Improvements in the data quality of a national BMI measuring programme

Abstract

Variations in data collection between collecting regions can affect the outcome measures. This study examines the impact of improvements in data collection on outcome measures in a national monitoring programme between 2007/2008 and 2010/2011. Multilevel analysis of 2007/2008 and 2010/2011 National Child Measurement Programme (NCMP) data estimated the relationship between body mass index (BMI) z-score and data collection variations within coordinating regions, while adjusting for individual-level and school-level factors. The total sample was 2 013 285 students from 17 279 primary schools in 152 coordinating regions in England. Data collection differences accounted for 31.2% of the regional variation in BMI z-score for Reception (aged 4-5 years) students in 2007/2008; this reduced to 12.6% in 2010/2011. For Year 6 (aged 10-11 years) students, it reduced from 5.3% in 2007/2008 to 2.4% in 2010/2011. Digit preference in the rounding of weight measurements showed the largest decreases, from 27.3 to 4.5% for Reception year pupils and from 4.2 to 1.0% for Year 6 pupils. This demonstrates that improvements in data collection variation between regions in the NCMP have led to improvements in data quality.