Search strategy
PubMed, Embase, the Cochrane Library, CINAHL, CNKI, CBM, WANFANG and ClinicalTrials.gov were searched for citations published in any language from inception to 15 December 2021, including terms for ondansetron AND pregnancy (see Supplementary Table S1 for detailed search strategy). There were no restrictions on the publication language and publication status. Additionally, we manually reviewed the reference lists from included studies and relevant systematic reviews.

Eligibility criteria and study selection

Eligible studies included any observational study conducted in humans (such as prospective cohort, case-control, nested case-control, or case-cohort design) that reported a measure of association (such as hazard ratios or incident rate ratios for prospective studies; or odds ratios for retrospective studies) between use of ondansetron during pregnancy and the risk of abnormal pregnancy outcome (still birth, preterm birth, congenital malformations, et al). A study was considered eligible if it met the following criteria: 1) A healthy or disease-matched control (either nausea and vomiting of pregnancy or hyperemesis gravidarum) group was included. These control groups should be unexposed to ondansetron but they might be exposed to either non-teratogenic drugs or antiemetics other than ondansetron; 2) The data reported were not overlapping with another study. If an overlap between two studies was detected, we preferred to include the one that received higher score regarding methodological quality. The exclusion criteria were animal studies, editorials and reviews. One reviewer assessed titles and abstracts of all studies identified through electronic searches. Potentially eligible studies were reviewed independently by a second reviewer, with discrepancies resolved by discussion; and when necessary, a senior author was consulted to reach consensus.

Data extraction

Pairs of authors (XC and MYS) independently extracted details of the study design, country of conduct, exposure and outcome assessment, participant characteristics, and statistical analyses (including adjustment for confounders) from included studies, with discrepancies resolved by discussion. We extracted risk estimates (hazard ratio, relative risks, or odds ratios) with their 95% confidence intervals from the multivariable-adjusted models with the most complete adjustment for potential baseline confounders. Authors were contacted for additional data, when necessary.

Risk of bias of individual study

We independently assessed the risk of bias of individual studies by paired reviewers (XC and MYS) using a modified version of the Newcastle-Ottawa scale on the basis of selection of study groups, comparability of groups, and ascertainment of exposure(s) or outcome(s).26-27 We rated each study as being of high methodological quality(low risk of bias) and low methodological quality (high risk of bias) based on the following criteria: 1) studies were considered to be high methodological quality if all seven questions were assigned to be low risk of bias, or only 1 of the 7 questions was assessed as“definitely no”, six questions were low risk of bias, or 2 of the 7 questions were assessed as “probably no”, other five questions were low risk of bias; 2) studies were considered to be low methodological quality if they did not meet the criteria for high methodological quality as detailed above. The authors were not blinded to the author names, institutions, results or journals of the publications. Any disagreements were resolved through subsequent discussion with another author (QYY).

Data synthesis and statistical analysis

We used odds ratios (ORs) and 95% confidence intervals (CIs) as a measure of the association between ondansetron and abnormal pregnancy outcomes. Due to the low risk of abnormal pregnancy outcomes, hazard ratios (HRs) and relative risks (RRs) were included in the same meta-analysis. Subsequently, we conducted a subgroup analysis to identify possible differences between studies using ORs and those using HRs/RRs. If an interaction test indicated significant differences between subgroups, we used the results from studies with ORs for assessing the association between ondansetron and abnormal pregnancy outcomes. We pooled ORs and 95% CIs through the inverse variance weighted method using random-effects meta-analysis.28 To confirm the specific type of outcomes, we classify the abnormal pregnancy outcomes. We examined statistical heterogeneity among the studies with Cochrane’s Q test and I2 value.29 Publication bias was assessed using the Begg’s rank correlation test at the p < 0.05 level of significance for all meta-analyses with at least 5 studies.30 We also carried out priori sensitivity analyses. All analysis were performed with Stata V.16.1 software (Stata Corp, College Station, Texas, USA).

Certainty of evidence assessment

Paired reviewers (XC, MYS) independently rated the certainty (quality) of evidence as high, moderate, low or very low using the grading of recommendations, assessment, development and evaluation (GRADE) approach considering risk of bias, imprecision, inconsistency, indirectness and publication bias.31 Any conflict was resolved through the adjudication of a third reviewer (QYY).