“To tall men I’m a midget and to short men I’m a giant; to the skinny ones I’m a fat man and to the fat ones I’m a thin man….. In fact I’m quite ordinary.” So says the Ordinary Man in Norman Juster’s Phantom Tollbooth. However, current figures suggest that, in terms of weight, ‘ordinary’ no longer applies to the majority of England’s adults, 62 per cent of whom are now either overweight or obese. Excess weight is a major cause of chronic conditions such as heart disease, type 2 diabetes, and some cancers, leading to premature mortality and avoidable ill health. The Government’s call to action on obesity includes a commitment to providing data at a local level to achieve a downward trend in excess weight in adults by 2020. So, an indicator on excess weight in adults was included in the Public Health Outcomes Framework and the Active People Survey used to ask adults about their height and weight in order to generate local estimates.
A previous post on this blog examined how behaviour change can be used to challenge public perceptions around obesity. In this post we take a look behind the scenes at how height and weight are measured and what can be done to improve data quality when they are not reported accurately.
Excess weight is defined as adults with a body mass index (BMI) of greater than or equal to 25kg/m2. To find your BMI, divide your weight in kilograms (kg) by your height in metres (m) then divide the answer by your height again – or you could use the NHS Choices calculator.
You could be forgiven for thinking that height and weight are easy to measure. Surely you mark height with a pencil against the kitchen wall and get out the tape-measure? And don’t you just stand on the bathroom scales to find your weight? This is true, but although they are relatively easy to measure it would still be prohibitively expensive to collect height and weight measurements from enough people to provide data for local areas. This means we have to rely on self-reported height and weight if we want to produce statistics below regional level.
Research has shown that adults consistently overestimate their height and underestimate their weight, and if you're middle-aged then you're even more likely to get the figures wrong. In PHE we were aware of these discrepancies, so we investigated how to monitor excess weight reliably. We looked at the height and weight data that are currently collected from two sources: the Active People Survey (APS), which provides figures at a local level; and the Health Survey for England (HSE) 2006 to 2010 which provides national figures. Both surveys asked participants for their height and weight. The HSE 2011 also measured survey participants to confirm the height and weight that people self-reported.
Our investigations showed the differences between self-reporting and being measured varied according to age and sex. The largest differences occurred in adults over the age of 50. We tested whether self-reported figures could be adjusted to give better estimates of likely actual height and weight. We found that applying adjustments to the APS figures produced results that were very similar to the measurements that the HSE took of survey participants and also provided a good fit to the 2006-2008 local authority level modelled estimates for obesity prevalence.
Does it really make a difference which figures you use? Well, the self-reported, or unadjusted, figures from the APS, were found to give a prevalence of obesity in adults of 15 per cent and prevalence of excess weight of 49 per cent. By adjusting the figures to take account of inaccuracies in self-reporting, a prevalence of 23 per cent was found for obesity and 64 per cent for excess weight. The adjusted figures are close to the 2012 HSE estimates that suggested a prevalence of 25 per cent for obesity and 62 per cent for overweight and obesity combined.
The problems become even more obvious at local authority level. If the APS figures were not adjusted to take account of reporting inaccuracies, it is possible that excess weight prevalence would be under-reported by around a third overall and would range from 31 per cent to 63 per cent for individual local authorities. The reasons for this wide local variation are unclear but may include factors such as the age and sex structure and the distribution of BMI across the local population. When the data are adjusted the prevalence of excess weight across local authorities ranges from 46 per cent to 76 per cent.
A potential difficulty with producing data on excess weight at a local level is the possible influence of local attitudes. For example, do adults in Birmingham report their weight any differently from adults in Norwich? Or is height viewed differently in Rotherham than in Southampton? There is no reason to believe this is so but, unless both measured and self-reported height and weight are taken at a local level, there is no way to know for certain.
After much investigation, we decided to use adjusted figures for adult height and weight to provide robust data on excess weight for local public health teams. The first figures were published in the data tool in February. We expect the next data to be released later this year, which will give an indication of how the results might vary year-on-year. PHE will be analysing this data over the coming months to better understand the patterns and trends in prevalence of overweight and obesity at national, regional and local level.
Want to know more about this measure? Here is further explanation (follow the link and choose indicator 2.12 Excess weight in adults from the drop-down list).
More importantly, what do the data look like for your local area? Check for yourself.
Further supporting indicators for underweight, healthy weight, overweight and obesity, as well as both adjusted and unadjusted data, are available on the PHE Obesity website.
So the next time you are asked for your height and weight, ask yourself: how accurate are your estimates?
3 comments
Comment by Bren posted on
Hello Heather, Clare, Caroline, and Hywell,
Thanks for the really thoughtful and rightly so questioning blog.
I thought the consideration given to, how accurate is the measurement of height and weight measurements in the first instance really powerful. Like any self reporting measure, it is about bias and what it may not tell us too.
I also liked the focus on prevalence and not on snap shot of what the level of obesity is not without then looking at prevalence data.
Would the principles be related to blood pressure measurements, although I accept this is one measure and reported reading? In so far as any bias on the self taking and calibration of blood pressure readings.
Thanks,
Bren.
Comment by Clare Griffiths posted on
Hello Bren. We are glad you found the blog post thought provoking.
We think the kind of adjustment we have used with the height and weight data would be possible with blood pressure readings in principle. However in practise this might prove difficult to do.
One issue is that the general public is probably less aware of their blood pressure than their height or weight. As a result the self-reported blood pressure estimates may vary more from the true value than we have found with estimates of height and weight. If the variation between the estimates and the real values is large, or if there is little in the way of any systematic variation (i.e. if the difference between the measured and self-reported blood pressure readings is not found to vary in relation to other factors such as age or sex) then it would be difficult to apply this sort of adjustment.
When producing the adjusted obesity figures we were also able to make use of the HSE 2011 dataset, which collected both measured and self-reported height and weight from the same individuals. Our analysis of this dataset helped confirm that the adjustment we planned to applying could be used to adjust height and weight measurements at an individual level. If we were to consider making adjustments to self-reported blood pressure readings we would ideally need a dataset which contained both self-report and measured data on blood pressure. Unfortunately there are no large datasets which currently contain this information for the English population which would be an additional barrier to producing such statistics.
Comment by Hannah posted on
http://bmi-calories.com/bmi-calculator.html presents an interesting global BMI statistic (average bmi for both males and females) by country and gender using an interactive geo-map chart with data from 177 countries. It's interesting to see how the US country estimated county-level mean BMI compares to the rest of the world.