities in North America that provide specialized
dementia care still have institutionalized settings and care structures. The housing
landscape has yet to shift to provide alternate options for dementia patients, though
several new models have been developed in North America and are being evaluated
106–
108
. For now, the current, traditional, institution-like LTC facility remains the prevailing
housing option.
Primary care experiences are also different once a move to a LTC facility is made.
Many LTC facilities have in-house physicians or specific family physicians responsible for
the care of the majority of the facility’s LTC residents. In these cases, the patient-
physician relationship developed while a patient lives in the community often ends
when the patient becomes a LTC resident; responsibility for primary care is transferred
to the physician practicing at the LTC facility to which the patient has been moved
109
.
In BC, only 54% of general practitioners provide residential care services with an 18%
28
increase in the average annual number of residential care visits over the past ten years
(2003/04-2012/13)
110
. Indeed, there has been a 13% drop in physicians providing
residential care services since 2003/04 despite a 10% increase in the number of
residential care patients and a 10% increase in the number of general physicians over
the past ten years
110
. This is reflected in the declining comprehensiveness of primary
care. Physicians practice in fewer care settings and have a more narrow scope of care
111–113
. A recent study demonstrated that the care of 90% of LTC residents in Ontario
was accounted for by only 50% of family physicians with a mean of 42 residents per
physician
114
. A similar concentration of residential care services by a small proportion
of physicians is seen in BC, though some local health authorities (such as Cowichan and
Prince George) do not appear to have this trend
110
. This trend toward more
concentrated residential care services plays havoc with continuity of care and may
potentially also affect quality of care (as suggested by the high rates of inappropriate
prescribing amongst this population)
115,116
.
2.5 American and Canadian Guidelines for Dementia Care
Over the past thirty years, there have been several position papers and guidelines
published in the US and Canada on evaluating people suspected of having dementia
and providing subsequent dementia care. These range from lengthy, detailed
recommendations based on consensus expert opinion and/or literature review(s),
5,7,117–
120
to short user-friendly flowcharts with brief references
121–123
. Their intent is to
provide guidance on the care of individuals with dementia, and they are aimed mainly
at primary care physicians who are the initial point of contact for these individuals. The
guidelines have undergone significant evolutions since first published in the 1980s. The
29
more recent offerings are based on a better understanding of the long-term social
implications of dementia, not just for the patients but also for those who care for them.
Canada first produced national guidelines in 1989 through a national consensus
conference bringing together experts in the field
7
. At the time, the focus of the
guideline was on accurate diagnosis of dementia with no attention to treatment or
management following that diagnosis. This consensus conference was subsequently
repeated in 1999, 2006 and more recently in 2012. The resulting guidelines (with those
from the 2012 conference still in process) now include treatment and management
recommendations. The result of the 3rd consensus conference (in 2006) was the
dissemination of a series of six papers: risk factors and prevention
124
, investigating and
diagnosing dementia
32
, defining and diagnosing mild cognitive impairment
125
,
managing mild and moderate dementia
4
, pharmacological and non-pharmacological
therapy
126
and finally, managing severe dementia
127
. These contained 146
recommendations that reached strong consensus with the experts.
Three of these papers are of particular interest with regard to dementia care
guidelines. The second paper, on the investigation and diagnosis of dementia, provides
explicit detail for clinical diagnosis while contextualizing the difficulties in differentiating
the dementias and severity levels
32
. Using a vignette, it outlines a detailed clinical
evaluation including a history from the patient (focused on the cadence of the illness,
vascular risk factors and other risk factors), collateral history from an informant, a
physical examination and a formal cognitive test. Recommendations on core laboratory
tests remain relatively unchanged from the previous guidelines (complete blood count,
thyroid stimulating hormone (TSH), serum calcium, electrolytes, fasting glucose), with
30
the addition of recommendations for B12 measures in all older adults. Testing of red
blood cell folate and serum folic acid levels is now no longer necessary since the
introduction of folic acid into Canadian grain in 1998. Recommendations for
neuroimaging (primarily computed tomography and magnetic resonance imaging) are
still made; however, the committee determined that there is insufficient evidence to
recommend the routine use of functional imaging (magnetic resonance spectroscopy),
collection of biomarkers or neuropsychological testing. More importantly, the
guidelines make a special recommendation that the results of tests, diagnosis and
management be conducted over a series of visits spanning several weeks in order to
provide appropriate time to accurately identify dementia and prepare the patient for a
diagnosis. The committee concludes with the identification of what members
considered the most important current knowledge gap -- what is known about
diagnosis and what is actually practiced in a general physician’s office. The latter is a
specific focus of this thesis.
The fourth paper in the series, on approaches to management of mild to
moderate dementia, is of importance because of its guidance in managing a long-term
progressive disease like dementia with which most family physicians have had very
limited experience
4
. As with previous papers, a vignette is used as an example of how a
case can be approached. Fifteen bulleted points highlight important recommendations
that touch on disclosure, need for referrals to specialists (because of uncertainty about
diagnosis, request for second opinion, need for assistance with pharmacotherapy,
expressed interest in research studies, and/or inability to appropriately manage patient),
assessment of safety risks (driving, financials, activities of daily living etc), referral to the
local Alzheimer’s Society chapter and community resources and, finally, managing and
31
responding to therapy, functional problems, behavioural symptoms of dementia and
co-morbidities. While the guidelines warn physicians that the needs of the patient and
caregiver evolve and will require monitoring for increasing cognitive, functional and
behavioural challenges, they do not provide detail on the frequency with which a
patient should be monitored or the psychosocial counselling that should be provided.
The guidelines also mention the importance of caregivers in the management of
patients with dementia; the subject of caregiver support, while mentioned, is rather
cursorily covered. The focus is on meeting with the caregiver regularly to assess the
status of the patient with dementia, with a particular focus on behaviour, and on finding
ways to deal with distressing problems regarding the patient. Multi-component
interventions such as providing education, counselling, support and respite for
caregivers is stated with no supporting details.
In the fifth paper regarding pharmacological and non-pharmacological
interventions, no non-pharmacological interventions are recommended outside of
physical exercise. Details on pharmacological interventions are provided, in particular
the effectiveness, selection and side effects of acetyl-cholinesterase inhibitors
126
.
Unlike Canada, the US has guidelines published by several different national
working groups and organizations which appear to work separately as opposed to
through a consensus process. For example, ‘The National Institute on Aging and the
Alzheimer’s Association Workgroup’ (commonly referred to as NINCDS-ADRDA), first
published guidelines in July 1984. These were updated in 2011, and both the original
and updated versions place explicit emphasis on clinical criteria
118,128
. They propose
different terminology for the classification of what they term probable AD dementia,
32
possible AD dementia and probable or possible AD dementia with evidence of AD
pathophysiology. ‘The Work Group on Alzheimer’s Disease and Other Dementias’
published a comprehensive guideline for diagnosis and management of dementia
which stressed the evolving complexity and required comprehensiveness of treating
dementia
5
. ‘The Quality Standards Subcommittee of the American Academy of
Neurology’ published three evidence-based reviews: Early detection of dementia: Mild
cognitive impairment
129
, Diagnosis of dementia
130
and Management of dementia
120
.
These were abstracted into guidelines by the American Geriatric Society Clinical Practice
Committee in 2003. Several other groups have released position papers regarding
clinical care for patients with dementia, notably the ‘American Association of Geriatric
Psychiatry’; however these were not guidelines and therefore were not reviewed
27
.
As outlined above, there is a deluge of national guidelines available on dementia
care, particularly in the US. Almost all the guidelines focus exclusively on the initial
diagnostic process which the literature has identified as a hurdle due to the low
diagnosis and treatment rate, particularly in the primary care setting (it is estimated that
less than 25% of individuals with Alzheimer’s Disease are diagnosed in Canada)
131
. The
guidelines across Canada and the US appear relatively complementary at least with
regard to the techniques and diagnostic processes, while differing on the classification
terminology. Unlike the US guidelines, the Canadian guidelines stress the diagnosis of a
more recently identified symptomatic, precursor phase to dementia, mild cognitive
impairment
125
. This is perhaps because the development of the cognitive test used to
identify mild cognitive impairment was pioneered in Canada
132
. Interestingly, in
contrast to Canadian guidelines, none of the US guidelines reviewed make reference to
specialist referrals or the conditions under which these referrals should be made. This
33
may be a reflection of the overall differences in the two healthcare systems, and in
particular the fact that patients in the U.S. are free to go to a specialist without a
primary care referral, so that primary care physicians play less of a quarterbacking role
south of the border.
2.5.1 Dementia Care Guidelines in British Columbia
For practitioners in Canada, both the US and Canadian guidelines are well
disseminated; however, most are written as lengthy research reports which are perhaps
daunting for busy practitioners with limited time. This may have prompted the
development of BC-specific dementia care guidelines which are more succinct, easy to
peruse and come with a decision support tool
122,133
. Additionally, BC did not cover
acetyl-cholinesterase inhibitors (AChI) until the launch of the Alzheimer’s Disease
Therapy Initiative in 2007; this may have also required the need for guidelines that take
this context into consideration
134
.
The BC guidelines were developed by a working group independent of the
Canadian Consensus Conference working group and reflect province-specific resource
availability and medical coverage
122
. The BC guidelines indicate that symptoms of
cognitive impairment should be suspected when there is a history that suggests
cognitive decline (emerging cognitive problems). This history might be derived either
from direct observation or reports from the individual’s social network. After a
comprehensive medical review and appropriate testing (physical exam, laboratory tests,
neuroimaging and administration of the standardized mini-mental examination) a
working diagnosis can be arrived at. With disclosure of that initial (or suspected)
34
diagnosis of dementia, physicians should have follow-up visits at least every six months
with the patient and their caregiver in which needs, concerns, safety planning, finances
and education should be discussed. Physicians should also be periodically reassessing
their patients and establishing a relationship with them and their family (and caregiver).
Outside of managing other co-morbidities and counselling, a trial of AChIs is
recommended for mild to moderate dementia, but patients using these therapies must
be monitored closely and frequently, particularly within the first three to six months.
Referral to a specialist (e.g. geriatric psychiatrists) is recommended when diagnosis or
management is problematic or uncertain, when patients or family members request a
referral, or when management issues are difficult. Finally, physicians are encouraged to
be aware of neglect and abuse, and to support patients to function as independently as
possible.
Acetyl-cholinesterase inhibitors were not covered under B.C.’s PharmaCare program
until the introduction of the Alzheimer’s Disease Therapy Initiative in October 2007.
134
.
Through this initiative, patients diagnosed with mild to moderate dementia were
eligible for coverage of their AChI medication (except for Memantine, a different class of
drug more recently introduced to the market for moderate to severe dementia) as long
as they enrolled in the program. The Alzheimer’s Disease Therapy Initiative was
designed primarily as a research study to address what the BC Ministry of Health saw as
a lack of clinical evidence for the effects of AChIs
135
. Physicians were required to assess
patients’ cognitive status every six months for the patients’ continued enrollment and
therefore coverage of their medication
45
. The study’s participant recruitment was
35
completed in October 2011 and the study reports that over 20,000 patients are enrolled
and will continue to receive coverage while in the study.
In 2007, the General Practice incentive program was introduced in BC
136
. The
program is meant to support and compensate general practice physicians for time
spent on providing guideline-consistent care in several areas including chronic care and
mental health. The Mental Health Initiative provides several new billing codes to
encourage physicians to take more time in their assessment and support of patients
with mental health illnesses, including dementia, as well as to participate in broader
care teams while accepting responsibility for providing longitudinal, coordinated care
for the patient
137
. While the Mental Health Initiative attempts to address many of the
facets of good dementia care by incentivizing physicians to take an in-depth record of
the patient’s history, conduct appropriate memory assessments and provide ongoing
counselling, a recent report from the Ministry of health demonstrates very slow uptake
of these incentives
138
. Billings for the specific codes (G14043 – GP Mental Health
Planning Fee, G14046 to 14048 – Mental Health Management Fee for ages 60 to 69, 70
to 79, 80+) have increased as have the expenditures on these services, however the
number of services billed indicates that these codes are not being consistently used by
physicians. There is also no reliable way of determining whether the services described
by the fee items are actually provided and whether physician practices have changed
due to these codes or whether services are being provided but being billed in more
traditional ways. For the purposes of this thesis it was not possible to assess more
specific physicians’ dementia care practice patterns using these Mental Health Initiative
billing codes and they will be excluded from any analysis.
36
2.6 Variations in Dementia Care
Variations in the use of clinical healthcare services have been reported in just
about any area examined
139–141
even after controlling for patient characteristics. In
dementia care, there is a large and growing literature available on variations in
pharmaceutical prescription use by individuals with dementia
115,116,142–145
; however less
is known about variations in the use of other dementia-care-related clinical services
such as cognitive assessments, counselling, referrals to specialists or community
services, frequency of prescriptions, monitoring, etc. These services are recommended
for providing appropriate dementia care and are described at length in dementia care
guidelines (as detailed in section 2.5 above). The use of guidelines have been shown to
reduce variation and maintain, or even improve quality of care
30
. Therefore large
variations in the use of these clinical services would be an indication of inconsistent care
across the province which could be associated with differences in care outcomes.
Variations in clinical service use by patients with dementia are often examined at
the individual-level. Of these, only a handful of studies have examined a fuller scope of
clinical dementia care services (though they are self-reported surveys), and comparing
across these studies reveals significant variation in the clinical services received by
dementia patients
25,26,64,146,147
. For example, Chodosh et al.
146
developed 18 dementia
care processes based on available dementia care guidelines. The types of processes
were comprehensive, ranging from assessments (cognitive status, activities of daily
living, behavioural problems etc.), treatments (care plans, advise to caregivers,
implementing non-pharmacological approaches, documenting medication side effects
and outcomes etc.), to education/support and safety, all of which are highly
37
recommended for receiving good dementia care. Adherence to the processes ranged
from 9 – 79% with 11 of the 18 processes scoring below 40% adherence. In fact, the
lowest percentages of patients to receive specific dementia care processes were in the
assessment and treatment category. However, almost 79% of patients were referred to
a community care support program. This is in contrast to Reuben et al.
25
who found
higher adherence to some assessment processes such as assessing cognition (69% of
patients) but again, low adherence in assessing functional status (20% of patients),
treatment processes (75%) and referrals to community support programs (0% of
patients).
Similarly, in one of the earlier studies that examined the use of healthcare
services by individuals with dementia, Callahan et al. demonstrated that less than 20%
of patients received any neuroimaging a year prior to or after being screened for
dementia and less than 50% received any diagnostic blood work to rule out reversible
dementias. More interesting is that Callahan et al. collated results from the patients’
Short Portable Mental Status Questionnaire and found that 76% of patients with
moderate to severe dementia had undocumented impairment as defined by the lack of
a dementia diagnosis anywhere in the outpatient file. Callahan’s results must be
contextualized by the time of publication. In 1995, dementia care guidelines were
available, but awareness was still being built and stigmatization was very strong. This
may help explain the very high percentage of undocumented impairment. Regardless,
the results from all these studies demonstrate widespread variation in dementia care
clinical service use and low adherence to guidelines, both of which continue to persist
to the present day.
38
While understanding variations in dementia care and the impact on outcomes is
necessary, understanding the factors that contribute to those variations is also
important as they may be modifiable by the healthcare system. There are several
possible explanations underlying these observed variations including system, provider
or patient characteristics. Physician practice patterns have been described previously as
contributing to variations in other healthcare areas
141,148–151
. In Canada, there are
63,000 physicians all of whom individually make daily clinical decisions that lead to an
average of $1.5 million worth of expenditures on health care services per year
152
.
Physicians therefore direct a vast amount of clinical healthcare resources and physician
practice style could contribute to variations in clinical healthcare services use by
individuals with dementia. Little research is available on physician practice patterns in
dementia care despite a recent study in which 95% of seniors reported having a primary
care physician
51
and estimates that almost 15% of Canadians 65 years or older suffers
from dementia
2
. On this premise, a systematic review of the literature was conducted to
better assess the current state of knowledge. This systematic review is presented in
Chapter 2.
2.7 Conceptual Framework
There is no single pathway of care that individuals with dementia follow, but
nearly all (if not all) journeys will include receipt of some clinical health services. The
conceptual framework described here outlines the characteristics and clinical services
that may shape care trajectories for patients with dementia.
39
In my conceptual framework (Figure 2.2), the horizontal arrow represents the
aggregation of the factors that ultimately contribute to the primary outcome being
measured, transitions. The characteristics on the left (green boxes: physician, individual
and system characteristics) are measurable characteristics that contribute both
warranted and unwarranted variations to the healthcare activities that are at the core of
the framework. Not included in this framework are characteristics not measurable using
administrative healthcare utilization data such as cultural beliefs, knowledge and
attitude about health and care use which can also affect the type of care received
22
.
The central core of the framework contains the two key healthcare activities being
measured, dementia care (based on processes of care outlined in best practice
guidelines), and primary care (services provided beyond the services measured by
dementia care). To the right is the primary outcome of interest, transitions experienced
by individuals with dementia. In this framework, transitions can be driven by receipt or
lack of dementia care or primary care. Care provisions are driven by physician, individual
and system characteristics.
40
Figure 2.2 – Conceptual framework guiding the thesis
Physician characteristics: Previous studies have indicated that physician age,
speciality, practice location (rural/urban), number of years in practice and type of
practice population (e.g. number of older patients and number of patients in practice
with dementia) are predictors of the type of dementia care provided
153,154
. Similar
physician variables were also associated with the provision of full service family
practice
112
, therefore their inclusion in this framework. As physicians direct a wide range
of clinical healthcare resources, their practice style will help shape trajectories of care for
people with dementia, and will ultimately shape outcomes of care. These characteristics
are depicted at the left hand side of the horizontal framework to indicate their
contribution both to the clinical services as well as to the primary outcome.
Individual characteristics: The variables used to measure individual demographic
characteristics are key predictors in the receipt/use of dementia and primary care as well
as, ultimately, care outcomes. I anticipate that inclusion of these characteristics such as
age, sex and neighbourhood socioeconomic status, will allow me to tease apart both
warranted and unwarranted variation. Another important characteristic is health status
41
(this measure is described in detail in section 4.4.1.1). Controlling for health status in a
regression model assessing variation in clinical services is central to being able to
interpret the magnitude and appropriateness of other variations which may be
unrelated to patient need; it is critical for being able to differentiate warranted and
unwarranted variations. As well, I measure for the presence of an informal caregiver,
described in more detail in Section 4.4.1.
System characteristics: The supply of general physicians, specialists and hospitals in a
region plays a central role in the type of care provided. For example, areas with fewer
specialists per capita will require primary care physicians to take on a larger role in
dementia care management for cases that might otherwise have been referred. As well,
areas where patients may have difficulty accessing laboratory or imaging facilities may
result in fewer laboratory and imaging tests being conducted for diagnostic purposes
155
. BC is divided into five geographic health authorities (Fraser, Vancouver Coastal,
Vancouver Island, Northern and Interior). The Fraser, Vancouver Coastal and Vancouver
Island health authorities contain major urban centres and a higher per capita supply of
specialists, general physicians and hospital beds. The influence of system capacity and
other factors such as this constitute the “system” characteristics within the framework.
The belief is that these factors will influence both the diagnostic and treatment/care
trajectories of patients with dementia.
Dementia care: Dementia-focused care will be defined and explored in detail in
Section 3.2.2 of Chapter 3 and Section 5.2.3 of Chapter 5, but briefly, it is care consistent
with guidelines for the diagnosis and management of dementia. These clinical services
are generally provided within the first two years of diagnosis and consist of services
42
required for diagnosis (laboratory tests, imaging, physical examination), symptom
management (pharmaceutical prescriptions), counselling immediately following the
diagnosis and referrals to specialists if required. Subsequent care falls within the
primary care category with regard to longitudinal management and any other services
outside of the ones mentioned.
Dementia care is a pivotal measure in this framework. Receipt of guideline-
consistent dementia care is thought to be associated with better outcomes for people
with dementia. Good dementia care emphasizes early diagnosis and management
which can reduce the number of hospitalizations experienced contributing to a lower
overall number of transitions. Measuring any unwarranted variations in receipt of
dementia care by individuals newly diagnosed with dementia can indicate inequality or
poor quality care, hence its inclusion in the core of the conceptual framework, to be
assessed as its own dimension as well as its association with outcomes.
Primary care: High quality primary care has been previously defined 22 and will be
measured by three characteristics, first contact with the patient, coordination and
referrals as needed, and continuity of care as previously defined
112
and described in
further detail in Section 7.2.3. High quality primary care has been shown to contribute
to better outcomes for patients, including better management of chronic conditions,
fewer preventable hospitalizations, shorter lengths of stay, reduced use of emergency
rooms and reduced risk of mortality
20,140
. Dementia is a chronic, progressive disease
that patients and their caregivers live with for several years. In order to meet the
longitudinal biopsychosocial needs for those patients over that time, particularly in the
context of co-morbidities, provision of primary care plays a key role. Primary care’s
43
inclusion in the center of the conceptual framework captures the importance of these
services with regard to longitudinal care for patients given that dementia care services
are primarily provided in the diagnosis year.
Primary Outcome: The final part of the framework suggests that the intensity and
scope of primary care, and dementia-focused care will affect a critical aspect of a
dementia patient’s life: the number and type of transitions experienced. The number of
transitions will be examined in more detail in 4.4.2, but briefly, transitions are defined as
a physical relocation from one place to another that involves at least one night’s stay.
Such transitions generally involve moves between home and a variety of care facilities,
or between care facilities such as respite care, hospital and LTC. A high number of
transitions in seniors diagnosed with dementia is particularly challenging due to the
stress of multiple unfamiliar environments. In the conceptual framework, this outcome
is shaped by the dementia care and primary care clinical service use as well as by
individual, physician and system factors. Good dementia care and primary care is
expected to positively influence this outcome, resulting in fewer transitions.
2.8 Primary Outcome Measure: Transitions and Consequences
For the purposes of the research reported here, a transition is defined as a situation
in which “an individual physically moves from one place to another and stays there for
at least one night”
10
. With an expanding and changing healthcare system, almost
everyone will experience at least one transition during his or her lifetime. For many
seniors, these transitions occur at various critical points during their trajectory of care,
often several times between different care settings. Age, gender and morbidity all
44
affect the number and type of transition (this is particularly true during the last few
years of life)
10,12
. Transitions between care settings pose a serious challenge to the
continuity of care and safety of the patient.
Care in multiple settings is often provided by physicians who do not have an
ongoing relationship with the patient and are therefore unaware of their complete
clinical or medication history
156
. Despite advances in technology, adoption of an
accessible, universal electronic database that spans across care settings is not yet in
sight and physicians continue to reduce their scope of practice, working primarily in
siloed, single settings
157,158
. This lack of continuity of care has led to evidence of
medication errors and quality deficiencies which pose significant threats to patient
safety
159
.
A study examining patients discharged from hospitals found that half of adult
patients experienced at least one error in medication continuity, diagnostic workup or
test follow-up
160
. Other studies find evidence of high adverse event rates after
discharge
14,161
with over half to 66% of events due to medication errors
161
. In another
study examining 30-day post-hospital care patterns, between 12 and 25% of all care
patterns were considered complicated and required re-admittance to higher intensity
care settings three weeks after discharge
13
. For patients experiencing a transition from
hospital to home, many are also unprepared for the challenges of self-care
responsibilities.
45
2.8.1 Transitions for Seniors
For seniors in particular, changes in health status can mean frequent transitions
between care settings. In the US, one in six nursing home residents are hospitalized in
any six month period
162
and as many as 80% of elderly people have an acute care
hospitalization just prior to their admission to a LTC facility
11,163,164
. Approximately 40%
of seniors living in the community are hospitalized every year
165
. These hospitalizations
can be precipitated by a variety of events including acute infections, acute episodes of
chronic illness and adverse events such as falls
166
.
Four system issues have been identified as particularly problematic in hospitalized
seniors: poor communication, preventable declines in health status, inadequate
discharge planning and major gaps in care during transfers to and from hospitals
167
.
One study showed that greater than 60% of chronically ill older adults living in the
community had medication errors while transitioning between hospital settings
14
; this
is particularly problematic considering the high risk of polypharmacy in seniors
168,169
.
Another study demonstrated increased mortality risk associated with transitions
15
. The
most significant consequence of these transitions for seniors, however, is preventable
hospital readmissions with as high as one fourth of all hospital readmissions of seniors
believed to have been preventable
16
.
2.8.2 Transitions for Seniors with Dementia
The challenges in medication, quality and continuity of care that affect the safety of
patients are compounded in dementia patients if no one is able to provide a verbal
history of their care or if they do not have a caregiver who can assist in their transitions
46
12,35
. The stress of unfamiliar environments is also particularly disruptive for people with
dementia for whom a stable environment is critical
11
. This can lead to additional
behavioural disruptions that may not be properly managed, trigger further transitions
and can lead to a preventable decline in health status in addition to the adverse
outcomes already experienced by seniors in transition
15
. Indeed, individuals with
dementia are significantly more likely to be hospitalized for all cause and ambulatory
care-sensitive conditions than non-dementia individuals
12,170
and experience a higher
number of transitions during the last two years of life
10
.
Dementia is a chronic and progressive disease with increasing intensity and need for
care over time. The clinical trajectories of these diseases have been well-documented
and the care needs of these patients should be anticipated by primary care doctors.
Guidelines for good dementia care stress the need for continuity and familiar
environments which cannot be consistent with high numbers of transitions
10,22,26
.
Family physicians are best positioned to support seniors to ensure effective care
coordination. However, seniors often have limited contact with physicians, particularly
once in a residential LTC facility
167,171
, while physicians have continued to reduce their
scope of care to working primarily in one setting which often excludes LTC
112
.
There is good evidence for the benefits of continuity of care. Studies have
demonstrated that good continuity of care is associated with fewer emergency room
visits
172,173
, fewer hospitalizations
172,174–176
and avoidable hospitalizations
38
, better
preventative care
177,178
, better medication adherence, better physician recognition of
medical problems
179
, and greater patient satisfaction
179
and communication
20
.
However, many of these studies have been conducted in adults and young children but
47
not in elderly populations and particularly not in vulnerable elderly populations such as
those with dementia
23
.
In summary, dementia is a common condition with increasing prevalence and
significant implications both for the person with dementia and for his / her caregiver(s).
Good evidence exists for health care that can help with both the diagnosis and ongoing
treatment of individuals with dementia. There is less existing evidence to help
understand how well those guidelines and care practices are followed in the actual
delivery of health care services, and what the implications might be for outcomes, most
specifically transitions. This thesis is aimed at filling in some of these knowledge gaps.
48
CHAPTER 3 – A SYSTEMATIC REVIEW OF VARIATIONS IN
DEMENTIA CARE PHYSICIAN PRACTICE PATTERNS
1, 2
3.1 Introduction
In Canada, primary care physicians are gatekeepers who are usually the first point of
contact for people with dementia
8
. Current guidelines for dementia care recommend
that in addition to providing a proper diagnosis, physicians develop a care program for
the patient and their caregiver(s), provide accurate information on the progression of
the disease, make referrals to local resources, and offer appropriate counseling on the
psycho-social impact of the disease
5–7,119
. Physicians therefore direct a wide range of
clinical healthcare resources.
Studies on physician practice patterns indicate variation in physician
methodologies and responses regarding dementia care processes
28,29
. There is a
substantial body of literature on physician prescription patterns with regard to
dementia
143,180–183
but in comparison, only a limited number of studies on other
physician dementia care practice patterns. While pharmacological management is
important, poor detection and general management means people with dementia may
not receive appropriate psychosocial interventions, routine monitoring or the
recommended multi-dimensional approach to dealing with increasing cognitive and
functional challenges
117
.
Best practice guidelines for dementia care have been established in the medical
community for several decades. The use of best practice guidelines have been shown to
1 A version of chapter 2 has been published. Sivananthan SN, Puyat JH, McGrail KM. Variations in self-reported
practice among physicians providing clinical care to individuals with dementia: a systematic review. Journal of
American Geriatrics, 2013. 61:(8) 1277-1285.
2
Copyright (2013) Wiley. Used with permission from publisher (John Wiley and Sons).
49
reduce variation and maintain, or even improve quality of care
30
, however there is no
existing comprehensive systematic review focused on physician practice patterns
associated with the care of people with dementia. It is important therefore to
determine to what extent actual practice, as reported in the literature, is consistent with
guideline-recommended care. My objective in this systematic review is as follows:
Research Question 1: Based on the current literature, to what extent is actual practice
consistent with guideline recommended care?
I hypothesize that there is wide variation in physician practice patterns with regard to
provision of dementia care, particularly discretionary dementia care processes.
3.2 Systematic Review Methods
I used the Centre for Reviews and Dissemination’s publication on guidance for
undertaking reviews in health care to develop the structure of this systematic review,
including the protocol and data extraction
184
(see Appendix A for study protocol). The
method also meets the criteria outlined by the Preferred Reporting Items for Systematic
Reviews and Meta-Analyses (PRISMA) statement for systematic reviews
185
.
3.2.1 Search Strategy
A search strategy was developed in consultation with a research librarian. Only English
language studies published as of March 1
st
, 2012 were included with no start date
limitation. I was deliberately broad in my search strategy to ensure I captured all
relevant studies. Eligible studies met all of the following criteria: 1) were experimental,
quasi-experimental (pre-post studies, interrupted time-series) or observational (case
50
control, cohort, cross-sectional) studies; 2) presented original empirical results; 3)
presented results at a physician-level; 4) measured physician care provided to seniors 60
years or older diagnosed with dementia; 5) included senile dementia, Alzheimer’s
Disease or vascular dementia under the term “dementia” as either a primary or
secondary diagnoses; 6) measured any clinical service/intervention or combination of
clinical services/interventions detailed as appropriate for dementia care management by
national and international guidelines;
5–7,119
and, 7) measured actual services provided
(vignette-based studies measuring behavioural intentions were excluded). Studies
focused exclusively on prescription patterns of AChIs, antipsychotics or psychotropic
medications were excluded as this is a vast body of literature and requires a separate,
specific review. Studies that assessed dementia care processes in specialist practice
settings only were also excluded as the practice patterns and patient demographics in
these settings are thought to be very different from that of general practices.
The databases searched were: Web of Knowledge, PubMed, Science Direct,
MedLine, PsychINFO, EMBASE and Google Scholar. The search strategy was adapted to
each database based on its thesaurus or Medical Subject Headings. An individual
search with each of the search terms was also conducted to ensure that no relevant
citations were missed. For Google Scholar, only a preliminary search was conducted for
the first 500 results to source any key grey literature and determine whether all key
citations were captured. References of key studies and all studies that met the inclusion
criteria after the abstract review were also scanned. While peer-reviewed and grey
literature was included in the original search, the final articles that met the inclusion
criteria were all peer-reviewed.
51
3.2.2 Data Collection and Data Extraction
I conducted the literature search on April 3rd, 2012. Another reviewer (Joseph Puyat)
and myself then conducted independent abstract and full-text reviews as outlined in the
PRISMA group recommended flowchart (Figure 3.1)
186
. The two lists of eligible studies
were compared at each round and disagreements resolved by discussion. Studies were
not eliminated on quality due to the limited number of eligible studies that met the
inclusion criteria.
For each eligible study, both reviewers independently extracted data. I
developed the data extraction tool and it was tested on three studies prior to
proceeding with the full data extraction. The following data items were obtained: study
year; setting (nursing home or community); study location; associated health networks;
study design; study purpose/research question; sample number; sample
representativeness; validation of survey; physician specialty; patient population;
physician characteristics; approach to screening; reports of diagnosis disclosure;
memory test; imaging; blood work; medical prescriptions; non-pharmacological
interventions; counselling; specialist referrals; community service referrals; and other.
Studies that conducted any form of regression or statistical analysis were noted and
their model covariates included.
I defined “dementia care processes” provided by physicians as diagnostic services
(memory test, imaging, blood work), management (pharmacological and non-
pharmacological interventions, counselling) and referrals (specialist referrals, community
service referrals). I developed definitions for each of these in order to standardize the
data for the studies that met the inclusion criteria. The definitions were grounded in
52
consensus guidelines
5,6,119
(except for mass screening for which there are no specific
recommendations) however, when studies diverged in their measurements of the
dementia processes (additional tests, techniques and frequency), the additional test and
lowest frequency were included in the definition. These are as follows:
Mass Screening – No guideline recommendations have yet been made regarding mass
screening. Study measurement included questioning >21% of patients 65 years and
older in the physician’s practice about memory problems
153
.
Disclosure – Guidelines recommend a physician discloses suspected diagnosis to a
patient and also to a family member when possible as soon as it is known
4,5
Diagnostic Services:
Diagnosis – The physician conducts a formal mental status or memory test to form an
initial diagnosis and for staging of dementia. Several tests are recommended by
guidelines including the Mini-Mental State Examination, Kokmen Short Test of Mental
Status, 7-Minute Screen, Memory Impairment Screen, Montréal Cognitive Assessment,
the DemTect, the General Practitioner Assessment of Cognition and the Behavioural
Neurology Assessment Short Form
32,121
. Study measurements included other mental
status test such as the Blessed information memory concentration test, short portable
mental status questionnaire, Washington University SDAT screening battery, Iowa
screening battery for mental decline, and Wechsler adult intelligence scale which were
not guideline recommended but used as a measure for formal diagnosis by some of the
studies
187
.
53
Imaging – A computed tomography (CT) scan or magnetic resonance image (MRI) of
the brain is conducted
5,32,121
Blood work – Guideline laboratory tests routinely ordered to determine the underlying
cause of dementia or for rule-out of other causes including a complete blood count,
TSH, serum calcium, electrolytes, fasting glucose, vitamin B12 levels and folate levels
5,32
.
Other laboratory tests measured by studies include chemical screening, sedimentation
rate, test for metabolic disorder, urine analysis and heavy metal screening
188
.
Management:
Intervention – All pharmaceutical interventions related to the treatment of dementia
(Donepazil, Rivastigmine, Galantamine, Tacerine, and Memantine). Also includes
prescriptions for behaviour and mood treatment such as antipsychotics, antidepressants
and psychotropics. Non-pharmacological interventions such as changes in ADLs,
environmental or communication changes were also included as interventions
5,32,121
.
Counselling – Provide counselling to patients with dementia and family members
regarding management and future planning. This includes financial planning, advanced
directives, end-of-life planning, respite care/caregiver stress, driving risks, management
of ADLs, a possible nursing home placement and general family counselling
4,5
.
54
Referrals:
Community Service – Referral to community services who assist patients and caregivers
to cope with dementia. Services include the Alzheimer’s Association, a support group,
an adult day centre, home health agency, respite care, area agency on aging, social
workers and recreational therapy
4,5
.
Specialist Referral – Referral to a Geriatrician, Neurologist, Psychiatrist, Geriatric
Psychiatrist, Neuropsychologist or to a Memory Clinic for further management
4,5
.
3.2.3 Data Analysis
The unadjusted proportion of physicians who reported use of each dementia care
process was either extracted, when available, or calculated, if possible. In cases where
double counting would occur because physicians report use of more than one
technique within a process (e.g. the use of more than one memory test), the technique
with the highest reported proportion was extracted.
3.3 Results
Through the initial search, 1,264 studies were identified. Of this, 1,222 studies
were excluded through a title review. Some of the excluded studies were centered on
caregiver practice patterns, others were qualitative studies, but the vast majority of
excluded studies were focused on physician prescription patterns in relation to
dementia care. For the full-text review, a kappa of 0.71 was reached indicating a formal
inter-rater agreement close to complete agreement (kappa=1)
189
(Figure 3.1). After
resolving differences in inclusion assessment through discussion and consensus, twelve
primary quantitative research studies met the final inclusion criteria
9,69,153–155,187,188,190–
55
194
. All the studies assessed actual services provided at a physician-level in patient
populations 60 years and older. Fifty percent of the studies focused exclusively on
primary care physicians encompassing general, internal and family medicine
practitioners
9,69,154,188,190,194
(Table 3.1). The other fifty percent of studies included
primary care physicians as well as specialists who were primarily geriatricians, as well as
neurologists and psychiatrists
153,155,187,191,193,195
. All twelve studies used cross-sectional
surveys, ten of which were either postal or web-based self-administered surveys
9,69,153,154,187,190,191,193–195
while the remaining two were interviewer administered semi-
structured surveys
155,188
. Four of the twelve studies used validated survey instruments
9,192–194
.
