a. Studies are sorted by mean exposure level (highest to lowest)

b. One PFDA study is not displayed because it used a different birth weight metric. Wang et al (2016) found an association between PFDA exposure and reduced birth weight in female infants (beta=-0.14, 95% CI -0.26--0.02 z-score unit per ln unit (ng/mL) PFDA increase)

c. Regression coefficients for continuous exposure expressed as change in birth weight (g) per increase in ln-transformed exposure concentration (ng/mL) or per IQR increase in exposure concentration. Lenters et al. 2016 per 2-SD increase ln-transformed exposure recalculated (divided by 2-SD); Robledo et al. 2015 per 1-SD increase recalculated (divided by SD); and Shi et al. 2015 log10 transformed recalculated (divided by ln10). Conversion factors and calculations

d. If a study presented regression coefficients for continuous exposure with multiple exposure units, only per IQR increase (Bach et al. 2016) or ln-transformed, if available (Li et al. 2017) are displayed

e. Robledo et al. 2015 regression coefficients for paternal serum PFDA are not displayed

f. Gyllenhammar et al. 2017 results displayed with Gyllenhammar et al. 2018 (both analyzed POPUP)