Longitudinal associations between accelerometry-assessed sleep clusters and obesity in adolescence and young adulthood: the German GINIplus and LISA birth cohorts study

Abstract Background Some studies have identified diverse sleep patterns in adolescents and adults by considering multidimensional objective sleep parameters. However, whether sleep patterns are consistent from adolescence to young adulthood and their long-term associations with obesity remain unknown. Purpose This study aimed to investigate the longitudinal associations of objectively measured and clustering-identified sleep patterns with body mass index (BMI) and overweight/obesity in adolescence and young adulthood, and to test whether these associations change over time or are modified by genetic risk. Methods A total of 1973 participants wearing 7-day accelerometers in either 15-year (n=1347, aged 14.3-16.4 years) and/or 20-year (n=1262, aged 19.5-22.4 years) follow-ups of German GINIplus and LISA birth cohorts were included. Sleep clusters were identified by K-means cluster analysis on 12 accelerometry-assessed sleep characteristics at each follow-up. BMI and overweight/obesity were measured by physical exams or self-/parent-reported questionnaires. Multivariable linear and logistic regression models with generalized estimating equations were tested. The interaction effects with time of follow-ups and polygenic risk scores (PRS, n=1087) for BMI were further examined. Results Although sleep characteristics changed from adolescence to young adulthood, such as shorter total sleep time (mean 7.2 to 6.6 hours), the consistent five sleep clusters were classified in both periods: good sleep, delayed sleep phase, sleep irregularity and variability, fragmented sleep, and prolonged sleep latency. Compared to the "good sleep" cluster, the "prolonged sleep latency" cluster showed significant associations with higher BMI (β=0.56, 95% confidence interval (CI) = [0.06, 1.05]) and increased odds of overweight/obesity (odds ratio (OR)=1.55, 95%CI= [1.02, 2.34]). PRS was associated with higher BMI (β=0.65, 95%CI= [0.39, 0.91]) and overweight/obesity (OR=1.52, 95%CI= [1.15, 2.00]), with no significant gene-sleep cluster interaction effect. Only among males, the interaction effects with time of follow-ups on overweight/obesity were observed in the "delayed sleep phase", "sleep irregularity and variability", and "fragmented sleep" clusters (p-interaction <0.05), showing more harmful effects with increasing age. Conclusions Five sleep patterns were identified consistently across adolescence and young adulthood, and the "prolonged sleep latency" pattern was associated with higher BMI and overweight/obesity, independent of genetic risk. For young adult males, "delayed sleep phase", "sleep irregularity and variability" and "fragmented sleep" patterns seemed to have more unfavorable impacts on overweight/obesity.