56
Figure 3.1 – Flowchart describing the approach used to identify all eligible studies
57
Table 3.1 - Characteristics of studies evaluating physician practice patterns in dementia care
Study Research Question Country Year Sample
(response
rate)
Study Design
(validated
instrument)
Sampling Strategy
(representative
sampling)
Physicians
speciality
assessed
Setting
Rubin Quantitative data on
state of PCPs diagnosis
and treatment of
dementia
USA 1987 50
(89%)
Interview
survey; (N)
Rockford Medical
Society; (Y)
Family, Internal Community,
Nursing Home
Somerfield What are physicians’
experience with
dementia, services
they provide and
factors for variation
USA 1991 53
(48%)
Postal survey;
(N)
Referrals to John
Hopkins, dementia
or cognitive
neurological clinic;
(N)
Family,
Internal,
Neurology,
Psychiatry
Community
Glasser What are practices of
rural physicians in
dementia care
USA 1993 102 (72%) Interview
survey; (N)
State directory,
medical society
records; (Y)
General,
Family,
Internal,
Specialists
Community
Bisset Can GPs care for
dementia patients in
community and their
view of current
resources
USA 1996 143 (41%) Postal survey;
(N)
Grampian state
directory; (Y)
General Community
Fortinsky How connected are
PCPs with community
services for dementia
care
USA 1998 255 (51%) Postal survey;
(N)
Academy of
Medicine of
Cleveland; (Y)
Family, Internal Community,
Nursing Home
Brown How do PCPs engage
in AD practice
behaviour
USA 1998 403 (27%) Postal survey;
(Y)
American Medical
Association; (Y)
General,
Family, Internal
Community,
Nursing Home
58
Study Research Question Country Year Sample
(response
rate)
Study Design
(validated
instrument)
Sampling Strategy
(representative
sampling)
Physicians
speciality
assessed
Setting
Cody Have educational
efforts affected PCPs
dementia care
USA 2002 142 (16%) Postal survey;
(Y)
State directory,
Annual Family
practice board
meeting; (N)
General Community
Cavalieri Do physicians give
advanced care
planning advice to
patients with AD
USA 2002 63
(23%)
Postal survey;
(N)
Kennedy Health
System; (N)
General,
Family,
Internal,
Geriatrics
Community
Fortinsky Comparing dementia
care practices of PCPs
in two states
USA 2009
422 (27%) Postal survey;
(N)
State directory; (Y) Family, Internal Community
Robinson What are the roles of
generalists and
specialists in dementia
care in Europe
France,
UK,
Germany,
Spain,
Italy
2010 500 Postal survey;
(Y)
Unknown;
(Unknown)
General,
Family,
Internal,
Specialists
Community
Baloch Does specialty or
geriatric training
correlate with comfort
in dementia care
management
USA 2010 134 (25%) Web-based
survey; (N)
Health Texas
Provider Network;
(N)
Family,
Internal,
Geriatric
Community
Cohen-
Mansfield
Comparing the
approach of MDs,
PhDs and NPs in
dementia related
behaviour
USA 2011 108 Web-based
survey; (Y)
American Medical
Directors’
Association; (Y)
All medical
practitioners
Nursing Home
59
There was one multi-jurisdictional study conducted across five countries in Europe
193
.
The other eleven studies were conducted in the US, but only two of these used
nationally representative samples
9,195
(Table 3.1, Figure 3.2 ).
Figure 3.2 – Geographical locations of eligible studies from the USA
*excludes one study based in Europe
The sample sizes for the studies varied from 50 to 500 practicing physicians. Fifty
percent of studies sampled from state directories
69,154,155,188,190,194
, 17% used national
directories
9,195
, 25% sampled from local hospital networks
153,187,191
and one study was
unknown
193
. Seven of the studies used representative sampling for the jurisdictions
they were surveying
9,69,154,155,188,190,192
and one study did not report on its sampling
strategy
193
. Ten of the studies reported response rates ranging from 16% to 89%
9,69,153–155,187,188,190,191,194
, though five of those studies reported rates below 27%
9,153,154,191,194
.
60
3.3.1 Variations in Dementia Care Processes
The scope of dementia care processes assessed by each study varied
considerably. The nine care processes were categorized based on current North
American consensus guidelines
6,7
. While ten of the study research questions identified
dementia care practice as the primary interest
9,153–155,187,188,190,193–195
, only four assessed
the full scope of care processes recommended by guidelines (at least one care process
within each of diagnostic, management and referral)
9,153,155,188
, three of which were
older studies (published prior to 2000) (Table 3.2). Most of these studies had low
response rates or fairly small sample sizes. All nine dementia care processes had wide
variations in the proportion of physicians who conducted each process (Table 3.2).
Mass screening for dementia is still highly debated so unsurprisingly, only two
studies assessed the proportion of physicians who screen regularly
153,190
. However the
proportion of physicians who screen for dementia appeared to be high at 68% and 78%.
Disclosure to a patient and family member had the least variation with consistently high
proportions of physicians (82% and 100%) reporting disclosure to both the patient and
family member (versus to only the patient or non-disclosure until diagnosis is certain) in
the two studies that assessed this process
153,194
.
Wide variation was seen in the formal memory testing process. The studies
reported higher proportions of physicians who assess mental status, but these were not
formal or written memory tests
9,153,155,187,188,190,194,195
. Of the eight studies that assessed
formal memory testing, six reported that the proportions of physicians who conducted
formal memory tests were less than 60%
9,155,187,188,190,195
(Table 3.2). Indeed, three of
these studies reported proportions below 15%
155,188,195
. The lowest proportion of
61
physicians using a formal assessment tool (4%) was reported in 2012
195
. Similarly, the
intervention process, which encompassed a wide range of pharmaceutical medications
both for dementia and consequent behavioural problems,
62
Table 3.2 - Variations in physicians practice patterns measured by proportions for each dementia care process
Study Guideline Recommended Dementia Care Processes
Diagnostic Management Referral
Memory
test
Imaging Blood
work
Intervention Counselling Community
service
Specialist referral
Rubin (1987); n=50 12% 80% 68% 48% 84% 26% 12%
Somerfield (1991); n=53 57% 58%
Glasser (1993); n=102 <10% 72%* 75% 33% 44% 18%
Bisset (1996); n=143 54% 70%
Fortinsky (1998)†; n=255 83%*
Brown (1998); n=403 40% 33% 36%* 47%* 21%
Cody (2002); n=142 96% 83%*
Cavalieri (2002)†; n=63 81%* 53%*
Fortinsky (2009); n=422 90%* 77%*
Robinson (2010); n=500 49% 39% 36%* 64%
Baloch (2010); n=134 84% 34%* 94%* 91%* 81%
Cohen-Mansfield (2011);
n=108
4% 91% 58%
*Indicates highest possible value for each process (e.g. Baloch et al., examined two imaging techniques, computer tomography
and magnetic resonance imaging. Some physicians use both techniques, so the technique with the highest proportion, computer
tomography, was taken)
†Studies whose research objective was not to measure all dementia care processes
No text, process not assessed
0 – 20%
21 – 40%
41 – 60%
61 – 80%
81 – 90%
63
reported proportions ranging from 33% to 91%
9,154,155,188,193,195
. Four of the six studies
reported proportions less than 50%
9,155,188,193
, which were primarily older studies and
may be a reflection of the change in the available pharmacological therapies.
Brain imaging and blood work processes are both recommended by consensus
guidelines to aid in diagnosis. The former recommendation applies only in some
circumstances and therefore is not a universal guideline. Imaging ranged from 33% to
80% in the proportion of physicians reported using it as a diagnostic tool
9,153,155,188
(Table 3.2). Physicians primarily made use of CT or MRI. All three studies that assessed
the blood work process reported consistently high proportions of physician utilization
at almost 75% and above
153,155,188
(Table 3.2).
The proportion of physicians who reported providing some form of counselling
153,155,188,191,193,194
was also consistently high with only two of six studies reporting less
than 80%
155,193
. Community service referrals on the other hand had a 3-fold variation
(26% to 83%)
9,69,154,188,191,193
with four of the studies reporting proportions below ~50%
9,188,191,193
, while specialist referral had almost a 7-fold variation (12% to 81%)
9,153,155,187,188,190,193,195
with newer studies (published after 2000) reporting higher
proportions.
3.4 Discussion
Existing literature shows wide variation in the proportion of physicians who report
conducting the dementia care processes recommended by guidelines. This was of
particular note for the formal memory testing process which had a 24-fold difference in
the proportion of physicians who reported conducting a formal test. Several factors
64
could have contributed to these care process variations including geography, resource
constraints, patient demographics and changes in both recommended practice and
approaches to practice over time
148,196
. The high heterogeneity of the included studies
prevented any aggregate of results or calculation of adjusted proportions. The study
protocol for this review used liberal definitions for each care process. This would tend to
push up the proportions reported here, and thus decrease variation. Given the range in
proportions reported, the observed variation in physician practice patterns is, if
anything, an understatement of ranges in actual patterns of practice.
Only four of the ten studies that identified dementia care as a primary research
objective included measures of the full scope of dementia care processes as identified
by guidelines. Given the geographic homogeneity of the studies (mainly US based) and
the identified research objective for each study, it was expected that the scope of
dementia care would be relatively synonymous across studies. The publication date of
the studies seems an unlikely explanation as almost all the studies that did measure the
full scope of dementia care were older studies (published prior to 2000), indicating that
the studies appear to be getting less comprehensive over time.
The range of publication dates should be taken into consideration when
examining the variation within each dementia care process, due to the shifting and
updating of guidelines and understanding of dementia over the years. Guideline
recommended dementia care was first developed in 1984 in the US, prior to the oldest
study in this review
128
. However, it contained no specific recommendations regarding
the management of dementia with an emphasis placed instead on correctly diagnosing
dementia. These guidelines were only recently updated
6
, however, other evidence-
65
based best practice guidelines published in the interim provided details on the
management of dementia, focusing on long-term counselling and appropriate referrals
120
. The more recently published studies reflect this shift in dementia care
understanding with higher proportions of physicians reporting counselling, community
service and specialist referrals while the older studies were reporting on practice
patterns in the absence of best practice guides.
The widest variation for any care process was noted in the formal memory testing
process (4 - 96%) which is the most mature of all the dementia care processes, having
guidelines and a large number of formally validated tools available for over three
decades
188
. It was surprising then that this was not the primary process used by
physicians to identify dementia. It may be that physicians are asking questions about
memory during the history-taking process, but formal tests are still necessary for a
diagnosis
188
.
Similarly, the specialist referral process had relatively wide variations in physician
proportions, but with more of a trend toward higher specialist referrals in newer studies.
This may be reflecting the shift in physician scope of practice over time
111
.
There is some contextual information (Table 3.1) that can aid in the interpretation
of the wide variation noted in the dementia care process. For example, Somerfield et al.
187
conducted their survey on physicians who had made at least one referral to a
dementia or neurology clinic. These physicians are assumed to be more knowledgeable
about dementia and may not be a representative sample, which could have contributed
to the higher rate of specialist referral. Glasser et al.
155
conducted interviews on
primary care physicians practicing in rural settings which contextualizes the lower
66
specialist referral rate. Finally, Cohen-Mansfield et al.
192
primarily focused on
physicians’ approach to dementia-associated behavioural issues in nursing homes
therefore those physicians who were surveyed are more likely to provide an intervention
which may explain the higher intervention rate. Given these differences in the available
studies that met the inclusion criteria, the general lower quality of existing studies and
the low number of studies, I am limited in my ability to draw firm conclusions about the
extent of the variation in the dementia care processes measured.
3.4.1 Limitations
There are some other limitations to this analysis. All the studies that met the
inclusion criteria were self-reported cross-sectional surveys with varying sampling
strategies, primarily small samples sizes, and low response rates which may be
associated with non-response bias and give rise to results that are less likely to
represent the target population. None contained data extracted from medical records
or administrative data, which would be a better indication of actual services provided.
Due to the limited number of studies on physician practice patterns in dementia care, I
made no further exclusionary decisions on the quality of the studies, which may have
contributed to the variation noted and influenced the quality of the results. Several
physician specialties were included in the practice pattern assessment but differences
among these specialities were not described as almost all the studies pooled specialists
and general physicians in their assessments and of those that conducted regression
analysis to determine the effect of speciality on the dementia care processes assessed,
only one study
153
noted a difference. Similarly, several studies did not describe what
was considered an intervention in their study measure process, therefore I was unable
67
to differentiate whether these were pharmaceutical or non-pharmaceutical interventions
and chose to pool the results.
Self-reported results only demonstrate immediate recall and are prone to recall
bias. However, since self-reported measures are susceptible to social desirability
response bias, participants often over-estimate their answers, yielding higher than
actual proportions
197
. The wide variation in each dementia care process that still
persists suggests that this bias is not a serious limitation. I was also unable to adjust for
the heterogeneity of the studies and only extracted unadjusted proportions. The use of
English only studies and primarily databases with peer-reviewed studies introduces the
possibility of language and publication bias to this review. Finally, while the MeSH
headings used in the search strategy appeared to index studies appropriately, they may
not have been consistently used therefore there is some likelihood of missed studies.
My search terms were broad precisely to mitigate this possibility.
3.5 Conclusion
There are wide variations, as reported in the literature, in practice among
physicians providing clinical care to individuals with dementia despite the availability
and dissemination of well-established best practice guidelines. These results are based
on a systematic review, but available studies that met the inclusion criteria all have
significant limitations as they are cross-sectional, self-reported surveys with often low
response rates, sample sizes and of varying research contexts.
As a counter to some of these limitations, I conducted an analysis of the receipt
of dementia care at a patient level, utilizing population-based health care services
68
utilization data (administrative data) in BC, described in detail in Chapter 5. Using these
comprehensive data at a patient level allows me to a) conduct population level analysis
that is not limited by sample size, response rates and is generalizable to the entire BC
population b) examine patient characteristics that will allow me to separate out
warranted and unwarranted variation and c) potential identify demographic
characteristics associated with variation that are modifiable.
Before moving to that study, the next chapter will first outline the analytical
methods used in this thesis, followed by the results of my analysis.
69
CHAPTER 4 – METHODOLOGY
4.1 The Health Services Research Lens
Populations grow and age and prices rise, which means healthcare services will
continue to need improvement in quality and efficiency to help control expenditure
growth. The magnitude and complexity of any task related to this within healthcare
services is daunting. Healthcare delivery encompasses several occupational groups that
work with and sometimes compete with each other; standardized processes are difficult
because of the uniqueness of each patient and the different types of coordination
required for high quality care Unlike other industries, doctors have considerable
autonomy in their decision-making and direct most of the resources used; and the
complexity of healthcare services is constantly evolving
198
. There are other pressures
as well, including several stakeholders’ influences such as government attempts at
financial control, local opinion, healthcare staff organizations, and the medical product
industry with its commercial stake
198
. These are the issues that the field of Health
Service Research (HSR) aims to address.
Health services research is a term used to encompass a wide variety of analytical
(qualitative and quantitative) methods drawing on several disciplines including
epidemiology, economics, geography, political science, statistics and sociology, while
also drawing on the clinical and biological sciences. HSR’s overarching goal is to
“provide unbiased, scientific evidence to influence health services policy at all levels so
as to improve the health of the public”
198
. Unlike a clinical lens, HSR often adopts a
population perspective to help answer more universal questions that support the
healthcare decision-making process. Health services research uses a variety of sources
70
of information, including surveys, interviews, focus groups, randomized trials, and
administrative data (collected primarily for other, usually administrative, purposes).
Administrative datasets within this sphere were first utilized in the early 1970’s
199
, but
are now a commonly accepted source of research data
200
.
4.1.1 Use of Administrative Databases
Administrative databases used for research purposes can be broadly defined as
containing pre-existing data that have been collected for different administrative or
operational purposes that are usually not research driven. Unlike primary data that are
collected specifically for research uses, the use of these data for research purposes is a
“secondary use”. There are a number of operational purposes underlying the creation
of these data sources, including registration or enrolment (e.g. for insurance coverage
purposes), payment (for the services of a health care professional, or for a prescription
drug or other health care product), or clinical care. The key advantage of using
administrative sources of data for research purposes is that they can provide already
available population-level information, thereby circumventing the cost and practical
barriers to creating purpose-built population-level data resources, and addressing
limitations inherently associated with studies using other sources of data such as
sampling bias and limited sample sizes
201
.
The majority of administrative databases used in health services research are
related to the delivery of services, so while covering an entire population, they can also
provide a view of people as they move through various components of a healthcare
system. This is particularly true in Canada, where universal coverage for some key
71
components of health care services creates population-based data resources as a by-
product of system operations.
Administrative data resources are not without challenges when it comes to
research applications. Availability, inclusions, frequency and quality can vary, since
these will all be affected by the primary purposes for which the data were assembled.
Therefore it is necessary to understand the details associated with how the data were
collected in order to be able to use them effectively for research purposes. For the
purposes of HSR, data from different sources can often be ‘linked’ in order to more fully
explore a research question, while preserving the integrity and privacy of individuals
and their information.
4.2 Administrative Data Sources
Individual patient-level, longitudinal data on BC’s four million residents is
available through Population Data BC (referred to as PopData BC henceforth). Data on
individuals and on caregiver-patient transactions are collected by the BC Ministry of
Health and used to create comprehensive, population-based health care data sets
suitable for research uses. The data provided to researchers contain unique study-
specific codes to prevent personal identification of either patients or physicians, while
still allowing linkages within and across files, to (approved) external files, and over time
202
. These data files include encounter, enrolment, clinical data and registries.
For this thesis, data were drawn from five internal databases available through
PopData BC and two external databases (all described below) which were linked to the
internal databases by programmers at PopData BC: 1) the Consolidation File; 2) the
72
Medical Services Plan (MSP) File 3); the Home and Community Care (HCC) File; 4) the
Hospital Separation File; 5) the Vital Statistics Deaths File; 6) the PharmaNet File; and 7)
the Practitioner File from the College of Physicians and Surgeons of BC. Data extracted
from these linked databases were used to develop the key variables and measures
required for my analysis as outlined in Table 4.1.
Table 4.1 - Key variables and data sources for analysis
Key Variables Data Source
Patient demographics and geography
Consolidation File
Dementia diagnosis, diagnosis of co-morbidities,
health status
MSP File, Hospital Separation File
Clinical service use (dementia care services,
longitudinal primary care services)
MSP File, Hospital Separation File
Home care and/or community service use
patterns
HCC File
Prescription patterns (acetyl-cholinesterase
inhibitor and/or antipsychotic)
PharmaNet
Death and cause of death
Vital Statistics Death File
Length of stay in the community prior to moving
to LTC
MSP File, HCC File
Hospitalizations
Hospital Separation File
Transitions Hospital Separation File, HCC File
Physician demographics and geography Practitioner File
In order to access these databases and publish results using de-identified data
extracted from them, Data Access Requests were filed through PopData BC, and the
proposed project was reviewed and approved by all relevant Data Stewarts. This data
access request process includes providing proof of ethics review, which was provided by
the University of British Columbia Office of Research Services Behavioural Research
73
Ethics Board, and proof of peer review, which was provided by the School of Population
and Public Health Thesis Screening Panel and the student’s supervisory committee.
4.2.1 The Consolidation File
The Consolidation file
203
is the central demographics file maintained by PopData
BC. It is a registry of all residents who meet eligibility criteria for BC health care
insurance. It has been augmented by PopData BC to include cases where individuals
receive services even in the absence of a valid registration. This file includes
demographic information for each individual regardless of whether the individual uses
healthcare services. The file is considered as complete a roster of individuals in BC as
possible, though there are known limitations including incomplete coverage of First
Nations populations and federal employees. For this thesis, demographic variables
used included year of birth, sex, and geographic location (health authority and health
service delivery area) and neighbourhood income quintiles. Neighbourhood income
quintile is a measure of the adjusted income per person equivalent, derived from postal
codes of patient’s residence and was used to develop a socioeconomic status (SES)
indicator variable since SES plays a key role in health service use
204
.
The consolidation file also has data on what are termed ‘economic families’,
which provides an indication of household composition and summarized by the ‘Shared
MSP id variable. In BC, premiums are levied for MSP coverage. The amount of the
premium is based on family size and income. The premium may be paid by employers
(as a benefit of employment), through public subsidy as in the case of people on
income assistance, or by individuals themselves. Lower income families are eligible for
74
either partial or full subsidies, with the level of subsidy determined by the previous
year’s family income.
The administrative implication of this premium system is that an economic family
is given one MSP contract number while each unique individual receives a Personal
Health Number. The combination makes it possible to determine the nature of the
economic family to which each person belongs
205
. For the purposes of this thesis, if a
person diagnosed with dementia belonged to an economic family of more than one
person, this was used as a proxy indicator of the existence of a potential informal
caregiver. This process is incomplete, in that it cannot be used to identify individuals
living with adult children who would have a separate MSP contract number. That is,
while older adults and their adult children may share the same physical location, they
do not constitute an economic family for the purposes of defining MSP contract
holders. In short, there is no way (using these data) to determine if a person has an
informal (family or non-family) caregiver who does not live with them or who lives with
them but has a different MSP contract.
4.2.2 The Medical Services Plan File
The MSP file
206
contains payment information for all care provided by physicians to BC
residents where the physicians are paid by the province on a fee-for-service basis. The
data contained therein describe services used, and each service record includes a
patient diagnostic code (using ICD version 9), which has been validated for research
uses of this genre)
207
. The service provision descriptors are fee codes -- five-digit codes
which indicate the insured service for which the practitioner was paid. These fee codes
75
can be aggregated into service code groupings. Service codes are two-digit codes that
indicate the type of service rendered by a practitioner, such as an office visit or a home
visit. Fee item codes can be retired, new fee items can be added, and the amount paid
associated with a fee item can change over time. Specific fee items used in this thesis
were cross-checked for their use and definition over the entire study period. Each
record in the MSP fee-for-service payment file also includes the date of each visit, the
total amount paid, a unique, study-specific physician identification number and the
physician specialty code. Specific fee item codes and the dates of visits were used to
create individual-level measures to assess if guideline-consistent dementia care and
longitudinal primary care had been provided.
It should be noted that MSP data do not include information on the use of
services provided by physicians paid by non-fee-for-service methods, that is, physicians
reimbursed through alternative payment arrangements (e.g. paid by salary or for a
specified block of time). These alternative payment arrangements have traditionally
represented less than 10 percent of total payments to physicians, but their relative
importance has been rising in recent years
208
, and varies by physician specialty and by
region (alternative payment arrangements are more common in rural and remote areas
of the province and among specialities such as psychiatry). This information gap means
that the analyses reported here are inevitably based on under-counts of incident
dementia and use of services for dementia by individuals who receive services from
these alternative payment providers.
76
4.2.3 The Home and Community Care File
Information on eligibility assessments and use of publicly-funded home care, home
support, assisted living, LTC services and adult daycare services is contained in the
Home and Community Care (HCC) file
72,209
. All residents who receive an assessment,
regardless of whether they are approved for care, are included in the file. The data for
the HCC file are derived from three other files: 1) the LTC advice file 2) the direct care
advice file and 3) the home support claims file. Each of these files contains detailed
information on the type of service (for example, LTC service, home nursing,
occupational therapy, physical therapy, community rehabilitation therapy, adult day
programs, and home support), and amount of service and subsequent transactions
which are tracked as service events for the period of time that the client receives care. A
client who receives several services throughout the course of care will have multiple
records indicating the different services and intensity of care received. Referrals for
assessments for HCC do not need to come from a physician, but can also come from
hospitals, social workers or the families (or patients) themselves, though the source of
the referral is not included in the HCC file. The HCC file also notably only has
information on publicly-funded care and therefore does not capture any service
information on privately-funded care or care provided by voluntary agencies.
Individuals may choose to purchase care privately because their assessment deems
them ineligible for public care, as a supplement to publicly funded care, or as a
substitute for that care.
77
4.2.4 The Hospital Separations File
The Hospital Separation file, also sometimes referred to as the Discharge Abstract
Database, records all inpatient and day surgery separations (discharges or deaths) from
acute care hospitals
210
. The file includes information on BC residents hospitalized in
other jurisdictions. The data in the file are extremely detailed and include information
on the date of admission and separation, the level of care received (differentiates day
surgery, acute, and rehabilitation hospitalization), diagnoses, procedures, interventions
and the most responsible physician during each stay. Visits to emergency rooms that
result in an inpatient admission are noted; however, other visits to and services received
while in the emergency room are not included. Prior to the 2001/2002 fiscal year,
hospitals in BC used ICD9 codes. Therefore both ICD9 and ICD10 codes are used in this
thesis.
4.2.5 The Vital Statistics Deaths File
The Vital Statistics Deaths file
211
is a record of all deaths in BC, including month, year of
death, place of death and underlying cause. It excludes all deaths of BC residents
outside of the province (death of non-BC residents that occur in BC are recorded in this
file but for the purposes of my thesis can be excluded based on previous healthcare
utilization). These records were used to determine if a person in the cohort died prior
to the end of the study period, in which case the records of that individual would be
right censored in any time-to-event analysis. The cause of death, while provided, was
not used for validation of death due to dementia as most people with dementia die due
to other causes that are exacerbated by dementia. Dementia is infrequently recorded as
78
a cause of death, though this practice is starting to change
212
. In any case, the cause of
death was not needed for the purposes of identifying incident cases of dementia.
4.2.6 PharmaNet
PharmaNet
213
is an online, real-time data system external to PopDataBC which records
all prescriptions and medical supplies dispensed in BC
214
. It includes information on
prescription drugs dispensed by community pharmacies in BC, through LTC settings,
emergency departments and hospital outpatient pharmacies dispensing medications for
patients to use at home. Exclusions are over-the-counter medications, cancer
medications dispensed by cancer agencies/centers, antiretroviral medications dispensed
by HIV centers, and medications dispensed while in hospital. Since it is housed at the
BC Ministry of Health, it is considered an external database from PopData BC’s
repository. The PharmaNet database includes information on all dispensed
prescriptions regardless of who pays. PharmaNet records contain detailed information
on patient demographics, the prescribing physician (including practitioner type and
speciality), the drug information number, its generic name, drug strength, its
therapeutic class, the date and quantity of medication dispensed, the number of days of
supply and a flag for special authority drugs which is pertinent for AChIs covered under
the Alzheimer’s Disease Therapy Initiative program mentioned previously.
Unfortunately the database does not contain any information on whether drugs
dispensed are actually ingested (i.e. there is no way to adjust for non-compliance).
The drugs of interest for this thesis, namely any AChIs, the NMDA-type glutamate
receptor inhibitor and antipsychotics, were identified through the Anatomical
79
Therapeutic Chemical (ATC) classification system which classifies drugs based on the
organ they act on
215
and their chemical subgroup (the 5
th
level of the classification).
This allowed identification of broad classes of medications used to treat dementia.
Pharmanet identifies drugs via drug identification numbers (DINs) which are 8-digit
numbers assigned by a computer system to each drug that has been approved for use
in Canada
216
. Therefore, I developed a cross-walk to map DINs into appropriate ATC
groups for the medications of interest (Appendix Table B.1).
4.2.7 The College of Physicians & Surgeons of BC File
The College of Physicians & Surgeons of BC file is the second external data file used for
this thesis. It contains information collected by the registering and licensing body for
physicians in BC; all registered and practicing physicians in the province of BC are
included. The records from the Registry file include information on physician age, sex,
year of graduation, place of training (Canada, US, Europe, Africa, Asia), specialty and
membership status code (indicating whether the physician is currently practicing).
Physician characteristics were used to control for practice style and its possible
contributions to the trajectories of care experienced by patients.
4.3 Study Design
This thesis was divided into four major research objectives:
1) Conduct a critical assessment examining the existing research literature on actual
physician practice patterns associated with the care of people with dementia and to
what extent those practice patterns are consistent with published guidelines
80
2) Examine population-based variations in receipt of guideline-consistent dementia care
and patient factors that are associated with that care
3) Examine transitions experienced longitudinally to establish patterns that can help
identify points of care when transitions are highest and the factors that contribute to
those transitions
4) Assess the association between receipt of guideline-consistent dementia care and/or
high quality primary care and the number of transitions experienced by individuals with
dementia
The broad approach uses administrative data with multivariate modelling. The model
selection for each objective is described in more detail in Section 4.4.3.
4.3.1 Identification of Study Cohort
For research objective 2, I used a combination of registry and utilization data
(described in detail in section 4.2 above). A cohort of seniors 69+ year olds with an
incident diagnosis of dementia in 2009/10 (described as cohort 2 in Figure 4.1) was
identified. Incidence was identified using a ten-year wash-in period (dating back to
1999/2000) during which no other diagnoses of dementia should have occurred
(described in more detail below). Since guidelines for dementia care focus primarily on
the measures required for a diagnosis, this study design allowed me to use the
incidence of dementia diagnosis to define the cohort and the period during which
guideline-consistent care is expected.
81
Figure 4.1 – Historical cohort study design
For research objectives 3 and 4, a longitudinal historical cohort methodology
was developed using registry and administrative data. A cohort of individuals 65+ years
old with an incident diagnosis of dementia in 2001/02 was identified and followed
forward (to 2010/11) and backward (one year pre-diagnosis, to 2000/01). This is shown
as cohort 1 in Figure 4.1. In this case incidence was identified using a two-year wash-in
period (dating back to 1998/99) during which no other diagnoses of dementia should
have occurred (described in more detail below).
Developing decision rules for defining the cohort was of critical importance to this
work. In any given year, there are approximately 500,000 seniors living in BC. Of these,
~15% can be expected to develop dementia
217
. Onset will occur at different times, in
different ways, and will be detected at different points in a life trajectory. The inclusion
criteria used here for the two (2001/02 and 2009/10 incident) cohorts were as follows:
All patients over the age of 65 in the year 2001/02 and over the age of 69 in the
year 2009/10
Registered as BC residents for the entire study period
Newly diagnosed with dementia in 2009/10 (objective 2) or 2001/02 (objectives 3
& 4)
Cohort 1 entry
Objective 2
Cohort 2 entry
Objective 3 & 4
00/01 01/02 10/11 09/10
82
The process of identifying incident cases is always dependent on available
information sources. In this case I interpreted “newly diagnosed with dementia” to
mean that record including a diagnosis associated with dementia as classified through
the International Classification of Diseases (ICD) could be found in the Medical Service
Plan data (ICD version 9 code) and the Hospital Separation data (ICD version 9 or 10
code). The specific codes are outlined in Tables 2.1 and 2.2. ICD-9 diagnosis codes for
senile dementia, vascular dementia, Alzheimer’s Disease and senility (290.xx, 294.9x,
294.1x, 331.xx excluding 331.83 for Mild Cognitive Impairment, 797.xx), and/or ICD-10
diagnosis codes (F00, F01, F02, F03, G30)
10,62,144
were used. Incident diagnosis was
defined in a way consistent with prior literature, as patients who had a minimum of two
physician diagnoses and/or one hospital diagnosis associated with dementia over a
two-year period
218
. This diagnosis requirement was used to increase specificity of
identifying confirmed cases of dementia while reducing rule-out diagnoses.
219
.
Administrative data have been demonstrated to be quite valid for the purposes of
identifying individuals with chronic conditions, including dementia
220
.
Often, cases of dementia can remain undetected for years. For the purposes of this
thesis, it was important to examine incident cases as identified in the health care system
because the focus is on appropriate dementia management, which is most significant
during the first year after diagnosis
221
and also because this study aims to describe the
longitudinal primary care services experience of dementia patients, which begins at the
point of diagnosis. A wash-in period of two years (for cohort 1) or eleven years (for
cohort 2) was used in order to capture true incident cases of dementia. In other words,
individuals had to meet the case definition as described and have no dementia
diagnoses in their medical or hospital records in the prior two or eleven years in order
83
to be included in the final cohorts. Cohort 1 was followed for a maximum of ten years
from the point of entry (or until death) and cohort 2 was followed for one year from the
point of entry.
4.4 Construction of Longitudinal Files
In any given year, each resident of B.C. can generate anywhere from zero to many
health care transaction records, which can appear in multiple different data files. The
data used here are taken from the seven files described above, covering a twelve year
period (from 1998/99 to 2010/11) for the entire population of BC aged 65+ in 2004/05.
The first step was to create an analytic record for each patient for each year of the study
period. For cohort 1, these records were from 1998/99 until 2010/11 or death
(whichever came first) including the two-year wash-in period. For cohort 2, the records
were from 1999/2000 to 2010/11, including the eleven-year wash-in period. A unique
study-specific patient id was used to link patient information across data files and years.
Annual fiscal year data files were produced containing comprehensive analytic
3
and
service use variables (described in Section 4.4.1 below) for each person. These annual
files were then merged to create two data files. The first data file contained information
on cohort 1, individuals who received an incident diagnosis of dementia in 2009/10, and
was used to assess receipt of guideline-consistent dementia care in the year after
diagnosis, addressing research objective 2. The second file was a longitudinal data file
for cohort 2, created to trace individuals who entered the cohort (i.e. received an
3
Analytic variables in this case are the independent variables used to describe patient, physician and system
characteristics
84
incident diagnosis of dementia) in 2001/02 and assess their healthcare transitions
backward and forward over time to address research objectives 3 and 4.
4.4.1 Analytic and Service Variables
Table 4.2 below lists the annual analytic, service and outcome variables defined
for each individual in the cohort, classified by the characteristics identified and
contextualized in the conceptual framework used to guide the study in Chapter 2,
Section 2.7.
Table 4.2 – Detail of variables measuring characteristics in conceptual framework
The covariates at the individual-level were divided into five categories: Individual
(or patient) descriptor variables, physician descriptor variables, system variables,
dementia care service use variables and longitudinal primary care service use variables
(Table 4.2). Each of these variables is described in further detail in Table 4.3 including a
Characteristic Description
Individual characteristics Age, sex, income category (SES), health status (number of
major ADGs or comorbidity), responsive behaviour
symptoms, caregiver status proxy (objective 3 & 4)
Physician characteristics Age, sex, speciality, practice years, # of patients 65+, # of
patients with dementia
System characteristics Health authority of residence or practice
Dementia Care Imaging, laboratory work, pharmaceutical prescriptions, in-
office examination, counselling, specialist referral (all as
defined in Chapter 4), home care assessment
Primary Care First contact (hospital or LTC visits from physician providing
plurality of care), continuity (proportion of all visits with
physician providing plurality of care), coordination & referral
(referrals for imaging & lab tests originating from physician
providing plurality of care)
85
definition of how they were constructed (if that is the case). Each subsequent Chapter
also describes their use in the respective Method sections. All variables were measured
annually for the duration of the study period (from 1998/99 to 2010/11) and for both
cohorts (except the Primary Care variables which were only measured for Cohort 1).
Individual descriptor variables: These are variables related to care use and outcomes
of care. Of these descriptor variables, income category, health status, behavioural
symptoms and the informal caregiver status proxy variable require further explanation.
Income category is provided in the consolidated file and is derived from the
neighbourhood income quintile data. It is used as a proxy indicator of SES, which plays
a key role in health service use
204
. Health status is measured in two ways, by the
number of multiple chronic conditions each individual has (used for objective 2) or by
the number of major aggregated diagnostic groups (ADGs) (used for objective 3 and 4)
and described in detail in Section 4.4.1.1 below. The behavioural symptoms variable is
derived from two variables indicating receipt of prescriptions of antipsychotics or
benzodiazepines. Individuals with dementia who also have behavioural symptoms often
have very different use of the healthcare system which needs to be accounted for.
