Abstract
BACKGROUND: Sleep problems are common among older adults and are associated with a wide range of adverse health outcomes. Concerns about pharmacological treatments have increased interest in non-pharmacological interventions; however, evidence comparing their relative effectiveness remains limited. METHODS: A systematic search was conducted in PubMed, Scopus, Embase, Web of Science, Cochrane Library, and CINAHL. Randomized controlled trials (RCTs) evaluating non-pharmacological interventions in adults aged ≥ 60 years published between 2000 and 2024 were included. Network meta-analyses were conducted using random-effects models to estimate standardized mean differences (SMDs) with 95% confidence intervals (CIs). P-scores were used to rank the efficacy of interventions. The protocol was registered in PROSPERO (CRD42024521492). RESULTS: Thirty-four RCTs involving 3078 participants and 21 interventions were included. Eleven interventions significantly improved sleep quality. Cognitive behavioral therapy for insomnia plus positive mood strategies (CBT-I+) showed the largest effect (P-score = 0.99, SMD = - 3.32, 95% CI - 4.59 to - 2.06), followed by cognitive behavioral therapy for insomnia (CBT-I) (P-score = 0.92, SMD = - 2.18, 95% CI - 3.04 to - 1.31). Subgroup analyses indicated that music therapy (MUS) was more effective among participants with PSQI < 10 (SMD = - 1.25, 95% CI - 1.85 to - 0.65), whereas CBT-I+ showed greater effects for those with PSQI ≥ 10 (SMD = - 5.48, 95% CI - 6.80 to - 4.16). By intervention setting, traditional Chinese health-promotion exercise (TCHPE) was more effective in home-based settings (SMD = - 1.55, 95% CI - 2.60 to - 0.50), whereas CBT-I+ showed greater effects in non-home settings (SMD = - 3.31, 95% CI - 4.57 to - 2.06). CONCLUSIONS: CBT-I+ was associated with the greatest improvements in sleep quality among older adults, particularly those with baseline PSQI ≥ 10 and in non-home settings. MUS showed greater benefits among older adults with baseline PSQI < 10, and TCHPE showed greater benefits in home-based settings. These findings support stratified, context-specific intervention selection. Given the low GRADE certainty, these results should be interpreted with caution.