Strengths and Limitations
The strengths of our study were including the large number of trials and their relatively sample sizes; comprehensive literature covered English and Chinese; involving all common acupuncture types (TA, EA, TEAS, WA and LA); more detailed subgroup analyses and meta regression to explore the heterogeneous; adherence to the intention to treat principle; as well as robustness of the results to sensitivity analyses. Compared with the previous and latest reviews published by Coyle ME et al (8 trials, English-language literature searched up to 3 April 2019) (22) and Xie ZY et al. (27 trials, literature searched up to December 2018) (15), we included the largest sample size and most comprehensive trials. Twenty trials were added in our reviews; eleven had positive results (6, 35, 39, 40, 46, 47, 49, 51, 52, 57, 77), and nine had negative results (30, 42-45, 53-56). In addition, we defined both CPR and LBR as primary outcomes, which were stricter than CPR alone used in almost all previous reviews. Compared to CPR, LBR is more likely to reflect the endpoint of IVF therapy and the most concerned outcome for patients.
A limitation of this review was the significant heterogeneous among trials. The acupuncture prescription and control type might be the important contributions to the heterogeneous. Although most acupuncture prescriptions were generally standardized based on the original protocol of Paulus et al (57), nearly every study would make appropriate adjustments to some extent (20). In addition, the type of control especially for the sham or placebo should be noted. Acupuncture point location and needle insertion or not are the main factors causing variance when we establish a sham or placebo acupuncture. In traditional Chinese medicine theory, acupoint is not a point and has its own domain. Even though a sham needle is applied away from the acupoint, it might have the similar treatment effect as the actual acupuncture. Even though noninvasive placebo acupuncture such as Streitberger acupuncture may also have certain effects because it stimulates the skin. Those have been illustrated by our results which suggested CPR and LBR in TA group were significant compared with no treatment, but no difference compared with sham or placebo acupuncture. Hence, placebo acupuncture might not always be an inert control and it is greatly difficult to establish an ideal placebo acupuncture.