Comparative Effectiveness of Sleep Apnea Screening Instruments During Inpatient Rehabilitation Following Moderate to Severe TBI

Objective: To determine the diagnostic sensitivity and specificity and comparative effectiveness of traditional sleep apnea screening tools in traumatic brain injury (TBI) neurorehabilitation admissions. Design: Prospective diagnostic comparative effectiveness trial of sleep apnea screening tools relative to the criterion standard, attended level 1 polysomnography including encephalography. Setting: Six TBI Model System Inpatient Rehabilitation Centers. Participants: Between May 2017 and February 2019, 449 of 896 screened were eligible for the trial with 345 consented (77% consented). Additional screening left 263 eligible for and completing polysomnography with final analyses completed on 248. Intervention: Not applicable. Main Outcome Measures: Area under the curve (AUC) of screening tools relative to total apnea hypopnea index≥15 (AHI, moderate to severe apnea) measured at a median of 47 days post-TBI (interquartile range, 29-47). Results: The Berlin high-risk score (receiving operating curve [ROC] AUC=0.634) was inferior to the Multivariable Apnea Prediction Index (MAPI) (ROC AUC=0.780) (P=.0211; CI, 0.018-0.223) and Snoring, Tired, Observed, Blood Pressure, Body Mass Index, Age, Neck Circumference, and Gender (STOPBANG) score (ROC AUC=0.785) (P=.001; CI, 0.063-0.230), both of which had comparable AUC (P=.7245; CI, −0.047 to 0.068). Findings were similar for AHI≥30 (severe apnea); however, no differences across scales was observed at AHI≥5. The pattern was similar across TBI severity subgroups except for posttraumatic amnesia (PTA) status wherein the MAPI outperformed the Berlin. Youden's index to determine risk yielded lower sensitivities but higher specificities relative to non-TBI samples. Conclusion: This study is the first to provide clinicians with data to support a choice for which sleep apnea screening tools are more effective during inpatient rehabilitation for TBI (STOPBANG, MAPI vs Berlin) to help reduce comorbidity and possibly improve neurologic outcome.