Furthermore, unresolved conceptual difficulties in measuring disability can make it difficult to compare and comprehend the implications of existing data. Some clinical measures of functional status, for instance, which were developed to follow individual patients in the course of treatment, are too detailed and require too much medical expertise and/or equipment to be used in a broad-based population survey (Guralnik et al., 1989a). Some issues are difficult to measure through respondents. For instance there can be confusion between one's ability to carry out a specific task and whether one gets assistance in carrying it out (Wilson and Drury, 1984).
Just as role expectations influence whether a functional limitation becomes a disability (Nagi, 1965), statistical measures of functional limitation can be affected by what people expect or are expected to be able to do. For instance, if health promotion messages convince an older woman that she should walk more often, a mild case of arthritis that did not limit her activity before could interfere with walking on some days. Despite a probable improvement in health status, she would legitimately then be counted by the NHIS among those with activity limitations. A similar effect explains part of the apparent increase in disability in the 1970s among middle-aged men. Improved chronic disease screening, increased opportunities for disability compensation, and changing societal norms about early retirement allowed many men to enter medical treatment and to retire earlier than had been possible, probably increasing their health prospects and quality of life. In the official statistics, however, it appears that disability has increased (Wilson and Drury, 1984).
Given these limitations, it is clear that comprehensive measures of health status and quality of life are needed to understand the full complexity of disability and the factors leading up to it. Such measures have been developed, but they have been applied primarily in clinical studies. Research on their extension to population-based surveys is underway (Erickson et al., 1989).
As was apparent in the life table analyses, there are very few data extant on transitions in the disabling process. Although one can estimate on a cross-sectional basis the relationship between chronic conditions and activity limitation in the NHIS, and between different measures of activity limitation and disability in the SIPP, it is not possible to say much about the transitions from particular chronic conditions to particular functional limitations to different types of disability, for example, work disability. To develop efficient prevention programs it is critical to know the likelihood and rate of a large number of these transitions and their associated risk factors.
Finally, it is apparent that the available data focus on the chronic health conditions, not on the underlying processes and events that lead to these conditions and ultimately to disability. For instance, intentional and unintentional injuries are a major cause of the mobility limitations that are so