Age-dependent association between early inpatient motor activity and subsequent length of stay in depressive disorders: a retrospective cohort study

Background Objective motor behavior may provide complementary information about psychiatric inpatient trajectories, but higher movement is not necessarily clinically favorable and may have different meanings at different ages. We examined whether the association between early inpatient motor activity and subsequent length of stay (LOS) varies across adulthood in depressive disorders. Methods In this retrospective cohort study, we analyzed 255 adults hospitalized with ICD-10 depressive disorders during 2024–2025. Motor activity was averaged across the earliest three valid actigraphy days within the first inpatient week, and a dynamic landmark at the end of the third valid day separated exposure assessment from subsequent LOS. Gamma generalized linear models with a log link tested interactions between standardized activity measures and age modeled continuously with a 3-degree-of-freedom restricted cubic spline. Benjamini-Hochberg false-discovery-rate correction was applied across four activity-by-age interaction tests. Results Age modified the adjusted associations of vector magnitude per minute (VM/min), steps/hour, and moderate-activity percentage with post-landmark LOS after false-discovery-rate correction, whereas the sedentary-percentage interaction was not significant. Per 1-SD higher VM/min, expected post-landmark LOS was 14.3% shorter at age 20 (95% CI, -22.0% to -5.9%), near null at age 40 (-0.1%; 95% CI, -14.4% to 16.6%), and 18.6% longer at age 60 (95% CI, 7.7% to 30.7%). The principal interaction was supported by alternative outcome models, spline specifications, medication-adjustment strategies, influence analyses, and bootstrap resampling. Conclusions The clinical information conveyed by early inpatient motor activity varied with age rather than following a uniform direction. Age should therefore be considered when interpreting actigraphy in depressive-disorder inpatients, although external validation is needed before clinical use.