Does the data show ‘epidemic’ obesity risks?

It is surprisingly difficult to find an answer to these questions in the ABS data. The Bureau publishes only age-stratified prevalence data from its samples; not basic statistics such as the mean and variance of the adult population BMI. I can’t tell from their publication how the BMI isdistributed across the whole population or how the ‘central tendency’ of that distribution changed between the measured sampling in 1995 and 2008. This is crucial information, however, because it is likely to show what is actually going on in the trend and whether the apparent increases in the population indices are due mainly (as I suspect) to the greater variance in the sample or due to an alarming ‘grossing-up’ of the average Australian (as some interested groups want us to believe).

People are living longer and BMI tends to increase with age, as the ABS prevalence data shows. So it’s quite possible that a larger proportion of the total population has a higher BMI just because there is a bigger ‘spread’ (variance) in the frequency distribution curve. If that were so, the ‘mean’ (central tendency) of the distribution may not have changed to the extent that the increase in the proportion of ‘overweight and obese’ in the sample seems to indicate. In other words, not so much ‘fatter on average’ as ‘fatter at the extremes’.

So let’s accept, for the moment, the premiss that there has been an explosion of fat. What can we do to check whether we should be alarmed? The obvious thing is to check whether the ‘risks’ that are said to be increasing are visible in the changing incidence of mortality. That is, has there been an alarming increase in actual deaths from the diseases that overweight and obesity are said to cause?

It turns out that there’s no apparent correlation between the claimed trend in the obesity risk-factor and actual deaths from obesity-associated diseases. Here’s the relevant information on mortality trends from the Australian Institute of Health and Welfare’s Australia’s Health 2008 publication.

“Australia’s mortality from cerebrovascular disease has been declining in recent decades… since the mid-1970s, consistent declines have been noted for both males and females. Age-standardised death rates for cerebrovascular disease fell by 37% (males) and 35% (females) over the period 1996–2005.”
“There have not been major changes in the death rate from diabetes (as an underlying cause) over the last 25 years… For males, the death rate rose by an average of 0.7% per year. In contrast, the rate for females fell by an average of 0.5% per year.”
“From 1997 to 2005, the mortality rate for CKD [chronic kidney disease] as the underlying cause of death fell from 13 to 11 deaths per 100,000 population.”

This absence of correlation between the obesity-trend and the mortality-trend in Australian now looks less mysterious, thanks to a unique and detailed United States study published in February by the senior statisticians of the United States Centers for Disease Control and Prevention. Using a very large sample of the U.S. population, and taking BMI and death data since 1988 into account, Flegal and Graubard found BMI is not significantly correlated with excess deaths from all-causes, from ‘obesity-related’ causes of death or from other causes. Looks like the same might be true in Australia.

But what about the disease burden? Excess mortality is the only final arbiter of the degree of risk, but the burden of disease on individuals and the community (individuals, taxpayers, carers) is also important. What’s going on there?

As far as I can tell from the semi-obscure way that ABS presents the data, the story for the disease burden is consistent with the much-less-than-alarming mortality data. I went back to the earliest report of the National Health Survey data available on the ABS site: as it happens, the 1995 summary report that contained the last measured BMI data. That report also contains some ‘comparison’ data on disease prevalence from 1989-90.

Here’s a summary of what I found:

Long-term disease incidence (self-reported), percent of population200819951998
Diabetes mellitus (A-type and B-type)5.72.21.1
All circulatory diseases (cardiac disease, hypertension, cerebro-vascular e.g. stroke)16.420.013.1

The diabetes incidence seems to have risen sharply—possibly because of more active diagnosis and because the determinative blood-sugar thresholds have been lowered in that time—but the incidence of circulatory disease has apparently fallen (while everyone was getting fat).

No Comments

Leave a Reply

Your email is never shared.Required fields are marked *