What is 'severe obesity'?

From the BBC...

A record number of primary school children are leaving school severely obese, according to new figures from Public Health England. 

Data for 2016/17 shows one in 25 10 to 11 year olds were severely obese.

That's more than 22,000 children, and the highest level since records began.

This bears more than a passing resemblance to a story reported by the BBC (and everyone else) back in May...

One in 25 children in England aged 10 or 11 are severely obese, new analysis has found.

Height and weight measurements show the number of children classed as severely overweight rose from 15,000 in reception to 22,000 by the time they leave primary school.

This was the first time we encountered 'severe obesity' among children in the UK because, as the Beeb noted at the time...

This is the first time the national child measurement programme data has included the severely obese category.

Today's story is a rehashing of the same figures courtesy of the glorified pressure group Public Health England, except that they have now retrospectively calculated the rate of 'severe obesity' in 2006/07. The figure in 2006/07 was 3.17 per cent while the figure in 2016/17 was 4.07 per cent.

Since the time series began in 2006/07, this allows them to conclude that 'severe obesity' is at its highest level since records began. PHE then use this as a justification for their extensive meddling in the food supply:

Dr Alison Tedstone, chief nutritionist at PHE, said:

"The rise in severe obesity and widening health inequalities highlight why bold measures are needed to tackle this threat to our children’s health."

As part of its work to reduce childhood obesity, PHE is working with the food industry to cut 20% of sugar from everyday products by 2020, and 20% of calories by 2024.

So what is 'severe obesity'? There is a clue in the Telegraph's report...

In total, 4.1 per cent of such children are now classed as “severely obese” - compared with just 0.4 per cent in 1990.

How do we know that 0.4 per cent of kids were 'severely obese' in 1990? We don't. Nobody ever inspected them, but in 1995 Tim Cole et al. argued that the 99.6th percentile was a 'reasonable definition' of 'superobesity' among children. Why? Because the body mass index of twenty year olds at the 99.6th percentile was 32.8.

Cole et al. then calculated the BMI of children at different ages in 1990. These formed the reference curves that have been used ever since to measure child obesity. If a child would have been in the 98th percentile in 1990 (ie. the heaviest two per cent), they are classed as obese. If they would have been in the 99.6th percentile (the heaviest 0.4 per cent), they are superobese.

Standard obesity reference curves for boys

There are two obvious problems with this. Firstly, a BMI of 32.8 is not 'superobese'. Obesity starts at 30, with 35 being the usual cut-off for severe obesity. Secondly, there is no reason to think that children of 10 or 11 will have the same rate of obesity as 20 year olds. Since obesity rises with age, we can be certain that rates of obesity are lower among children and are much lower among young children. We don't know how much lower because - incredibly - nobody has ever bothered to find out. Instead, we use reference curves based on young adults.

Most children who are classified as obese are not obese. This can be easily demonstrated, not least by the fact that most children who are classed as obese miraculously stop being obese when they become adults and we start measuring obesity properly. Look at the drop off between those aged 11-15 years and those aged 16-24 years!

I have written about this at length here and here, but have not discussed 'severe obesity' before because the figures were rarely mentioned. Why are they being mentioned now? I suspect it is because, using the normal measure, child obesity peaked in 2004 and has been lower in every subsequent year. This undermines the narrative of 'spiralling' child obesity that is used to justify government intervention in the nation's diet.

And so Public Health England have found a measure that actually is rising: 'severe obesity'. By definition, there were 0.4 per cent of kids at or above the 99.6th percentile in 1990, whereas four per cent of today's kids would have been in the 99.6th percentile in 1990. This is a tenfold increase and should not be lightly dismissed. The numbers are small but the rise is real.

But are they 'severely obese'? Would they be diagnosed with 'severe obesity' by a clinician? In some cases, yes. This group obviously includes the fattest children in the country. But it is highly likely that some of them would be classed as merely obese or overweight if they were physically examined. Given that many of the 20 year olds in the 99.6th percentile were not severely obese in 1990, it is almost certain that many - perhaps most - of the 10 or 11 year olds in the 99.6th percentile were not 'superobese' then and are not 'severely obese' now. In some cases, they may not be obese at all.

It would be more accurate to describe kids in the 99.6th percentile as 'probably obese' rather than 'severely obese'. The 99.6th percentile is a much better cut-off for obesity than the 98th percentile, let alone the ludicrous 95th percentile which the government uses to estimate the national prevalence of child obesity. But to call it 'severe obesity' is a stretch.

Whatever we call it, the numbers are much smaller than for the fictitious category of childhood 'obesity'. The idea that more than 20 per cent of 11-15 year olds are obese is for the birds, but it is perfectly plausible that 4 per cent of children this age would benefit from losing weight. Efforts should focus on that small minority rather than on the entire population.

UPDATE

I've done a bit of media on this today, including Five Live. You can listen here from 38 minutes.