All randomized controlled trials (RCTs) using CHM with the basic function of tonifying the kidney in the IVF-ET treatment of PCOS were included. PCOS was diagnosed according to the European Society for Human Reproduction and Embryology (ESHRE) and American Society for Reproductive Medicine (ASRM)-sponsored PCOS Consensus Workshop criteria (the Rotterdam criteria) [13]. The therapeutic intervention was a combination of IVF-ET and CHM, including any dosage form. The medication time of using CHM was included before oocyte extraction and during the overall IVF-ET cycle. The control group was IVF-ET, blank or placebo. The results measurements included primary and secondary measurements. The primary measurements were the clinical pregnancy rate (clinical pregnancy was defined as if there is at least one gestational sac or fetal heart beat, as confirmed by vaginal ultrasound) and/or live birth rate (delivery of one or more living infants with more than 20 weeks of gestation). The secondary measurements were the abortion rate and ovarian hyperstimulation syndrome (OHSS) incidence. Non-RCTs, retrospective studies, observational studies, animal experiments, reviews and individual case reports were excluded.

We searched PubMed (1948-January 2021), EMBASE (1974-January 2021), the Cochrane Library (1900-January 2021), MEDLINE (1966-January 2021), China National Knowledge Infrastructure (CNKI, 1982-January 2021), Wanfang Data (1982-January 2021), and Chinese Science and Technology Journal Database (CSTJ, 1989-January 2021). There were no restrictions on the language or publication type in any search.

The following terms were used as Medical Subject Headings (MeSH) terms or free terms: (polycystic ovary syndrome; PCOS; polycystic ovary morphology; ovary polycystic disease; oligohypomenorrhea; amenorrhea; hirsutism; Stein-Leventhal syndrome; sclerocystic ovarian degeneration) AND (in vitro fertilization; fertilization in vitro; test-tube fertilization; test-tube baby; intracytoplasmic-sperm-injection; fresh embryo transfer; frozen embryo transfer; embryo implantation; egg collection; assisted reproductive technology) AND (Traditional Medicine; Traditional Chinese Medicine; Chinese Medicine Compound; Chinese Herbal Medicine).

Two reviewers independently (C.L. and M.D.) carefully examined the titles and summaries of the electronic searcher, and obtained the full manuscripts of all quotations that might meet the predefined inclusion criteria. We have obtained a total of 220 studies from computer retrieval. After screening the titles and abstracts, 184 articles ineligible for the eligibility criteria were excluded. Further detailed full-text review eliminated articles with incomplete outcome measures, nonuse of CHM for tonifying the kidney, or only ovulation stimulation. Any disagreement regarding the inclusion was resolved by a third reviewer (F.W.) through discussion or arbitration. Finally, 10 RCTs were selected for the meta-analysis [14, 15, 16, 17, 18, 19, 20, 21, 22, 23]. The process of exclusion and inclusion criteria is shown in Fig. 1.

Fig. 1.

The process and results of study selection for the systematic review and meta-analysis.

Two investigators (C. L. and M. D.) independently carefully examined the data from all qualified articles. After examining the manuscripts, those that met the inclusion criteria were included as qualified studies and were subjected to data abstraction. Relevant data collected included the year of publication, study country, sample size, study design, treatment course of intervention and outcome measurements.

The risk of bias was assessed by two independent reviewers (C.L. and M.D.) using the Cochrane risk of bias assessment tool recommended by the Cochrane handbook [24]. Random sequence generation, allocation of concealment, blinding, incomplete outcome data, selective reporting, and other biases were evaluated with three potential responses: low risk, high risk and unclear risk.

Any differences between reviewers were resolved by a third reviewer (F.W.) through discussion or arbitration.

Statistical analysis was performed with RevMan software, version 5.4.1 (The Nordic Cochrane Centre, The Cochrane Collaboration, London, UK, 2020). The continuous results were presented as the mean differences (MDs) with 95% confidence intervals (CIs), while dichotomous outcomes were presented as odds ratios (ORs) with their 95% CIs. Heterogeneity was investigated by the chi-square test and I${}^2$ statistic. When heterogeneity exists, we use a fixed-effects model (I${}^2$50%, p $>$ 0.05); otherwise, adopt a random-effects model. We conducted four subgroup analyses: dosage forms, medication time, type of embyro transfer and fresh embyro transfer time. In addition, we performed a sensitivity analysis to examine the effects of the exclusion of each study. We used forest plots for heterogeneity assessment of therapeutic effects and funnel plots to evaluate the publication bias.

This systematic review and meta-analysis included 10 RCTs [14, 15, 16, 17, 18, 19, 20, 21, 22, 23]. The basic characteristics of the included studies were recorded and are shown in Table 1 (Ref. [14, 15, 16, 17, 18, 19, 20, 21, 22, 23]). The specific drug composition and dosage of each study were shown in the Table 2 (Ref. [14, 15, 16, 17, 18, 19, 20, 21, 22, 23]). All of the included studies were published between 2008 and 2020 in China. A total of 663 PCOS patients were assessed: 336 patients received CHM treatment, while 327 patients received only IVF-ET treatment without CHM. All of the 10 included studies only described the clinical pregnancy rate, but neither the live birth rate nor pregnancy complications.

