TABLE 15.

Key epidemiological studies relating As exposure with birth weight.

Reference study population design	Outcome definition	Population size (n)	Arsenic exposure a	Results	Additional information/confounders
Included in 2009 EFSA Opinion
Hopenhayn et al. (2003) Chile Cohort	BW	844 (442 from Antofagasta and 444 from Valparaiso)	Water: Antofagasta: 30–40 μg/L Valparaiso: < 1 μg/L	−57 g in Antofagasta (95% CI ‐123, +9)	Aggregated w‐As data Adjusted for many potential confounders, including GA
Yang et al. (2003) Taiwan Cohort	BW	18,259 (3872 from As‐exposed townships, and 14,387 from similar reference townships)	Water: In As‐exposed townships 30% > 50 μg/L	−29 g (95% CI –44, –14) in As‐exposed townships	Aggregated w‐As data. Adjusted for some potential confounders
Huyck et al. (2007) Rural Dacca, Bangladesh CS/cohort (samplings early pregnancy and around birth)	BW, LBW	49 live births	Water: max Q1: 0.5, Q2: 1.3, Q3: 9.0, Q4: 734 μg/L. Maternal hair 0.14–3.3, toenail 0.2–6.2 (both t‐As μg/g)	−194 g per μg/g As in hair No association between BW and water‐As or nail‐As OR for LBW versus hair As around 2 but CI included 1	LBW = < 2750 (=median in these newborns). Poor correlations between w‐As and most biomarkers Adjusted for a number of potential confounders including GA
Rahman et al. (2009) Matlab Bangladesh Cohort	BW	1578 singleton live births	Maternal u‐tiAs: median 95 μg/L	No overall association, but −1.7 g per μg/L in u‐tiAs at 0–100 μg/L. Mean BW 2.6 kg	Adjusted for many potential confounders. Exposure from water (70% > 10 μg/L, see Rahman et al., 2006)
Myers et al. (2010) Inner Mongolia, China CS	BW in term births	9890 singleton births 1996–1999 with As measured in the wells of the sub village	Water: 66% < 20, 20% 21–50, 7% 50–100, 7% > 100 μg/L. Data from 1991 to 1997 in well‐water registry	No association between BW and water‐As categories	w‐As data were averaged by sub village. No potential confounders were detected for BW. Women were generally health and well‐nourished
Not included in the 2009 Opinion
Kwok et al. (2006) Rural Bangladesh, 261 villages in 3 districts CS	LBW in term (≥ 37 week) births	2006 live singleton births with As measured in the wells of the villages	Water: 2003 samples, mean about 70 μg/L, six categories < 10 μg/L to > 300 μg/L	Adjusted OR 1.00 for LBW (< 2500 g) in full term births BW not analysed as continuous variable	Water samples (for each household), birth data for 2002, and covariates were collected in 2003 from Community Centres' registers + maternal interviews (2003). Most births at home
Gelmann et al. (2013) Romania CS	LBW	38 singleton live births with LBW and normal BW from As‐in‐water exposed (19) and unexposed (19) villages	Water: means in exposed 54 μg/L and in unexposed 1 μg/L. u‐iAs 2–90 μg/L in exp and 0–1 μg/L in unexposed	Among exposed those with LBW had higher u‐tiAs than those with normal BW	Pilot study with too few cases
Chou et al. (2014) Taiwan Cohort	BW, LBW	299 live births, 21 (7%) LBW (< 2500 g)	Maternal u‐tiAs. Median 22 μg/g Cr (90% DMA)	Significant adjusted association (multiple linear regression) between u‐MMA and BW. No association with LBW	The association btw u‐MMA and BW disappeared after further adjustment for DNA damage (N7‐MeG)
Laine et al. (2015) Mexico CS	BW	199 singleton live births, only 4 with LBW	Water: < 0.5–236, median 13 μg/L. Maternal u‐tiAs median 23 μg/L	w‐As, u‐tiAs, u‐DMA: null U‐MMA and U‐%MMA had a significant adjusted negative slope	Mean u‐DMA 43 μg/L in women reporting seafood and 31 μg/L in the others. The association with BW seemed to be at least partly caused by an association with GA
Bloom et al. (2016) Romania CS	BW	122 live singleton births	Water: mean 4.1 μg/L	No overall association between As and BW	Adjusted for potential confounders. An inverse association was found in the 29 smoking mothers
Gilbert‐Diamond et al. (2016) USA, New Hampshire Cohort	BW	706 live singleton births	Maternal u‐tiAs. Median 3.4 μg/L	No association between u‐tiAs, MMA or DMA and BW	General additive models adjusted for some potential confounders including gestational age
Kile, Cardenas, et al. (2016) Rural Bangladesh Cohort	BW	1153 live singleton births	Water: < 0.5–1400, median 2.3 μg/L Maternal toenail t‐As 0.2–35 μg/g, median 1.5 μg/g	Significant adjusted inverse association (multiple linear regression) between water As and BW, and between nail‐As and BW	Structural equation modelling suggested that almost all the association was mediated by an association with GA
