Association between maternal exposure to mercury during pregnancy and birth weight: a systematic review and meta-analysis

Abstract

Background

Epidemiological research has suggested that exposure to mercury in pregnancy may negatively affect birth weight (BW). However, the results were inconclusive. This meta-analysis aimed to provide a quantitative summary of evidence for the relation of prenatal mercury exposure to BW.

Methods

We performed a systematic search in the PubMed and Scopus databases until March 2024. The pooled effects of prenatal mercury exposure on birth weight (BW) were assessed using a random-effects model, using the standardized regression coefficients (β) along with their corresponding 95% confidence intervals (CI). Stratified analysis by type of specimen, sample size, method of mercury assessment, Hg concentration, and trimester of sampling was conducted to investigate possible sources of heterogeneity.

Results

Forty-one studies, including 128,487 participants, were analyzed. In the overall analysis, no significant relationship was revealed between prenatal mercury exposure and BW (β= -0.002, 95%CI: -0.003 to 0.0001; P = 0.06) with substantial heterogeneity (I² = 63.0, P = 0.001). However, in the stratified analysis, exposure to mercury was inversely linked to the neonatal BW in studies on placental exposure (β= -0.144, 95%CI: -0.272 to -0.016; P = 0.02) and exposure at delivery (β= -0.010, 95%CI: -0.020 to -0.002; P = 0.01) and at the third trimester (β= -0.0003, 95%CI: -0.0005 to -0.0001; P = 0.004) of pregnancy. Mercury was also negatively associated with BW in studies that measured mercury using atomic absorption spectroscopy (β= -0.010, 95%CI: -0.020 to -0.001; P = 0.02). Furthermore, blood Hg levels ≥ 2.09 µg/L (β = -0.029, 95% CI: -0.052 to -0.006; P = 0.01) and placental Hg levels ≥ 10 µg/kg (β = -0.193, 95% CI: -0.293 to -0.094; P = 0.001) were significantly associated with lower BW.

Conclusion

This meta-analysis revealed that mercury exposure may be negatively associated with birth weight, especially when higher concentrations are present in the blood and placenta, as well as during the late stages of pregnancy, which significantly correlates with lower neonatal BW.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12887-025-06323-y.

Background

Low birth weight (LBW) is identified as a significant contributor to under-five mortality. LBW affects 15.5% of all neonates worldwide, of which 96% are from less-developed countries [1]. LBW can result from prematurity, intrauterine growth restriction, or a combination of both factors. LBW is linked to many unfavorable health consequences, such as impaired growth and cognitive development, increased odds of neonatal mortality, and higher susceptibility to non-communicable diseases later in life [2]. The etiology of LBW is not clearly understood, but socioeconomic and demographic factors, genetics, inadequacies in prenatal care, complications during pregnancy, and intake of illicit drugs, caffeine, alcohol, and tobacco are linked to the risk of LBW [3, 4]. In addition, exposure to heavy metals during pregnancy has been linked to adverse birth outcomes, primarily due to the well-established toxicity of heavy metals [5].

Mercury (Hg) is an extremely toxic environmental contaminant and is recognized as one of the top 10 chemicals of major public health concern according to the World Health Organization (WHO) [6]. Coal-burning power plants and the chemical industry are identified as major sources of Hg emissions [7]. Once released into the atmosphere, Hg can eventually find its way into water supplies where it undergoes conversion to methylmercury through microbial processes, leading to widespread contamination of aquatic ecosystems [8]. The main way humans are exposed to methylmercury is by consuming contaminated fish and shellfish[9].. Skin-lightening creams and dental amalgams are other significant sources of inorganic and elemental Hg, respectively [10]. Hg can have toxic effects on the skin and eyes as well as the immune, digestive, nervous, and respiratory systems. During the process of organogenesis, the developing fetus is particularly vulnerable to Hg exposure. This is attributed to the capacity of Hg to easily pass through both the placenta and the blood-brain barrier [7], which results in unfavorable impacts on intrauterine growth and neurodevelopment of the fetus [9]. Nevertheless, the relationship between maternal Hg exposure and birth weight (BW), as the continuous measurement of neonatal weight at birth, is not well-identified. Epidemiological studies examining the relationship between maternal Hg exposure and offspring BW have produced inconsistent results. Arinola et al. followed sixty-eight women during their first trimester of pregnancy through to delivery and found that higher maternal blood Hg levels were significantly associated with reduced birth weight [11]. A cross-sectional study of 1,578 women aged 16 to 50 in Saudi Arabia revealed that higher Hg concentrations in the maternal placenta were linked to lower neonatal birth weight, while Hg levels in umbilical cord blood showed no significant correlation with birth weight [12]. Moreover, a cross-sectional study by Ou et al. found that elevated urinary Hg levels during pregnancy were linked to decreased birth weight [13]. In contrast, several studies have reported no significant association between Hg levels in cord blood [14], hair [15], whole blood [16], and urine [17] with the birth weight of newborns.

The contradictory evidence of studies could result from the heterogeneity in factors such as sample size, method of Hg assessment, the type of specimen used, and the trimester during which the samples were collected. Moreover, maternal age is a known factor influencing BW [18] and may modify the effect of Hg exposure on fetal growth, thus may contribute to the heterogeneity across the studies. Therefore, this meta-analysis was conducted to assess the relationship between maternal Hg exposure and BW of neonates.

