U.S. Life Span, Obesity, Elevation and Temperature

I had a conversation with a friend a few months back about the effect of cold temperatures on life expectancy; bottom line, she thought the colder the place you live, the longer you live. She theorized that this might have something to do with brown fat storage and/or thyroid function. I was skeptical, but curious.

Consequently I did a (very) little digging into questions of thyroid levels and brown fat and their impact on body mass index (BMI) and life expectancy. That adult humans even have brown fat at all was only accepted in the last few years, so there won't be brown fat maps floating around as yet. As for THR, TSH, or T3/T4 levels, I would guess that people living in cold climates would have higher T3/T4, at least seasonally (this is probably what drives the "blood thickening" that you lose when you move to Florida, for example). However after a brief search I wasn't able to find any solid studies relating thyroid levels to life expectancy and/or climate.

But we still have amateur epidemiology! Recall, the original discussion was about whether there is an effect of climate on obesity and/or life expectancy; let's not worry for now about how it might be mediated. (The following hyperlinks go to data sources.) I threw together some scatter plots comparing average U.S. state temperature and average U.S. state elevation to average state BMI and average state lifespan. I also used per capita income and % black population as a comparison for the effects of poverty. Instead of tarting up some boring scatter plots I'll just show the R-values:

Relationship	R-value	Correlation
Obesity-Ave Temp	0.47833	Positive
Obesity-Elevation	0.642028	Negative
Obesity-PCI	0.225389	Negative
Obesity-% Black	0.648228	Positive

Relationship	R-value	Correlation
Life Expect-Ave Temp	0.631585	Negative
Life Expect-Elevation	0.292404	Positive
Life Expect-PCI	0.583952	Positive
Life Expect-% Black	0.724155	Negative

So, what do we see here?

1) For both obesity and life expectancy, % black state population by state is a better predictor than PCI, average temp or altitude.

2) However, for obesity, % black population is just barely a better predictor than altitude. Low pO2 makes you skinny. No surprise there.

3) Interestingly, while altitude almost tied for best predictor of obesity, it is the weakest predictor here of life expectancy. People at altitude are a lot skinnier on average but don't live that much longer on average.

4) The situation is reversed for income. The richer a state, the less obese it is, although the relationship is weak. However, the richer a state, the longer-lived it is, with a substantially stronger relationship than with obesity.

5) There is a positive correlation between obesity and temperature, and a negative one with life expectancy and temperature. My friend would argue that this supports her position. However, people are not distributed evenly among states: Southern and South-central states (that both have a higher black population and are not as wealthy) make up 17 of the 20 warmest states, so it's difficult to tell what's going on, especially since those were better predictors anyway. A real study would look at economically and ethnically-matched individuals at least. While I'm ordering up studies from real epidemiologists to satisfy my curiosity, I'd still like to see average T3/T4 levels by average temperature and average January temperature.

Added later: from Marginal Revolution - I'd always been curious why obesity would suddenly become a problem in the second half of the twentieth century, when a lot of the production problems had made large volumes of calories cheap before. Turns out there's no mystery and it was already a problem a century ago.