We thank Drs Goya and Cabero for their comments. Our trial involved 1180 unselected twin pregnancies randomized at 20 ⁺⁰ to 24 ⁺⁶ weeks’ gestation to cervical pessary vs expectant management; there was no significant difference between the groups in spontaneous birth < 34 weeks (13.6% vs 12.9%), which was the primary outcome of the study. In both groups there was an inverse correlation between the cervical length at randomization and the rate of spontaneous birth at < 34 weeks, which was not significantly different between the 2 groups.

The pessary was removed at < 34 weeks in 22.3% (131 of 588) of pregnancies, including 100 before elective delivery for medical indications, such as severe preeclampsia, preterm labor or preterm prelabor rupture of membranes, and 31 for patient request.

Subsequently, there was birth at < 34 weeks in 91% of the first group and 23% of the patient request group; in the latter group, the rate of preterm birth was not significantly higher than that in the total group treated with pessary placement. Consequently, in most cases the pessary was removed because the patients were about to deliver and such delivery could not have been prevented had the pessary been left in situ.

In the paper we acknowledge that a potential limitation of our study is that many research team doctors were involved in the insertion of the pessary and they did not receive supervised training in doing so. It is therefore not possible to be certain that there was appropriate insertion in all cases. However, the same was also true for the study of Liem et al, and therefore, this cannot be the explanation for the differences in the results of the 2 trials.

Serial measurements of cervical length and use of tocolytics and steroids would not have influenced the results of the trial because women in the control group received the same obstetrical care as those in the pessary group. Progesterone was administered to 2 patients in the control group; had this not been done and both patients delivered preterm, this would not have affected the conclusions of the study.

Randomization resulted in 2 groups with similar maternal and pregnancy characteristics.