Detecting central sleep apnea in adult patients using WatchPAT—a multicenter validation study

Abstract

Study objectives: To assess the accuracy of WatchPAT (WP—Itamar-Medical, Caesarea, Israel) enhanced with a novel systolic upstroke analysis coupled with respiratory movement analysis derived from a dedicated snoring and body position (SBP) sensor, to enable automated algorithmic differentiation between central sleep apnea (CSA) and obstructive sleep apnea (OSA) compared with simultaneous in-lab sleep studies with polysomnography (PSG). Methods: Eighty-four patients with suspected sleep-disordered breathing (SDB) underwent simultaneous WP and PSG studies in 11 sleep centers. PSG scoring was blinded to the automatically analyzed WP data. Results: Overall WP apnea-hypopnea index (AHI; mean ± SD) was 25.2 ± 21.3 (range 0.2–101) versus PSG AHI 24.4 ± 21.2 (range 0–110) (p = 0.514), and correlation was 0.87 (p < 0.001). Using a threshold of AHI ≥ 15, the sensitivity and specificity of WP versus PSG for diagnosing sleep apnea were 85% and 70% respectively and agreement was 79% (kappa = 0.867). WP central AHI (AHIc) was 4.2 ± 7.7 (range 0–38) versus PSG AHIc 5.9 ± 11.8 (range 0–63) (p = 0.034), while correlation was 0.90 (p < 0.001). Using a threshold of AHI ≥ 15, the sensitivity and specificity of WP versus PSG for diagnosing CSA were 67% and 100% respectively with agreement of 95% (kappa = 0.774), and receiver operator characteristic (ROC) area under the curve of 0.866, (p < 0.01). Using a threshold of AHI ≥ 10 showed comparable overall sleep apnea and CSA diagnostic accuracies. Conclusions: These findings show that WP can accurately detect overall AHI and effectively differentiate between CSA and OSA.