Background: Clinical trials measure exacerbations of COPD inconsistently. Our objective was to determine if different methods for ascertaining and analyzing COPD exacerbations lead to biased estimates of treatment effects.
Methods: Information on the methods used to count, analyze and report COPD exacerbation rates was abstracted from clinical trials of long-acting bronchodilators or long-acting bronchodilator/inhaled steroid combination products published between 2000-2006. Data from the Optimal Therapy of COPD Trial was used to illustrate how different analytic approaches can affect the estimate of exacerbation rates and their confidence intervals.
Results: Twenty-two trials (17,156 patients) met the inclusion criteria and were reviewed. None of the trials adjudicated exacerbations or determined independence of events. Fourteen of 22 studies (64%) introduced selection bias by not analyzing outcome data for subjects who prematurely stopped study medications. Only 31% of trials used time-weighted analyses to calculate mean number of exacerbations/patient-year and only 15% accounted for between-subject variation. In the Optimal COPD trial the rate ratio for exacerbations/patient-year was 0.85 when all data were included in a time-weighted analysis but was overestimated as 0.79 when data for those who prematurely stopped study medications was excluded and was further overestimated as 0.46 when a time-weighted analysis was not conducted. P values ranged from 0.03 to 0.24 depending on how exacerbations were determined and analyzed.
Conclusions: Clinical trials have used widely different methods to define and analyze COPD exacerbations and this can lead to biased estimates of treatment effects. Future trials should strive to include blinded adjudication and assessment of the independence of exacerbation events, and trials should report time-weighted, intention-to-treat analyses with adjustments for between-subject variation in COPD exacerbations.
- statistical methods
- systematic review
Statistics from Altmetric.com
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online appendix