Methods

This review was reported following the guidelines outlined by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [19].

Search strategy

The PubMed and Scopus databases were systematically searched for all studies published before March 2024, using terms related to Hg and BW. The search strategy is presented in Table S1. The search for relevant studies was limited to those published in the English language. Furthermore, we screened the reference lists of relevant studies to identify any additional publications that could be included. The identified studies were first imported into EndNote 7 software, where duplicates were removed. The screening process involved two stages: initially, two independent reviewers examined the titles and abstracts to exclude clearly irrelevant studies; subsequently, the same reviewers evaluated the full texts of the remaining articles against the predefined inclusion criteria. Any disagreements were resolved through discussion or, if necessary, consultation with a third reviewer.

Eligibility criteria

Two separate researchers evaluated the eligibility of studies and disagreements were solved by a discussion among all authors. Studies with the following criteria were included: (1) participants were pregnant women and their infants, (2) the exposure was Hg in biological samples of pregnant women and the outcome was BW of infants as the continuous measurement of neonatal weight at birth, (3) studies with case-control, cohort, and cross-sectional design, and (4) provided standardized beta coefficients for linear regression models (β) and their 95% confidence intervals (CI) or sufficient information to compute this data. We excluded studies with irrelevant exposure/outcome, dietary and environmental exposure to Hg, reviews, animal studies, letters, editorials, studies with overlapped subjects, and studies with incomplete data.

Data extraction and quality assessment

The following information was collected from the selected studies: publication year, study location, the author’s name, sample size, type of specimen, method of statistical analysis, variables adjusted for in the analysis, method of Hg assessment, maternal age, trimester of exposure, and the reported effect estimates (β and 95% CI). Some studies assessed Hg exposure using different biological samples, such as urine, amniotic fluid, erythrocytes, hair, blood, placenta, and cord blood, as well as at various stages of pregnancy, including the first, second, and third trimesters, and at delivery. Given that both the type of biological sample and the timing of exposure are among the most significant sources of heterogeneity, performing subgroup analyses based on these factors is essential. Therefore, we extracted all effect sizes accordingly to enable subgroup analyses by sample type and timing of exposure.

The quality of the eligible publications was assessed with the use of a modified version of the New castle Ottawa Scale (NOS) developed for cross-sectional studies, which assigns a maximum score of 9. Studies that scored ≥ 7 and ≥ 3 to 7 were considered to be of high and moderate quality, respectively [20]. The NOS for cross-sectional studies focuses on three main domains: selection (4 points), comparability (2 points), and outcome (3 points). Each criterion was applied based on predefined standards consistent with NOS guidelines to ensure an objective and transparent risk of bias evaluation. The selection domain includes the representativeness of the sample (random sampling or inclusion of all eligible pregnant women from the target population), sample size (≥ 200 subjects), and response rate (≥ 95% or demonstrated comparability between respondents and non-respondents in key characteristics such as age and body mass index (BMI)). The comparability domain examines whether the study controls for potential confounding factors (age and BMI) as well as any additional covariates. The exposure and outcome domain evaluates the use of validated measurement tools for assessing Hg concentrations (using the same laboratory tests for all subjects) and birth weight (medical records or standardized measurements), along with the appropriateness of statistical analysis (regression models adjusting for confounders, reporting effect sizes and confidence intervals).Two independent reviewers extracted the data. In case of any disagreement, discussions were held to resolve them. During the data extraction and quality assessment process, disagreements between the two independent reviewers primarily involved differences in interpreting study eligibility, extracting specific numerical data such as effect sizes or confidence intervals, and rating certain Newcastle-Ottawa Scale items, particularly those requiring subjective judgment like comparability and exposure assessment. In total, disagreements occurred in approximately 14% of the studies reviewed (n = 6 studies) which were resolved through discussion and consensus. When consensus could not be reached, a third reviewer was consulted to make the final decision.

Statistical analysis

Standardized regression coefficients (β) with corresponding 95%CIs were used as effect size for the relationship between maternal Hg exposure and BW of infants. The included studies reported effect sizes using different analytical approaches (standardized regression coefficients, unstandardized regression coefficients, spearman’s correlations) which required some pre-processing to standardize the magnitude of the observed effects in each study. Standardized regression coefficients (β) were extracted directly if reported in the original studies. If Pearson or Spearman correlation analysis was done in a study, the correlation coefficient r was considered as β, and its standard error (SE) was calculated based on the following relationship:. If an unstandardized regression coefficient was reported in the study, the regression coefficient was converted to the β and the corresponding SE using the following formulas: b and SE (b), where SD(X), SD(Y), “b” and SE(b) demonstrated the standard deviation of exposure variable, the standard deviation of response variable, the unstandardized regression coefficient, and standard error for b. This procedure was done to homogenize the effect sizes before performing the meta-analysis. The heterogeneity across publications was tested using the Cochran’s Q-test and I[2] index [21, 22]. I² values of < 25% indicated low heterogeneity, 25–50% indicated moderate heterogeneity, 50–75% indicated substantial heterogeneity, and 75–100% indicated extreme heterogeneity [23]. Due to the expected heterogeneity among the included studies, the effect sizes were combined using a random effects model. Stratified analysis by type of specimen, sample size, method of Hg assessment, and trimester of sampling was conducted to investigate possible sources of heterogeneity. Furthermore, subgroup analysis was conducted based on the median of Hg concentrations in different biological samples, including cord blood (˂ 4.4 vs. ≥ 4.4 µg/L), blood (˂ 2.09 vs. ≥ 2.09 µg/L), placenta (˂ 10 vs. ≥ 10 µg/kg), hair (˂ 0.58 vs. ≥ 0.58 µg/g), and erythrocytes (˂ 2.14 vs. ≥ 2.14 µg/kg) to examine the effect of Hg concentration on the pooled effect sizes. To investigate the influence of maternal age on the relationship between Hg exposure and birth weight, a meta-regression analysis was performed based on the age of participants. The meta-regression was performed by including the mean maternal age reported in each study as a continuous covariate in the meta-regression model to assess whether variations in maternal age across studies explained heterogeneity in the Hg–birth weight association. Additionally, a sensitivity analysis was conducted to assess the stability of the pooled results by excluding individual studies and examining the impact on the overall findings. Publication bias was evaluated using Egger’s test, where a p-value < 0.05 indicated significant publication bias [24, 25]. All statistical analyses were conducted using Stata 13.0 software (StataCorp LP, College Station, TX).

Results

Study selection and characteristics

The systematic search conducted in databases yielded a total of 1,756 publications. After removing duplicates (426 in total) and publications that were not relevant based on their titles and abstracts (1,262 in total), 68 articles remained for full-text screening. Out of these, 27 more studies with specific reasons presented in Fig. 1 were excluded. Ultimately, the meta-analysis included a total of 41 unique studies with 54 effect sizes [1, 6–9, 11, 13–17, 26–55], encompassing a combined sample size of 128,487 participants. In cases where studies reported different effect sizes in their stratified analyses, we collected all available effect sizes for those studies. The studies were published between 2010 and 2023. The mean age of subjects was between 20 ± 20.74 and 34.5 ± 4.8 years. The type of sample was cord blood in 15 studies (16 effect sizes) [7–9, 14, 15, 17, 28, 29, 31, 32, 35, 39, 44, 48, 54], whole blood in 17 studies [7, 11, 13, 16, 31, 34, 36, 37, 41–43, 45–47, 51, 52, 55], placenta in 5 studies [9, 13, 29, 38, 40], hair in 6 studies [1, 15, 26, 27, 30, 33], erythrocytes in 4 studies [6, 31, 49, 53], urine in 3 studies [13, 17, 50], and amniotic fluid in one study [29]. Moreover, 21 studies (28 effect sizes) measured Hg concentrations using atomic absorption spectroscopy (AAS) and 21 studies (24 effect sizes) used inductively coupled plasma-mass-spectrometry (ICP-MS), while the method of Hg assessment was not reported in 2 studies. Additionally, samples were obtained at delivery in 20 studies (26 effect sizes) [7, 8, 11, 13, 14, 17, 28, 30–33, 35, 38–40, 44, 48, 51, 52, 54], at the third trimester in 10 studies (14 effect sizes) [1, 6, 26, 27, 29, 37, 43, 46, 47, 53], at the second/third trimesters in one study [15], at the second trimester in 6 studies (7 effect sizes) [9, 45–47, 50, 55], and at the first trimester in 3 studies [34, 41, 49]; time of sampling was not reported in 3 studies [16, 36, 42]. The studies by Lee et al. [47] and Ashrap et al. [46] reported effect sizes for the second and third trimesters separately; both effect sizes were extracted for these studies. According to the NOS criteria, the quality of the included studies was assessed as moderate to high, with scores ranging from 4 to 9 (Table S2). Further details and characteristics of the included studies are presented in Table 1.

Fig. 1.

Flow diagram of the study

Table 1.

Characteristics of eligible studies

Author	Year	Country	Study design	Sample size	Type of sample	Mercury assessment	Mercury concentration (mean ± SD)	Maternal age (mean ± SD)	Sampling time	Adjustment
Lee	2023	South Korea	Prospective cohort	344	Cord blood	AAS	7.35 µg/L (range: 1.81–28.2)	33.5 ± 12.I4	At delivery	Newborn’s sex, living area, maternal age group, gestation day, delivery experience, maternal body mass index (BMI), alcohol consumption, and average exposure time to secondhand smoke during pregnancy
Takatani	2022	Japan	Prospective cohort	93,739	Blood	ICP-MS	4.2 ± 2.49 µg/L	31.3 ± 4.9	At delivery	Maternal age, pre-pregnancy BMI, alcohol consumption and smoking status, income, education, gestational age, sex, and parity
Zhao	2023	China	Prospective cohort	292	Blood	ICP-MS	0.69 ± 0.29 µg/L	28.0 ± 2.9	Second trimester	Maternal age, the pre-pregnancy BMI, passive-smoking history, weight gain during pregnancy, neonatal sex, maternal education and annual family income
Kobayashia	2019	Japan	Prospective cohort	15,444	Blood	ICP-MS	3.66 (IQR: 2.59–5.18) µg/L	30.9 ± 4.9	Second trimester	Maternal age, BMI before pregnancy, parity, smoking during pregnancy, drinking during pregnancy, education level, annual household income, pregnancy-induced hypertension, gestational diabetes mellitus, vaginal delivery/cesarean section, infant gender, gestational age, and selenium levels
Al-Saleh	2014	Saudi Arabia	Cross-sectional	250	Cord blood Placenta	AAS	Cord blood: 2.876 µg/L Placenta: 31 µg/kg	28.7 ± 5.8	Second trimester	Newborn’s gender and third trimester BMI
Drouillet-Pinard	2010	France	Prospective cohort	151	Hair	AAS	0.52 ± 2.6 µg/g	28.7	Third trimester	Adjusted for centre, maternal age and height, smoking during pregnancy, parity, gestational length, delay between birth and anthropometric measures, and newborn’s sex
Ding	2013	China	Prospective cohort	258	Blood Cord blood	AAS	Blood: 0.91 ± 0.37 µg/L Cord blood: 1.60 ± 0.72 µg/L	27.8 ± 4.8	At delivery	Gestational age, parity, infant sex, pre-pregnancy BMI, weight gain during pregnancy, maternal age, household monthly income, and smoking during pregnancy
Eguchi	2019	Japan	Prospective cohort	62	Blood	ICP-MS	1.0 ± 0.57 µg/L	33.3 ± 3.38	Third trimester	Age, BMI, infants gender, and gestational age
Hadjer	2020	Algeria	Prospective cohort	10	Cord blood	ICP-MS	1.25 ± 0.33 µg/L	NR	At delivery	Crude
Arinola	2018	Nigeria	Cross-sectional	68	Blood	AAS	7.98 ± 0.93 µg/L	29.48 ± 4.5	At delivery	Family income, maternal age, and education level
Baldewsingh	2020	Suriname	Prospective cohort	198	Hair	AAS	0.63 µg/g (IQR 0.36–1.09)	26.3 ± 7.26	Third trimester	Maternal age, ethnic background, household income and village location
Chen	2014	USA	Prospective cohort	50	Blood Cord blood Erythrocytes	ICP-MS	Blood: 0.32 µg/L (95%CI: 0.24–0.42) Cord blood: 0.21 µg/L (95%CI: 0.16–0.29) Erythrocytes: 2.35 µg/kg (95%CI: 1.82–3.03)	27.6 ± 6.4	At delivery	Crude
Gundacker	2010	Austria	Prospective cohort	53	Hair	AAS	0.18 µg/g (IQR: 0.11–0.42)	30 ± 6.5	Third trimester	Crude
García-Esquinas	2013	Spain	Cross-sectional	140	Cord blood	AAS	6.72 µg/L (95%CI: 5.74–7.87)	31.1	At delivery	Newborn’s sex, gestational age and maternal age, and maternal fish consumption
Guo	2013	China	Prospective cohort	213	Cord blood	AAS	1.85 ± 1.16 µg/L	26.4 ± 4.3	At delivery	Gestational age, parity, infant gender, pre-pregnancy BMI, weight gain during pregnancy, maternal age, household income, smoking during pregnancy, second-hand smoke exposure during pregnancy
Gustin	2020	Sweden	Prospective cohort	581	Erythrocytes	ICP-MS	1.8 ± 1.3 µg/kg	31 ± 4.7	Third trimester	Gestational age, infant sex, maternal early-pregnancy BMI, maternal education, pre-pregnancy smoking
Kozikowska	2013	Poland	Cross-sectional	40	Cord blood Placenta Umbilical Cord Amniotic fluid	AAS	Cord blood: NR Placenta: 10 µg/kg (range: 4–104) Umbilical Cord: 8 µg/kg (range: 3–64) Amniotic fluid: 0.5 µg/L (range: 2–11)	19–38	Third trimester	Crude
Kosik-Bogacka	2018	Poland	Cross sectional	91	Placenta	AAS	10 ± 10 µg/kg	18–45	At delivery	Crude
Kim	2017	Korea	Prospective cohort	1147	Blood	AAS	3.43 ± 1.85 µg/L	18–45	Third trimester	Gestational age at birth, maternal age, maternal education, infant sex, BMI at pregnancy
Lee	2019	Korea	Cross-sectional	466	Blood	NR	4.47 ± 1.97 µg/L	29.6	Second trimester	Crude
							4.47 ± 1.97 µg/L		Third trimester
