Health Care Service Provision Over the Palliative Care Trajectory

Transcription

1 Health Care Service Provision Over the Palliative Care Trajectory by Lisa Masucci A thesis submitted in conformity with the requirements for the degree of Master of Science Department of Health Policy, Management and Evaluation University of Toronto Copyright by Lisa Masucci 2011

2 Health Service Provision Over the Palliative Care Trajectory Abstract Lisa Masucci Master of Science Department of Health Policy, Management and Evaluation University of Toronto 2011 Health system restructuring combined with the preferences of palliative care recipients to be cared for at home has lead to a shift in the delivery of care from the hospital to the home setting. An analysis was conducted on five main home-based palliative care health service components: home-based nurse visits, home personal support worker visits, home-based physician visits, ambulatory physician visits, and other ambulatory and home-based visits. First, we assessed the proportion of total cost associated with the main services at different time points over the palliative care trajectory. Second we examined the socio-demographic and clinical factors that predict the propensity and intensity of service use, using a two-part model. The results suggest that the greatest contributor to the total cost of home-based palliative care was personal support worker visits, followed by nurse visits. The regression analysis revealed that patient age as well as functional status most often predicted health service use. ii

3 Acknowledgments This thesis would not have been possible without the dedication, support, and mentorship of my thesis supervisor, Dr. Peter C. Coyte. I am very grateful for the trust he has placed with me in executing this thesis and the many things I have learnt from him throughout these years of work. I am indebted to Dr. Denise N. Guerriere for providing me with guidance and mentorship from the start of my program until the very end. I thank her for her insightful comments and providing me with an abundance of knowledge surrounding palliative care. Many thanks are also extended to Brandon Zagorski for the abundant statistical guidance and support, and Dr. Larry Librach for aid in the preliminary stages of this work. Lastly, I would like to thank my mom and dad for their support and encouragement throughout this process. iii

4 Table of Contents Table of Contents Acknowledgments... iii! Table of Contents... iv! List of Tables... vii! List of Appendices... x! 1! Introduction and Objectives... 1! 1.1! Study Objectives... 3! 2! Literature Review and Search Strategy... 6! 2.1! Literature on Palliative Health Service Use... 7! 2.1.1! Total Costs of Home-Based Palliative Care Services... 7! 2.1.2! Costs Reported on the Individual Components of Care... 11! 2.1.3! Cost of Service Use Over the Palliative Care Trajectory... 13! 2.1.4! Summary... 16! 2.2! Literature on the Predictors of Health Service Utilization... 17! 2.2.1! Predictors of Health Service Utilization... 18! 2.2.2! Summary of the Literature on the Predictors of Health Service Utilization... 26! 3! Research Design and Methods... 28! 3.1! Participant Recruitment... 28! 3.2! Data Collection and Questionnaire Design... 29! 3.3! Variables... 32! 3.3.1! Dependent Variables... 32! 3.3.2! Independent/Explanatory Variables... 38! 3.4! Study Population, Statistical Methods and Data Analysis... 40! iv

5 3.5! Statistical Methods and Data Analysis... 41! 3.5.1! Wilcoxon Signed-Ranks Test... 41! 3.5.2! Analyses of the Predictors of Service Use... 43! 3.6! Summary... 48! 4! Results... 50! 4.1! Participant Recruitment... 50! 4.2! Costs of Each Service Category... 52! 4.3! Predictors of Health Service Utilization 0-90 days until Death... 61! 4.3.1! Health Service Utilization Statistics... 61! 4.3.2! Propensity to Use Home-Based Personal Support Worker Visits... 63! 4.3.3! Intensity of Home-Based Personal Support Worker Visits... 64! 4.3.4! Propensity to Use Home-Based Nurse Visits... 65! 4.3.5! Intensity of Home-Based Nurse Visits... 66! 4.3.6! Propensity of Home-based Physician Visits... 67! 4.3.7! Intensity of Home-Based Physician Visits... 68! 4.3.8! Propensity of Ambulatory Physician Visits... 69! 4.3.9! Intensity of Ambulatory Physician Visits... 70! ! Propensity of Other Ambulatory and Home-Based Visits... 70! ! Intensity of Other Ambulatory and Home-Based Visits... 71! ! Summary of Results... 71! 5! Discussion... 73! 5.1! Total Costs... 73! 5.2! Physician Visits... 74! 5.3! Nurse Visits... 76! 5.4! Personal Support Worker visits... 77! 5.5! Proportion of Costs Over The Palliative Care Trajectory... 77! v

6 5.6! Predictors of Health Service use... 81! 5.7! Predictors of Home-Based Personal Support Worker Visits... 81! 5.8! Predictors of Home-Based Nurse Visits... 82! 5.9! Predictors of Home-Based Physician Visits... 84! 5.10!Predictors of Ambulatory Physician Visits... 86! 5.11!Predictors of Other Ambulatory and Home-Based Visits... 87! 5.12!Strengths and Limitations... 87! 5.13!Policy Implications... 90! 5.14!Directions for Future Research... 93! 6! Conclusion... 95! Tables and Figures... 98! References ! Appendices ! Appendix A ! Terminology ! Appendix B ! Ambulatory and Home Care Record (AHCR) ! Appendix C ! The Palliative Performance Scale (PPS) ! vi

