Congenital anomaly and perinatal outcome following blastocyst‐ vs cleavage‐stage embryo transfer: systematic review and network meta‐analysis

ABSTRACT Objectives To compare the reported rate of any congenital anomaly and perinatal outcome of pregnancy following blastocyst‐ vs cleavage‐stage embryo transfer using a pairwise meta‐analysis and to evaluate the same outcomes following fresh‐blastocyst, frozen‐blastocyst, fresh‐cleavage or frozen‐cleavage embryo transfer using a network meta‐analysis. Methods A literature search was performed in PubMed, Scopus and CENTRAL and registers for ongoing studies, from inception to February 2022, for randomized controlled trials (RCTs) with any sample size and observational studies including at least 100 live births per group, comparing the rates of any congenital anomaly and perinatal outcome of pregnancy following fresh/frozen embryo transfer at cleavage (day 2–3) vs blastocyst (day 5–7) stage. Risk ratios (RRs) along with their 95% CIs were pooled via a random‐effects model meta‐analysis. Within a frequentist network meta‐analysis framework, outcomes of all four treatment modalities (i.e. fresh‐blastocyst, fresh‐cleavage, frozen‐blastocyst, frozen‐cleavage) were compared further. Any congenital anomaly constituted the primary outcome, whereas preterm delivery (delivery < 37 weeks), low birth weight (LBW; < 2500 g), gender of the neonate (male), perinatal death and healthy neonate (defined as liveborn neonate, delivered at term, weighing ≥ 2500 g, surviving for at least 28 days postbirth and without any congenital anomaly) were considered as secondary outcomes. Subgroup analyses by plurality (liveborn singleton vs multiple pregnancy) were conducted in the pairwise and network meta‐analyses. The risk of bias was assessed using the RoB2 tool for RCTs and the ROBINS‐I tool for non‐randomized studies. Certainty of evidence was assessed using GRADE. Results Through the literature search, 550 studies were retrieved and 33 were included in the systematic review. We found no significant difference in the risk for any congenital anomaly between blastocyst‐ and cleavage‐stage transfer (RR, 0.80 (95% CI, 0.63–1.03); 10 studies; n = 192 442; I 2 = 85.5%). An increased probability of a male neonate was observed following blastocyst‐ vs cleavage‐stage transfer (RR, 1.07 (95% CI, 1.06–1.09); 18 studies; n = 227 530; I 2 = 32.7%). No significant differences in other secondary outcomes or significant subgroup differences between liveborn singletons and multiple pregnancies were observed. The network meta‐analysis showed a significantly lower risk for LBW following frozen‐blastocyst vs fresh‐blastocyst (RR, 0.76 (95% CI, 0.60–0.95)) or fresh‐cleavage (RR, 0.74 (95% CI, 0.59–0.93)) transfer. Frozen‐blastocyst transfer was associated with an increased risk for perinatal death compared with the fresh‐cleavage method (RR, 2.06 (95% CI, 1.10–3.88)). The higher probability of a male neonate following blastocyst transfer remained evident in the network comparisons. All outcomes were assessed to be of very‐low certainty of evidence. Conclusions Current very‐low certainty of evidence shows that there may be little‐to‐no difference in the risk for congenital anomaly or adverse perinatal outcome of pregnancy following blastocyst‐ vs cleavage‐stage embryo transfer, although there was a slightly increased probability of a male neonate following blastocyst transfer. When considering cryopreservation, frozen‐blastocyst transfer was associated with a reduction in the risk for LBW compared with both fresh‐transfer modalities, and fresh‐cleavage transfer may be associated with a reduction in the risk for perinatal death compared with frozen‐blastocyst transfer. High‐quality RCTs with separate data on fresh and frozen cycles and consistent reporting of culture conditions and freezing methods are mandatory. Individual participant data meta‐analyses are required to address the substantial inconsistency resulting from current aggregate data approaches. © 2022 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.


