I modelled child obesity cut-offs as derived from parental perceptions, which differed greately from cut-offs based on objective measures. This paper generated a large amount of media interest.
There is a known disconnect between when parents recognise obesity in their own child, and how experts define obesity based on objective measurements of height and weight (expressed as age-specific BMI centiles). The expert BMI cut offs, typically used in the UK, are based off the centile points at which obesity is assumed to have a negative impact on health (either now or in the future) from a distribution of BMIs representative of children in the early 1990’s. This is why a histogram of BMI centiles of children now will not be centred on the 50th centile - as much more than half of kids today have a BMI centile above 50.
The discrepancy between how parents and doctors define obesity has not been explained simply. A fundamental problem in tackling obesity is that as more and more of a child’s friends have excess weight, parents also appear to recognise obesity at higher and higher BMI centiles.
In the study I was working on at LSHTM, led by Dr Sanjay Kinra, we had BMI of all children aged 4-5 and 10-11 from 5 primary care trusts that was routinely collected by school nurses, alongside a questionnaire where parents told us whether they believed their child was underweight, healthy weight, overweight or very overweight. While researching a methodological problem for another analysis, I came across a paper in Stats Med on how to take paired continuous and multinomial data, build a multinomial model, and back calculate the ‘cut points’ that would say at what level of the continuous predictor (BMI centile), the multinomial categories (parent reported weight status) were most likely. Effectively, I could calculate weight status cut offs on the BMI centile scale that represent how parents see their own child.
Measured BMI centile was matched with parent classification of weight status in 2976 children. Parents become more likely to classify their children as underweight when they are at the 0.8th centile or below, and overweight at the 99.7th centile or above. Parents were more likely to underestimate a child’s weight if the child was black or South Asian, male, more deprived, or the child was older. These values differ greatly from the BMI centile cut-offs for underweight (2nd centile) and overweight (85th).
Black, J.A., Park, M., Gregson, J., Falconer, C.L., White, B., Kessel, A.S., Saxena, S., Viner, R.M., Kinra, S. (2015) Child obesity cut-offs as derived from parental perceptions: cross-sectional questionnaire. British Journal of General Practice, e234-49. doi:10.3399/bjgp15X684385
This paper was covered by multiple news outlets, with the LSHTM press office identifying 267 online articles about the paper. Below is a selected list of links to articles about the paper I first authored, followed by a video by the BBC summarising the findings.