Finally, the informal caregiver status proxy variable was used for objective 3 and 4 and
was derived from the shared MSP id variable in the Consolidated file. The shared MSP
id indicates all family members currently enrolled in the MSP who belong to an
‘economic family’ and share a household (Described in detail in Section 4.2.1). If a
person with dementia belongs to an economic family, my assumption is that any family
member(s) living in their household will likely function as a caregiver. Sensitivity
analysis measuring this against the caregiver status and marital status variables (more
86
definite measures of having a caregiver) from the HCC data indicated good
concordance. The former is used in analyses because the latter are available only for
the subset of the population who receive home and community care services.
Physician variables: These were previously described in Chapter 2, Section 2.7, and
identified as variables describing physician practice patterns controlled for in Objectives
3 and 4.
System variables: The health authority variable identifies the geographic location in
which the patient resides and most likely receives care. Based on the health authority it
is possible to determine further descriptive characteristics of that system. There are five
geographic health authorities in BC. Three of these, Fraser, Vancouver Coastal and
Vancouver Island, contain major urban centres and a higher supply of specialists/family
physicians. The Northern and Interior health authorities are considered more rural
health authorities because they lack major urban centres and have a lower supply of
specialist/family physicians. The Fraser health authority has the largest population of
the five, but is geographically the smallest. It is also the most urbanized. The
Vancouver coastal health authority includes three major population areas, an urban core
(Vancouver, Richmond and the North Shore), a moderately urban area (Howe Sound,
Powell River and the Sunshine coast) and a more sparsely populated area (Bella Coola
and Central Coast). Its population pyramid has a significant bulge of 20-40 year olds
compared to the overall population due primarily to the city of Vancouver. Vancouver
Island is less populated and is home to a high number of retirees, reflected in its
population pyramid, who reside in the urban areas of Greater Victoria and Nanaimo.
The Interior health authority encompasses almost a quarter of BC’s land mass with a
87
much lower population density and with a small urban centre. Its population includes a
larger proportion of younger children and older adults compared to BC overall. Lastly,
the Northern is the largest geographic area but is the smallest health authority by
population size. It is the least urbanized and has a smaller proportion of older adults
and larger proportion of younger children.
Dementia Care variables: The dementia care service use variables were developed
based on best practice guidelines for dementia care published by the BC Ministry of
Health and the National Canadian Consensus Conference
7,122
and described in Chapter
4. These were categorized into six categories: recommended laboratory testing (blood
work), imaging, physical examination, prescriptions, counselling and referral. I was
unable to measure provision of memory tests, non-pharmacological interventions or
community service referrals as there was no information on these in the files to which I
had access. The imaging category only includes head CT as information on MRI was not
available in data files to which I had access. The dementia care service use variables
were measured for the year of the first dementia diagnosis and the following year in
keeping with guidelines that indicate these services should be provided during the
period immediately following initial diagnosis.
Primary Care variables: The primary care service variables were measured annually for
the year of diagnosis. Three main characteristics were defined as predictors of high
quality primary care, these were: access, long-term person-focused care and
coordination of care (described in more detail in Chapter 7 and summarized in Table
7.1). First access was measured in several ways including: the number of settings care
was provided, receiving care while in hospital or LTC from the physician providing
88
plurality of care. Long-term person-focused care was measured through continuity of
care, while coordinated care was measured by the proportion of all referrals to
specialists and for laboratory testing originating from the assigned primary care
physician. Patients were assigned to the primary care physician who provided plurality
of care, defined as the physician most responsible for each patient and therefore
receiving most of the patient’s visits (a more detailed description is provided in Section
4.4.1.2 below).
4.4.1.1 Health Status
In addition to basic demographics, heath status needed to be characterized for
each patient in order to control for the effects of other chronic conditions on the
outcome variables. Health status was proxied using the Johns Hopkins Adjusted Clinical
Group Case-Mix system, a risk adjustment methodology
222
. This system uses diagnoses
that a patient accumulates from physician and hospital visits over a specific period of
time (in this case, a year), to determine a clinical complexity/severity category. Each
diagnosis is classified into one of 32 ADGs, which are defined based on clinical similarity
(severity, persistence, reoccurrence) and expected health care service utilization (e.g.
follow-up visits, specialist referrals). Each individual over the course of the year can
then have 1-32 different ADGs assigned. These aggregations are further refined by the
system to reflect the concept of major conditions, with eight ADGs identified as “major”,
including time limited major conditions, chronic unstable conditions, major adverse
injuries and malignancy
222
.
Combinations of ADGs, along with age and sex, are used to map patients into
one of about one-hundred mutually exclusive Adjusted Clinical Groups (ACGs). Each
89
ACG category is a measure of the patient’s expected need for health services over a
subsequent period (based on services actually received during the measurement
period). ACGs are mutually exclusive and exhaustive. The system has been validated for
use with BC administrative data
223
. For simplicity, the mutually exclusive ACGs are
combined by the system to measure six levels of overall morbidity (non-users to very
high users), termed resource utilization bands (RUBs). For this thesis, only one variable
derived from the John Hopkins Adjusted Clinical Group Case-Mix system was used for
measurement of health status, the number of major ADGs individuals had upon entry
into the cohort.
Within my health status variable cluster, I used a second measure of morbidity --
the number of chronic diseases. A large proportion of individuals with dementia often
have several other chronic conditions
224
which can both affect the care they receive at
the point of dementia diagnosis and longitudinally, and the outcomes of care. This
variable was derived as follows: I identified ten chronic conditions of high prevalence in
the older-adult population: diabetes mellitus, stroke, hypertension, chronic kidney
disease, liver disease, arthritis/osteoarthritis, congestive heart failure, coronary vascular
disease, chronic obstructive pulmonary disease and cancer,
64,65
as well as depression
(commonly found in dementia patients)
225
. A variable based on a simple count of co-
morbid conditions was then created at entry into the cohort.
4.4.1.2 Physician Attribution Process
In both dementia care and primary care, physician (or practice) characteristics
(Described in Table 4.2) have been shown to influence a patient’s trajectory of care.
Attributing these physician variables depends on identifying a single most responsible
90
primary care physician for each patient. In the absence of rostering in BC, this implies
using an attribution process in which individuals are “assigned” to a physician for the
purposes of analysis.
For objectives 3 and 4, the assignment of a physician was done annually based
on the physician providing plurality of care for each patient in that fiscal year. The
physician providing plurality of care was defined as the physician who received the
highest proportion of unique visits during the fiscal year. In the event of a tie, the
assignment was made to the physician who billed for the most fee items for that
patient. Exceptions to these rules were institutional visits as these were measured as
service units recording the number of days on which the practitioner saw that person.
Limitations to an assignment like this is that for some physicians, the patient ‘visit’ may
actually be only a patient management activity or form fee (such as renewing a
prescription over the phone) and did not actually involve a face-to-face contact. No
charge referrals, which occur when a patient does not come in to the office for a visit
but receives a renewal of a referral to a specialist (referrals to specialists must come
from the primary care physician and in BC expire after 6 months), were excluded prior to
counting visits.
Sensitivity analysis was conducted to determine what percentage of patients
received their dementia diagnosis from the same physician to whom they were assigned
through the plurality of care process described. I found that this occurred for 70% of
patients who had an incident physician MSP diagnosis code for dementia. Additionally,
analysis was conducted to determine the median number of patients with dementia in
the cohort treated by the same physician (to determine if clustering needs to be
91
accounted for). Each physician cares for a small number of patients (the median is 2),
eliminating concerns of clustering of patients within providers.
92
Table 4.3 - Key analytic and service variables
Variable Data Source Description
Patient Variables
Age Consolidated Age in years
Age Group in 2001/02 Derived Age variable categorized 65-74, 75-84, 85+ years
Sex Consolidated Female or Male
Socioeconomic status Consolidated Neighbourhood income measured through Quintile of adjusted income per
person equivalent, 1-5 indicating lowest to highest income
Death Vital Stats Year and month of death
Behavioural symptoms flag Derived Indicator flag for behavioural issues. Composite variable developed from
PharmaNet indicating antipsychotic and/or benzodiazepine prescriptions
Adjusted Clinical Group MSP Indicator of health status described in Section 4.4.1
Resource Utilization Band MSP Uses ACGs to create coarser non-overlapping groups of morbidity levels (0-non
user, 1-healthy, 2-low, 3-moderate, 4-high, 5-v high)
# Major Aggregated Diagnostic Groups MSP Diagnoses are classified into 1 of 32 Aggregated Diagnostic Groups based on
clinical similarity and expected health care service utilization. Eight of these
groups are considered ‘major’
Has diabetes MSP Described in further detail in section 4.4.1.1
Has hypertension MSP
Has congestive heart failure MSP
Has chronic kidney disease MSP
Has liver disease MSP
Has arthritis/osteoarthritis MSP
Has cancer MSP
Has depression MSP
Has cardiovascular disease MSP
Has stroke MSP
Has chronic obstructive pulmonary
disease
MSP
# Multiple Chronic Diseases Derived Count of the number of chronic diseases previously defined in section 4.4.1.1
Incidence of dementia flag Derived Indicates if incident diagnosis of dementia in that fiscal year
Dementia diagnosis location MSP Codes location that patient received dementia diagnosis, hospital or physician
office
93
Variable Data Source Description
Caregiver status proxy Derived Derived from the shared MSP id variable in the Consolidated file indicating if
the individual shares a household described in section 4.4.1
Physician/Practice Variables
Age College Physician age in years
Sex College Female or Male
Date of graduation College Year of graduation from medical school
Years of practice Derived Derived from date of graduation and membership
Place of graduation College Indicates if physician is foreign or Canada trained
Speciality College Indicates if physician has geriatric, psychiatric or neurological training
# patients 65+ MSP Indicates if physician has large number of patients 65+
# patients 65+ with dementia MSP Indicates the number of patients with dementia physician provides care to
System Variable
Health Authority Consolidated Indicator of region of residence, 1-5 based on health authority delineations
Physician Practice location College Health authority physician practices in
Dementia Care Service Variables Defined in detail in Chapters 3 and 5
Laboratory testing MSP Tests recommended by guidelines to exclude other reasons for dementia-like
symptoms
Complete hematology profile
Serum electrolytes
Serum calcium
Serum glucose
Thyroid stimulating hormone
B12 vitamin
Imaging MSP Imaging recommended by guidelines to assist with diagnosis
Head computed tomography
Prescriptions Pharmanet Identified using DINs that were converted to ATCs described in section 4.2.6
Acetylcholinesterase inhibitor
Recommended antipsychotics
All other antipsychotics
Trazodone for sleep disorders
Care Management MSP Management processes during diagnosis defined by guidelines
Complete examination in office
94
Variable Data Source Description
Individual counselling in-office
Referral
Referral to specialist re. dementia MSP Referral to a Geriatrician, Neurologist, Psychiatrist, Neuropsychologist,
Geriatric Psychiatrist
Primary Care Service Variables Defined in detail in Chapter 7
Continuity
# of GP visits MSP Number of visits to primary care physician
# of GP visits with attributed physician Derived Number of visits with the primary care physician identified as providing
plurality of care in the year
Proportion of total visits with attributed
physician
Derived Used as Continuity measure, indicates of all visits, the proportion that occurred
with the attributed physician the patient is attributed to in the year
Coordinated & Referral Care
# lab referrals MSP Number of referrals for laboratory tests
# lab referrals with attributed physician Derived Number of lab referrals originating from the primary care physician identified
as providing plurality of care in the year
# specialist referrals MSP Number of referrals to specialists
# specialist referrals with attributed
physician
Derived Number of specialist referrals originating from the primary care physician
identified as providing plurality of care in the year
Proportion of total lab referrals from
attributed physician
Derived Used as Coordination measure, indicates of all lab referrals, the proportion
that originated from attributed physician in the year
Proportion of total specialist referrals
from attributed physician
Derived Used as Coordination measure, indicates of all specialist referrals, the
proportion that originated from attributed physician in the year
Access
# days care received outside office hours MSP Indicates # days care was received outside office hours
# days care in home MSP Indicates # days care was received at home
# days care in hospital MSP, DAD Indicates # days care was received in hospital
# days care in LTC MSP, HCC Indicates # days care was received in LTC
# alternate settings of care Derived Indicates the number of settings outside of the office that care was provided
by attributed physician (includes home, hospital, LTC, Assisted Living)
Received visit from physician plurality in
hospital flag
Derived Used as Access measure, indicates if patient received visit from attributed
physician while in hospital
Received visit from physician plurality in
LTC flag
Derived Used as Access measure, indicates if patient received visit from attributed
physician while in LTC
Received service from physician plurality
outside office hours flag
Used as Access measure, indicates if patient received service from attributed
physician outside of office hours
95
Variable Data Source Description
Outcome Variables
Flag for hospitalization DAD Indicates if person had at least 1 hospitalization that year
Number of hospitalizations DAD Total number of all hospitalizations (includes acute, rehab, ER etc.)
Total hospital days accrued DAD Total number of days accrued in hospital over the year. If admittance and
discharge happen on same day, counted as 1 day.
Mean length of stay in hospital Derived Mean consecutive number of days spent in hospital that year
LTC Facility flag HCC Indicator if person resides in a LTC facility at beginning of year
New LTC Facility flag HCC Indicates if person moved into LTC facility during the year
Second move to LTC facility HCC Indicates if person moved to a second LTC facility
Length of stay in community Derived Tabulates the number of days an individual remains in the community prior to
institutionalization. Derived from HCC data
# of transitions annually Derived Tabulates the number of transitions between home, hospital, respite care,
assisted living and LTC. From MSP and Discharge abstract data
Total number of transitions Derived Used as outcome variable. Tabulate the total number of transitions over entire
study period
# of months alive annually Derived Used as outcome variable. Tabulates the total number of months the person is
alive based on Vital Stats records
# of transitions per year alive Derived Used as outcome variable. Average of the transitions experienced per year
alive
96
4.4.2 Outcome Variables
For this thesis the primary outcome variable, the number of transitions
experienced, was defined as a physical move from one location to another with a stay of
at least one night at the destination location, based on the recorded dates of admission
and discharge
10,12
. The identification of a “transition” was verified based on a move
that accounted for the departure and return locations. Moves that resulted in a return
location different from the original departure location would count as two transitions,
so long as the “at least one overnight stay” criterion was met for each of the two
apparent destination locations. For example, an individual who lives at home, is
admitted to the hospital with an overnight stay and returns home will have had one
transition; an individual who is admitted to the hospital from home, stays at least one
night in the hospital, and then is discharged to a LTC facility will have had two
transitions. An exception to this rule is for individuals who experience a move from
their home to a LTC facility or Assisted Living and then back home. In these cases, it is
rare that an individual who needs the level of assistance that a LTC facility provides
would be able to live at home alone again. It is more likely that they have moved back
to the community with an adult child or other caregiver, or into a private facility for
which no data were available. This type of move was counted as two transitions
because daily living circumstances changed twice.
The number of transitions for each patient in the data sets were assessed for
each year and summed over the trajectory of the patient’s care during the study period
as demonstrated in Figure 4.2.
97
Figure 4.2 – Example of transitions that can be experienced over study period
Hospitalizations of any kind including day surgeries with discharges that occur
the following day, respite care and moves to and between LTC facilities or Assisted
Living facilities were included. However, in-patient moves within the same ward or
institution were excluded
10
. The data cover all public hospitals, residential care facilities
and assisted living facilities.
4.4.2.1 Long-Term Care Facility Transition
Approximately 25,000 seniors 65 years and older in BC were referred and initially
assessed through the HCC system each year during the study period (Figure 4.3). Home
and community care services provide additional support, allowing individuals to remain
in the community longer, but also conduct assessments for those who cannot remain in
the community to determine a needed level of care including possible moves to an
Assisted Living or residential LTC facility. An initial assessment can result in a number of
different outcomes. For example, some individuals may be found ineligible for publicly-
funded support, in which case they may choose to ‘stay put’, or move into a private
facility, data from which are not captured in the data sets available for my analyses.
98
Those who were considered eligible for public support after an initial assessment could
go on to receive home-care services, could be deemed eligible for LTC, and from 2004
onward, could be considered eligible for Assisted Living (2004 was when public funding
for Assisted Living began in BC)
226
.
While Figure 4.3 offers a snapshot in time of this assessment process, individuals
could have been assessed at any time during the study follow-up period; some had
multiple assessments and some had already been assessed prior to the start of the
study. For the purposes of this thesis, understanding the assessment process for HCC
services is necessary in the context of measuring individuals’ first transition into a LTC
facility. I created two variables, one indicating if a person was already living in a LTC
facility at the point of entry into the cohort, and the other indicating the year of a
person’s first transition into a LTC facility. Measurements regarding transitions into a
LTC facility could only be made for the subpopulation that received an assessment from
HCC services and moved into a publicly-funded LTC facility of Assisted Living.
Figure 4.3 – Pathways to publicly funded home and community care support
99
4.4.3 Analytic Models
The analytic methods used for this thesis varied for each objective and are
therefore described in detail within the Method section of each chapter though brief
descriptions of model selection are provided here. The analytic models described
below detail the dependent and independent variables that will be used in the analytic
methods to answer each research question.
Research question 2b: Is there variation in receipt of guideline-consistent care, and if
so, what are the patient characteristics most strongly associated with this variation?
A multivariate logistic (or logit) regression model is utilized when the response variable
of interest (y) is dichotomous rather than continuous. This allows me to estimate the
probability (p) associated with my response variable (y) for multiple discrete explanatory
(independent) variables. In this case, each dementia care process is a dichotomous
response variable (received the process of care, yes or no). I estimate the probability
that any patient/individual characteristic affects the likelihood that individuals will
receive each particular dementia care process. Modelling this probability (p) using a
logistic function, means the log odds of the response variable (y) (the
logarithm of the odds of success or in this case, receiving the care process) is modeled
as a linear combination of all the following discrete patient characteristics:
ln(dementia care process) = β0 + β(age) +β(sex) +β(income) +β(health status) +
β(health authority)
100
Research question 3c & 3d: What are the types of transitions experience and what
factors influence these transitions?
What factors influence the number of transitions during the peak points of care and
overall?
To understand what patient characteristics most influence the number of transitions
experienced (other than death), over and above physician practice patterns, both system
and physician characteristics were controlled for in a negative binomial model, Poisson
model or zero-inflated Poisson model. All models are part of the family of Generalized
Linear Models and are used for analyzing count data. The choice of particular model
and link function was based on the distribution of the particular outcome of interest.
While both Poisson and negative binomial regression control for the “rare event” nature
of a transition, the negative binomial model’s assumptions better fit the transitions data.
Negative binomial regression does not assume an equal mean and variance, therefore
correcting for over-dispersion (when the variance is greater than the mean) in my data
through an extra parameter. Regardless, a preliminary Poisson model was run and the
Pearson Chi square Goodness-of-fit test assessed for every outcome (the assumptions
being: the probability of an event is proportional to the time of observation, no two
events can occur simultaneously, the event rate is constant through time and that the
events occur independently). Time was offset to account for the number of months of
follow-up in the study.
Of the four outcome variables assessed, two were modelled using negative binomial
analysis: 1) transitions in the year of diagnosis; and 2) total transitions experienced over
the study period. The count of LTC moves in the year of diagnosis was modelled using
101
Poisson regression. Lastly, the count of hospitalizations in the year of diagnosis was
modelled using a zero-inflated Poisson analysis due to an excess of zero counts. The
zero-inflated Poisson model operates in two parts, modelling the count of
hospitalizations using Poisson regression and the excess zeros using a logit model. The
log of the response variable (y) (the logarithm of the odds of success or in this case,
having a transition) is modeled as a linear combination of all the following discrete
patient characteristics, controlling for death and physician characteristics:
Research question 4a & 4b: Is receipt of guideline-consistent dementia care in the
year of diagnosis associated with the number of transitions the patient experience in
that year?
Is the receipt of high quality primary care associated with the number of transitions the
patient experiences in the year in which dementia is first diagnosed?
First, the odds of receiving guideline-consistent dementia care based on the location of
diagnosis (community vs. hospital) was assessed to determine whether it should be an
explanatory variable that should be included in the larger model examining the rate of
transitions. A multivariate logistic (or logit) regression model was used with the
response variables (y), which was each dementia care process (as previously described
in research question 2b), and an explanatory variable indicating diagnosis location
(while controlling for patient characteristics).
Rate of Transition= β0 + β(age) +β(sex) +β(income) +β(health status) + β(caregiver) +
β(responsive behaviour) + β(health authority) + β(end-of-life) + β(GP yrs of practice)
+ β(# patients 65+) + β(# dementia patients)
ln(dementia care process)= β0 + β(age) +β(sex) +β(income) +β(health status) +
β(health authority) + β(dx location)
102
The same model selection process conducted for research question 3 was applied to
research question 4a and 4b. The response variable (y) of the number of transitions
experienced in the year of diagnosis was modelled using negative binomial regression
in two separate models with explanatory variables, dementia care and primary care
(after controlling for patient, physician and system characteristics).
Rate of Transition= β0 + β(age) +β(sex) +β(income) +β(health status) + β(health
authority) + β(caregiver) + β(dx location) + β(GP yrs of practice) + β(# patients 65+)
+ β(# dementia patients) + β (primary care)
Rate of Transition= β0 + β(age) +β(sex) +β(income) +β(health status) + β(health
authority) + β(caregiver) + β(dx location) + β(GP yrs of practice) + β(# patients 65+)
+ β(# dementia patients) + β (dementia care)
103
CHAPTER 5 – CARING FOR DEMENTIA: A POPULATION-BASED
STUDY EXAMINING VARIATIONS IN GUIDELINE-CONSISTENT
CARE
4
5.1 Introduction
The systematic review described in Chapter 3 identified wide variation in the
application of guideline-recommended practice among physicians providing clinical
care to individuals with dementia
31
. This physician practice variation is indicative of
inconsistent care received by patients with dementia and may result in differences in
care outcomes.
As highlighted in the systematic review, available studies examining this question
have significant study limitations. All were chart extractions or self-reported surveys of
physicians or patients with low response rates and/or small sample sizes
31,146,153,155,195
.
Enlisting administrative data for this question offers one way to avoid the sample,
geographic, and response rate limitations of previous studies. By conducting this
analysis at a patient level, I will be able to examine potentially modifiable patient
characteristics associated with any variation and be able to separate out warranted and
unwarranted variation.
The administrative data described in Chapter 4 provide a unique opportunity to
compare patterns of care received with those recommended by dementia care
guidelines. My research questions therefore are as follows:
4
A version of chapter 5 has been accepted. Sivananthan SN, Lavergne KM, McGrail KM. Caring for dementia: A
population-based study examining variations in guideline-consistent care. Alzheimer’s & Dementia 2015.
104
Research Question 2a: What proportions of individuals newly diagnosed with
dementia in BC receive each of the guideline-consistent dementia care processes?
Research Question 2b: Is there variation in receipt of guideline consistent care, and if
so, what are the patient characteristics most strongly associated with this variation?
I hypothesize that a high proportion of individuals will receive guideline-consistent
laboratory and imaging processes of dementia care as these are more regularly
administrated, however a lower proportion of individuals will receive all other processes
of care as these are more discretionary. Secondly, I hypothesize that there will be
variation in receipt of guideline-consistent care by location and health status, however
variation by other patient characteristics will indicate potential inequity and lack of
quality of care.
5.2 Methods
I used the health care administrative data described in Chapter 4 to conduct a
population-based study in BC. Detailed analysis was conducted on the study
population for fiscal year 2009/10, though data dating back to 2000/01 was utilized for
the purpose of identifying incidence of dementia, described in more detail below.
5.2.1 Data Sources
Five of the seven databases described in Chapter 4 were used for this analysis: 1)
the central consolidation file 2) the medical services plan (MSP) payment file 3) Hospital
abstract data 4) Vital statistics death file and 5) the PharmaNet file.
105
5.2.2 Study Population
Cohort 2, described in Chapter 4 section 4.3.1 was used for this component of
the project. Briefly, these were all community-dwelling seniors in BC, 69 years and older
at the time of diagnosis, who were incident cases of dementia in 2009/10 or 2010/11. A
case was considered ‘incident’ if there were no dementia diagnoses during a 10-year
wash-in period. Of note, the guidelines themselves, while differentiating the types of
dementias in their pathophysiology, do not make distinctions in the diagnostic and
subsequent care processes they recommend therefore combining the various types of
dementias would not greatly impact my measure of guideline concordance.
Individuals were excluded if they were not registered for MSP coverage for the
whole of 2009/10, or if they lived in a LTC facility or died at any point during the one-
year follow-up period. The study focused on community-dwelling seniors, as those who
live in a LTC facility are known to have different patterns of primary care service use
113,114
. A full year of follow-up was used to provide adequate opportunity to assess
receipt of guideline-consistent care. Sensitivity analysis demonstrates that patients
who died during the year had very different service use patterns with generally lower
rates of receipt of most care processes (Appendix Table C.1, C.2A, B).
5.2.3 Recommended Dementia Care Guidelines
I used the most updated BC Dementia Care Guidelines, released in July 2007, as
they reflect province-specific resource availability, were readily available to physicians
practicing in BC, and incorporate recommendations from the Canadian Consensus
group national guidelines
7,122
. Dementia care processes that were previously
106
developed
31
and described in Chapter 3 section 3.2.2, were used to characterize
different types of care received by patients with dementia: laboratory blood work,
imaging, pharmacological interventions, counselling and specialist referrals. The care
processes can be divided into those that are recommended for all newly diagnosed
individuals (laboratory tests, counselling), those that are recommended in some
circumstances (imaging, pharmaceuticals, physical exam, specialist referrals) and those
that are contra-indicated (benzodiazepines). Therefore, I did not expect 100%
adherence as circumstances may vary for patients. My main interest, instead, was in
identifying any systematic variation in these care processes by patient characteristics.
The data used for this study did not include information that would have permitted
measurement of the provision of memory testing, provision of magnetic resonance
imaging, or referral to community services. Service use was measured during the year
of diagnosis in order to capture service provided during the lead up to diagnoses
captured and in keeping with guidelines that indicate these services should be provided
both as part of and after a confirmatory diagnosis of dementia
122
. The definition of
each care process is provided below.
Diagnostic Services:
Laboratory tests – These are tests routinely ordered to determine the underlying cause
of dementia and/or for rule-out of other causes (e.g. medication-induced dementia-like
symptoms). The tests include a complete blood count, serum electrolytes, serum
calcium, serum glucose, TSH and vitamin B12 levels.
Imaging – Not routinely indicated but a CT scan of the brain is recommended under
certain circumstances such as if the patient is under 60 years of age, has abrupt onset or
107
rapid progression, and has a history of cancer or if there is a history of urinary
incontinence or gait disorder.
Physical Examination – A physician exam should be conducted to identify neurological
abnormalities and other medical conditions that could be contributing to the cognitive
decline
Prescriptions:
Pharmaceutical Intervention – All prescriptions related to the treatment of dementia
were examined (Donepazil, Rivastigmine, Galantamine and Memantine). Prescriptions of
antipsychotics for behavioural and psychological symptoms are recommended only
when alternate therapies are inadequate, symptoms are severe, or there is a risk of harm
to the patient and others.
Benzodiazepines are the only class of drugs that the BC guidelines indicate should not
be used due to high potential for adverse events. Given this and the fact that I was
interested in assessing if changes in antipsychotic drug use might be related to
concurrent changes in benzodiazepines, utilization of both was assessed.
Management:
Counselling – Counselling visit for a minimum of 20 minutes provided to patients with
dementia and family members during which patient/caregiver guides, support
resources, care management, advance and safety planning can be discussed.
Specialist Referral – Referral to a Geriatrician, Neurologist or Psychiatrist is
recommended when management or diagnosis is problematic.
108
5.2.4 Explanatory Variables
I included the following patient demographic variables as possible contributors
to variations in provision of dementia care: age, sex, income quintile, number of
multiple chronic diseases and location (health authority).
The number of chronic diseases was defined in two different ways. One counted
the occurrence of the twelve most prevalent chronic conditions in seniors
65
measured
by the presence of at least two physician diagnoses over a two-year period, consistent
with prior literature on chronic condition ascertainment
218
. The chronic conditions
included hypertension, depression, arthritis or osteoarthritis, cancer, congestive heart
failure, diabetes, cardiovascular disease, stroke, chronic obstructive pulmonary disease,
chronic kidney disease, liver disease. I used a 10-year assessment period counting back
from the date of dementia diagnosis to ensure complete capture of prevalent disease.
A second measure validated in the literature, the count of major adjusted diagnostic
groups (mADGs), was characterized for each patient using the John Hopkins Adjusted
Clinical Group Case-Mix system as previously described in Chapter 4, section 4.4.1.1.
This measure of mADGs was used to test the sensitivity of my measure of co-morbidity
and no difference was found between the count of prevalent chronic conditions versus
number of mADGs therefore the count of prevalent chronic conditions was used as a
better adjustment for rule out diagnostic testing.
5.2.5 Statistical Analysis
Characteristics of the patient cohort and dementia care processes are described
using counts and percentages. Each dementia care process was a dichotomous
109
response variable (received the process of care, yes or no), therefore I estimated the
probability that patient characteristics (i.e. age, gender, income, co-morbidity and
geography) affected the likelihood that individuals will receive each particular dementia
care process using multivariate logistic regression analysis. All variables were retained
in the model regardless of significance and coefficients displayed as odds ratios. All
variables were tested for collinearity. Model fit was assessed using Hosmer-Lemeshow
goodness-of-fit test. Analyses were performed using STATA, version 10
227
.
5.3 Results
The study population was limited to community dwelling seniors because those who
live in a LTC facility are known to have attenuated service use once admitted. Separate
analyses of seniors living in a LTC facility revealed that they receive a significantly lower
proportion of guideline-consistent dementia care processes compared to non-residents
(Table 5.1). This is particularly apparent for the imaging, counselling and specialist
referral dementia care processes where residents are unlikely to leave the LTC facility for
a CT scan (18% received a CT scan) or referral to a specialist (1% received a referral). An
extremely low proportion of residents receive in-office examinations (4%) and in-office
individual counseling (5%); this is not un-expected given that the majority of residents
of LTC facilities are seen by facility physicians. Given these differences in primary care
utilization patterns, seniors living in a LTC facility or newly admitted to a facility in
2009/10 were excluded from the analysis reported here.
110
Table 5.1 – Proportion of seniors 69+ years of age residing in a LTC facility receiving
guideline-consistent care
Dementia Care
Process
Service Measured Lives in LTC facility
(n=1107)*
Laboratory
Complete hematology profile 73%
Serum electrolytes 70%
Serum calcium 13%
Serum glucose 39%
Thyroid stimulating hormone 54%
B12 vitamin 34%
Imaging Head computed tomography 18%
Counselling &
Specialist
Referral
Complete examination in-office 4%
Individual counselling in-office 5%
Referral to specialist re. Dementia 1%
*Excludes individuals who died
In 2009/10, 9820 individuals were newly diagnosed with dementia in BC. Of these, 8
were not registered in BC for the full year, 1328 were excluded due to death during the
one-year follow up period and 1439 were residents in a LTC facility or newly entered a
LTC facility and were excluded. This left a study population of 7045 individuals (Figure
5.1).
Figure 5.1 – Flow of study participants
Individuals newly diagnosed
with dementia in 2009/10
(n=9820)
Final inclusion in analysis
(n=7045)
Excluded (n=2775)
Zero days registered (n=8)
Died during year of diagnosis
(n=1328)
Lives in a LTC facility (n=1439)
111
5.3.1 Final Study Participants
Patients in the community dwelling dementia cohort had a mean age of 83±6.41,
were primarily female (59%) and had a mean income quintile of 2.91±1.60 (where 1
indicates low and 5 indicates high, Table 5.2). Over 50% had three or more chronic
diseases other than dementia. The five most prevalent chronic conditions were
hypertension (72%), arthritis/osteoarthritis (39%), cancer (32%), depression (28%) and
diabetes (26%) (Table 5.2). The study population had a mean number of 2.51±1.29
major ADGs, which indicates this is a high-needs population (1 indicates low and 5
indicates high, Table 5.2). The dementia cohort was fairly evenly distributed across the
four larger health authorities, with the fifth contributing 3.3% of the cohort, consistent
with the small population in this health authority.
112
Table 5.2 – Baseline characteristics of community-dwelling individuals newly diagnosed
with dementia in 2009/10
Characteristics Patients with
Dementia
(n=7,045)
Age in years (mean+ SD) 82.71 + 6.41
69-80 37.5%
81-90 51.3%
91+ 11.2%
Sex*
Male 41.2%
Female 58.7%
Income Quintile (mean+ SD)* 2.91 + 1.60
1- Lowest income 24.3%
2- Lower income 21.0%
3- Medium income 18.8%
4- Higher income 17.2%
5- Highest income 17.5%
Multiple other chronic diseases (not including
dementia)
0 chronic diseases 5.6%
1 chronic disease 18.2%
2 chronic diseases 25.8%
3+ chronic diseases 50.4%
Number of major ADGs (mean+ SD) 2.51 + 1.29
Treatment Prevalence of major chronic diseases
Hypertension 72.0%
Depression 27.6%
Arthritis or Osteoarthritis 38.7%
Cancer 31.6%
Congestive heart failure 19.6%
Diabetes 25.7%
Cardiovascular disease 18.3%
Stroke 8.4%
Chronic obstructive pulmonary disease 8.0%
Delirium 7.7%
Chronic kidney disease 6.2%
Liver disease 0.7%
Health Authority*
Interior 22.5%
Fraser 27.3%
Vancouver Coastal 25.0%
Vancouver Island 21.7%
Northern 3.3%
*Missing data: Sex Unknown=13, Income Unknown=98, Quintile HA Unknown=17
113
The majority (75%) of patients were determined to be newly diagnosed incident
dementia cases using family physician payment records showing a minimum of two
MSP claims associated with ICD-9 dementia codes (Figure 5.2). Nine percent of patients
received the designation as a result only of a hospital diagnosis; the remaining 16%
received both a hospital diagnosis and a diagnosis on at least one MSP claim. For 71%
of those who received their diagnoses through the family physician payment claim
route, both initial MSP claims originated with the same family physician. Of those who
received a hospital diagnosis and a minimum of two MSP claims, 59% of incident cases
were based on MSP claims from the same physician. Receiving a diagnosis of dementia
in hospital may be indicative of patients whose dementia went unrecognized in the
community and therefore may have more severe dementia.
Figure 5.2 – Location of diagnosis of dementia (Cohort 2)
114
5.3.2 Variations in Dementia Care Processes
Overall, a high percentage of the cohort received most of the guideline-consistent
laboratory tests (between 74 to 89%, except calcium and B12) and imaging (58%
received a head CT) (Figure 5.3A) within a year of diagnosis. Under half of the cohort
received a prescription for an AChI (40%) or antidepressant (43%) (Figure 5.3B).
However, over a quarter of the cohort received an antipsychotic (28%), suggested for
use only when other non-prescription therapies have been exhausted. A further 26%
receive a benzodiazepine, which is specifically not recommended for seniors with
dementia. A low percentage received referrals to a specialist (9%), or complete in-
office examinations (32%).
