Study selection
Studies were eligible for inclusion if they were observational cohort or case-control studies. Case reports or series were excluded from the present review. We included studies that involved pregnant women regardless of study setting. The exposures of interest were the infancy parameters of presently pregnant women and were defined as follows: LBW, birth weight <2500 g 13; small for gestational age (SGA), birth weight <10th percentile for the given gestational age, stratified by sex, using the average weight of gestational age 30; and preterm birth, gestational age of <37 weeks 31. When data on both birth weight and gestational age were reported, we extracted data on birth weight in preference. The comparator group comprised women who were not born small or born at full term.
The outcome of interest was GDM, as defined by the International Association of Diabetes Pregnancy Study Groups (IADPSG), World Health Organization (WHO), American Diabetes Association or Endocrine Society, or International Classification of Diseases 11th revision (ICD-11) or earlier 32-36. If studies used other definitions, they were included in the present review; however, we removed them to assess the robustness of the pooled estimates. For studies that reported LBW, preterm birth, or SGA as a risk factor in pregnant women without reporting the association with GDM, we contacted study authors to acquire estimates of such associations, where available. These additional estimates were included in the present analysis, provided they were measures of an association between at least one of the exposure factors and the outcome of interest.
Two investigators independently screened article titles and abstracts to shortlist relevant studies; subsequently, the same sets of authors assessed the full text for study eligibility. In cases where data were incomplete and precluded study eligibility assessment, we contacted study authors with requests for clarification. Multiple publications were assessed together; the record with the most complete data was included in the present review.
Data extraction and quality assessment
Two investigators independently extracted data from all included studies, using a pilot-tested, uniform data extraction sheet. Any discrepancies between reviewers were resolved through consensus between two reviewers or arbitration by a third reviewer, as required. For studies that compared three or more exposure groups, we contacted study authors to obtain data comparing two groups of interest. In cases where this approach was unsuitable, we extracted the relevant data, as reported, and performed subgroup comparisons between the two groups subdivided by specific thresholds (i.e., birth weight 2500 g, <10th percentile, and gestational age 37 weeks for LBW, SGA, and preterm birth, respectively), as this approach may have resulted in conservative effect estimates. The same authors who performed data extraction also independently assessed the risk of bias in each study, using the Quality In Prognosis Studies (QUIPS) tool 37. We prospectively identified the following candidate confounders: age, obesity, smoking status, socioeconomic status, diabetes mellitus before the index gestation, and family history of diabetes 38, 39.