Progesterone in Women with Bleeding in Early Pregnancy: A Bayesian re-analysis

A randomized trial assessed the effect of progesterone in preventing pregnancy loss. The incidence of live births after at least 34 weeks of gestation (primary outcome) was 74.7% (1513 of 2025 women) in the progesterone group, and 72.5% (1459 of 2013 women) in the placebo group. Using standard frequentist methods, the authors concluded that the treatment effect was not statistically significant, p=.08. I performed a Bayesian re-analysis to assess the probability that progesterone was effective at varying effect size thresholds. The probability that progesterone was on average effective ranges from very high to negligible, all at low effect size thresholds.