INTRODUCTION
Cleavage-stage embryo transfer following controlled ovarian hyperstimulation for assisted reproduction used to be the most common method of embryo transfer. Extending embryo culture to the blastocyst stage (day 5-7) has been suggested to reduce the risk of multiple pregnancy without compromising the pregnancy rate [1][2][3] . Theoretical advantages of longer cell culture include the improved selection of higher-quality embryos, better embryo-endometrium synchronization and reduced uterine contractility [4][5][6][7] . However, the extended embryo culture increases the exposure of the embryo to cell culture conditions and may increase epigenetic alterations 8 . Clinical evidence also remains conflicting; while some studies report perinatal complications to occur more frequently in pregnancies following embryo transfer at the blastocyst stage [9][10][11][12][13] , others have reached different conclusions [14][15][16][17][18] . There is currently no high-quality evidence to support the use of either strategy 2 .
Since the publication of previous systematic reviews 9,[19][20][21] , additional original studies have emerged 16,18,[22][23][24][25] , including those with a randomized design 26,27 . Moreover, previous meta-analyses merged data from fresh-and frozen-cycle embryo transfers, thus not evaluating the effects of cryopreservation and developmental stage at the time of transfer on their results 19,28,29 . Other limitations included the observational design of the eligible studies 28 and the fact that variations in assisted reproduction procedures, including culture conditions and media used, may have affected the final results 17,30 . Even though the most recent systematic review investigated explicitly the potential impact of cryopreservation on perinatal outcome via subgroup analysis, it was not designed to compare all the available modalities 31 . Finally, previous reviews 9, [19][20][21]31 focused primarily on singleton pregnancy; however, large cohorts 14,18 providing data on infants from multiple pregnancy have emerged.
The objective of this systematic review and meta-analysis was to compare the rate of any congenital anomaly and the perinatal outcome of singleton and multiple pregnancy resulting from blastocyst-vs cleavage-stage embryo transfer. This work further aimed to compare the same outcomes between pregnancies resulting from fresh-blastocyst, frozen-blastocyst, fresh-cleavage and frozen-cleavage embryo transfers by using a frequentist network meta-analysis framework.

Protocol and registration
The protocol followed the updated Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines 32 and was registered with PROSPERO in July 2020 (registration number: CRD42020179040) 33 .

Eligibility criteria
Eligible studies recruited women with a delivery following fresh-or frozen-embryo transfer at the cleavage or blastocyst stage. The comparisons were made between pregnancies resulting from embryo transfers at the cleavage (day 2-3) vs blastocyst (day 5-7) stage. Randomized controlled trials (RCTs) with any sample size and observational studies (i.e. cohort or case-control studies) including at least 100 live births following cleavage-stage embryo transfer and 100 live births following blastocyst-stage embryo transfer were considered eligible.
The primary outcome was the risk of any congenital anomaly, either external or internal, minor or major, as defined by the World Health Organization (WHO) 34 and the Centers for Disease Control and Prevention 35 . The rationale for merging the categories of major and minor congenital anomalies into one outcome (defined as any congenital anomaly) included practical reasons related to statistical power.
The published protocol regarding birth weight and gestational age at delivery outcomes was elaborated. Given the limited data availability, the risk of VLBW/LBW vs NBW/HBW/VHBW was compared between blastocyst and cleavage groups. LBW and VLBW were merged into a single category (i.e. LBW < 2500 g) and recorded as an event, and all other categories (NBW, HBW and VHBW) were used as the comparator. In the same way, VPTD and PTD were merged into a single-event category (i.e. gestational age at delivery < 37 weeks or PTD).

Literature search
The following databases/registers were screened: MED-LINE (accessed through PubMed), Scopus, CENTRAL, ClinicalTrials.gov and WHO ICTRP. No date restrictions were applied. The following search algorithm was used: ''embryo* AND (cleav* OR ''day 2'' OR ''day 3'') AND (blastoc* OR ''day 5'' OR ''day 6'') AND (''Congenital Abnormalities''[MeSH] OR perinatal OR ''birth weight*'' OR ''preterm birth*'')''. The search algorithm was adjusted for each database while maintaining a common overall architecture. The date of the initial search was 6 June 2021. The literature search was repeated on 6 February 2022 to prevent the potential omission of any recently published eligible studies.
Retrieved records underwent semiautomatic deduplication using EndNote 20 (Clarivate Analytics, London, UK) 36 . The title and abstract of all retrieved articles were screened by two independent reviewers (C.S. and M.P.). After exclusion of studies deemed irrelevant, the remaining articles were accessed in full and those fulfilling the eligibility criteria were included in the study. Disagreements were resolved by consensus. The reference lists of systematic reviews and retrieved articles were also searched for additional papers, and no language restrictions were applied.