115
Figure 5.3A – Percentage of seniors receiving guideline-consistent lab testing and imaging
116
Figure 5.3B – Percentage of seniors receiving guideline-consistent prescription and dementia care management processes
117
In multivariate analysis, after adjusting for patient demographic characteristics,
there is a consistent decreasing gradient in receipt of guideline-consistent dementia
care with age (Figure 5.4A-C; Appendix C.3A, B). Seniors 91 years and older had
significantly lower odds of receiving a specialist referral (0.24 CI 0.14-0.40), an AChI (0.33
CI 0.26-0.42), the serum glucose test (0.35 CI 0.28-0.44), a complete examination in-
office (0.38 CI 0.30-0.49), individual counselling in-office (0.41 CI 0.33-0.50), a head CT
(0.51 CI 0.42-0.63) and all other laboratory tests compared to seniors 69-80 years old. A
comparable pattern was noted for seniors 81 to 90 years old compared to the reference.
Figure 5.4A – Adjusted odds ratio of receiving calcium, B12, laboratory tests and computed
tomography scans
118
Figure 5.4B – Adjusted odds ratio of receiving acetyl-cholinesterase inhibitors, antipsychotics and
non-recommended benzodiazepines
Figure 5.4C – Adjusted odds ratio of receiving a physical exam, counselling or specialist referral
119
Similar to age, there is a gradient by income. After adjusting for other
characteristics, those in the medium, higher and highest income categories had higher
odds of receiving individual counselling in-office (1.15, 1.20 and 1.31 respectively)
compared to those in the lowest income category (Figure 5.4C). Individuals in the
highest income category had higher odds of receiving a specialist referral (1.51 CI 1.18-
1.95) with a 4% difference in proportion compared to those in the lowest income
category. They also had higher odds of receiving an AChI (1.34 CI 1.14-1.56), but lower
odds for all other prescriptions of antipsychotics (0.79 CI 0.67-0.93), trazodone (0.71 CI
0.54-0.92) and antidepressants (0.85 CI 0.73-1.00) (Figure 5.4B). Of note, individuals
with dementia and other co-morbidities had significantly higher odds of receiving
antidepressants, particularly those with three or more co-morbidities (4.10 CI 3.18-5.28)
compared to individuals with dementia and no co-morbidities.
Despite having the smallest population, the Northern health authority had the
highest proportion of individuals receiving laboratory tests and head CTs compared to
the other health authorities (Figure 5.3A) with higher odds for both processes compared
to the Interior health authority (Figure 5.4A). The Fraser health authority, which contains
a major urban centre, had a significantly higher proportion of seniors receive an AChI
(43%) with a 12% difference and an odds ratio of 1.61 (CI 1.40-1.85) compared to the
reference (Figure 5.3B, Figure 5.4B). Conversely, individuals in the Fraser (0.78 CI 0.68-
0.91), Vancouver Coastal (0.81 CI 0.70-0.94) and Vancouver Island (0.84 CI 0.71-0.97)
health authorities (all with major urban centres) had lower odds of receiving an
antipsychotic compared to the Interior health authority reference. The Fraser (0.83 CI
120
0.71-0.96) and Vancouver Coastal (0.69 CI 0.59-0.81) health authorities also had a lower
odds for receiving a benzodiazepine.
5.4 Discussion
There is significant variation in the proportion of individuals newly diagnosed with
dementia who receive guideline-consistent dementia care. In a universal healthcare
system like Canada’s where there are no explicit financial barriers to care, this variation
is still associated with income and geography, as well as age and sex. These results are
consistent with those reported elsewhere
31,155,192
. However, to the best of my
knowledge, this is the first study to use administrative data to capture provision of
clinical services at a population-level to all seniors newly diagnosed with dementia and
to assess the full scope of dementia care processes.
I find that the magnitude of variation for what may be considered essential dementia
care processes, such as regularly administered laboratory exclusionary tests (except
calcium and vitamin B12) and head CTs, is comparatively small. As expected, at least
60% of patients received these care processes. This is also not surprising as over 75% of
the individuals had two or more chronic diseases and would have received many of the
laboratory tests or a CT scan as part of standard diagnostic protocol in cases involving
multiple concurrent chronic conditions. However, this is not the case for more
discretionary but equally important processes, such as a complete examinations in-
office, in-office counselling and specialist referrals. It is expected that dementia care
management processes such as these, particularly counselling, should have a smaller
121
magnitude of variation. Counselling is essential for lowering anxiety and preparing the
patient and family for what to expect in light of such a devastating diagnosis. Given
that physician-reported barriers to provision of dementia care highlight insufficient
time, difficulty in accessing specialists, and lack of expertise as key constraints, the
magnitude of variation is unsurprising
70,71
.
Age significantly influences many of the dementia care processes assessed in this
study. Given the increased probability of concurrent multiple chronic disease with
advanced age, physicians may be less inclined to do extensive laboratory test or CTs on
patients as they get older due to complications and unnecessary hardship. However
this rationale does not hold for the absence of counselling or specialist referrals for
those of advanced age. Indeed, these processes should be provided, particularly in the
case of multiple co-morbidities, to better explore and alleviate potential medication
interactions, complex symptoms and develop strategies to cope
4
. One possibility is
that of age discrimination in practitioners’ attitude toward providing early diagnosis and
broader interventions to older individuals. This has been described previously as a
barrier to provision of good dementia care with physicians’ negative perceptions in
diagnosing dementia being associated with therapeutic nihilism, stigma and
ageism
41,228
. Second is the lack of time to appropriately support older individuals with
more complex symptoms and difficulty accessing specialists with limited availability
41,71
.
An alternate possibility is that the severity of dementia at diagnosis, which I was unable
to measure, may play a role in mediating the relationship between age and the
management care processes.
122
I also find a pattern of inequality, in which seniors newly diagnosed with dementia in
higher income categories had significantly higher odds of receiving individual
counselling and referrals compared to those of lower income. However it is impossible
to know if this pattern arises because of patient health seeking behaviour where those
with higher education are associated with asking for services
204
or because of a bias in
physician behaviour. Insofar as income is unrelated to patient need for services, these
observations deserve further attention.
Variations in use of medications recommended for dementia are more nuanced.
Consistent with earlier work, a lower percentage of seniors newly diagnosed with
dementia receive recommended AChIs
143
. This discretionary prescription practice may
be related to cautions both in the guidelines and literature around the small to modest
benefits of AChI over short durations, the side effects and potential polypharmacy
associated with their use in individuals with higher co-morbidity
229–231
. Receipt of
AChIs is influenced by income with individuals in higher income categories having
higher odds of receiving AChIs. Coverage of AChI in BC is only available under a
program designed to provide coverage of AChIs to patients with mild to moderate
Alzheimer’s Disease who also agree to participate in a research study to assess the
clinical effectiveness of AChIs
134
. Therefore, for all other individuals with dementia, use
of AChIs would be an out-of-pocket expense, explaining the income related variation.
In contrast, I find a higher than expected percentage of seniors with dementia
receiving antidepressants, antipsychotics and non-recommended benzodiazepines. The
off-label use of antipsychotics and benzodiazepines to treat behavioural and
123
psychological symptoms of dementia has been shown to have low efficacy at best and
to introduce significantly increased risk of adverse effects at worst
232,233
. Use of atypical
antipsychotics in seniors with dementia comes with a black box warning, but there is
also no evidence that typical antipsychotics and benzodiazepines mitigate the risk of
death in dementia any better than atypical antipsychotics
231
. The guidelines
recommend extreme caution when prescribing antipsychotics and only when alternate
options to control these behavioural symptoms such as environmental (identifying
potential triggers during meal or bath times) or psychosocial interventions (music, pet
therapy, light exercises) are inadequate
122
. I measure use of these drugs among
patients in the early stages of a diagnosis of dementia, during which neuropsychiatric
symptoms have generally not fully manifested, who should not require the use of these
therapeutics, that is, unless diagnosis has been delayed and the dementia has
progressed in severity or these drugs prescribed for a different condition. Yet I find that
over one quarter of seniors newly diagnosed with dementia receive an antipsychotic
(28%) and/or benzodiazepine (26%). Additionally, individuals in low-income categories
and living within health authorities with smaller urban centres (and less access to
specialists/family physicians) were more likely to receive antipsychotic treatment. I
noted a similar geographic trend with the variation in benzodiazepine use. Use of these
drugs should be sporadic, short-term and infrequent in this population. These trends
are an indication of larger issues within a healthcare system with drugs substituting for
other forms of more appropriate non-pharmacological care.
124
5.4.1 Limitations
There are some limitations to this analysis. I am not able to account completely
for patient need, including assessing dementia severity; nor can I completely account for
psychiatric illness, co-morbidities, or patient preferences for care, health seeking
behaviours or health goals. To the extent that variation is the result of clinical
assessment of need (which may be the case considering age, sex and co-morbidities)
variations may be appropriate. However the full magnitude of variation by age, as well
as those observed by income and geography, are unlikely to be explained by patient
need or preference and merit further investigation. I was also unable to measure
cognitive testing which is a key component of the diagnostic process and has been
shown previously to have significant variation
31
. Administrative data cannot shed light
on the important behavioural management aspects of dementia care. Lastly, I was
unable to measure referrals to community care, mental health or speciality services
which are invaluable resources for both patients and their caregivers to meet the long-
term psychosocial management needs of dementia. These services are able to provide
specialized education and support to patient and physicians, though I do not anticipate
that this should affect receipt of other guideline-recommended services.
Programs for pharmaceutical coverage in BC also complicate interpretation of
findings with respect to income and pharmaceutical use. For example, Fair Pharmacare
is an income-related program which could contribute to the income-related variation in
the use of other drugs assessed in this study. While the BC guidelines and national
Canadian guidelines have almost identical recommendations, unlike the BC guidelines,
the national guidelines do recommend the short-term use of benzodiazepines for
125
behavioural-related symptoms. In both cases, recommendations are based on
management of dementia alone, despite the fact that co-morbidity is high in this
patient population.
5.5 Conclusion
The systematic review from Chapter 2 demonstrated that self-reported dementia
care practices amongst physicians have wide variability, particularly for use of a formal
memory test, therapeutics and making referrals to specialists or community services.
The findings here, based on analysis of administrative data, are consistent with those
earlier conclusions. I found wide variation in use of pharmacological interventions,
particularly those not recommended for dementia patients, and in specialist referrals
along with in-office examinations and counselling. In particular, I observe patterns of
antipsychotic and benzodiazepine prescription that are indicative of their use as
substitutions for other types of non-pharmaceutical care. Encouragingly, several other
dementia care processes such as exclusionary blood work and imaging had relatively
high proportions of use with less variation.
There is no expectation that everyone will receive all guideline-recommended
care because of differences in needs based on personal circumstances and the severity
of dementia when it is diagnosed. Of more interest are the variations I find in use of
services across patient characteristics. These differences unrelated to patient need say
more about equity in health care services use. Patterns of inequality by age, income
126
and geography may signal barriers to care, particularly for the management of dementia
care processes such as counselling and referral to specialists.
These results add to our understanding of actual physician dementia practice
patterns and highlight the gap between actual practice and guideline-recommended
care. This has important implications as it corroborates the qualitative literature around
barriers to providing good dementia care including time constraints during
appointments, problematic attitudes resulting in age discrimination, lack of
interdisciplinary teams and difficulty accessing community resources
41,71
resulting in
patterns of inequity. Physician education that goes beyond the pathophysiology and
pharmacological treatment of dementia as a disease and instead emphasizes dementia
as a chronic, complex condition which can benefit from timely diagnosis, comprehensive
biopsychosocial treatment and management would play an essential role in reframing
physicians’ approach. At a policy level, these results indicate the need to recognize that
primary care physicians alone may not be able to provide adequate care and the need
for integration across care systems to adequately support patients. The next step in this
analysis will be to examine the association between receipt of dementia care processes
and health outcomes. Chapter 6 will provide more detailed descriptives on the health
outcome of interest, transitions, before I model associations in Chapter 7.
127
CHAPTER 6 – IDENTIFYING POINTS OF CARE WHEN
TRANSITIONS ARE HIGHEST AND THE FACTORS THAT
CONTRIBUTE TO THEM
6.1 Introduction
A transition is defined as a physical move to a formal acute or community-based
healthcare institution other than the home with a stay of at least one night
10
. A greater
number of transitions between care settings poses a serious challenge to continuity of
care and the safety of the patient and is associated with medication errors, preventable
hospital readmissions and increased mortality risk
13–17
. Safety challenges are
compounded by symptoms of dementia such as disorientation and behavioural
outbursts, and these are potentially exacerbated and if a caregiver is not present to
assist in the patient’s transitions
12,35
. Particularly important is the fact that the overall
number of transitions experienced is higher for people with dementia compared to
those without dementia
37
. Guidelines for good dementia care stress the need for
continuity and familiar environments which cannot be consistent with a higher number
of transitions
10,11,22,26
.
Given the adverse outcomes associated with transitions, particularly for dementia
patients, surprisingly few studies have examined the longitudinal dynamic of transitions
between home, hospital, respite care and LTC facilities experienced by individuals with
dementia
10,12,37
and only one has done so beginning in the year in which a diagnosis of
dementia was first established
37
. Individuals newly diagnosed with Multiple Sclerosis
and Parkinson’s disease have been shown to experience peaks in healthcare utilization
128
and transitions in the year of diagnosis
234–236
. Individuals with dementia also shows
peaks in use, but this has not been examined specifically in the year of diagnosis
37
.
Research literature on end-of-life has demonstrated that individuals living in the
community with dementia experience 32% more care transitions than those without
dementia in the two years before death
10
. As outlined in my conceptual framework in
Chapter 2, Section 2.7, age, gender, morbidity
10,15,237
, behavioural symptoms associated
with dementia
238
and the presence of a caregiver
35,100
are thought to affect the number
and type of transitions, Yet none of the studies that have longitudinally described
transitions have examined these factors.
Examining transitions experienced longitudinally has the potential to identify
points of care when transitions are highest and the factors that contribute to those
transitions. Understanding these patterns can help assess the relationship between
receipt of care and outcomes and can identify possible points of intervention to reduce
transitions. My research questions are:
Research Question 3a: What is the rate of transitions experienced over the study
period and what are the points of care during which transitions are highest?
Research Question 3b: Are there individuals who consistently experienced a high
number of transitions and if so what are their characteristics?
Research Question 3c: What are the types of transitions experienced and what factors
influence these transitions?
Research Question 3d: What factors influence the number of transitions during the
peak points of care and overall?
129
I hypothesize that:
i) The number of transitions experienced will be higher at the point of diagnosis and at
the end-of-life compared to the time period after diagnosis and before their end-of-life;
ii) The individuals who consistently experience a high number of transitions will have a
high number of co-morbidities in addition to dementia and behavioural symptoms of
dementia compared to those with a low number of transitions;
iii) Hospitalizations will drive transitions, but a large proportion of the cohort will also
move to LTC. Age, morbidity and behavioural symptoms will be the factors associated
with hospitalizations or a move to LTC. The lack of a caregiver will be significantly
associated with a move to LTC but not hospitalizations;
iv) The factors associated with a higher number of transitions will be age, gender (male),
morbidity, behavioural symptoms of dementia and the lack of a caregiver both during
the peak points and overall.
6.2 Methods
I analyzed eleven years of retrospective health care administrative data (April
2000/01 to March 2011) for a cohort of individuals newly diagnosed with dementia in
2001/02.
6.2.1 Data Sources
All seven databases described in Chapter 4 were used for this analysis. That
includes: 1) the central consolidation file 2) the medical services plan (MSP) payment file
130
3) Hospital abstract data 4) Vital statistics death file 5) Home and Community Care data
6) PharmaNet file 7) the College of physicians and surgeons of BC file.
6.2.2 Study Population
Cohort 1, described in Chapter 4 section 4.3.1 was used for this component of the
project. Briefly, these were all seniors in BC, 65 years and older, with a new diagnosis of
dementia in fiscal year 2001/02, followed backward one year (April 2000) and forward
until death or the end of the study period (March 2011). A wash-in period of two years
was used to ensure no other dementia diagnoses occurred and that new incident cases
were being captured
5
. Individuals were excluded if they were not registered for health
care coverage for the whole study period (prior to death) or if they were not an incident
case in 2001/02, i.e. if they had dementia diagnoses during a 2-year wash-in period
(April 1999 – April 2001).
6.2.3 Primary Outcome – Transitions
Care transitions are defined as a physical move from one location to another with
a stay of at least one night. Only transitions between among, publicly-funded respite
care, Assisted Living, LTC and hospitals were included. Outpatient care or moves
between in-patient wards of the same facility were excluded, consistent with prior
literature
10,37
. These variables were calculated on an annual fiscal year basis for each
year of the study period (2000/01 – 2010/11) during which the participants were alive.
5
While the cohort was only followed backward one year for analysis, incidence was determined based on a two year
wash-in period, accessing data from 1999/2000
131
6.2.4 Explanatory Variables
I examined the association between the number of transitions experienced and
the following patient variables: age; sex; neighbourhood income quintile; health status
(measured by number of major ADGs, described in detail below); caregiver proxy status
(identified using a shared MSP id variable in the consolidation file, described in detail in
Chapter 4, Section 4.2.1); health authority of residence (described in detail in Chapter 4,
Section 4.4.1); and the presence of treated behavioural symptoms (measured by a
prescription of an antipsychotic or benzodiazepine, described in Section 4.4.1). Further
analysis demonstrated that 45% of those in the cohort prescribed a benzodiazepine in
the year of diagnosis had not received a prescription for a benzodiazepine at any point
in the three previous years. This suggests the prescription in the year of diagnosis was
likely related to the dementia diagnosis (data not shown). Fifty-five percent had at least
one prior prescription and 28% a prescription in all three years preceding diagnosis,
indicating long-term use for reasons unrelated to dementia (such as anxiety or
insomnia).
In addition to patient-level variables, I adjusted for physician practice variables
including the number of years the physician had been practicing, the number of
patients over the age of 65 in the practice, and the number of dementia patients in the
practice. These were attributed to the patient-level data based on the physician
providing plurality of care, described in detail in Section 4.4.1.2.
Patient health status was defined in two different ways. As previously described,
these were to 1) count the occurrence of the ten most prevalent chronic conditions in
seniors
65
measured by the presence of at least two physician diagnoses for that
132
condition over a two-year period; and 2) count the number of major conditions
(adjusted for in regression analysis), using the Johns Hopkins Adjusted Clinical Group
Case-Mix system, described in Chapter 4, Section 4.4.1.1.
6.2.5 Statistical Analysis
Characteristics of the patient cohort are described using counts and percentages.
The mean annual number of transitions experienced during the entire study period,
total number of care transitions, average rate of transitions per year alive, percentage of
participants who experienced transitions, hospitalizations, LTC moves, second LTC
moves and the length of stay in hospital are reported.
The data were analyzed using negative binomial regression to account for over-
dispersion for two outcomes: a) the number of transitions experienced in the year of
diagnosis and b) the number of transitions experienced over the entire study period.
Zero-inflated Poisson analysis and standard Poisson analysis were used for the
remaining two outcomes respectively: c) the number of hospitalizations experienced in
the year of diagnosis; and d) the number of moves between LTC facilities experienced in
the year of diagnosis.
The choice of models in each case was determined empirically. The mean and
variance for each outcome variable was first assessed to determine if they fit Poisson
assumptions of the mean being equal to the variance. A histogram of each outcome
variable was graphed to determine its distribution. An exploratory Poisson regression
model was run for all four outcomes and the Pearson Chi-Square goodness-of-fit test
133
used to measure the distribution of the outcome variable. When the Pearson
goodness-of-fit test results were significant (indicating that the distribution of the
outcome variable differs significantly from a Poisson distribution), other models were
considered. Larger standard errors and wider confidence intervals were also assessed as
those indicate the Poisson model did not fit the data.
If the distribution of the outcome variable showed an excess of zero counts, a
zero-inflated Poisson regression model was employed with the option of robust
standard errors. The Vuong test was used to measure if a zero-inflated model was
better than an ordinary Poisson regression model. In other cases, a negative binomial
model was used and the Likelihood Ratio test was used to measure if this distribution
was a more appropriate fit than Poisson (a large test statistic and an alpha significantly
different from zero would suggest the outcome is over-dispersed and a simpler Poisson
distribution is not sufficient). Finally, in all cases, model fit was determined by the
adjusted coefficient of determination (R
2
), indicating how much of the variance in the
data is accounted for by the model. The data was formatted with time, counted as
months from the diagnosis to death. All models adjusted for individual characteristics,
death and physician practice variables except for when modelling the total number of
transitions experienced over the study period as it was not possible to accurately
attribute a single physician who provides plurality of care over the entire study period.
Analyses were performed using STATA, version 10
227
.
134
6.3 Results
6.3.1 Baseline Characteristics
In BC, 6876 individuals were newly diagnosed with dementia in 2001/02. The
cohort had a mean age of 82.3 ± 6.96 years upon entry, was primarily female (62%) and
had a median average income quintile of 2.8 ± 1.5 (where 1 indicates low and 5
indicates high, Table 6.1). Over one-third of the cohort received their diagnosis of
dementia in hospital, either through a dementia hospital code alone or a combination
of hospital and one MSP code (37%) (Figure 6.1). Just under a quarter (23%) of the
cohort had no evident chronic disease other than dementia, while 44% had two or more
chronic diseases with an average of 2.6 ± 1.3 major ADGs (including dementia),
indicating a somewhat high needs population. At the time of entry, 75% of the cohort
were living in the community and were fairly evenly dispersed across the health
authorities, except for the Northern health authority which contained 3.8% of the
cohort, consistent with its small population size. Within the first year 18% of the cohort
died and more than 50% died by year 4 (Figure 6.2).
135
Table 6.1 – Baseline characteristics of individuals newly diagnosed with dementia in
2001/02
Characteristics Patients with Dementia
(n=6,876)
Age in years (mean+ SD) 82.3 + 6.9
65-75 17.4%
76-85 47.9%
86+ 34.7%
Sex*
Male 37.7%
Female 62.0%
Income Quintile (mean+ SD)* 2.8 + 1.5
1- Lowest income 24.8%
2- Lower income 19.4%
3- Medium income 16.8%
4- Higher income 15.6%
5- Highest income 17.1%
Multiple chronic disease (not including dementia)
0 chronic diseases 23.1%
1 chronic disease 32.9%
2 chronic diseases 25.3%
3+ chronic diseases 18.7%
Number of major ADGs (mean+ SD) 1.5 + 0.7
Treatment Prevalence of major chronic diseases
(other than dementia)
Hypertension 36.4%
Depression 21.3%
Arthritis or Osteoarthritis 18.1%
Cancer 16.8%
Congestive heart failure 14.8%
Diabetes 13.9%
Cardiovascular disease 12.3%
Stroke 5.3%
Chronic obstructive pulmonary disease 4.2%
Chronic kidney disease 1.3%
Liver disease 0.3%
Living in a LTC Facility
Yes 11.2%
No 88.8%
136
Characteristics Patients with Dementia
Health Authority*
Interior 18.9%
Fraser 26.1%
Vancouver Coastal 26.7%
Vancouver Island 24.4%
Northern 3.8%
Figure 6.1 – Location of diagnosis of dementia (Cohort 1)
*Missing data: Sex unknown=26, Income unknown=430, Health Authority unknown=45
137
Figure 6.2 – Survival curve of cohort
6.3.2 Aggregate Healthcare Use and Transitions
Table 6.2 shows aggregate healthcare use during the entire period. On average,
participants survived a relatively short period after diagnosis (median 4.6 years),
experiencing a mean of 3.7 (± 2.9) transitions, or a rate of about one transition per year
alive (0.9 ± 0.8 transitions) (Table 6.2). Almost 90% of the cohort experienced a
hospitalization at some point during their trajectory, with a median of 32 accrued
hospital days.
138
Table 6.2 – Transitions and healthcare use during the entire study period
Use Total Sample
(n=6,876)
Total transitions (mean ± SD; median) 3.7 ± 2.9; 3.0
Total years alive (mean ± SD; median) 5.2 ± 3.2; 4.6
Transitions per year alive (mean ± SD) 0.9 ± 0.8
Participants who experienced a transition 94.6%
Participants who died (%) 83.9%
Participants with any hospital stay (%) 87.5%
Total hospital days accrued (mean ± SD; median) 50.1 ± 58.0; 32.2
Patients with LTC use (%) 57.6%
When the cohort was stratified by the rate of transitions experienced per year
alive, 5% experienced no transitions while 1% experienced 4 or more transitions per
year alive (Table 6.3). Participants who experienced 4 or more transitions per year alive
had shorter survival post-diagnosis than those with fewer transitions (2.0 ± 1.2 years),
had higher needs (3.6 ± 1.2 major ADGs), a majority had behavioural symptoms of
dementia (70% received a prescription for an antipsychotic or benzodiazepine), 42% had
a caregiver and only 29% remained in the community over the study period.
Conversely, participants who experienced no transitions during their life time survived
on average for 5.3 ± 3.6 years, had fewer major ADGs (1.7 ± 1.0), only 34% received a
prescription for an antipsychotic or benzodiazepine, 31% had a caregiver and 62%
remained in the community over the study period. Examined further, participants
categorized by the average rate of transitions experienced over their time alive
consistently remained in that category even when examining total transitions they
experienced in each year (Figure 6.3). For example, individuals categorized into the high
average transition category (4+ transitions) had a high number of total transitions
139
(between 3 to 6 transitions) in each year that they survived, not just an overall high
average (Figure 6.3).
Table 6.3 – Cohort characteristics stratified by rate of transitions
No
transits
(n=375)
≤1
transit/yr
(n=4279)
>1-≤2
transits/yr
(n=1586)
>2-≤4
transits/yr
(n=572)
4+
transits/yr
(n=64)
% of cohort 5% 62% 23% 8% 1%
Age (mean ± SD) 82.9 ± 8.0 83.0 ± 7.0 82.7 ± 6.7 82.8 ± 6.6 82.5 ± 7.8
Gender (% female) 69% 65% 56% 53% 53%
Years alive (mean ± SD) 5.3 ± 3.6 6.2 ± 3.1 3.6 ± 2.2 2.6 ± 1.3 2.0 ± 1.2
No. major ADGs 1.7 ± 1.0 2.3 ± 1.2 3.1 ± 1.3 3.6 ± 1.3 3.6 ± 1.2
Prescription of antipsychotic or
benzo (%)
34% 43% 56% 62% 70%
Has a caregiver (%) 31% 38% 35% 38% 42%
Remained in community (%) 62% 52% 26% 20% 29%
Figure 6.3 Transitions rate in each year for those with <1.0, 1.0-1.9, 2.0-3.9 and 4.0+
average transitions per year for 2001/02-2003/04, by year of death (up to 2003/04)
140
6.3.3 Transitions during Year of Diagnosis
Over the study period, individuals experienced a high number of transitions
during the year of diagnosis, regardless of survival time (Figure 6.4A). Sixty-six percent
of the cohort experienced at least one transition, while 17% of the cohort experienced 3
or more transitions that year (Appendix Table D.1). The transitions were primarily driven
by hospitalizations (Figures 6.4B, 6.4C).
Figure 6.4A – Transition rates over time for those diagnosed in 2001/02, by year of death
141
Figure 6.4B – Hospitalization rates over time for those diagnosed in 2001/02, by year of
death
Figure 6.4C – Other transitions over time for those diagnosed in 2001/02, by year of death
Over 60% of the cohort experienced a hospitalization in the year of diagnosis
with a length of hospital stay of 41.7 ± 49.5 days (median 25 days). Over one third of the
cohort was identified based on dementia hospital codes only, or a hospital and one MSP
142
code (Figure 6.1). Of this group, only 15% received dementia as a principal diagnosis
when hospitalized (Appendix Table D.2). The remainder received a secondary diagnosis
of dementia but were hospitalized for other reasons with the most frequent three being
a fractured femur (9%), pneumonia (5%) and heart failure (4%) (Appendix Table D.2).
Excluding this group, the average number of transitions in the year of diagnosis was 1
where 49% of the cohort diagnosed in the community experienced at least 1 transition
and 12% experienced 3 or more transitions (data not shown).
Looking a little more closely at those who did not receive their initial diagnosis of
dementia in hospital, a quarter made their first move to LTC in the year they received
their diagnosis. Of these movers, those who experienced a hospitalization had a mean
length of stay in hospital of 51.4 days while those who did not move to LTC that year
had a mean length of stay in hospital of 29.7 days (data not shown). In comparison, for
those who received their initial dementia diagnosis in hospital, 36% made their first
move to LTC in the year they received their diagnosis and had a mean hospital length of
stay of 58.6 days (data not shown). Those in this group that did not move to LTC had a
mean hospital length of stay of 29.3 days (data not shown).
The risk of moving to a LTC facility was highest early in the care trajectory,
specifically during the year of diagnosis, and diminished markedly over time (Figure 6.5).
Indeed, the longer individuals survived post diagnosis, the less likely they were to end
up in LTC, though survivorship bias needs to be considered as individuals with a better
health status may be more likely to survive longer and remain in the community.
143
Figure 6.5 – Number and proportion of those eligible moving from community to LTC, by
years since diagnosis
About half of those who remained in the community experienced a
hospitalization, while 1 in 5 of those in the community experienced at least one re-
hospitalisation (Figure 6.6). Among those who newly moved to LTC in the year of
diagnosis, only 19% experienced no other transition while most patients had two (45%),
three to five transitions (30%), or more (5%) transitions including re-hospitalizations and
moves between LTC facilities (Figure 6.6). Of the 11% who were already in LTC when
they received their diagnosis, just under half had no transitions, while almost 1 in 4
individuals were hospitalized. Over 10% experienced re-hospitalizations and 11% were
transitioned to a different care facility, despite being in LTC when they received their
diagnosis. Of the 18% of the cohort that died in the year of diagnosis, the majority
experienced a transition (55%) and a further 36% experienced re-hospitalizations (Figure
6.6). Stratified by location, half of the cohort remained in the community during the
year of diagnosis.
144
Figure 6.6 – Frequency of different transition types experienced by newly diagnosed
dementia patients
6.3.4 Transitions at End-Of-Life
Participants who survived longer over the ten-year study period had fewer
transitions and hospitalizations over time (Figures 6.4A, B). However, regardless of
survival time, participants not in LTC facilities experienced a marked increase in
hospitalizations in the year prior to and the year of death, often exceeding
hospitalizations in the year of diagnosis (Figure 6.7A). Further, those with a high
number of major ADGs experienced a significant increase in hospitalizations regardless
of whether they lived in a LTC facility or in the community, whereas those who had no
major ADGs other than dementia did not experience this increase at end-of-life. This
suggests that it is not dementia alone, or even primarily, that drives hospitalizations at
end-of-life (Figure 6.7B).
145
Figure 6.7A – Hospitalization rate over time for those diagnosed in 2001/02, by LTC
status and year of death
Figure 6.7B – Hospitalization rate over time for those diagnosed in 2001/02, by LTC status
and number of major ADGs
146
6.3.5 Factors Associated with Transitions
Multivariate analyses showed that the patient factors associated with a higher
rate of transitions in the year of diagnosis (after adjusting for physician practice patterns
and death) were: poor health status, age (older experienced more transitions, except the
oldest old), residing in health authorities with small urban centres, and at least one
prescription for a benzodiazepine or antipsychotic prescription drug (Table 6.4).
Multivariate analysis for the overall rate of transitions experienced over the entire study
period demonstrated that age was no longer significant, but all previously identified
factors associated with transitions in the year of diagnosis still were, with the addition of
gender (being male).
Looking at hospital transitions more specifically, poor health status, residing in
health authorities with small urban centres and receipt of a prescription for a
benzodiazepine were all associated with a higher rate of hospitalizations in the year of
diagnosis (Table 6.4). Neither age nor gender was significant. For transitions between
LTC facilities, poorer health status and living in rural health authorities continued to be
associated with a higher number of transitions between LTC facilities. However, age
(being older) and receipt of a prescription of benzodiazepine or antipsychotic drug were
also associated with transitions between LTC facilities. The presence of a caregiver was
associated with fewer transitions between LTC facilities (Table 6.4).
147
Table 6.4 – Multivariate regression analysis examining patient and practice factors
associated with transitions experienced
Transitions in
Year of
Diagnosis
Transitions over
Years Alive
NB1
Hospitalizations
in Year of
Diagnosis
Moves
between LTC in
Year of
Diagnosis
PATIENT VARIABLES
Number Major ADGs
None Reference Reference 1.4 (1.3-1.4)
‡ NB2
Reference
1-2 2.6 (2.0-3.4)
‡
1.4 (1.2-1.5)
‡
1.2 (0.9-1.6)
3-4 5.8 (4.5-7.5)
‡
2.3 (2.0-2.5)
‡
1.6 (1.2-2.2)
‡
5+ 8.1 (6.2-10.5)
‡
3.1 (2.7-3.5)
‡
1.8 (1.3-2.5)
‡
Age
65-75 Reference reference Reference Reference
76-85 1.1 (1.0-1.2)
‡
1.0 (0.9-1.1) 1.0 (0.9-1.1) 1.3 (1.2-1.6)
‡
86-95 1.2 (1.1-1.2)
‡
1.0 (0.9-1.1) 1.0 (0.9-.1.1) 1.7 (1.5-2.0)
‡
96+ 1.0 (0.9-1.2) 0.9 (0.8-1.1) 0.9 (0.7-1.1) 1.7 (1.3-2.3)
‡
Sex
Female Reference Reference Reference Reference
Male 1.0 (0.9-1.0) 1.1 (1.1-1.2)
‡
1.0 (0.9-1.0) 1.0 (0.9-1.1)
Health Authority
Interior Reference Reference Reference Reference
Fraser 1.0 (0.9-1.1) 0.9 (0.9-1.0) 1.0 (0.9-1.1) 1.1 (1.0-1.2)
Vancouver Coastal 1.3 (1.2-1.4)
‡
1.2 (1.1-1.3)
‡
1.2 (1.1-1.3)
‡
1.2 (1.0-1.4)
†
Vancouver Island 1.0 (0.9-1.1) 0.9 (0.9-1.0) 1.0 (0.9-1.1) 0.9 (0.8-1.1)
Northern 1.3 (1.2-1.5)
‡
1.3 (1.2-1.4)
‡
1.2 (1.1-1.4)
‡
1.0 (0.7-1.3)
Income Quintile
1- Lowest Reference Reference Reference Reference
2- Low 1.0 (0.9-1.1) 1.0 (0.9-1.0) 1.0 (0.9-1.1) 1.0 (0.9-1.2)
3- Medium 1.0 (0.9-1.1) 1.0 (0.9-1.0) 1.0 (0.9-1.1) 0.9 (0.8-1.1)
4- High 1.0 (0.9-1.1) 1.0 (0.8-1.0) 0.9 (0.9-1.0) 1.0 (0.9-1.1)
5- Highest 0.9 (0.9-1.1) 0.9 (0.8-0.9)
‡
0.9 (0.9-1.0) 0.8 (0.7-0.9)
‡
Prescription for benzodiazepine 1.2 (1.1-1.3)
‡
1.2 (1.1-1.3)
‡
1.2 (1.1-1.3)
‡
1.3 (1.2-1.4)
‡
Prescription for antipsychotic 1.1 (1.1-1.2)
‡
1.1 (1.0-1.1)
‡
1.1 (1.0-1.1) 1.6 (1.5-1.8)
‡
Presence of a caregiver 0.9 (0.9-1.0) 0.9 (0.9-1.0) 0.9 (0.8-1.0) 0.8 (0.7-0.9)
‡
6.4 Discussion
To the best of my knowledge, this is the first time administrative data have been
used to assess longitudinal patterns of transitions experienced by individuals newly
diagnosed with dementia. My data demonstrate that while individuals newly diagnosed
‡ - P value <0.001, † - P value <0.01,* - P value <0.05
NB1: Practice variables were not included in this model
NB2: Was modelled using a zero-inflated Poisson analysis with the number of major ADG variable as continuous
All models controlled for death and physician practice patterns except as noted in NB1
148
with dementia in B.C. in this particular year did not experience as high a number of
transitions overall as individuals in the US
37
(11.2±10.2), they did still experience on
average a transition per year alive, primarily to an acute care hospital. Over the ten-year
trajectory, I was able to identify two points of time during which a high number of
transitions were experienced: 1) the year of diagnosis and 2) the year of death. My
results corroborate others who have demonstrated that end-of-life is associated with
higher healthcare utilization and transitions
10,239
; however, this is the first time that the
year of diagnosis has been identified as a period of high transitions and disruption for
those with dementia.
