Computer modeling informs study design: vaginal estrogen to prevent mesh erosion after different routes of prolapse surgery.
INTRODUCTION AND HYPOTHESIS: Many clinicians use perioperative vaginal estrogen therapy (estradiol, E(2)) to diminish the risk of mesh erosion after prolapse surgery, though supporting evidence is limited. We assessed the feasibility of a factorial randomized trial comparing mesh erosion rates after vaginal mesh prolapse surgery (VM) versus minimally invasive sacral colpopexy (MISC), with or without adjunct vaginal estrogen therapy. METHODS: A Markov state transition model simulated the probability of 2-year outcomes of visceral injury, mesh erosion, and reoperation after four possible prolapse therapies: VM or MISC, each with or without estrogen therapy (E(2)). We used pooled estimates from a systematic review to generate probability distributions for the following outcomes after each procedure: visceral injury, postoperative mesh erosion, and reoperation for either recurrent prolapse or mesh erosion. Assuming different assumptions for E(2) efficacies (50 and 75 % reduction in erosion rates), Monte Carlo simulations estimated outcomes rates, which were then used to generate sample size estimates for a four-arm factorial trial. RESULTS: While E(2) reduced the risk of mesh erosion for both VM and MISC, absolute reduction was small. Assuming 75 % efficacy, E(2) decreased the risk of mesh erosion for VM from 7.8 to 2.0 % and for MISC from 2.0 to 0.5 %. Total sample sizes ranged from 448 to 1,620, depending on power and E(2) efficacy. CONCLUSIONS: The required sample size for a trial to determine which therapy results in the lowest erosion rates would be prohibitively large. Because this remains an important clinical issue, further study design strategies could include composite outcomes, cost-effectiveness, or value of information analysis.
Weidner, AC; Wu, JM; Kawasaki, A; Myers, ER
Volume / Issue
Start / End Page
Electronic International Standard Serial Number (EISSN)
Digital Object Identifier (DOI)