The association of objectively and subjectively measured modifiable lifestyle factors with internalizing problems: The role of genetic confounding and shared method variance bias

BackgroundSleep duration and physical activity have been associated with internalizing problems, such as depression, in observational studies. However, genetic confounding and measurement error may introduce bias. We assessed genetic confounding in the associations of sleep duration and physical activity with internalizing problems using both device-based and questionnaire assessments to estimate shared genetic risk across different assessment methods in adolescents. MethodsIn this preregistered study embedded in the Adolescents Brain Cognitive Development cohort, we included unrelated European adolescents with both self-reported and Fitbit-measured sleep duration devices data (N = 2283) and data on moderate to vigorous physical activity (days/week, N = 2772). Adolescents had a mean age of 12-year-old (SD = 0.65), with roughly 48% female. The internalizing problem scores were derived from self-reports of the Brief Problem Monitor A genetic sensitivity analysis was conducted to assess genetic confounding by combining polygenic scores and molecular-based heritability estimates of internalizing problems. ResultsLonger sleep duration was associated with lower internalizing problems using both self-reported (-0.15-SD, 95% CI, -0.19 to -0.11-SD) and objective (-0.10-SD, 95% CI, -0.15 to -0.06-SD) assessments. More frequent moderate/vigorous physical activity was associated with lower internalizing problems using both self-reported (-0.09-SD 95% CI, -0.13 to -0.05-SD) and device-based (-0.06-SD, 95% CI, -0.09 to -0.01-SD) assessments. A higher internalizing polygenic score was associated with more internalizing problems and shorter self-reported sleep duration but not device-based assessed sleep duration. Substantial genetic confounding (81%) was found between self-reported sleep duration and internalizing problems, predominantly among boys. No clear evidence of genetic confounding was found in the association between device-based measured sleep duration and internalizing problems. Similarly, no genetic confounding was observed for measures of physical activity in relation to internalizing problems. ConclusionThe observed negative relationship between reported child sleep duration and internalizing problems may be partly due to genetic confounding, particularly among boys. This genetic influence likely captured some measurement error (i.e., shared method variance) in associations where exposure and outcome were assessed by self-reports. Observational studies relying on self-reports may overestimate the impact of adolescent sleep duration on internalizing problems, especially among boys. Questionnaire assessment of physical activity was less impacted.