Adaptive propensity score procedure improves matching in prospective observational trials

• Main Authors: Dorothea Weber, Lorenz Uhlmann, Silvia Schönenberger, Meinhard Kieser
• Format: Article
• Language: English
• Published: BMC 2019-07-01
• Series: BMC Medical Research Methodology
• Subjects: Adaptive design; Clinical Trials; Sample size recalculation; Matched cohort; Prospective matching
• Online Access: http://link.springer.com/article/10.1186/s12874-019-0763-3

Abstract Background Randomized controlled trials are the gold-standard for clinical trials. However, randomization is not always feasible. In this article we propose a prospective and adaptive matched case-control trial design assuming that a control group already exists. Methods We propose and discuss an interim analysis step to estimate the matching rate using a resampling step followed by a sample size recalculation. The sample size recalculation is based on the observed mean resampling matching rate. We applied our approach in a simulation study and to a real data set to evaluate the characteristics of the proposed design and to compare the results to a naive approach. Results The proposed design achieves at least 10% higher matching rate than the naive approach at final analysis, thus providing a better estimation of the true matching rate. A good choice for the interim analysis seems to be a fraction of around 12 $\frac {1}{2}$ to 23 $\frac {2}{3}$ of the control patients. Conclusion The proposed resampling step in a prospective matched case-control trial design leads to an improved estimate of the final matching rate and, thus, to a gain in power of the approach due to sensible sample size recalculation.