Prevalence and prediction of exercise-induced oxygen desaturation in patients with chronic obstructive pulmonary disease.

BACKGROUND Previous studies with small sample sizes reported contradicting findings as to whether pulmonary function tests can predict exercise-induced oxygen desaturation (EID). OBJECTIVE To evaluate whether forced expiratory volume in one second (FEV(1)), resting oxygen saturation (SpO(2)) and diffusion capacity for carbon monoxide (DLCO) are predictors of EID in chronic obstructive pulmonary disease (COPD). METHODS We measured FEV(1), DLCO, SpO(2) at rest and during a 6-min walking test as well as physical activity by an accelerometer. A drop in SpO(2) of >4 to <90% was defined as EID. To evaluate associations between measures of lung function and EID univariate and multivariate analyses were used and positive/negative predictive values were calculated. Receiver operating characteristic curve analysis was performed to determine the most useful threshold in order to predict/exclude EID. RESULTS We included 154 patients with COPD (87 females). The mean FEV(1) was 43.0% (19.2) predicted and the prevalence of EID was 61.7%. The only independent predictor of EID was FEV(1) and the optimal cutoff value of FEV(1) was at 50% predicted (area under ROC curve, 0.85; p < 0.001). The positive predictive value of a threshold of FEV(1) <50% was 0.83 with a likelihood ratio of 3.03 and the negative predicting value of a threshold of FEV(1) ≥80% was 1.0. The severity of EID was correlated with daily physical activity (r = -0.31, p = 0.008). CONCLUSIONS EID is highly prevalent among patients with COPD and can be predicted by FEV(1). EID seems to be associated with impaired daily physical activity which supports its clinical importance.