Objective: To validate a cardiorespiratory polygraphy system (BITMED NGP 140) by comparing it to conventional polysomnography in the diagnosis of sleep apnea-hypopnea syndrome.
Patients and methods: Polysomnography and cardio-respiratory polygraphy were performed simultaneously on 103 consecutive patients referred because of suspected sleep apnea-hypopnea syndrome. The Bland and Altman method and intraclass correlation coefficients were used to assess agreement between the 2 methods of measurement. Receiver operating characteristic (ROC) curves were used to calculate the yield of cardiorespiratory polygraphy compared to that of conventional polysomnography.
Results: Ninety-two valid studies were obtained for 72 men and 20 women (mean [SD] age: 52.4  years). By the Bland and Altman method, the difference between the respiratory event index obtained with the BITMED NGP140 and the apnea-hypopnea index (AHI) obtained by conventional polysomnography was 7.6 (13.2) in the manual analysis (95% confidence interval [CI], 4.9-10.4) and 12 (15.3) (95% CI, 8.8-15.3) in the automatic analysis. For a corrected AHI (AH/total time in bed) mean differences were -2.2 (5.9) and 2.4 (8.2) for manual and automatic analysis. The intraclass correlation coefficients were greater than 0.94. The areas under the ROC curves of the respiratory event index were greater than 0.97 for all cut points. For an AHI of 30 or higher, the best cut-off point determined by manual cardiorespiratory polygraphy analysis was 27 (sensitivity, 98% and specificity, 98%). For the different cut-off points cardiorespiratory polygraphy correctly classified between 92% and 98% of patients in both the manual and automatic analyses.
Conclusions: The BITMED NGP140 had good agreement with conventional polysomnography for the measurement of respiratory events and provided high diagnostic yield.