7 List of Tables Table 1: Characteristics which may predict Health Service Use Table 2a: Frequency of Publicly Financed Ambulatory Service Use (n=135) Table 2b: Frequency of Publicly Financed Home-Based Service Use (n=135) Table 3: Service Costs of Ambulatory Visits Table 4: Service Costs of Other Ambulatory and Home-Based Visits Table 5: Characteristics of Patients and Caregivers Table 6: Number of Patients in Each Time Interval, Relative to Death Table 7: Mean Cost per Patient per Service Category over 30 days Table 8: Aggregate Unconditional Costs Per Patient Per 30 days Table 9: Wilcoxon Signed-Rank Test for Difference in Unconditional Costs between Time Intervals 0-30 and (n=31) Table 10: Wilcoxon Signed-Rank Test for Difference in Unconditional Cost between Time Intervals and (n=36) Table 11: Chi-Square Analysis Comparing 28 patients to the rest of the sample Table 12: Number of Reports of Service Use by Service Category in the Final 3 months of Life Table 13a: Summary Results of Full Logistic Regression Model Table 13b: Summary Results of Reduced Logistic Regression Model Table 13c: Summary Results of Full Ordinary Least Squares Regression Model Table 13d: Summary Results of Reduced Ordinary Least Squares Regression Model vii

8 Table 14a: Probability that a Patient has at Least One Home-Based PSW Visits 0-90 days Prior to Death: Logistic Regression, Full Multivariable Table 14b: Probability that a Patient has at Least One Home-Based PSW Visit 0-90days Until Death: Logistic Regression, Restricted Model Table 15a: PSW Ordinary Least Squares Regression of Predictors of Service Use for those Using Services 0-90 days until death, Full Multivariable Table 15b: PSW Ordinary Least Squares Regression of Predictors of Service Use for those Using Services 0-90 days until death, Restricted Multivariable Table 16: Probability that a Patient has at Least One Home-Based Nurse Visit 0-90 days Until Death: Logistic Regression, Full Multivariable Table 17a: Nurse Ordinary Least Squares Regression of Predictors of Service Use for those Using Services 0-90 days until death, Full Multivariable Table 17b: Ordinary Least Squares Regression of Predictors of Nursing Service Use for those Using Services 0-90 days prior to death, Restricted Multivariable Table 18a: Probability that a Patient has at Least One Home-Based Physician Visits 0-90 days Until Death: Logistic Regression, Full Multivariable Table 18b: Probability that a Patient had at Least One Home-Based Physician Visit 0-90 days Prior to Death: Logistic Regression, Restricted Multivariable Table 19a: Home-Based Physician Visits Ordinary Least Squares Regression of Predictors of Service Use for those Using Services 0-90 days until death, Full Multivariable Table 19b: Ordinary Least Squares Regression of Predictors of Home-Based Physician Service Use for those Using Services 0-90 days prior to death, Restricted Multivariable viii

9 Table 20a: Probability that a Patient has at Least one Ambulatory Physician Visit 0-90 days Until Death: Logistic regression, Full Multivariable Table 20b: Probability that a Patient has at Least One Ambulatory Physician Visit 0-90 days Until Death: Logistic Regression, Restricted Multivariable Table 21: Ambulatory Physician Visits Ordinary Least Squares Regression of Predictors of Service Use for Those Using Services 0-90 days Until death, Full Multivariable Table 22a: Probability that a Patient has at Least One Other Ambulatory and Home-Based Visit 0-90 days Until death: Logistic Regression, Full Multivariable Table 22b: Probability that a Patient has at Least One Other Ambulatory or Home-Based Visit 0-90 days Until Death: Logistic Regression, Restricted Multivariable Table 23: Other Ambulatory and Home-Based Visits Ordinary Least Squares Regression of Predictors of Service Use for those Using Services 0-90 days Until Death, Full Multivariable ix

10 List of Appendices Appendix A: Terminology Appendix B: Ambulatory and Home-Care Record (AHCR) Appendix C: Palliative Performance Scale (PPS) x

