[Evaluation, correction and impact of non-response in studies of childhood obesity].

OBJECTIVE: To evaluate and correct the impact of non-response in the prevalence of underweight, overweight and obesity in children aged 6 to 15 years old using silhouette scales.

METHOD: Cross-sectional study carried out in 2013 among 8,145 Galician schoolchildren aged 6-15 years old. The students who agreed to participate were weighed and measured and, based on body mass index, the prevalence of underweight, overweight and obesity was estimated. Teachers rated all students using silhouette scales. The valuations were used to estimate the prevalence corrected by non-response. Using the Bayes theorem, participation rates were estimated according to weight status.

RESULTS: The participation rate was 92.3% in the 6 -to 11-year-old group, and 90% in the 12- to 15-year old age group. In both groups, the prevalence of underweight and overweight were similar between participants and non-participants. However, obesity was higher among non-participants, especially at 12 to 15 years of age (6.3% vs. 12.2% ; p < 0.05). The prevalence did not change when corrected by the teacher's valuation. The participation rate of obese students was lower than the overall rate (82% vs. 90% at 12 to 15 years old; p < 0.05).

CONCLUSIONS: The presence of participation bias, which was greater at 12-15 years old, was confirmed. However, the impact of the bias on prevalence was negligible due to the high participation rate. In obesity studies with objective measures, it is essential to quantify non-participation, as well as to assess its impact and correct it.