Study selection and data collection
Two authors (V.K. and C.S.) assessed independently the obtained records for eligibility. Agreement regarding potential relevance was reached by consensus. Full-text copies of the relevant papers were obtained, and another set of reviewers (D.M.T., V.K., I.B. and M.P.) independently extracted the items of interest using prepiloted forms, agreed upon by all authors. Inconsistencies were discussed by the reviewers until consensus was reached or by discussion with a third author (W.P.M.). If more than one study was published on the same cohort with identical endpoints, the report containing the most comprehensive information on the population was included to avoid overlapping data (Appendix S1). When the information of interest was not provided, the authors of the original studies were contacted via email, using prepiloted forms that included the requested data.
Prespecified items for which data were sought were: name of first author, date of publication, location of study, published protocol (yes/no), funding (yes/no), mean age and body mass index of participants, parity, type of infertility, smoking (pack/years), alcohol consumption (glasses per week), number of participants in each treatment arm, oxygen pressure (low ≤ 6% or atmospheric) in which blastocysts were cultured, number of congenital anomalies in each arm, gestational age at delivery, birth weight, gender, healthy neonate at home and number of perinatal deaths in each arm. Estimates for the contingency table were obtained from the reported risk ratio (RR) using commonly used formulas in the meta-analysis literature 37 .

Risk of bias
Two sets of authors (W.P.M., D.M.T. and V.K., I.B.) assessed independently the risk of bias of eligible studies. For RCTs, the risk of bias was assessed using the risk of bias (RoB) 2 tool 38 . RCTs were classified as having low risk of bias, risk of bias of some concerns or high risk of bias. The ROBINS-I tool was used to assess the risk of bias within each non-randomized study 39 . The non-randomized observational studies were classified as having low, moderate, serious or critical risk of bias. Regarding RCTs, blinding was not considered as a likely factor to influence the risk of performance and detection bias for the reproductive outcome 40 . Visualization of the final assessments was performed in line with the common practice of presenting bar charts for the distribution of the judgments in each domain 41 .

Summary measures and assessment of heterogeneity
Continuous variables were presented as mean ± SD, median (interquartile range) or median (range). RRs along with their 95% CIs comparing blastocyst-vs cleavage-stage embryo transfer were provided for all outcomes. Pairwise meta-analysis was performed by fitting a random-effects (DerSimonian-Laird) model, as the true effect size should not be assumed to be the same across studies, and this model incorporated the observed heterogeneity among studies, obtaining more conservative Outcome following blastocyst-vs cleavage-stage embryo transfer 15 CIs 37 . Reporting was separate for randomized and non-randomized studies.
Interstudy heterogeneity was assessed by calculating the Cochran's Q statistic and the inconsistency index (I 2 ), with values > 50% indicating at least substantial heterogeneity 42 .

Small-study effects
As a test for potential small-study effects, funnel plots for all outcomes are presented 43 . For plots including at least 10 studies, formal testing of asymmetry was also performed (Egger's test; P < 0.1 indicating significance) 37,43 . When funnel plot asymmetry was detected, contour-enhanced funnel plots were created to investigate potential reasons for the observed asymmetry 37 . If publication bias was considered a plausible explanation for the asymmetry and substantial heterogeneity was absent 37 , the non-parametric trim-and-fill method was employed to correct for it 44 .

Subgroup analysis
Considering recent data on the potential impact of oxygen concentration during embryo culture on resulting pregnancy outcome 45,46 , a subgroup analysis by oxygen pressure during embryo culture was planned at protocol stage.
A post-hoc subgroup analysis stratified by plurality of pregnancies resulting in live births (i.e. liveborn singletons vs multiples) was conducted. P-values < 0.1 indicated significant between-subgroup differences.

Network meta-analysis
With respect to all outcomes, the following arms were compared within a frequentist network meta-analysis framework: fresh-blastocyst, fresh-cleavage, frozenblastocyst and frozen-cleavage embryo transfers. Network geometry was visualized 47 . Global and local tests for inconsistency were applied to evaluate for transitivity 48 . The global approach for overall inconsistency was computed according to the type of between-treatment comparison for all cases, and the values were then used to test for global linearity using the Wald test 49,50 . The local test was based on the node-splitting method by Dias et al. 51 , implementing the symmetrical alternative proposed by White 50 . Log-transformed RRs were used as effect measures and were subsequently back-transformed to RRs for easier interpretation of results. Pooled RRs in each comparison set and pooled overall RRs were visualized via network forest and interval plots. Cumulative rankings for identifying superiority among the four arms, in terms of all investigated outcomes, were calculated and were then visualized using rankograms 52 . Treatment was considered superior if it had a higher cumulative probability of decreasing the risk of the investigated outcome, except for the healthy neonate outcome. If adequate data to create connected networks were available, separate post-hoc network meta-analyses of data on liveborn singletons and multiples were performed.
Statistical significance was set at a two-tailed P-value < 0.05. All analyses were conducted using Stata version 13 (StataCorp. LLP, College Station, TX, USA) using the metan and network packages.