11 1 1 Introduction and Objectives Over the past decade, Canadians have witnessed a shift in the setting where health care services are delivered, with more resources being provided in the home setting. This change is partially attributable to the restructuring of the health delivery system such as hospital bed closures, limits to health care budgets, and the advanced use of technology within the home setting (MacAdam 2000). In addition, the preference of patients to be cared for within their own home rather than in an institutional setting has influenced this shift (MacAdam 2000). Designating financial resources to the community has not only resulted in a larger number of patients accessing palliative care services, but also an increase in the participation and collaboration of communitybased physicians and nurses in the care of terminally ill patients (Bruera, and Sweeney 2002). Home-based palliative care programs provide palliative support services through community care teams (Carstairs 2000). These teams consist of a combination of physicians, nurses, personal support workers, spiritual counselors, and/or case managers who work collaboratively to provide patient-centered care (see Appendix A for definition). In addition, informal caregivers, who are family and friends, may also aid in providing care. The purpose of palliative care is to improve the quality of life of patients and their families, through the prevention and relief of suffering by means of early identification, assessment and management (WHO 2010). The overall goals of home-based palliative care programs are to provide components of palliative care in the home setting and to allow patients to remain at home as long as they choose. Relatively little research has been conducted on the utilization and costs of home-based palliative care services. In order to develop standards of care across the provinces, variation in the use and cost of various professionals involved in the provision of home-based palliative care over the palliative care trajectory must be measured. The palliative care trajectory represents the time

12 2 from which a patient is enrolled in a palliative care program until death. This period of time is characterized by a reduction in physical performance, and a decline in the ability to engage in self-care (Murray et al. 2005). The symptoms often experienced by palliative patients include pain, bowel disturbance (e.g., nausea, constipation, vomiting, diarrhea), loss of appetite, dyspnea and confusion (Marshall et al. 2008). The Commission on the Future of Health Care in Canada reports that estimates of home-based palliative care service utilization in Canada, and costs associated with these services, are nonexistent. Two influential reports on the health care system in Canada, the Romanow (2002) and Kirby (2002) reports, have highlighted the need for home-based palliative care to be covered as an insured service under the Canada Health Act (Kirby 2002; Romanow 2002). The Canada Health Act (CHA) outlines five principles: public administration, comprehensiveness, universality, portability and accessibility. The provincial and territorial health care plans must comply with these principles in order to qualify for full federal funding through the Canada Health Transfer (1985). As these two reports highlight, home-based health services are not privileged under the CHA, and thus the principles of the Act do not apply. The provinces independently decide which palliative care services to finance and the amount (i.e., hours) of these services that are provided. In the absence of a national framework for the provision of home-based palliative care services, potential inter-provincial variations and inconsistencies in the frequency, organization, and delivery of palliative care exist. For example, a 2008 report on Cancer Control in Canada reported that only six out of the ten provinces and three territories had explicit policies concerning the provision of nursing and personal support worker services 24 hours a day, seven days a week. Further, nine out of ten provinces and three territories had policies that promoted a team-based approach to palliative care. Given these

13 3 inconsistencies, there is a need to evaluate palliative care service delivery in Canada (Miller, and Morrison 2008). The challenges that exist within the home care environment in Canada include the lack of a coherent national strategy, inadequate funding, lack of human resources, and caregiver burden (Carstairs 2005). Therefore, it is imperative to examine the variations in the provision of homebased palliative care services. With further information, provinces may be able to develop standards of care and improve the allocation of human health services. 1.1 Study Objectives The two main purposes of this study are to: 1) assess the proportion of total cost associated with the main services used by home-based palliative care patients, at different time points over the palliative care trajectory; and 2) examine the socio-demographic and clinical factors that predict the propensity and intensity of service use. This study will assess five main home-based palliative care service components: home-based physician visits, nurse visits, personal support worker visits, ambulatory physician appointments, and other home-based and ambulatory appointments. Further description and rationale for the consideration of these service components will be discussed in Chapter 3. The Andersen and Newman (2005) framework of health service utilization was used as a conceptual framework to guide the selection of the determinants of health service use. The Andersen and Newman (2005) framework considers both the societal and individual determinants of health service utilization. Within this framework, an individual s utilization of health services is a function of three characteristics: predisposing characteristics, enabling characteristics and needs-based characteristics. Predisposing characteristics are the social and cultural characteristics which exist prior to a patient s illness. These characteristics consist of

14 4 demographic (age, gender), social structural (education, occupation, social networks, ethnicity, and culture), and health beliefs (attitudes or beliefs about medical care, physicians, and disease). These characteristics can be seen as indirectly influencing health care use among patients. For example, patients in different age groups may have different severity of illness and thus different patterns of health care use. The second characteristic is classified as enabling, which considers the ability of a patient to obtain medical services. Enabling characteristics include family resources such as income level, health insurance status, and whether or not a patient has a regular source of care and the accessibility to that source of care. Enabling characteristics also include community factors within which the patient lives. For example, the availability of health care professionals and facilities as well as wait times may affect a patient s receipt of health services. Finally, the third category of this model is needs-based (or recently re-named illness level ), which is both the perceived and evaluated illness level of the patient (Andersen, and Newman 1973), and relates to patients perceptions of their own general health and functional status as well as how they experience symptoms of illness. Evaluated need is professional judgment about a patient s health status and his/her need for medical care. In addition to the Andersen and Newman (2005) framework for health service utilization, the selection of potential predictors of service use was also influenced by the availability of data and the empirical literature, which will be discussed in section 2.6. Table 1 outlines the characteristics that were chosen as potential predictors of the propensity and the intensity of service use in this thesis. The propensity of service use is defined as an individual s inclination or probability to use a certain service (i.e., whether an individual used a service or not). The intensity of service use is defined as the amount of a certain service a patient uses. Within this thesis, the intensity of service use will be represented as costs in 2010 CDN dollars and as a continuous variable.

