Validação do System for Observing Fitness Instruction Time (SOFIT) para avaliar o nível de atividade física em escolares do ensino fundamental durante a aula de educação física

Abstract

Physical activity provides several health benefits. Physical Education classes are the space in which children develop their motor skills for active behavior throughout life. The quantification and monitoring of physical activity is necessary to assess the behavior of individuals as well as the effect of promotion programs and the contribution of contexts, mainly by precise and accessible methods. The aim of this study was to test the concurrent validity of the System for Observing Fitness Instruction Time (SOFIT) with accelerometry to assess the sedentary behavior and physical activity of elementary school students during Physical Education classes. This is a validation study with cross-sectional design. Six municipal schools in the urban area of São José dos Pinhais were intentionally selected. In each school, a class of 1st, 3rd and 5th year was selected and within each class, six students (three of each sex) to use the accelerometer, totaling 108 children. Sedentary behavior and physical activity during Physical Education classes were assessed using the direct observation method (SOFIT) and objectively by accelerometry (wGT3X-BT) simultaneously. The classes were videotaped and later analyzed, synchronizing the observations in 15-second epochs. Sedentary behavior and physical activity were categorized by three different cutoff points, suggested by Evenson and collaborators (2008), Puyau and collaborators (2002) and Romanzini and collaborators (2014), and by SOFIT. The Kruskal-Wallis test and Bonferroni correction were used to compare the counts values in each of the SOFIT categories. To test the association between the two instruments, Spearman’s correlation was used. Finally, to detect the classificatory capacity of the methods, sensitivity, specificity, Receiver Operating Characteristic Curve (ROC Curve), percentage of simple agreement and Cohen’s Kappa tests were used. All analyzes were performed using SPSS 23.0 software (p≤ 0.05). The most observed activity by SOFIT was standing (34.9%) which corresponds to light in the other classifications, light by the criteria of Evenson (53.9%) and Romanzini (40.9%) and sedentary (42.7 %) by the Puyau criteria. The accelerometry values were significantly different between the SOFIT categories except between lying and standing for the vertical axis and sitting and standing for the magnitude vector. The correlation between SOFIT and accelerometry was moderate and ranged from 0.591 to 0.602 depending on the classification model and the axis of the accelerometry. The discriminatory capacity of the different cutoff points was better for identifying sedentary behavior and vigorous activities, but weak for detecting mild and moderate physical activities. SOFIT and accelerometry have concurrent validity, but they cannot correctly discriminate between light and moderate activities.