After removal of 70 duplicates, a total of 216 citations were hit. We excluded 206 articles by reviewing titles and abstracts as they were considered non-relevant screened. We retrieved a further two articles through hand-searching. Therefore, 10 remaining studies were selected to undergo further full-text review. Among these studies, 5 were excluded for absence of quantitative data. In total, we included a total of two RCTs and three Retrospective cohort studies involving 5905 live-born infants in the quantitative synthesis. Study selection process is shown in PRISMA flow diagram (Fig.1).
Study characteristics and bias assessment
Five studies assessing neonatal outcomes of singleton pregnancies after TLI versus SI meeting the inclusion criteria. Studies were conducted in various regions including Hungary, Spain, America, England and China. Both in vitro fertilization (IVF) or intra cytoplasmic sperm injection (ICSI) cycles were performed in TLI and standard culture groups in our study. Further, both day 3 embryos and blastocysts transferred were included in our meta-analysis. The age of patients and mature oocytes might lead to bias in the results, so in one of our included studies , we only selected data matched the ages and oocytes of the TLI group. Among the 3 eligible retrospective cohort studies, we use NOS for quality assessment. The main trial and patient characteristics of the included studies were summarized in Table 1. Finally, Fig.6 showed a funnel plot of four studies included in this meta-analysis that reported low birth weight. All studies lay inside the 95% CIs, with an even distribution around the vertical, indicating no obvious publication bias.
Gestational weeks at delivery
There were four studies analyzing gestational weeks at delivery in singleton pregnancies between TLI group and SI group, including 2399 live births in the TLI group, and 1845 live births in the SI group and no significant difference was observed (Mean difference=0.09; 95%CI, -0.18 to 0.35; I2=33%; P=0.52; nTLI=2399, nSI=1845; Fig.2A).
Preterm birth (PTB, <37 weeks)
Four studies (two RCTs and two retrospective cohort studies) reported PTB, including 2399 live births in the TLI incubation group, and 1845 live births in the SI group. There were no significant differences between the two groups with the overall odds ratio (OR) for preterm birth was 0.98 (95%CI, 0.78 to 1.24; I2=0%; P=0.86; nTLI=2399, nSI=1845; Fig.2B).
Early preterm birth (early PTB, <32 weeks)
Two RCTs and two retrospective cohort studies reported early PTB, but a RCT (Insua et al. ) was excluded for its early PTB was defined as before 34 gestational weeks, not fulfilled our criteria which was defined as before 32 weeks. In total, 2250 singleton live births were included in the TLI incubation group and 1723 singleton live births in the SI group. There were also no significant differences between the two groups (OR=0.41; 95%CI, 0.11 to 1.50; I2=50%; P=0.18; nTLI=2250, nSI=1723; Fig.2C).
Live birth weight
Four studies (Bingxin et al.  Kovacs et al. ; Insua et al. ; Mascarenhas et al. ) were included for meta-analysis to investigate overall impact on the singleton live birth weight (BW), low birth weight (LBW, <2,500g), and very low birth weight (VLBW, <1,500g). Overall, our meta-analysis included 2399 live births in the TLI group, and 1845 in the SI group. The results showed no significant differences on birth weight in TLI group comparing to standard group (Mean difference=71.61; 95%CI, -30.55 to 173.77; I2=62%; P=0.17; nTLI=2399, nSI=1845; Fig.3A).
Low birth weight (<2,500g)
Four studies reported low birth weight, including 2399 singleton live births in the TLI group, and 1845 singleton live births in the SI group. The overall OR was of 0.85 (OR=0.85; 95%CI, 0.64 to 1.13; I2=0%; P=0.26; Fig.3B) when comparing LBW rate between different groups. Our result showed that TLI group would not increase risks of LBW compared with SI group in fresh ET (Embryo transfer, ET) cycles.
Very low birth weight (<1,500g)
4244 babies were reported about VLBW, including 2399 in the TLI group, and 1845 in the SI group. But no significant difference was observed between the TLI group and the SI group in fresh cycles (OR=0.58; 95%CI, 0.17 to 1.96; I2=50%; P=0.38, nTLI=2399, nSI=1845; Fig.3C). The results of fresh ET cycles indicated that TLI group would not increase the risk of VLBW rate.
As for macrosomia rate, only two studies involving 3913 live-born infants were included. Our meta-analysis showed that TLI group would not increase risks of macrosomia rate compared with SI group in fresh ET cycles (OR=1.11; 95%CI, 0.84 to 1.47; I2=0%; P=0.48, nTLI=2216, nSI=1697; Fig.4A).
With respect to the risk of congenital malformations, three studies involving 5252 live-born infants were included. We found no differences in incidence of congenital malformations between the TLI and SI groups (OR=1.19; 95%CI, 0.66 to 2.14; I2=0%; P=0.56; nTLI=2971, nSI=2281; Fig.4B).
Sex ratio analysis of newborn babies also showed that there was no difference between the two groups (OR=2.16; 95%CI, 0.53 to 8.81; I2=96%; P=0.28; nTLI=2024, nSI=1567; Fig.5).