We assessed the validity of using a sheet-shaped body vibrometer (SBV) as a portable monitoring device for obstructive sleep apnea (OSA) screening. Seventy consecutive patients with suspected OSA underwent simultaneous in-laboratory polysomnography (PSG) and SBV. We evaluated the screening accuracy of the respiratory event index (REI) obtained with the SBV, using the REI based on either the estimated total sleep time (REI_eTST) or time in bed (REI_TIB); these were compared to the apnea-hypopnea index (AHI) obtained via PSG. Bland-Altman plots indicated that the mean difference between REI_eTST and AHI was lower than that between REI_TIB and AHI (1.2 ± 19.8 vs. 6.5 ± 16.8). For AHI ≥ 15, the sensitivity and specificity at an optimal REI_eTST of 17.0 were 90.9% and 76.9%, whereas those at an optimal REI_TIB of 15.9 were 86.4% and 80.8%, respectively; moreover, for AHI ≥ 30, these values at an optimal REI_eTST of 26.0 were 89.5% and 88.2%, whereas those at an optimal REI_TIB of 23.8 were 84.2% and 92.2%, respectively. The optimal cutoff values of REIs for AHI of ≥ 5 were markedly different from those for AHI obtained via PSG (REI_eTST, 14.9; REI_TIB, 15.0), but close to those for AHI of ≥ 15; both had good sensitivities and specificities. REIs obtained via SBV performed well in moderate-to-severe, but not mild, OSA screening; REI_eTST showed a slightly higher sensitivity and a relatively closer value to the AHI obtained via PSG when compared to REI_TIB. We consider the SBV less acceptable for screening mild cases than more severe cases.
View full abstract