Biological plausibility

In a recent review in Science, Oeppen and Vaupel (1) documented the steady linear growth in life expectancy over the past 160 years, increasing by approximately 2.5 years for each decade between 1840 and 2000. Oeppen and Vaupel argue that there is no evidence that the rate of increase is slowing down; that is, there are no demographic data on which to estimate a "maximum" human life span achievable at some future date. If these same trends continue, then average life span for both women and men will exceed 100 years later in this century.

A commonly used test of an epidemiologic finding is biological plausibility—does the reported finding "make sense," given our current understanding of how the body works? This is not an infallible test. We can sometimes be misled by its application. But a lack of biological plausibility tends to stimulate a rigorous search for potential biases that might have produced a spurious association. In similar fashion, demographic analyses and projections should have clinical plausibility.

Much of the public discussion about the aging boom proceeds from the unspoken assumption that the old people of tomorrow will be pretty much the same as the old people of today. When projections are made of 4 million United States centenarians in 2050, the discussions seem to assume that those 4 million people will be just like the 80,000 centenarians alive today. To give one example, projections for future prevalence of Alzheimer's disease and other chronic illnesses account for projected increases in life expectancy but assume no change in age-specific incidence of the disease.

Such assumptions are not clinically plausible. Life expectancy increases because people are healthier. What does "healthier" mean? I think most clinicians would assume that a population becomes healthier because the prevalence of acute and chronic diseases decreases, for whatever reasons. This, in turn, produces longer lives. Thus, as life expectancy increases, individuals at any given age—say age 75—will look younger than did 75-year-olds of an earlier era.

Since 1960, life expectancy has increased by approximately 10 years. One should then assume that the average 75-year-old today looks like the average 65-year-old of 1960. Similarly, if we want to picture the 4 million centenarians in 2050, we should think of our current patients in their late 80s, while people in their late 80s in 2050 might resemble those who are 75 today.

Everything I have stated above is a massive oversimplification. While it is an oversimplification, I think it explains at least 80% of the relationship between increasing life expectancy and patterns of health and disability. Robine and Michel have provided the tools to move beyond that oversimplification, and similar oversimplifications by other authors, to try to come to a more complete understanding of how increasing life spans might affect rates of morbidity and disability. What Professors Robine and Michel (2) have done in their thoughtful discussion of population aging is to systematically work through how different clinically plausible theories would affect patterns of disability, and then test those theories against the available evidence. They have identified four elements that affect the disability/mortality relationship. This should assist further discussions. What I found particularly interesting is their emphasis on cultural determinants of health and disability. Both health and disability have a subjective component, and are thus affected by beliefs and expectations. These vary regionally and temporally, as well as by individual characteristics. Thus, "good health" is an interaction of an individual's physiologic state with his or her own expectations of what good health should feel like. Forecasting health and disability then takes on some of the same uncertainties as economic forecasts, which depend so much on individuals' beliefs and behaviors. That subjective component means that, just as in economics, we will always be much better in explaining what has happened than describing what will happen.