LAUSR

The apnea hypopnea index and percentage of periodic breathing detected by the ventilator machine are often used by sleep doctors to evaluate whether sleep apnea has been adequately treated or need further interventions. There are concerns about the accuracy of this autodetection. Patients with sleep apnea who were treated with positive airway pressure at the Beth Israel Deaconess Medical Center (Boston) and tracked by the EncoreAnywhere system were included. The machine detected AHI(AHIm) and PB(PBm) were extracted from the first week data in every month from the start of use. The manual scored AHI(AHIs) and PB(PBs) were calculated from the last waveform graph during every month. The apnea hypopnea index as well as periodic breathing in 1st, 2nd, 3rd,6th month AHIm, AHIs, PBm and PBs were compared respectively. A total of 128 patients were included. The mean age was 56.5 and 66% of them were male. In the first month, the mean AHIs was significantly higher than AHIm, 16.27 vs. 5.36, p<0.001. There was also a large difference between percentage of PBs and PBm, 15.55% vs. 1.96, p<0.001. 78% patients whose AHIm <5 were actually has AHIs >5. The Kappa value for the AHIm and AHIs were 0.074, p=0.069; the value of PBm and PBs was 0.216, p=0.015. In the 2nd, 3rd and 6th months, the mean difference between AHIs and AHIm was 10.58, 10.68, 10.12, respectively. The mean difference between PBs and PBm was 12.32%,11.53%,and 9.18%. Autodetection of respiratory events consistently under-estimates the severity of residual events. Mean differences remained stable over 6 months. Caution is recommended when attributing non-apnea reasons for residual symptoms in patients with apparently low machine estimated AHI and PB. This study is supported by American Academy of Sleep Medicine Foundation, category-I award to RJT