Direct vs. Indirect Measures of the Need for Health Care

The most precise and appropriate approach to identify need for medical care in RHA populations would be the detailed measurement of specific morbidity in these populations. An inventory of specific morbidity measures would include, for example, incidence rates of acute morbidity such as injury or respiratory and gastrointestinal infection, and prevalence rates of chronic disorders such as cardiovascular disease, diabetes and arthritis. At this time in Manitoba, data that describe these direct measures of morbidity and that have been validated for accuracy are not available for Regional Health Authority populations.

An alternate approach to the direct measurement of need for medical care would rely upon generic measures of health status rather than specific measures of morbidity. Examples of generic measures would include measures of perceived health status, measures of disability or measures of health status obtained from instruments such as the SF-36 or the recently developed Health Utility Index, which is included in the National Population Health Survey. Generic measures have the advantage of providing a common summary measure of health status for a population, thus capturing the impact of a wide range of diseases and illnesses. Generic measures have been shown to be sensitive to change over time in the health of populations. The primary obstacle to using these measures in a need-based funding formula is that they require data obtained from large-scale survey samples. Manitoba Health's recent commitment to substantially increase the Manitoba sample size for the 1998 National Population Health Survey will provide the opportunity to obtain Regional Health Authority estimates of need for health care based on direct measures of health status.

Indirect measures of need for medical care include mortality measures and measures of the social and economic circumstances of populations. Mortality measures are defined as indirect primarily because a relatively rare event occurring to a minority of the population is used to characterize the health of the entire population. Measures of the social and economic status of a population can be considered indirect indicators of health status, in that a large body of evidence demonstrates that at all stages of the age course, individuals with fewer economic, educational and social resources have poorer health status. Both mortality data and socioeconomic status measures are routinely available for the Manitoba population. Mortality measures can be developed for single year periods, while socioeconomic status measures, derived from the federal census, are available at 5 year intervals.

It is important to make one additional distinction between direct and indirect measures of need. Direct need indicators are measured at the individual level, and indirect measures at the population level measures of need based on direct indicators can be related at the individual level to measures of the use of health care services.

Methodological Issues in Selecting Measures of Need for Health Care

A population measure of need for medical care should have the following characteristics:


1. Transparency and Face Validity
   
   A need indicator should make sense, both to health care professionals and managers and to the general public.
   
   
2. Precision
   
   The precision of a need indicator is determined by two characteristics; 1) the degree to which the indicator can be measured reliably, and 2) the stability of the need estimate, which is a function of the prevalence of the characteristic in the population and the size of the sample from which the estimate is obtained.
   
   Measurement reliability focuses on the control of random measurement error. An example of random measurement error would be an error in the Regional Health Authority residential classification of a death, or the classification as a diabetic of an individual person without diabetes. In general, mortality data and many measures derived from the census are thought to have less measurement error than indicators developed from administrative health care records or self-reported survey responses.
   
   Estimates derived from very small samples of populations have very wide confidence intervals. Similarly, the confidence intervals around estimates obtained from population-based data sources are also determined by the size of the population: small populations have more unstable estimates than large populations. Finally, the estimates for rare events, such as infant mortality, are more unstable than conditions such as diabetes that occur more frequently.
   
   
3. Independence
   
   Indicators of need should not be susceptible to efforts to manipulate the data which might misrepresent the actual health status of the population.
   
   
4. Efficiency, Accessibility and Currency
   
   Finally, the measurement of need should reflect the efficient use of resources and should be based on accepted methodologies which can be understood by Regional Health Authority Boards, managers and constituencies. Consistent with the face validity criteria, the need measure should be as current as possible.