Over-reading, or data quality control assessment, identified 12% of multiple breath washout measurements lacked technical accuracy or raised quality concern issues. These study findings, published in Respiratory Medicine, indicate that the inclusion of invalid data decreases the validity of multiple breath washout measurements in clinical trials involving adults with bronchiectasis.
Researchers of the BRONCH-UK Clinimetrics study (ClinicalTrials.gov Identifier: NCT02468271) evaluated the effect over-reader assessment has on the quality of multiple breath washout results and its measurement outcomes, including lung clearance index, lung clearance index coefficient of variation, and tidal volume coefficient of variation, to ensure accurate results in clinical trials involving a range of respiratory diseases.
Multiple breath washout measurements were collected at 3 sites naive to multiple breath washout testing and at 2 sites considered expert in multiple breath washout testing. All sites used Ecomedics Exhalyzer® D and standard cystic fibrosis operating procedures. An over-reader team assessing the validity and quality of the measurements evaluated all test results.
Data from 98 patients with idiopathic or postinfectious bronchiectasis yielded 250 multiple breath washout tests. Of the total measurements taken, 10 tests were not completed and 30 tests were unacceptable for analysis: 5 were disqualified by the site-reader, and 25 were disqualified by the over-reader. The excluded tests did not belong to patients with a core characteristic or a specific level of disease severity. The disqualified tests had a significantly larger lung clearance index coefficient of variation compared with acceptable tests (mean [SD] CV%, 7.9 [2.3] vs 2.7 [0.2], P =.004).
The site-reader results of lung clearance index, lung clearance index coefficient of variation, and tidal volume coefficient of variation were not significantly different from the over-reader results. The sites naive to multiple breath washout testing had a significantly larger change in lung clearance index than the sites with testing experts (mean [SD] change, 0.16 [0.39] vs 0.05 [0.15], P =.047). In addition, the sites naive to testing benefitted from over-reader assessment throughout a 12-month time frame, indicating that experience in multiple breath washout testing improved the accuracy of the results.
The researchers concluded that involving an over-reader for multiple breath washout measurements in clinical trials requires many resources but produces study outcomes with greater validity. While there were not significant changes in results for measurement outcomes in this study after over-reader assessments disqualified 12% of the tests, these results “would have incorrectly been included in the database for analysis.”
O’Neill K, Lakshmipathy GR, Ferguson K, et al. Quality control for multiple breath washout tests in multicentre bronchiectasis studies: experiences from the BRONCH-UK clinimetrics study. [published online October 30, 2018]. Respir Med. doi:10.1016/j.rmed.2018.10.030