We know the hurricane is coming. Now we are trying to decide just how fast the wind is blowing. The article by Robine and Michel (1) offers a useful and provocative synthesis of discrepant findings. As such, it makes an important academic contribution. The question is whether all this agonizing will affect policy decisions. Whatever the final shape of the morbidity and disability projections, they will be dwarfed in the United States by the demographic shifts of an aging baby boomer population. The storm is coming. We do not have to debate estimated wind speeds to know that we need to take aggressive action now. Hurricane commentators claim that knowing the storm's intensity can be critical in preparing for it. But the gale wind forecasts have done little to spur us to action. We are still lolling on the beach listening to weather reports.
Even if the changes in health status do yield relative reductions in disability, we will have many more disabled people to contend with. Not only will there be more people, but those who survive will have the opportunity to contract more disabilities (2).
The accuracy of forecasts can have important policy implications. For example, earlier analyses have shown that the differences in measures of disability used can have a substantial impact on the size of cost estimates for long-term care (3), but even there the differences were among big effects. A set of projections for the impact of medical miracles over the next several decades offers a frightening glimpse into the social havoc that an even-more-aged population could create (4).
The medical model of disability works through disease, but the social model emphasizes the role of the environment. The causal pathway here is not yet clear. A variety of factors can change in both contexts. Economic developments can encourage people to see themselves as disabled. Changing expectations can reset the threshold for disability. On the other hand, from a medical perspective, a more compelling case would require a clearer linkage between morbidity and disability. No one has yet argued that prevention goes directly to disability without passing through disease. However, the causal pathway has not been well elucidated, especially with regard to changes in either health behavior practices or better treatment. No consistent patterns can be seen, despite the dramatic changes in some areas. The decline in mild disability but not severe disability is consistent with the fall in heart disease, but what about the fall in stroke rates, which should reduce severe disability? On the other hand, the treatment for dementia, a major source of disability, has not changed much.
Those who want to make a case for more or better geriatric care, or even for better chronic care (5), might want to seek support in evidence of a shift in patterns of morbidity and disability; but these are complex causal models to build on the basis of finding simple (or not so simple) associations. A number of factors can reasonably compete for a share of the credit: lifestyle, public health, medical care, pharmacological developments, and affluence.
On the other hand, because disability is associated with earlier death, the effects on cumulative health expenditures are offsetting (6). Cast the other way, effective prevention that leads to healthier extended life expectancy may not cost more than an earlier disabled death.
Some philosophers of science take pains to distinguish between a theory and an explanation (7). The work offered here seems to fit more closely into the latter category. It is nonetheless useful as a device for accounting for what appear to be discrepant findings, but it is also conveniently flexible enough to accommodate these discrepancies. Indeed the authors recognize the tentative nature of their integrating theory. The concept of disability transition fits nicely with the original idea of demographic transition, but that model seems more an explanation than a theory. The research agenda should be designed not simply to describe and compare but to look for antecedents and to test pathways that can help to elucidate a credible causal pattern.