15 5 Table 1: Characteristics that may predict Health Service Utilization Predisposing Characteristics Enabling Characters Needs-Based Characteristics! Age of patient! Age of caregiver! Sex of patient! Sex of caregiver! Socio-economic status! Living Arrangement! Functional status This chapter has outlined the main purpose and framework used in this thesis. Chapter two will review the literature addressing the costs of home-based palliative care, the variation in costs over time, and the predictors of the propensity and intensity of service use. Chapter three will describe the data and methods used in the empirical analysis. The study results will be presented in Chapter four. Chapter five will discuss the implications of the key findings as they relate to the literature, limitations of the study, as well as policy implications and areas for further research. Lastly, Chapter six will provide an overall conclusion to this thesis.

16 6 2 Literature Review and Search Strategy A systematic literature review was conducted to identify studies focusing on the use and cost of home-based services among palliative patients. The following databases were searched: PubMed, MEDLINE, EMBASE, Web of Science, International Bibliography of Social Sciences, CINAHL, The Cochrane Collaboration, and Google Scholar (period= ). The key search terms used were: palliative AND home visit(s); palliative AND home care AND costs; home care AND illness trajectories; home-based care AND palliative visit(s); end-of-life care AND home-visit(s); health care utilization AND palliative OR hospice care; home care AND resource utilization AND palliative; palliative care AND health service use. These search terms were also used in conjunction with the terms predictor(s) or determinant(s). In addition, several journals that were directly relevant to the thesis subject were also examined; these include Palliative Medicine, Journal of Palliative Care, Journal of Pain and Symptom Management, Health Economics, and the Journal of the American Medical Association. Grey literature was also searched on the Ontario Ministry of Health and Long-Term Care website as well as the Ontario Homecare Association website. Finally, a snowballing technique was implemented whereby references cited in the relevant articles were examined and included if they proved relevant to the research topic. This search strategy resulted in the identification of 1,499 studies pertaining to palliative care health service utilization and 320 studies focusing on home-based palliative care. Out of these studies, 30 examined the costs of home-based palliative care and after further review, 26 of these studies were selected for inclusion. Through the review of the references used in these studies as well as specific journals, another eight articles were selected for inclusion. The search for determinants or predictors of home-based palliative care service use resulted in the identification of 56 studies, with 13 reviewed and 10 of these 13 selected for inclusion. Two additional studies were identified from reference lists of retrieved papers.

17 7 2.1 Literature on Palliative Health Service Use The literature review is divided into the following sections: Section reports studies addressing the total overall costs of palliative service use; reports studies which examine the costs each individual component of care; reports studies which examine the costs of palliative service use over the palliative care trajectory; and section reports studies which measured the predictors of the propensity and intensity of health service use Total Costs of Home-Based Palliative Care Services The costs associated with home-based palliative care services have been assessed in studies which compare the costs of home-based palliative care services to standard or conventional care (Brumley, Enguidanos, and Cherin 2003; Hughes et al. 1992; Serra-Prat, Gallo, and Picaza 2001; Shnoor et al. 2007). Home-based palliative care programs are those which provide interdisciplinary care comprised of some variation of palliative care physician specialists, nurses, personal support workers, and/or family physicians. Standard or conventional care has been described in the literature as care provided in a hospital or ambulatory setting with various levels of home care being provided depending on patients needs. Within these studies, it has been found that home-based palliative care programs are a cost-effective alternative to conventional care provided within the hospital or ambulatory setting. Nine studies have examined the total cost of home-based palliative care services (Bruera et al. 2000; Brumley et al. 2003; Brumley et al. 2007; Chan et al. 2001; Dumont et al. 2009; Fassbender et al. 2005; Gomez-Batiste et al. 2006; Serra-Prat et al. 2001; Shnoor et al. 2007). Of these studies, the majority do not report the professional service costs separate from the total cost of health care use (Brumley et al. 2003; Brumley et al. 2007; Chan et al. 2001; Dumont et al. 2009; Gomez-Batiste et al. 2006; Serra-Prat et al. 2001; Shnoor et al. 2007). Thus, it is