Almberg et al. (2017) USA, Ohio CS	LBW and VLBW (< 1500 g) in term infants	428,804 live singleton births from records 2006–2008. In term births 2.9% LBW and 1.2% VLBW	W‐As aggregated by 88 counties and year: 0.5–12 μg/L, median 1.5 μg/L. Based on 2968 measurements in 975 water systems	Adjusted ORs for LBW (1.00) and VLBW (0.99) were normal	ORs adjusted for many potential confounders from the records ORs were increased in some sub analyses excluding counties with many private wells
Rahman et al. (2017) Rural Bangladesh Cohort	BW	1180 live singleton births, same study base as Kile, Cardenas, et al. (2016)	See Kile, Cardenas, et al. (2016), but median w‐As here 2.2 μg/L and nail‐As 1.2 μg/g	Associations: See Kile, Cardenas, et al. (2016)	Mediation analyses suggested that most of the association was mediated by GA, but among the smallest infants there was also a direct effect
Lin et al. (2019) Rural Bangladesh Cohort	BW	1057 live singleton births, Same study bas as Kile, Cardenas, et al. (2016) and Rahman et al. (2017)	Toenail As 0.04–47, median 1.2	BW inversely associated with protein, fat and fibre intake, but positively associated with carbohydrate intake. Nail‐As not significantly associated with BW, when adjusted for macronutrients, and nail‐As was not a significant mediator for BW	Seems to weaken the conclusions from Kile, Cardenas, et al. (2016) regarding BW versus nail‐As, where macronutrients were not adjusted for. But still association between nail‐As and GA; the latter was a mediator to BW in Kile, Cardenas, et al. (2016)
Bozack et al. (2020) Rural Bangladesh Cohort	BW	Subset of Kile, Cardenas, et al. (2016): 413 births with % methylation of DNA cord blood at 3 CpG sites in the gene body of DNA methyltransferase 3 alpha	See Kile, Cardenas, et al. (2016)	Significant inverse association between nail‐As and BW.	See Kile, Cardenas, et al. (2016), Rahman et al. (2017) and Lin et al. (2019). Here structural equation modelling also includes DNA methylation in cord blood.
Howe, Farzan, et al. (2020) USA, Los Angeles Cohort	BW	Births with maternal: 116 (hair‐As) 100 (blood As) 167 (u‐iAs)	Medians: Hair: 0.01 μg/g Blood: 0.67 μg/g u‐tiAs: 5.7 μg/L	No association with u‐tiAs	In urine additionally 0.5 μg/L AsB
Howe, Claus Henn, et al. (2020)	BW	262	Median u‐tiAs 5.8 μg/L	No significant association with BW in relation to GA	Note that then a possible impact of As on BW via GA is not taken into account
Shih et al. (2020) USA, seven areas Cohort (NCS)	BW	212	Total As (median 7.8 and DMA median 3.4 μg/L) in maternal urine	No association btw u‐DMA and BW	High LODs: u‐tiAs 1.25, DMA 1.7, MMA 0.9. All had u‐tiAs >LOD and 77% had DMA > LOD, but for other species > 60% < LOD
Bulka et al. (2022) USA nationwide Cross‐sectional	BW	3,305,090 term births in 3105 counties	The probability of As in water in four categories (≤ 1, > 1, ≤ 5, > 5, ≤ 10, > 10) was estimated in 1 × 1 km grids, based on about 20,000 As measurements in private wells. Estimates were then aggregated on county level	Adjusted beta (grams) per 10% increase in probability of exceeding the following thresholds: ≤ 1, > 1 −3.1 (−4.0, −2.2) ≤ 5, > 5 −2.8 (−4.2, −1.3) ≤ 10, > 10 −3.9 (−6.1, −1.7)	Adjusted for proportion of private wells, maternal age, race/ethnicity, marital status, education, smoking during pregnancy, pre‐pregnancy BMI, rurality/urbanicity, annual concentration of PM < 2.5 μm in the county

Abbreviations: As, arsenic; AsB, arsenobetaine; BMI, body mass index; BW, birth weight; CI, confidence interval; Cr, creatinine; CpG, cytosine‐phosphate‐guanine; CS, cross‐sectional study; DMA, sum of dimethylarsinous acid and dimethylarsinic acid; DNA, deoxyribonucleic acid; EFSA, European Food Safety Authority; GA, gestational age; iAs, inorganic arsenic; km, kilometre(s); LBW, low birth weight; LOD, limit of detection; MMA, sum of monomethylarsonous acid and monomethylarsonic acid; n, number; N7‐MeG, N7‐methylguanine; NCS, National Children's Study; OR, odds ratio; PM, particulate matter; t‐As, total arsenic; u‐DMA, urinary DMA; u‐iAs, urinary inorganic arsenic; u‐MMA, urinary MMA; USA, United States of America; u‐tiAs, total urinary iAs (sum of iAs and methylated its metabolites MMA and DMA); VLBW, very low birth weight; w‐As, water arsenic.

^a iAs in urine includes MMA and DMA, but not AsB.