RESUMO
BACKGROUND: Wide participation in electronic surveys and reliable reporting of anthropometry can serve to reduce costs associated with monitoring of obesity among adolescents where resources are limited. We conducted a single school pilot study among Caribbean adolescents to assess use of electronic surveys and whether face to face encouragement could promote enrollment. In addition, we assessed students' ability to reliably report simple anthropometry. METHODS: Students were provided with access to an electronic survey on anthropometry and food preferences regarding school-based food offerings. Responses to survey questions were presented as percentages. A sample of students also had their heights and weights measured after reporting these measures from memory. Intra-class correlation coefficients were used to assess reliability among measurers and Bland-Altman plots, consistency between student reported and recorded anthropometric measures and Support Vector Machine to assess robustness of anthropometry prediction models. RESULTS: Response rate to the electronic survey was low (9%). Students were able to interpret questions; open-ended options were inappropriately used 13% of the time. Post survey qualitative responses indicated displeasure with use of school-associated e-mail addresses. Concerns with confidentiality were expressed as well as preference for completion of surveys during school time. Students reliably reported anthropometry most measures fell within the 95% CI of Bland-Altman plots. SVM classified with a prediction accuracy of 95%. Estimates of overweight from recorded and reported measures were similar. CONCLUSIONS: Adolescents are able to report simple anthropometry, and this can be used to help with monitoring of growth and overweight. Although they are capable of competently completing electronic surveys, school-based email is an ineffective contact tool. In-person school-based contact and administration of surveys are preferred. Adolescents can reliably report simple anthropometry that can be utilized for estimation of overweight/obesity prevalence. This method can be widely applied.