Assessment of bias.

a: Evaluated as moderate risk in secondary analysis (low risk in primary analysis) due to incomparable treatment effect. b: Studies only eligible for secondary analysis. c: In the study of Andersen (2013), the OR was adjusted by maternal age, number of previous miscarriages, income and education. d: In the study of Le guyen, the OR was adjusted by maternal age, long-term illnesses, parity and multiple pregnancy. e: In the study of Meeraus (2015), the Hazard Ratio was adjusted by maternal age, Townsend quintile, year of delivery, smoking, alcohol problems, obesity, illicit drug use, treatment of chronic medical conditions and potentially neurologically-damaging infection during pregnancy. f: In the study of Muanda (2017), cases and controls were matched by gestational age and year of pregnancy; in the analysis of specific macrolides, the ORs were adjusted by 11 covariates, e.g. maternal age, education level, chronic comorbidities, maternal infections (urinary tract infection, respiratory tract infection, bacterial vaginosis and sexually transmitted infections) and prior exposure to antibiotics.