Title

Predictor of exertional hypoxemia: Dlco Vs. Dlco/va. a retrospective study.

Document Type

Conference Proceeding

Publication Date

2017

Publication Title

Am J Respir Crit Care Med

Abstract

Diffusion capacity for carbon monoxide (DLCO) is used to measure the transfer of gas across alveolo-capillary membrane of the lung. It is unknown if this is the best test to predict hypoxemia as it does not account for alveolar volume. As such, the DLCO to VA ratio has been used to account for alveolar volume, which has been thought to be a more useful predictor of hypoxemia. However, prior studies looked only at certain subgroups of patients and often looked at tests that were separated by a fair bit of time. We investigated which test (the DLCO or the DLCO/VA) best predicts hypoxemia during exercise test (6-min walk or O2 dosing tests) done in patients with a variety of pulmonary disorders. Methods: We reviewed our pulmonary function tests (PFTs) database to identify patients that had PFTs and exercise test done on the same day. We evaluated diffusion capacity based on percent predicted of normal diffusion capacity. Hypoxemia was scored as present if the lowest oxygen saturation during exercise was ≤88%. Logistic regression was used to assess the predictive ability of both the DLCO and the DLCO/VA. We compared these predictors using the area under the curve of the ROC curve. The 2 AUCs were compared statistically using the nonparametric approach of DeLong. We then picked an optimal cutoff using Youden's index so that the test characteristics of this cutoff could be evaluated. Results: Between March, 2013 and September, 2015, 1028 patients were identified that had both PFTs and exercise test on the same day. The DLCO and DLCOVA had an ROC AUC of 0.76 and 0.69 respectively, which were significantly different (p<0.0001. Figure). To further investigate the predictive ability of DLCO vs DLCO/VA, we looked at the AUC for subsets of patients with the following diagnoses: COPD, CHF, ILD, Sleep apnea, and Pulmonary Hypertension. In all these subgroups, DLCO had a statistically significantly higher AUC than DLVA. Based on the optimal cutoff point for DLCO (45% predicted), and the DLVA (70 % of predicted), sensitivity (58.5% vs. 52%), specificity (83% vs. 80%), positive predictive value (64% vs. 58%) and negative predictive value (79.5% vs. 75.7%) were higher for DLCO as opposed to DLCO/VA, respectively. [Graph Presented]. Conclusion: Both DLCO and DLCO/VA can be used to identify patients at risk for exertional hypoxemia, but the DLCO is a better predictive test than DLCO/VA.

Volume

195

This document is currently not available here.

Share

COinS