Sleep duration, depressive symptoms, and postoperative delirium: a mediation analysis

Purpose The mechanism linking postoperative sleep duration to postoperative delirium (POD) remains unclear. This study examined the potential indirect pathway involving depressive symptoms (assessed via the PHQ-9) in the association between postoperative sleep duration and POD. Methods We included 500 patients aged ≥ 65 years undergoing unilateral total hip or knee arthroplasty under general anesthesia. The associative pathways among sleep duration, depressive symptoms, and POD were assessed using covariate-adjusted three-step linear probability models with heteroscedasticity-consistent (HC3) robust standard errors, validated via 1000-resample bootstrapping and the Sobel test. Results Postoperative sleep duration had significant total (β=-0.174, 95% CI -0.190, -0.158) and direct (β=-0.045, 95% CI -0.057, -0.032) associations with POD. An indirect pathway involving depressive symptoms was statistically significant (β=-0.129, 95% CI -0.146, -0.113). Adjusted paths for sleep duration→PHQ-9 (β=-2.660, 95% CI -2.913, -2.407) and PHQ-9→POD (β=0.049, 95% CI 0.045, 0.052) were significant. Sensitivity analysis showed E-values of 1.67, 1.26, and 1.53 for total (OR = 0.840), direct (OR = 0.956), and indirect (OR = 0.879) pathways, respectively. Conclusion Depressive symptoms are indirectly associated with the relationship between postoperative sleep duration and POD, alongside a direct independent association of sleep duration. Concurrent assessment and management of sleep and depressive symptoms are essential for optimizing perioperative care.