Table 1.
Characteristics of studies included in the systematic review and meta-analysis.
| Sources (study design) | Location | Study population | Arsenic concentration | Outcome studied | Confounders adjusted for | Total score on NOS | ||
|---|---|---|---|---|---|---|---|---|
| Marker for exposure | Exposure contrast | Range, median, or mean | ||||||
| Fei et al. 2013a,b (PCO) | New Hamsphire (USA) | 133 pregnant women | Arsenic levels in urine | NA | Not reported | Birth weight | Infant sex, maternal age, gestational age | 7/9 |
| Guan et al. 2012b,c (CS) | Dalian (China) | 125 mother–infant pairs | Arsenic levels in maternal and cord blood | Arsenic-affected area (590 μg/L) vs. arsenic-free area | Not reported | Birth weight | Maternal age, BMI, parity, gestational age at delivery, maternal education, maternal secondhand smoke exposure, infant sex | 5/9 |
| Cherry et al. 2010c,d (RCO) | Gonoshasthaya Kendra villages (Bangladesh) | 934 infant mortality occurring in designated area, 2001–2003 | Arsenic levels in tube-well water | ≥ 50 μg/L vs. < 10 μg/L | 0.05–166 μg/L | Infant mortality | First pregnancies, others with no formal education, mothers designated as destitute | 7/9 |
| Myers et al. 2010c,d (RCO) | Bayingnormen (Mongolia, China) | 9,890 singleton deliveries of mothers | Arsenic levels in tube-well water | > 50 μg/L vs. ≤ 50 μg/L | UD–1,200 μg/L | Birth weight, preterm delivery, stillbirth, and neonatal mortality | Maternal age, gravidity, infant sex for the analysis of birth weight and maternal age, gravidity, infant sex adequacy for the analysis of preterm delivery, stillbirth, and neonatal mortality | 7/9 |
| Rahman et al. 2010b,c (PCO) | Matlab district (Bangladesh) | 2,924 pregnant women | Arsenic levels in urine | 249–1,253 μg/L vs. < 33 μg/L (spontaneous abortion)268–2,019 μg/L vs. < 38 μg/L (stillbirth)268–2,019 vs. < 38 μg/L (infant mortality) | UD–1,253 μg/L | Spontaneous abortion, stillbirth, infant mortality | No significant confounder was found | 7/9 |
| Rahman et al. 2009b,c (PCO) | Matlab (Bangladesh) | 1,578 women with single births | Arsenic concentrations in urine | ≥ 100 μg/L vs. < 100 μg/L | 6–978 μg/L | Birth weight | Asset score, BMI, height, age, education, season, gestational age at birth, sex of infant | 8/9 |
| Cherry et al. 2008c,d (RCO) | Villages in 13 subdistricts (Bangladesh) | 30,984 pregnancies and outcomes | Average arsenic concentrations in hand-pumped well water | ≥ 50 μg/L vs. < 0.10 μg/L | UD–81 μg/L | Stillbirth | Age, sex, previous pregnancy, previous stillbirth, low socioeconomic status, maternal education, paternal education, maternal smoking, mother high BP, mother edema, gestational age, birth weight, home delivery | 8/9 |
| Sen and Chaudhuri 2008c,d,e (CS) | Villages located in North 24 Parganas district (states of West Bengal) | Pregnancy outcomes of 240 married women | Arsenic levels in tube-well water | 600 μg/L vs. < 10 μg/L | 10–600 μg/L | Spontaneous abortion and stillbirth | None | 2/9 |
| Huyck et al. 2007b,c (PCO) | 42 villages in Sirajdikhan Upakila of Munshigani district (Bangladesh) | 49 women ≥ 18 years of age | Arsenic levels in maternal hair at first visit | ≥ 2.70 μg/g vs. < 0.28 μg/g | 0.14–3.28 μg/g | Birth weight | Gestational age at first prenatal visit, maternal weight gain, birth gestational age, and activity level during pregnancy | 7/9 |
| Rahman et al. 2007c,d (PCO) | Matlab (Bangladesh) | 29,134 pregnancies identified by the HDSS in 1991–2000 | Arsenic levels in tube-well water | ≥ 409 μg/L vs. < 10 μg/L | Median, 224 μg/L | Fetal loss, infant mortality, neonatal | Age, parity, education, and socioeconomic status | 7/9 |
| Ahamed et al. 2006a, 2006bc,d,e (CS) | Eruani village (Bangladesh) | 56 pregnancy outcomes in women of reproductive age | Arsenic levels in tube-well water | Exposed area (501–1,200 μg/L) vs. control area | 501–1,200 μg/L | Spontaneous abortion and stillbirth | None | 1/9 |
| von Ehrenstein et al. 2006c,d (CS) | 21 villages in West Bengali (south 24-Parganas district) (India) | 202 married women, 20–40 years of age | Arsenic levels in tube-well water | ≥ 200 μg/L vs. < 50 μg/L | Mean = 101.7 μg/L | Spontaneous abortion, stillbirth, neonatal mortality, infant mortality | Mother’s age at child’s birth, BMI, maternal education, education of the head of the household, and type of housing material | 3/9 |
