Long-term accuracy of DLCO devices

Introduction: Accurate diffusing capacity for carbon monoxide (DLCO) devices are required for clinical use and research studes; however, 25% of instruments do not meet accuracy standards (Jensen, R. et al. Eur Resp J 2009; 33:828-34). Objectives: We aimed to determine long-term accuracy of DLCO devices used in the COPDGene study. Methods: DLCO measurement was performed at all 18 COPDGene study sites from 2013 to 2020 using a ndd EasyOne Pro instrument (Zurich, Switzerland) that incorporates automated calibration and linearization of gas sensors and a flow sensor that does not require calibration. Each instrument was tested with a DLCO simulator (Hans Rudolph, Kansas City, MO) at three different simulated DLCO levels to verify and validate each instrument’s accuracy in 2013 and again in 2020. Comparisons were made between each instrument’s measured DLCO and the "targets" set by the DLCO simulator with acceptable accuracy defined as a difference Results: Mean differences between the three simulated targets and measured DLCO from 2013 were 0.561, 1.122 and -0.400 mL/min/mmHg. In 2020 these differences were 0.922, -0.416 and -0.378 mL/min/mmHg. No statistically significant difference was found between 2013 and 2020 in the average of the differences from simulator targets: Mean differences and drift reported as mL/min/mmHg for 2013 was 0.180 and for 2020 was 0.069; seven year drift was 0.112. Concluison: The 18 DLCO devices used in the COPDGene study using automated calibration and linearization of gas sensors and a calibration-free flow sensor produced consistently accurate, stable measurements in DLCO over seven years.