One in four adults in England are currently estimated to obese. This figure has tripled since the 1980s and follows similar trends internationally. This is important since excess body weight is associated with multiple adverse health outcomes including Type II Diabetes, Stroke, Osteoarthritis and Depression. This places considerable burden on health services costing the NHS over £5 billion annually. If trends continue, this figure is set to increase.
When analysing trends in body weight, previous studies typically only considered changes in either the prevalence of obesity or in the average (mean) body mass index (BMI). However, these metrics only tell us one part of the story. Focusing just on the average BMI value, for example, will only tell us how mean BMI has changed over time and therefore ignores what is happening throughout the whole range of BMI values. The middle can easily hide differences above and below it, and hence it is important to also analyse changes throughout the whole distribution to understand population-level trends.
Our study sought to illuminate how the distribution of BMI has changed over time through examining the dispersion and distribution of BMI between 1992 and 2013 in England. Data were taken from the Health Survey for England. We measured dispersion of BMI through calculating the standard deviation (a measure of how spread out the values are). To examine the distribution of BMI, we calculated the median (middle value), 5th (i.e. the value at which the bottom 5% lowest values are under – the lower end of the distribution) and 95th (the upper end of the distribution) percentiles of BMI.
We found that for males and females, the dispersion of BMI values has increased over the period suggesting that the distribution had become wider. Calculating the median, 5th and 95th percentiles allowed us to pick apart what was driving this (Figure 1). Median BMI increased between 1992 and 2000, before it started to decline and level off by 2013. The lower end of the distribution remained flat, whereas the upper end increased throughout the whole period. These findings highlight that simply focusing on the median would lead us to believe that there has been a slowing down of the trend of rising body weight, when in fact the upper end of the distribution has also increased.
Figure 1: Estimated median, 5th and 95th percentiles of body mass index for England, 1992–2013.
Surprisingly, these patterns remained consistent when we disaggregated them by socio-economic status (measured by level of education). Each social group witnessed increased dispersion, caused by growth at the upper end of the distribution. This suggests that something else than socio-economic differences is at play in driving this diverging trend.
Our results support an emerging international evidence base that has shown similar trends of dispersion and a growing upper end of the distribution for the US and in some low- and middle-income countries. These findings are important because it is the upper end of the distribution of BMI values where individuals are at the highest risk of developing conditions such as Type II Diabetes or Stroke. As this part of the distribution continues to shift upwards, we may expect growing prevalence of obesity-related health conditions. The slowing down of median BMI hailed by some public health officials may not be quite the success it appears at first.
The question that remains now is why? What is driving these changes? We are not sure at the moment, but the increasing international evidence of this phenomena might indicate that something wider is occurring in our societies. Opening a dialogue on the matter is only the first step, but an important one to understanding this mystery.
Answers on a postcard please.
Green, MA, Subramanian, SV, Razak, F, 2016, Population level trends in the distribution of body mass index in England, 1992-2013, Journal of Epidemiology and Community Health, http://jech.bmj.com/content/early/2016/02/16/jech-2015-206468.full.