| Milton et al. 2005c,d (CS) | 29 villages in Comilla district, 2 villages in the Chandpur district, 43 villages in the Chaudanga district (Bangladesh) | 533 ever-married women, 15–49 years of age | Arsenic levels in tube-well water | > 50 μg/L vs. ≤ 50 μg/L | UD–1,710 μg/L | Spontaneous abortion, stillbirth, and neonatal mortality | Height, history of hypertension and diabetes, and age at first pregnancy for neonatal mortality | 3/9 |
| Mukherjee et al. 2005c,d,e (CS) | Murshidabad (West Bengal, India) | 17 married women of reproductive age (18–40 years) with at least 1 pregnancy | Arsenic levels in drinking water | Exposed area (401–1,474 μg/L) vs. nonexposed area (< 3 μg/L) | 401–1,474 μg/L | Spontaneous abortion and stillbirth | None | 1/9 |
| Rahman et al. 2005c,d,e (CS) | Jalangi block (India) | 13 married women of reproductive age (18–40 years) | Arsenic levels in drinking water | Women in exposed areas (501–1,474 μg/L) vs. women in control area (< 10 μg/L) | Not reported | Spontaneous abortion and stillbirth | None | 1/9 |
| Chakraborti et al. 2003c,d,e (CS) | Semria Ojha Patti village of Ara (Bhoipur, India) | 16 women | Arsenic levels in tubes-well water | 463–1,025 μg/L vs. 7–39 μg/L | 7–1,025 μg/L | Stillbirth | None | 1/9 |
| Guo et al. 2003a,d (CS) | Villages in Wuyan County (Inner Mongolia, China) | 224 women | Arsenic levels in well water | Exposed area (43 μg/L) vs. nonexposed area (9.6 μg/L) | Not reported | Spontaneous abortion | Sex, age, smoking and alcohol consumption | 3/9 |
| Hopenhayn et al. 2003a,d (PCO) | Antofagasta and Valparaiso (Chile) | 844 singleton mothers 18–45 years of age | Arsenic levels in water | 40 μg/L vs. < 1 μg/L | 32.9–52.7 μg/L | Birth weight | Location, calendar time, arsenic exposure | 6/9 |
| Yang et al. 2003a,d (RCO) | 18 villages in 4 township in Lanyang Basin (Taiwan) | 18,259 singleton births | High arsenic–exposed community used as a surrogate | Exposed area (UD–3,590 μg/L) vs. nonexposed area | UD–3.59 ppm | Preterm delivery, birth weight | Maternal age, marital status, maternal education, infant sex | 6/9 |
| Ahmad et al. 2001c,d (CS) | Village of Samta in thana Sharsha, Jessore district; village of Katiarchar in Sadar thana, Kishorgonj district (Bangladesh) | 192 married women of reproductive age (15–49 years) | Arsenic levels in tube-well water | > 50 μg/L vs. ≤ 0.2 μg/L | 200–450 μg/L | Spontaneous abortion, stillbirth, and preterm birth | Socioeconomic status, education, and age at marriage | 3/9 |
| Hopenhayn-Rich et al. 2000c,d (RCO) | Antofagasta and Valparaiso (Chile) | Mortality of infants, 1950–1996 | Arsenic levels in public water | > 50 vs. 5 μg/L | 40–860 μg/L | Fetal mortality, neonatal mortality, | Location, calendar time, arsenic exposure | 6/9 |
| Ihrig et al. 1998c,d (C‑;C) | Bryan, TX (USA) | 119 case babies, and 267 control babies | Arsenic levels estimated from airborne emissions | > 100 vs. 0 ng/m3 | Not reported | Stillbirths | Maternal age, race/ethnicity, parity, income group, exposure as a categorical variable, and exposure–race/ethnicity interaction | 7/9 |
| Aschengrau et al. 1989 (C-C)a,d | Boston, MA (USA) | 286 cases, 1,391 controls | Arsenic levels in public drinking water | 1.4–1.9 μg/L vs. UD | UD–19 μg/L | Spontaneous abortion | Water source, maternal age, educational level, history of prior spontaneous abortion | 7/9 |
| Abbreviations: BMI, body mass index; BP, blood pressure; C-C, case–control study; CS, cross-sectional study; HDSS, health and demographic surveillance system; NA, not applicable; NOS, Newcastle–Ottawa Scale; PCO, prospective cohort study; RCO, retrospective cohort study; UD, undetected. aStudies examining low to moderate arsenic dose in the general population. bStudies examining high arsenic dose in the general population. cStudies applying biomarkers/individual-level data. dStudies applying group/ecological data. eStudies that did not control for potential confounders. | ||||||||