Male infertility has become a global health issue. It has been reported that semen counts have declined over the past 50 years . According to the results of Agarwal et al. , at least 30 million men worldwide are infertile, with the highest rates seen in Africa and Eastern Europe. IUI has long been considered as the first-line treatment for male infertility. The published pregnancy rates with IUI in NC cycles vary from 0-20.5%, while in stimulated cycles this rate varies from 3.9-13.6% per cycle  . In our study, the pregnancy and live birth rates in NC cycles were 12.2% and 11.1% per cycle, respectively, while the corresponding values for COS cycles were 13.1% and 11.9%, respectively. Despite the higher rate seen in COS cycles, no statistically significant differences were observed. To assess whether the semen quality influenced the outcome of either treatment, we divided the study population into three groups: couples with TPSMC < 5 × 106; couples with TPSMC 5-10 × 106; and couples with TPSMC ≥ 5 × 106. Similar results were observed in all subgroups.
It has also been suggested that COS could overcome subtle ovulation disorders that cannot be detected by routine testing [17, 20]; to some extent, multifollicular growth is associated with increased pregnancy rates, and IUI with COS is always preferred.
A 1999 randomised trial  showed that couples treated with superovulation had higher pregnancy rates than did couples without stimulation after four IUI cycles (33% vs. 18%). However, it should be noted that the stimulation in this study was intense (150 IU follicle-stimulating hormone as the initial dose), and the multiple pregnancy rates were high, which is unacceptable. In the present study, couples were advised to cancel the cycle if more than three dominant follicles ≥16 mm were present.
Another retrospective cohort study by Huang et al.  suggested that, in an IUI programme for unexplained or mild male factor infertility, ovarian stimulation with LE may significantly increase live birth rates while controlling multiple pregnancy rates. In that study, compared with NC (6.2%), live birth rates were significantly higher in IUI cycles stimulated with CC (8.9%), LE (9.4%), and gonadotropins (9.5%). The multiple pregnancy rates in the NC, CC, LE, and gonadotropin groups were 0.7%, 4.6%, 1.3%, and 3.9%, respectively. However, as proposed by those authors, female age, duration of infertility, and first cycle proportion were significantly different among the four groups, which was a limitation of that study.
An overall decline in fertility with advancing age has been well documented . Dovey et al.  also reported a significant reduction in the efficacy of CC/IUI with advancing female age. In a prospective, randomised, parallel study , the authors found that female age was the only factor that influenced the chance of success in couples with idiopathic subfertility or male subfertility. Therefore, the lower pregnancy and live birth rates of the NC cycle reported in the study of Huang et al. might be related to older female age.
In addition, Liu et al. concluded that ovarian stimulation in ovulatory women undergoing IUI appeared to have a limited role, with the exception of LE and HMG. In their study, when all ovarian stimulation protocols were combined, no differences in pregnancy or live birth rates were found in ovulatory women. However, further analysis of the impact of the different ovarian stimulation protocols showed that the live birth and pregnancy rates of ovulatory women undergoing IUI treatment stimulated with LE combined with HMG were significantly higher than those in an NC (12.2% vs. 7.6%, and 16.8% vs. 9.3%, respectively). However, in that retrospective study, four types of infertility aetiologies (endometriosis, tubo-peritoneum, male factor, and unexplained infertility) were included. It should be noted that the efficacy of ovarian stimulation might differ from infertility aetiologies. As it has been reported the addition of active ovulation management (CC+HCG) increased the pregnancy rate per IUI cycle in unexplained infertility; however, in couples with male infertility, this protocol may not be beneficial . In our study, only the couples with male infertility were included, and we believe the results might be more generalizable.
