Wednesday, July 15, 2009

The way to a man's heart is .... through simpler nutrition labeling?

A headline today from the BBC: "Lower IQ 'a heart disease risk'". This is a report of a paper in the European Heart Journal in which researchers looked at the life experience of more than 4000 US Vietnam vets and concluded that IQ alone explains 20% of the risk of heart disease. They controlled for known risk factors such as smoking, diet, exercise and socioeconomic status (heart disease risk is already well-known to be higher among people with lower income and education levels, and smoking, diet and exercise are associated with SES as well), and still found an IQ effect.

Clearly, however, if this association is real, low IQ is not the direct 'cause' of heart disease--unless someone truly believes that 'brain' genes affect heart function such as clogging of arteries! Instead, IQ is a confounder, a measured variable that is not directly causal but is correlated in the sample with some unmeasured truly causal exposure factor, be it genetic or otherwise.

But confounder for what? The lead researcher suggests that the problem may be that people with low IQ may have trouble heeding health advice. It needs to be simpler and easier to understand.

"For instance, we often read about how some types of alcohol are good for you while others, or even the same ones, are not. The messages can be difficult to interpret, even by knowledgeable people."

This is an interesting quote because it (presumably unwittingly) points out that health advice is often contradictory, not to mention highly cultural. And thus impossible to interpret meaningfully, no matter your IQ. Some alcohol is good for you--except when it's not. Hard to think how you could make that into advice that's easier to follow. The other interesting bit about this quote is that the alcohol he's talking about is red wine, more generally the drink of choice among the middle and upper classes.

That aside, what could IQ be a marker of? Many people suspect IQ scores to be nothing but cultural markers--indeed, it's well-known that they vary by race, and have been changing rapidly over recent time, for largely cultural reasons. So IQ would be a marker of race, but the BBC story doesn't mention race, because the paper itself doesn't, either. Unbelievable!

One can hardly imagine an aware US epidemiologist who would not think right off the bat that race may well be the real risk factor here, for which IQ is a correlated marker. Heart disease risk is higher among African Americans, and presumably, if this particular study group follows known trends in IQ scores, African Americans would be more heavily represented in the lower IQ scoring/higher heart disease risk group. Thus, IQ is merely a marker for race here, and the risk associated with race.

But, let's run with the idea that intelligence might actually be a risk factor for heart disease. Numerous studies have been done to try to identify genes 'for' IQ, but with little replicable success. But let's suppose there are genes 'for' intelligence--indeed, if the IQ/heart disease association is real, mapping studies looking for genes associated with heart disease should have at least identified IQ genes. (But determining whether genes are IQ genes or not would be difficult because 70-80% of genes, no matter where else they are expressed or what their known function(s), are expressed in the brain--are they all 'brain genes' or genes for intelligence?). Indeed, genes 'for' poor circulatory function may affect ability to study and learn, so the reverse should also occur: IQ mapping should pick up cardiovascular genes!

Further, if this association is real, it raises the question of causality. But what 'causes' AIDS? Is it the HIV virus? But, HIV/AIDS rates are highest among the poor, so was Thabo Mbeki, the former president of South Africa, at least partly correct when he insisted that HIV/AIDS is caused by poverty? So if low IQ is a marker of race, and SES, do poverty and racial discrimination cause heart disease?

This study was carried out by British researchers. One wonders if their less nuanced understanding of American race and class issues erroneously led them to conclude that an effective way to lower heart disease is to make nutrition labeling easier to understand. Or could they be from such a middle-class environment as to be insensitive to these kinds of issues?

Or could it be that the general operating notions of causation, based as they are on stereotypical study designs that look for statistical associations, are a barrier to understanding?

As usual, genetic causation today and genetic causation as the result of evolution are similar. If natural selection favors some trait, then any genetic variation that gets the favored state will be favored. If variation in IQ genes led our ancestors to pick low-fat fruit (say), then those genotypes would be favored by selection just as fat-metabolizing genes would. If they could be detected by the kinds of searches for evidence of selection that many are doing these days, much experimental effort could be wasted trying to show how those genes were involved in lipid metabolism.

This is a subtle world today, and it's been that way throughout our ancestry.

Science is based on cause and effect, which is not identical to correlation and effect. The meaning of 'cause' has been central to philosophical thinking since Aristotle. Perhaps we should be less driven by methodology and pay more careful attention to that in science today.

No comments: