Impact of preoperative sleep disturbance on perioperative neurocognitive disorders in older adults undergoing major non-cardiac surgery: A multicenter prospective cohort study

BACKGROUND: Perioperative neurocognitive disorders (PND) significantly affect recovery in older adult surgical patients. However, whether untreated preoperative sleep disturbance (SD) independently contributes to postoperative cognitive dysfunction (POCD) remains unclear. METHODS: This multicenter prospective cohort study involving 535 patients aged ≥60 years undergoing major non-cardiac surgery (≥2 h) was conducted between June 30, 2024 and May 31, 2025. Patients were stratified by Pittsburgh Sleep Quality Index (PSQI): SD (PSQI>7, n = 288) or non-SD (PSQI≤7, n = 247). The primary outcome was POCD assessed on postoperative days 7, 30, 90, and 180. Secondary outcomes included postoperative delirium (days 1-3), 15-item quality of recovery (QoR-15) scores, and insomnia severity (days 30, 90, and 180). Generalized estimating equations identified independent predictors of POCD. RESULTS: Compared with non-SD patients, SD patients showed significantly increased POCD risk on postoperative day 7 (41.7 % vs. 27.1 %; RR = 1.44, 95 %CI 1.16-1.79; P < 0.001), day 30 (36.1 % vs. 18.2 %; RR = 1.73, 95 %CI 1.33-2.25; P < 0.001), day 90 (25.7 % vs. 13.0 %; RR = 1.66, 95 %CI 1.22-2.25; P < 0.001), and day 180 (19.4 % vs. 8.9 %; RR = 1.75, 95 %CI 1.21-2.52; P < 0.001). Preoperative SD was associated with increased delirium risk (29.9 % vs. 18.6 %; RR = 1.43, P = 0.006), poorer QoR-15 scores (difference = 8, P < 0.001), and persistent insomnia (34.7 % vs. 13.8 % at day 180, P < 0.001). PSQI≥10 optimally predicted POCD risk (sensitivity 71.8 %, specificity 69.4 %). CONCLUSIONS: Untreated preoperative SD independently predicts both early delirium and POCD after major non-cardiac surgery. A PSQI≥10 effectively identifies high-risk older adult patients, emphasizing preoperative sleep optimization as a potential strategy to mitigate postoperative cognitive impairment.