Is There A Correlation Between BMI and Cognitive Decline?

A recent (and IMHO, fascinating) study published in the journal Neurology suggests that heavier people score lower on cognitive tests and have a higher risk of dementia later in life than their thinner counterparts.

I go into this post with the full knowledge that perhaps:
a) people will attack me just for drawing attention to this research
b) people will attack the quality of the research
c) people will get pissed at the results
d) people will get pissed at me
e) and various other permutations of people getting their panties in a wad.

That's their right, of course.

HOWEVER, this is my blog and I'll post what I please--and I happen to find this line of research both provocative and important. So, the content of this post is under the fold, proceed at your own risk.

The goal of these researchers' was simple: to determine whether body mass index (which is a ratio of your height to weight and, yes, does have its flaws) is associated with cognitive function and cognitive decline in healthy men and women. The reason they wanted to investigate this question is that previous links have been found between dementia and obesity, however they are not clear enough to provide a causal link between being overweight and cognitive problems. It is important to note that they used healthy participants whose only difference was whether they were of normal BMI or obese BMI (greater than or equal to 30). All subjects were free of dementia at the start of the study.

About 3300 French people participated in the study, 1660 men and 1576 women, who were either age 62, 52,42, or 32 at the study's start. During routine medical examinations (required by French employment law for all workers), participants were given cognitive tests. In addition, participants aged 62 were screened for cognitive decline. Data was collected in 1996, and then five years later in 2001.

Data Collection:Data was collected in the form of medical history (self-reported), education level, lifestyle/family information, occupation, exercise level, stress levels, blood pressure, medications, as well as a calculation of BMI. Four cognitive tests were used: a word-recall test (short term memory), the Wechsler Adult Intelligence Survey, a selective attention test, and a delayed free recall test of material learned earlier (long term memory). Baseline (1996) levels were compared to 5-year levels (2001), and subjected to chi-square and t tests to prove significance.

Results: (this is the part that some folks aren't gonna like)
At the baseline, cognitive performance was higher when BMI was lower, and this correlation persisted after adjusting for age, sex, education, physical activity, and region of residence. The delayed-recall task (long term memory) was the most highly associated with BMI, however the intelligence survey and word recall tests were all significantly correlated. These results persisted after adjusting the results for the other factors listed above. Higher BMI changes were also associated with poorer performance on the selective attention test at the follow-up timepoint.
"Whatever the test, we found a negative association between BMI at the baseline and the improvement of cognitive performance.....In this analysis, the association between BMI and decline in [the intelligence test] remained significant after multivariate adjustment."


The authors formulate several conclusions from their data:
1) Higher BMIs are associated with poorer cognitive performance in a linear way
2) Higher BMIs at baseline were associated with a higher level of cognitive decline

My thoughts are this: while correlations are not causations, this study is quite compelling because it is the first that has followed a large number of participants longitudinally on several cognitive tests, and that adjusts for a host of physical/social/economic factors which could have been considered confounds. Furthermore, this study included participants from the south of France which are overall much healthier than the general Westernized-country population. I would love to see this study replicated in the USA, although I would wager than a much larger impact would be seen---so in that case, these results might be considered too mild.

Why would a higher BMI mediate cognitive decline? There are a few hypotheses. One postulates a direct, negative action of excess fat tissue on neuronal tissue through adipocyte-produced chemicals (see Bray 2004). Leptin, for example, is thought not only to regulate food intake but also plays a role in the learning and memory process. This is supported by leptin knockout mice models, which have changes in the morphology of brain structures (like the hypothalamus) and cognitive impairment. Another hypothesis is that the atherosclerosis and vascular disease which often accompanies obesity has a negative impact on brain function. It is well established that cardiovascular disease is associated with obesity and dementia (see Decarli 2004). And also, diabetes (and insulin resistance) is associated with cognitive decline. Recently, a study demonstrated a pathology of neuronal plaques (a symptom of Alzheimer's) in insulin-resistant mice (see Ho 2004). In any event, the preponderance of evidence is becoming undeniable (although the human capability of denial does astound me): obesity is not good for your brain.

Source: Cournot et al. 2006. Relation between body mass index and cognitive function in healthy middle-aged men and women. Neurology. Oct (1), 67; 1208-1214.

More like this

From the Telegraph: The new five-year study of more than 2,200 adults claims to have found a link between obesity and the decline in a person's cognitive function. The research, conducted by French scientists, which is published in this month's Neurology journal, involved men and women aged between…
Healthy aging is characterized by a gradual decline in cognitive function. Mental processes such as attention, memory and the ability to process information are at their peak when people are in their 30s and 40s, but as we get older, we find it increasingly difficult to focus on relevant…
This is the second of a series that examines menopause, hormone therapy, and consequences of each. Today's installment looks at the makeup of common hormones on the market, physiological consequences of hormone loss and replacement, and the Women's Health Initiative (WHI) studies. A slight emphasis…
THE dangers of obesity are very well known. Being overweight is associated with an increased risk of coronary heart disease and stroke, the two leading causes of death in the Western world. Gout is more common in overweight people, with the risk of developing the condition increasing in parallel…

Well, for starters, I think this is exactly right, but the sample population IS French!
(Gallic shoulder shrug and puff of smoke...)