18 8 impossible to establish within these studies what percentage of the total cost was attributable to services provided by each of the various professionals. In addition, two studies that do report the cost of the various service components (Bruera et al. 2000; Fassbender et al. 2005), do not examine the costs of all professionals involved in providing care. Each of these studies are outlined in the subsequent paragraphs. Of the nine studies, three were conducted in Canada (Bruera, and Sweeney 2002; Dumont et al. 2009; Fassbender et al. 2005). One of these Canadian studies assessed the total cost of homebased palliative care as well as the total costs of physician visits among a group of cancer patients of a new palliative program (Bruera et al. 2000). The researchers reported that the total public cost of care (which included the cost of acute-care facility visits, overhead costs associated with running the palliative care program, physician billings, and physician specialist billings) was reduced from $12,191,577 to $10,730,888 (CDN, 1996) following the introduction of an interdisciplinary team-based palliative care program. This reduction in total cost was attributable mainly to the reduced use of acute-care facilities (i.e., hospitalizations). In contrast, physician billings, which included both ambulatory and home-based visits, increased from $427,731 for all patients within the study to $537,342 (CDN, 1996) (Bruera et al. 2000). These costs do not represent the total cost of care, since only physician costs were included and costs of other health services, such as nurse or personal support worker visits were not considered. In addition, the costs of home-based physician visits and ambulatory physician visits were reported together, and therefore the costs attributable directly to physician home-based visits remain unknown. Similarly, in another Canadian study, Fassbender et al. (2005) examined the total cost of physician services per year for a sample of terminally ill cancer patients receiving care from a

19 9 home-based palliative care program. However, again, this study did not separate physician visits into ambulatory and home-based visits. The overall total cost of palliative care in the last year of life per patient was $28,093 (CDN, 2000). Physician visits accounted for 10% of this cost and home care services accounted for 6% of total costs. The remainder of costs were attributable to residential hospice care (8%), nursing home care (6%), hospital care (67%) and prescription medications (3%) (Fassbender et al. 2005). This study does not indicate the percentage of the total cost attributable to other professional services. Therefore, it is not known whether the utilization of physician services was low because of other types of professionals providing care. The third Canadian study prospectively evaluated the costs for patients with a terminal illness registered in a home-based palliative care program (Dumont et al. 2009). The mean total public cost was $13,146 (CDN), which included hospitalizations, ambulatory and home-based care, medications, and medical equipment. The mean total cost for home-based care per patient was $3,227 (CDN) (18.7% of cost/patient), whereas the mean total cost for ambulatory care per patient was $1,420 (CDN) (7.9% of cost/patient). This study provides some insight into the division of costs among health care settings; however, it did not provide an assessment of the costs associated with each of the professionals. As well, it does not assess the costs over the palliative care trajectory. Of the five remaining studies assessing total costs, two were conducted in Spain (Gomez-Batiste et al. 2006; Serra-Prat et al. 2001), two in the United States (Brumley et al. 2003; Brumley et al. 2007) and one in Israel (Shnoor et al. 2007). One of the Spanish studies prospectively assessed the total cost of home care services among palliative cancer patients over a 16-week period (Gomez-Batiste et al. 2006). The other Spanish study retrospectively assessed overall costs of home-based care and ambulatory care over a one month period (Serra-Prat et al. 2001). The

20 10 services that were examined in this study consisted of hospital stays, emergency visits, outpatient visits, and home-based visits; they did not separate out any specific health professional visits. The American study by Brumley (2007) provides the costs of home-based palliative care services for those diagnosed with chronic pulmonary disease (COPD), congestive heart failure (CHF), or cancer. This study prospectively assessed costs of those who were admitted to a palliative care program and died during the course of the study. The total cost of care was presented as a mean aggregate cost ($12,760) for physician visits, emergency department visits, hospital days, skilled nursing facility days and home-based palliative care; costs for individual components of care were not provided. Similarly, another study conducted by Brumley et al. (2003), provided the overall mean total cost of home-based palliative care ($7,990). In this particular study, the costs were based on the amount of care provided by various health care professionals. These costs, however, did not include facility and administrative/overhead costs in the calculations and therefore are very conservative estimates. The last study which assessed the total costs of care was conducted in Israel (Shnoor et al. 2007). This study retrospectively examined the overall costs of cancer patients registered in a homebased palliative care program in the last two months of life. The average overall per patient cost reported ($4,761) was based on the cost of medications, laboratory and diagnostic tests and ambulatory and home-based services. Other than physician visits, this study did not specify which professionals services were included in the total cost. To date, the empirical literature has devoted little attention to home-based costs provided by the various health professionals. Within all nine of the previously mentioned studies, the total costs were not separated by all the key service components. It is imperative to determine the amount of care being received by different professionals and in which setting of care. This information

21 11 will aid in determining how much of each service should be provided by health care administrators. In addition, these studies have not provided health service utilization information at various time points over the palliative care trajectory. The costs were reported for one time point and as a result, an understanding of the utilization behavior of cancer patients from the time points relative to death remains unknown Costs Reported on the Individual Components of Care Three studies (Guest et al. 2006; Johnson et al. 2009; Witteveen et al. 1999) reported palliative care costs by individual service components. One of these three studies was conducted in Canada and examined the costs of health service utilization for palliative cancer patients within the last year of life (Johnson et al. 2009). These patients were enrolled in a home care pilot program which included care by a palliative physician, palliative care coordinator, palliative care nurses, and a hospice and psychological support group. The aggregate family physician cost per patient for both ambulatory and home-based visits was $ ($2007, CDN). The total aggregate cost per patient for all types of physician specialties was $1, ($2007, CDN), making this the largest cost component. The second largest cost component was home-based personal support workers with a cost of $ ($2007, CDN). Home-based nursing visits cost per patient was $21.85 ($2007, CDN) making this category one of the least utilized services. This study did not separate ambulatory and home-based physician visits, and therefore, cost contributions between the two remain unknown. In addition, comparing the costs of home-based personal support worker visits and nurse visits to home-based physician visits is not possible. Witteveen and colleagues (1999) prospectively assessed the costs of home-based and hospital palliative care services among patients diagnosed with end-stage cancer in The Netherlands. Within this study, specific groups of cancer patients were sampled in that patients in the study

