Concordance of Early and Late End Points for Community-acquired Bacterial Pneumonia Trials

Abstract

Abstract Background While there are ongoing regulatory convergence efforts, differences remain in primary end points recommended for community-acquired bacterial pneumonia (CABP) trials. The US Food and Drug Administration (FDA) recommends assessing CABP symptom resolution at an early time point (3–5 days after randomization). Other regulatory agencies recommend assessing overall clinical response at a later time point (5–10 days after therapy ends). Methods We analyzed participant-level data from 6 recent CABP trials submitted to the FDA (n = 4645 participants) to evaluate concordance between early and late end-point outcomes. We used multivariate logistic regression to identify factors associated with discordance. Results Early and late end-point outcomes were concordant for 85.6% of participants. The proportions of early end-point responders that ultimately failed and early end-point nonresponders that ultimately succeeded were similar (6.0% vs 8.4%, respectively). Early end-point response was highly predictive of late end-point success (positive predictive value, 92.9%). Multivariate logistic regression identified early end-point responders/late end-point failures as less likely to be obese and more likely to be infected with Chlamydophila pneumoniae or Staphylococcus aureus, have received antibacterial drug therapy prior to randomization, and have severe chest pain at baseline. The most common investigator-provided reasons for failure among early end-point responders/late end-point failures were receipt of nonstudy antibacterial drug therapy and loss to follow-up. Conclusions Early and late end-point outcomes were highly concordant. These data may be useful in the continuing efforts to reach international regulatory convergence on CABP clinical trial design recommendations.