Background: Despite high prevalence in the stroke population, sleep apnea is underdiagnosed. Obstructive sleep apnea is associated with poor cardiovascular outcomes and treatment with continuous positive airway pressure has been shown to lower blood pressure. No standard exists for screening patients who present to the hospital with acute stroke. We assessed three screening tools, the Epworth Sleepiness Scale (ESS), Berlin Questionnaire (BQ), and STOP-BANG Questionnaire (STOP-BANG), along with the use of a portable sleep study device for evaluation of sleep apnea. Overnight polysomnography (PSG) was performed on a subset of patients on outpatient follow up.
Methods: Patients admitted to the stroke unit at our hospital, over nine months, were screened for sleep apnea using the three instruments, ESS, BQ, and STOP-BANG. The patients were evaluated with a portable sleep study device, ApneaLink Air (ResMed, USA), prior to discharge. Respiratory effort, respiratory flow, pulse oximetry, and oxygen saturation were recorded and sleep apnea was determined by apnea-hypopnea index (AHI) ≥ 5. Predictions from the screening tools were compared to the portable sleep study and overnight PSG results. Sensitivity and specificity testing were used to assess the validity and reliability of the tools.
Results: Sleep questionnaires were administered on 37 patients who underwent an overnight sleep study. Portable studies were used to evaluate 33 patients, and 13 PSGs were performed. Obstructive sleep apnea was diagnosed in 20 (69%) and central sleep apnea in 9 (31%). Cheyne-Stokes pattern breathing was observed in 2 (5%). Mean AHI was 18.3 + 21.8/hr and maximum AHI was 105.8/hr. Sensitivity for the ESS, BQ, and STOP-BANG were 0.39, 0.66, and 0.83 and specificity for these tools were 0.26, 0.33, and 0.29, respectively. In patients who underwent the portable sleep study and overnight PSG, 9/10 (90%) of the studies were concordant.
Conclusions: The STOP-BANG questionnaire, administered to hospitalized stroke patients, had high sensitivity and low-moderate specificity in our study, compared to two other commonly used screening tools. Further, the feasibility of using an unattended inpatient portable sleep study on stroke inpatients is demonstrated.