The spike in transitions during the year of diagnosis occurs even after accounting for
end-of-life or newly moving to a LTC facility, and occurs regardless of survival time or
care location (i.e. living in the community or LTC). Transitions in this year are primarily
hospitalizations, followed by multiple moves between LTC facilities for those who reside
in one. Some of the moves between multiple LTC facilities may have been influenced by
a policy implemented in BC in April 2002 which created a shift to needs-focused
residential care access, a dimension of which included the expectation that individuals
accept the first available LTC bed in any health authority, within 48 hours of notification
240
. Individuals could subsequently move to their preferred location following processes
each health authority had in place. It is also likely that once an individual received a
diagnosis of dementia, they were moved to a different LTC facility better equipped to
meet their needs.
149
Similar increases in hospitalizations, healthcare service use and costs have been
identified within the year of diagnosis for individuals with other chronic, degenerative
diseases
234–236
. Dementia is not an acute condition nor is it effectively managed in
acute hospital settings, so why the increase in hospitalizations and extended length of
stay? Examining those who received their incident diagnosis of dementia in hospital
indicated that for the majority, dementia was not the primary reason they were
hospitalized, yet a large percentage of the cohort received an initial diagnosis of
dementia in hospital (37%). It is possible that the symptoms of dementia went
unrecognized by the primary care physician for a while, were not communicated to the
patient and caregiver or was simply not coded by the primary care physician so that it
was only eventually recorded in the hospital record. Regardless, it appears other
adverse events are what prompt the initial acute hospitalization
62,165
. Certain underlying
conditions such as stroke can increase the risk of dementia, or other conditions that
develop with the onset of dementia (difficulty swallowing, increasing the risk of
pneumonia, increased risk of a fall) can trigger a hospitalization
170,241
. This was
paralleled in the cohort I examined where a fall, pneumonia and stroke were the top
three reasons for the initial hospitalization in which dementia was subsequently
diagnosed. Lastly, the vast majority of individuals newly diagnosed with dementia are
multimorbid. The literature indicates that dementia is generally not diagnosed early so
that when diagnosis does occur, the primary symptoms of deteriorating global
cognition and verbal fluency would already be pronounced. This can significantly
inhibit the individual’s ability to self manage their chronic conditions, locate or even
communicate symptoms which could trigger acute hospitalizations and
150
rehospitalizations
170,242
. The long stays in hospital may be because individuals with
dementia take longer to recover and require more rehabilitation, particularly after the
effect of a changed environment which can be particularly distressing for individuals
with dementia
236
. But it is more likely that the long stays are also because the hospital
determines that they can no longer go home so they remain in hospital until they are
discharge to a LTC facility
240
.
Another unanswered question is why the peak in hospitalizations in the year of
diagnosis but not other years (other than end-of-life)? First, other studies have
demonstrated an association of incident dementia with increased hospitalizations
170
.
One explanation is that a large proportion of the cohort moved to a LTC facility during
the year of, and the year after diagnosis. The LTC facilities have a primary care team
that can provide appropriate care and are trained to recognize symptoms that patients
with dementia cannot communicate, decreasing the need for hospitalizations in
subsequent years. As well, receiving the diagnosis and becoming more familiar with the
disease, what to anticipate and perhaps being connected with HCC services can assist
patients, their caregivers and physicians to better manage the dementia and the effects
of dementia on co-morbidities after the year of diagnosis. For those with dementia and
their caregivers, the year of diagnosis is an overwhelming period of stress during which
major decisions about care need to be made in addition to grappling with the long-
term implications of a diagnosis. Each new transition experienced increases the risk of
medication errors, miscommunication, preventable readmissions and mortality
15,156,160,167
. This is therefore, an opportunity for a point of intervention during which
patients can be referred to appropriate services and provided support to ensure
151
continuity while minimizing the number of transitions (and in particular acute
hospitalizations) experienced. Studies that reported on the introduction of a care
transition intervention for dementia patients newly discharged from hospital or residing
in LTC facilities found sharply reduced rehospitalisation rates, correlated with lower
hospital costs
243,244
. Similar interventions need to be considered during the year of
diagnosis. In addition, providing better education to caregivers of dementia about the
trajectories of dementia, its effect on co-morbidities and how to anticipate and react to
events prior to a crisis are essential.
A second spike in transitions is noted at end-of-life. I find that regardless of survival
period post-diagnosis, people in the community with dementia start experiencing an
increase in hospitalizations the year prior to death and then a sharp spike in the number
of hospitalizations in the year of death. This is in marked contrast to those in LTC.
These results mirror findings from Finland
10
. The literature indicates poor prognosis and
limited life expectancy for people with end-stage dementia following acute illness and
aggressive medical treatment
245
. The majority of people with dementia have expressed
a wish to die at home without major medical interventions
246
and yet for those in the
community with dementia, hospitalization in the year of death continues to occur, a
venue in which decision-making is more rushed, care is fragmented and aggressive
interventions to prolong life are more readily performed
247,248
. These results indicate
that the care at end-of-life for those with dementia follows the same general patterns as
those in the general population despite recognizing that they require different care
239
.
152
The reasons for the spike in acute hospitalizations at end-of-life were not examined
but are likely multifaceted and may not be specific to individuals with dementia. Two-
thirds of all deaths in Canada occur in hospital despite most people’s wish to die at
home
91
. The proportions of individuals with dementia who die and access palliative
care is low even though such care is associated with better quality of care at end-of-
life
249,250
. Once hospitalized, individuals with dementia are more likely to be given
active, acute care treatment at end-of-life instead of palliative care and they receive
inadequate pain control
251
. This could be because, as previously highlighted, it can be
difficult for physicians to appropriately assess and treat symptoms in individuals with
dementia so that when a decline in health occurs, it triggers a crisis and hospitalization.
Individuals with dementia living in the community nearing end-of-life may have
caregivers receiving inadequate home support who are ill equipped to handle the last
stages of a person dying with dementia, particularly if they are multimorbid
252
. Many
caregivers of individuals with dementia report high levels of stress, poor health and
burnout
55,82,91
. Caregivers have also reported needing adequate, realistic information
about issues such as pain management, navigating the system, accessing respite care
and palliative care
252
. Lastly, individuals with dementia living in the community are less
likely to have advanced care orders, including do-not-resuscitate and do-not-
hospitalize
253
. Given this, it is even more important to provide patients and caregivers
better education and user-friendly information about advanced care planning and
palliative care options. Physicians need to be provided tools to begin discussions
around end-of-life early as part of comprehensive dementia care management, similar
to discussions that occur during cancer care.
153
Of note is the overall survival time post-diagnosis for patients with dementia. I find
a median of nearly 5 years survival post-diagnosis (note, this is not post-onset of
dementia as there is no accurate way to measure this). This finding corroborates existing
evidence
37,254–256
. While this may seem to be a long period with enough time for
physicians to discuss end-of-life options, given the progressive degenerative nature of
dementia, discussing end-of-life options early, even within the year of diagnosis, would
provide individuals with dementia better opportunity to make decisions about their care
before they are no longer capable of making such decisions
36,257
.
I find that higher morbidity, living in more rural health authorities, and having
behavioural symptoms associated with dementia are strongly associated with the
number of transitions experienced both in the year of diagnosis and transitions overall
in the study period, after adjusting for physician practice style. This parallels
observations from other studies about the importance of these factors
15,238
. Morbidity
plays the biggest role. Indeed, patients with dementia and a high burden of other
sources of morbidity experienced the highest number of hospitalizations overall and at
end-of-life regardless of whether they lived in a LTC facility or the community. LTC is
considered a better/more appropriate care setting for this complex population than the
community. These patients receive better medical support to manage their conditions
and such facilities should be better equipped to provide good end-of-life care
33
. Yet, I
find that the hospitalizations for people with a high burden of morbidity in LTC are
similar in frequency and pattern to those in the community. Both of these are
significantly higher than for those who only have dementia. This suggests that the spike
in hospitalizations at end-of-life is being driven by co-morbidities combined with the
154
dementia, not by the dementia alone. More needs to be done to understand what
supports need to be in place to manage and minimize the transitions experienced by
this complex population and particularly why these patients are being hospitalized while
in LTC.
Interestingly, the presence of a caregiver has a weak association with fewer moves
between LTC facilities in the year of diagnosis. This is in the direction hypothesized, and
the fact that it is not more strongly significant may reflect that the proxy variable used
as a measure of caregiver status is imperfect and likely underestimates the effect of
having access to an informal caregiver. It is likely that the lower transitions and LTC
moves associated with individuals in the highest income categories are an indication of
the use of private facilities which I lack information on.
6.4.1 Limitations
This study faced several limitations. The reasons for institutionalization are varied,
and include availability of a caregiver, difficult to manage behavioural symptoms, and
progression and severity of the dementia
246
. I was unable to separate out and measure
these factors using the administrative data available for this study. I was unable to
measure the location of death and whether patients receive palliative care while in
hospital or in LTC, though previous studies have indicated that older residents with
dementia who die in LTC or hospital generally do not receive palliative care
247,258,259
.
Despite the availability of universal, publicly funded healthcare services and LTC, there
are several private LTC facilities in the province for which there is no publicly available
155
information. In order to have a longer follow-up period, I identify patterns of diagnosis
beginning in 2001/02, however, these patterns may be different in more current years.
Finally, there is the potential for omitted variable bias given the type of data I had
available. However, the variables identified as being significantly associated with
transitions in the models have been validated through other quantitative and qualitative
studies, and as long as any omitted variables are uncorrelated with the independent
variables included, my regression analysis should still produce unbiased estimates.
6.5 Conclusion
For people with dementia for whom familiar environments and routines are
important, frequent transitions (particularly to a hospital environment) are especially
distressing and can cause further disorientation, and decline
11,12
. As well, each new
transition increases a multitude of risks. The spike in transitions during the year of
diagnosis highlights an unsettling situation affecting a frequently distressing period for
individuals with dementia and their caregivers, but also suggests a useful target for
interventions that might be effective in reducing such transitions. In my next chapter I
explore in further detail whether receiving what is considered guideline-consistent
dementia care and/or high quality primary care is associated with fewer transitions
experienced.
156
CHAPTER 7 – GUIDELINE-CONSISTENT DEMENTIA CARE AND
HIGH QUALITY PRIMARY CARE – IS THERE AN ASSOCIATION
WITH TRANSITIONS?
7.1 Introduction
The chronic nature of dementia requires care that provides accessibility, care
coordination and continuity in order to meet the ongoing biopsychosocial needs of
people with a primary diagnosis of dementia
18,19
. These are also the attributes of high
performing primary care -- there is good evidence that these result in better
management of chronic conditions, fewer preventable hospitalizations, shorter lengths
of stay, reduced use of emergency rooms and reduced risk of mortality
20–22
, However,
many of the studies that have examined primary care have focused on adults and young
children but not on elderly populations and particularly not on vulnerable elderly
populations such as those with dementia
23
.
In the previous chapter I identified the year of initial diagnosis and the year of death
as crucial time points during which individuals with dementia experience a peak in the
number of transitions. Guidelines for dementia care outline diagnostic and treatment
services that are (or should be) heavily concentrated in the first year after diagnosis.
The hypothesis is that adherence to guidelines will lead to better outcomes, but studies
that have assessed physician barriers to implementing these guidelines indicate
skepticism among healthcare providers regarding the strength of this association
41
.
Therefore, my research questions are as follows:
157
Research Question 4a: Is receipt of guideline-consistent dementia care in the year of
diagnosis associated with the number of transitions the patient experiences in that year?
Research Question 4b: Is receipt of high quality primary care associated with the
number of transitions the patient experienced in the year in which dementia was first
diagnosed?
I hypothesize that holding all else constant, receipt of guideline-consistent dementia
care, and appropriate primary care that encompasses continuity of care and
coordination will contribute to fewer transitions. I also hypothesize that receipt of
guideline-consistent dementia care will be associated with receiving high quality
primary care as measured by continuity of care.
7.2 Methods
The same cohort described in chapter 6 was used in this section of the analysis.
Here, I focused on the year of diagnosis as this was the year in which I measured receipt
of guideline-consistent care. As well, this was when the cohort experienced a high
number of transitions and when the largest proportion of the cohort moved to LTC or
died.
7.2.1 Data Sources and Study Population
The same data sources and study population as described in Chapter 6 were
utilized in this analysis. For consistency and comparison with Cohort 2 from Chapter 5,
the main cohort in this analysis excluded individuals who died during the year of
158
diagnosis and/or lived in LTC at the time of diagnosis. However, separate focused
sensitivity analyses were conducted on two groups, 1) individuals who newly moved to
LTC in the year of diagnosis (referred to as sub cohort 1 as these individuals were
included in the main cohort) and 2) individuals who died during the year of diagnosis.
7.2.2 Primary Outcome - Transitions
The same primary outcome as described in Chapter 6, Section 6.2.3 was used.
Briefly, care transitions are defined as a physical move from one location to another with
a stay of at least one night. Only transitions between the community, respite care,
Assisted Living, LTC and hospitals were included. Outpatient care or moves between in-
patient wards of the same facility were excluded; this approach was consistent with prior
literature
10,37
.
7.2.3 Measures of Guideline-Recommended Dementia Care & High Quality
Primary Care
I examined the association between receipt of 1) guideline-consistent dementia
care and 2) high quality primary care, with the number of transitions experienced.
Guideline-consistent dementia care was described in Chapter 5 (see also Sivananthan,
Lavergne, & Mcgrail, 2013) but briefly, the guidelines were derived from the most
updated BC Dementia Care Guidelines, released in July 2007, as they reflect province-
specific resource availability, were readily available to physicians practicing in BC, and
incorporate recommendations from the Canadian Consensus group national guidelines
7,122
. The care processes I expected to find in guideline-consistent care, in the first year
159
of diagnosis (2001/02), were: laboratory exclusion tests (any one of blood count, serum
electrolytes, serum calcium, serum glucose, TSH and vitamin B12); a CT scan of the
brain; prescription of an AChI, antipsychotic or benzodiazepine, trazodone or
antidepressant; a physical examination; counselling; referral to a specialist (Geriatrician,
Neurologist or Psychiatrist); and receipt of a HCC assessment.
Receipt of high quality primary care was based on a definition developed by
Starfield et al.
22. Three dimensions of primary care were measured: 1) access 2) long-
term person focused care and 3) coordinated care from an assigned primary care
physician. Patients were assigned to the primary care physician who provided the
highest proportion of unique visits during the fiscal year of initial diagnosis. In the
event of a tie, the physician who billed the largest number of fee items (essentially visits)
was assigned as the most responsible source of primary care.
The following attributes were measurable within the context of those dimensions (they
are also summarized in Table 7.1):
Access - Defined as care from the assigned primary care physician in alternate settings
(e.g. hospital) or outside regular office hours. Access was measured as three binary
variables indicating receipt (or not) of a visit from the assigned primary care physician
outside office hours or while the patient was in hospital or, for the sub cohort, while in
LTC. Receipt of house calls from the assigned physician was extremely low and
therefore excluded as a measure.
160
Long-term person-focused care –Defined as providing longitudinal care to patients
across the spectrum of their needs. I was unable to operationalize ‘person-focused’ care
with administrative data, however, continuity is an essential requisite to person-focused
care since it is impossible to understand the values and context of a patient without
following them over time. Therefore provision of continuity of care serves as a
reasonable marker for ‘person-focused’. This was measured as the number of visits with
assigned primary care physician, divided by total number of visits patient had that
year
261
. This measure of continuity of care was chosen because it was easier to interpret
and operationalize.
Coordinated care – Defined as the coordination of care across care settings, including
referrals to specialists and for laboratory testing. This was measured by the proportion
of all referrals to specialists and for laboratory testing originating from the assigned
primary care physician.
161
Table 7.1 – Description of primary care dimensions measured
PRIMARY CARE DIMENSIONS
ACCESS
Received hospital visit from
assigned physician
Flag of billings in hospital by assigned physician
while patient was in hospital
patient level
Received LTC visit from
assigned physician
Flag for billings in LTC by assigned physician
while patient was in LTC
patient level
Received visit from assigned
physician outside office
hours
Flag for billings indicating services were
provided outside of office hours by assigned
physician
patient level
LONG-TERM PERSON FOCUSED CARE
Continuity of care # visit with assigned physician divided by total
number of visits patient had in the year
patient level
COORDINATED CARE
Proportion of specialist
referrals originating from
assigned physician
# outpatient specialist referrals originating from
assigned physician divided by total number of
specialist referrals in the year
patient level
Proportion of laboratory
referrals originating from
assigned physician
# outpatient laboratory referrals originating
from assigned physician divided by total
number of laboratory referrals in the year
patient level
7.2.4 Controlled Explanatory Variables
All explanatory variables described in Chapter 6, Section 6.2.4 were used for
adjusted for. Briefly, baseline patient characteristics included age, sex, income quintile
(measured at the neighbourhood level), health status (measured by number of major
ADGs as previously described), caregiver proxy status (identified using a shared MSP id
as previously described in Section 4.2.1), health authority, and receiving a dementia
diagnosis in hospital (versus the community). These are described further in section
7.2.5. Physician practice characteristics adjusted for included the number of years the
physician had been practicing, the number of patients over the age of 65 in the practice
162
and the number of dementia patients in the practice. These were based on the
physician providing plurality of care, described in detail in Section 4.4.1.2.
7.2.5 Statistical Analysis
The proportion of the cohort to receive guideline-consistent dementia care and
high quality primary care were described using counts and percentages for the main
cohort and sub-cohort. The association between location of the incident dementia
diagnosis (hospital versus community) and receipt of dementia care was examined via
logistic regression to determine whether the location was a variable of importance to be
included in the final model.
The data were analyzed using negative binomial regression for three models: the
association between 1) receipt of guideline-consistent dementia care and transitions 2)
receipt of high quality primary care and transitions 3) receipt of both guideline-
consistent dementia care and high quality primary care and transitions. Model selection
and fit followed the same process as described in Chapter 6, Section 6.2.5. When both
guideline-consistent care and high quality primary care were examined in a model, I
tested for collinearity in the measures related to laboratory and specialist referrals.
Chapter 6 showed that the highest number of deaths and moves to LTC was in
the year of diagnosis, a period during which transitions may be modifiable by primary
care. Therefore sensitivity analyses were conducted to examine the relationship
between receipt of high quality primary care and transitions for individuals who newly
moved to a LTC facility (sub-analysis 1, conducted on a sub cohort) in the year of
163
diagnosis or those who died and were excluded from the main cohort (sub-analysis 2)
(summarized in Table 7.2). For sensitivity, these analyses were also conducted in the
year following diagnosis (2002/03) to see if the same trends were observed. Analyses
were performed using STATA, version 10
227
.
Table 7.2 – Summary of sub-analyses and sub-cohorts
Analytic Model Cohort description
Model A Rate of Transition= β0 + β(age) +β(sex) +β(income)
+β(health status) + β(health authority) + β(caregiver) + β(dx
location) + β(GP yrs of practice) + β(# patients 65+) + β(#
dementia patients) + β (dementia care)
Main cohort
Model B Rate of Transition= β0 + β(age) +β(sex) +β(income)
+β(health status) + β(health authority) + β(caregiver) + β(dx
location) + β(GP yrs of practice) + β(# patients 65+) + β(#
dementia patients) + β (primary care)
Main cohort
Model C Rate of Transition= β0 + β(age) +β(sex) +β(income)
+β(health status) + β(health authority) + β(caregiver) + β(dx
location) + β(GP yrs of practice) + β(# patients 65+) + β(#
dementia patients) + β (dementia care) + β (primary care)
Main cohort
Model D (sub-
analysis 1)
Rate of Transition= β0 + β(age) +β(sex) +β(income)
+β(health status) + β(health authority) + β(caregiver) + β(dx
location) + β(GP yrs of practice) + β(# patients 65+) + β(#
dementia patients) + β (primary care)
Sub-cohort of
individuals who newly
move to LTC
Model E (sub-
analysis 2)
Rate of Transition= β0 + β(age) +β(sex) +β(income)
+β(health status) + β(health authority) + β(caregiver) + β(dx
location) + β(GP yrs of practice) + β(# patients 65+) + β(#
dementia patients) + β (primary care)
Individuals who died
(excluded from main
cohort)
7.3 Results
7.3.1 Descriptive Analyses
After exclusion for death in the year of diagnosis (n=1,231) and living in a LTC
facility at the beginning of the year (n=771), 4,711 individuals newly diagnosed with
dementia remained in the main cohort. The main cohort experienced a median of 1
164
transition in the year of diagnosis, while 27% of those who experienced a transition had
3 or more transitions. Almost 60% of individuals in the main cohort experienced at least
1 hospitalization, half of whom (53%) were diagnosed with dementia in hospital (Table
7.3). In comparison, the sub-cohort (those who newly moved to LTC) had a median of 2
transitions (including their move to LTC), almost 50% had 3 or more transitions, and
80% experienced at least 1 hospitalization (Table 7.2). Individuals who died in the year
of diagnosis also had a median of 2 transitions in the year of diagnosis, 30%
experienced 3 or more transitions and 97%of that cohort experienced at least 1
hospitalization.
7.3.2 Receipt of Guideline-Consistent Dementia Care and High Quality Primary
Care Process
For the main cohort, in the year of diagnosis, a high percentage of the cohort
received laboratory tests (90%) and half received imaging (50%) consistent with
guideline dementia care processes (Table 7.2). One-third of the cohort was prescribed
an AChI, while between one-third and one half of the cohort received an antidepressant
(46%), antipsychotic (39%) or benzodiazepine (37%), used for behavioural symptoms of
dementia. However, an extremely low percentage of the main cohort received any
discretionary dementia care processes such as counselling (5%), physical examination
(4%) or a referral to a specialist (15%). Just over 60% received a HCC assessment, which
would determine eligibility for home care or LTC services; referral for this assessment
does not necessarily get initiated by a physician, so the proportion referred by
physicians would undoubtedly be lower than 60%.
165
Primary care was measured through three processes (Table 7.3). Access to a
physician in alternate settings demonstrates that individuals in the main cohort received
care, on average, in 2 settings. A high proportion of the main cohort that experienced a
hospitalization in the year of diagnosis received a visit from their physician (72%) and
58% of the cohort that newly moved to a LTC facility received a visit from their physician
(Table 7.3). Long-term person-focused, or continuity of care, was high in the cohort in
the year of diagnosis, with on average 75% of visits being with the assigned physician.
Broken down further, 16% of the cohort had all visits with the same physician while 21%
of the cohort had 50% or less of their visits with the assigned physician. With respect
to coordinated care, 59% and 63% of patients respectively, received all their specialist
and all their laboratory referrals from the same physician.
With regard to the three dimensions of high quality primary care for individuals
who newly moved to LTC during the year of diagnosis, individuals in that sub-cohort
received care in a median of 3 settings, had lower continuity of care (median 67% of
physician visits with the ‘assigned’ primary care physician) but otherwise did not appear
to differ significantly from the main cohort. A higher proportion of these individuals
(73%) received all their specialist referrals from the assigned physician (Table 7.3).
Individuals who died in the year of diagnosis (and were excluded from the main cohort)
also received care in a median of 3 alternate settings and a very high percentage
received a visit from the assigned physician when in hospital or LTC (84% and 80%
respectively) (Table 7.3).
166
Table 7.3 –Dementia care and primary care use in year of diagnosis
Receipt of Services Main cohort
(n=4711)
% of cohort Sub cohort
(newly moved to
LTC) (n=1326)
People who died
†
(n=865)
Number of transitions experienced (median) 1 transition 2 transitions 2 transitions
Experienced at least 1 transition 2940 62.4% 79.3%* 96.3%
Experienced 3+ transitions 791 26.9% 48.7% 30.0%
Experienced at least one hospitalization 2699 57.3% 79.3% 96.3%
Experienced 2+ hospitalizations 1076 39.9% 38.1% 43.5%
DEMENTIA CARE
Laboratory testing 4217 89.5%
CT scan 2386 50.6%
Prescription for AChI 1529 32.5%
Prescription for antipsychotics 1828 38.8%
Prescription for benzodiazepines 1736 36.8%
Prescription for trazodone 326 6.9%
Prescription for antidepressants 2146 45.6%
Counselling 250 5.3%
Physical examination 182 3.9%
Referrals for dementia 714 15.2%
Home & community care assessment 2907 61.7%
PRIMARY CARE
Access
# alternate care settings (median) 2.0 3.0 3.0
Received hospital visit from assigned
physician (for those hospitalized) 1943 72.0% 72.5% 84.1%
Received LTC visit from physician assigned
physician (for those newly in LTC) 769 58.0% 58.0% 80.3%
Long Term Person-Focused Care
Proportion of all visits from assigned
physician (median) 75.1% 66.8% 67.3%
Received 100% of care from assigned
physician 767 16.3% 9.4% 23.9%
Received 50% or less of care from assigned
physician 245 21.2% 31.5% 26.9%
Coordinated Care
Received 100% of specialist referrals from
assigned physician 1226 59.0% 72.6% 75.6%
Received 100% of laboratory referrals
originating from assigned physician 2406 62.7% 61.7% 75.3%
*NOTE: All individuals moved to LTC which was counted as a transition, therefore this is a measure of an additional transition
†NOTE: These are all individuals who died and were excluded from the main cohort
Individuals who received a diagnosis of dementia in the community had a much
higher likelihood of receiving laboratory testing, acetyl-cholinesterase inhibitors,
counselling, a physical examination and referral to a specialist for dementia, all
167
processes associated with guideline-consistent dementia care (Table 7.4). They were
also less likely to receive a prescription for an antipsychotic or benzodiazepine.
Table 7.4 – Odds of receiving guideline-consistent care based on receiving a diagnosis in
the community (relative to receiving an initial diagnosis in hospital)
Dementia Care Odds CI
Laboratory testing 2.4 1.9-2.9
CT scan 0.9 0.8-1.1
Acetyl cholinesterase 1.9 1.7-2.2‡
Antipsychotics 0.7 0.4-0.9‡
Benzodiazepines 0.8 0.7-0.9‡
Trazodone 0.9 0.7-1.2
Antidepressants 1.1 1.0-1.3
Counselling 3.0 2.1-4.2‡
Physical examination 2.4 1.6-3.4‡
Referrals for dementia 1.5 1.3-1.9‡
HCC assessment 0.3 0.3-0.4‡
‡ - P value <0.001, † - P value <0.01,* - P value <0.05
Controlling for age, sex, income, health authority, health status
7.3.3 Regression Analysis
Preliminary analysis indicated that variables used to measure the primary care
dimension of access appeared to have occurred mainly in emergency situations.
Receipt of care outside of office hours, while in hospital and while in LTC were all
significantly associated with a higher number of transitions. If at all, it would have been
anticipated that receipt of these services would be associated with fewer transitions
unless they were provided in emergency situations during which patients are already
experiencing multiple transitions. The use of a physician-level variable indicating
provision of care outside office hours was tested for sensitivity (as it was hypothesized a
physician-level variable that measures access would be a better indication of a
physician’s willingness to provide care outside office hours, without necessarily being a
proxy for care in emergency situations) and was not significant. The measure of access
168
as a dimension of high quality primary care was therefore excluded from subsequent
regression analysis.
In multivariate analysis, after controlling for patient and practice characteristics
and for receiving a diagnosis of dementia in hospital; receiving guideline-consistent
laboratory testing, an acetyl-cholinesterase inhibitor, counselling, a physical
examination and a referral to a specialist for dementia were associated with fewer
transitions (Table 7.5, model A). However, receiving a prescription for an antipsychotic
or benzodiazepine, and a HCC referral were associated with a higher number of
transitions.
Similarly, controlling for patient and practice characteristics, higher continuity of
care was highly associated with fewer transitions in the year of diagnosis. Receiving a
higher proportion of specialist referrals from the assigned physician was also associated
with fewer transitions (Table 7.5, Model B). When receipt of high quality primary care
and guideline-consistent dementia care were both assessed in one model, the same
processes of care that were independently associated with transitions remained
significant, with only the continuity of care measure dropping slightly in the strength of
its association (Table 7.5, Model C). Due to collinearity between receipt of laboratory
tests/ specialist referrals in guideline-consistent dementia care and coordination of
referrals for laboratory tests/specialists in high quality primary care, the coordination
variables used to measure high quality primary care were dropped in this model as they
were more indirect measures.
169
The sensitivity analysis that focused on individuals who newly moved to LTC in
the year of diagnosis demonstrated that higher continuity of care is associated with a
fewer number of transitions, but receiving a higher proportion of laboratory referrals
from the assigned physician was associated with a higher number of transitions (Table
7.5, Model D). This same association was seen in the sub-analysis on individuals who
died in the year of diagnosis (Table 7.5, Model E).
170
Table 7.5 – Multivariate regression analysis examining association of dementia care and primary care with transitions
experienced in the year of diagnosis
Transitions
(Model A)
Transitions
(Model B)
Transitions
(Model C)
Transitions -
sub analysis 1
(Model D)
Transitions -
sub analysis 2
(Model E)
PRIMARY CARE VARIABLES
Continuity of care 0.4 (0.4-0.5)‡ 0.5 (0.4-0.6)‡ 0.6 (0.5-0.8)‡ 0.6 (0.4-0.7)‡
Coordination of specialist referrals 0.9 (0.8-0.9)‡ - 1.1 (1.0-1.2) 1.0 (0.9-1.2)
Coordination of lab referrals 1.0 (0.9-1.1) - 1.1 (1.0-1.2)* 1.1 (1.0-1.3)*
DEMENTIA CARE VARIABLES
Lab 0.9 (0.8-1.0)† 0.9 (0.8-1.0)*
CT scan 1.1 (1.0-1.1)* 1.1 (1.0-1.1)
AChI 0.7 (0.6-0.7)‡ 0.7 (0.6-0.7)‡
Antipsychotics 1.2 (1.1-1.3)‡ 1.2 (1.1-1.2)‡
Benzodiazepines 1.2 (1.2-1.3)‡ 1.2 (1.2-1.3)‡
Trazodone 1.1 (1.0-1.2) 1.1 (1.0-1.2)
Antidepressants 1.1 (1.0-1.1)* 1.1 (1.0-1.1)
Counselling 0.8 (0.7-0.9)‡ 0.8 (0.7-0.9)†
Physical examination 0.7 (0.6-0.9)‡ 0.7 (0.6-0.7)‡
Referrals for dementia 0.9 (0.8-1.0)† 0.9 (0.8-1.0)†
HCC assessment 2.2 (2.0-2.4)‡ 2.1 (2.0-2.3)‡
Sub analysis 1: sub-cohort of individuals who newly moved to LTC; Sub analysis 2: individuals who died
‡ - P value <0.001, † - P value <0.01,* - P value <0.05
Controlling for age, sex, income, health authority, health status, behavioural symptoms, receipt of diagnosis in hospital, physician practice characteristics
171
7.4 Discussion
Receipt of guideline-consistent dementia care and high quality primary care, are
independently associated with fewer transitions across care settings for patients newly
diagnosed with dementia in the year of diagnosis. While, in general, guidelines
introduced and followed lead to more uniformity
30
and concordance with best practice
guidelines has been shown to be associated with better outcomes
24–26
, previous studies
have consistently demonstrated a lack of concordance between best-practice
recommendations for dementia care and actual practice, leading to under-diagnosis
and poor management
8,28,29,119,262,263
. Barriers to physician adoption of dementia
guidelines include a myriad of issues such as a perceived lack of evidence for better
outcomes
41
. These data demonstrate, for the first time, that receipt of some guideline-
consistent dementia care, independent of continuity of primary care, are associated with
fewer care transitions. This is important in this population because transitions are
associated with poorer outcomes.
Fewer transitions are the result of fewer hospitalizations and fewer physical
moves between LTC facilities or Assisted Living facilities during the year of diagnosis.
Given the peak in transitions, and particularly the high number of hospitalizations
during the year of diagnosis (as demonstrated in Chapter 6), the finding here
demonstrates that guideline-consistent dementia care is associated with fewer
transitions during this chaotic period. This is not only consistent with my hypothesis,
but reinforces the importance of finding ways to improve the uptake of care guidelines
for this population. This is particularly true for receipt of discretionary dementia care
management processes such as counselling and a referral to a specialist for dementia.
172
It is important to acknowledge that I am not able to show a causal relationship with this
analysis; however my hypothesis was based on literature which has demonstrated that,
for example, counselling better prepares patients and their caregivers in terms of what
to expect given a diagnosis and communicates techniques to handle symptoms,
potentially reducing the need for hospitalization
53,127,264
Notably, receiving a prescription for an antipsychotic or benzodiazepine is
associated with a higher number of transitions. I was unable to determine whether
individuals who were prescribed these medications had a clinical indication for their
receipt, though the literature indicates low lifetime prevalence of most psychiatric
illnesses that would have required a prescription of this type
265,266
. My results are
consistent with the literature which indicates that 1) individuals with behavioural
symptoms of dementia experience a higher number of transitions across care facilities
238,246
and 2) receipt of prescriptions for antipsychotics or benzodiazepines is associated
with a higher risk of adverse events
232
. Of note, accelerated cognitive and memory
decline are predictors of hallucinations and psychosis
267
. Given that I was unable to
measure functional status or severity of dementia, it is possible that the acuity of
dementia at the point of diagnosis was higher in individuals receiving these
prescriptions for behavioural symptoms, and that it is this, rather than the prescriptions,
that was contributing to a higher number of transitions. As well, cognitive, functional
and behavioural changes are considered primary stressors leading to institutionalization
246
, and I was unable to capture these within the data available to me.
Receiving a HCC referral was also associated with a higher number of transitions.
This may be because receiving the referral means individuals are more likely to be
173
assessed, which would allow them access to LTC and contribute to a transition.
Interestingly, individuals diagnosed with dementia in the community were less likely to
receive a HCC referral than those newly diagnosed in hospital. While these referrals do
not need to originate from a physician, it is possible that 1) primary care physicians
practicing in the community are unaware of the referral process and not referring their
patients, 2) patients who are diagnosed in hospital have a poorer health status/severe
dementia and are referred, or 3) the hospitalization itself brings patients to the
attention of the system and they are therefore more likely to be referred.
Similar to guideline-consistent dementia care, I found receipt of high quality
primary care, measured through higher continuity and coordination of care, to be
associated with fewer transitions. Given that these are attributes of a high performing
primary care system, with good evidence for better outcomes of care, the results are not
surprising. However this is a contribution to the continuity of care literature, because
that literature lacks studies on elderly vulnerable populations such as those with
dementia. Further, this strong association with higher continuity exists at the end-of-
life and for individuals newly moving to a LTC facility even after controlling for
morbidity and other relevant factors. However, higher coordination of laboratory
referrals was associated with more transitions. This may be because at both those care
points (moving to LTC and end-of-life) receiving higher coordination of laboratory
referrals, similar to the measure of access, is provided under emergency circumstances.
Newly moving to a LTC facility is a major life event for an elderly person with
dementia; having high continuity of care would ensure that the individual’s care needs
are recognized and met and multiple medications are reconciled early in the transition.
174
Similarly, as demonstrated in Chapter 6, end-of-life is a period during which individuals
with dementia experience a second peak in transitions. This can be due to a variety of
factors. Often, individuals with dementia are multi-morbid and these other conditions
are exacerbated (and masked) by the dementia. As well, pain management is difficult to
ascertain and control in individuals with dementia
268,269
which can trigger
hospitalizations during which more extreme, life-prolonging interventions are taken.
