An alternative explanation would be Geoffrey Miller's idea that IQ is a general 'genetic fitness' indicator - so that the higher the IQ the better are a person's genes (on average); and the less likely more-intelligent people are to get almost any kind of disease (which is what has been found by the same Scottish group who did this work).

if miller's mutational load indicator can explain this, it be interesting to see how elevated mortality varies as a function of age. if, the issue is how IQ effects the choices one makes, and that the cumulative effect of the choices leads to pathology later in life, then the difference would be back-loaded. if it is due to variance in mutational load, then the propensity toward illness should be clear early on in life so the difference at any given time would be less in terms of mortality (since it would be distributed over the years). i'll let others think of the design for testing this model....

It could be due to differential responding to a few extreme tail events (that can cause death but with the right action can be averted); like having nitroglycerine at hand and using it when needed.

I don't know much about stats and such, but wouldn't you have to control for factors such as race, IQ and coronary heart disease (CHD) factors?

I believe that African-Americans suffer from CHD at a significantly higher rate than other racial groups and that they are more likely to suffer a fatal heart attack. African-Americans also have a lower IQ on average, and since IQ and socio economic status have a strong correlation, wouldn't the study be skewed?

There is another explanation: an underlying physiological difference that simultaneously improves IQ and cardiovascular health. I suggest mitochondrial activity as this ubiquitously active factor.

Another possibility is that intelligent people get better health care because they search it out. I probably didn't die of lymphatic cancer in 1997 because I was the first person in the world with my type of lymphoma to be treated with Rituxan, now the world's top-selling cancer drug. I found out about the clinical trial because I was on the Internet back in January 1997, actively researching for trials.

Where would one place hereditary conditions such as Diabetes that have a strong correlation with Heart Disease? In my case, the maternal side has a consistent Diabetic history together with anemia, and more than one distinguished ancestor kicked the bucket because of heat failure. On the father's side OTOH we have solid country genes with the secret sauce for long life. But it pales in comparison to the maternal line when it comes to academic achievement, which remains lopsided even after considering the difference in socioeconomic conditions. (My maternal grandpa was a NTSE topper, a Huge Deal in the subcontinent) I guess its aberrational.

fractal storm, you need to control for genetic background.

Michael Marmot in his book The Status Syndrome argues that persons with low social standing are more stressed because of that and that this affects their life expectancy. This is valid also for monkeys. The ones with low rank are more stressed. Interesting book.

Roy Frye, a pathologist at University of Pittsburgh, has formulated a hypothesis that attempts to explain some of the correlations between ethnicity, cognitive ability, socioeconomic position, and health/mortality.

A pdf of "Cognitive epidemiology of ethnic health and the CHRM2 vagal vigour hypothesis" is freely available here on the Nature Precedings website.

Frye's paper studies how mortality varies between local regions of California with regard to cognitive ability, ethnicity, and socioeconomic position.

Razib said it would "be interesting to see how elevated mortality varies as a function of age".

The supplemental figures in Frye's paper present abundant data showing patterns of age-range variation in mortality in relationship to variations in ethnicity and cognitive ability.

Frye's claim that variation in both IQ and health/mortality reflect pleiotropic effects linked to the rs8191992 SNP variation in the 3'UTR of the CHRM2 is now somewhat dubious because recent studies of large cohorts (Lind et al in Behav Genet and a meeting abstract from McGue's group at Minnesota) have cast doubt on the claim (from three prior studies) that IQ is linked to SNP variation in the CHRM2 gene.

How might we measure or estimate how much of the IQ:health correlation is based on better understanding of health issues, vs other stuff like IQ as a marker for mutational load or lousy developmental environment or what-have-you? It seems like one place to look is at people with more or less knowledge or training about medicine.

For example, if we could somehow compare health outcomes for doctors (people who graduated from medical school) with those of some comparable professionals (similar IQ and economic/social status), that would give us some handle on this. If medical knowledge is an important advantage in health, then doctors ought to get a large benefit in life expectancy. Obviously, there are a lot of confounding variables here that would make this kind of comparison hard, though. (Among other things, depending on the kind of doctor you are, you're probably around sick people all the time, meaning you get exposed to more bad stuff. And I don't think there are many jobs with the stress level of, say, a surgical residency. And so on.)

Maybe a better observational study would be between members of the same profession. The intuition is that when more useful information is available to you, higher IQ should be associated with better outcomes. So, what you'd expect is that the correlation between IQ and health outcomes should be biggest for doctors, somewhat smaller for nurses (who aren't exposed to as much detailed medical training as doctors, but who still get a lot), and still smaller for people entirely outside the medical field. (I suspect you have to control for medical specialty, though--the life of a emergency room doctor is probably different enough from the life of a dermatologist to need to be controlled for, and since different specialties pay very differently and are harder or easier to get into, they probably correlate with IQ.)

Is it possible that there's causation in the other direction? That is, if you are naturally healthy, your brain develops better and you end up smarter.