Validity and reliability of a telephone survey of physical activity in Brazil.

Introduction: Physical activity in Brazil is assessed by the Telephone-based Risk Factor Surveillance System for Chronic Diseases (VIGITEL). Objectives: To evaluate the reliability and concurrent validity of VIGITEL physical activity indicators. Methods: For the reliability study, 305 individuals responded to VIGITEL interviews, which were repeated with intervals of 7 - 15 days, in Belo Horizonte, 2013. The evaluated indicators included "sufficiently active on leisure time," "active in transportation," "inactive in four domains of physical activity (leisure, work, transportation, and housework)," and "watching TV for long periods." Kappa coefficient (k) was used to measure agreement between both interviews. For concurrent validity assessment, the same subjects also responded to the Global Physical Activity Questionnaire (GPAQ), used as reference method for comparison of VIGITEL indicators. Comparison was assessed by measures of sensitivity, specificity, positive (PPV), and negative predictive values (NPV). Results: Reliability study showed substantial agreement for active individuals in leisure time (k = 0.70) and inactive individuals (k = 0.64). The agreement of watching TV for long periods was moderate (k = 0.56) and the activeness in transportation showed fair agreement (k = 0.35). In comparison to the reference method, sensitivity ranged from 54.8 to 67.7 in frequency of inactive, active in leisure time and work. Transportation domain was represented by 11.9 of sensitivity. Specificity ranged from 72.0 to 91.2 among four domains of physical activity and inactive. Conclusion: Physical activity questionnaire used by the surveillance system seems to be reliable in all domains, except in transportation questions. VIGITEL was considered comparable to GPAQ in most aspects of physical activity evaluation.