Something to be Thankful For

Back in February, I wrote that “some news is good news.” The news in question came from the Framingham study and it showed that the incidence of dementia had been falling in the US by 20 percent each decade since the 1980s. However, I was concerned about the generalizability of the finding since the total number of people identified as having dementia was 371—and they were all from Framingham, MA. Now another, larger study points to a similar decrease. That’s something to be grateful for this Thanksgiving season. 

Before we get too excited—or complacent—let’s consider a few caveats. The new study was based on data from the Health and Retirement Study, a large, nationally representative cohort study that’s been going on since 1992. It looked at the prevalence of dementia in all those in their survey who were over age 65 and lived either in the community or a nursing home in one or both of two sample years, 2000 and 2012. That turned out to be 21,057 people, of whom a total of about 2000 developed dementia. So far, so good: a bigger sample, drawn from the entire country, with a respectable sized group of individuals with dementia. 

But now for the problems. Dementia was diagnosed using a modified version of the Telephone Interview on Cognitive Status, a measure involving a 27-point scale. Sounds good, except that the full Telephone Interview on Cognitive Status is not an established way to diagnose dementia, and the abridged version used by the investigators is even less well established. And indeed, when compared to a different test which the authors regard as the gold standard, the telephone interview correctly classified dementia in 78 percent of respondents. The other 22 percent of people were either falsely diagnosed as having dementia or incorrectly assessed as normal. That’s worrisome. The point of the study was to compare the prevalence of dementia in 2000 and 2012, which is awfully difficult to do if you can’t accurately determine prevalence.

Still, the findings of a decline in prevalence from 11.6 percent in 2000 to 8.6 percent in 2012 (corrected for the change in age and sex distribution of the population) are consistent with those of the Framingham Heart Study and of a British study. They show a 24 percent decline in prevalence of the disease, despite an increase in obesity and diabetes during the same period. And, as with the earlier studies, increases in education and improvement in control of cardiovascular risk factors (high blood pressure, smoking, and diabetes) are associated with the fall in dementia.

So maybe it’s really true. Maybe the risk for each us individually is not quite as bleak as I have been suggesting in this blog. On the other hand, the projection is that by 2050, there will be 83.7 million people age 65 or older. If even 8.6 percent of them have dementia, as suggested in the current study, that’s over 7 million people. Unless we find a cure soon, which doesn’t seem terribly likely, we’re still going to be faced with an enormous public health problem.  

Beyond Doctoring

I’ve long been amazed by the legerdemain that went into deciding what Medicare will cover and what it won’t. I’m not talking about decisions made in the past decade about what procedures to pay for, by and large rational decisions that have been based on a careful analysis of the evidence supporting their efficacy. I’m talking about some of the most basic aspects of Medicare, such as its exclusion of long term care. Now I recognize that the main concern of those who crafted the 1965 legislation was to provide some kind of health insurance for older people without busting the budget. To achieve this end, they decided to distinguish between things that are medical (which Medicare would ostensibly cover) and things that are not (which it wouldn’t). What that distinction has meant is that housing, transportation, diet, and all kinds of other nominally social goods are off limits for Medicare coverage. A new study by Elizabeth Bradley and her colleagues at Yale shows just how arbitrary—and often counterproductive—such a conceptual divide actually is.

Following up on their groundbreaking work in which they showed that countries with higher social service spending relative to health care spending had better health outcomes, the study team compared the performance of the 50 states (and the District of Columbia) over a 10-year period, from 2000-2009. They defined the extent of each state’s investment in social services by calculating the ratio of social service plus public health spending (on education, income support, nutritional assistance, housing, transportation, and the environment) to the state’s total government health care spending (Medicare plus Medicaid). Then they examined the relationship between this ratio and eight health outcomes (including the percent of the population that is obese, has asthma, or has functional limitations, and mortality rates for heart attack, lung cancer, and diabetes). What they found is that states with higher ratios of social to health spending had significantly better health outcomes (in 7 of the 8 domains).

