Guest essay by Eric Worrall
A new study claims a warmer climate is associated with higher rates of diabetes. But even if the authors are right, obesity, diet, age and heredity remain the dominant factors.
Diabetes incidence and glucose intolerance prevalence increase with higher outdoor temperature
Lisanne L Blauw, N Ahmad Aziz, Martijn R Tannemaat, C Alexander Blauw, Anton J de Craen, Hanno Pij, Patrick C N Rensen
Objective Rising global temperatures might contribute to the current worldwide diabetes epidemic, as higher ambient temperature can negatively impact glucose metabolism via a reduction in brown adipose tissue activity. Therefore, we examined the association between outdoor temperature and diabetes incidence in the USA as well as the prevalence of glucose intolerance worldwide.
Research design and methods Using meta-regression, we determined the association between mean annual temperature and diabetes incidence during 1996–2009 for each US state separately. Subsequently, results were pooled in a meta-analysis. On a global scale, we performed a meta-regression analysis to assess the association between mean annual temperature and the prevalence of glucose intolerance.
Results We demonstrated that, on average, per 1°C increase in temperature, age-adjusted diabetes incidence increased with 0.314 (95% CI 0.194 to 0.434) per 1000. Similarly, the worldwide prevalence of glucose intolerance increased by 0.170% (95% CI 0.107% to 0.234%) per 1°C rise in temperature. These associations persisted after adjustment for obesity.
Conclusions Our findings indicate that the diabetes incidence rate in the USA and prevalence of glucose intolerance worldwide increase with higher outdoor temperature.
Read more: http://drc.bmj.com/content/5/1/e000317
The data seems quite noisy. For example, consider the following graph (Table 1 in the study) of diabetes incidence vs change in temperature.
The state of South Carolina (average annual temperature 63F) shows a strong correlation between diabetes and temperature, but the state of Louisiana (average annual temperature 69F) shows a strong negative correlation. Arizona (average annual temperature 75F) also shows a negative correlation.
In addition, the impact of the factors the study attempted to adjust for is likely significantly larger than the impact demonstrated by the study.
For example, the following, from a 2007 study of body mass index vs diabetes, shows a strong relationship between body mass and diabetes, ranging from below 5% for skinny people to around 25% for very fat people – a far more pronounced effect than the small difference claimed by the climate study above.
Note the substantial disagreement between the two studies in the graph above, of the impact of obesity on diabetes rate.
I’m not saying the authors are necessarily wrong, they appear to have made a credible attempt to tease out a small effect from a noisy data set. But even if they are right, factors other than temperature are far more significant predictors of whether someone is likely to suffer from diabetes.