P-693 Retrospective evaluation of the effect of the implementation of the DIVINE dose-calculator in daily practice on ovarian hyperstimulation treatment risk and live birth rate

Abstract

Abstract Study question Does the implementation of the DIVINE dose-calculator in routine care decrease treatment risk due to ovarian hyperstimulation whilst maintaining treatment efficacy? Summary answer The implementation of the DIVINE dose-calculator significantly decreases treatment risks caused by ovarian hyperstimulation, without affecting treatment efficacy in terms of live birth. What is known already Meta-analyses show that tailoring the gonadotropin starting dose did not affect live birth rates, but reduced treatment risks. The ORT iOS IPD-MA study group used individual participant data to develop and validate a model for ovarian stimulation in IVF/ICSI: the DIVINE dose-calculator. This model scored adequate for assessing hyperstimulation treatment risk and poor for predicting live birth. It provides clinicians an overview of risk percentages per gonadotropin starting dose, which aids in selecting an appropriate, personalized gonadotropin starting dose. This model should be implemented in routine care to confirm the true clinical effect on treatment risk and live birth results. Study design, size, duration Retrospective cohort study using pseudonymized data in two fertility centers in The Netherlands. The dose-calculator was implemented on 15th February 2021. Data from 36 months before and 12 months after implementation were collected. The dose-calculator included female age, AMH and GnRH-type in order to select a starting dose between 100-225 IU. Baseline characteristics and fresh cycle treatment outcomes were extracted from the patient file, checked and cleaned. Missing data were imputed 50 times. Participants/materials, setting, methods Subfertile women younger than 38 years, starting their first IVF/ICSI treatment cycle with follitropin alfa or beta were included. Couples who did not plan to receive a fresh embryo transfer, women diagnosed with PCOS and couples undergoing limited insemination or PGT were excluded. Intention-to-treat and per protocol analyses were completed. Inverse propensity weighting was applied to balance confounders. Weighted regression was then performed using general linear models or ordered logistic regression models. Main results and the role of chance 493 women were included, 401 in the standard group and 92 in the dose-calculator group (per protocol analysis). There was no difference in live birth rate before and after implementation: 25.4% vs 25.0% respectively (OR: 1.01, 95% CI: 0.6 – 1.71). Ovarian hyperstimulation treatment risk clearly decreased after the implementation of the dose-calculator from 11.3% in the pre-implementation period to 4.3% in the post-implementation period (OR: 0.27, 95% CI: 0.08 – 0,91), of which OHSS incidence declined from 6.8% to 0% after implementation. Pre-implementation, women stimulated 1.0 day shorter as compared to the post-implementation period (MD, CI: 0.25 – 1.75), and used 200 IU gonadotropins less per cycle (MD, 95% CI: 48.68 – 352.11). The cancellation rate due to low ovarian response were equal pre- versus post-implementation (OR: 0.94, 95% CI: 0.45 – 1.98) as well as the cancellation rate due to high ovarian response (OR: 0.56, 95% CI: 0.11 – 2.84). Post-implementation 1.05 less oocytes were collected during follicle aspiration (MD, 95% CI: -2.51 – 0.41), which resulted in 0.41 less usable embryos (MD, 95% CI: -1.04 – 0.17). The usable embryo per oocyte was comparable (MD: 0.003, 95% CI: -0.061 – 0.067). The intention-to-treat analysis confirmed these results. Limitations, reasons for caution Results are preliminary and based on single center data. We expect to finalize the analyses including approximately 100 patients from the second center and a cost-effectiveness analysis before ESHRE 2023. Unfortunately, cumulative cycle results including cryopreserved embryo transfers and subsequent treatment cycles are not available. Wider implications of the findings The dose-calculator aids in optimizing ovarian stimulation, without compromising live birth prospects. It is an easy to use model, which substantially reduces risks caused by ovarian hyperstimulation. A free of charge mobile application will be developed to stimulate implementation. Future research should address cumulative cycle results and temporal/geographical validation. Trial registration number Not applicable