Accuracy of the end-expiratory lung volume measured by the modified nitrogen washout/washin technique: a bench study

The functional residual capacity (FRC) determines the oxygenating capacity of the lung and is heavily affected in the clinical context of the acute respiratory distress syndrome. Nitrogen-wash-in/wash-out methods have been used to measure FRC. These methods have rarely been validated against exactly known volumes. The aim of the study was to assess the accuracy and precision of the N₂ washout/washin method in measuring FRC, by comparing it with set volumes in a lung simulator.

We conducted a diagnostic bench study in the Intensive Care Unit and Radiology Department of a tertiary hospital in Switzerland. Using a fully controllable high fidelity lung simulator (TestChest®), we set the functional residual capacity at 1500 ml, 2000 ml and 2500 ml and connected to the GE Carestation respirator, which includes the nitrogen washout/washin technique (INview™ tool). FRC was then set to vary by different levels of PEEP (5, 8, 12 and 15 cmH₂O). The main outcome measures were bias and precision of the TestChest® when compared to the results from the washout/washin technique, according to the results of a Bland Altman Analysis. We verified our findings with volumetric computed tomography.

One hundred and thirty-five nitrogen-wash-in/wash-out measurements were taken at three levels of F_IO₂ (0.4, 0.5, 0.6). The CT volumetry reproduced the set end-expiratory volumes at the Simulator with a bias of 4 ml. The nitrogen-wash-in/wash-out method had a bias of 603 ml with acceptable limits of agreement (95% CI 252 to - 953 ml). Changes were detected with a concordance rate of 97%.

We conclude that the TestChest® simulator is an accurate simulation tool, concerning the simulation of lung volumes. The nitrogen wash-in/wash out method correlated positively with FRC changes, despite a relatively large bias in absolute measurements. The reference volumes in the lung simulator verified with CT volumetry were very close to their expected values. The reason for the bias could not be determined.