gms | German Medical Science

Overweight or obesity during childhood or adulthood is increasing seen as the major public health problem of our times. There is debate about the ‘real’ (as opposed to mythical) public health impact of the current obesity epidemic, even amongst those who agree that there is a problem. For example, recent estimates of the number of deaths attributable to obesity in the USA vary from 112,000 to 400,000 per annum. However, the evidence that being obese, particularly in the presence of being physically unfit, is detrimental to ones health is strong. Suggestions to the contrary in part relate to a failure to fully appreciate important biases in observational epidemiology. Two sources of bias are likely to be particularly important in studies assessing the association of body mass index with future health outcomes and might explain why some studies fail to demonstrate strong effects. First, reverse causality may weaken any true association. Secondly, smoking is an important confounder that might mask any true effect of overweight or obesity. Using already published data and some novel findings the impact of these two sources of bias will be demonstrated.

Whilst a number of randomised trials of interventions aimed at preventing obesity and treating (with respect to weight loss) individuals who are obese, very few of these have progressed to examine the long term effects on obesity related morbidity and mortality. This is important both in terms of clarifying causal associations but also in determining the most effective interventions for the prevention and treatment of obesity and its related morbidity. However, with the increasing recognition that some important determinants of obesity are exposures occurring in the intrauterine / infancy / childhood periods the ideal trials require commitment and imagination by the research community and funders to establish appropriately designed life course randomised trials.

For some potential early life risk factors it may be appropriate to use the principles of Mendelian Randomization, which is based on the plausible proposition that the association between a disease and a genetic polymorphism that mimics the biological link between a proposed exposure and a disease is not generally susceptible to the reverse causation or confounding that may distort interpretations of conventional observational studies, to provide evidence for or against causality. A specific example related to the proposed fetal overnutrition hypothesis as a cause of obesity will be presented.