Certainty of evidence
The certainty of evidence was evaluated using the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) approach, which takes into consideration study limitations, indirectness, imprecision, inconsistency and publication bias 53,54 .

Subgroup pairwise meta-analysis
Two studies 71,72 reported on the oxygen conditions during culture, and in both studies, it was set at 5%. Due to the lack of variability in oxygen conditions, this perprotocol subgroup analysis was not performed.
No significant between-subgroup differences (singleton vs multiple pregnancy) were observed with respect to congenital anomaly, PTD, LBW and gender of the neonate ( Figure S2a  could be conducted. The results of the analysis focusing on singletons only were similar to those of the analysis of the overall population including both singleton and multiple pregnancy.

Network meta-analysis
Five studies 10,18,29,60,70 were stratified by both the use of cryopreservation (i.e. fresh vs frozen cycle) and embryo developmental stage at transfer (i.e. blastocystvs cleavage-stage transfer) and led to connected networks ( Figure S3). Node sizes indicated fewer studies reporting on frozen cycles. The body of evidence was consistent in all network meta-analyses, except for the meta-analysis on any congenital anomaly ( Figures S4a-e Figure S6b). When considering only multiple pregnancy, fresh-cleavage transfer reduced significantly the risk of PTD compared with frozen-cleavage, fresh-and frozen-blastocyst transfers (n = 14 116 deliveries, four studies 12,18,68,69 ; Figure S7a). Additionally, frozen-blastocyst transfer increased significantly the risk for PTD compared with fresh-blastocyst transfer in multiple pregnancy.

Figure 3
Pairwise meta-analysis forest plot showing risk ratio with 95% CI for any congenital anomaly, according to whether pregnancy was conceived following cleavage-or blastocyst-stage transfer. Weights are from random-effects model; continuity correction was applied to studies with zero cells. Only first author's name is given for each study. DL, DerSimonian-Laird random-effects meta-analysis model.  Frozen-blastocyst transfer increased significantly the risk of perinatal death when compared with fresh-cleavage transfer (RR, 2.06 (95% CI, 1.10-3.88)). All other comparisons showed no significant differences among the interventions (n = 45 333 infants/fetuses, four studies 10,55,64,69 ; Figures 4d, S5d). The results did not change when only singletons were analyzed (n = 44 461 infants/fetuses, three studies 10,55,64 ; Figure S6d), whereas data on multiple pregnancy were inadequate to create a connected network.

Ranking of treatments
Relative ranking of treatments with respect to all outcomes (considering both singleton and multiple pregnancy) via calculation of cumulative ranking probabilities and visualization through rankograms is available in Figures S8a-e. Potential superiority was demonstrated by the fresh-blastocyst arm (followed by the frozenblastocyst arm) in reducing the risk of any congenital anomaly in liveborn neonates, fresh-and frozen-cleavage arms in reducing the risk of PTD, frozen-blastocyst arm (followed by frozen-cleavage) in decreasing the risk of LBW, fresh-cleavage arm (followed by frozen-cleavage) in reducing the risk of perinatal death, and fresh and frozen-cleavage arms in reducing the probability of higher male births. These findings should be interpreted in combination with the effect estimates reported above.

Small-study effects
The funnel plots for all outcomes are presented in Figures S9a-e. The analysis on the healthy neonate outcome included only two studies; therefore, a funnel plot was not created. Visual inspection and the formal test (Egger's regression, when performed) demonstrated no major funnel plot asymmetry with respect to the rest of the investigated outcomes.

Certainty of evidence
All outcomes were classified as having very-low certainty of evidence due to limitations of the included studies. Certainty of evidence on the congenital anomaly (95% CI, 0.63-1.03; I 2 = 85.5%), PTD (95% CI, 0.92-1.14; I 2 = 95.0%) and LBW (95% CI, 0.87-1.05; I 2 = 95.2%) outcomes was downgraded further due to both imprecision and inconsistency. Certainty of evidence on the healthy neonate (I 2 = 96.2%) outcome was downgraded further due to inconsistency, while certainty of evidence on the perinatal death outcome was downgraded further due to imprecision only (95% CI, 0.78-1.22).