These generally result in poor outcomes for elderly patients with dementia, particularly
at end-of-life
245
. Having high continuity of care at end-of-life would allow the
physician who is aware of the patient’s history, medications and perhaps end-of-life
wishes to: assess and manage symptoms to ensure early interventions are provided
before symptoms worsen; communicate with the caregiver and provide support as to
what to expect; and ensure the wishes of the patient are followed at end-of-life (if they
had been previously documented). They would therefore be providing informational,
management and relational continuity during essential periods of care
40
.
I hypothesized that higher continuity of care is the pathway through which better
guideline-consistent dementia care is provided. Yet my analysis demonstrates that the
two appear to be independent effects. The strength of the association between
guideline-consistent dementia care and transitions remained the same even after the
continuity of care measure was introduced; it was in fact the strength of the continuity
of care measure that dropped (though only slightly). This indicates that 1) receiving
guideline-consistent dementia care alone is important and 2) providing better
continuity is not the mechanism by which guideline-consistent care appears to occur;
the effects are separable. This goes counter to my hypothesis, but is not unrealistic for
actual practice. Many reasons have been identified for why physicians do not provide
175
guideline-consistent dementia care, but this does not mean that the physician is not
providing good continuity of care. Continuity would allow the physician to still monitor
the patient, notice any decline or worsening of symptoms and potentially also monitor
the health of the caregiver, which are predictors of institutionalization and
hospitalizations.
Finally, receiving a diagnosis of dementia in hospital is not optimal. People who
received their diagnosis in hospital had a higher number of transitions, and were less
likely to receive guideline-consistent dementia care. Those who received their diagnosis
of dementia in hospital also had higher odds of receiving a prescription for an
antipsychotic or benzodiazepine, both of which are used to control the behavioural
symptoms of dementia. It was, however, not possible to determine the mechanism
through which this occurs, for example which came first, being more likely to receive
these prescriptions because these individuals were hospitalized, or being hospitalized
because of symptoms. As previously discussed, it is possible that individuals who
received prescriptions for antipsychotics or benzodiazepines had a higher acuity of
dementia and are therefore more likely to be hospitalized. Regardless, trends over time
(comparing between Cohort 1 and Cohort 2, Figure 5.2 and Figure 6.1 respectively)
indicate a shift toward diagnosis in the community.
7.4.1 Limitations
There are several limitations to the analysis reported in this Chapter. First, these
results demonstrate association, not causality. My model cannot establish,
unequivocally, an order of cause and effect. For example, high transitions and
176
disruptions may make it difficult for physicians to provide guideline-consistent
dementia care. Indeed, the transitions themselves may prompt investigation of cause
that leads to a dementia diagnosis. This would be consistent with the high proportion
of patients receiving a diagnosis of dementia in hospital. Other studies will need to
study causality more directly, but if implementing better guideline-consistent dementia
care results in fewer transitions, it could be a significant cost saving and life-improving,
strategy. A potential method to estimate causality could be the use of an instrumental
variable to control for unobserved confounding and measurement error, thereby
allowing for the possibility of causal inferences from observational data. The key would
be determining a good instrument for this purpose. Given this, there is the potential for
omitted variable bias due to the type of data I had available. However, the variables
identified as being significant in the models have been validated through other
quantitative and qualitative studies and as long as any omitted variables are
uncorrelated with the independent variables included, my regression analysis should
still produce unbiased estimates
I was unable to determine if individuals who were prescribed antipsychotics or
benzodiazepines had a clinical indication for their prescriptions; however the prevalence
of schizophrenia (0.87% lifetime prevalence), bipolar I disorder (0.35% lifetime
prevalence), major depressive disorder with psychotic features (0.35%), general
psychosis (1.7%) and neurotic disorders (2.4%) is low
265,266
. While there may be clinical
indications for prescriptions of antipsychotics other than the behavioural/psychosis
symptoms of dementia, I would not expect those indications to explain all of the
prescriptions. I was unable to measure the severity of dementia at diagnosis given the
177
nature of the administrative data to which I had access. This, obviously, could influence
the number and types of transitions experienced by individuals.
7.5 Conclusion
The results of Chapter 6 showed that the year of diagnosis is critical. The
analyses here are consistent with, and suggestive of, the importance of increasing
uptake of guideline-consistent dementia care. Previous studies have indicated that
physicians perceive a lack of evidence demonstrating the value of implementing
guideline-consistent dementia care
41
. This study now contributes to that body of
literature, demonstrating a clear association between guideline-consistent dementia
care and outcomes. As well, these results provide evidence for the continued value of
high quality primary care in a complex population during key points in time of a
patient’s care trajectory when gaps in continuity usually occur.
178
CHAPTER 86,7 - CONCLUSION
8.1 Summary and Key Contributions
This thesis had four main objectives:
1) Conduct a critical assessment examining the existing research literature on physician
practice patterns associated with the care of people with dementia and to what extent
those practice patterns are consistent with published guidelines
2) Examine population-based variations in receipt of guideline-consistent dementia care
and patient factors that are associated with that care
3) Examine transitions experienced longitudinally to establish patterns that can help
identify points of care when transitions are highest and the factors that contribute to
those transitions
4) Assess the association between receipt of guideline-consistent dementia care and/or
high quality primary care and the number of transitions experienced by individuals with
dementia
A systematic review of the literature demonstrated that there is wide variation in
self-reported physician practice patterns with regard to provision of guideline-
consistent dementia care. The review also highlighted significant limitations of existing
research in that all the studies that met the inclusion criteria were self-reported survey
6
A version of chapter 8 has been accepted. Sivananthan SN. Short End of the Stick? Health Care Inequity in a
Vulnerable Population of Individuals with Dementia. Canadian Geriatrics Society Journal of CME 2014
7
Copyright (2014) AndrewJohn. Used with permission from publisher(Andrew John Publishing)
179
designs with varying sample strategies and small sample sizes. This led to my second
objective which was to use population-based administrative data, organized at a
patient-level, to address the study limitations seen in the systematic review, but also to
examine potentially modifiable patient-related factors associated with any variation as a
way to differentiate warranted and unwarranted variation. My results corroborated what
was seen in the systematic review, with wide variation in receipt of guideline-consistent
dementia care processes. My research also showed that this variation was associated
with income, geography, age and sex of dementia patients, indicating potential
inequities in care. The next step in my analysis was to examine if there is an association
between receiving these dementia care processes and my primary health outcome,
transitions between living/care locations. Transitions were chosen as an outcome
variable of particular interest for this population because the literature demonstrates
that dementia patients fare better with minimal physical disruption and changes to
routine. A greater number of transitions between care settings poses a serious
challenge to the continuity of care and the safety of the patient and is associated with
medication errors, preventable hospital readmissions and increased mortality risk
13–17
.
Existing literature lacks in-depth descriptions of longitudinal patterns of transitions
experienced by individuals newly diagnosed with dementia. My research illustrated that
over a ten-year follow-up of a cohort newly diagnosed with dementia, transitions are
highest during the year of diagnosis and at end-of-life. In both cases, the primary driver
is hospitalizations. Finally, I found that, holding all else constant, receipt of guideline-
consistent dementia care is associated with fewer transitions in the year of diagnosis,
independent of receipt of high quality primary care.
180
These findings are significant in that they demonstrate the lack of concordance
between guideline-recommended care and actual practice using data and a method
adapted to address the biases that limited the generalizability of previous studies. From
a policy-guidance perspective, these findings also highlight patterns of inequity in
receipt of dementia care unrelated to need, while demonstrating the value of providing
guideline-consistent dementia care in the year of diagnosis during which patients
experience the highest number of transitions and flux in care. This is particularly timely
as Canada, a member of the G8 Summit on Dementia in 2013
270
, recently increased its
research investment in the prevention, diagnosis and care of dementia
271
and co-
hosted a Global Dementia Legacy event to explore the value of academia-industry
collaborations in September 2014
272
. BC specifically, was one of the first provinces to
recognize the need for a provincial dementia strategy which the government released in
the form the Dementia Action Plan in November 2012
273
. Advances based on the three
priorities identified 1) support prevention and early intervention, 2) ensure quality
person-centred dementia care and 3) strengthen system capacity and accountability is
still ongoing. The Action Plan outlines high-level strategic goals to address gaps in care
which are supported by the evidence in this thesis.
This concluding chapter is organized into three sections. First I provide a brief
summary of the primary findings of the research chapters (Chapters 3, 5-7), their
contribution to knowledge in the field contextualized by our current understanding of
dementia care, and potential policy implications that flow from them. Second I
comment on the strengths and limitations of the overall thesis, extending beyond the
specific limitations already addressed in each research chapter. Finally, I discuss future
strategies that can build on the evidence presented here.
181
8.2 Research Findings
8.2.1 Variations in Guideline-Consistent Dementia Care – Lesson Learned from a
Systematic Review
Guidelines are created based on existing evidence and are meant to indicate
approaches to practice that are likely to produce the best results for individual patients.
They are published to encourage specific processes of care and to create more
uniformity in practice with the expectation that this will improve patient care and
outcomes
30
. Guidelines are particularly important in care for dementia patients
because of the evolving nature of our understanding of optimal dementia care and
because assessment and subsequent treatment is provided by primary care physicians
who often have limited knowledge of changes to guidelines and sometimes limited
experience with dementia patients
6,7,41
.
In Chapter 3, I systematically review the existing literature on physician practice
patterns associated with the care of people with dementia, through a lens of
consistency with published guidelines. My results demonstrate wide variations in the
proportion of physicians who report providing guideline-consistent dementia care. For
example, I saw a large difference in the percentage of physicians who report providing a
formal memory test, fundamental for an accurate diagnosis of dementia. Similar
variations were seen in the use of guideline-recommended therapeutics, referrals to
specialists and referrals to community services, the first two of which assist in managing
the more difficult symptoms of dementia. There was less variation in other services
182
assessed, and encouragingly, exclusionary blood work and provision of counselling are
generally widely used.
The large variations in the dementia care processes surfaced by this systematic
review are eye-opening as they are an indication that patients are not receiving
appropriate, equitable or in some cases adequate care despite widely available
guidelines. This in turn can have significant consequences for the accuracy of their
diagnosis, access to resources, management and their future outcomes of care. This is
the first systematic review to summarize and examine the variations reported in the
literature. These results also have important implications as they corroborate the
qualitative literature around barriers to providing good dementia care which include
inadequate time to fully explore issues during appointments, challenging behavioural
symptoms, problematic physician attitudes toward providing care to individuals with
dementia resulting in age discrimination, lack of interdisciplinary teams to appropriately
address the multifaceted psychosocial issues and difficulty accessing community
resources resulting in patterns of inequity
41,70,71,165,269
.
While these results add to our understanding of actual physician dementia
practice patterns, and highlight the gap between actual practice and guideline-
recommended care, there are some reasons for caution. There are several factors
unrelated to physician practice decisions or styles that could contribute to the variations
reported in this literature. These might include, for example, geography, patient
demographic characteristics and changes in guideline recommendations over time. The
heterogeneity of study populations and settings made it impossible to assess these
factors in the review.
183
8.2.2 Persistent Variation – Getting Past Self-Report using Administrative Data
As a counter to some of the limitations previously highlighted in the systematic
review, in Chapter 4, I conducted an analysis of the provision of dementia care utilizing
population-based health service administrative data from BC.
My population-based study of community-dwelling seniors demonstrated that a
high percentage of B.C. dementia patients had been receiving guideline-consistent
diagnostic care processes. This finding corroborated trends surfaced in my systematic
review. However, on the treatment side, use of pharmacological interventions such as
anti-psychotics (typical and atypical) and benzodiazepines, which are generally not
recommended in this population (in which such prescribing is often contra-indicated),
was frequent. The overall rates for discretionary dementia care management processes
(specialist referrals, counselling and in-office examination) were found to be generally
low. More importantly, I found that variations in dementia care management processes
were influenced by income, age and geography.
Limiting the analyses to community-dwelling seniors makes the conclusions
particularly compelling from a policy perspective, as many seniors have expressed a
wish to remain in the community longer. Physician practice patterns are very different
in LTC, where rates of pharmacological intervention tend to be higher since many of the
patients have complex neuropsychiatric symptoms that prevent them from living in the
community safely and there is a larger team providing primary care services
114
. The
frequent use of antipsychotics and benzodiazepines was not anticipated for community-
dwelling seniors; this has serious policy and health consequences due to an association
184
with falls, increased risk of cerebrovascular adverse events and, worse, increased risk of
mortality
232,233,274,275
.
Variations in the guideline-consistent dementia care management processes
were influenced by income and age. Canada functions under a universal, first-dollar
coverage of hospital and physician services which means there are no explicit financial
barriers to access those components of care. The variation related to age may be due
to the possibility of age discrimination in practitioners’ attitudes toward providing early
diagnosis and broader interventions to older individuals. This has been described
previously as a barrier to provision of good dementia care, with physicians’ negative
perceptions in diagnosing dementia being associated with therapeutic nihilism, stigma
and ageism
41
. Other possible reasons for age-associated differences in care processes
is the lack of time that physicians have to appropriately support older individuals with
more complex symptoms, and age-related difficulty accessing specialists with limited
availability
71
. However, insofar as income and age are unrelated to patient need for
service, my results indicate potential inequity and require further attention.
8.2.3 Transitions – The Year of Diagnosis and Disruption
The prevalence of multiple moves, or transitions, is high in the elderly
33,34
and
poses a serious challenge to the continuum of care and the safety of the patient as
demonstrated by medication errors, quality deficiencies, preventable hospital
readmissions and increased mortality risk
13–17
. Three previous studies have assessed
transitions longitudinally, ranging from eighteen months
12
to seven years
37
. However,
no previous literature has tracked patients from the year of diagnosis and followed
185
them over such an extensive period (ten years), nor have any of the longitudinal studies
examined the factors that contribute to transitions.
My contributions to the literature are three. First, I demonstrate that the year of
diagnosis and the year of death are the points in care when transitions are highest. This
is important information for clinical and policy decision-makers. End-of-life literature
long ago identified the year of death (and often the year prior to death) as a period of
high transitions with fragmented care and rushed decision-making often leading to
overly aggressive interventions to prolong life
10,239,248,269
. Understanding the end-of-life
care trajectory and attempting to put in place better quality indicators and measures to
ensure individuals’ wishes are respected is still a work in progress. What is new here is
that my study also identifies the year of diagnosis as an important point in the care
trajectory, with the high number of transitions representing a second (and of course
earlier) potential point of intervention for decision-makers.
Second, the transitions occurring during the year of diagnosis were dominated
by hospitalizations, though moves from community to LTC are also highest in this year.
Primary care physicians control many of the services provided in the year of diagnosis
and, given the right resources, can connect patients with appropriate support services
early to reduce the number of transitions. Increased home support, better integration
of primary and community care and expanding community-based palliative care are
potential strategies described in more detail later. Particularly in the year of diagnosis,
higher morbidity, living in health authorities with smaller urban centres and having
behavioural symptoms of dementia are strongly associated with the number of
186
transitions experienced. These are factors that allow the identification and targeting of
patients at high risk of multiple transitions, for early intervention.
Finally, the focus of most guideline-consistent dementia care processes is during
the year of diagnosis. As the results from chapter 7 demonstrate, receiving guideline-
consistent dementia care is associated with fewer transitions. This highlights the
importance of finding ways to support physicians in providing better guideline-
consistent care that can mitigate transitions during the year of initial diagnosis.
8.2.4 The Effectiveness of Guideline-Consistent Dementia Care
Given the importance of the year of diagnosis in a dementia patient’s life,
guidelines play a key role as their emphasis is on accurate diagnosis and management.
The diagnostic processes obviously cluster in the year of initial diagnosis. But to a
significant extent, decisions and patterns related to management are also set in motion
in that year. While guidelines have been shown to be associated with better outcomes
for patients
24–26
, studies have also demonstrated hesitancy on the part of physicians in
implementing these guides
41
and a consistent lack of concordance with best practice
recommendations leading to under-diagnosis and poor management
8,28,29,119,262,263
. My
work in the previous chapter adds further evidence of the value of following guideline-
consistent care. My results also confirm that continuity of care is highly associated with
fewer transitions, regardless of whether it is the year of diagnosis, death, or move to
LTC.
While these results demonstrate association not causality, there is a hypothesized
causal pathway. Guideline-consistent dementia care encourages early diagnosis (with
187
the use of diagnostic tools such as lab tests and imaging), which would allow for early
management of dementia. Early and appropriate management both through guidelines
and high quality primary care would allow physicians to educate patients and their
caregivers on what to expect before the dementia has progressed too far in severity.
They can then monitor that progress, adjusting management of dementia and any co-
morbidities before patients reach crisis, which is what triggers multiple hospitalizations.
For example, some recommended management process from dementia guidelines
include: reconciliation of prescriptions which can prevent medication-error induced
hospitalizations; appropriate counselling which helps patients and caregivers
understand the prognosis of dementia, anticipate further decline, the effects of
dementia on self-management of other chronic diseases and develop strategies to deal
with these issues before they become acute; referrals to specialists who can provide
additional monitoring and specialized care in difficult cases so that patients do not seek
that care in emergency rooms; and referrals to community services that can assist
patients and caregivers to manage care before caregivers reach a point of crisis.
Continuity of care becomes particularly important in being able to monitor the patient,
knowing their medical history and therefore being able to address issues before it
reaches the point where a patient is hospitalized.
A number of policy implications emerge. Recent data indicate that patients with
a diagnosis of dementia account for a large proportion of alternate level of care (ALC)
hospitalizations and ALC hospital days
276,277
. ALC hospitalizations describe patients in
acute care beds waiting to be moved to a different, more appropriate, care setting, most
frequently residential care
51
. Prolonged stays in ALC are associated with serious
adverse outcomes, including functional decline and infections; they are particularly
188
traumatic for individuals with dementia
277
. As previously discussed, early diagnosis and
management of dementia based on guidelines could conceivably address some of these
patients’ (and caregivers’) needs prior to a crisis that triggers an acute hospitalization.
Therefore, guideline-consistent dementia care could not only ensure better outcomes
for patients but could also reduce the burden of dementia patients on ALC beds.
Interventions to reduce transition are likely to have the most effect if aimed at this
diagnosis period.
8.3 Strengths and Limitations
Each research chapter discusses the limitations associated with the
methodological approach, data and interpretation of results. Therefore, here I will limit
my discussion to the strengths and limitations of the overall thesis.
A key strength of the analyses described here is the use of health care
administrative data for the entire population of BC newly diagnosed with dementia, and
the ability to follow a cohort over a period of eleven years. This allowed me to
circumvent the biases inherent in other studies examining guideline-consistent
dementia care because of their predominant use of self-reported survey data with
limited sample sizes, limited geographic coverage and varying sampling strategies. The
population-based nature of my research helps with external validity, meaning my results
are more generalizable, at least to other provinces, and to other countries with similar
public health care benefit structures that might affect access to care for patients with
dementia. As well, the ability to follow a cohort over eleven years allowed me to
assess trends over the entire arc of a patient’s care trajectory. The median survival time
189
of individuals with dementia in my cohort, from point of diagnosis, was nearly 5 years,
and a ten-year follow up period enabled complete capture of all points between
diagnosis and death for 84% of the cohort (i.e. only 16% of the original 2001-diagnosed
cohort was alive at the end of the study period).
Another important strength in this study design was the ability to identify
individuals newly diagnosed with dementia. By being able to focus on this crucial point
in the trajectory, I was able to measure receipt of guideline-consistent dementia care
during the year of diagnosis and highlight the peak in transitions that occurred during
this period at a population level. The use of administrative data allowed me to trace the
healthcare utilization patterns of this cohort from diagnosis and up to nine years post
diagnosis, while giving me a wealth of patient, physician and system characteristics to
explore and control for in my analysis to tease out specific associations.
Despite the advantages of using administrative data, they also bring with them
some important limitations. Administrative data are subject to transcription and other
quality errors, and missing information, in addition to problems with loss of subjects
which occur regardless of the source of data. Nevertheless, the linked data sources I
used have been shown to have good validity
278
. I was not able to measure all of the
dementia care processes included in the dementia care guidelines, particularly the use
of formal memory tests and the specific role of physicians in referrals to HCC services
which play such a key supportive role in the lives of individuals with dementia and their
caregivers. I also did not have access to data that would have given me in-depth
demographic and functional data which would have been valuable in determining living
arrangements, caregiver status and dementia severity. These variables have been
190
shown to play mediating roles in dementia care. However, previous work has also
shown that the presence of a caregiver has a larger role to play with regard to receipt of
home care, remaining in the community and institutionalization
100,279,280
, but not with
regard to transitions that are primarily driven by hospitalizations or receipt of guideline-
consistent care
281
. These would be expected to be more heavily influenced by physician
decisions. I was able to develop a proxy variable for caregiver status which was used for
a sensitivity analysis described in Chapter 4, Section 4.4.1. The results of that analysis
allowed me to proceed with subsequent analyses, confident that caregiver status was
not going to create any significant omitted variable biases.
All the healthcare utilization data I used are for services provided by physicians
paid through the fee-for-service system. This excludes services provided by physicians
paid through alternative payment arrangements. While in general less than 10 percent
of physicians are paid exclusively through non-fee-for-service methods, this does mean
that my analyses likely under-count the incidence of dementia and are missing the care
utilization data for those patients.
8.4 Future Directions
Specific policy recommendations based on the result of each research objective
have been described in Section 8.2, however there are some overarching observations
and recommendations that are important to highlight at this juncture.
My findings indicate differences in actual care provided and wide variations in
receipt of guideline-consistent dementia care that are influenced by age, income and
geography. The lack of concordance despite evidence of better outcomes for patients
191
has continued to be a major issue and several factors have been previously identified
41
.
I also provide evidence that transitions peak in the year of diagnosis and that receiving
guideline-consistent dementia care in particular is associated with better outcomes, i.e.
fewer transitions.
There would appear to be four recommendations that fall rather naturally out of
the findings reported here. The first relates to physician training. If we are to see more
physicians adopt guideline-consistent care for dementia patients, a reframing and
refreshing of physician education may be necessary. Exposure needs to go beyond the
pathophysiology and pharmacological treatment of dementia as a disease. Instead
education should emphasize dementia as a chronic, complex condition which can
benefit from timely diagnosis, comprehensive biopsychosocial treatment and
management in the context of the patient’s other conditions. Second, a shift in focus
for guidelines needs to occur. Guidelines would be more effective if they provided
more detail around long-term management. For example, physicians should be
provided with evidence-based, relevant environmental interventions for behavioural
symptoms which are often the most difficult management aspect of dementia
117,282
.
Studies have demonstrated that verbal redirection, reassurance, reduced stimulus, and
determining triggers and eliminating them are valuable interventions. But these often
require more frequent monitoring and interdisciplinary team resources for effectiveness
282,283
. Another example is the management of dementia when patients have multiple
major co-morbidities (more often the rule rather than the exception); this tends to be a
barrier to implementing guideline-consistent care. The field has not yet fully
recognized the complexity of providing care in this context, but examples from other
fields such as cancer care could be useful where for example reliable and valid
192
instruments that assess outcomes and care for older cancer patients with pre-existing
co-morbidities are being developed
284–286
.
Third, these results indicate the need to recognize that primary care physicians
alone may not be able to provide adequate dementia care and that more resources are
needed to support them
8
. Patients living at home require a cooperative and
coordinated response from physicians, HCC and social services
287
. Fortunately, several
promising models of care have been piloted and may provide a useful foundation. Case
management programs that provide individualized, flexible responses to people’s needs
by integrating care across clinical and social services with high intensity support have
been shown to be particularly effective when appropriately implemented
78
. These case
management programs often include collaborative, interdisciplinary teams with the
primary care physician at the core and including other service providers such as social
workers, psychiatrists and advanced practice nurses who assist with assessing and
monitoring care. Collaborative, interdisciplinary teams have been piloted in multiple
clinical trials and have yielded better outcomes relating to incidence of adverse
behavioural symptoms, quality of patient care, caregiver health scores and adherence to
guidelines
18,26,288–290
.
Another example is the increased role of skilled staff such as physician assistants,
care coordinators and nurse practitioners in situations of workforce or financial
limitations. After some training, these skilled staff work in a dyad with the physician to
provide better case management and referral services
165,291,292
. This is thought to be
particularly effective in more rural settings in the absence of access to specialists or
community resources
293
. Last are programs like First Link, used by the Alzheimer’s
193
Society of Canada
21,294
. These programs work with physicians to connect people newly
diagnosed with dementia with services offered through the Alzheimer’s Society. In this
case, community service providers actively work with the physician to refer patients to
the program, making them aware of services they can refer patients to. Since many
patients do not contact community resources (despite a referral from the physician)
until in crisis, the First Link program removes the onus of responsibility from the patient
by reaching out first and linking patients with programs as soon as possible.
Preliminary analysis has demonstrated that primary care physicians had improved
knowledge on community support services available, better relationships/partnerships
between physicians and community resource staff, and more effective coping by
caregivers of individuals with dementia
295
.
The fourth recommendation is to further explore interventions that are effective
at reducing transitions, particularly during the year of diagnosis. To be clear, not all
transitions are bad and some transitions are certainly necessary. However, not
developing care sensitive long-term plans for individuals can result in more transitions
than are necessary. Several interventions have been implemented that have been
shown to be promising at reducing unnecessary transitions. For example, improving
early, advanced care planning that ensures appropriate care consistent with individuals’
wishes (e.g. medical orders for life sustaining treatment or hospitalizations) has been
shown to be effective both at reducing transitions and deaths in hospital
176,293,296,297
.
This could prove to be particularly important for individuals with dementia (most of
who are also multi-morbid) at end-of-life, but also perhaps earlier on. Given the
predictable cognitive decline associated with dementia, discussing scope of care, care
settings and appointing a substitute decision maker while the person with dementia is
194
still able to contribute to these decisions could mean better quality of care and a lower
burden on the caregiver who is often forced to make decisions that leave feelings of
guilt and depression
80
.
A second example of an intervention that may decrease transitions is the use of
patient navigators. Patient navigators are skilled staff such as registered nurses or
trained educators who act as the patient’s advocate and personal guide while the
patient is in hospital. They coordinate patient care, communicate with the patient, their
family and physician and help oversee care transitions both within the hospital and
post-discharge. The program was initially implemented in cancer care to improve
access to care for underserved patients but has now been adopted by several other
units in chronic and complex care
298
. One hospital that implemented the program for
their complex inpatient care unit (the majority of whom are elderly patients with
dementia) reported shorter lengths of stay in hospital and lower hospital 30-day
readmission rates
299
. Using patient navigators during the year of diagnosis of dementia
has the potential to significantly influence the number of transitions experienced in that
year (i.e. reduce rates of re-hospitalization) as well as the quality of care and should be
more carefully explored. The value of integrating care across settings cannot be
emphasized enough. My results and those of the other studies cited in this thesis
highlight how essential it is to provide continuity of care across care settings.
8.5 Areas for Further Inquiry
While I highlight some policy implications from my research in section 8.4 above,
there are still many further areas of inquiry that would help add to our understanding of
195
patterns and trajectories of care for people with dementia. I highlight a few of these
here.
First, the outcome of interest for this thesis was the number of transitions for
individuals, examining a 10-year follow-up period from initial diagnosis of dementia, or
until death. This is not the only potential outcome of interest. Future research could
examine effects on the quality of life of patients, as well as the quality of life and health
outcomes for caregivers. Another important outcome is length of time spent at home
(i.e. not in LTC) after diagnosis. I was not able to pursue that outcome because of a lack
of comprehensive information on functional and cognitive status, living arrangements
and presences of informal caregivers, all of which are known to affect an individual’s
ability to remain in the home
82,300,301
. Indeed, future research that is able to incorporate
details on the ongoing functional and cognitive status of individuals with dementia and
indicators of social support would greatly enhance the conclusions that can be drawn.
One possible source of this information is the interRAI suite of minimum data sets,
which are standardized assessment tools designed to gather information for both
assessment and care planning
302
. InterRAI data collection has been mandated since the
mid to late 2000s in BC’s Home and Community Care Offices for assessment for both
home care and LTC.
Second, there may be interest in understanding more than just the number of
transitions, but also the time between transitions and factors that influence that timing.
Alternatively, understanding the timing and kinds of transitions that are associated with
increased risk of functional decline and medical complications may have useful policy
196
planning implications. Again, this line of inquiry would be aided with a fuller
understanding of patients’ living arrangements, functional and cognitive status.
Finally, there is more research that could be done on the best approaches and
supports to increasing provider awareness of appropriate dementia care, and in
particular the effects of community supports that are available such as the First Link
program
295
. These programs are thought to be particularly important for providing
caregivers with additional support and information that physicians may be unable to
provide due to limited time, knowledge or training.
8.6 Conclusion
The results of my research demonstrate that the need to better understand the
patterns of care experienced by individuals with dementia is paramount to providing
more appropriate, strategic support to the physicians who manage their care as well as
the patients and caregivers themselves. The BC Ministry of Health and the Federal
Ministry of Health have both identified dementia as a growing concern for Canadians
who are looking for a responsive healthcare system that provides sustainable, quality
care
270,272,273
. My research illustrates patterns of inequity in receipt of dementia care
unrelated to need, while demonstrating the value of providing guideline-consistent
dementia care (as well as continuity of care) in the year of diagnosis during which
patients experience the highest number of transitions and flux in care. This is evidence
that the current system of care for individuals with dementia is not adequately
addressing their needs. If the BC Ministry of Health is truly committed to improving the
health system for all individuals with dementia and their caregivers while incorporating
197
new approaches to service delivery
273
, an evidence-based approach that provides
targeted resources to physicians that encourages guideline-consistent dementia care
and focusing on the year of diagnosis may prove to be a potentially valuable starting
points of intervention.
198
BIBLIOGRAPHY
1. Ferri, C. P. et al. Global prevalence of dementia: a Delphi consensus study. Lancet 366,
2112–2117 (2005).
2. The Alzheimer Society of Canda. Rising Tide: The Impact of Dementia on Canadian
Society. 1–65 (2010).
3. Blennow, K., de Leon, M. J. & Zetterberg, H. Alzheimer’s disease. Lancet 368, 387–403
(2006).
4. Hogan, D. B. et al. Diagnosis and treatment of dementia: 4. Approach to management of
mild to moderate dementia. Can. Med. Assoc. J. 179, 787–93 (2008).
5. Work Group on Alzheimer’s Disease and Other Dementias. Practice Guidelines for the
Treatment of Patients With Alzheimer’s Disease and Other Dementias. 1–86 (2010).
6. Jack, C. R. et al. Introduction to the recommendations from the National Institute on
Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer's
disease. Alzheimers. Dement. 7, 257–62 (2011).
7. Chertkow, H. et al. Diagnosis and Treatment of Dementia: Introducing a series based on
the third canadian consensus conference on the diagnosis and treatment of dementia.
Can. Med. Assoc. J. 178, 6–11 (2008).
8. Boustani, M., Schubert, C. & Sennour, Y. The challenge of supporting care for dementia in
primary care. Clin. Interv. Aging 2, 631–6 (2007).
9. Brown, C. J., Mutran, E. J., Sloane, P. D. & Long, K. M. Primary Care Physicians’ Knowledge
and Behavior Related to Alzheimer's Disease. J. Appl. Gerontol. 17, 462–479 (1998).
10. Aaltonen, M., Rissanen, P., Forma, L., Raitanen, J. & Jylhä, M. The impact of dementia on
care transitions during the last two years of life. Age Ageing 41, 52–7 (2012).
11. Mead, L. C., Eckert, J. K., Zimmerman, S. & Schumacher, J. G. Sociocultural aspects of
transitions from assisted living for residents with dementia. Gerontologist 45 Spec No,
115–23 (2005).
12. Cohen, C. A. & Pushkar, D. Transitions in care. Lessons learned from a longitudinal study
of dementia care. Am. J. Geriatr. Psychiatry 7, 139–46 (1999).
13. Coleman, E. A., Min, S., Chomiak, A. & Kramer, A. M. Posthospital Care Transitions:
Patterns, Complications, and Risk Identification. Health Serv. Res. 39, 1449–1466 (2004).
14. Forster, A. J., Murff, H. J., Peterson, J. F., Gandhi, T. K. & Bates, D. W. Adverse drug events
occurring following hospital discharge. J Gen Intern Med. 20, 317–323 (2005).
199
15. Aneshensel, C. S., Pearlin, L. I., Levy-Storms, L. & Schuler, R. H. The Transitition From
Home to Nursing Home Mortality Among People With Dementia. J. Gerontol. 55B, S152–
162 (2000).
16. Coleman, E. A. & Boult, C. Improving the quality of transitional care for persons with
complex care needs. J. Am. Geriatr. Soc. 51, 556–7 (2003).
17. King, B. J. et al. The Consequences of Poor Communication During Transitions from
Hospital to Skilled Nursing Facility: A Qualitative Study. J. Am. Geriatr. Soc. (2013).
doi:10.1111/jgs.12328
18. Boustani, M., Sachs, G. & Callahan, C. M. Can primary care meet the biopsychosocial
needs of older adults with dementia? J. Gen. Intern. Med. 22, 1625–7 (2007).
19. Chodosh, J. et al. Effect of a dementia care management intervention on primary care
provider knowledge, attitudes, and perceptions of quality of care. J. Am. Geriatr. Soc. 54,
311–7 (2006).
20. Blumenthal, D., Mort, E. & Edwards, J. The efficacy of primary care for vulnerable
population groups. Health Serv. Res. 30, 253–73 (1995).
21. Spenceley, S. M., Sedgwick, N. & Keenan, J. Dementia care in the context of primary care
reform: an integrative review. Aging Ment. Health 1–14 (2014).
doi:10.1080/13607863.2014.920301
22. Starfield, B. Primary Care: Balancing Health Needs, Services, and Technology. 286 (Oxford
Printing Press, 1998).
23. Worrall, G. & Knight, J. Continuity of care for older patients in family practice: how
important is it? Can. Fam. Physician 52, 754–5 (2006).
24. Wenger, N. S. et al. A practice-based intervention to improve primary care for falls,
urinary incontinence, and dementia. J. Am. Geriatr. Soc. 57, 547–55 (2009).
25. Reuben, D. B. et al. Assessing care of vulnerable elders--Alzheimer’s disease: a pilot study
of a practice redesign intervention to improve the quality of dementia care. J. Am. Geriatr.
Soc. 58, 324–9 (2010).
26. Vickrey, B. G. et al. The Effect of a Disease Management Intervention on Quality and
Outcomes of Dementia Care: A randomized, controlled trial. Ann. Intern. Med. 145, 713–
26 (2006).
27. Lyketsos, C. G. et al. Position statement of the American Association for Geriatric
Psychiatry regarding principles of care for patients with dementia resulting from
Alzheimer disease. Am. J. Geriatr. Psychiatry 14, 561–72 (2006).
28. Boise, L., Camicioli, R., Morgan, D. L., Rose, J. H. & Congleton, L. Diagnosing Dementia:
Perspectives of Primary Care Physicians. Gerontologist 39, 457–464 (1999).
29. De Lepeleire, J. et al. The primary care diagnosis of dementia in Europe: an analysis using
multidisciplinary, multinational expert groups. Aging Ment. Health 12, 568–76 (2008).
200
30. Jong, J. D. De, Groenewegen, P. P., Spreeuwenberg, P., Schellevis, F. & Westert, G. P. Social
Science & Medicine Do guidelines create uniformity in medical practice ? 70, 209–216
(2010).
31. Sivananthan, S. N., Puyat, J. H. & McGrail, K. M. Variations in self-reported practice among
physicians providing clinical care to individuals with dementia: a systematic review. J. Am.
Geriatr. Soc. 61, 1277–1285 (2013).
32. Feldman, H. H. et al. Diagnosis and treatment of dementia: 2. Diagnosis. Can. Med. Assoc.
J. 178, 825–836 (2008).
33. Oakes, S. L. et al. Transitional Care of the Long-Term Care Patient. Clin. Geriatr. Med. 27,
259–271 (2011).
