Travis’ Note: Today’s post comes from PhD Student Ash Routen. You can find out more about Ash and his work at the bottom of this post.
Consistent with the majority of developed countries, a significant proportion of children here in the UK are overweight or obese (around 30% of 10-11 year olds as of 2010). How do we know this? Well, since 2005 the UK Department of Health have been operating the ‘National Child Measurement Programme’ (NCMP) a nationwide public health surveillance initiative, which to the best of my knowledge is the biggest in Europe. Annually over one million children (aged 4-5 and 10-11 years) have their height and weight measured by teams of school nurses, these measures are then used to calculate body mass index (BMI), and then BMI determined weight status. This data is used to inform local planning and delivery of weight intervention/healthy lifestyle services, and to produce national overweight and obesity prevalence figures.
As BMI differs between genders and increases with age during childhood, the NCMP categorise the children’s weight status as underweight, healthyweight, overweight or obese, by comparing their BMI to children the same age and gender using a BMI growth reference chart. So, children above the 91st percentile (91% of all children used in the reference sample) are defined as overweight, and above the 98th as obese. In most regions, the child’s weight status is fed back to the parent’s and children by letter, with information on local weight management initiatives if they are categorised as underweight, overweight, or obese. Of course you can imagine the furore of some parent’s and thus the NCMP has attracted quite a lot of negative media attention (e.g.http://www.telegraph.co.uk/health/children_shealth/7514267/Letter-to-fat-four-year-old-prompts-complaint-from-obesity-group.html) as a result. As such there is great onus on ensuring the quality of data and identifying any sources of potential ‘error’, which include human (i.e. reliability of the nurses measurements), technical (i.e. reliability of the scales/height measuring device) and biological ‘error’(i.e. both daily and monthly variation in BMI).
I was interested in the impact of the time of day when the measurements are conducted. The nurses follow a standardised protocol, but can take the measurements at any time of day (and indeed the month of measurement may also vary). We know that our weight fluctuates throughout the day, and that we shrink a little after rising due to gravity pushing us back down! What we didn’t know was if combined variation in these measures would result in a change in BMI. Who cares right? they would be at no more or less risk of adverse health if their BMI shifts a little…but could it be enough for those whom are on the cusp of a BMI weight category (e.g. 90.5 percentile) to be differently classified due to the time of day they are measured?
What did we do?
To investigate this issue we took a sample of 74 children (aged 10-11 years) and measured their height and weight in the morning (0900-1045 hr) and again in the afternoon (1300-1500). From this we calculated their BMI, BMI percentile and weight status category using two set’s of BMI percentile cut-off’s, namely clinical cut-off’s (overweight: 91st and obese: 98th) as used by the NCMP and clinicians, and population monitoring cut-off’s as used mainly by researchers (85th and 95th centiles).
What did we find?
Not surprisingly in the afternoon all the children were shorter (-0.5 cm), however only girls were heavier (+0.1 kg), and BMI (+0.12 kg.m2), and BMI percentile was greater (+2.5 centiles) in all children. In relation to weight status categories there were no shifts in the number of people in each category from morning to afternoon, but on an individual level there were some interesting findings. When applying the clinical BMI cut-off’s we saw that one girl moved from healthyweight to overweight, and using the population monitoring cut-off’s , two girls moved from the healthyweight to overweight category, and one moved from the overweight to obese category with BMI increases of only 0.30, 0.55 and 0.26 kg/m2, respectively.
What are the implications?
We saw that it only takes a height loss of about 1 cm and an increase in weight of about 150g to shift a girls BMI category if they are near to the cut-off threshold. As only a few individuals changed (and this was a small sample) the results may seem inconsequential. However both on an individual and national level there may be some impact. Nationally, comparison of prevalence data (and thus future direction of resources) between schools and regions (and this is where I speculate) may be may be clouded if they measure a greater proportion of their children either in the morning or the afternoon. Whilst the extrapolation of the present observations using the clinical BMI cut-off’s (which the NCMP use in parental feedback) to the potential impact on the national NCMP data is tenuous, it is worthy of consideration. As the time of day when measurements are taken is not standardised, or recorded by the NCMP it could be supposed that 50% of the measurements taken are performed in the morning and 50% in the afternoon. If one in every 27 (3.7%) healthy weight girls (as we found in our sample) were on the cusp of overweight in samples measured in the morning they could well have been categorised as overweight had they been measured in the afternoon. Out of the 162,640 healthyweight girls measured by the NCMP in 2009/10 this would represent 6017 girls being classified as overweight instead of healthyweight, which hinders a ‘true’ assessment of prevalence data.
Arguably of more importance is that potentially 6017 parents and children would be informed that their child is overweight, due to their misfortune of being measured in the afternoon as opposed to the morning. We know that children labelled as overweight may be at greater risk of stigmatisation, teasing and anxiety; it is not unimaginable therefore that such a letter could trigger unhealthy activity and dietary habits and unnecessary parental intervention. For all the useful information such screening programmes provide us researchers, we must be cognisant of the discourse surrounding the issue of childhood obesityand consider the impact of such surveillance programmes on individual children and families (see Michael Gard’s work for a thought provoking viewpoint: http://bod.sagepub.com/content/13/4/118.extract).
What can we do?
In our paper we conclude that arguably, to increase data reliability the time of day in which the measurements are performed should be standardized (to either morning or afternoon) by the NCMP and indeed any public health surveillance programme that does not standardise measurement – this would at least ensure that all children are treated equitably. We do not have one ‘true BMI’, but fluctuate about a mean value on a daily and weekly basis. Therefore for the purposes of comparison, and analysis of trends year-on year, we should choose either to measure in the morning or afternoon. However, on the individual level it appears wise to ensure that children are measured in the morning to avoid unfavourable shifts in weight category and associated psychosocial implications of labelling. Standardisation of the timing of taking the measurements is one simple revision to the procedures of such surveillance programmes that could help to limit the impact of at least one potential ‘error’ variable.
About the author: Ash Routen is in the final months of his doctoral studies at the University of Worcester, UK examining the impact of pedometer interventions on habitual PA in kids, with an interest in the assessment of body composition and objective physical activity measurement in kids. He can be found on Twitter @AshRouten.
Routen, A., Edwards, M., Upton, D., & Peters, D. (2011). The impact of school-day variation in weight and height on National Child Measurement Programme body mass index-determined weight category in Year 6 children Child: Care, Health and Development, 37 (3), 360-367 DOI: 10.1111/j.1365-2214.2010.01204.x