Patient and study design
We retrospectively analyzed the IUI-D cycle performed in the reproductive medicine center of Northwest Women’s And Children’s Hospital, China, from January 2014 to December 2016. The data were collected from the medical records of couples. The study protocol was approved by the Ethics Committee for the Clinical Application of Human Assisted Reproductive Technology of Northwest Women’s and Children’s Hospital. All methods were performed in accordance with the Measures for the Administration of Human Assisted Reproductive Technology.
In our center, IUI-D is performed for male factor infertility (azoospermia or severe oligospermia). Before treatment, all women underwent a detailed history and physical examination. All women had a salpingography to confirm at least single tubal patency. Both the natural cycle and stimulated cycle were included. The recorded parameters were mainly related to the woman, including age, duration of infertility, pregnancy history, number of attempts, tubal patency, Cos protocol, endometrium thickness and types, number and diameter of mature follicles on the day of insemination, I-O interval and donor semen quality.
IUI-D Procedure and I-O Interval Evaluation
Transvaginal ultrasound and serum LH and E2 tests were performed to monitor ovulation. For natural cycles, the ultrasound check started on the eighth day of the cycle. For ovulation stimulation cycles, the test started on the fifth day of the cycle. The COS cycle was stimulated by clomiphene citrate, gonadotropins or clomiphene citrate plus gonadotropins. The initial dose was 50 mg/day for clomiphene citrate (CC) (days 5–9) or 75 IU/day for gonadotropins and was modified according to the ovarian response. The ovulation trigger was given with an injection of 10,000 IU HCG when the follicle was ≥18 mm but the serum LH was <35 IU/L.
An ultrasound check was carried out between 8:00 a.m. and 9:00 a.m. every one to five days depending on the growth speed of the follicle. When the leading follicle was larger than 14 mm, the patients started the test for urinary LH; if the test was positive or the leading follicle was larger than 18 mm, serum estrogen (E2) and LH were quantified (except for those who refused blood tests). If the serum LH was ≥35 IU/L (defined as a spontaneous LH rise), E2 and LH were retested three hours later. If the serum LH was <35 IU/L, we continued to test serum LH and E2 the next day for natural cycles, but for stimulating cycles, we administered an HCG injection when the follicle was ≥18 mm. Meanwhile, we increased the frequency of the ultrasound test (Figure 1).
According to the change in E2 and LH, the HCG injection time and the ultrasound results, the insemination was arranged at 4:00 p.m. or 9:00 a.m. Thus, the I-O interval could be identified according to multiple ultrasound checks and the time of insemination (Table 1). In total, there were three types of I-O intervals: ≥ +19 h (Insemination preceded ovulation by more than 19 hours), -1 h ~ +19 h (Insemination preceded ovulation by fewer than 19 hours), and -19 h ~ -1 h (Ovulation preceded insemination by 1 hour to 19 hours; Table 1).
Processing of Donor Sperm
The donor sperm samples were supplied by Shaanxi Province Human Sperm Bank and guaranteed by the National Health and Family Planning Commission (NHFPC) of the People’s Republic of China. Sperm donors were screened strictly in accordance with NHFPC standards. Generally, eligible sperm had a minimum concentration of 60 × 106 per ml, progressive motility of 60%, and normal morphology of 4%. A proper match between patients and donors in racial and ethnic features, as well as blood type, were guaranteed. Before IUI, the frozen sperm sample was thawed fully and then centrifuged at 300g for 20 minutes, using a two-step discontinuous density gradient in a 45% and 90% Pure Sperm-100 platform. The semen samples were examined after thawing as well as after optimization according to the WHO standard . The volume of washed sperm sample used for insemination was 0.5 mL.
Intrauterine Insemination and Luteal Phase Support
Insemination was performed by one of our center’s gynecologists. The prepared sperm was gently inserted within 1 cm of the fundal extend of the uterine cavity using a soft catheter. The patient then rested for 10-15 minutes in a supine position. Daily treatment with 200 mg micronized progesterone or 20 mg dydrogesterone was used for 15 days after IUI-D. Some patients with a history of recurrent spontaneous abortion received an injection of 2000 IU human chorionic gonadotropin three times (every three days).
Diagnosis of pregnancy
The serum β-hCG concentration was quantified approximately 16 days after insemination. For women who had positive serum β-hCG, ultrasound confirmation of pregnancy was carried out two weeks later. A clinical pregnancy was defined after sonographic evidence of the gestational sac was observed. Live birth was defined as a live-born delivery at least 28 weeks after IUI-D.
The observations of any variables were not completely independent of each other when all IUI-D cycles were included. Consequently, classical statistical analyses with the assumption that samples are independent could not be used for the entire data from all cycles. In this case, generalized estimating equations (GEEs) that allow analysis of correlated observations  were used to evaluate the effects of these variables on IUI outcome, as in previous studies [8,14]. The outcome measure response variable was whether pregnancy existed per cycle. The outcome measure used as a response variable was whether pregnancy existed per cycle. For independent samples, two-group comparisons were performed by two-tailed Student’s t-test or Mann-Whitney U tests for continuous variables (expressed as the means ± standard deviations (SDs)) or by the χ2 test for categorical variables (expressed as frequencies and percentages). Moreover, stepwise multivariate logistic analysis was used to construct a predictive model for the clinical pregnancy rate and live birth rate in independent samples. The initial analysis included the variables shown in Tables 2-5. Variables were removed stepwise when the Wald test P-value for a given variable was over 0.05. Only statistically signiﬁcant variables were included in the ﬁnal model. All analyses were performed using the software Statistical Package Social Science (SPSS) 22.0. For all statistical tests, P < 0.05 defined statistical signiﬁcance.