It's striking that the variability in spending on health care (as a percentage of GDP) across the states is considerable, ranging from less than 4 percent in Colorado, Utah, and Wyoming to nearly 10 percent in Maine, West Virginia, and Missouri. Likewise, the variability in spending on social services and public health is dramatic, going from about 12 percent to over 20 percent. The net effect is that the allocation of resources between social services and health care differs substantially from one part of the country to the next.

It’s a complicated study and I’m sure that methodology mavens will have a field day with it. But the attempt to assess the contribution of social supports to outcomes is so reasonable and the results are so striking that we have to take very seriously the idea that social factors are a major determinant of health and well-being. I’m convinced this is particularly true in older people, whose quality of life is at least as affected by where they live and their ability to find meaning in life as it is by their physical ailments. I suspect that this study is as important as work by Michael Marmot showing that health worsens as people descend the social ladder—not just because of income inequality, but also because of discrepancies in social status. If we want to foster good health, which the World Health Organization defines as “a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity,” we need to focus on relationships and housing as well as on drugs and devices. And for older people, that may mean user-friendly computers and better assisted living facilities rather than a left ventricular assist device or a new monoclonal antibody.

Some News is Good News

The Framingham Heart Study has been ongoing since 1948 and it continues to provide answers, or at least insights, into all kinds of important medical questions. The latest news from Framingham is that the incidence of dementia, the rate at which new cases develop, has been declining over a period of 30 years, and that’s good news indeed.

It’s not exactly news, as several other epidemiologicstudies have pointed in the same direction, but many of those studies were either methodologically flawed or, though suggestive, did not yield statistically significant results The Framingham Study has the great virtue of using consistent, standard diagnostic criteria for dementia; it has monitored the cognitive status of the people it’s been following since 1975 (with even more extensive monitoring beginning in 1981); and it makes use of a special “dementia review panel,” which includes a neurologist and neuropsychologist to evaluate the evidence in every case of possible dementia. Because the Framingham Study collects all sorts of other data as well, including blood pressure readings, cholesterol levels, body-mass index, and information about diabetes, heart disease, cigarette smoking, and education, the investigators are able to control for all these factors.

The conclusion: since 1977, there has been a decline in the incidence of dementia averaging 20 percent per decade. But there are a few caveats. This study identified a total of 371 cases of dementia. That’s it. The 371 cases were scrupulously identified from among just over 5000 study participants, but it’s nonetheless a fairly small number of people. And if you look at the rate at which new cases of dementia were identified by “epoch,” you will see that the rate went from 3.6/100 in the first epoch to 2.8/100 in the second period, to 2.2/100 in the third period, to 2.0/100 in the final epoch. That is, the rate has fallen steadily but the most dramatic decline took place years ago. Things seem to be leveling off.

And there’s another issue that’s worth dwelling on: if you separate out the cases of Alzheimer’s disease from the cases of vascular dementia, you find that the rate of new cases of vascular dementia has fallen significantly, but there has been no statistically significant fall in the rate at which Alzheimer’s disease develops. This is not entirely surprising as we have made enormous inroads in cutting the rate of heart disease, in large measure by controlling blood pressure, treating high cholesterol, and persuading people to stop smoking—or better yet, never start. Exactly the same risk factors affect the development of heart disease and vascular dementia: prevent one, and you’ll prevent the other. But we have made no progress at all in preventing Alzheimer’s disease, which is thought to be responsible for the majority of cases of dementia.

The good news—and it really is good news—is that the progress in preventing heart disease has spilled over into the prevention of vascular dementia. The not-so-good news is that most dementia isn’t due to vascular disease. We should still vigorously combat vascular dementia, which after all causes some cases of dementia, and which also interacts with Alzheimer’s disease, in those unlucky people who are afflicted with both, to produce symptoms that are worse than would be expected from the combination of the two conditions. But there’s still a long way to go.