Tuesday, June 24, 2008

This is all very confusing

The Times (June 20th) along with 407 others (according to Google News) reports that

Australians are...leading the way as the heavyweight champions of the world - with arguably a greater proportion of obese citizens than even the notoriously supersized Americans

60% of the Australian population is overweight, apparently. That's nine million people, according to the study, 'Australia's Future Fat Bomb' (seriously).

The Age, at least, points out that the BMI index is a crude tool for measuring how overweight someone is (it doesn't allow for any differentiation between muscle and fat).
Neither The Financial Times, The Telegraph, The Mirror, or The Times note this. None of the newspapers mention that the sample used to estimate the figure was highly unrepresentative, as it consisted of 14, 000 people who attended a National Blood Pressure Screening Day, that is, 14, 000 people who were worried enough about their blood pressure to have it checked; or 14, 000 of whom a higher than average proportion could probably be assumed to be overweight, as the two conditions are correlated (or, more importantly, commonly known to be correlated, such that if you are overweight you're more likely to be worried about your blood pressure). Also, the age distribution of the sample may have been skewed such that people of middle age and older were over-represented.
The report itself is obfucatory on this issue - it states that 'close to 14, 000' Australian adults aged 18 to 95 participated - close to? The authors have no qualms in printing figures like 7, 352, 529 (number of Australian men aged 18 and over) and 2, 412, 192 (the number of Australian women 'likely overweight'), so why so coy when it comes to piddling tens of thousands? They don't give figures for each of the age ranges they examine, except for the 45-64 age group, of which there were 5, 873. It makes me wonder if perhaps the older age groups (45+) were heavily over represented in the sample, something the hard figures, had they been provided, would have made clear thus (further) undermining the claim that 60% of the Australian population aged 18 and more is overweight.
Further, according the the Australian National Health Survey 2004-2005, the age group 45 - 64 has the highest proportion of people who are overweight (50% women, 72% men aged 45 to 54; 58% women, 72% men aged 55-64 compared with 28% women and 36% men aged 18-24). If this group, and those older (56% women and 58% men overweight aged 65-74 - though they're roughish figures as the graph isn't entirely clear) made up the main body of the sample of 14, 000, then it raises serious questions about the usefulness of this study in estimating the proportion of the Australian population that is actually overweight.
Page 10 of the study notes that 'Although the full methods and results of this study are in preparation for publication they have been accepted for presentation by Dr Carrington at the prestigious European Society of Cardiology Scientific Meeting in Munich (September 2008)'. Which is fair enough, but doesn't seem like a good enough reason to hold back on either methods or results before releasing the report on the world with the schlocksome title 'Australia's Future Fat Bomb'. Like most things, if and when the methods used and results obtained are questioned by the authors' academic peers, it won't be picked up by the press, and the uncertain figure of 9 million fat Australians will have taken on the status of commonly accepted fact.

No comments: