Early-life cadmium exposure and child development in 5-year-old girls and boys: a cohort study in rural Bangladesh

Abstract

Background: Cadmium is a commonly occurring toxic food contaminant, but health consequences of early-life exposure are poorly understood.

Objectives: We evaluated the associations between cadmium exposure and neurobehavioral development in preschool children.

Methods: In our population-based mother-child cohort study in rural Bangladesh, we assessed cadmium exposure in 1,305 women in early pregnancy and their children at 5 years of age by measuring concentrations in urine (U-Cd), using inductively coupled plasma mass spectrometry. Children's IQ at 5 years of age, including Verbal (VIQ), Performance (PIQ), and Full-Scale IQ (FSIQ), were measured by Wechsler Preschool and Primary Scale of Intelligence. Behavior was assessed by the Strengths and Difficulties Questionnaire (SDQ).

Results: In multiple linear regression models, adjusted for sex, home stimulation, socioeconomic status (SES), and maternal and child characteristics, a doubling of maternal U-Cd was inversely associated with VIQ (-0.84 points; 95% confidence interval: -1.3, -0.40), PIQ (-0.64 points; -1.1, -0.18), and FSIQ (-0.80 points; -1.2, -0.39). Concurrent child U-Cd showed somewhat weaker association with VIQ and FSIQ, but not PIQ. Stratification by sex and SES indicated slightly stronger associations with PIQ and FSIQ in girls than in boys and in higher-income compared with lower-income families. Concurrent U-Cd was inversely associated with SDQ-prosocial behavior and positively associated with SDQ-difficult behavior, but associations were close to the null after adjustment. Quantile regression analysis showed similar associations across the whole range of each developmental outcome.

Conclusion: Early-life low-level cadmium exposure was associated with lower child intelligence scores in our study cohort. Further research in this area is warranted.

Figure 1

Associations of maternal U-Cd (A) and concurrent (children’s) U-Cd at 5 years of age (B) with FSIQ at 5 years of age. Solid lines represent Lowess-moving average curves; dashed lines represent fitted curves of FSIQ and U-Cd (log₂-tansformed), adjusted for age at testing, tester, sex, birth order, birth weight, HAZ (5 years), HOME, maternal body mass index (early pregnancy), maternal IQ, SES, and maternal or concurrent urinary arsenic (log₂-transformed). Twenty subjects are not included; 5 subjects with FSIQ < 50 points and 15 subjects with FSIQ > 100 points.

Figure 2

Multivariable-adjusted associations between (A,B) maternal U-Cd (log₂-transformed) and (C,D) concurrent U-Cd (log₂-transformed) with developmental outcomes at 5 years of age after stratification by sex (A,C) or SES (B,D). Estimates with 95% CIs represent the change in outcome score with a doubling of U-Cd exposure adjusted for age at testing, tester, birth order, birth weight, HAZ (5 years), HOME, maternal body mass index (early pregnancy), maternal IQ, maternal or concurrent U-As (log₂-transformed).

Figure 3

Multivariable-adjusted cadmium-related differences in child development (with 95% CIs), calculated by comparing the model predicted developmental measures of children with maternal (A) and concurrent (B) U-Cd and U-As ≥ 95th percentile (maternal U-Cd 2 µg/L and U-As 524 µg/L; concurrent U-Cd 0.63 µg/L and U-As 364 µg/L) to those ≤ 5th percentile (maternal U-Cd 0.18 µg/L and U-As 18 µg/L; concurrent U-Cd 0.078 µg/L and U-As 17 µg/L). Estimates with 95% CIs are adjusted for, beside cadmium and arsenic, age at testing, tester, sex, birth order, birth weight, HAZ (5 years), HOME, maternal body mass index (early pregnancy), maternal IQ, and SES.