There were three dosage forms in the treatment groups: decoction, granules and capsules. The medication time of using CHM could be divided into two types: before oocyte extraction and during the overall process of the IVF-ET cycle. Only one study [20] used placebo as the control group, and the rest of the studies used blank controls. In terms of the selection of downregulating protocols for IVF-ET, one study [23] used the GnRH antagonist long protocol, and the rest of the studies used the GnRH agonist long protocol.

Totally ten articles were included in this systematic review and meta-analysis, which were all single-center studies. We measured the risk of bias in each of the included studies. Randomization was applied in all of trials, among which 8 studies [14, 15, 16, 17, 19, 20, 22, 23] described the specific randomization methods. None of the articles used allocation concealment, two of which [18, 21] based allocation on the patient’s visit date and outpatient number as a random scheme, which might have led to a high risk for selection bias. Only one of the included studies [20] was conducted in a single-blind manner of the participants. There was no significant risk of bias in the selective reporting and other biases. The summary of the assessment of risk of bias is shown in Figs. 2,3.

Fig. 2.

Risk of bias figure.

Fig. 3.

Summary of risk of bias.

For the clinical pregnancy rate, data can be obtained from all 10 included studies (n = 663). There was no significant statistical heterogeneity among the studies (test for heterogeneity, p = 0.99, I${}^2$ = 0). Using the fixed-effects model, the pooled data showed significant efficacy of CHM used during IVF-ET compared with the control group (OR = 2.41, 95% CI: 1.73–3.35, p 0.01; Fig. 4). Furthermore, the publication bias was analyzed via funnel plot (Fig. 5), which revealed no significant evidence of symmetry.

Fig. 4.

Forest plot of the clinical pregnancy rate.

Fig. 5.

Funnel plot of the effect sizes for CHM in IVF-ET.

Concerning pregnancy complications, only abortion and OHSS were reported in the ten included studies. Firstly, for the abortion rate, data can be obtained from 4 of the 10 included studies (n = 125). There was no significant statistical heterogeneity in these studies (test for heterogeneity, p = 0.92, I${}^2$ = 0). Using a fixed-effects model, the pooled data showed that CHM did not significantly reduce the occurrence of abortion compared with the control group (OR = 0.64, 95% CI: 0.23–1.81, p = 0.40; Fig. 6). Secondly, for the OHSS incidence, data can be obtained from 9 of 10 included studies (n = 599). There was no significant statistical heterogeneity among the studies (test for heterogeneity, p = 0.92, I${}^2$ = 0). Using the fixed-effects model, the pooled data showed that CHM significantly reduced the development of OHSS compared with the control group (OR = 0.31, 95% CI: 0.18–0.55, p 0.01; Fig. 7).

Fig. 6.

Forest plot of the abortion rate.

Fig. 7.

Forest plot of the OHSS incidence.

Since the literature concerning the abortion rate and the OHSS incidence was inadequate and the clinical pregnancy rate was the primary measurement of this study, we conducted a subgroup analysis on the clinical pregnancy rate. To further explore whether the dosage forms of CHM will affect the clinical pregnancy rate, we conducted a subgroup analysis of different dosage forms of CHM. Since only one article used capsule forms, so it was not considered for further subgroup analysis. We compared decoction with granules of CHM during IVF-ET and found no significant difference between the two dosage forms (decoction: OR = 2.36, 95% CI: 1.56–3.57, p 0.01; granules: OR = 2.89, 95% CI: 1.56–5.34, p 0.01, Fig. 8).

Fig. 8.

Subgroup analysis of the different dosage forms of CHM.

We conducted another subgroup analysis of medication time for using CHM during IVF-ET. We found that using CHM before oocyte extraction could significantly improve the clinical pregnancy rate; however, the effects of taking CHM throughout the entire process of IVF-ET had no incremental improvement (before oocyte extraction: OR = 2.36, 95% CI: 1.56–3.57, p 0.01; overall the process of IVF-ET: OR = 1.81, 95% CI: 0.86–3.80, p = 0.12, Fig. 9).

Fig. 9.

Subgroup analysis of the CHM medication time.

About the type of embryo transfer, frozen embryos were selected and routine IVF followed by full embryo freezing and embryo transfer on the third or fifth day of the artificial cycle in two of the included articles [22, 23]. The rest of the articles chose fresh embryo transfer. We conducted one subgroup analysis concerning the selection of embryo transfer methods, and another subgroup analysis based on transfer time of fresh embryo cycles. We found a higher rate of clinical pregnancy rate in the fresh embryo transfer compared with frozen embryos (frozen embryo: OR = 1.81, 95% CI: 0.86–3.80, p = 0.12; fresh embryo: OR = 2.67, 95% CI: 1.85–3.87, p 0.01, Fig. 10). Meanwhile, there was no significant difference on clinical pregnancy rates during fresh embryo transplantation time, either 2–3 days or 18–22 hours (2–3 days: OR = 2.73, 95% CI: 1.59–4.69, p 0.01; 18–22 hours: OR = 2.67, 95% CI: 1.31–5.43, p 0.01, Fig. 11).

Fig. 10.

Subgroup analysis of the type of embryo transfer.

Fig. 11.

Subgroup analysis of fresh embryo transfer time.