Summary of evidence
The present systematic review demonstrates that there may be little-to-no difference in the risk of any congenital anomaly or adverse perinatal outcome in the newborns following blastocyst-vs cleavage-stage embryo transfer when the effect of cryopreservation is not taken into account. The network meta-analysis supported the findings of the pairwise meta-analysis concerning the primary outcome. When considering the overall included population and singletons only, it further showed a lower risk of LBW in the frozen-blastocyst compared with the fresh modalities; in multiple pregnancy, the risk was reduced following both frozen-blastocyst and fresh-cleavage transfers when compared with fresh-blastocyst transfer. The fresh-cleavage modality demonstrated potential superiority in reducing the risk for perinatal death when compared with frozen-blastocyst transfer. The higher probability of a male neonate following blastocyst transfer was demonstrated by the pairwise and network meta-analyses. In the multiple-pregnancy group, fresh-cleavage transfer reduced significantly the risk of PTD compared with all other transfers, and frozen-blastocyst transfer increased significantly the risk compared with the fresh-blastocyst one. All other comparisons revealed non-significant differences between the interventions.

Clinical significance of this study with respect to that of previous studies
A considerable number of original studies on the topic 16,18,[22][23][24]28,59 have emerged since the publication of the latest relevant systematic reviews [19][20][21] . A key challenge encountered by former studies was the inability to explore concurrently the effect of developmental stage at the time of transfer and the impact of cryopreservation on the outcome of interest. Two former systematic reviews 19,21 attempted to address this issue. A recent, more up-to-date systematic review has adopted a similar approach with more available data 31 . The present work aimed to evaluate studies that reported on outcomes of interest in a stratified fashion to create connected networks of four interventions and synthesize separately data on multiple pregnancy.
The concurrent consideration of the effect of cryopreservation and the stage of embryo transfer when conducting comparisons is of high clinical value. The freeze-all strategy potentially results in lower odds of development of ovarian hyperstimulation syndrome and comparable cumulative pregnancy, perinatal and neonatal outcomes when compared with conventional strategies [73][74][75] . Extending in-vitro culture to the blastocyst stage is based primarily on previously reported higher live-birth rates, especially in good-prognosis patients 2,6 , but such an effect has been recently questioned 20 . Pregnancy outcomes may be altered when combining certain freezing protocols with specific developmental stages 2,76 ; this observation reinforced the need for the network comparisons of the present meta-analysis.

Interpretation and comparison with other systematic reviews
In accordance with most former systematic reviews 6,20,31 , we found no significant difference in the risk of any With respect to PTD, previous meta-analyses have indicated a slightly higher risk in the blastocyst-transfer group 6,9,[19][20][21] . Former studies have attributed this association to potential genetic or epigenetic changes in trophodermal cells during the extended culture 13,76,77 , defective implantation 6 or the association of male gender with preterm birth 9,78 . The most up-to-date work 31 demonstrated a higher risk of VPTD following fresh-blastocyst compared with fresh-cleavage transfer 21,31 . Similar findings on PTD were reported after synthesis of adjusted data 31 ; this evidence was derived from subgroup analyses (frozen-blastocyst vs frozen-cleavage, and fresh-blastocyst vs fresh-cleavage), yet the overall comparison indicated no significant difference, similar to this study 31 . Notably, the data of most reviews applied to singleton pregnancy, while the present work considers both singleton and multiple pregnancy. However, the subgroup analyses by plurality of pregnancy indicated no significant between-subgroup differences. When considering singletons only, the results were in accordance with those of the overall mixed population.
Although the pairwise analysis demonstrated no significant difference in the risk of LBW, in agreement with the findings of the majority of previous papers 9,20,21,31 , a network comparison demonstrated that the risk was significantly lower following frozen-blastocyst vs both fresh-transfer modalities. In contrast, a slightly higher risk associated with the blastocyst stage was observed by one systematic review, which lacked data on frozen cycles 6 , while in another, no differences were noted between the subsets 23 . This may be related to certain reported advantages of frozen-over fresh-embryo transfer 2,20,79,80 , although current evidence suggests that the potential effect of frozen transfer on the long-term health of newborns needs further investigation 81,82 .
We also noted that fresh-cleavage transfer was associated with a lower risk of perinatal death compared with the frozen-blastocyst modality. A higher risk for perinatal mortality in the blastocyst group was indicated by one systematic review 20 , supporting the network findings of this study. When considering the comparison between any blastocyst and any cleavage technique, however, no significant differences have been observed between groups 31 .
A previous systematic review focusing on fresh-embryo transfer only reported a potential association between blastocyst transfer and offspring sex ratio skewed in favor of males 83 . This was supported by the findings of our work. Male embryos have been reported to develop more rapidly in animal species, increasing their probability of being selected for transfer at the blastocyst stage 83 .
Finally, the performed relative ranking of treatments with respect to all outcomes revealed the potential superiority of blastocyst transfer in reducing the risk of any congenital anomaly, cleavage transfer in reducing the risk for PTD, perinatal death and higher incidence of live male births, and frozen modalities in decreasing the risk of LBW. Nevertheless, interpretations of findings may not always be feasible based on current available evidence and should be interpreted in combination with the reported effect estimates.