An increasing number of studies [4, 15-17] reported that IUI with or without COS for male infertility has no significant differences in the pregnancy rate, which is consistent with the results of two recent Cochrane reviews [10, 23]. Furthermore, in our study, to eliminate the potential influence of repeated cycle data and different diagnostic criteria (WHO 4th and 5th versions) on the results, we only included the first cycle and performed a subgroup analysis between cycles before 2014 and from 2014 onwards, which revealed similar results. In addition, the cumulative live birth rates after three IUI cycles with a similar protocol were 23% for NC-IUI and 25% for COS-IUI (P=0.837, data not shown). Therefore, we advocate that for male infertility, IUI with COS did not significantly increase the pregnancy and live birth rates, regardless of the diagnostic criteria for male infertility.
Two patients who received IUI with COS had multiple pregnancies (one twin and one heterotopic pregnancy, both with HMG protocol, and in both two follicles were generated) in our study. A trend toward higher multiple pregnancy rates was observed in the COS-IUI group, when compared to that in the NC-IUI group, but the difference was not statistically significant (P=0.091). However, when we further compared the multiple pregnancy rates between the NC-IUI group and ≥ 2 follicles group, a significant difference was found (0 vs. 16.7%, P=0.033). Few studies have focused on reporting multiple pregnancies of IUI with or without COS in male infertility.
A meta-analysis  included 14 studies reporting on 11 599 IUI cycles in patients with various types of subfertility and showed that the pregnancy rates increased from 8.4% to 15% with COS when multifollicular growth was achieved as compared to monofollicular stimulation, and the multiple pregnancy rates increased from 3.7% to 17% per conceived cycle. Compared with monofollicular growth, when stimulating two, three, and four follicles, the pregnancy rates increased by 5, 8, and 8%, while the risk of multiple pregnancies increased by 6, 14, and 10%, respectively. Those authors advocated that IUI with COS should not aim for more than two stimulated follicles; one stimulated follicle should be the goal if safety is the primary concern, whereas two follicles may be acceptable after careful patient counselling. Recently, data from another systematic review and meta-analysis  did not support the routine use of gonadotropins for ovarian stimulation in IUI for women with unexplained infertility, as the increased pregnancy rate is associated with a larger increased risk of multiple gestations.
Additionally, Evans et al.  also reported that the mean clinical pregnancy rate per IUI ranged from 13.0% with one mature follicle to 19.6% with five mature follicles. The pregnancy rate increased modestly with the presence of each additional mature follicle (adjusted odds ratio: two follicles vs. one: 1.3, 95% CI 1.2–1.4; three follicles vs. one: 1.4, 95% CI 1.3–1.5; four follicles vs. one: 1.5, 95% CI 1.4–1.7; and five follicles vs. one: 1.6, 95% CI 1.4–1.9). However, a greater significantly increased risk of multiple pregnancies was seen with the presence of each additional mature follicle (adjusted odds ratio: two follicles vs. one: 3.5, 95% CI 2.7–4.4; three follicles vs. one: 5.6, 95% CI 4.4–7.1; four follicles vs. one: 7.2, 95% CI 5.6–9.4, and five follicles vs. one: 8.6, 95% CI 6.2–11.8). There was only a 1.9% increase in singleton pregnancies per IUI regardless of mature follicle number (from 12.4% with one follicle to 14.3% with five follicles).
Compared with singleton pregnancies, multiple pregnancies are associated with many pregnancy complications, which is unacceptable . Therefore, the aim of fertility treatment is shifting from focusing on the pregnancy rate to the birth of healthy singletons .
Our study had several limitations. First, the retrospective nature of our study might have led to potential inherent bias. In addition, we only included cycles in which IUI was performed. Cycles with more than three dominant follicles ≥16 mm were cancelled in all groups and were not included in this study. The data from these cycles may favour IUI with COS. Finally, the time span of this study was 10 years, which means the diagnostic criteria about male infertility changed, since the WHO manual (5th version) is widely used as a source of standard methodology for semen collection, analysis, and preparation. However, other elements of the IUI practice did not change. We also performed an analysis stratified for different criteria to eliminate the potential influence on the results as much as possible.