Agreement Between Fitbit And Actigraph Measures Of Free-living Physical Activity In Adolescent Latinas

Consumer wearables are increasingly used as an intervention tool, but their validity as a physical activity measurement tool is still unclear, particularly in adolescent girls and other understudied demographic groups. PURPOSE: This study aimed to explore the minute- and day-level agreement between objective physical activity measures among adolescent Latinas. METHODS: Data were taken from a clinical trial for a physical activity intervention for Latina teens. Participants wore a Fitbit Inspire HR and ActiGraph GT3X+ for two overlapping weeks, one at baseline at one at six-months. ActiGraph data were classified using two different cutpoints (Freedson (Adult) and Treuth), which were aligned with Fitbit data to produce 3,502,011 matched minutes of wear across 136 adolescent girls ages 13-18 years (M = 15.73). Confusion matrices and agreement statistics were calculated for each pairwise comparison (Freedson v Treuth v Fitbit) to determine 1) agreement in minute-level physical activity intensity classification, 2) binary minute-level MVPA classification, and 3) binary day-level classification of meeting physical activity guidelines (752 valid days). Ordinal Logistic Regression (OLR) was performed at the minute-level to identify significant classification differences between devices, reported as an odds ratio. RESULTS: Freedson and Treuth cut points showed strong agreement for minute-level physical activity intensity classification (BA = 0.92, AC1 = 0.97), while Fitbit showed fair to moderate agreement (BA = 0.74, AC1 = 0.33-0.34) with comparable findings for binary minute-level classifications. When classifying a participant as having met daily physical activity guidelines, Freedson and Treuth cutpoints showed moderate agreement (BA = 0.84, AC1 = 0.33), while the Fitbit showed fair to moderate agreement with the ActiGraph (BA = 0.87-0.91, AC1 = 0.23-0.47). OLR findings suggest Fitbit is more likely to classify a minute at a higher intensity level (p < 0.0001). CONCLUSION: ActiGraph cut points showed strongest agreement at minute-level comparisons, with more disagreement when data are aggregated to the day level. Overall Fitbit showed fair to moderate agreement with ActiGraph in adolescent girls. Further research on the types of physical activity may provide insight into areas of disagreement. Supported by: This study is supported by a grant from the National Institutes of Health/National Institute of Nursing Research: R01NR017876.