|Background: The demand for palliative care in Canada is expected to grow in coming years due to a confluence of factors that include population aging, a more gradual dying process, emphasis on the early provision of palliative care, as well as the expansion of the scope of palliative care beyond cancer. In particular, there is emphasis on the provision of palliative care in home settings as individuals prefer to remain at home for as long as possible, and for health systems, the potential reduction in end of life care costs. The proportion of Canadian decedents who received palliative home care services prior to death is low, however, and may reflect an inadequate supply of palliative home care services resulting from the dearth of information required to support health system planning activities. Specifically, the need to understand the characteristics, service utilization patterns and costs of palliative home care clients has been emphasized in reports on palliative care in Canada.
Objectives: This dissertation sought to address information gaps on palliative home care in Ontario, Canada by: 1) characterizing palliative home care clients and examining their service utilization patterns; 2) estimating formal service-related palliative home care costs, informal care costs, and combined formal and informal care costs; 3) identifying predictors of formal, informal and combined care costs, with particular emphasis on the Palliative Performance Scale (PPS).
Methods: A sample of palliative home care clients was drawn from home care clients in Ontario assessed using the interRAI Palliative Care (PC) between 2011 and 2017 (n=68,731). For each client, their first interRAI PC assessment was selected and linked to home care referral information and service billing/payment records of up to 181 days post-assessment contained in Health Shared Services Ontario’s Client Health Related Information System. Descriptive statistics were used to characterize these clients and describe their patterns of palliative home care service utilization. A subsample of clients on service for five or more weeks post-assessment and with a PPS score was then drawn from the original sample (n=39,072). Mean weekly costs of formal palliative home care services over the first five weeks of service post-assessment were estimated for this sample, as were informal care costs and combined formal and informal care costs. Predictors of these costs were then identified using linear and logistic regression models. Candidate variables considered as predictors of cost were organized based on Andersen and Newman’s framework on health service utilization, and informed by existing literature and by one-on-one interviews with individuals familiar with the management or provision of palliative care in Ontario. These variables were obtained from clients’ assessment records that included PPS scores and interRAI PC scales and items.
Results: A large majority of palliative home care clients had a cancer diagnosis (85.0%) and over half had an expected prognosis of less than six months (60.7%), with half dying during the follow-up period. Almost all (93.4%) clients had at least some loss of function based on the Instrumental Activities of Daily Living-Activities of Daily Living (IADL-ADL) Functional Hierarchy Scale, and moderate to high health instability based on the Changes in health, End-stage Disease Signs and Symptoms (CHESS) Scale. Services were initiated within days of assessment and utilization of services was particularly great in the first week post-assessment. Essentially all clients received case management and nursing services, while the most frequently used services were for nursing and personal support. For clients who were discharged during the follow-up period, increasing service use could be observed with closer proximity to discharge.
The estimated mean weekly formal palliative home care cost over the first five weeks of service was $352.91 (2020 Canadian dollar [CAD]). Linear regression using generalized linear models found the PPS to be significant in predicting formal care costs and explained 29.4% of variation in costs. By comparison, interRAI scales and items corresponding to components of the PPS explained 31.2% of cost variation. Expanded models containing interRAI PC scales and items both related and unrelated to components of the PPS had an explained variance of 37.7%, and addition of the PPS to this model led to a small increase in the explanation of cost variation (39.7%). Logistic regression models used to assess the probability of being a ‘high’ cost palliative home care cost client performed well (c-statistic between 0.81 and 0.94) and identified similar characteristics to those identified as predictors of cost in linear regression. Across models, characteristics that were consistently significant in predicting formal care costs included clients’ region of residence (Local Health Integration Network [LHIN]), having a live-in caregiver, being designated as an end of life home care client, having an informal caregiver who reported being unable to continue in care activities, having family and friends who felt overwhelmed by the client’s illness, poorer functioning (PPS and IADL-ADL Functional Hierarchy Scale), greater health instability (CHESS), altered mode of nutritional intake, the presence of dyspnea, presence of fatigue, greater pain, need for and use of intravenous (IV) medications, and bladder and bowel incontinence.
Estimates of weekly informal care costs were between $565.24 and $693.55, and ranged between $918.15 and $1,046.45 for combined mean weekly formal and informal care costs for the first five weeks of service post-assessment (all 2020 CAD). As a share of the total combined care costs, between 61.6% and 66.3% of care costs were for informal care. The generalized linear model of informal care costs and interRAI PC scales and items explained 41.4% of variation in costs. Addition of the PPS to this model increased variance explanation to 42.1%. A large number of characteristics were identified as significant predictors of informal care costs, although clients’ marital status, LHIN of residence, poorer function (IADL-ADL Functional Hierarchy Scale), greater fatigue, presence of expressions of unrealistic fears, and greater bowel incontinence had particularly large effect sizes. The association between formal care costs and informal care costs was also examined and the two types of costs were found to be complementary. In regard to the combined mean weekly formal and informal care costs, variance explanation using interRAI PC scales and items was 51.7%, and increased to 53.3% with the addition of the PPS. A large number of characteristics were also found to be predictive of combined care costs. Characteristics with particularly large effect sizes included LHIN of residence, function, need for altered mode of nutritional intake, dyspnea, fatigue, and daily use of intravenous medications.
Conclusions: This dissertation provides a detailed description of palliative home care clients in Ontario, and their home care service utilization patterns and associated costs. The PPS appears to be a reasonable predictor of formal, informal and combined palliative home care costs. A number of other client characteristics were also identified as being predictive of these costs with substantial overlap in the characteristics identified for each type of cost. The majority of these characteristics can be categorized as illness (need) factors within Andersen and Newman’s framework on health service utilization indicating that cost variations are influenced primarily by clients’ clinical and/or functional needs. Caregiver characteristics, including indicators of distress, were also predictive of formal, informal and combined costs indicating the importance of caregiver capacity to provide care in the care of palliative home care clients. One application of findings from this dissertation may be to inform on future development of a case-mix system for palliative home care as variance explanation of palliative home care costs observed in this dissertation were high, suggesting that the classification of palliative home care clients into groups with relatively similar clinical characteristics and levels of resource utilization is possible.