TCS Daily

Fat, Flabby and Forgetful?

By John Luik - May 16, 2005 12:00 AM

They're back! Not long after the furor caused by the Centers for Disease Control's revised "figures" which showed that not 400,000 but about 26,000 Americans died annually from obesity and that being overweight might well be protective against premature death, the fat police are at it again with their scary headlines about the dangers of fat. This time the story is perhaps even more dramatic since it is about how middle age obesity "causes" late life dementia. So the supposedly bad news isn't just that being fat will likely kill you before you reach old age, it's also that if by some chance you do survive an early death you can worry about spending your golden years crazy.

That's the startling news from gerontological epidemiologist Rachel Whitmer and colleagues at the Kaiser Permanente Division of Research whose study, "Obesity in middle age and future risk of dementia: a 27 year longitudinal population based study" was published in the British Medical Journal. Supported by funding from the US National Institutes of Health, Whitmer followed 10,276 people from their mid-forties, in 1964, until 2003. From 1994 until 2003, 713 of these individuals were diagnosed with dementia, including Alzheimer's disease. Of these 713, the prevalence of dementia was "significantly higher for those who were obese or overweight at mid-life." According to Whitmer, obese people (BMI>30) were 74% more likely than normal weight individuals to develop some from of dementia, while the overweight (BMI 25-29.9) were 35% more likely to develop it. Hence all those frightening headlines, earnest talk show musings, and very happy proponents of massive government interventions against fat. As Philip James, head of the International Obesity Task Force, told the Associated Press "This adds another major reason for concern about the obesity problem and it now unfolds yet another area where... we have to say, 'for God's sake, we better get cracking.'"

But before we get, as it were, cracking, perhaps the Whitmer study deserves a quieter and more careful reading in order to determine whether millions of middle-aged and overweight Americans -- roughly half of the middle-aged -- really are destined for an Alzheimer's cursed old age. To begin with, it is important to understand exactly what the study is, namely a small-scale, longitudinal epidemiological study that is being interpreted as saying that middle age + obesity=old age dementia. Such a study, however, is unable by its very nature to produce a cause and effect conclusion in which middle aged obesity causes old age dementia. That's because risk factor epidemiological studies -- which is what this is -- are statistical exercises, not laboratory experiments. They are not carefully controlled clinical trials in which two groups are identical except for one controlled variable. Rather they are purely observational studies in which the researcher does not control exposure to either the independent variable -- in this case obesity -- nor to the possibly dozens of confounding variables that might influence the conclusions which are drawn. As Robert Heaney of Creighton University writes "... since observational studies cannot adequately control for other factors -- known and unknown -- that may obscure or exaggerate the sought for relationships, they can never establish causal relationships...." These risk factor observational studies tell us nothing more than how likely it is that an association between two things, like being obese and having dementia, is real, as opposed to chance.

The world is full of associations, for example, between the number of births in certain European countries and the prevalence of storks, between the price of beer and the salary of priests in Chicago, between the wearing of shoes and the incidence of malaria. But that doesn't mean that these associations represent causal connections in which storks cause births or a lack of shoes causes malaria. As the authors themselves tell us "Obesity and overweight in middle age as measured by body mass index and skin-fold thickness were strongly associated with risk of dementia in later life." Notice those words, "strongly associated with risk." This isn't a study that presents any scientific evidence that being overweight or obese CAUSES dementia -- whatever the hysterical headlines might say. It is a study which has simply found that there was an association between midlife obesity and latter life dementia. Indeed, as in many instances of association, there might well be another risk factor that produces both midlife obesity and dementia.

Second, it is important to keep in mind just how small this study really is. In all there were 713 persons who were diagnosed with some form of dementia, 354 of whom were either obese or overweight. But 359 of those suffering from dementia were either normal weight or underweight, a fact that goes unnoted by the authors. In fact, even with the women, where the authors tell us there is a strong association between obesity and dementia, 237 subjects who had dementia were normal weight or underweight, while just 159 were overweight or obese, a very curious finding for such an allegedly strong connection between obesity and dementia.

Third, there are enormous gaps in this study. The only time the participants received anything like a comprehensive medical examination was from 1964-1973 when they were in their forties. From 1973 until 1994 we know nothing about these participants in terms of health status in general or more importantly their exposure to dementia risk factors. It is unclear from the study whether, for instance, the original health histories included family history for dementia, a factor which increases the risk for some forms of dementia fourfold. More importantly, at the time of their initial medical examination, none of the participants was assessed for cognitive functioning to determine whether they suffered from early onset dementia.

