Multicenter randomized, controlled trial of a preterm birth prevention program. Collaborative Group on Preterm Birth Prevention.

OBJECTIVE

The results of a randomized, controlled trial of a preterm birth prevention program at five centers with primarily low-income populations are presented.

STUDY DESIGN

Pregnant women at high risk for preterm labor were independently randomized into intervention and control groups at each of five centers (2395 women). Specially trained staff instructed women in the intervention group to recognize early signs of preterm labor and to notify the staff should any sign of preterm labor occur, and women came in weekly for pelvic examinations after 20 to 24 weeks' gestation. Because the intervention had very different and contrary effects on preterm birth rates across these sites, the analysis focused on patient risks and on measures of the process of care as possible explanations for the differences in outcomes.

RESULTS

The observed spontaneous preterm birth rates, when averaged for all sites, were not lower in the intervention group than in the control groups (16.1% vs 15.4% for < 37 completed weeks' gestation, 11.9% vs 10.9% for < 36 completed weeks). There was substantial heterogeneity of program effects between centers (p < 0.01 for homogeneity test statistic). The differences in intervention effects between centers were explainable only in part by patient risk characteristics (p < 0.10 for homogeneity test statistic). The only intermediate measure of the process of care that tended to differentiate among sites with positive intervention effects was the rate of hospital admission for preterm labor. Sites with elevated admission rates in the intervention group versus the control group tended to have negative intervention effects on outcomes.

CONCLUSION

Because the preterm birth prevention program did not show a reliable benefit and because the reasons for varied outcomes are not understood, the use of the program cannot be recommended for predominantly low-income populations.