Answers are grounded in this paper and cite their sources.
Purpose To balance the benefits and risks of different controlled ovarian stimulation (COS) protocols, we evaluated the individualized treatment effect (ITE) of the GnRH antagonist protocol versus the follicular phase depot GnRH-a long protocol, corresponding to the cumulative live birth rate (CLBR). Methods A retrospective cohort study included 17,571 IVF/ICSI cycles from January 2016 to December 2022. Propensity score matching was performed and a causal forest model was trained to estimate the ITE. SHAP analysis quantified feature contributions, and generalized additive models were used to visualize feature effects. Results In this exploratory analysis, the depot GnRH-a long protocol demonstrated superior treatment effects compared to the antagonist protocol (ITE range: −0.199 to −0.034). However, higher ITE values correlated with increased OHSS incidence in agonist cycles. An exploratory counterfactual analysis suggested that reallocating patients with ITE > −0.115 to the antagonist protocol would reduce CLBR by 6.08 percentage points (64.30% vs. 58.23%) but decrease OHSS rate by 0.31 percentage points (1.61% vs. 1.30%). In the exploratory SHAP analysis, basal estradiol (E2) appeared as the strongest modifier: the antagonist protocol tended to be favored in patients with low basal E2, high BMI, older age, or extreme AMH levels. Conclusions Although confirmation in prospective randomized studies is required, the follicular phase depot GnRH-a long protocol may achieve higher CLBR than the GnRH antagonist protocol and was accompanied by elevated OHSS risk. In this exploratory analysis, the ITE estimates suggested a potential trade-off between CLBR and OHSS risk. Basal E2 may be considered as an additional marker when selecting a controlled ovarian stimulation protocol.
No takes yet. Share an insight, caveat, or question.
Ma et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: