Criterion validity and agreement of the Brazilian version of the sedentary behavior questionnaire against the ActiGraph accelerometer

Abstract

Objective: This study aimed to assess the criterion validity of the Sedentary Behavior Questionnaire in college students by comparing it with data from an accelerometer. Methods: A convenience sample of 80 university students (73.7% female), from a public university in Brazil participated in the study. Participants wore an ActiGraph (wGT3X-BT) accelerometer for 7 consecutive days and completed the Sedentary Behavior Questionnaire via Google Forms. Additionally, to evaluate the reliability of the questionnaire, 20 participants completed the questionnaire again after a two-week interval for test-retest analysis. Descriptive analysis was performed for both continuous and categorical variables. Data distribution was evaluated using the Kolmogorov- Smirnov test. The paired t-test was performed to compare the reference method with the questionnaire, the intraclass correlation coefficient to assess reliability of the Sedentary Behavior Questionnaire, the Spearman’s correlation coefficient to verify the relationship between methods, and the Bland-Altman´s plot to assess the agreement between the values of both methods. Results: High test-retest reliability was found for weekdays (0.78), weekends (0.89), and total values (0.90). No significant statistical correlations were found in total values between the Sedentary Behavior Questionnaire and ActiGraph for cutoff points of <100 cpm (rho = -0.033, p = 0.774) and <150cpm (rho = 0.140, p = 0.216). Moreover, the questionnaire underreported sedentary behavior when adopting a cutoff point of <100 cpm and overreported it when the cutoff point was <150 cpm. Conclusions: The Sedentary Behavior Questionnaire showed high reliability for measuring sedentary behavior in a 14-day test-retest time period. However, a weak correlation and low agreement were found between the questionnaire and Actigraph, with the self-report overestimating and underestimating the sedentary behavior. Further research should be carried out to improve the correlation and agreement between the Sedentary Behavior Questionnaire and measuring devices, reducing the frequently attributed biases.