Different analysis methods of Scottish and English child physical activity data explain the majority of the difference between the national prevalence estimates

• Main Authors: Chloë Williamson, Paul Kelly, Tessa Strain
• Format: Article
• Language: English
• Published: BMC 2019-02-01
• Series: BMC Public Health
• Subjects: Children; Physical activity; Surveillance
• Online Access: http://link.springer.com/article/10.1186/s12889-019-6517-7

Abstract Background The percentages of children in Scotland and England meeting the aerobic physical activity (PA) recommendation differ greatly according to estimates derived from the respective national health surveys. The Scottish Health Survey (SHeS) usually estimates over 70% meeting the recommendation; Health Survey for England (HSE) estimates are usually below 25%. It is plausible that these differences originate from different analysis methods. The HSE monitors the percentage of children in England that undertake 60 min of moderate-to-vigorous PA on each day of the week (‘Daily Minimum Method’ (DMM)). The SHeS monitors the proportion that undertakes at least seven sessions of moderate-to-vigorous PA, with an average daily duration ≥60 min in Scotland (‘Weekly Average Method’ (WAM)). We aimed to establish how much this difference in analysis methods influences prevalence estimates. Methods PA data from 5 to 15 year olds in the 2015 HSE and SHeS were reanalysed (weighted n = 3840 and 965, respectively). Two comparable pairs of estimates were derived: a DMM and WAM estimate from the HSE not including travel to/from school, and WAM estimates from the HSE and the SHeS including travel to/from school. It is not possible to calculate a DMM estimate from the SHeS due to questionnaire design. Results were presented for the total samples, and by sex and age sub-groups. Results The HSE WAM estimate was 31.7 (95% CI: 30.2–33.3) percentage points higher than the DMM estimate (54.3% (95% CI: 52.6–56.0) and 22.6% (95% CI: 21.2–24.1) respectively). The magnitude of this difference differed by age group but not sex. When comparable WAM estimates were derived from the SHeS and the HSE, the SHeS was 11.8 percentage points higher (73.6% (95% CI: 69.8–77.1) and 61.8% (95% CI: 60.2–63.5) respectively). The magnitude of this difference differed by age group and sex. Conclusions The results indicate that the difference in the analysis method explains the majority (approximately 30 percentage points) of the difference in the child PA prevalence estimates between Scotland and England (leaving approximately 12 percentage points representing true differences or related to questionnaire differences). These results will help national surveillance determine how to increase comparability between the U.K. home nations.