Fourth, risk-factor epidemiological studies are judged on how reliable their associations are -- in this case how strong is the linkage between obesity and latter dementia. The authors tell us that "obesity and overweight in middle age...were strongly associated with the risk of dementia", but this simply isn't the case. The strongest association that they found, between obese women and the risk of dementia (adjusted for mid-life and late life co-morbidity) was 2.07, which counts as a weak association. Most of their associations between obesity and risk of dementia were between one and two, which are considered very weak associations. In short, the statistical evidence that the connection between obesity and dementia is a real one as opposed to a chance finding is weak.

But that isn't the end of this study's troubles. Not only are their findings weak, they only apply to one sex -- women, though you have to read a considerable way into the study to discover this. In fact, if you didn't understand the language of epidemiology you would be unlikely to find this out at all. As the authors note quickly in passing: "overweight men had a non-significant 30% increase in risk" which means there was no increased risk of dementia for men who were overweight or indeed obese at middle age.

And this creates additional problems. One of the things that helps strengthen the case for a genuine association is what epidemiologists call biological plausibility. In other words there has to be some biologically compelling reason an alleged cause -- in this case mid-life obesity -- is associated with dementia only in women and not in men. The fact that the authors were unable to provide a biologically plausible reason obesity is not associated with dementia in men further weakens their case.

Biological plausibility also undermines the study's conclusions in another way. The authors controlled for such risk factors as high cholesterol, diabetes, hypertension, heart disease and stroke which are often linked with obesity and which some have suggested might be associated with dementia. Yet even after controlling for these conditions, there was still supposedly an association between obesity and dementia according to the authors. This means there is no recognized biologically plausible explanation for the link between obesity and dementia. And given that the link between obesity and heart disease, stroke and even diabetes is tenuous, the claim that obesity now somehow causes dementia is decidedly improbable.

Nor is the study's validity helped out by the fact that it is contradicted by much of the established research on dementia, particularly Alzheimer's disease. Take for example the work on risk factors for Alzheimer's disease. These risk factors include age, family history, gene mutations on chromosomes 21, 14, and 1, which are associated with early onset of the disease, while the gene ApoE4 has been identified with the more common late onset form of the disease. Additional risk factors include Down syndrome and head injury. Possible risk factors might include Type 2 diabetes, heart disease, high blood pressure and stroke. Notice that obesity is not a risk factor. Again, there is no indication in the study that the authors controlled for any of these established risk factors except for age, a fact which further undermines their alleged association. Some dementia research has suggested that elevated homocysteine levels might be a risk factor and that antioxidants might have a protective role. Both homocysteine and antioxidants are connected with life-style factors such as diet and physical activity yet the study had not information from participants on these factors.

More importantly, the evidence about the alleged connection between dementia and obesity and fat is extremely limited and contradictory. For example, the Personnes Agees Quid study in 2003 found no relationship between BMI and the incidence of dementia and the Rotterdam study found no increased risk of Alzheimer's with increased cholesterol levels. And the previous study that looked at weight and Alzheimer's disease in women that is cited by the authors found that there was no significant difference in BMI between women who developed Alzheimer's before age 79 and those who did not, nor any significant difference in BMI between women who developed other forms of dementia and those who did not. All of which further weakens the author's claims about the link between midlife obesity and latter dementia.

Finally, the lead author, Rachel Whitmer, published another article on dementia and so-called midlife risk factors in January of this year ("Midlife cardiovascular risk factors and risk of dementia in late life" Neurology, 2005) that uses the same population base of the Kaiser Permanente medical care program of northern California. Only in that study, she and her co-authors concluded that such things as hypertension, high cholesterol, diabetes and smoking "substantially increases risk of late-life dementia." That list is interesting in itself since it mixes real diseases -- hypertension and diabetes -- with mere risk factors like high cholesterol and smoking. It looks as if dementia is becoming a bit like smoking in that everything is somehow connected to it. Simply take your pick of the Whitmer studies as to whether it is obesity, smoking, diabetes, hypertension or high cholesterol that leads to late life dementia.

So just why did this study, which proves nothing about the relationship between obesity and dementia, receive so much fear-inducing media attention, both in the US and internationally, particularly right after the obesity numbers debacle? The answer is to be found in a combination of the general scientific illiteracy of the media and the determination by those who claim that obesity is killing us by the millions to prove their case. Whitmer and her colleagues make it easy for a deadline pressured reporter and her headline writer to do their work with such claims as "Failure to contain the present epidemic of obesity may accentuate the expected age related increase in dementia." And what could be more headline grabbing than linking America's favorite obsession -- weight -- with one of its greatest worries -- the risk of dementia, which is neatly summarized in the author's conclusion that "obesity in middle age increases the risk of future dementia...." Never mind the boring technical stuff of whether the study's data actually supports its claims, it is the claims that are useful. The result is another new and terribly useful "fact" not only to be endlessly and mindlessly repeated but, at the right moment, to be used to further the war on fat. And that's what makes all the difference.

John Luik is a health policy analyst currently writing a book about obesity in America.



TCS Daily Archives