22 12 were receiving either parenteral administration of fluids or analgesics in the home. This study provided the aggregate cost of home-based care provided by family physicians, a team that consisted of medical oncologists, specialist nurses, hospital services (inpatient and ambulatory), and home-based nurse services. It was found that the home-based nurse services contributed to the majority of professional service costs within the home (67%). When examining the overall total costs of home-based and hospital care, 95% of total costs were made up of hospitalizations, laboratory and radiology services, and outpatient treatment. Information about this team was missing the involvement of personal support workers and palliative physicians. As well, general practitioners were trained to use the technology within the home, and it is unknown if they were able to provide the same psychological and emotional support as a palliative physician specialist trained within this area of care. This indeed would impact the frequency of physician service use within the home. A British study retrospectively examined the home-based utilization of general practitioner visits, palliative physician visits, and district nurse visits for cancer patients receiving opioid treatment (Guest et al. 2006). As well, ambulatory palliative physician visits, hospitalizations, general practitioner surgery visits, and the use of medications were also measured. The results indicated that the mean overall cost to the health care system varied by cancer type with colon cancer patients having the lowest costs and ovarian cancer patients having the highest costs. Home-based physician visits accounted for only a small component of aggregate costs, with very few patients using this service (range: 0 - <1% depending on cancer diagnosis) and hospitalizations accounted for the majority of total service costs (range: 35-77% depending on cancer diagnosis). The utilization of other services were reported as follows: surgeon visits (range: 6-15% of total cost), nurse visits (range: 8-12% of total cost), ambulatory palliative physician visits (range: 0-2%). The cost of home-based nurse visits was much higher than that

23 13 of home-based general practitioner visits. However, the researchers suggest that the low proportion of costs attributed to home-based general practitioner visits is a result of the underrecording of home-based visits within the database used. This study is limited in scope as it focuses only on individuals who are receiving opioids. Thus, this study may not be generalizable to those with cancer who are not receiving this class of medications. These three studies did not assess the costs of the all the key health professionals separately. Witteven et al. (1999), assessed the costs of home-based nurse visits, general practitioner visits, and hospital care (inpatient and ambulatory), however, they did not report the cost of personal support worker visits or ambulatory physician visits. Guest et al. (2006) reported home-based general practitioner visits, palliative physician visits, and nurse visits as well as ambulatory visits to a palliative physician, hospitalizations, general practitioner surgery visits, but did not examine personal support worker visits. Lastly, Johnson et al. (2009) examined personal support worker visits, nursing visits, and family physician visits (home-based and ambulatory); however, they did not provide costs for ambulatory and home-based physician visits separately Cost of Service Use Over the Palliative Care Trajectory Thus far, the literature reviewed herein has focused on the assessment of one particular time point in the palliative care trajectory. It is important to examine health service utilization patterns of patients at various time points over their trajectory in order to obtain a true understanding of the costs of care. This is essential because health administrators need to determine and ensure sufficient funds for service provision and because costs may vary over time. Three studies have assessed the costs of home-based palliative care at multiple time points over the palliative care trajectory (Maltoni et al. 1997; Mukamel, Bajorska, and Temkin-Greener 2002; Tamir, Singer, and Shvartzman 2007). Tamir et al. (2007) assessed the utilization of

24 14 health services by cancer patients at three time points leading up to death (one year, three months, and one month prior to death) and found that the costs of health services decreased as death approached. Hospitalizations made up the majority of total costs at each time point, with the mean total cost per patient being $4,414 (U.S.) (62% of total) in the last year of life, $2,083 (71% of total) in the last 3 months of life, and $849 (75% of total) in the last month of life. The mean total cost of physician visits also decreased closer to death with the mean cost per patient being $16 in the last year of life, $7 in the last 3 months of life, and $2 in the last month of life (Tamir et al. 2007). It is unknown however, whether these physician visits were at home or in the ambulatory setting. Conversely, the other two other studies (Maltoni et al. 1997; Mukamel et al. 2002) reported that the use of palliative care services increased as patients approached death. Mukamel et al. (2002) retrospectively examined variations in ambulatory and home-based services for elderly patients in the last three years prior to death. The researchers assessed a program that combined acute and long-term care for frail elderly who stayed in the program until death. The researchers examined three time points in the two years preceding death (1 month, 12 months and 24 months prior to death). The results indicated that utilization more than doubled in the last month before death ($5,882, US) compared with utilization at 12 and 24 months ($2,600, US). Utilization changed only minimally for periods far from death and beginning at the seventh month, total utilization began to exhibit an increasing trend. By examining the individual components of care, it is evident that most of the total increase in cost was attributable to an increase in hospital use, nursing home use, use of home care, social services, and primary physician care. For all of these services (except hospital care) there was an increase in costs by 30-50% in the last month before death relative to 12 months before death. Hospital care increased by almost 250% in the last month before death. Although this study provided the change in total cost and the

