Total gonadotropin dose is not associated with live birth or embryo quality in fresh cycles at equivalent levels of ovarian response: a retrospective cohort study stratified by ovarian response

Introduction Whether the total gonadotropin (Gn) dose used for controlled ovarian stimulation is independently associated with live birth and embryo quality remains controversial, largely because the dose is individualized according to patient prognosis, which generates confounding by indication. We examined this association within homogeneous levels of ovarian response. Methods This single-center retrospective cohort study analyzed 3,711 fresh autologous embryo transfer cycles (female age ≤35 years, follicular-phase long protocol, January 2016 to June 2025); cycles with trigger-day progesterone ≥1.5 ng/mL were excluded because fresh transfer was canceled per center policy. Cycles were stratified by oocyte yield into poor (<10, n=1,346), normal (10–15, n=1,582), and high (>15, n=783) response. The exposure was the total Gn dose. The primary outcome was the fresh embryo transfer live birth rate; secondary outcomes included clinical pregnancy, early miscarriage, and morphological embryo-quality indicators. We modeled dose–response relationships with restricted cubic splines (RCS), compared dose quartiles within strata using sequentially adjusted logistic regression, controlled the false discovery rate (FDR) by the Benjamini–Hochberg method, and prespecified five sensitivity analyses. Results In the overall cohort, the total Gn dose showed neither a linear (P=0.871) nor a nonlinear (P=0.701) association with live birth. Compared with the lowest dose quartile (Q1), fully adjusted odds ratios for Q2, Q3, and Q4 were 1.18 (95% CI: 0.90–1.55), 1.08 (0.78–1.49), and 1.34 (0.92–1.95), and no stratum showed a significant association. In the normal-response stratum, the highest quartile was associated with reduced live birth in the crude model (OR=0.70, P=0.036), but the association attenuated and disappeared with sequential adjustment (aOR=1.08, P=0.757), which quantitatively illustrates how confounding by indication arises and how it can be corrected. No secondary outcome remained significant after FDR correction (minimum q=0.137), and all sensitivity analyses agreed with the main analysis. Conclusions Within the same level of ovarian response, the total Gn dose was not significantly associated with fresh embryo transfer live birth or morphological embryo quality. Within the dose range studied (700–8,025 IU; interquartile range 1,650–2,725 IU), escalating the Gn dose to improve embryo quality or live birth lacks evidential support, and moderate dose increases in patients with reduced ovarian sensitivity showed no signal of harm.