Tuesday, 6 July 2010

Computer says "lose weight, fatty"

[trigger warning]

The government continues its campaign to give every child their very own unhealthy relationship with food. The complaints about their plan only make the news - as in this case from February, or this one yesterday - when the child is fit and active and only just above the threshold at which the letters are triggered, but of course that doesn't mean that the letters actually make sense.

According to this earlier article [trigger warning: full of unchallenged fat hatred], letters are sent out to parents after measurements of 4-5 year old children and later of 10-11 year old children.

As a general principle, the National Child Measurement Programme is a good idea. Getting this sort of broad statistical information - and being able to compare it across years and decades - lets you track changes in the population and adjust accordingly (by making door frames taller as height increases, for instance)

This particular application of it, on the other hand, is absolutely terrible.

There are so many things wrong with the idea it's hard to know where to start, but for example:

  1. BMI was designed as a population statistic measure. Since it's being used in a population statistics assessment, that's not unreasonable. However, it was never designed to be applied to individuals and has a number of extensively documented flaws for those who try.
  2. Among those flaws is that it is designed for adult humans rather than the rapidly varying body shapes of children in early puberty or pre-puberty growth. Even if it was semi-meaningful for adults, it makes no sense for children at all.
  3. Children tend to gain weight in early puberty anyway, which then often gets converted into energy to gain height later on. That's what's supposed to happen.
  4. The letters seem to be sent out based on a point cut-off. So in the February article, we have:

    ... said that for a young girl, this meant she was 1% outside the healthy category and could be at risk of heart disease and cancer.

    There is not - obviously to anyone with any grasp of anatomy, medicine, or even basic statistics - any sudden step change in risks as you cross the arbitrary and suspiciously round number of the heavy edge of the government-mandated weight zone. (It's furthermore - as I've mentioned previously - not even clear that the risk gradient is upwards at this point). It's a hopelessly broken idea anyway, but applying this sort of step cut-off is absurd.
  5. Equating health with weight and nothing else is as usual totally wrong, which is just obvious enough that the "but my child does ten different forms of regular exercise" stories make the news, but not quite obvious enough in UK culture that the rest of the logical conclusions get made.
  6. As usual, it's going to completely ignore anyone who can't exercise for whatever reason, whether that is disability, a lack of local facilities, lack of time, or something else.
  7. While there is a correlation in older adults between weight and some health conditions, it's not yet been shown that there the weight causes the conditions: for many it's more plausibly a symptom. Furthermore, of course, there's no known correlation between weight as a child and gaining these conditions in later life, certainly no correlation between weight as a young child and childhood incidence of these conditions, and people's body shapes and BMIs gradually change - sometimes quite significantly - over the years anyway.
  8. The advice given, as the articles make clear, is going to be counter-productive. Exercise is all very well, but only for those people who can actually do it. Putting children on diets - or worse, permanent "weight-loss" surgery - will usually actively harm them, either through childhood malnutrition (which is strongly correlated with a bunch of nasty health conditions), or causing an eating disorder, or both).

I wonder if the coalition could be tempted into making an "efficiency saving" in the NHS and stopping sending out these letters. At about 1.8 million children in the relevant age bands, "most" of whom are surveyed and presumably receive a letter, that's potentially quite a bit of money saved.

Making a rough guess that the cost of sending the letter, not counting any costs that would have been incurred by the useful bits of the programme, is about £1 for the envelope, bulk postage rate, paper, printer ink, additional wear and tear on equipment, and staffing costs, that's around 1 million pounds a year that could be spent on either hospital equipment and staff, or on an effective and useful public health initiative.

The coalition so far hasn't said much on the "obesity epidemic" and neither party mentioned it in their manifestos. The Health Secretary rightly criticises the approach of Jamie Oliver, which seems to have been as counterproductive here as when he was temporarily exported to the USA. So perhaps there's a chance that this policy could be dropped.