Real World Effects of Medications for COPD - a UK Population-Based Non-Interventional Cohort Study With Validation Against Randomised Trial Results

Abstract

Background: COPD treatment is informed by RCT results, which may not apply to all patients. Non-interventional studies include groups excluded from trials, but their use to estimate treatment effectiveness is unproven.

Objectives: To use (1) individual trial data and (2) trial results to validate non-interventional methods for assessing COPD treatment effectiveness.

Methods: Individual patient data from the COPD TORCH trial were 1:1 matched to people with COPD in the UK Clinical Practice Research Datalink (CPRD) between January 2000 and January 2017. The results of analysing exacerbation rate in this trial-analogous non-interventional data were then compared to the TORCH trial results for Seretide (vs no seretide and vs salmeterol). Cohort methodology with propensity score techniques was used to adjust for potential confounding.

Results: There were 2773 people per group for the Seretide vs no Seretide analysis and 939 people for the Seretide vs salmeterol analysis. The propensity score matched exacerbations rate ratio (Seretide vs no seretide) was 1.25 (95% CI 1.14-1.37), compared to 0.75 (0.69-0.81) for the TORCH Seretide vs placebo analysis. For the Seretide vs salmeterol analysis, the rate ratio was 0.84 (0.73-0.97), compared to 0.88 (0.81-0.95) in TORCH.

Conclusions: While we were unable to replicate COPD placebo-controlled trial results, our results indicate that it is possible to replicate active comparator COPD trial results in non-interventional data. The methodology applied here will be used for (active comparator) studies of COPD treatment effects for groups underrepresented in trials.