Criteria for considering studies for this review
We performed a pairwise meta-analysis.
Types of studies
We included intervention studies in the form of randomized controlled trials and nonrandomized controlled trials that compared sequential embryo transfer to single transfer protocols.
Types of participants
Participants were suffering from infertility and underwent IVF/ICSI protocol.
Types of interventions
One of the interventions was sequential embryo transfer and the control interventions was single transfer (details of protocols are shown in Table 1).
Types of outcome measures
Primary outcomes:
- Clinical pregnancy rate [10]
- Live birth rate [10]
Secondary outcomes:
- Multiple pregnancies rate
- Miscarriage rate
Data collection and analysis
Selection of studies
Studies screening was carried out by two independent researchers (LC, WG) by reading article titles and abstracts for inclusion or exclusion. Any differences between the parties regarding the research were resolved through discussion. When there are any irresolvable differences, the records will be evaluated by a third author (FW).
Data collection process
Data were extracted by one reviewer (WG), and checked by a second (LC). For each included study, the following information would be collected: study name; design; methods; setting and time period; information about the participants; stimulation protocols; drop-outs; interventions; outcomes, including clinical pregnancy rate, live birth rate, multiple pregnancies rate, and miscarriage rate.
Search methods for identification of studies
This study was based on the PRISMA guidelines for systemic review and meta-analysis [11]. The electronic databases used were Cochrane Library, EMBASE and MEDLINE, from 2011 to October 2021 without limitation of language, region, or publication type. We included intervention studies including randomized controlled trial, non-randomized studies with comparison groups (NRCTs), prospective controlled study, retrospective cohort study. The treatment group include double-embryo transfer or sequential transfer. The inventions of the control group included two or more embryos in day-2/day-3 or two blastocysts in day-4/5/6 (details of protocols are shown in Table 1). The strategies for electronic search at the database used a combination of (MeSH): (((((((Double-Embryo Transfer) OR (sequential transfer)) OR (consecutive transfer)) AND ((IVF) OR (ICSI))) NOT (sequential culture)) NOT (elective single embryo transfer)) NOT (single embryo transfer).
We excluded studies as followed: (1) self-controlled study; (2) conferences, review articles, books, notes, thesis, editorial, posters, letters, case reports; (3) overlapped data sets, unreliable extracted data and paragraphs only abstract available.
Assessment of risk of bias in individual studies
Quality of studies
For randomized controlled trials, we use the Cochrane collaboration tools to assess the risk of bias[12], including selection bias (random sequence generation), selection bias (allocation sequence concealment), performance bias (blinding of participants and personnel performance), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting), and other bias. The reviewers rated the quality of the included studies as low risk, unclear risk or high risk.
For non-randomized controlled studies, we use Newcastle-Ottawa Scale (NOS) to assess the quality[13]. The NOS contains eight items. The NOS ranges from zero to nine stars as follows: selection of the study group (up to 4points), comparability of cohorts (up to 2 points) and ascertainment of outcome (up to 3 points). Poor-quality studies less than four points, medium-quality studies between four to six points and high-quality studies achieve more than seven points.
Data synthesis
All data were processed by Review Manager (version 5.2). Risk ratio (RR) and 95% confidence intervals (CIs) for variables with dichotomous data was used for RCTs and odds ratios (ORs) for nonrandomized studies. For these variables, the Mantel-Haenszel method was used to calculate the weighted summary RR. For continuous data, mean difference (MD) was calculated and corrected for sample bias. Randomized and non-randomized studies had similar (homogeneous) estimates of effect sizes, so we pooled the results[14]. We use GRADE-pro [15] to make the ’Summary of findings’ table. We summarized and graded the certainty of the evidence for critical outcomes clinical pregnancy rate, live birth rate, multiple pregnancies rate, and miscarriage rate.
Investigation of heterogeneity
Higgins I2 values[16] were used to assess statistical heterogeneity between studies, and values of I2 25% were indicative of low heterogeneity. When I2≤50%, the fixed effects model was used; when 50%≤I2≤75%, the random effects model was used for analysis. I2 of our results were all homogeneous according to the chi-square test and I2≤50%, as a result fixed-effect model was used.
Sensitivity analysis
For all outcomes such, we examined the sensitivity versus risk of bias (selection bias, performance bias, attrition bias, detection bias, selective reporting and reporting bias).