by Patrick Basham and John Luik
This article appeared in the British Medical Journal on February 2, 2008.
The claims, both in the media and in professional publications, about an epidemic, its causes, consequences, and cure often exceed the scientific evidence and mistakenly suggest an unjustified degree of certainty. The fact that cases are "clearly above normal expectancy" anchors the concept of an epidemic. In this respect, describing obesity as an epidemic is subject to two difficulties.
Firstly, it is difficult to determine normal expectancy. Much of the data on overweight and obesity are limited, equivocal, and compromised in terms of extent and the reliability of the measurements and the populations sampled. In the US, for example, data about population weights date from only 1960. Several pieces of evidence, however, suggest that the contemporary situation may be close to, rather than in excess of, normal.
The earliest national survey shows that in 1960 45% of the US population was overweight, accordingto sex specific weight for height tables (corresponding to a body mass index of 25 to <30).1 In the 1970s, 22% of US men aged 18-19 were overweight compared with 16.7% of boys aged 12-19 in 2002.2 Fogels ongoing work in various countries on the relation between health, mortality, nutrition, and technology suggests that as populations grow healthier, prosperous, and long lived they gain in height and weight.3
Moreover, current data are highly equivocal in their support for claims of an epidemic. For example, the average population weight gain in the US in the past 42 years is 10.9 kg or 0.26 kg a year.4 Yet, between 1999-2000 and 2001-2002, according to the National Health and Nutrition Examination Survey, there were no significant changes in the prevalence of overweight or obesity among US adults or in the prevalence of overweight among children.2
Secondly, the determination of the categories of normal, overweight, and obese is entirely arbitrary and at odds with the underlying evidence about the association between body mass index and mortality, a fact that destroys the indexs scientific pretensions and diagnostic value. The bands for overweight and obesity in the US, for example, are the product of the 1997 National Institutes for Health task force report on the prevention and treatment of obesity that supposedly links these bands to increased risk of death. However, the study on which the report is based does not support these linkages.5 It found that the death risks for men with a body mass index of 19-21 were the same as those for men who were overweight and obese (29-31).
The studys findings are not unusual. Flegal and colleagues found the weight group with the lowest death rate was overweight,6 while Gronnigers analysis found negligible differences in risk of death among people with body mass values from 20 to 25.7 Even where there are significant associations, the risks are so modest as to be highly suspect. For example, whereas the reported lung cancer risks for smokers are typically 10-15 times higher than for non-smokers, the death risks for overweight and obese people are in many instances closer to 0.5-1.75 above those for people with normal weight.8
Despite the supposedly abnormal levels of overweight and obesity, life expectancy continues to increase. According to the UK Office for National Statistics, the current life expectancy of 77.2 years for men and 81.5 for women will rise by 2031 to 82.7 and 86.2, respectively.9
Moreover, the association of overweight and obesity with higher risks of disease is equally unclear, partly because of the multifactoral character of these diseases. Increases in overweight and obesity have been paralleled by falls in US total cardiovascular mortality and mortality from coronary heart disease and stroke, as well as in prevalence of hypertension and hypercholesterolemia.101112
Several factors justify scepticism about the link between non-insulin dependent diabetes mellitus and overweight and obesity. They include the absence of compelling direct evidence that excess fat is the cause of insulin resistance, the fact that the link fails four of the Austin Bradford Hill criteria for causality, and that increased physical activity and dietary changes reduce diabetes risk in advance of, or in the absence of, weight loss.13
There is considerable evidence that most fat adults were not fat children.14 Moreover, the thousand families cohort study found both little consistency between childhood overweight and adult obesity and no net increase in adult risk of disease for overweight children or teenagers. Nor did childhood thinness protect against either adult obesity or coronary vascular disease.14 15
Some in the public health community believe that deliberate exaggeration or, indeed, misrepresentation of the risks of diseases or certain behaviours or our capacity to prevent or treat them on a population-wide basis is justified, if not demanded, in the interests of health. Since many of the exaggerations come from people who understand the scientific uncertainties around overweight and obesity, it seems that these individuals have adopted such an approach to the obesity epidemic. The unwelcome implications of this for science policy and for evidence based medicine dwarf those of any obesity epidemic, real or imagined.