Apnea Obstructiva

302

Thorax 2000;55:302–307

Automated analysis of digital oximetry in the diagnosis of obstructive sleep apnoea
Juan-Carlos Vázquez, Willis H Tsai, W Ward Flemons, Akira Masuda, Rollin Brant, Eric Hajduk, William A Whitelaw, John E Remmers

Division of Respiratory Medicine, Department of Medicine, University of Calgary, Calgary, Alberta, Canada T2N 4N1 J-C Vázquez W H Tsai W W Flemons AMasuda R Brant E Hajduk W A Whitelaw J E Remmers
Correspondence to: Dr J E Remmers Received 9 June 1999 Returned to authors 9 July 1999 Revised manuscript received 17 September 1999 Accepted for publication 9 December 1999

Abstract Background—The gold standard diagnostic test for obstructive sleep apnoea (OSA) is overnight polysomnography (PSG) which is costly in terms of time and money.Consequently, a number of alternatives to PSG have been proposed. Oximetry is appealing because of its widespread availability and ease of application. The diagnostic performance of an automated analysis algorithm based on falls and recovery of digitally recorded oxygen saturation was compared with PSG. Methods—Two hundred and forty six patients with suspected OSA were randomly selected for PSG andautomated oV line analysis of the digitally recorded oximeter signal. Results—The PSG derived apnoea hypopnoea index (AHI) and oximeter derived respiratory disturbance index (RDI) were highly correlated (R = 0.97). The mean (2SD) of the diVerences between AHI and RDI was 2.18 (12.34)/h. The sensitivity and specificity of the algorithm depended on the AHI and RDI criteria selected for OSA casedesignation. Using case designation criteria of 15/h for AHI and RDI, the sensitivity and specificity were 98% and 88%, respectively. If the PSG derived AHI included EEG based arousals as part of the hypopnoea definition, the mean (2SD) of the diVerences between RDI and AHI was –0.12 (15.62)/h and the sensitivity and specificity profile did not change significantly. Conclusions—In a population of patientssuspected of having OSA, oV line automated analysis of the oximetry signal provides a close estimate of AHI as well as excellent diagnostic sensitivity and specificity for OSA.
(Thorax 2000;55:302–307) Keywords: sleep apnoea; polysomnography; oximetry; portable monitor

able, with sensitivities ranging from 40% to 100%.3–8 In 240 consecutive patients Séries et al found that oximetry had a 98%sensitivity for diagnosing OSA (apnoea hypopnoea index (AHI) >10/h) but a specificity of only 48%.8 We have previously described a monitor which used snoring (primarily) and oxygen saturation (secondarily) to identify respiratory events.9 In 129 patients referred for assessment of sleep apnoea who were evaluated with simultaneously performed PSG, the sensitivity and specificity ranged from 84% to 90% andfrom 95% to 98%, respectively, depending on the AHI diagnostic criterion employed. However, these findings were limited by possible bias since a prospective consecutive sample of patients was not used and the interpretative algorithm was developed retrospectively. In a subsequent unpublished pilot study the diagnostic performance of this monitor was improved by modifying the oximetry analysisalgorithm and eliminating the snoring signal from the analysis process entirely. We prospectively validated the new automated analysis algorithm (AAA) in a randomly selected group of patients referred to a university-based tertiary sleep referral centre. Methods Patients were recruited from the Alberta Lung Association Sleep Centre, which is the major sleep referral centre for southern Alberta. Allreferrals were considered eligible unless they met one or more of the following exclusion criteria: age 4% (type A) and (2) either a fall in oxygen saturation of >4% or an arousal (type B). Oxygen desaturation was defined as a decrease in oxygen saturation of >4% (at nadir) followed by an increase to within 1% of the prehypopnoea baseline value. A reduction in thoracoabdominal movement was defined as...