Marques	2013	Brazil	Prospective cohort	1433	Hair	AAS	5.36 µg/g (range:0.73–24.14)	20 ± 20.74	At delivery	Gestation age, maternal age, number of pregnancies, income, mother education
Bashore	2014	USA	Prospective cohort	191	Cord blood Urine	ICP-MS	Cord blood: 1.49 µg/L (IQR:0.9–2.64) Urine: 0.35 µg/g Creatinine (IQR:0.11–0.78)	18–45	At delivery	Age, education attainment, racial/ethnic group, and living with partner/spouse
Bank-Nielsen	2019	Greenland	Prospective cohort	509	Blood	ICP-MS	5.6 ± 6.6 µg/L	27	NR	Age, BMI, alcohol during pregnancy, cotinine, parity, n-3/n-6 ratio, and region
Ashrap	2020	Puerto Rico	Prospective cohort	556	Blood	ICP-MS	1.2 ± 1.7 µg/L	26.7	Third trimester	Maternal age, maternal education, pre-pregnancy BMI, and exposure to secondhand smoking
							1.2 ± 1.7 µg/L		Second trimester
Ou	2015	China	Prospective cohort	50	Placenta Blood Urine	AAS	Placenta: 27.95 ± 10.88 µg/kg Blood: 2.36 ± 0.94 µg/L Urine: 0.76 ± 0.64 µg/g creatinine	29.2 ± 3.2	At delivery	Residential location, maternal age, gestational length, parity, infant sex, maternal after-delivery anthropometry, paternal and parental anthropometry
Tang	2016	China	Cross-sectional	103	Cord blood	ICP-MS	23.41 ± 13.41 µg/L	27.04 ± 2.70	At delivery	Adjusted by maternal BMI, maternal age, education, newborn gender, number of abortions, parity, and weight gain
Serme-Gbedo	2016	Canada	Prospective cohort	349	Blood	AAS	0.62 µg/L (IQR: 0.38–1.01)	28.8 ± 4.4	First trimester	Crude
Shih	2021	USA	Prospective cohort	125	Blood	ICP-MS	0.58 µg/L (Range: 0.11–5.32)	NR	NR	adjusted for maternal age, race/ethnicity, education, income, smoking status during pregnancy, number of prior livebirths, BMI, and infant sex
Ricketts	2017	Jamaica	Cross sectional	207	Placenta	AAS	0.64 ± 0.5 µg/kg	29 ± 6	At delivery	Crude
Valvia	2017	Faroe Islands	Prospective cohort	604	Hair Cord blood	AAS	Hair: 2.21 µg/g (IQR: 1.30, 4.03) Cord blood: 11.9 µg/L (IQR: 7.2, 20.9)	29.2 ± 5.2	Second and third trimesters	Maternal age at delivery, education, parity, pre-pregnancy BMI, smoking during pregnancy and child sex
Vigeh	2018	Japan	Prospective cohort	334	Blood	AAS	6.06 ± 3.81 µg/L	34.5 ± 4.8	First trimester	Gestational age
Wells	2015	USA	Cross-sectional	271	Cord blood	ICP-MS	0.94 µg/L (95% CI: 0.84, 1.07)	25.7 ± 6.29	At delivery	Gestational age, demographic/medical history, infant sex, maternal age, primiparity, prepregnancy BMI, maternal race, maternal smoking, maternal pregestational and gestational hypertension, and maternal pregestational and gestational diabetes, cord serum selenium
Xu	2022	Argentina	Cross-sectional	596	Blood	ICP-MS	0.35 µg/L (95%CI: 0.32–0.38)	28.8 ± 6.6	At delivery	Maternal age, parity, pre-pregnancy BMI, smoking, education and gestational age
Taylor	2016	UK	Cross-sectional	4044	Blood	ICP-MS	2.09 ± 0.99 µg/L	28.0 ± 4.9	NR	Maternal educational attainment, age, parity, height, BMI, sex of baby, gestational age at delivery, smoking, alcohol consumption, maternal selenium
Tatsuta	2017	Japan	Cross-sectional	489	Cord blood	AAS	10.2 µg/L (IQR: 4.5–23.8)	31.5 ± 4.3	At delivery	Gestational age, parity, BMI before pregnancy, smoking/drinking habits
Rahman	2021	USA	Prospective cohort	1391	Erythrocytes	ICP-MS	2.92 µg/kg (95%CI: 2.74–3.11)	32.3 ± 4.7	First trimester	Gestational age at delivery, maternal age, education, pre-pregnancy BMI, parity, smoking status, race/ethnicity, and household income, infant sex
Rahbar	2015	Jamaica	Cross-sectional	100	Cord blood	ICP-MS	4.4 ± 2.4 µg/L	29.6 ± 6.75	At delivery	Socioeconomic status and maternal education
Howe	2022	USA	Prospective cohort	1002	Urine	ICP-MS	1.22 ± 1.94 µg/g creatinine	28 ± 6.75	Second trimester	Maternal age, pre-pregnancy BMI, maternal education, parity, gestational age at urine collection, in utero tobacco smoke exposure, and cohort
Lozano	2019	Spain	Cross-sectional	1249	Cord blood	AAS	11.20 ± 8.91 µg/L	30.7	At delivery	Crude
Gustin	2023	Sweden	Prospective cohort	515	Erythrocytes	ICP-MS	1.5 µg/kg (range: 0.01–11)	30 ± 6.5	Third trimester	Parity, maternal education, pre-pregnancy smoking, gestational age at birth, and infant sex, and parity, maternal education, and pre-pregnancy smoking
Govarts	2016	Belgium	Cross-sectional	230	Hair	ICP-MS	0.255 µg/g (95%CI: 0.230–0.283)	NR	At delivery	Gestational age, child’s sex, smoking of the mother during pregnancy, parity and maternal prepregnancy BMI