34. Boling, P. A. Care Transitions and Home Health Care. Clin. Geriatr. Med. 25, 135–148
(2009).
35. Wolff, J. L., Boyd, C. M., Gitlin, L. N., Bruce, M. L. & Roter, D. L. Going it together:
persistence of older adults’ accompaniment to physician visits by a family companion. J.
Am. Geriatr. Soc. 60, 106–12 (2012).
36. Finucane, T. E., Bellantoni, M. & Ouslander, J. G. Strangers in strange lands: the serial
transfer of individuals with severe dementia. J. Am. Geriatr. Soc. 61, 1804–5 (2013).
37. Callahan, C. M. et al. Transitions in care for older adults with and without dementia. J. Am.
Geriatr. Soc. 60, 813–20 (2012).
38. Menec, V. H., Sirski, M., Attawar, D. & Katz, A. Does continuity of care with a family
physician reduce hospitalizations among older adults? J. Health Serv. Res. Policy 11, 196–
201 (2006).
39. Nyweide, D. J. et al. Continuity of care and the risk of preventable hospitalization in older
adults. JAMA Intern. Med. 173, 1879–85 (2013).
40. Haggerty, J. L. et al. Continuity of care: a multidisciplinary review. BMJ 327, 1219–21
(2003).
41. Aminzadeh, F., Molnar, F. J., Dalziel, W. B. & Ayotte, D. A Review of Barriers and Enablers
to Diagnosis and Management of Persons with Dementia in Primary Care. Can. Geriatr. J.
15, 85–94 (2012).
42. Alzheimer, A. über eine eigenartige erkrankung der hirnrinde. Allg. Zeitschrift für
Psychiatr. und Psych. Medizin 64, (1907).
43. Harvey, R., Skelton-Robinson, M. & Rossor, M. The prevalence and causes of dementia in
people under the age of 65 years. J Neurol Neurosurg Psychiatry 74, 1206–9 (2003).
44. Schneck, M. J. Vascular dementia. Top. Stroke Rehabil. 15, 22–6 (2008).
45. Drummond, D. C. Diagnosing and Treating Alzheimer’s Disease. Depression 1–20 (2008).
201
46. Khachaturian, Z. S. Revised criteria for diagnosis of Alzheimer’s disease: National Institute
on Aging-Alzheimer's Association diagnostic guidelines for Alzheimer's disease.
Alzheimers. Dement. 7, 253–6 (2011).
47. The World Health Organization. World Health Organization’s International Classification
of Diseases. (2013). at
48. O’Malley, K. J. et al. Measuring diagnoses: ICD code accuracy. Health Serv. Res. 40, 1620–
39 (2005).
49. BC Ministry of Health. Diagnostic Code Descriptions (ICD9). at
50. The American Medical Association. The Differences Between ICD-9 and ICD-10. 1–4
(2012). at
51. Canadian Institute for Health Information. Health Care in Canada, 2011: A Focus on
Seniors and Aging. Health Services Research 1–162 (2011).
52. Sink, K. M., Covinsky, K. E., Barnes, D. E., Newcomer, R. J. & Yaffe, K. Caregiver
characteristics are associated with neuropsychiatric symptoms of dementia. J. Am. Geriatr.
Soc. 54, 796–803 (2006).
53. Kutzleben, M. Von, Schmid, W., Halek, M., Holle, B. & Bartholomeyczik, S. Community-
dwelling persons with dementia: What do they need? What do they demand? What do
they do? A systematic review on the subjective experiences of persons with dementia.
Aging Ment. Health 16, 378–390 (2013).
54. Kane, R. L. a. & Rosalie A Kane. What Older People Want From Long-Term Care, And How
They Can Get It. Health Aff. 20, 114–127 (2001).
55. Dang, S., Badiye, A. & Kelkar, G. The dementia caregiver--a primary care approach. South.
Med. J. 101, 1246–51 (2008).
56. Canadian Institute for Health Information. Caring for Seniors With Alzheimer’s Disease
and Other Forms of Dementia. 1–21 (2010).
57. Bynum, J. Characteristics, costs, and health service use for medicare beneficiaries with a
dementia diagnosis: report I: medicare current beneficiary survey. (2009).
58. Albert, S., Giled, S., Andrews, H., Stern, Y. & Mayeux, R. Primary care expenditures before
the onset of Alzheimer’s disease. Neurology 59, 573–8 (2002).
59. Hill, J. W. et al. Alzheimer’s disease and related dementias increase costs of comorbidities
in managed Medicare. Neurology 58, 62–70 (2002).
60. Oremus, M. & Aguilar, S. C. A systematic review to assess the policy-making relevance of
dementia cost-of-illness studies in the US and Canada. Pharmacoeconomics 29, 141–56
(2011).
202
61. Herrmann, N. et al. The relation between disease severity and cost of caring for patients
with Alzheimer disease in Canada. Can. J. psychiatry 55, 768–75 (2010).
62. Bynum, J. P. et al. The Relationship Between a Dementia Diagnosis, Chronic Illness,
Medicare Expenditures, and Hospital Use. J. Am. Geriatr. Soc. 52, 187–194 (2004).
63. Yang, Z., Zhang, K., Lin, P.-J., Clevenger, C. & Atherly, A. A longitudinal analysis of the
lifetime cost of dementia. Health Serv. Res. 47, 1–19 (2011).
64. Callahan, C., Hendrie, H. & Tierney, W. Documentation and evaluation of cognitive
impairment in elderly primary care patients. Ann. Intern. Med. 122, 422–9 (1995).
65. Schubert, C. C. et al. Comorbidity profile of dementia patients in primary care: are they
sicker? J. Am. Geriatr. Soc. 54, 104–9 (2006).
66. Guijarro, R. et al. Impact of dementia on hospitalization. Neuroepidemiology 35, 101–8
(2010).
67. Canadian Consensus on Diagnosis and Treatment of Dementia Working Group. 3rd
Canadian Consensus Conference on Diagnosis and Treatment of Dementia
Recommendations. in 1–28 (2007).
68. Wolfs, C. A. G., Kessels, A., Dirksen, C. D., Severens, J. L. & Verhey, F. R. J. Integrated
multidisciplinary diagnostic approach for dementia care: randomised controlled trial. Br. J.
Psychiatry 192, 300–5 (2008).
69. Fortinsky, R. H. How Linked Are Physicians to Community Support Services for Their
Patients With Dementia? J. Appl. Gerontol. 17, 480–498 (1998).
70. Koch, T. & Iliffe, S. Rapid appraisal of barriers to the diagnosis and management of
patients with dementia in primary care: a systematic review. BMC Fam. Pract. 11, 52
(2010).
71. Hinton, L. et al. Practice constraints, behavioral problems, and dementia care: primary
care physicians’ perspectives. J. Gen. Intern. Med. 22, 1487–92 (2007).
72. McGrail, K. M. et al. Home Health Services in British Columbia - A portrait of users and
trends over time. October 1–84 (2008).
73. Karlawish, J. H. T. et al. Caregivers’ preferences for the treatment of patients with
Alzheimer’s disease. Neurology 55, 1008–14 (2000).
74. Abetz, L. et al. Alzheimer’s disease treatment: assessing caregiver preferences for mode of
treatment delivery. Adv. Ther. 26, 627–44 (2009).
75. Brega, A. G., Jordan, A. K. & Schlenker, R. E. Practice Variations in Home Health Care.
Home Health Care Serv. Q. 22, 41–64 (2003).
76. Jansen, I. Residential Long-Term Care in Canada: Our Vision for Better Seniors’ Care.
Education And Training 1–101 (2009).
203
77. The British Columbia Government. Helping Seniors Live Well - BC Seniors’ Guide. 1–160
(2009).
78. Somme, D. et al. Analysis of case management programs for patients with dementia: A
systematic review. Alzheimers. Dement. 1–11 (2012). doi:10.1016/j.jalz.2011.06.004
79. Molzahn, A. E., Gallagher, E. & McNulty, V. Quality of Life Associated with Adult Day
Centers. J. Gerontol. Nurs. 35, 37–46 (2009).
80. Nolan, M. & Dellasega, C. “I really feel I”ve let him down’: supporting family carers during
long-term care placement for elders. J. Adv. Nurs. 31, 759–67 (2000).
81. Robinson, C. A., Reid, R. C. & Cooke, H. A. A home away from home: The meaning of
home according to families of residents with dementia. Dementia 2–20 (2010).
doi:10.1177/1471301210381679
82. Grunfeld, E., Glossop, R., McDowell, I. & Danbrook, C. Caring for elderly people at home:
the consequences to caregivers. CMAJ 157, 1101–5 (1997).
83. Provincial Dementia Service Framework Working Group. BC Dementia Service Framework.
Group 1–157 (2007).
84. Cohen, M., Murphy, J., Nutland, K., Ostry, A. & Care, C. Continuing Care Renewal or
Retreat: BC Residential and Home Health Care Restructuring 2001-2004. 1–54 (2005).
85. Canadian Centre for Policy Alternatives. Cuts to Residential & Home Health Care in BC. 1–
4 (2005).
86. BC Ombudsperson et al. The Best of Care: Getting It Right for Seniors in British Columbia
(Part 1). 2, 1–254 (2009).
87. Penning, M. J., Brackley, M. E. & Allan, D. E. Home care and health reform: changes in
home care utilization in one Canadian province, 1990-2000. Gerontologist 46, 744–58
(2006).
88. Hollander, M. J. Final report of the study on the comparative cost analysis of home care
and residential care services. (2001).
89. Chappell, N. L., Dlitt, B. H., Hollander, M. J., Miller, J. A. & McWilliam, C. Comparative costs
of home care and residential care. Gerontologist 44, 389–400 (2004).
90. Coyte, P. C. & McKeever, P. Home care in Canada: passing the buck. Can. J. Nurs. Res. 33,
11–25 (2001).
91. Health Council of Canada. Seniors in need, caregivers in distress: What are the home care
priorities for seniors in Canada?. Health (San Francisco) 1–28 (2012).
92. Case, D. Contributions of journeys away to the definition of home: an empirical study of a
dialectical process. J. Environ. Psychol. 16, 1–15 (1996).
204
93. Sixsmith, A. & Sixsmith, J. Transitions in home experience in later life. J. Archit. Plann. Res.
8, 181–191 (1991).
94. Rowles, G. D. Evolving images of place in aging and “aging in place.” Generations 17, 65–
70 (1993).
95. Golant, S. M. Conceptualizing time and behavior in environmental gerontology: a pair of
old issues deserving new thought. Gerontologist 43, 638–48 (2003).
96. Frank, J. Semiotic use of the word “home” among people with Alzheimer’s disease: A plea
for selfhood?. 171–197 (Springer, 2005).
97. Aminzadeh, F., Dalziel, W. B., Molnar, F. J. & Garcia, L. J. Symbolic meaning of relocation to
a residential care facility for persons with dementia. Aging Ment. Health 13, 487–96
(2009).
98. McDowell, I. et al. Patterns of Caring for People with Dementia in Canada. Can. J. Aging
13, 470–487 (1994).
99. Assisted Living Registrar of British Columbia. Assisted Living, Information for Applicants.
Office 1–50 (2009).
100. Yaffe, K. et al. Patient and caregiver characteristics and nursing home placement in
patients with dementia. JAMA 287, 2090–7 (2002).
101. Wattmo, C., Wallin, A. K., Londos, E. & Minthon, L. Risk factors for nursing home
placement in Alzheimer’s disease: a longitudinal study of cognition, ADL, service
utilization, and cholinesterase inhibitor treatment. Gerontologist 51, 17–27 (2011).
102. Nolan, M. R., Davies, S., Brown, J., Keady, J. & Nolan, J. Beyond person-centred care: a new
vision for gerontological nursing. J. Clin. Nurs. 13, 45–53 (2004).
103. Alzheimer’s Society of British Columbia. Guidelines for Care: Person-centred care of
people with dementia living in care homes. 1–65 (2011).
104. Mead, R. The Sense of An Ending. The New Yorker 92 – 103 (2013).
105. Jenkins, C. & Smythe, A. Reflections on a visit to a dementia care village. Nurs. Older
People 25, 14–9 (2013).
106. Miller, A., Voyer, C. & Puim, T. Creation of a landmark: The Residential Care Redesign at
Boundary Hospital Community Health Centre. Long Term Living Magazine 10, 10–13
(2009).
107. Gnaedinger, N., Robinson, J., Sudbury, F. & Dutchak, M. Renovating the built environment
for dementia care: lessons learned at the Lodge at Broadmead in Victoria, British
Columbia. Healthc. Q. 10, 76–80 (2007).
108. Smith, R., Mathews, R. M. & Gresham, M. Pre- and postoccupancy evaluation of new
dementia care cottages. Am. J. Alzheimers. Dis. Other Demen. 25, 265–75 (2010).
205
109. Levy, C., Palat, S.-I. T. & Kramer, A. M. Physician practice patterns in nursing homes. J. Am.
Med. Dir. Assoc. 8, 558–67 (2007).
110. The General Practice Services Committee Residential Care Working Group. GPSC
Residential Care Project: Current State (draft). 1–56 (2014).
111. Chan, B. T. B. The declining comprehensiveness of primary care. Can. Med. Assoc. J. 166,
429–34 (2002).
112. Lavergne, R. et al. Defining and Measuring Full Service Family Practice in BC , 1991 – 2006.
1–15 (2013).
113. Katz, P. R., Karuza, J., Kolassa, J. & Hutson, A. Medical practice with nursing home
residents: results from the National Physician Professional Activities Census. J. Am.
Geriatr. Soc. 45, 911–7 (1997).
114. Lam, J. M., Anderson, G. M., Austin, P. C. & Bronskill, S. E. Family physicians providing
regular care to residents in Ontario long-term care homes Characteristics and practice
patterns. Can. Fam. physician 58, 1241–1248 (2012).
115. Chen, Y. et al. Unexplained variation across US nursing homes in antipsychotic prescribing
rates. Arch. Intern. Med. 170, 89–95 (2010).
116. Rochon, P. a et al. Variation in nursing home antipsychotic prescribing rates. Arch. Intern.
Med. 167, 676–83 (2007).
117. Sadowsky, C. H. & Galvin, J. E. Guidelines for the Management of Cognitive and
Behavioral Problems in Dementia. J. Am. Board Fam. Med. 25, 350–366 (2012).
118. McKhann, G. M. et al. The diagnosis of dementia due to Alzheimer’s disease:
Recommendations from the National Institute on Aging-Alzheimer's Association
workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers. Dement. 7,
263–9 (2011).
119. Waldemar, G. et al. Recommendations for the diagnosis and management of Alzheimer’s
disease and other disorders associated with dementia: EFNS guideline. Eur. J. Neurol. 14,
e1–26 (2007).
120. Doody, R. S. et al. Practice parameter: Management of dementia (an evidence-based
review): Report of the Quality Standards Subcommittee of the American Academy of
Neurology. Neurology 56, 1154–1166 (2001).
121. AGS Clinical Practice Committee. Guidelines Abstracted from the American Academy of
Neurology’s Dementia Guidelines for Early Detection, Diagnosis, and Management of
Dementia. J. Am. Geriatr. Soc. 51, 869–873 (2003).
122. British Columbia Ministry of Health & British Columbia Guidelines & Protocol Advisory
Committee. Cognitive Impairment in the Elderly – Recognition, Diagnosis and
Management. Guidelines 1–30 (2007).
206
123. American Academy of Neurology. American Academy of Neurology Guideline Summary
for Clinicians: Detection, Diagnosis and Management of Dementia. 1–2 (2004).
124. Patterson, C. et al. Diagnosis and treatment of dementia: 1. Risk assessment and primary
prevention of Alzheimer disease. CMAJ 178, 548–56 (2008).
125. Chertkow, H. et al. Diagnosis and treatment of dementia: 3. Mild cognitive impairment
and cognitive impairment without dementia. Can. Med. Assoc. J. 178, 1273–85 (2008).
126. Hogan, D. B. et al. Diagnosis and treatment of dementia: 5. Nonpharmacologic and
pharmacologic therapy for mild to moderate dementia. Can. Med. Assoc. J. 179, 1019–
1026 (2008).
127. Herrmann, N. & Gauthier, S. Diagnosis and treatment of dementia: 6. Management of
severe Alzheimer disease. CMAJ 179, 1279–87 (2008).
128. McKhann, G. et al. Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-
ADRDA Work Group under the auspices of Department of Health and Human Services
Task Force on Alzheimer's Disease. Neurology 34, 939–939 (1984).
129. Petersen, R. et al. Practice parameter: early detection of dementia: mild cognitive
impairment (an evidence-based review). Report of the Quality Standards Subcommittee
of the American Academy of Neurology. Neurology 56, 1133–42 (2001).
130. Knopman, D. et al. Practice parameter: diagnosis of dementia (an evidence-based review).
Report of the Quality Standards Subcommittee of the American Academy of Neurology.
Neurology 56, 1143–53 (2001).
131. Sternberg, S. A., Wolfson, C. & Baumgarten, M. Undetected dementia in community-
dwelling older people: the Canadian Study of Health and Aging. J. Am. Geriatr. Soc. 48,
1430–4 (2000).
132. Nasreddine, Z. S. et al. The Montreal Cognitive Assessment, MoCA: a brief screening tool
for mild cognitive impairment. J. Am. Geriatr. Soc. 53, 695–9 (2005).
133. British Columbia Ministry of Health. Cognitive Impairment in the Elderly Flow Sheet.
Dementia 6009, 1 (2006).
134. British Columbia Ministry of Health. Alzheimer’s Disease Drug Initiative. (2007). at
135. British Columbia Ministry of Health. Alzheimer’s Drug Therapy Initiative Qs & As for
Research Program. Administrator 1–4 (2008).
136. General Practice Services Committee. FP Incentive Program. (2009). at
137. GP Services Committee. Mental Health Incentives. (2011).
138. British Columbia Ministry of Health. MSP Fee-For-Service Payment Analysis 2008/2008 to
2011/2012. 1–119 (2012).
207
139. Hartford, K., Roos, L. L. & Walld, R. Regional Variation in Angiography, Coronary Artery
Bypass Surgery and Percutaneous Transluminal Coronary Angioplasty in Manitoba , 1987
to 1992: The Funnel Effect. Med. Care 36, 1022–1032 (1998).
140. Einarsdóttir, K., Preen, D. B., Emery, J. D., Kelman, C. & Holman, C. D. J. Regular primary
care lowers hospitalisation risk and mortality in seniors with chronic respiratory diseases.
J. Gen. Intern. Med. 25, 766–73 (2010).
141. Roos, N. P., Flowerdew, G., Wajda, a & Tate, R. B. Variations in physicians’ hospitalization
practices: a population-based study in Manitoba, Canada. Am. J. Public Health 76, 45–51
(1986).
142. Ahronheim, J. C., Mulvihill, M., Sieger, C., Park, P. & Fries, B. E. State practice variations in
the use of tube feeding for nursing home residents with severe cognitive impairment. J.
Am. Geriatr. Soc. 49, 148–52 (2001).
143. Herrmann, N. et al. A population-based study of cholinesterase inhibitor use for
dementia. J. Am. Geriatr. Soc. 55, 1517–23 (2007).
144. Maneno, M. K. et al. National patterns of dementia treatment among elderly ambulatory
patients. J. Natl. Med. Assoc. 98, 430–5 (2006).
145. Rattinger, G. B., Mullins, C. D., Zuckerman, I. H., Onukwugha, E. & Delisle, S. Clinic visits
and prescribing patterns among Veterans Affairs Maryland Health Care System dementia
patients. J. Nutr. Health Aging 14, 677–83 (2010).
146. Chodosh, J. et al. Caring for patients with dementia: how good is the quality of care?
Results from three health systems. J. Am. Geriatr. Soc. 55, 1260–8 (2007).
147. Wenger, N. S. et al. The quality of medical care provided to vulnerable community-
dwelling older patients. Ann. Intern. Med. 139, 740–7 (2003).
148. Maclean, L. C., Richman, A. & Maclean, C. Physician Practice Style Variations. Socio-Econ
Plann Sci. 23, 361–371 (1989).
149. Noble, M. New practice atlas outlines variations in health care provided across Ontario.
CMAJ 151, 218–9 (1994).
150. Chang, C.-H., Stukel, T. A., Flood, A. B. & Goodman, D. C. Primary care physician workforce
and Medicare beneficiaries’ health outcomes. JAMA 305, 2096–104 (2011).
151. Diehr, P., Cain, K., Ye, Z. & Abdul-Salam, F. Small area variation analysis. Methods for
comparing several diagnosis-related groups. Med. Care 31, YS45–53 (1993).
152. Canadian Institute for Health Information. National Health Expenditure Trends, 1975 to
2010. 1–172 (2010).
153. Baloch, S., Moss, S. B., Nair, R., Tingle, L. & Moss, B. Practice patterns in the evaluation and
management of dementia by primary care residents, primary care physicians, and
geriatricians. Proc. Baylor Univ. Med. Cent. 23, 121–125 (2010).
208
154. Fortinsky, R. H., Zlateva, I., Delaney, C. & Kleppinger, A. Primary care physicians’ dementia
care practices: evidence of geographic variation. Gerontologist 50, 179–91 (2009).
155. Glasser, M. Alzheimer’s disease and dementing disorders: Practices and experiences of
rural physicians. Am. J. Alzheimers. Dis. Other Demen. 8, 28–35 (1993).
156. Coleman, E. A. & Boult, C. Falling through the cracks: challenges and opportunities for
improving transitional care for persons with continuous complex care needs. J. Am.
Geriatr. Soc. 51, 549–55 (2003).
157. Dubin, S. When the patient suffers: optimization of transitions between care facilities.
Geriatr. Nurs. 28, 298–300 (2007).
158. Anderson, M. & Helms, L. Communication between continuing care organizations. Res
Nurs. Heal. 18, 49–57 (1995).
159. Coleman, E. A. & Berenson, R. A. Lost in Transition: Challenges and Opportunities for
Improving the Quality of Transitional Care. Ann. Intern. Med. 140, 533–536 (2004).
160. Moore, C., Wisnivesky, J., Williams, S. & McGinn, T. Medical errors related to discontinuity
of care from an inpatient to an outpatient setting. J Gen Intern Med 18, 646–651 (2003).
161. Forster, A. J., Murff, H. J., Peterson, J. F., Gandhi, T. K. & Bates, D. W. The Incidence and
Severity of Adverse Events Affecting Patients after Discharge from the Hospital. Ann.
Intern. Med. 138, 161–174 (2003).
162. Intrator, O. et al. Hospitalization of nursing home residents: The effects of states’
Medicaid payment and bed-hold policies. Health Serv. Res. 42, 1651–1671 (2007).
163. Smallegan, M. There was nothing else to do: Needs for care before nursing home
admission. Gerontologist 25, 364
164. Boise, L., Neal, M. B. & Kaye, J. Dementia assessment in primary care: results from a study
in three managed care systems. J. Gerontol. A. Biol. Sci. Med. Sci. 59, M621–6 (2004).
165. Naylor, M. D., Kurtzman, E. T. & Pauly, M. V. Transitions of elders between long-term care
and hospitals. Policy. Polit. Nurs. Pract. 10, 187–94 (2009).
166. Hutt, E., Ecord, M., Eilertsen, T. B., Frederickson, E. & Kramer, A. M. Precipitants of
emergency room visits and acute hospitalization in short-stay medicare nursing home
residents. J. Am. Geriatr. Soc. 50, 223–9 (2002).
167. Naylor, M. D. Transitional care of older adults. Annu. Rev. Nurs. Res. 20, 127–147 (2003).
168. Reason, B., Terner, M., Moses McKeag, A., Tipper, B. & Webster, G. The impact of
polypharmacy on the health of Canadian seniors. Fam. Pract. 29, 427–32 (2012).
169. Somers, M. et al. Quality use of medicines in residential aged care. Aust. Fam. Physician
39, 413–6 (2010).
209
170. Phelan, E., Borson, S., Grothaus, L., Balch, S. & Larson, E. B. Association of Incident
Dementia with Hospitalizations. J. Am. Med. Assoc. 307, 165–172 (2012).
171. Ouslander, J. G. et al. Potentially avoidable hospitalizations of nursing home residents:
frequency, causes, and costs: [see editorial comments by Drs. Jean F. Wyman and William
R. Hazzard, pp 760-761]. J. Am. Geriatr. Soc. 58, 627–35 (2010).
172. Hong, J. S., Kang, H. C. & Kim, J. Continuity of care for elderly patients with diabetes
mellitus, hypertension, asthma, and chronic obstructive pulmonary disease in Korea. J.
Korean Med. Sci. 25, 1259–71 (2010).
173. Christakis, D. A., Mell, L., Koepsell, T. D., Zimmerman, F. J. & Connell, F. A. Association of
lower continuity of care with greater risk of emergency department use and
hospitalization in children. Pediatrics 107, 524–9 (2001).
174. Gill, J. M. & Mainous, A. G. The role of provider continuity in preventing hospitalizations.
Arch. Fam. Med. 7, 352–7
175. Weiss, L. J. & Blustein, J. Faithful patients: the effect of long-term physician-patient
relationships on the costs and use of health care by older Americans. Am. J. Public Health
86, 1742–7 (1996).
176. McGregor, M. J., Pare, D., Cox, M. & Brasher, P. Correlates of a “do no hospitalize”
designation: in a sample of frail nursing home residents in Vancouver. Can. Fam.
physician 56, 1158–64 (2010).
177. McDaniel, D. B., Patton, E. W. & Mather, J. A. Immunization activities of private-practice
physicians: a record audit. Pediatrics 56, 504–7 (1975).
178. Christakis, D. A., Mell, L., Wright, J. A., Davis, R. & Connell, F. A. The association between
greater continuity of care and timely measles-mumps-rubella vaccination. Am. J. Public
Health 90, 962–5 (2000).
179. Blankfield, R. P., Kelly, R. B., Alemagno, S. A. & King, C. M. Continuity of care in a family
practice residency program. Impact on physician satisfaction. J. Fam. Pract. 31, 69–73
(1990).
180. Alessi-Severini, S. et al. Utilization and costs of antipsychotic agents: a Canadian
population-based study, 1996-2006. Psychiatr. Serv. 59, 547–53 (2008).
181. Amuah, J. E. et al. Persistence with cholinesterase inhibitor therapy in a population- based
cohort of patients with Alzheimer’s disease. Pharmacoepidemiol. Drug Saf. 19, 670–679
(2010).
182. Olsson, J. et al. Quality of drug prescribing in elderly people in nursing homes and special
care units for dementia: a cross-sectional computerized pharmacy register analysis. Clin.
Drug Investig. 30, 289–300 (2010).
183. Joyce, T., Shadid, G., Li, H., Offord, K. & Agerter, D. A Survey of Family Physicians and
Psychiatrists Psychotropic Prescribing Practices and Educational Needs. Psychiatry Prim.
Care 20, 360–367 (1998).
210
184. Centre for Reviews and Dissemination. Systematic Reviews: CRD’s Guidance for
Undertaking Reviews in Health Care. 1–294 (2009).
185. Liberati, A. et al. The PRISMA statement for reporting systematic reviews and meta-
analyses of studies that evaluate health care interventions: explanation and elaboration. J.
Clin. Epidemiol. 62, e1–34 (2009).
186. Moher, D., Liberati, A., Tetzlaff, J. & Altman, D. G. Preferred reporting items for systematic
reviews and meta-analyses: the PRISMA statement. PLoS Med. 6, e1000097 (2009).
187. Somerfield, M. R., Weisman, C. S., Ury, W., Chase, G. A. & Folstein, M. F. Physician Practices
in the Diagnosis of Dementing Disorders. J Am Geriatr Soc 39, 172–175 (1991).
188. Rubin, S. M., Glasser, M. L. & Werckle, M. A. The examination of physicians’ awareness of
dementing disorders. Geriatr Soc 35, 1051–1058 (1987).
189. Landish, R. J. & Koch, G. G. The Measurement of Observer Agreement for Categorical
Data. Biometrics 33, 159–174 (1977).
190. Bisset, A. F. & MacPherson, I. A. Patients with dementia: the view from general practice in
Grampian. Health Bull. (Raleigh). 1, 32–36 (1996).
191. Cavalieri, T. A., Latif, W., Ciesielski, J., Ciervo, C. & Forman, L. How physicians approach
advance care planning in patients with mild to moderate Alzheimer’s disease. J. Am.
Osteopath. Assoc. 102, 541–544 (2002).
192. Cohen-Mansfield, J., Jensen, B., Resnick, B. & Norris, M. Assessment and treatment of
behavior problems in dementia in nursing home residents: a comparison of the
approaches of physicians, psychologists, and nurse practitioners. Int. J. Geriatr. Psychiatry
27, 135–45 (2011).
193. Robinson, L., Vellas, B., Knox, S. & Lins, K. Clinical Practice Patterns of Generalists and
Specialists in Alzheimer’s Disease: What are the differences, and what difference do they
make? J. Nutr. Heal. Aging 14, 545–552 (2010).
194. Cody, M., Beck, C., Shue, V. M. & Pope, S. Reported practices of primary care physicians in
the diagnosis and management of dementia. Aging Ment. Health 6, 72–76 (2002).
195. Cohen-Mansfield, J. & Jensen, B. Physicians’ perceptions of their role in treating
dementia-related behavior problems in the nursing home: actual practice and the ideal. J.
Am. Med. Dir. Assoc. 552–557 (2008).
196. Berger, J. T. The influence of physicians’ demographic characteristics and their patients'
demographic characteristics on physician practice: implications for education and
research. Acad. Med. 83, 100–5 (2008).
197. Arnold, H. J. & Feldman, D. C. Social Desirability Response Bias in Self-Report Choice
Situations. Acad. Manag. J. 24, 377–385 (1981).
198. Black, N. Health services research: saviour or chimera? Lancet 349, 1834–36 (1997).
211
199. Wennberg, J. & Gittelsohn. Small area variations in health care delivery. Science 182,
1102–8 (1973).
200. Virnig, B. A. & Mcbean, M. Administrative Data for Public Health Surveillance and
Planning. Ann. Intern. Med. 22, 213–30 (2001).
201. Finlayson, E. & Berkmeyer, J. Research Based on Administrative Data. Surgery 145, 610–
616 (2009).
202. Population Data BC. Population Data BC. at
203. BC ministry of Health [creator] (2012): Consolidation File (MSP Registration & Premium
Billing). V2. Population Data BC [publisher]. Data Extract. MOH (2012).
204. Williams, D. Socioeconomic Differentials In Health: A Review and Redirection. Soc.
Psychol. Q. 53, 81–99 (1990).
205. BC Ministry of Health. MSP Premiums. (2012). at
206. BC ministry of Health [creator] (2012): Medical Services Plan (MSP) Payment Information
File. V2. Population Data BC [publisher]. Data Extract. MOH (2012). at
207. Hu, W. Diagnostic Codes in MSP Claim Data, Summary Report. (1996).
208. Canadian Institute for Health Information. Physicians in Canada: The Status of Alternative
Payment Programs, 2005-2006. 1–44 (2008).
209. BC ministry of Health [creator] (2012): Home & Community Care (Continuing Care). V2.
Population Data BC [publisher]. Data Extract. MOH (2012). at
210. BC ministry of Health [creator] (2012): Discharge Abstract Database (Hospital
Separations). V2. Population Data BC [publisher]. Data Extract. MOH (2012). at
211. BC Vital Statistics Agency [creator] (2012): Vital Statistics Deaths. V2. Population Data BC
[publisher]. Data Extract. BC Vital Statistics Agency (2012). at
212. Todd, S., Barr, S. & Passmore, A. P. Cause of death in Alzheimer’s disease: a cohort study.
QJM 106, 747–53 (2013).
213. BC ministry of Health [creator] (2012): PharmaNet. V2. Ministry of Health [publisher]. Data
Extract. MOH (2012).
214. BC Ministry of Health. BC Ministry of Health PharmaNet. at
212
215. The World Health Organization. WHO Collaborating Centre for Drug Statistics
Methodology: Structure & Principle of the Anatomical Therapeutic Chemical Classification
System. (2011). at
216. Health Canada. Health Canada Drug Product Database. (2013). at
217. The Alzheimer Association of Canada. A New Way of Looking at the Impact of Dementia
in Canada. (2012).
218. Broemeling, A. & Watson, D. Chronic conditions and co-morbidity among residents of
British Columbia. Distribution 1–46 (2005).
219. Sivananthan, S. N., Peterson, S., Lavergne, R., Barer, M. L. & McGrail, K. M. Designation,
diligence and drift: understanding laboratory expenditure increases in British Columbia,
1996/97 to 2005/06. BMC Health Serv. Res. 12, 472 (2012).
220. Lix, L. et al. Defining and Validating Chronic Diseases: An Administrative Data Approach.
Health (San Francisco) 2006, 1–217 (2006).
221. Health, B. M. of. Cognitive Impairment in the Elderly – Recognition , Diagnosis and
Management. Dementia (2008).
222. Manitoba Centre for Health Policy. Concept: Adjusted Clinical Groups (ACG) - Overview.
(2010). at
223. Reid, R. J., MacWilliam, L., Verhulst, L., Roos, N. & Atkinson, M. Performance of the ACG
case-mix system in two Canadian provinces. Med. Care 39, 86–99 (2001).
224. Doraiswamy, P. M., Leon, J., Cummings, J. L., Marin, D. & Neumann, P. J. Prevalence and
Impact of Medical Comorbidity in Alzheimer’s Disease. Journals Gerontol. Ser. A Biol. Sci.
Med. Sci. 57, M173–M177 (2002).
225. Korczyn, A. & Halperin, I. Depression and Dementia. J. Neurol. Sci. 283, 139 – 142 (2009).
226. McGrail, K. M. Long-term care as part of the continuum. Healthc. Pap. 10, 39–43 (2011).
227. STATA Data Analysis & Statistical Software. (2007).
228. Kane, M. N. Awareness of ageism, motivation, and countertransference in the care of
elders with Alzheimer’s disease. Am. J. Alzheimers. Dis. Other Demen. 17, 101–109 (2002).
229. Takeda, A. et al. A systematic review of the clinical effectiveness of donepezil,
rivastigmine and galantamine on cognition, quality of life and adverse events in
Alzheimer’s disease. Int. J. Geriatr. Psychiatry 21, 17–28 (2006).
230. Hansen, R. A. et al. Efficacy and safety of donepezil, galantamine, and rivastigmine for the
treatment of Alzheimer’s disease: a systematic review and meta-analysis. Clin. Interv.
Aging 3, 211–25 (2008).
213
231. Tricco, A. C. et al. Efficacy and safety of cognitive enhancers for patients with mild
cognitive impairment: a systematic review and meta-analysis. CMAJ 185, 1393–401
(2013).
232. Meeks, T. W. & Jeste, D. V. Antipsychotics in Dementia: Beyond “Black Box” Warnings.
Curr. Psychiatr. 7, 51–65 (2008).
233. Puyat, J. H., Law, M. R., Wong, S. T., Sutherland, J. M. & Morgan, S. G. The Essential and
Potentially Inappropriate Use of Antipsychotics Across Income Groups: An Analysis of
Linked Administrative Data. Can. J. psychiatry 57, 488–495 (2012).
234. Asche, C. V, Singer, M. E., Jhaveri, M., Chung, H. & Miller, A. All-cause health care
utilization and costs associated with newly diagnosed multiple sclerosis in the United
States. J. Manag. Care Pharm. 16, 703–12 (2010).
235. Marrie, R. A., Yu, N., Wei, Y., Elliott, L. & Blanchard, J. High rates of physician services
utilization at least five years before multiple sclerosis diagnosis. Mult. Scler. 19, 1113–9
(2013).
