Sunday, May 4, 2014

The Australian Paradox: Part 1

For those who don’t know, the Australian Paradox essentially says that there has been a “consistent and substantial decline in total refined or added sugar consumption by Australians over the past 30 years”, during which time the prevalence of obesity in Australia has tripled.  And concludes that “the findings challenge the implicit assumption that taxes and other measures to reduce intake of soft drinks will be an effective strategy in global efforts to reduce obesity”.  In other words: sugar down, obesity up, therefore sugar isn’t responsible 

Rory Robertson has already really thoroughly critically analysed the Australian Paradox.  Also in February, Wendy Carlisle did an investigation of the Australian Paradox for Background Briefing. 

In this post I’m just going to look at the results and save other points for the following post 

Obesity has increased over the last 30 years.  No surprises there

This is the FAO data, which has gotten its Australian data from the ABS Apparent Consumption of Foodstuffs up until 1999 as the ABS ceased that dataset after 1999.  In the years after 1999 there’s a flat line for ‘refined sucrose’ with a very low year-to-year variation.  It looks like the FAO made up the data for 2000 onwards (see here) 

Another issue is that when calculating apparent consumption of sugar, the ABS data didn’t include sugar contained in imported highly processed foods.  Between 1988 and 2010, imports of highly processed food containing sugar increased to a much greater extent than exports.  If the data on imports was added to the FAO dataset it would suggest an increase in sugar availability [1]* 

* On Background Briefing one of the authors says “my paper has not been criticised by any scientist”

Both these graphs suggest sugar intake has increased in both adults (F3) and children (F4)

In the discussion the authors say that per capita sales of sugar sweetened beverages (SSBs) decreased by 10% between 1994 and 2006, but per capita sales of SSBs increased by 30% and the reduction of 10% refers to market share.  This was corrected after background briefing in February, although the authors said this change, and another two, have no material impact on the conclusions of our paper”.  When your paper is supposed to look at trends in sugar consumption I don’t see the point of mentioning market share between SSBs and diet drinks (DDs) 

Volume Sold (L/person/year)

Market Share (%)

* The numbers in these tables are rough to make it easy for you to see the relationship between volume sold and market share (you don’t need a calculator)

If you wanted to look at 30 years trends I don’t understand the point of figure 6 as it only lists data from 1997-1998 to 2005-2006 (8 years).  In relation to this figure the authors say “overall, there was a decrease in sugar contribution from nutritively sweetened carbonated soft drinks to the Australian food supply, amounting to 12,402 tons (~600 g per person*, Figure 6) from 2002 to 2006.”.  That sounds like a lot of sugar, but on closer inspection, this impressive figure only amounts to an average reduction of 0.42 g/person/day** 

* Corrected here 

** 12.4 billion grams / 20 million people / 1461 days

It’s interesting that this data in children suggests no difference or a decrease in SSBs from 1995 to 2007, while F4 shows a more than doubling in SSBs over the same period and F5 shows a 30% increase in volume of SSBs sold among both adults and children

Bonus graph!  This one comes from The Australian Paradox Revisited.  Like the ABS data, the issue here is that this doesn’t include imports of sugar and sugar in processed foods, which have increased over time [1] 

Taking the results at face value: 

·         F2: Decrease in apparent sugar consumption
·         F3: Increase in sugar intake among adults
·         F4: Increase in sugar intake among children (overall)
·         F5: Increase in the volume SSBs sold
·         F6: No difference in sugar from SSBs between 1997-1998 to 2005-2006
·         F7: Decrease in SSBs consumption by children
·         B1: Increase in availability of refined sugar 

No Change
F3, F4, F5, B1
F6 (limited trend)
F2, F6

On balance, the results presented in the Australian Paradox suggest that sugar and SSB consumption has more likely increased rather than decreased.  This is even while ignoring issues with the data such as not accounting for sugar from increased imports of processed sugary foods and SSBs (F2, B1) 

In the discussion the authors say there has been a “consistent and substantial decline in total refined or added sugar consumption by Australians over the past 30 years”, which is clearly inconsistent with their own results and you could only make that conclusion if you looked at the FAO data and ignored all the other results 

* F4 and F7 suggest a decline in sugar and SSB consumption respectively between 1995 and 2007.  F1 suggests the incidence of obesity in children increased between 1985 and 1997 and between 1997 and 2004.  Being quite dissatisfied with the authors’ referencing and citations, I looked for information on trends in childhood obesity and came across this meta-analysis that found the trend in prevalence of childhood obesity plateaued from 1996 onwards [2].  If the data in children and the meta-analysis is true then sugar intake in children decreased between 1995 and 2007, which is the timeframe when the prevalence of obesity in children plateaued, suggesting there is no Australian paradox in children


  1. By and large if you are poor you are more likely to buy from a constantly changing variety of cheap, because ersatz, imported drinks and confectioneries. In NZ these are often not labelled in English and probably not tabulated in any food availability calculations; many give the impression of having slipped past sleeping regulators.
    Thus not only is NZ intake of sugars likely to be unrepresentative of consumption patterns of the poorest people, the intake of trans fats (very low in the normal food in a respectable supermarket, hence in estimates of the NZ diet) is likely to be underestimated too.
    These are ersatz versions of junk foods, so God only knows what dyes and other non-food items also go into them.

    1. That's not good. As Rory points out, the task of counting sugar in imports would be huge. Provided we are equally as bad at FFQs now as we were before, I would place more emphasis on that, especially if you could do a second study to see what people actually ate so as make adjustments