Wednesday, August 13, 2008

The fat kid fumble

Clumsy children more likely to become obese as adults, study finds. Physical control and coordination in childhood and adult obesity: longitudinal birth cohort study by Walter Osika and Scott M. Montgomery was published yesterday in the BMJ. It's part of the National child development study in Great Britain, a longitudinal cohort study of children born between 3rd and 9th March 1958 and living in Great Britain. The study in question here gathered assessment data on hand control, coordination and general clumsiness, as assessed by teachers when the cohort was 7 and 11. Data were available for 7990 children at age 7 and 6875 at age 11. An association was found between teacher reports of 'clumsiness' and subsequent obesity (as measured by Body Mass Index) at age 33. It reads:
Among 7990 cohort members at age 7 years, teachers reported that poor hand control, poor coordination, and clumsiness "certainly applied" more often among those who would be obese adults.

Similar results were found for the cohort measured at age 11 years.
The authors caution:
Many other environmental or individual characteristics could explain the associations.

and go on to suggest, as an example, that individuals with poor motor control may be less likely to exercise, thus increasing the risk of obesity. They conclude that while the study adds to the body of evidence associating poorer cognitive functioning with obesity, neither this study nor any other which has found such a link, can tease out how or why this should be. They conclude:
Rather than being explained by a single factor, an accumulation throughout life of many associated cultural, personal, and economic exposures is likely to underlie the risks for obesity and some elements of associated neurological function.


265 words. And it took me all of twenty minutes, including the time taken to read the paper, to write. The Daily Mail gives it 221 words. I'll reproduce them in full:

Clumsy children are more likely to be obese in later life because they exercise less, research says.

A study that started 50 years ago found youngsters with poor hand control and co-ordination are far more prone to piling on the pounds during adulthood.

This puts them at higher risk of heart attack, stroke and diabetes.

The findings, published online by the British Medical Journal, found children who performed worse in tests assessing their cognitive and physical function were more likely to be obese by the age of 33.

Those who were obese were 57 per cent more likely to have suffered poor hand control aged seven, more than twice as likely to have suffered poor co-ordination and almost four times as likely to have been clumsy.

The study adds to evidence of a link between poorer cognitive function in childhood and obesity and Type 2 diabetes in adulthood.

Researchers said cognitive impairment in obese adults was assumed to be a consequence of obesity.

However, the study indicates obese adults and those with Type 2 diabetes may already have had lower levels of cognitive function in childhood 'consistent with a subtle developmental impairment'.

The research, whose authors are from the Karolinska Institute in Stockholm and Imperial College, London, is based on 11,042 people taking part in the ongoing National Child Development Study in Great Britain, which began in 1958.

And again, this time with nimby commentary:

Clumsy children are more likely to be obese in later life because they exercise less, research says.
fair enough

A study that started 50 years ago found youngsters with poor hand control and co-ordination are far more prone to piling on the pounds during adulthood.

This puts them at higher risk of heart attack, stroke and diabetes.

The findings, published online by the British Medical Journal, found children who performed worse in tests assessing their cognitive and physical function were more likely to be obese by the age of 33.
Why not just say association? What's wrong with it? 11 letters, 1 more than 'more likely' and without the connotations and grammatically questionable see-sawing between temporalities.

Those who were obese were 57 per cent more likely to have suffered poor hand control aged seven, more than twice as likely to have suffered poor co-ordination and almost four times as likely to have been clumsy.
This is where it gets seriously wonky. The odds ratios reported in the study were 1.57 (95% confidence interval 1.13 to 2.20; P=0.008) for poor hand control, 2.30 (1.52 to 3.46; P<0.001) for poor coordination, and 3.91 (2.61 to 5.87; P<0.001) for clumsiness (in the age 7 cohort); and 0.88 (0.81 to 0.96; P=0.003) for copying designs, 0.84 (0.78 to 0.91; P<0.001) for marking squares, and 1.14 (1.06 to 1.24; P<0.001) for picking up matches (a higher score indicates poor function in this test) for the age 11 cohort. Now, it should be clear, first, that the measures used to assess the two cohorts were different, and so can't be aggregated. We'll assume Daily Mail Reporter knew this, and chose to focus on the age 7 cohort. How can we assume this? The figures: the 57% figure came from one of two places. The first option is a misreading of 1.57 as 0.57 (perhaps (s)he blinked while reading). The second option comes from one of the study's results tables, which contains the number 5.7 as the incidence of teacher reports that the child 'certainly' had poor hand control. At the top of the table, and in the body of the text, it is explained that the figures in the table are percentages. Taking a look at the hard numbers, we see that there were 912 obese 33 year olds in the study, of whom 52 were reported to 'certainly' have poor hand control age 7. 52/912 x 100 gives us 5.7%. 5.7, not 57, and certainly not 57% more likely. Going back to option one, the odds ratio of 1.57, that's not 57% more likely in anyone's book

The study adds to evidence of a link between poorer cognitive function in childhood and obesity and Type 2 diabetes in adulthood.

Researchers said cognitive impairment in obese adults was assumed to be a consequence of obesity.
No they didn't

However, the study indicates obese adults and those with Type 2 diabetes may already have had lower levels of cognitive function in childhood 'consistent with a subtle developmental impairment'.
The study does not indicate this. This line comes from the introduction, and references two previous studies: Chandola T, Deary IJ, Blane D, Batty GD. Childhood IQ in relation to obesity and weight gain in adult life: the national child development (1958) study. Int J Obes 2006;30:1422-32 and Olsson GM, Hulting AL, Montgomery SM. Cognitive function in children and subsequent type 2 diabetes mellitus. Diabetes Care 2008;31:514-6. The line in question even has hyperlinks to the part of the reference section the studies are cited in, saving the bother of scrolling down the page to find them.

The research, whose authors are from the Karolinska Institute in Stockholm and Imperial College, London, is based on 11,042 people taking part in the ongoing National Child Development Study in Great Britain, which began in 1958.

Most of the words and phrases needed to report on medical research are right there in any given study. Often, the authors helpfully include a little three or four line 'Conclusion' after the discussion section that outlines the findings and puts them in context. Mangling figures does not add to the public's understanding of a study. It simply adds to the word count, pushing out space for the all important caveats and contextualisation needed in any science story. Including them for the sake of appearing learned and sciencey is about as convincing as donning a white coat, a pair of novelty thick rimmed glasses and mussing your hair up to look like Einstein.

2 comments:

crankynick said...

My guess is that the "57%" figure came from the 1.57 ratio - after all 1.57 is 57% more than 1, right?

healthy distrust said...

It's not a 1:57 ratio, though, it's an odds ratio of 1.57, which, in the study's results table, is a ratio of 1.13 to 2.2, adjusted. The actual percentage difference between 'certainly' having poor hand control in the obese vs non obese groups was 5.7-3.4, or a 2.3% difference. There's no way to translate this into a meaningful 'more likely' figure, as the incidence of obesity in the general population wasn't calculated in this study (and the authors don't make this claim, it should be noted). So, The Daily Mail was misleading on two counts: first, they extrapolated the data to the population at large; second, they, perhaps as you did, misread an odds ratio of 1.57 as being 1:57, which is something quite different.