Development of a quality assessment algorithm for multiple breath nitrogen washout

Abstract

Introduction: There is a need to integrate quality control (QC) into multiple breath washout (MBW) tests to facilitate accurate interpretation of results. We aimed to develop an algorithm to automate application of QC criteria for MBW. Methods: We defined quantitative measures of quality into a series of algorithms [Jensen R. et al. PLoS ONE; 2016] and applied these to MBW data collected on the Exhalyzer® D (EcoMedics AG). The performance of the algorithm was compared to independent assessment by experienced reviewers. Results: Anonymized MBW data (15,070 trials; 2,866 tests) in healthy children, and subjects with Cystic Fibrosis, were retrospectively evaluated using the algorithm. Seventy percent of trials were categorized as accept/exclude, whereas 30% were flagged for manual review. The algorithm agreed with the reviewer’s decision for 88% of trials. The lung clearance index was similar between the algorithm and reviewers (Δ -0.01, limits of agreement -0.31, 0.29) and resulted in a similar proportion of acceptable test occasions (86.4% vs. 83.6%). An equal number of trials flagged for review were subsequently classified as accept/exclude by reviewers. Conclusion: QC criteria can successfully be applied to MBW data using an algorithm and only requires review in approximately 30% of trials. The agreement between the algorithm and manual review was similar to previously observed inter-reviewer agreement [Jensen R. et al. PLoS ONE; 2016]; disagreement may reflect subjective interpretation.