Frontiers in Public Health · Oct 31, 2026
Hypothetical Reallocation of Screen-Based Sedentary Time to Physical Activity and Sleep in Relation to Cardiometabolic and Mental Health in Young Adults: A Compositional Isotemporal Substitution Analysis
By Zhihao Du, Guanyu Lu
Background: Sleep, physical activity, and sedentary behavior are co-dependent within a finite 24-hour day. Screen-based sedentary behavior is prevalent in young adults and combines prolonged sitting with digital-media exposure, but conventional regression does not directly model time-for-time trade-offs. We examined model-based differences in cardiometabolic and mental-health outcomes associated with hypothetically reallocating screen-based sedentary time to sleep, light physical activity (LPA), or moderate-to-vigorous physical activity (MVPA). Methods: In 2,450 young adults with complete 7-day ActiGraph wGT3X-BT accelerometry and ecological momentary assessment data, daily time was represented as a five-part composition: sleep, LPA, MVPA, non-screen sedentary behavior, and screen-based sedentary behavior. Isometric log-ratio multivariable models, compositional isotemporal substitution, and restricted cubic splines estimated hypothetical reallocations. Outcomes were Patient Health Questionnaire-9 (PHQ-9), Generalized Anxiety Disorder-7 (GAD-7), Perceived Stress Scale-10 (PSS-10), homeostatic model assessment of insulin resistance (HOMA-IR), triglycerides (TG), and fasting plasma glucose (FPG). Estimates are cross-sectional and model-based, not causal. Results: The 24-hour budget was 435.2 min/day sleep, 220.5 LPA, 38.6 MVPA, 362.4 non-screen sedentary behavior, and 383.3 screen-based sedentary behavior. Reallocating 30 min/day from screen-based sedentary behavior to MVPA was associated with lower PHQ-9 (-1.78; 95% CI -2.31 to -1.25), GAD-7 (-1.34; -1.86 to -0.82), PSS-10 (-2.15; -2.85 to -1.45), HOMA-IR (-0.38; -0.52 to -0.24), TG (- 0.18 mmol/L; -0.26 to -0.10), and FPG (-0.16 mmol/L; -0.23 to -0.09). Reallocation to sleep showed larger modeled differences for psychological than metabolic outcomes; reallocation to LPA produced smaller but generally favorable cardiometabolic estimates. Continuous curves suggested diminishing marginal differences for screen-to-MVPA reallocation, with descriptive inflections near 22.5 min/day for PHQ-9 and 20 min/day for HOMA-IR; the screen-to-sleep GAD-7 curve plateaued near 32 min/day. Subgroup estimates were exploratory. Conclusions: Within this cross-sectional compositional framework, reallocating screen-based sedentary time to MVPA was associated with the broadest favorable pattern across psychological and cardiometabolic outcomes, whereas sleep and LPA showed outcome-specific associations. These hypothetical reallocations should not be interpreted as causal effects or individualized dose prescriptions. The results do not establish screen-specific associations relative to non-screen sitting, and subgroup plots do not demonstrate interaction effects. Prospective and intervention studies are needed.