Results
Figure 1 presents the flow of studies through the present review selection process. After screening 15,281 records, 59 records representing 44 studies were assessed for eligibility based on the full text. Finally, 18 studies including 827,382 participants were included in the qualitative synthesis; 15 studies including 825,622 participants were included in quantitative synthesis. Supplementary Table S3 lists all excluded studies with reasons for exclusion. We did not find any ongoing or unpublished studies by searching study registries. By contacting authors, we obtained unpublished data from two studies23, 25.
Table 1 shows the characteristics of the included studies. Nine studies (810,197 participants) used population-based samples, 2 (6,915 participants) were multicentre studies, 6 (9,439 participants) were single-centre studies, and 1 (831 participants) did not specify the study setting. Supplementary Table S4 shows the details of the inclusion and exclusion criteria of the included studies. All studies were conducted in high-income countries, mostly between the late 1990s and early 2010s. The studies included participants of non-Hispanic White, Hispanic, African, Asian, or Indian descent. Two studies (28,722 participants) only included women about to deliver their first child. Two studies (140,714 participants) compared pregnant women born preterm and at full term, 9 (216,439 participants) compared women born with and without LBW, and 4 (468,469 participants) compared women born with and without SGA status. The remaining 3 studies (1,760 participants) only compared the mean birth weight of women with or without GDM. Figure S1 presents a summary of study quality assessment using the QUIPS tool37. The overall quality of the included studies was moderate to low, mainly due to uncontrolled confounders.