AbstractBACKGROUNDThe aromatase inhibitor letrozole is increasingly recommended for ovulation induction, as it is more effective with fewer side-effects than other agents. But many clinicians are reluctant to use the drug for fertility treatment due to a strong-label warning against its use, which warns about congenital malformation risk to the foetus in women seeking pregnancy.OBJECTIVE AND RATIONALEThe aim of this study was to determine the risks of congenital malformations and pregnancy loss with letrozole compared with clomiphene primarily, and with other fertility drugs and natural conception.SEARCH METHODSA systematic review and meta-analysis using PRISMA harms guidelines. We searched MEDLINE, EMBASE and other sources from inception until January 2020, with the MeSH words for ‘letrozole’ and pregnancy OR foetal/neonatal outcome. We included studies reported on congenital malformations in foetuses born to mothers conceived after fertility treatment, with letrozole versus clomiphene, placebo, gonadotrophins, metformin, natural conception or other agents, from randomised trials, comparative cohort studies and non-comparative observational cohorts. Quality of the studies was assessed using Cochrane risk of bias tool and Newcastle Ottawa Scale. The McMaster tool was used to assess the quality of reported harm for foetal congenital malformations in the studies. We compared the absolute risk of events using risk difference measures and pooled the findings using a fixed-effect model. We evaluated the statistical heterogeneity using forest plots and the I2 statistic and funnel plot to assess publication bias. We assessed the strength of evidence for congenital malformation and pregnancy loss as per the GRADE recommendations and with the Fragility index.OUTCOMESWe included 46 studies (18 randomised trials; 21 comparative cohorts; 7 non-comparative cohorts). Overall 2.15% (101/4697; 95% CI 1.7 to 2.5) of babies conceived on letrozole for fertility treatment had congenital foetal malformations. We did not observe a significant increase in congenital malformations with letrozole versus clomiphene in the randomised trials (risk difference (RD) 0.01, 95% CI −0.02, 0.03; I2 = 0%; 14 studies) and found a significant reduction in the cohort studies (RD −0.02, 95% CI −0.04, −0.01; I2 = 0%, 11 studies). The fragility index was 44% (7/16) (either an increase in the intervention arm or a decrease in control arm was needed to alter the results). The risks of pregnancy loss were not increased with letrozole versus clomiphene in the 14 randomised trials (RD −0.01, 95% CI −0.06, 0.04; I2 = 0%), and the risks were reduced in the six cohort studies (RD −0.09, 95% CI −0.17, −0.00; I2 = 68%). The GRADE quality of evidence was low to moderate for congenital malformations and pregnancy loss. We did not find any increased congenital malformation risk with letrozole versus gonadotrophins, natural conception or natural cycle ART, but the number of studies was small.WIDER IMPLICATIONSThere is no evidence that letrozole increases the risk of congenital foetal malformation or pregnancy loss compared with clomiphene, natural conception or other fertility agents, to warrant warning against its use. Given its therapeutic benefits and lack of evidence of harm to the foetus, clinicians should consider letrozole as first-line agent for ovulation induction.