Subgroup analyses
Since there was statistical heterogeneity for IVF outcomes across studies (CPR: I2 =47.4, p <0.001; LBR:I2 =62.9, p <0.001), subgroup analyses and random-effects meta-regression were utilized. In subgroups classified by treatment type, studies used TA was found to be statistical heterogeneity (CPR: I2 =61.5, p <0.001; LBR: I2 =68.3, p =0.001) (Table 1 ). Therefore, the meta-regression analyses were restricted to TA methods. With regard to TA and TEAS on CPR, control type (no acupuncture, sham or placebo) was found to be the potential source of heterogeneity (p = 0.043, respectively) (Table 2 ). When compared with no treatment, TA did improve CPR (18 trials, OR=1.85, 95%CI=1.51 to 2.28). However, no significant difference was found between TA and sham control (penetration: 4 trials, OR=1.89, 95%CI=0.93 to 3.82; non-penetration: 6 trials, OR=1.04, 95%CI=0.76 to 1.43). Similar findings were observed for LBR results, TA group was significantly higher than that from no treatment (10 trials, OR=1.36, 95%CI=1.07 to 1.73), but not from sham control group (9 trials, OR=1.42, 95%CI=0.95 to 2.11). While according to acupuncture sessions, TA performed for 2 or more than 3 times showed significant effect on CPR (17 trials, OR=1.51, 95%CI=1.18 to 1.94; 9 trials, OR=1.99, 95%CI=1.52 to 2.60; respectively), while performed for 1 session found no significance (2 trials, OR=1.47, 95%CI=0.34-6.39). It was also an increasing improvement for TEAS if three or more treatments were administered (5 trials, OR=2.34, 95%CI=1.58 to 3.48) (Tables 2-3 ). The additional results of the effects of different acupuncture methods by baseline CPR, OPR or LBR of no treatment group were shown inFigure S4 . Studies with lower control group rates of CPR showed larger effects of TA (OR=2.14, 95%CI=1.63 to 2.81) than that with higher control group rates of CPR (OR=1.61, 95%CI=1.22 to 2.13).