However, Mon Cherie, I have to agree, this just makes sense at an intuitive level... have you ever been to a Wal-Mart on a Saturday afternoon? Lots of "High BMI Indexes", and not a lot of High IQ quotients, if you know what I mean.

Reading between the lines, the report says to me that smarter people are thinner,(because they realize that too much fat is bad, and they are smart enough and care enough to keep off the extra calories and weight?) Dumb people on the other hand, either don't know or don't care, so, c'est la vie, eh?

Excuse me now, I have exceeded my snarkiness quota for the day, and must put my finger down my throat to maintain at least a semblance of my long-gone blinding intelligence.
(well, intelligent for an ex-buckeye anyway)

While I tend to agree (my impression over time is that overall health is positively correlated with overall mental capability, at least from what I've personally [unscientifically] observed and studies I've read about), I have to wonder what happens to the data when you factor out the especially high BMI's that were in the study (i.e. if you cut off the outliers, is the trend still as strong? Is "somewhat overweight" generally correlated with "somewhat reduced ability to memorize words"? )

Although I agree with the conclusion of the study purely based on my own observations and bias, I have to wonder what the actual numbers were. Are the numbers statistically significant and does the plot graph above show that. The vertical measure words(n) ranges from 2 to 16 while the BMI ranges from 15 to 45. Does the cognitive unit of measure differ greatly from 2 to 16? If you look at the trend line what is the drop off really saying? While the difference in BMI from 15 to 45 is huge (no pun intended)is the drop in cognitive function from say 9 to 7 really significant. I wonder what the results would show if you studied a group of people with a high level of cognitive function and then measured their BMI. Just food for thought (OK, that pun was fully intended)

Did they control for participation in the National Football League? That outfit is brimming with high-BMI but low-adipose specimens evincing enormous levels of cognitive impairment from an early age, which coaches exploit to strategic advantage.

I can't get access to the full article, but the correlation coefficients they report in the abstract are pretty small, except maybe for the heaviest people. I'm surprised they didn't look at other medical factors in their model (if, in fact they didn't). As you say, diabetes is associated with cognitive decline, and if an individual is very overwieght, they will eventually develop diabetes, assuming they live long enough. The data could therefore be explained just in terms of cognitive decline in diabetics, which would show up in the overweight population. As these measurements were taken at medical checkups, I'd also be surprised if data on diabetes and other conditions weren't available. It would be a much more informative study to have a regression model that included these factors in the first place, rather than hypothesizing about them in the discussion.

Of course, if I'm wrong about what exactly they did, please correct me - it just sounds like that from the abstract and your post...

As a fat smart guy (got the dissertation 50, working on losing it again), I wouldn't be surprised if it has something to do with loss of liver function due to fatty liver disease.

Obesity's probably infectious (an adenovirus; mouse models support this). So higher BMI could a proxy for poorer immune system. This would also make the person more susceptible to infections that affected cognitive ability directly or indirectly.

The figure you show is very weak support for any hypothesis. There are not enough measurements at the high BMI end (I have no access to the full article).

For your research, do not forget that France has a record high alcohol consumption. Alcohol has well documented direct effects on both BMI and cognitive abilities. This statistics might just measure a secondary correlation of alcohol effects.

The interesting question, once a correlation has been established, would be whether it's a causative effect or just that, for example, people who are genetically inclined to be overweight are also inclined to suffer from dementia. So in the first case presumably the overweight person might be able to reduce his/her risk by losing weight, if it's the latter case we wouldn't expect that to make any difference.

While the data may well offer a weak correlation coefficient, that *graph* looks like a puff of smoke with a line drawn through it!

By David Harmon (not verified) on 12 Oct 2006 #permalink

As a person-of-high-BMI who makes a living pushing word stuff around, I had to laugh at that ridiculous graph. Tell us another one.

By speedwell (not verified) on 13 Oct 2006 #permalink

The people who wrote this study must fat, because they're pretty stupid. That is the awesomest plot I have ever seen: a few outliers cause the entire trend. They just drew a line through an uncorrelated scatter.

"All subjects were free of dementia at the start of the study."

So how many subjects were demented after the study?

@ sam: My macaw would like to show you how well she can count, but she needs a few extra fingers as props. Can she borrow a handful of yours?