Key points are not available for this paper at this time.
Background: Obstructive sleep apnea (OSA) is frequent but underrecognized after stroke, worsening prognosis, recurrence, and mortality. Polysomnography is rarely feasible in acute care, and existing screening tools have limited accuracy. We aimed to identify simple OSA clinical predictors to improve risk stratification in stroke patients. Methods: In this prospective study, 116 consecutive acute stroke patients (mean age 73 years, 57% male) underwent standardized clinical evaluation, Berlin Questionnaire, Epworth Sleepiness Scale (ESS), and home sleep apnea test during hospitalization. OSA was defined as apnea–hypopnea index (AHI) ≥ 15. Logistic regression identified independent predictors; the model’s performance was assessed by accuracy, sensitivity, specificity, and ROC curves. Results: OSA was diagnosed in 42 patients (36%). OSA patients showed higher NIHSS at admission (p = 0.048) and higher ESS scores (p = 0.047), but similar vascular risk factors and stroke subtypes compared to non-OSA patients. In a multivariate analysis, age (OR 1.05; 95% CI 1.00–1.10; p = 0.036) and witnessed apneas (OR 6.20; 95% CI 1.31–29.22; p = 0.021) were OSA independent predictors. The two-variable models achieved 72.9% accuracy, 90.3% specificity, 41.2% sensitivity, Nagelkerke R2 = 0.223, and AUC = 0.739 (p < 0.001), outperforming both the Berlin Questionnaire (AUC 0.596) and ESS (AUC 0.616). Conclusions: A simple model based on age and witnessed apneas reliably identified stroke patients at high risk for OSA, with good discriminative performance and higher accuracy than standard questionnaires. Its high specificity supports targeted allocation of sleep studies in resource-limited acute settings, potentially improving early detection, secondary prevention, and care pathways after stroke.
Figorilli et al. (2025) studied this question.