Abstract
PURPOSE: Sleep-disordered breathing (SDB) is associated with increased risk for cardiovascular morbidity and mortality and for sleepiness-related accidents, but >75 % of the patients remain undiagnosed. We sought to determine the diagnostic accuracy of ECG-based detection of SDB when used for population-based screening. METHODS: All male workers, mostly truck drivers, of a transport company (n = 165; age, 43 ± 12 years) underwent standard attended overnight polysomnography. Cyclic variation of heart rate (CVHR), a characteristic pattern of heart rate associated with SDB, was detected from single-lead ECG signals during the polysomnography by a newly developed automated algorithm of autocorrelated wave detection with adaptive threshold (ACAT). RESULTS: Among 165 subjects, the apnea-hypopnea index (AHI) was ≥5 in 62 (38 %), ≥15 in 26 (16 %), and ≥30 in 16 (10 %). The number of CVHR per hour (CVHR index) closely correlated with AHI [r = 0.868 (95 % CI, 0.825-0.901)]. The areas under the receiver operating characteristic curves for detecting subjects with AHI ≥5, ≥15, and ≥30 were 0.796 (95 % CI, 0.727-0.855), 0.974 (0.937-0.993), and 0.997 (0.971-0.999), respectively. With a predetermined criterion of CVHR index ≥15, subjects with AHI ≥15 were identified with 88 % sensitivity and 97 % specificity (likelihood ratios for positive and negative test, 30.7 and 0.12). The classification performance was retained in subgroups of subjects with obesity, hypertension, diabetes mellitus, dyslipidemia, and decreased autonomic function. CONCLUSIONS: The CVHR obtained by the ACAT algorithm may provide a useful marker for screening for moderate-to-severe SDB among apparently healthy male workers.