Contribution of pulse oximetry in relation to respiratory flow events in a home-based approach aimed at diagnosing obstructive sleep apnea

OBJECTIVE: To compare pulse oximetry with manual analysis against all signals of respiratory polygraphy. MATERIAL AND METHODS: This retrospective study estimated sensitivity (S), specificity (Sp) and positive/negative likelihood ratio (LR+/-) of the oxygen desaturation index (ODI-test) and apnea-hypopnea index (AHI-reference). RESULTS: 3854 patients (61.5% men) were included. Age, BMI, Epworth sleepiness scale and AHI were: 55 years (44-65), 30.9kg/m(2) (27-36), 7 points (4-11), and 14 events/hour (6-25), respectively. 18% showed an AHI <5 events/hour, 34% = 5 and <15, 27% = or > 15 and < 30, and 31% > 30. The S, Sp, and LR+/- of ODI for AHI = 5 events/hour was 93%, 92%, 12 and 0.08 with an accuracy of 93%. For AHI = 15 events/hour, the values were: S 94%, Sp 94%, LR+ 15 and LR- 0.06 and 94% accuracy (r((2)) Spearman: 0.92). CONCLUSION: In a population at a high risk for OSA, home-based pulse oximetry had a diagnostic accuracy > 90% when is compared against all respiratory signals obtained from simplified home sleep testing.