25 15 percentage change in the cost of individual components of care, it did not provide the cost of the individual components of care at any of the time points. As a result, it is difficult to determine the differences in cost between each component of care. As well, the study focused on a specific managed program that is geared towards frail elderly individuals with various ailments; therefore, it may not be generalizable to a population of those diagnosed with cancer. The second study by Maltoni et al. (1997) also reported that palliative care service use increased as an individual approached death. This Italian study reported the costs of a multidisciplinary home-based palliative care team comprising a medical oncologist, palliative specialist, nurse, psychologist, and voluntary assistant. Costs among cancer patients at two time points over the palliative trajectory (14 weeks prior to death and 1 week prior to death) were measured. The average cost per patient of a physician home visit increased from 60 (EUR, 1995) 14 weeks prior to death to 80 (EUR, 1995) one week prior to death. As well, the average cost per patient for nurse home visits increased from 39 (EUR, 1995) to 59 (EUR, 1995) during the same time points. This increase in the cost was statistically significant. The total average cost reported was (EUR, 1995) 14 weeks prior to death and (EUR, 1995) 1 week prior to death. This change in the overall cost was a 23.1% increase, which was found to be statistically significant. The researchers found that the increase in the total cost was mainly attributable to the intensification of nursing care, which accounted for 21.8% increase in the total cost from 14 weeks prior to death to 1 week prior to death. A possible explanation for this finding is that patients close to death need more symptom support, which is something that mainly nurses provide. These results, however, should be interpreted with caution as this study was conducted in Italy, which is known to have a very limited healthcare system that varies throughout the country. It is important to examine service use in Canada where the operation and coordination of healthcare teams may be different.

26 Summary Sections critically appraised the empirical literature that examined the utilization of healthcare services among palliative patients. Four studies compared the cost of standard palliative care to home-based palliative care (Brumley et al. 2003; Hughes et al. 1992; Serra-Prat et al. 2001; Shnoor et al. 2007). The majority of studies have assessed the total overall costs associated with health service use and do not report costs by the various components of care (Bruera et al. 2000; Brumley et al. 2003; Brumley et al. 2007; Chan et al. 2001; Dumont et al. 2009; Fassbender et al. 2005; Gomez-Batiste et al. 2006; Serra-Prat et al. 2001; Shnoor et al. 2007). It is imperative to assess the various components of care that are involved in the delivery of home-based palliative care in order to determine the degree to which services are being utilized, as well as, the relationship between the utilization of each service. In addition, among these studies, only three were conducted in Canada (Bruera et al. 2000; Dumont et al. 2009; Fassbender et al. 2005). Two of the Canadian studies (Bruera et al. 2000; Fassbender et al. 2005) examined the costs of the introduction of a new palliative home-based program, while one (Dumont et al. 2009) evaluated the costs of patients with a terminal illness registered in a palliative care program. Again, these studies do not examine the various components of care and therefore one cannot assess which services were utilized to a greater degree than others. More extensive research needs to be conducted on the costs of home-based palliative care in Canada to highlight the resource utilization trends within the Canadian health care context. Three studies have examined the various individual components of care (Guest et al. 2006; Johnson et al. 2009; Witteveen et al. 1999). Both Guest et al. (2006) and Witteveen et al. (1999) found that hospital costs comprised the majority of overall palliative care costs and that the cost of home-based nurse visits was higher than that of general practitioners. In contrast, Johnson et

27 17 al. (2009) found that home-based nurse visits made up the least utilized category. Within their study, the majority of costs were attributed to physician specialty visits. Again, these studies did not examine the utilization of all the key health professionals within the palliative care setting as well as the costs at various time points over the palliative care trajectory. This is important as utilization patterns may change as death approaches. Of those studies that did examine the utilization of palliative care services over time, two studies reported that the utilization of palliative care services such as hospital, nurse, and physician services increases closer to death (Maltoni et al. 1997; Mukamel et al. 2002) while one reported that the utilization of physician and hospital services decreases closer to death (Tamir et al. 2007). This area of research must be examined further in order to understand why these discrepancies exist. This next section will concentrate on a body of literature, which examined the predictors of health care service utilization. This section describes the Andersen and Newman theoretical framework and divides the literature into the predisposing, enabling, and needs based variables inherent within the framework. In addition, the literature is divided by predictors of the propensity and intensity of service use. 2.2 Literature on the Predictors of Health Service Utilization As described in section 1.1, the Andersen and Newman (2005) theoretical framework for health care utilization was used as a conceptual framework for this thesis. This framework served as a guide in the selection of relevant variables to be included in the analysis. Within this framework, there are three sets of characteristics: predisposing, enabling, and needs based which may effect health service utilization.

