Postplacental insertion of the levonorgestrel intrauterine device after cesarean delivery vs. delayed insertion: a randomized controlled trial

Objective

This trial was designed to compare levonorgestrel intrauterine device (LNG-IUD) use at 1 year after delivery between women randomized to postplacental insertion at the time of cesarean delivery and delayed insertion 4–8 weeks after delivery.

Study design

This randomized controlled trial was conducted at two urban medical centers. Eligible pregnant women with planned cesarean deliveries were randomized to immediate postplacental insertion during cesarean or delayed insertion after 4–8 weeks. We used intention-to-treat analysis for the primary outcome of LNG-IUD use 12 months after delivery.

Results

Forty-two women were randomized, 20 into the postplacental group and 22 in the delayed group. Although confirmed use of the LNG-IUD 12 months after delivery was higher in the postplacental group (60.0% vs. 40.9%, p=.35), this difference was not statistically significance. Expulsion was significantly more common in the postplacental group (20.0% vs. 0%, p=.04). There were significant differences between the two sites in baseline population characteristics, follow-up and expulsion. The trial did not answer the intended question as it was halted early due to slow enrollment.

Conclusions

Our results show higher expulsion after postplacental insertion compared to delayed insertion but suggest similar IUD use at 12 months. Moreover, it provides valuable lessons regarding a randomized controlled trial of postplacental LNG-IUD placement due to the challenges of estimating effect size and the nature of the population who might benefit from immediate insertion.

Implications

Postplacental insertion of an IUD may improve use of highly effective contraception during the postpartum period. While our results suggest higher expulsion after postplacental insertion compared to delayed insertion and similar IUD use at 12 months, our trial was insufficient to definitively test our hypothesis.