HOW DOES NEED COUNTThere is a trend across Canada to transfer more responsibility for health care to the regional level. Manitoba has recently established Regional Health Authorities (RHAs) to manage the delivery of health care services. Consensus is growing that levels of funding for these regional authorities should be based on a measure of their relative need for health care resources. There is, however, no consensus about how to define need for health care services or how to measure it.
IN ALLOCATIONS FOR HEALTH CARE?
This project by the Manitoba Centre for Health Policy (MCHP) was undertaken to consider approaches for factoring need into funding allocations to different regions. It preceded-and contributed to-a project to develop a funding methodology for the Regional Health Authorities.
The two projects are similar in some ways and different in others. This project attempted to consider how one would factor need into regional funding for a single sector - physician services - where we have fairly comprehensive data. This is a particularly challenging area because the distribution of physician services does not appear to be well matched with need. Several approaches to factoring in a population's need for services were reviewed. This report, therefore, provided useful information for the challenges of building needs-based funding formulas for RHAs.
By contrast, the RHA funding methodology project considered the issue of allocating funding across a much broader set of services, including some services for which there is very little data about how they are used by individual residents of Manitoba. On the advice of a committee of experts, MCHP developed a framework for funding that took into account limitations in data and drew on an understanding of what we had already learned from earlier studies, including this one. This latter project is called "Needs-based Funding for Regional Health Authorities: A Proposed Framework" and will be released later in 1997.
What Determines Need for Health Care?To understand what we learned from this project, it is important to place the study in context. The literature on population health suggests that a broad range of factors are associated with a population's need for health care, but three categories are most important. They are:
the demographic mix of people according to age and gender; socio-economic characteristics, specifically those that are risk factors for poor health, such as unemployment, education levels, proportion of single-parent families and percentage of people living in poor housing; the health of residents, often measured by premature mortality (death) rates.
The MCHP study examined these factors in various combinations in relation to visits to physicians that took place outside of hospitals in 1993/94.
Age and GenderWe began by looking at age and gender because they are among the most important factors associated with need for health care. For example, pregnant women and infants need more care than other people; and the elderly need more than the young. To understand how requirements differ because of these factors, we divided the province into 58 Physician Service Areas (PSAs) based on grouping physicians in relation to the population they serve. The patient population in each PSA was itself divided into 21 age and two gender categories. By calculating an average utilization pattern for each category, we could estimate the average number of physician visits that each category in each PSA would require. Given the number of people in each category, we could then add these up to estimate the number of physician visits each PSA would need based on its population mix of age and gender.
But when we looked at the allocations based on these demographic calculations, we found the formula had a serious problem. Areas in which more people, on average, die younger are said to have a higher premature mortality and can be thought of as having higher health needs. It was discovered that the age/gender formula allocated areas with lower premature death rates (and presumably better health) more physician visits per capita than they were currently using. On the other hand, it gave areas with higher mortality rates (and poorer health) fewer visits. Therefore, using age and gender alone, needier areas would have had their services reduced. Based on these findings, the age/gender formula was rejected as the sole measure of need.
Socio-Economic Characteristics and Premature MortalityMCHP has already done considerable work in linking socio-economic characteristics to health status. The Socio-Economic Risk Index (SERI), developed by MCHP, shows a close relationship between poor health and high unemployment, low education levels, high proportion of single-parent families and poor housing. The study next combined the age/gender formula with the socio-economic index to see how they would work in combination to allocate health care resources according to need. The "fit" was relatively good. Socio-economic factors are clearly related to patterns of physician visits, and the effect varies over the different age/gender categories.
When the formula was tested against premature mortality, the result this time went in the right direction-areas with higher death rates and poorer health were allocated more physician visits, and healthier areas got fewer. But the differences in visit levels were not uniformly related to differences in premature mortality.
The study concluded that an additional adjustment was required. The best model is one that uses age/gender and socio-economic factors for a first-stage adjustment, and then adds an additional adjustment to take into account differences in premature mortality rates across the PSAs. The greater the difference between an area's premature death rate and the provincial mean rate (the mean rate is in the middle), the greater the adjustment. Areas with higher than average premature mortality would be allocated additional resources, and those below would have their allocation reduced.
In addition, there was an aspect of socio-economic characteristics that required more attention. Dropping out of high school will not necessarily affect one's health immediately. Nor will being unemployed for a month or two. Instead, these factors tend to influence health over a period of years. Therefore, just adding up the proportion of the population that has low education levels or is unemployed on the day of a Census may not be as accurate an indicator of health status as we would like.
To address this issue, and also to overcome any problems with sampling errors and rounding off statistics in the Census, the model was redeveloped to provide a longer-term view of the population. The most recently available Census data (1991) were combined with data from the previous Census (1986) to achieve a rolling average over time for an area. This broadened the base of analysis and allowed us to build what should be a more valid and stable model over time.
What Did We Conclude?The model developed for this project requires an enormous amount of data. Because of this, it cannot at this time be extended to many other aspects of care where data are sparse (for example, home care). Where adequate data are not available, simpler approaches, such as that recommended for the RHA funding methodology, must be used.
In spite of these limitations, MCHP believes development of the model for this project represents a major step forward. It proposes a tentative solution to the complex question: How many more health care resources should be allocated to a region for each factor that increases need (for instance, the proportion of young children or the proportion of poor people who live there)?
The model holds promise for the future as we collect more comprehensive data on which to consider needs-based funding allocations. In the present, where data are available, it provides a useful framework for understanding whether services are provided and used according to need and what would be required to move in this direction. For instance, it provided a useful tool for supporting needs-based planning for physician services.
The most important implication of applying the model to physician services was that there should be a significant reallocation of resources to fund physician services to regions outside of Winnipeg. That is because northern and rural regions were shown to have higher relative needs for health care that are not matched by correspondingly higher levels of spending on physician services.
For the longer term, this preliminary description of a novel and relatively complex approach to allocation of health care resources has highlighted some important issues and provided an opportunity for discussion and refinement. Hopefully, the discussion will cast further light on identifying the fairest and most accurate method of funding based on need for health care.
Summary written by Norm Frohlich, Charlyn Black and Cheryl Hamilton, based on the report: Issues in Developing Indicators for Needs-Based Funding: by Norman Frohlich and Keumhee Chough Carrière.
Questions About This Report?
Return to Report Summaries Directory
Web page design by RJ Currie: email@example.com