28 Predictors of Health Service Utilization Predisposing Characteristics The predisposing component includes the individual characteristics of the patient and caregiver which existed before the onset of the illness. Andersen and Newman (2005) propose that individuals with certain characteristics are more likely to use health services, despite these characteristics not being directly responsible for health service use. A predisposing characteristic that has been studied in the literature is patient age. Six studies have examined the relationship between patient age and the likelihood of using various palliative care services (Burge, Lawson, and Johnston 2005; Gagnon et al. 2004; Grande et al. 2006; Gray, and Forster 1997; Jakobsson et al. 2007; Miller et al. 2003). Grande et al. (2006) investigated the relationship between patient age and the likelihood of using palliative nurse specialist care among a sample of patients with various palliative diagnoses. Patients who received home-based palliative nurse specialist care were younger than those who did not receive this service. Similarly, a study assessing the likelihood of using specialist palliative home-based nursing services within the home, among individuals dying of cancer, found that patients who received these services were significantly younger than those who had not received this service (Gray, and Forster 1997). Conversely, the remaining four studies (Burge et al. 2005; Gagnon et al. 2004; Jakobsson et al. 2007; Miller et al. 2003) reported a positive relationship between these two variables. One study examining the probability of health service use during the last three months of life reported that as the age of the patient increased, the likelihood of having a home and hospital based general practitioner visit also increased (Jakobsson et al. 2007). The researchers suggest that those in the

29 19 older age groups experienced more problems and symptoms as a result of their chronic illness and old age. This study, however, did not separate home-based and hospital-based physician visits and examined predictors of utilizing these services together. Therefore, it is unknown whether age predicts physician visits at each site. In addition, a study by Burge et al. (2005) examined the likelihood of cancer patients having at least one home-based family physician visit. The results indicated that older age (75 years and older) was associated with an increased likelihood of having at least one home-based family physician visit (OR =1.41). The remaining two studies (Gagnon et al. 2004; Miller et al. 2003) have also demonstrated that there is a positive relationship between receiving any palliative care service and the age of the patient. Gagnon et al. (2004) examined the probability of receiving palliative care service use among a group of women diagnosed with breast cancer within the last six months of life. It was found that those in the older age groups (more than 70 years of age) were more likely to receive palliative care services compared to women who were years of age. As well, younger women (less than 50 years of age) were less likely to receive palliative care services compared to women years of age. Another study by Miller et al. (2003) examined the probability of receiving continuous hospice care which consisted of a minimum of 8 hours of home care in one day with care provided by a nurse for more than half of the palliative care period. The results demonstrated that patients younger than 65 had a lower probability of using continuous hospice home care than those aged 65 to 84. Patient age has also been examined within studies that measured predictors of the intensity of service use (Chan et al. 2001; Mukamel et al. 2002; Shnoor et al. 2007; Tamir et al. 2007). Tamir et al. (2006) found that patients in older age groups (above 65 years) had significantly lower costs compared to younger patients during the last year of life. This was found across all

30 20 three time points assessed (1, 3, and 12 months before death) and this difference was more pronounced as patients were closer to death. The researchers hypothesized that higher expenditure among the younger age groups may be explained by more aggressive oncology treatment being given to younger patients. Similarly, three other studies found that patients in older age groups had significantly lower costs than those in younger age groups (Chan et al. 2001; Mukamel et al. 2002; Shnoor et al. 2007). One of these studies examined the total cost of care and found that costs among patients 59 years of age or younger were much higher than those of patients 60 years of age or older (Shnoor et al. 2007). The second study found patients older than 85 years of age utilized less health care services overall than individuals 55 to 85 years of age at all three time periods examined (1,12, and 24 months before death) (Mukamel et al. 2002). The last study found that patient age was negatively related to the overall health care costs per observed day (Chan et al. 2001). Perhaps this is because with younger patients there is more of a focus on sustaining life, even despite the underlying premise of palliative care to provide psychological and emotional support as well as pain management. In addition to patient age, the age of the caregiver has also been reported to affect palliative health care utilization. Caregiver age has been found to be a predictor of receiving nursing care, as well as, palliative nurse specialist care. Grande et al. (2006) found that the age of the caregiver was negatively associated with the probability of utilizing both types of nursing care; the likelihood of service utilization decreased as caregiver age increased. The researchers postulated that younger caregivers may seek additional help due to life stage factors, such as having young children and being responsible for two separate households (their own and their family member s) (Grande et al. 2006). As well, younger caregivers may also be effective at lobbying for care. To the best of our knowledge, no palliative studies have been found which examine the caregiver age as a predictor of the intensity of health service use.