This retrospective cohort study was conducted after approval by the ethics committee of the First People’s Hospital of Yunnan Province (No. KHLL2020-KY013). Patients undergoing their first cycle with controlled ovarian stimulation using a GnRH analog and fresh embryo transfer (ET) at the university-affiliated hospital from January 2017 to May 2020 were included. The exclusion criteria were as follows: 1) the presence of polycystic ovary syndrome (PCOS), luteinized unruptured follicle syndrome, and other endocrinology disorders; 2) the use of oral contraceptives in the last 3 months; 3) a self-reported history of a family genetic disorder or an abnormal chromosomal karyotype; and 4) incomplete medical records. The clinical information was deidentified. Simple randomization was used to select the agonist or antagonist regimen. However, for patients with normal ovarian reserve and the desire to have more oocytes retrieved, the agonist tended to be used. For patients with a potential high ovarian response or low anti-Müllerian hormone (AMH) level (less than 2 ng/mL), antagonists tended to be used. Finally, pituitary suppression was initiated after the patient learned about the difference between agonists and antagonists and signed the informed consent form.

The agonist pituitary suppression regimen was initiated at the mid-luteal phase. A total of 1.25 mg of the long-acting triptorelin acetate (Diphereline, Ipsen Pharma Biotech, Signes, France; Decapeptyl, Ferring GmbH, Kiel, Germany) was injected intramuscularly (IM). After approximately 14 days, when the estradiol (E2) level was < 50 pg/mL, the endometrial thickness was < 5 mm, and no ovarian cysts were noted, COS was initiated by an FSH (rFSH, Gonal-F, Merck-Serono, Aubonne, Switzerland) subcutaneous (SC) injection once a day. Ultrasound and serum hormone examinations were conducted from the 6th day of COS. When there were two dominant follicles with a diameter of > 18 mm, the serum LH, E2, and progesterone levels and the endometrial thickness (EMT) were recorded. Then, 5,000 IU of exogenous human chorionic gonadotropin (hCG) was injected at 10 PM to trigger the ovulation process.

In the flexible antagonist regimen, exogenous FSH (rFSH, Gonal-F, Merck-Serono, Aubonne, Switzerland) was SC injected daily from the 2nd day of menses to start COS. From the 4th day, when the E2 level was > 300 pg/mL, the diameter of the dominant follicle was > 14 mm or the LH level was > 10 mIU/mL, a 0.25 mg SC injection of cetrorelix acetate (Cetrotide, Merck Europe B.V., Idron, France) or ganirelix acetate (Orgalutran, Merck Sharp & Dohme B. V., Ravensburg, Germany) was injected daily until the diameters of the two dominant follicles were > 18 mm. Subsequently, the ovulation trigger was administered by injecting exogenous hCG.

Thirty-six hours after the ovulation trigger was administered, transvaginal ultrasound-guided puncture and oocyte retrieval were performed. According to male sperm motile measurements, conventional IVF or ICSI was conducted. The statuses of the zygote and embryo were monitored and recorded. The presence of two pronuclei (2PN) on the first day was considered normal fertilization. The number of 2PN divided by the number of oocytes retrieved was defined as the 2PN rate. On the 3rd day, the original 2PN embryos, with 6–8 blastomeres and cell debris < 20%, were considered the optimal choice for ET. β-hCG was tested on the 14th day after ET. Luteal support was continued to the 8th week if the β-hCG test was positive.

Baseline characteristics, including age, body mass index (BMI), AMH level, and antral follicle count (AFC), were recorded. Primary or secondary infertility, causes of female infertility, history of parturition or miscarriage, and methods of fertilization (conventional IVF versus ICSI) were recorded. The LHOTD was collected on the day of the ovulation trigger, and all serum hormones were measured by a UniCel DxI 800 Access Immunoassay System (Beckman Coulter, CA, USA). Medical records were also reviewed to document the levels of estradiol and progesterone, as well as the endometrial thickness, on the ovulation trigger day.

After ET, follow-up visits were cancelled if the β-hCG test was negative. If the β-hCG test was positive, a clinical pregnancy test was conducted at the 5th–6th week. The detection of a viable sac(s) was defined as a clinical pregnancy. An extrauterine sac was defined as an ectopic pregnancy. Pregnancy loss during the first trimester was defined as early pregnancy loss. If pregnancy continued past the 12th week, it was defined as an ongoing pregnancy. The primary outcomes of the study were the clinical pregnancy and live birth rates.

Statistical analyses were conducted with SPSS (version 26.0, Armonk, NY, USA). A P value of < 0.05 via a two-tailed test was considered statistically significant. The cycles were assigned into tertile groups based on the LHOTD in each agonist and antagonist regimen (T1, T2, and T3). To compare the three LHOTD groups, normally distributed continuous data, such as age, BMI and the number of oocytes retrieved, are expressed as the mean ± standard deviation (SD) and were compared by one-way ANOVA. Nonnormally distributed continuous data, such as the AMH level, AFC, indicators on ovulation trigger day and 2PN rate, are expressed as the median (25th–75th percentiles) and were compared by the Kruskal–Wallis test. Categorical data, such as gravidity and parity, medical history, fertilization method, the number of embryos transferred and clinical outcomes, are demonstrated as counts (percentage) and were analyzed by the chi-square (χ2) test. The multiple pairwise comparison P value was adjusted by the Bonferroni method. A stepwise progressive multivariate regression model [11] was introduced to assess the effect of the LHOTD on the clinical pregnancy and live birth rates. Finally, a total of 3 models were developed to account for the important information of each IVF-ET treatment stage as comprehensively as possible to eliminate confounding factors. Model 1 included age, BMI, the AMH level, and the AFC. In Model 2, primary infertility, the cause of female infertility, and history of parturition or miscarriage were included. In Model 3, we added the ovulation trigger day indicators, including progesterone, E2, EMT, the number of oocytes retrieved, fertilization method, and the number of embryos transferred, on the basis of Model 2. Mantel‒Haenszel stratification analysis [12] was used to demonstrate that the LHOTD of the antagonist regimen was not correlated with the clinical outcome.