236. Hobson, D. E. et al. Healthcare utilization in patients with Parkinson’s disease: a
population-based analysis. Parkinsonism Relat. Disord. 18, 930–5 (2012).
237. Aaltonen, M., Forma, L., Rissanen, P., Raitanen, J. & Jylhä, M. Transitions in health and
social service system at the end of life. Eur. J. Ageing 7, 91–100 (2010).
238. Horttana, B. M., Ahlström, G. & Fahlström, G. Patterns of and reasons for relocation in
dementia care. Geriatr. Nurs. (Minneap). 28, 193–200 (2002).
239. Teno, J. M. et al. Changes in End-Of-Life Care for Medicare Beneficiaries: Site of Death,
Place of Care, and Health Care Transitions in 2000, 2005, and 2009. J. Am. Med. Assoc.
309, 470–477 (2013).
240. Cohen, M., Tate, J. & Baumbusch, J. An Uncertain Future for Seniors: BC’s restructuring of
home and community health care, 2001-2008. 1–71 (2009).
241. Lyketsos, C. G. Prevention of unnecessary hospitalization for patients with dementia: the
role of ambulatory care. JAMA 307, 197–8 (2012).
242. Lin, P.-J., Fillit, H. M., Cohen, J. T. & Neumann, P. J. Potentially avoidable hospitalizations
among Medicare beneficiaries with Alzheimer’s disease and related disorders. Alzheimers.
Dement. 9, 30–8 (2013).
243. Ouslander, J. G. et al. Interventions to reduce hospitalizations from nursing homes:
evaluation of the INTERACT II collaborative quality improvement project. J. Am. Geriatr.
Soc. 59, 745–53 (2011).
244. Coleman, E. A., Parry, C., Chalmers, S. & Min, S.-J. The care transitions intervention: results
of a randomized controlled trial. Arch. Intern. Med. 166, 1822–8 (2006).
245. Morrison, R. S. Survival in End-Stage Dementia Following Acute Illness. JAMA 284, 47
(2000).
214
246. Luppa, M., Luck, T., Brähler, E., König, H.-H. & Riedel-Heller, S. G. Prediction of
institutionalisation in dementia: a systematic review. Dement. Geriatr. Cogn. Disord. 26,
65–78 (2008).
247. Mitchell, S. L., Kiely, D. K. & Hamel, M. B. Dying with advanced dementia in the nursing
home. Arch. Intern. Med. 164, 321–6 (2004).
248. Gozalo, P. et al. End-of-life transitions among nursing home residents with cognitive
issues. N. Engl. J. Med. 365, 1212–1221 (2011).
249. Sampson, E. L., Ritchie, C. W., Lai, R., Raven, P. W. & Blanchard, M. R. A systematic review
of the scientific evidence for the efficacy of a palliative care approach in advanced
dementia. Int. Psychogeriatrics 17, 31–40 (2005).
250. Birch, D. & Draper, J. A critical literature review exploring the challenges of delivering
effective palliative care to older people with dementia. J. Clin. Nurs. 17, 1144–63 (2008).
251. Hinkka, H. et al. Decision making in terminal care: a survey of finnish doctors’ treatment
decisions in end-of-life scenarios involving a terminal cancer and a terminal dementia
patient. Palliat. Med. 16, 195–204 (2002).
252. Bunbrack, J. The information needs of informal caregivers involved in providing support
to a critically ill loved one. 1–19 (2005).
253. Mitchell, S. L., Morris, J. N., Park, P. S. & Fries, B. E. Terminal care for persons with
advanced dementia in the nursing home and home care settings. J. Palliat. Med. 7, 808–
16 (2004).
254. Christina, W. et al. A Reevaluation of the Duration of Survival after the Onset of Dementia.
N. Engl. J. Med. 344, 1111–1116 (2001).
255. Larson, E. B. Survival after Initial Diagnosis of Alzheimer Disease. Ann. Intern. Med. 140,
501 (2004).
256. Brookmeyer, R., Corrada, M. M., Curriero, F. C. & Kawas, C. Survival Following a Diagnosis
of Alzheimer Disease. Arch. Neurol. 59, 1764 (2002).
257. Rose, K. M. & Palan Lopez, R. Transitions in Dementia Care: Theoretical Support for
Nursing Roles. Online J. Issues Nurs. 17, 1–18 (2012).
258. Ahronheim, J. C., Morrison, R. S., Baskin, S. A., Morris, J. & Meier, D. E. Treatment of the
dying in the acute care hospital. Advanced dementia and metastatic cancer. Arch. Intern.
Med. 156, 2094–100 (1996).
259. Mitchell, S. L., Teno, J. M., Intrator, O., Feng, Z. & Mor, V. Decisions to forgo hospitalization
in advanced dementia: a nationwide study. J. Am. Geriatr. Soc. 55, 432–8 (2007).
260. Sivananthan, S. N., Lavergne, M. R. & Mcgrail, K. M. Caring for dementia: a population-
based study examining variations in guideline-consistent care. Alzheimers. Dement.
215
261. Roos, N. P., Carriere, K. & Friesen, D. Factors influencing the frequency of visits by
hypertensive patients to primary care physicians in Winnipeg. Can. Med. Assoc. J. 159,
777–783 (1998).
262. Pimlott, N. J. G., Siegel, K., Persaud, M., Slaughter, M. N. S. & Hollingworth, G.
Management of dementia by family physicians in academic settings Résumés de
recherche. Med. Care 52, 1108–1109 (2006).
263. Wilcock, J. et al. Concordance with clinical practice guidelines for dementia in general
practice. Aging Ment. Health 13, 155–61 (2009).
264. Brodaty, H., Green, A. & Koschera, A. Meta-Analysis of Psychosocial Interventions for
Caregivers of People with Dementia. J. Am. Geriatr. Soc. 51, 657–664 (2003).
265. Perälä, J. et al. Lifetime prevalence of psychotic and bipolar I disorders in a general
population. Arch. Gen. Psychiatry 64, 19–28 (2007).
266. Copeland, J. R. et al. Range of mental illness among the elderly in the community.
Prevalence in Liverpool using the GMS-AGECAT package. Br. J. Psychiatry 150, 815–23
(1987).
267. Paulsen, J. S. et al. Incidence of and risk factors for hallucinations and delusions in
patients with probable AD. Neurology 54, 1965–71 (2000).
268. Flegel, K. People with dementia need better pain management. C. Can. Med. Assoc. J.
185, 1195 (2013).
269. Sachs, G. A., Shega, J. W. & Cox-Hayley, D. Barriers to excellent end-of-life care for
patients with dementia. J. Gen. Intern. Med. 19, 1057–63 (2004).
270. UK Department of Health. G8 summit on dementia. (2013). at
271. CBC News. National Dementia Research Initiative Gets $31.5M Federal Funding. (2014). at
272. Canadian Institutes of Health Research. Global Action Against Dementia - Second Global
Dementia Legacy Event. (2014). at
273. BC Ministry of Health. The Provincial Dementia Action Plan for British Columbia. 1–24
(2012).
274. Sink, K. M., Holden, K. F. & Yaffe, K. Pharmacological treatment of neuropsychiatric
symptoms of dementia: a review of the evidence. JAMA 293, 596–608 (2005).
275. BC Ministry of Health. A Review of the use of antipsychotic drugs in British Columbia
residential care facilities. Statistics 1–47 (2011).
276. Institute for Clinical Evaluative Sciences. Health System Use by Frail Ontario Seniors: An
in-dept examination of four vulnerable cohorts. 1–145 (2011).
216
277. Canadian Institute for Health Information. Alternate Level of Care in Canada. 1–20 (2009).
278. Williams, J. & Young, W. Inventory of studies on the accuracy of Canadian Health
Administrative databases. (1996).
279. Tsuji, I., Whalen, S. & Finucane, T. E. Predictors of nursing home placement in community-
based long-term care. J. Am. Geriatr. Soc. 43, 761–6 (1995).
280. Cohen, C. a, Pringle, D. & LeDuc, L. Dementia caregiving: the role of the primary care
physician. Can. J. Neurol. Sci. 28 Suppl 1, S72–6 (2001).
281. Golden, A. G., Tewary, S., Dang, S. & Roos, B. a. Care management’s challenges and
opportunities to reduce the rapid rehospitalization of frail community-dwelling older
adults. Gerontologist 50, 451–8 (2010).
282. Cohen-Mansfield, J. Nonpharmacologic treatment of behavioral disorders in dementia.
Curr. Treat. Options Neurol. 15, 765–85 (2013).
283. Gitlin, L. N., Hodgson, N., Jutkowitz, E. & Pizzi, L. The cost-effectiveness of a
nonpharmacologic intervention for individuals with dementia and family caregivers: the
tailored activity program. Am. J. Geriatr. psychiatry 18, 510–9 (2010).
284. Deppermann, K.-M. Influence of age and comorbidities on the chemotherapeutic
management of lung cancer. Lung Cancer 33, S115–S120 (2001).
285. Extermann, M. Measurement and impact of comorbidity in older cancer patients. Crit. Rev.
Oncol. Hematol. 35, 181–200 (2000).
286. Yancik, R., Ganz, P. A., Varricchio, C. G. & Conley, B. Perspectives on Comorbidity and
Cancer in Older Patients: Approaches to Expand the Knowledge Base. J. Clin. Oncol. 19,
1147–1151 (2001).
287. Kodner, D. L. The quest for integrated systems of care for frail older persons. Aging Clin.
Exp. Res. 14, 307–13 (2002).
288. Callahan, C. M. et al. Effectiveness of collaborative care for older adults with Alzheimer
disease in primary care: a randomized controlled trial. J. Am. Med. Assoc. 295, 2148–57
(2006).
289. Cherry, D. L. et al. Interventions to improve quality of care: the Kaiser Permanente-
alzheimer’s Association Dementia Care Project. Am. J. Manag. Care 10, 553–60 (2004).
290. Geldmacher, D. S. & Kerwin, D. R. Practical Diagnosis and Management of Dementia Due
to Alzheimer’s Disease in the Primary Care Setting: An Evidence-Based Approach. Prim.
Care Companion CNS Disord. 15, 1–20 (2013).
291. Morgan, D., Innes, A. & Kosteniuk, J. Dementia Care in Rural and Remote Settings: A
Systematic Review of Formal Paid Care. Maturitas 68, 17–33 (2011).
292. Callahan, C. M. et al. Redesigning systems of care for older adults with Alzheimer’s
disease. Health Aff. (Millwood). 33, 626–32 (2014).
217
293. Konetzka, R. T., Spector, W. & Limcangco, M. R. Reducing hospitalizations from long-term
care settings. Med. care Res. Rev. 65, 40–66 (2008).
294. Alzheimer’s Society of Canada. First Link Program. at
295. McAiney, C., Hillier, L. & Stolee, P. First Link Saskatchewan Demonstration Project: Final
Evaluation Report. 1–117 (2011).
296. Molloy, D. W. et al. Systematic implementation of an advance directive program in
nursing homes: a randomized controlled trial. JAMA 283, 1437–44 (2000).
297. Molloy, D. W. & Guyatt, G. H. A comprehensive health care directive in a home for the
aged. Can. Med. Assoc. J. 145, 307–311 (1991).
298. Dohan, D. & Schrag, D. Using navigators to improve care of underserved patients: current
practices and approaches. Cancer 104, 848–55 (2005).
299. Kwan, J. L., Bell, C. M., Morgan, M. W. & Stewart, T. E. Inpatient Patient Navigator Program
Reduces Length of Stay. The Harvard Business Review 1 (2013).
300. The World Health Organization. Dementia: A Public Health Priority. 1–112 (2012).
301. Canadian Healthcare Association. Home Care in Canada: From the Margins to the
Mainstream. Canadian Healthcare Association 1–147 (2009).
302. Morris, J. et al. Resident Assessment Instrument (RAI): RAI-MDS 2.0 User’s Manual,
Canadian Version. 1–317 (2010).
218
APPENDIX A, CHAPTER 3
A.1 Objective of Study Protocol
The objective of this review is to assess the evidence for variations in physician practice
patterns in caring for people with dementia in a systematic way.
Objective 1: Are there variations in clinical services provided to seniors?
Objective 2: Are there variations in referral services for additional care (e.g. home care,
adult day care, respite care etc.) provided to community dwelling seniors?
A.2 Method
A.2.1 Study Design
This systematic review will include all quantitative study designs including experimental,
quasi experimental (pre-post studies, interrupted time-series) and observational (case
control, cohort, cross-sectional) studies. While randomized control trials are the optimal
design, it is not anticipated that there will be many conducted in this field, therefore
quasi-experimental and observational studies will also be included. Editorials,
commentaries and letters that do not report information on original empirical results
will not be included. Qualitative studies will also be indexed, but separated.
A.2.2 Population/Comparators
General physicians providing care to seniors 60 years or older diagnosed with dementia,
both primary and secondary diagnoses. Specialist providing dementia care will also be
evaluated separately if possible. If it is possible to investigate sub-populations of
patients, I will do so (i.e. Alzheimer's disease, Parkinson's disease, Frontal Temporal
Dementia, Vascular Dementia) as well as differentiate between home care and
residential care patient population when possible.
A.2.3 Interventions
Any clinical service or combination of clinical services detailed as appropriate for
dementia care management (hereafter referred to as ‘dementia care processes’). These
are all services outlined by national and international guidelines such as physical
checkup, history taking, formal memory testing, neuropsychological testing, disclosure
of diagnosis, imaging, blood workups, prescriptions for dementia, depression,
behavioural challenges, non-pharmaceutical therapeutics, environmental interventions,
counseling and referrals to specialists. Referral to services for additional care or support
219
will also be included such as home and community services, respite care, support
groups, Alzheimer’s association, adult day centres etc.
A.2.4 Outcomes
Any quantitative clinical outcome measures (e.g. rates, prevalence, odds ratios, rate
ratios etc.), both unadjusted and adjusted.
A.2.5 Inclusion and Exclusion Criteria
Inclusion criteria:
All English language studies
Peer reviewed and grey literature
Published prior to March 1
st
, 2012
Quantitative experimental, quasi-experimental and observational studies
Exclusion criteria:
Published studies not reporting on original empirical results
Conference abstracts, book chapters
Studies focused exclusively on prescription patterns of acetyl-cholinesterase
inhibitors, antipsychotic or psychotropic medications
Studies focused on caregiver patterns of care
Physician practice patterns assessed for variability analyzed at a patient-level
Presenile dementia or age 60 and below
Non-primary care setting
Only specialist practice patterns
Vignettes (i.e. Only assess actual services provided)
A.3 Review Process
A.3.1 Search Method
I will employ a two-stage search strategy for this review. In consultation with a research
librarian, I will develop a search strategy for the following electronic subscription
databases for primary studies: Web of Knowledge database (ISI/Thompson), PubMed,
Science Direct, MEDLINE (Ovid), PsychINFO and EMBASE (Ovid). To expand my search
for grey literature, I will do a preliminary search in Google Scholar for the first 500
results. The search strategy will be adapted to each database based on its thesaurus or
Medical Subject Headings. An individual search with each of the search terms will also
be conducted to ensure that no relevant citations are missed. The concept of practice
patterns and dementia will be combined with filters for language (English only) and
publication dates where possible.
220
To increase the probability of capturing all potentially relevant studies, I will perform a
secondary search by looking at the reference list of key studies that met my inclusion
criteria.
A.3.2 Data Collection and Extraction Procedure
After consultation with a librarian, the literature search will be conducted by reviewer 1
(SS). The search results will be screened by title for potential inclusion in the review
(SS). All studies that meet inclusion criteria by title will be included for abstract review.
Abstracts will be extracted by reviewer 1 (SS) and the abstract review will be conducted
independently by both reviewers (SS and JP). Results will be compared and
disagreements resolved by discussion until consensus is reached.
Reviewer 1 (SS) will then obtain full-texts of all articles that met inclusion criteria after
abstract review and store them in the reference manager Mendeley’s shared online
folder, accessible only by both reviewers. Full-texts will then be reviewed to determine
if studies continue to meet inclusion criteria. Particular attention will be paid to the
participants (type of physicians assessed), level of analysis (patient or physician-level),
methodology, dementia care processes assessed and the outcome reported. Only
studies indentified by both reviewers will be included for data extraction. If consensus
cannot be reached, a third reviewer can be requested for independent assessment.
A data extraction tool will be developed by reviewer 1 (SS) to extract all relevant data.
This tool will be tested by both reviewers independently on three studies drawn at
random from the final list of included studies. Once the tool’s effectiveness has been
agreed upon by both reviewers, it will be used on the remaining studies. Independent
data extractions will be conducted by each reviewer who will then meet to compare and
discuss the extracted data until consensus is reached.
A.3.3 Critical Appraisal
Reviewer 1 (SS) will conduct a critical appraisal of the results of each study. The quality
of the studies may not be assessed if there are a limited number of studies. The
following criteria will be considered:
The magnitude of difference in the outcome measurement across studies
The potential for bias (classification, selection, measurement bias)
Publication year and the availability of best practice guidelines
221
APPENDIX B, CHAPTER 4
Table B.1 - Cross-walk mapping a sample of drug identification numbers to
corresponding anatomical therapeutic chemical groups for medications of interest
Drug Type Drug Identification Number Anatomical Therapeutic
Chemical Group
Acetyl-cholinesterase
Inhibitor (excluding
Memantine)
02232043, 02232044
02269457, 02269465
02242115, 02242116
02242117, 02242118
02266733, 02316943
02316951, 02316978
02344807,02348950
02349116, 02366487
N06DA02
N06DA03
N06DA04
N06DX01
Memantine 02260638, 02320908
02321130, 02324067
02344807, 02348950
02349116, 02366487
N06DX01
Antipsychotics 00346780, 00346799
00346802, 00346810
02229250, 02229269
02229277, 02229285
02299038, 02299046
02299054, 02299062
02025280, 02025299
02025302, 02025310
N05AH01
N05AH03
N05AH04
N05AX08
Trazodone 00579351, 00579378
00702277, 00824135
01937227, 01937235
02053187, 02053195
N06AX05
Benzodiazepines 00012874, 00013285
00013293, 00013757
00020915, 00020923
00020931, 00134325
00231363, 00295698
00402680, 00402737
N05BA01
N05BA02
N05BA04
Antidepressants 00010448, 00353868
00353876, 00425265
00010464, 00010472
00010480, 00021504
N06AA01
N06AA02
(Full table available upon request)
222
APPENDIX C, CHAPTER 5
Table C.1 – Baseline characteristics of individuals newly diagnosed with dementia in
2009/10 who died during study period
Characteristics Patients with Dementia
(n=1,328)
Age in years (mean+ SD) 85.90 + 6.61
69-80 21.8%
81-90 53.5%
91+ 24.7%
Sex*
Male 51.0%
Female 48.9%
Income Quintile (mean+ SD)* 2.78 + 1.52
1- Lowest income 24.5%
2- Lower income 24.4%
3- Medium income 19.9%
4- Higher income 15.0%
5- Highest income 15.2%
Multiple chronic disease (not including dementia)
0 chronic diseases 3.6%
1 chronic disease 12.3%
2 chronic diseases 21.6%
3+ chronic diseases 62.4%
Number of major ADGs (mean+ SD) 3.72 + 1.27
Treatment Prevalence of major chronic diseases
Hypertension 71.0%
Depression 25.0%
Arthritis or Osteoarthritis 35.6%
Cancer 42.0%
Congestive heart failure 38.2%
Diabetes 28.1%
Cardiovascular disease 25.1%
Stroke 12.2%
Chronic obstructive pulmonary disease 11.9%
Chronic kidney disease 10.8%
Liver disease 0.2%
Health Authority*
Interior 28.8%
Fraser 24.3%
Vancouver Coastal 26.7%
223
Characteristics Patients with Dementia
(n=1,328)
Vancouver Island 3.76%
Northern 3.3%
*Missing data: Sex Unknown=1, Income Unknown=11, Quintile HA Unknown=1
224
Table C.2A – Percentage of individuals who died during study period who received laboratory testing and imaging processes
outlined in guidelines stratified by patient characteristics
Laboratory Testing Imaging
Patient Characteristics Complete
hematology
profile
Serum
electrolytes
Serum calcium Serum glucose Thyroid
stimulating
hormone
B12 vitamin Head
computed
tomography
Age
69-80 63 60 26 39 36 19 66
81-90 58 55 19 38 38 20 57
91+ 50 50 14 26 30 15 46
Sex*
Female 53 52 19 32 35 17 53
Male 60 57 20 39 36 19 60
Income Quintile*
Lowest 52 50 15 33 34 17 57
Low 58 55 19 37 37 20 57
Medium 57 55 22 30 35 13 53
High 59 57 22 43 38 21 63
Highest 62 58 21 37 34 22 50
Multiple Diseases
0 diseases 37 21 5 11 24 16 50
1 disease 43 34 14 22 24 15 46
2 diseases 57 57 19 39 43 23 51
3+ diseases 61 60 22 38 36 18 61
Health Authority*
Interior 54 52 14 27 31 15 44
Fraser 60 59 22 41 41 20 60
Vancouver Coastal 54 50 17 37 35 17 65
Vancouver Island 62 61 27 37 36 24 62
Northern 56 51 28 39 31 13 44
225
Table C.2B – Percentage of individuals who died during study period who received prescriptions and dementia care management
processes outlined in guidelines stratified by patient characteristics
Prescriptions Counselling and Specialist Referral
Patient
Characteristics
AChI Antipsychotics Trazodone Antidepressants Benzodiazepines examination
in-office
counselling
in-office
Referral to
specialist
Age
69-80 9 24 6 35 26 10 25 4
81-90 14 24 6 29 24 9 20 5
91+ 7 28 5 25 23 9 11 2
Sex*
Female 11 24 6 30 26 9 18 2
Male 12 26 6 28 22 10 19 5
Income Quintile*
Lowest 13 28 6 25 22 10 18 2
Low 9 21 6 32 26 10 17 3
Medium 9 23 7 34 25 10 20 2
High 12 21 2 27 23 10 20 6
Highest 13 33 8 28 26 3 20 6
Multiple Diseases
0 diseases 8 26 3 21 18 3 11 5
1 disease 11 23 4 18 18 9 11 6
2 diseases 14 28 6 24 25 13 20 3
3+ diseases 11 25 6 26 34 8 20 3
Health Authority*
Interior 11 31 8 35 33 8 19 2
Fraser 12 15 3 27 17 8 16 5
Vancouver
Coastal
13 30 6 28 22 8 17 4
Vancouver Island 9 22 5 28 27 14 27 3
Northern 13 26 8 23 16 13 8 3
226
Table C.3A - Proportional odds ratio for factors associated with receiving guideline-consistent care from multivariate modelling
Adjusted Odds Ratio (95% Confidence Interval)
Laboratory Tests Imaging
Independent
Variables
Complete
hematology
profile
Serum
electrolytes
Serum calcium Serum glucose Thyroid
stimulating
hormone
B12 vitamin Head computed
tomography
Age
69-80 1.00 1.00 1.00 1.00 1.00 1.00 1.00
81-90 0.85 (0.67 - 1.08) 0.89 (0.73 - 1.08) 0.81 (0.70 - 0.93)† 0.63 (0.53 - 0.74)‡ 0.75 (0.63 - 0.90)† 0.82 (0.71 - 0.94)† 0.85 (0.74 - 0.98)*
91+ 0.48 (0.36 - 0.65)‡ 0.68 (0.52 - 0.89)* 0.48 (0.39 - 0.60)‡ 0.35 (0.28 - 0.44)‡ 0.48 (0.38 - 0.61)‡ 0.61 (0.50 - 0.75)‡ 0.52 (0.42 - 0.63)‡
Sex
Female 1.00 1.00 1.00 1.00 1.00 1.00 1.00
Male 0.91 (0.7 - 1.19) 0.93 (0.75 - 1.16) 0.76 (0.65 - 0.89)‡ 0.99 (0.82 - 1.21) 0.67 (0.54 - 0.82)‡ 0.74 (0.63 - 0.86)‡ 1.24 (1.06 - 1.46)†
Income Quintile
1- Lowest 1.00 1.00 1.00 1.00 1.00 1.00 1.00
2- Low 1.14 (0.92 - 1.43) 1.17 (0.96 - 1.43) 1.26 (1.09 - 1.46)† 1.16 (0.99 - 1.37) 0.95 (0.80 - 1.12) 1.26 (1.09 - 1.45)† 1.03 (0.89 - 1.19)
3- Medium 1.18 (0.94 - 1.49) 1.19 (0.97 - 1.46) 1.11 (0.96 - 1.30) 1.13 (0.95 - 1.33) 1.01 (0.85 - 1.21) 1.13 (0.98 - 1.31) 0.90 (0.77 - 1.04)
4- High 1.12 (0.88 - 1.42) 1.01 (0.82 - 1.23) 1.11 (0.95 - 1.30) 1.07 (0.90 - 1.27) 1.07 (0.89 - 1.28) 1.08 (0.93- 1.25) 1.00 (0.86 - 1.17)
5- Highest 1.33 (1.04 - 1.71)* 1.23 (1.00 - 1.51)* 1.22 (1.04 - 1.42)† 1.10 (0.92 - 1.30) 1.18 (0.98 - 1.42) 1.44 (1.24 - 1.68)‡ 0.94 (0.81 - 1.09)
Multiple Chronic
Diseases
0 diseases 1.00 1.00 1.00 1.00 1.00 1.00 1.00
1 disease 1.17 (0.83 - 1.64) 1.72 (1.32 - 2.24)‡ 1.05 (0.83 - 1.33) 1.07 (0.83 - 1.38) 0.95 (0.71 - 1.25) 0.91 (0.72 - 1.15) 1.07 (0.86 - 1.34)
2 diseases 1.28 (0.92 - 1.78) 1.98 (1.54 - 2.56)‡ 1.03 (0.82 - 1.29) 1.14 (0.89 - 1.46) 0.95 (0.72 - 1.24) 0.84 (0.67 - 1.05) 1.20 (0.97 - 1.50)
3+ diseases 1.44 (1.05 - 1.98)* 2.83 (2.21 - 3.63)‡ 1.21 (0.98 - 1.51) 1.19 (0.95 - 1.51) 0.92 (0.70 - 1.20) 0.80 (0.64 - 0.98)* 1.51 (1.23 – 1.87)‡
Health Authority
Interior 1.00 1.00 1.00 1.00 1.00 1.00 1.00
Fraser 1.92 (1.54 - 2.41)‡ 1.79 (1.48 - 2.17)‡ 2.06 (1.79 - 2.38)‡ 2.07 (1.77 - 2.41)‡ 2.17 (1.84 - 2.56)‡ 1.79 (1.56 - 2.05)‡ 1.30 (1.13 - 1.49)‡
Vancouver
Coastal
1.04 (0.85 - 1.27) 0.91 (0.76 - 1.09) 1.45 (1.25 - 1.69)‡ 1.32 (1.14 - 1.53)‡ 1.23 (1.05 - 1.43)† 1.31 (1.14 - 1.51)‡ 1.63 (1.41 - 1.87)‡
Vancouver Island 1.64 (1.30 - 2.07)‡ 1.55 (1.27 - 1.89)‡ 2.44 (2.09 - 2.83)‡ 1.76 (1.50 - 2.07)‡ 1.75 (1.47 - 2.08)‡ 2.16 (1.87 - 2.50)‡ 1.47 (1.27 - 1.70)‡
Northern 2.01 (1.18 - 3.41)† 2.05 (1.31 - 3.23)† 3.80 (2.85 - 5.07)‡ 2.24 (1.57 - 3.21)‡ 2.37 (1.60 - 3.51)‡ 2.43 (1.82 - 3.25)‡ 1.55 (1.16 - 2.06)†
‡ - P value <0.001 † - P value <0.01 * - P value <0.05
227
Table C.3B – Proportional odds ratio for factors associated with receiving guideline-consistent prescriptions and dementia care
management processes, from multivariate modelling
Adjusted Odds Ratio (95% Confidence Interval)
Prescriptions Management
Indepen-
dent
Variables
AChI Antipsychotics Trazodone Antidepressant Benzodiazepine† Complete
examination in-
office
Individual
counselling in-
office
Referral to
specialist re.
Dementia
Age
69-80 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
81-90 0.82 (0.71 - 0.94)† 1.08 (0.93 - 1.26) 0.81 (0.64 - 1.03) 0.75 (0.65 - 0.86)‡ 0.89 (0.76 - 1.04) 0.74 (0.64 - 0.85)‡ 0.72 (0.63 - 0.82)‡ 0.61 (0.48 - 0.77)‡
91+ 0.33 (0.26 - 0.42)‡ 1.14 (0.91 - 1.42) 0.76 (0.53 - 1.08) 0.53 (0.43 - 0.66)‡ 1.01 (0.81 - 1.26) 0.41 (0.30 - 0.49)‡ 0.41 (0.33 - 0.50)‡ 0.24 (0.14 - 0.40)‡
Sex
Female 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
Male 1.15 (0.73 - 1.00)* 0.92 (0.82 - 1.16) 0.93 (0.71 - 1.21) 0.66 (0.56 - 0.77)‡ 0.69 (0.57 - 0.82)‡ 1.15 (0.98 - 1.35) 0.79 (0.68 - 0.93)† 1.30 (1.03 - 1.65)*
Income
Quintile
1- Lowest 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
2- Low 1.15 (0.99 - 1.33) 0.92 (0.79 - 1.07) 0.77 (0.60 - 0.98)* 0.93 (0.80 - 1.07) 0.94 (0.80 - 1.10) 0.93 (0.79 - 1.08) 1.09 (0.95 - 1.26) 1.04 (0.80 - 1.34)
3-
Medium
1.28 (1.10 - 1.49)‡ 0.85 (0.72 - 0.99)* 0.83 (0.65 - 1.06) 0.92 (0.79 - 1.07) 0.97 (0.83 - 1.15) 0.97 (0.83 - 1.13) 1.15 (1.00 - 1.34)* 1.24 (0.96 - 1.60)
4- High 1.20 (1.03 - 1.41)* 0.89 (0.75 - 1.04) 0.77 (0.59 - 1.00)* 0.90 (0.77 - 1.04) 0.99 (0.84 - 1.18) 1.09 (0.93 - 1.27) 1.20 (1.03 - 1.40)* 1.19 (0.92 - 1.56)
5- Highest 1.34 (1.14 - 1.56)‡ 0.79 (0.67 - 0.93)† 0.71 (0.54 - 0.92)† 0.85 (0.73 - 1.00)* 0.97 (0.82 - 1.15) 1.02 (0.87 - 1.20) 1.31 (1.13 - 1.53)‡ 1.51 (1.18 - 1.95)‡
Multiple
Chronic
Diseases
0 diseases 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
1 disease 0.96 (0.76 - 1.21) 1.06 (0.82 - 1.37) 1.24 (0.78 – 1.98) 1.64 (1.26 - 2.13)‡ 1.21 (0.91 - 1.60) 0.76 (0.60 - 0.97)* 1.04 (0.83 - 1.30) 0.80 (0.54 - 1.16)
2 diseases 0.94 (0.76 - 1.17) 1.20 (0.93 - 1.53) 1.37 (0.88 - 2.15) 2.29 (1.78 – 2.96)‡ 1.41 (1.07 - 1.85)* 0.88 (0.70 - 1.11) 1.06 (0.85 - 1.33) 0.85 (0.60 - 1.22)
3+
diseases
0.60 (0.49 - 0.74)‡ 1.21 (0.95 - 1.53) 1.79 (1.17 - 2.75)† 3.87 (3.03 – 4.95)‡ 1.81 (1.39 - 2.36)‡ 0.85 (0.67 - 1.05) 1.11 (0.89 - 1.37) 1.021(0.72 - 1.43)
Health
Authority
Interior 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
Fraser 1.61 (1.40 - 1.85)‡ 0.78 (0.68 - 0.91)‡ 0.78 (0.61 - 0.99)* 0.89 (0.77 - 1.02) 0.83 (0.71 - 0.96)† 1.31 (1.13 - 1.52)‡ 0.94 (0.82 - 1.07) 1.90 (1.50 - 2.42)‡
Vancouver
Coastal
0.89 (0.77 - 1.03) 0.81 (0.70 - 0.94)† 1.10 (0.87 - 1.39) 0.81 (0.70 - 0.93)† 0.69 (0.59 - 0.81)‡ 1.19 (1.02 - 1.38)* 0.83 (0.72 - 0.96)† 1.83 (1.43 - 2.34)‡
228
Adjusted Odds Ratio (95% Confidence Interval)
Prescriptions Management
Indepen-
dent
Variables
AChI Antipsychotics Trazodone Antidepressant Benzodiazepine† Complete
examination in-
office
Individual
counselling in-
office
Referral to
specialist re.
Dementia
Vancouver
Island
1.17 (1.01 - 1.36)* 0.83 (0.71 - 0.97)* 0.80 (0.62 - 1.03) 0.85 (0.73 - 0.98)* 0.89 (0.76 - 1.05) 1.64 (1.41 - 1.92)‡ 1.22 (1.06 - 1.41)† 0.77 (0.58 - 1.04)
Northern 0.83 (0.62 - 1.11) 0.99 (0.73 - 1.34) 0.61 (0.35 - 1.08) 0.81 (0.60 - 1.08) 0.89 (0.65 - 1.22) 0.89 (0.65 - 1.22) 0.95 (0.71 - 1.25) 0.36 (0.17 - 0.79)‡
‡ - P value <0.001 ‡ - P value <0.01 * - P value <0.05
229
APPENDIX D, CHAPTER 6
Table D.1 - Longitudinal health care use
Year (-1) Year 0 Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 Year 8 Year 9
2000/01 2001/02 2002/03 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 2010/11
Total cohort at beginning of
year (n) 6,876 6,876 5,788 4,828 3,970 3,281 2,683 2,153 1,719 1,372 1,105
Cohort survival (%) 100% 100% 84% 70% 58% 48% 39% 31% 25% 20% 16%
HOSPITAL USE
No hospital stay (n) 5,356 2,619 4,098 3,675 3,032 2,591 2,192 1,780 1,435 1,151 919
Had hospital stay (n) 1,520 4,257 1,690 1,153 938 690 491 373 284 221 186
Cohort with hospital stay (%) 22% 62% 29% 24% 24% 21% 18% 17% 17% 16% 17%
Mean number hospital
admissions for those
hospitalized (mean ± SD) 1.56 5.1 2.2 2.38 2.18 2.26 1.73 1.71 2.02 1.79 1.82
Mean LOS at hospital for those
hospitalized (mean ± SD)
23.10 ±
40.42
41.71 ±
49.52
28.52 ±
41.23
23.08 ±
33.93
20.74 ±
29.67
22.81 ±
34.03
20.28 ±
30.88
18.44 ±
25.66
18.20 ±
23.25
17.80 ±
27.80
18.79±
28.61
TRANSITIONS
Experienced a transition (n) 1,750 4,570 2,096 1,493 1,231 1,090 598 464 487 283 196
Cohort who experienced
atleast one transition (%) 25% 66% 36% 31% 31% 33% 22% 22% 28% 21% 18%
Cohort who experienced 3+
transitions (%) 4% 17% 6% 4% 5% 4% 2% 3% 3% 2% 2%
Total transitions (mean ± SD) 0.41 ±
0.88
1.34 ±
1.40
0.62 ±
1.05
0.51 ±
0.98
0.52 ±
1.00
0.52 ±
0.98
0.34 ±
0.80
0.34 ±
0.82
0.43 ±
0.81
0.30 ±
0.69
0.27 ±
0.69
230
Table D.2 – Principal diagnoses for patients newly diagnosed with dementia in hospital only
Principal Diagnosis Proportion of
Patients
(n=2499)
Dementia 14.95%
Fracture of femur 8.96%
Pneumonia 4.80%
Heart Failure 3.68%
Stroke 2.92%
Rehabilitation care 2.76%