We thank Macafee et al for their interest in our study. They emphasize important issues, with which we in general agree. In our manuscript we highlighted the limitations of the meta-analysis, including that about half of the original studies did not adjust for confounders, and because of the stigma associated with abortion, previous procedures may have been underreported in the case and control groups. Lack of adjustment for confounders is indeed an important limitation. Approximately 18 of the 36 included studies (references 24−27, 29−35, 37−39, 44−47) did adjust for some confounders, and most found an association with surgical termination and preterm birth, even after they adjusted for confounders. We also call for future research and for well-designed randomized trials. This call for more research in the effect of uterine evacuation on future pregnancies is probably our strongest recommendation. We acknowledge that medical and surgical abortion are incredibly safe procedures and seek to know the true impact that abortion may have on future pregnancies in prospective trials. We think that patient preference for the type of abortion experience should help guide the decision-making and that women should be given the choice between a surgical and a medical approach.

Our meta-analysis, based on the available literature, included more than 1,000,000 women; it suggests that previous surgical uterine evacuation is an independent risk factor for spontaneous preterm delivery and that women with history of surgical abortion have about twice the odds of preterm birth in the subsequent pregnancy compared with women without such a history. We are hopeful that our study will inspire prospective research to determine which method of termination results in the lowest risk of preterm birth in future pregnancies.

We agree that the odds ratios (ORs) provided by Zhou et al are much greater (OR 19.51, 95% confidence interval. 17.61−21.61) than the other studies. The percentage of women are reported in Table 3 of the original study (774/1775 vs 2377/62,350); however, the weight of the study on the pooled meta-analysis data is less than 4%. Moreover, the OR is much lower after confounders were adjusted (OR 1.89; 95% CI, 1.70−2.11; Figure 7 of the meta-analysis ).

Finally, we would like to emphasize that increasing the use of highly effective contraceptive methods in women who desire them should be an important solution to the persistent problem of health disparities of unplanned and teen pregnancies in the United States and would lead to improvements in women’s and children’s health, including a decrease in the incidence of preterm birth. Individual-level access barriers such as providers’ misconceptions and gaps in technical training as well as patients’ lack of awareness can be addressed directly by professional medical organizations, health care training programs, and others.