NR not reported, AAS atomic absorption spectroscopy, ICP-MS inductively coupled plasma-mass-spectrometry, BMI body mass index, IQR intra-quartile range, CI confidence interval

Meta-analysis

Pooling all the effect sizes using the random effects model did not reveal a significant association between prenatal Hg exposure and BW (β= −0.002, 95%CI: −0.003 to 0.0001; P = 0.063) with a significant heterogeneity (I² = 63.0, P = 0.001) (Fig. 2). In the stratified analysis by sample size, sample type, sampling time, and the method of Hg assessment, exposure to Hg was inversely linked to the neonatal BW in placental exposure (β= −0.144, 95%CI: −0.272 to −0.016; P = 0.024; I² = 81.6, P for heterogeneity = 0.001), exposure at delivery (β= −0.010, 95%CI: −0.020 to −0.002; P = 0.015; I² = 73.0, P for heterogeneity = 0.001) and at the third trimester (β= −0.0003, 95%CI: −0.0005 to −0.0001; P = 0.004; I² = 0.0, P for heterogeneity = 0.706) of pregnancy, as well as in studies that measured Hg using the AAS (β= −0.010, 95%CI: −0.020 to −0.001; P = 0.022; I² = 57.1, P for heterogeneity = 0.001). When subgroup analysis was conducted based on Hg concentrations, it was found that blood Hg levels ≥ 2.09 µg/L (β = −0.029, 95% CI: −0.052 to −0.006; P = 0.012; I² = 70.3, P for heterogeneity = 0.001) and placental Hg levels ≥ 10 µg/kg (β = −0.193, 95% CI: −0.293 to −0.094; P = 0.001; I² = 0.0, P for heterogeneity = 0.672) were significantly associated with lower BW(Table 2).

Fig. 2.

Overall meta-analysis of the association between mercury exposure in different biological samples during pregnancy and birth weight

Table 2.

Subgroup analysis for the association of mercury exposure during pregnancy and birth weight

		Test of association		Test of heterogeneity
Subgroups	Effect sizes included	β (95%CI)	P	I2	P
Overall	54	−0.002 (−0.003 to 0.0001)	0.063	63.0	0.001
Sample size
˂500 participants	39	−0.003 (−0.003 to 0.0001)	0.063	69.3	0.001
≥ 500 participants	15	−0.005 (−0.020 to 0.010)	0.072	27.7	0.152
Sample type and Hg concentration
Cord blood (Overall)	16	−0.010 (−0.044 to 0.024)	0.561	47.1	0.016
˂ 4.4 µg/L	7	−0.002 (−0.048 to 0.043)	0.920	65.7	0.008
≥ 4.4 µg/L	8	−0.021 (−0.086 to 0.044)	0.523	27.9	0.202
NR	1	−0.022 (−0.332 to 0.288)	0.881	-	-
Blood (Overall)	17	−0.038 (−0.145 to 0.069)	0.062	73.0	0.001
˂ 2.09 µg/L	8	−0.025 (−0.050 to 0.001)	0.484	77.0	0.001
≥ 2.09 µg/L	9	−0.029 (−0.052 to −0.006)	0.012	70.3	0.001
Placenta (Overall)	5	−0.144 (−0.272 to − 0.016)	0.024	81.6	0.001
˂ 10 µg/kg	2	−0.098 (−0.293 to 0.096)	0.320	88.0	0.004
≥ 10 µg/kg	3	−0.193 (−0.293 to − 0.094)	0.001	0.0	0.672
Hair (Overall)	6	−0.003 (−0.020 to 0.012)	0.682	0.0	0.570
˂ 0.58 µg/g	3	−0.005 (−0.020 to 0.010)	0.527	0.0	0.532
≥ 0.58 µg/g	3	0.033 (−0.034 to 0.100)	0.334	0.0	0.501
Erythrocytes (Overall)	4	−0.012 (−0.083 to 0.060)	0.739	51.3	0.106
˂ 2.14 µg/kg	2	−0.0001 (−0.0001 to 0.0.001)	0.13	0.0	0.982
≥ 2.14 µg/kg	2	−0.118 (−0.448 to 0.212)	0.48	83.7	0.013
Amniotic fluid	1	−0.003 (−0.313 to 0.307)	0.982	-	-
Urine	3	−0.002 (−0.010 to 0.003)	0.371	75.5	0.012
Sampling time
At delivery	26	−0.010 (−0.020 to −0.002)	0.015	73.0	0.001
Third trimester	14	−0.0003 (−0.0005 to −0.0001)	0.004	0.0	0.706
Second/third trimesters	1	0.008 (−0.102 to 0.118)	0.882	-	-
Second trimester	7	−0.033(−0.133 to 0.068)	0.523	63.0	0.001
First trimester	3	−0.045 (−0.158 to 0.068)	0.437	74.2	0.026
NR	3	0.029 (−0.088 to 0.031)	0.340	0.0	0.869
Mercury assessment
AAS	28	−0.010 (−0.020 to −0.001)	0.022	57.1	0.001
ICP-MS	24	−0.001 (−0.002 to 0.001)	0.571	67.7	0.001
NR	2	−0.236(−0.546 to 0.073)	0.136	24.8	0.245

NR not reported, AAS atomic absorption spectroscopy, ICP-MS inductively coupled plasma-mass-spectrometry, Hg mercury

Sensitivity analysis

The sensitivity analysis showed that none of the individual studies significantly affected the pooled effect size. To assess the robustness of our results, we also conducted a sensitivity analysis by excluding studies that contributed the most weight to the meta-analysis. After removing the studies by Gustin et al. [6], Bashore et al. [17], and Ou et al. [13], which had the highest weights, the pooled effect size was β = −0.02 (95% CI: −0.03 to −0.004; P = 0.02), indicating that the findings remain consistent and reliable.

Meta-regression and publication bias

The association of Hg with BW was not affected by maternal age (B = −0.003, SE: 0.007, P = 0.70) (Fig. 3). The results of the Egger’s test indicated a significant publication bias (P = 0.005) (Fig. 4).

Fig. 3.

