Friday 11 January 2013

Obesity and mortality revisited

The obese cost public health systems less in the long term because they die earlier.

And now it looks like those who are merely overweight rather than highly obese actually live longer than those who are of normal weight. JAMA reports that only categories 2 and 3 in the obesity rankings are associated with higher mortality risk; the overweight have a mortality risk of 0.94 compared with normal-weight individuals. From the study:

This study presents comprehensive estimates (derived from a systematic review) of the association of all-cause mortality in adults with current standard BMI categories used in the United States and internationally. Estimates of the relative mortality risks associated with normal weight, overweight, and obesity may help to inform decision making in the clinical setting.

The most recent data from the United States show that almost 40% of adult men and almost 30% of adult women fall into the overweight category with a BMI of 25 to less than 30.111 Comparable figures for Canada are 44% of men and 30% of women112 and for England are 42% of men and 32% of women.113

According to the results presented herein, overweight (defined as a BMI of 25-<30) is associated with significantly lower mortality overall relative to the normal weight category with an overall summary HR of 0.94. For overweight, 75% of HRs with measured weight and height and 67% of HRs with self-reported weight and height were below 1. These results are broadly consistent with 2 previous meta-analyses114115 that used standard categories. In a pooled analysis of 26 observational studies, McGee et al114 found summary relative risks of all-cause mortality for overweight of 0.97 (95% CI, 0.92-1.01) for men and 0.97 (95% CI, 0.93-0.99) for women relative to normal weight.
If the healthists keep talking about high proportions of the population who are overweight and obese, but it's only the heaviest cohort within that group that experience increased mortality risk, and that group is only a small portion of the overall category we keep hearing about in the paper, why add the overweight and the category 1 obese to the tallies? To get bigger numbers and fuel perception of a crisis. Timandra Harkness explains [HT: @cjsnowdon]:
The reason this unassuming paper drew howls of outrage was the same as the reason the benefits of moderate alcohol intake are never noted without criticism: it spoils the headline health message that Fat is Bad.

Even worse, it blows the cover on the great myth – that an epidemic of Bad Fatness is sweeping the developed world. By including the dangerously obese, the innocuously tubby and the healthily plump in one category, ‘overweight including obese’, 60 per cent of the English population are labelled as potentially At Risk.

Being At Risk means these people need guidance and protection from their own vulnerable state, from the temptations of our obesogenic world and the frailties of their own sugar-addicted brains. At such a time of national peril, no measure is too extreme.

But less than a quarter of English adults are obese, according to new figures released just before Christmas, a fraction almost unchanged since 2007. And the ‘morbidly obese’ category – BMI over 40, the ones for whom it really might be worth shedding a few pounds, medically speaking – also remains steady since 2009 at 2.5 per cent of the UK population.

If only one in 40 of us is in significant weight-related danger, why do the other 97.5 per cent of us need to be protected by the state against sugary cereals and fizzy drinks? Could it be because only a few of us have fallen, but all of us are in peril? Weak, foolish and easily led astray, we need to be frightened back on to the right path. Thus Tam Fry, spokesman for the National Obesity Forum - who has called for children to be monitored from birth for signs of obesity - told the Independent: ‘If people read this and decide they are not going to die… they may find themselves lifelong dependents on medical treatment for problems affecting the heart, liver, kidney and pancreas – to name only a few.’
I suppose that an alternative hypothesis would be that some folks find the obese to be aesthetically displeasing and prefer to base policy interventions on a purportedly paternalistic basis than on an externality-via-aesthetics argument. The latter is more economically defensible, provided that we expect Hollywood et al get things roughly right about what sorts of actors more people prefer looking at, but harder to defend in popular forum. I'd also expect that since the morbidly obese suffer wage and health penalties already, any incremental Pigovean aesthetic tax added to the mix wouldn't substantially affect things.


  1. I thought it had long been known that (i) the BMI 25 cutoff had been pulled out of thin air and (ii) there's not a lot of action in terms of health outcomes below BMI 30. Everything else is aesthetics, politics, and psychology. Perhaps Nietzsche got it slightly wrong: The demise of Christianity did not leave a vacuum in terms of morals, but it did leave one in terms of reasons for despising people, which is now filled by public health-related arguments. In the future, we may be looking back on the period ca. 1970-2000 as the golden age of tolerance.

  2. I suspect at least part of this result is the ridiculous nature of the BMI system, where only rakes can scrape into the normal weight category. A six foot person clocks in as overweight as soon as they hit 85kg.

  3. So the BMI cut-offs then serve the same role as the rather low threshold needed to be a problem drinker: you can inflate the number of people with a problem, while citing the effects that obtain for a small minority at the tail of that group.

  4. This study is flawed because it relies on the fundamentally flawed BMI model. Just about any athlete -even a recreational gym-goer- will be classified as overweight or obese. It is dangerous to conclude that the overweight live longer than "normal" people when you consider that there is zero distinction between fat and muscle mass.

    If you conducted a study on the basis of body-fat percentage it would no doubt return very different results.

  5. What I get from the study is that I should apply more skepticism to claims around obesity epidemics when cited numbers include those who are counted in the "overweight" range.

  6. Clearly, in the case of health statistics and manufacturing epidemics to torment the population with, it really is a case of the tail wagging the dog!

  7. I demand a subsidy for my bacon consumption.

  8. That's the trick. Similar: The definition, and use, of the term "binge drinking" by alcohol researchers.