Reliability and variability of assessed sports/play features in a multi-national study: The International Study of Childhood Obesity, Lifestyle and the Environment (ISCOLE)

Purpose: To examine reliability and variability of objectively-measured aspects of the school physical activity environment across 247 schools in 12 countries, representing both the developed and developing world.

Methods: Audits of the school physical activity and nutrition environment were conducted by trained data collectors for all schools participating in the International Study of Childhood Obesity, Lifestyle, and the Environment (ISCOLE). This analysis presents results pertaining to features supporting sports/play. The availability and quality of eight features supporting sports/play were assessed as present and functional, present but non-functional, and not available. Interrater reliability was assessed in 52 of the schools, representing a minimum of 5% (median=8%) of the country-specific school samples. The amount of variation occurring within sites vs. between sites (ICC) was assessed using multilevel (random-effects) logistic regression models.

Results: Audits of the school environment were conducted in 246 schools attended by 10-year-old children (4 to 29 schools per country). For the eight sports/play features, item-level inter-rater reliability (kappa) ranged from 0.80 to 0.99. Of the eight sports/play features assessed, two to six showed low variability (present in <10% or >90% of schools) within the country-specific school samples. Across all features, there was significant variability across countries in average availability of features (ICC range: 35% to 76%).

Conclusions: Features of the school physical activity environment can be reliably assessed across diverse country settings. However, it is unclear whether the items assessed in the current instrument will have sufficient within-country variability to correlate with student physical activity levels.