Meta-regression analysis for the effect of maternal age on the association between mercury exposure during pregnancy and birth weight The points/circles represent individual study estimates of the effect of age on the association between mercury level (exposure) and birth weight (outcome). The size of each point/circle corresponds to the weight of the study in the meta-analysis, with larger circles indicating studies that contribute more weight, typically due to larger sample sizes or lower variance. The circled points highlight the studies with the highest weight in the meta-analysis as key outliers

Fig. 4.

Funnel plot for publication bias

Discussion

This meta-analysis aimed to explore the relationship between Hg exposure during pregnancy and neonatal BW. The findings revealed that placental exposure to Hg may inversely be linked to neonatal BW. This negative association was observed in late pregnancy (at delivery and at the third trimester), but not in early pregnancy (at the first and second trimester). The association of Hg with BW was not affected by maternal age. This study identified that higher Hg exposure, specifically blood levels ≥ 2.09 µg/L and placental levels ≥ 10 µg/kg, is significantly associated with reduced BW. The placental Hg association was notably stronger (β = −0.193) than the blood Hg association (β = −0.029). This suggests that exceeding these Hg thresholds, particularly in the placenta, significantly impacts fetal growth. However, there was significant heterogeneity across the studies, which may limit the reliability of the findings. The levels of Hg were significantly different across the studies. Blood Hg ranged from 0.32 to 7.98 µg/L; levels ≥ 2.09 µg/L were significantly associated with reduced BW. Placental concentrations of Hg ranged from 0.64 to 31 µg/kg wet weight, with a median of 10 µg/kg. Thresholds ≥ 10 µg/kg were linked to a significant reduction in BW. Cord blood Hg varied widely (0.21–23.41 µg/L), while hair Hg ranged from 0.18 to 5.36 µg/g. Urine Hg concentrations ranged from 0.35 to 1.22 µg/g creatinine. Diffidence in the levels of Hg could play a role in the heterogeneity across the studies. The placenta accumulates both inorganic Hg (from maternal dental amalgams) and methylmercury (from fish consumption). However, heterogeneity in measurement protocols (e.g., wet vs. dry weight, sample processing) limits cross-study comparability, which may be involved in the observed heterogeneity. Standardized methods are needed to validate placental Hg as a reliable exposure biomarker. Our findings showed that placental Hg, directly interfacing with fetal circulation, had a stronger negative association with birth weight (β = −0.193) than maternal blood Hg (β = −0.029). Associations were significant at higher concentrations, suggesting that low Hg exposure may not affect birth weight. Geographic context is critical because Hg exposure varies widely depending on local environmental contamination, dietary habits, and industrial activities [56]. For example, coastal or fishing communities often report higher methylmercury levels due to fish consumption, whereas regions with artisanal gold mining or industrial pollution exhibit elevated inorganic Hg exposure [57]. These differences affect both the magnitude of Hg detected in biological media and the Hg species involved, influencing toxicity profiles. Consequently, location-specific factors significantly contribute to heterogeneity across studies, highlighting the need for tailored public health interventions based on regional exposure sources.

The findings of this meta-analysis align with previous studies that have reported a negative association between prenatal Hg exposure and BW [11, 40]. In line with these findings, other studies have also indicated a negative association between prenatal Hg exposure and child growth after birth [58]. Elevated prenatal Hg exposure was linked to reduced weight starting from 18 months of age and a lower height z-score from 5 years onwards [59]. Contrarily, several other studies have not found substantial evidence supporting an association between prenatal Hg exposure and BW [51, 55]. The discrepancies observed in the findings of previous studies could be attributed to variations in population characteristics, time of sampling, type of biological samples, and methods of measuring Hg exposure. The previous meta-analysis by Pan et al. [60] also reported a significant positive association between Hg exposure during pregnancy and the risk of LBW in infants, defined as BW < 2500 g (odds ratio = 1.079, 95% CI: 1.032–1.128). While Pan et al. [60] focused on LBW as a categorical outcome, our meta-analysis examined BW as a continuous variable and summarized the results using pooled linear regression coefficients. Additionally, the analysis by Pan et al. [60] was limited by a small number of included studies and did not explore the sources of heterogeneity across the studies. In contrast, our study incorporated a larger number of studies and conducted various subgroup analyses by Hg concentrations, time of Hg exposure, method of Hg assessment, sample size, and type of sample to identify potential origins of heterogeneity, thereby enhancing the robustness and reliability of our findings.

The findings of the present meta-analysis suggested that Hg exposure during pregnancy may be related to lower BW. The reasons why exposure to Hg in late pregnancy is inversely linked to neonatal BW, but exposure to Hg in early pregnancy is not related to BW, are not entirely clear. However, this may be related to the timing of fetal development and the accumulation of Hg in the placenta over time [61, 62]. The later stages of pregnancy are critical for fetal growth and development, and exposure to Hg during this time may have a greater impact on fetal growth and BW than exposure during earlier stages of pregnancy [63]. Moreover, Hg can accumulate in the placenta over time, and exposure to Hg in late pregnancy may result in higher levels of Hg in the placenta, which can then affect fetal growth and development [61].

