Simultaneous laboratory-based comparison of ResMed Autoset with polysomnography in the diagnosis of sleep apnoea/hypopnoea syndrome

Abstract

ResMed Autoset (AS) is a simplified diagnosis system for obstructive sleep apnoea/hypopnoea syndrome (OSAS) based on the respiratory flow/time relationship by pressure variation measured through simple nasal prongs. A multicentre prospective trial was used to compare AS and polysomnography (PSG) for diagnosing 95 patients, with suspected OSAS. Physicians gave a pretest probability of the patient having OSAS. The apnoea/hypopnoea index (AHI) was compared between the two methods of diagnosis for the whole population and for subgroups according to the pretest probability. Twenty-four patients had AHI < 15 events x h(-1) on PSG and 19 AHI 15-30, and 52 patients had AHI > or = 30. Correlation between AHI assessed by AS and PSG was r=0.87 for total sleep time (TST), p<0.0001. A Bland and Altman plot gave an agreement between the two methods of +/-40%. For a threshold of AHI > or = 15 events x h(-1) to diagnose OSAS, AS has a sensitivity of 92%, specificity of 79%, positive predictive value of 93% and negative predictive value of 76%. With a pretest probability > or = 80%, sensitivity and positive predictive value were 98 and 100% respectively. Of six false negative, four had a high pretest probability (> 80%) or Epworth score > or = 10. Using these parameters as a criterion for proceeding to PSG after a negative AS study would mean that two apnoeic patients (AHI 20 and 17 events x h(-1) by PSG) would escape detection. The Autoset is useful for the detection of obstructive sleep apnoea but with high pretest probability and a negative Autoset result polysomnography should be performed.