Strengths and limitations
This is the first systematic review to investigate the effect of both the stage of embryo transfer and cryopreservation on the rate of any congenital anomaly and perinatal outcome by directly and indirectly comparing outcomes of all possible treatment combinations within a frequentist network meta-analysis. Although a meta-analysis of RCTs would be preferred, it could not be performed given the limited availability of such studies; hence, the meta-analysis included observational studies with a sample of at least 100 live births per group. Few of the 33 included studies were judged to have a low overall risk of bias, while all outcomes were assessed as having very-low certainty of evidence and were downgraded by the limitations of the included studies, inconsistency and/or imprecision. Inconsistency was noted in the congenital anomaly network; therefore, findings should be interpreted with caution. Due to the lack of individual patient data or inconsistent aggregate data reporting, we were unable to account for potential confounders like freezing methods, culture conditions or day of blastocyst transfer.

Conclusion
Current very-low certainty of evidence shows that there may be little-to-no difference in the risk of any congenital anomaly and adverse perinatal outcome following blastocyst-vs cleavage-stage embryo transfer, though there might be a slightly increased probability of a male newborn following blastocyst transfer. Network meta-analysis demonstrated a significantly lower risk of LBW following frozen-blastocyst transfer vs both fresh modalities, a significantly lower risk of perinatal death following fresh-cleavage vs frozen-blastocyst transfer and a significantly lower risk for PTD of multiple pregnancies following fresh-cleavage transfer compared with all other modalities. High-quality RCTs with separate data on fresh and frozen cycles, and explicit, consistent reporting of culture conditions and freezing methods are mandatory in future research. Individual participant data meta-analyses may account for potential confounders and resolve the substantial inconsistency associated with current aggregate data approaches.

SUPPORTING INFORMATION ON THE INTERNET
The following supporting information may be found in the online version of this article: Appendix S1 Studies with potentially duplicate populations after comparisons of study centers and study periods Figure S1 Pairwise meta-analysis forest plots showing risk ratios for preterm delivery (< 37 weeks) (a), low birth weight (< 2500 g) (b), perinatal death (c), male neonate (d) and healthy neonate (e). DL, DerSimonian-Laird random-effects meta-analysis model.

Figure S4
Network sidesplitting of nodes (local test on inconsistency), examining differences between direct and indirect evidence (measure of effect: risk ratio, log scale) for any congenital anomaly (a), preterm delivery (< 37 weeks) (b), low birth weight (< 2500 g) (c), perinatal death (d) and male neonate (e). A symmetrical alternative of the Dias et al. 51 method was employed.

Figure S6
Network meta-analyses forest and interval plots for singleton pregnancies showing risk ratio for any congenital anomaly (a), preterm delivery (< 37 weeks), low birth weight (< 2500 g) (c), perinatal death (d) and male neonate (e).

Figure S7
Network meta-analyses forest and interval plots for multiple pregnancies showing risk ratio for preterm delivery (< 37 weeks) (a) and low birth weight (< 2500 g) (b). Only data on preterm delivery and low birth weight were sufficient to create connected networks.

Figure S8
Network rankograms and cumulative ranking probabilities for identifying superiority of treatments in reducing the risk for any congenital anomaly (a), preterm delivery (< 37 weeks) (b), low birth weight (< 2500 g) (c), perinatal death (d) and male neonate (e). This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.