Hg exposure during pregnancy occurs through fish consumption, contaminated water in areas with high levels of industrial pollution, occupational exposure in certain industries, such as mining, dentistry, and chemical manufacturing, and skin-lightening creams [9, 10, 64]. Policy- and society-level interventions to reduce Hg exposure should include enforcing stricter regulations on industrial Hg emissions and waste management to limit environmental contamination. Additionally, public health policies should improve food safety standards by monitoring and advising on Hg levels in seafood, while implementing community-wide education programs to raise awareness about Hg sources and safe consumption practices, especially for vulnerable groups such as pregnant women. These combined efforts can effectively reduce Hg exposure at the population level and protect maternal and child health. Moreover, as a precautionary measure, it is generally advised to pregnant women to avoid consuming large fish such as swordfish, tilefish, shark, king mackerel, as these species are known to have high Hg levels [65]. Instead, they should choose fish that are low in Hg, such as shrimp, salmon, and catfish, and limit their consumption to 8–12 ounces per week [66]. Pregnant women who work in industries with potential Hg exposure should talk to their employer about safety precautions and consider switching to a different position or task during pregnancy. Also, providing safe water [67] and recommending to avoid skin lightening creams [68] during pregnancy are among the strategies to decrease the burden of Hg exposure in pregnant women.

The negative association between Hg exposure during pregnancy and BW can be explained by the molecular mechanisms of Hg toxicity. Hg is a toxic heavy metal that can cross the placenta and accumulate in the fetal blood circulation [7]. It can interfere with fetal growth and development by disrupting various cellular processes, including DNA synthesis, cell division, and protein synthesis [54]. Hg can also cause oxidative stress in the placenta and fetal tissues, which can damage cells and tissues and lead to inflammation and damage to DNA, proteins, and lipids [9]. This oxidative stress can impair the function of various cellular pathways involved in fetal growth and development [9, 69]. Furthermore, Hg can interfere with the function of essential enzymes and proteins that are involved in the regulation of fetal growth [70]. For example, it can inhibit the activity of enzymes that are responsible for the metabolism of essential nutrients, such as folate, which is crucial for fetal development [71]. Additionally, Hg can disrupt the endocrine system, affecting the production and function of hormones that play a key role in regulating fetal growth. In this line, Hg can affect the function of the thyroid gland, which plays a critical role in fetal growth and development [72]. Hg can inhibit the uptake of iodine by the thyroid gland, which can lead to hypothyroidism and impaired fetal growth [73, 74]. Hg can disrupt the structure and function of the placenta, which is responsible for delivering nutrients and oxygen to the developing fetus, leading to restricted fetal growth and lower BW [75]. Hg accumulates in the placenta, and on average, the placental Hg concentration has been reported to be more than twice the maternal blood concentration [62]. These mechanisms may justify the results of the present meta-analysis showing the inverse association of placental Hg and BW.

The study’s strengths encompass its relatively high number of the included studies, high sample size, and the implementation of subgroup analysis, meta-regression analysis, and sensitivity analysis to identify potential sources of heterogeneity. Nevertheless, some limitations of the current meta-analysis should be acknowledged. First, a significant heterogeneity was detected across the studies, which complicates the interpretation of pooled results. To better characterize heterogeneity, we used a random effects method because it accounts for variability both within and between studies, providing a more conservative and generalizable estimate. The sensitivity analysis showed that none of the individual studies significantly affected the pooled effect size. It also demonstrated consistent results after excluding high-weight studies, showing the reliability of the findings. Subgroup analysis revealed that the differences in the time of sampling, type of sample, method of Hg assessment, Hg concentration, and sample size of the studies were the sources of the observed heterogeneity. However, difference in maternal age was not a source for the heterogeneity. Differences in sampling timing can affect Hg levels due to biological fluctuations, while diverse sample types and assessment methods may vary in sensitivity and specificity. Additionally, smaller sample sizes can increase variability and reduce precision by reducing statistical power. Future studies should standardize these variables or carefully account for them to improve consistency and reduce heterogeneity across the studies. Second, there was remarkable evidence of publication bias in the present study. The search strategy was restricted to publications with English language, which may have ignored non-English language articles, leading to incomplete retrieval. Third, although the results of the majority of the studies were adjusted for the potential covariates, the effect sizes of some studies were from the raw estimates without adjusting for potential covariates, which is another limitation in this study. Moreover, due to the observational nature of the analyzed studies, causal results cannot be inferred from this meta-analysis. The quality of included studies varied and presented various aspects of risk of bias, which may introduce bias and affect the validity of our findings. Finally, potential reporting bias cannot be ruled out, as studies with non-significant findings are less likely to be published. These factors should be considered when interpreting the results and highlight the need for cautious application of the findings.

In conclusion, this meta-analysis provided evidence for the inverse association of Hg exposure during pregnancy and BW. However, a significant degree of heterogeneity was detected among the studies, suggesting that the findings should be interpreted with caution. This finding suggests that pregnant women should be cautious about their exposure to Hg, especially in the later stages of pregnancy and follow the recommended safety strategies to reduce their exposure during pregnancy.

Abbreviations

BW

Birth weight

CI

Confidence intervals

LBW

Low birth weight

WHO

World Health Organization

NOS

New castle Ottawa Scale

Authors’ contributions

S Z: Study design. G W and L L: Search strategy, manuscript preparation, study selection and data extraction. D M and R G: Literature assessment. YZ: Data curation and writing-original draft preparation.

Funding

This research received no specific grant.

Data availability

The study is based on extracting data from published articles and all data are included in the report.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.