Abstract
BACKGROUND: Disturbances in sleep-wake homeostasis (Process S) and circadian rhythm (Process C) are common precipitants of delirium, especially among older hospitalized adults. We conducted a systematic review and meta-analysis to test whether four sleep-modulation agents-melatonin, ramelteon, suvorexant, and lemborexant-lower delirium incidence or shorten its duration in hospitalized patients, with stratified analyses by drug class, age, and surgical status.
METHODS: We systematically searched PubMed, Embase, and CENTRAL through March 2025. We included randomized controlled trials (RCTs) and observational studies assessing delirium prevention with melatonin, ramelteon, suvorexant, and lemborexant in hospitalized adults (≥ 18 years), compared to placebo or standard care. Data synthesis was performed separately for RCTs and observational studies using random-effects models. Meta-regression was used to explore effect modifiers. Risk of bias was assessed using RoB2/ROBINS-I tools. Certainty of evidence was graded using the GRADE assessment.
RESULTS: Thirty-seven studies (27 RCTs, 10 observational) comprising 7845 patients were included. Among RCTs, melatonin (RR 0.94; 95% CI 0.72-1.22) and ramelteon (RR 0.63; 95% CI 0.39-1.03) showed no significant effect on delirium incidence, whereas orexin receptor antagonists were associated with a lower risk (RR 0.55; 95% CI 0.35-0.87). Evidence for a class difference was inconsistent across analytic approaches: a subgroup heterogeneity test suggested differential effects (interaction-p = 0.09), but the meta-regression found no between-class difference (p = 0.14). No other specific test for subgroup differences was statistically significant in RCTs. Meta-regression confirmed patient setting as a significant modifier in observational studies, but not in RCTs.
CONCLUSION: Sleep-wake pharmacotherapies may reduce incident delirium in hospitalized adults. In randomized trials, melatonin and ramelteon did not significantly reduce delirium incidence, whereas dual orexin receptor antagonists showed a possible benefit, but the meta-regression did not demonstrate a reliable between-class difference, and the evidence remains limited. Adequately powered randomized trials across inpatient settings are needed to clarify any true differences and define clinical relevance.