The use of Latent Class Analysis (LCA) to Assess Children's Movement Behaviours Measured by Accelerometer and Self-report

Latent Class Analysis (LCA) is a statistical method that can help researchers interested in better understanding the movement behaviors (MB) of children, based on the analysis of the level of physical activity (PA) and sedentary time (ST). This study aimed to evaluate and compare two models LCA (one for the accelerometer and one for the 24-hour recall) that represent the MB of children. A cross-sectional study involving 101 10-year-old Brazilian children. The classes were based on vigorous PA (VPA), moderate PA (MPA), light PA (LPA), and sedentary time (ST). To assess these behaviors, a 24-hour recall and an accelerometer were used. The accelerometer was used during four days. The time spent on each of the MB was categorized dichotomously, based on the 25th and 75th percentiles. Thus, "adequate" times were considered when the ST was below 25thP, the LPA and MPA were above 25thP and the VPA was above 75thP. LCA was used to model the variable "MB." For each latent class model (accelerometer and recall), two classes were found: "Adequate MB" and "Inadequate MB." Regardless of the method (accelerometer or self-report), the values of the "Inadequate MB" class had higher prevalence. Self-report predicted higher PA and lower ST compared to the accelerometer. The model based on accelerometry revealed that girls were 2.11 times more likely to belong to the "Inadequate MB" class when compared to boys. LCA was a multivariate statistical method that allowed the integrated evaluation of parameters that represent the MB analyzed by device-based and self-report methods.