Results
We identified 57 eligible Cochrane systematic reviews and 1 IPD meta-analysis focussing on primary or secondary PTB prevention in pregnant women containing 373 potentially eligible RCTs (Figure 1). These Cochrane reviews were published between 2006 and 2019 (median year 2017). From the 373 eligible RCTs, we were able to include 350 RCTs for data extraction (Figure 1) coming from 56 reviews (See Table S1 for overview of SRs and RCTs included). These 350 RCTs were published between 1967 and 2019, with a gradual increase in publications over time untill 2015, followed by lower numbers of studies in the years 2016-2019 due to an expected delay in the uptake of RCTs in systematic reviews (Figure S1). The 350 RCTs randomised a total of 400,903 participants in all continents, with a higher cumulative number of trials performed in North America (n=92 trials, n=82,241 randomised women) and Europe (n=93 trials, n=56,653 women) and an increase of trials coming from Asia (n=71 trials, n=114,573 women) with a steep rise after the millennium (Figure S1).
There were 108 RCTs published before 2000 (including 122,742 randomised women), 104 RCTs between 2000-2009 (including 118,928 randomised women) and 138 RCTs between 2010-2019 (including 159,233 randomised women) showing no significant difference in randomised women per published study between the three time frames. General characteristics of the total RCTs sample and stratified for publications between the three time frames are shown in Table 1. Some potential increase over time was seen in the number of multicenter trials (93/212 [43.9%] before 2010 and 74/138 [53.6%] ≧2010; RR 1.22, 95%CI 0.98-1.52, with a stable median of 5 to 6 participating centers). A substantial increase was observed in the reporting of primary outcomes (139/212 [65.6%] before 2010 and 120/138 [87%] ≧2010; RR 1.33, 95%CI 1.18-1.49), power calculations (141/212 [66.5%] before 2010 and 109/138 [79.0%] ≧2010; RR 1.19, 95%CI 1.04-1.35) and use of a Data Safety Monitoring Board (43/212 [20.3%] before 2010 and 42/138 [30.4%] ≧2010; RR 1.50, 95%CI 1.04-2.17) (Table 1). Stratification by low/high impact factor and by publication year before 2010 and ≧2010 among high impact factor publications are shown in supplementary material (Table S2).
The median sample size in the 350 trials was 224 (IQR 106 to 803; Table 1). Median inclusion time was 2.3 years (range 1 month to 10.8 years). From the total of 400,903 randomised women, data was analysed for 363,417 (90.6%) women.