Sleep Disturbances, Metabolic Markers, and Outcomes After Stroke: A Retrospective Cohort Study in a Tertiary Hospital

Background: Stroke remains a leading cause of death and long-term disability worldwide. Sleep disturbances are increasingly recognized as potential factors influencing recovery after stroke. Therefore, in this study we examined the associations of sleep disturbances and metabolic markers with post-stroke outcomes. Methods: We conducted a retrospective study of adult patients with stroke-related presentations. The primary analysis included 270 patients with follow-up mRS data. Extracted variables included demographics, vascular risk factors, stroke subtype, imaging findings, sleep features, and selected laboratory markers. Functional outcome was classified as favorable, mRS 0-2, or unfavorable, mRS 3-6. Recurrent stroke burden was analyzed as 0-1 versus ≥2 documented events. Associations and predictive performance were assessed using group comparisons, LASSO logistic regression, and random forest models with repeated 10-fold cross-validation. Results: Among 270 stroke patients, 214 (79.3%) had favorable outcomes and 56 (20.7%) had unfavorable outcomes. Sleep disturbances were common, especially nocturnal awakenings (59.6%), increased sleep apnea risk (44.4%), circadian rhythm disturbances (28.9%), and insomnia (23.7%). Unfavorable outcomes were linked to older age, cardio-aortic embolism, large vessel/cortical stroke, abnormal vascular imaging, insomnia, and lower HDL. In LASSO analysis, age, steno-occlusive/atherosclerotic imaging, cardio-aortic embolism, and insomnia predicted unfavorable outcome, while HDL was protective. For recurrent stroke, small artery occlusion and hypertension with diabetes were retained. In predictive modeling, the best random forest model showed good discrimination AUC values of 0.791 ± 0.0126. Conclusions: Poorer stroke outcomes were associated with vascular factors, insomnia, and low HDL; recurrent events were mainly associated with small artery occlusion.