Measuring outcomes in fertility trials: can we rely on clinical pregnancy rates?

Abstract

<h4>Objective</h4>To assess whether the estimates of treatment effect in randomized clinical trials (RCTs) in reproductive medicine differ when either clinical pregnancy or live birth is used as the outcome measure.<h4>Design</h4>Metaanalysis.<h4>Setting</h4>We analyzed RCTs in reproductive medicine found in systematic reviews published in the Cochrane Library that reported on both clinical pregnancy and live birth.<h4>Patient(s)</h4>Subfertile couples.<h4>Intervention(s)</h4>For each individual RCT, data on clinical pregnancy and live birth were extracted.<h4>Main outcome measure(s)</h4>We compared the outcome of each study by calculating a kappa-statistic (statistically significant treatment effective or not) and by comparing the odds ratio by calculating the ratio of the odds ratios (ROR).<h4>Result(s)</h4>We found 67 systematic reviews, of which 42 reported on pregnancy and live birth. These 42 reviews included 654 RCTs, of which 143 (22%) reported both on pregnancy and live birth. The pregnancy loss rates in the treatment and control groups were comparable. Of the 143 RCTs, the conclusion based on pregnancy rate and live birth rate was comparable (kappa value of 0.81; 95% confidence interval [CI], 0.68-0.94). The odds ratios estimating treatment effect from pregnancy and live birth were also comparable (ROR, 1.01, 95% CI 0.9 to 1.12).<h4>Conclusion(s)</h4>Only a minority of randomized clinical trials in reproductive medicine report on live birth. Conclusions on the effectiveness of a treatment based on either clinical pregnancy or live birth as endpoints are comparable.