Prenatal blood lead levels and Birth Weight: a Meta-analysis study

Abstract

Purpose

Lead, a known toxic metal, causes several adverse reproductive effects, including low birth weight. Fortunately, the exposure level has sharply decreased during the recent decades, but a definitive safe level did not introduce for pregnant women yet. The current meta-analysis study aimed to conduct a quantitative estimation of maternal and umbilical cord blood lead effects on birth weight.

Methods

Two researchers have independently searched the scientific literature for retrieving related studies using the PRISMA criteria for data extraction. Twenty-one full-text articles were selected from primary 5006 titles, limited by the English language and published between 1991 and 2020 on humans.

Results

The pooled mean of maternal and umbilical cord blood lead levels were 6.85 µg/dL (95% CI: 3.36–10.34) and 5.41 µg/dL (95%CI: 3.43–7.40), respectively. The correlation coefficient analysis showed a significant inverse association between the mean maternal blood lead level and birth weight, which was confirmed by Fisher Z-Transformation analysis (-0.374, 95% CI: -0.382, -0.365, p < 0.01). In addition, a significantly lower birth weight (∆: 229 gr, p < 0.05) was found in the relatively high level of maternal blood lead than in low-level exposure (> 5 µg/dL vs. ≤ 5 µg/dL, respectively).

Conclusion

In short, the present study findings suggest an increasing maternal blood lead levels could be a potential risk factor for reducing birth weight. Thus, pregnant women should avoid lead exposure, as much as possible.

Supplementary information

The online version contains supplementary material available at 10.1007/s40201-022-00843-w.

Introduction

Lead, a well-known toxic heavy metal without any health benefit, ubiquitously dissipated throughout the earth. Human can be encountered to this heavy metal form industrial activities (i.e., lead glazed, ceramic pottery, battery manufacture, lead mining, and smelting), polluted environmental (i.e., water, soil, and air), and food chines [1–3]. In addition to environmental sources, women can be endogenously exposed to highly releases bones lead during pregnancy [4, 5]. Thereafter, the blood lead partially passes the placenta, with higher rate after the first trimester [6], to reach approximately 85% of the maternal blood concentration in the fetus blood [7–9].

In recent years, intensive activities and new legislations cause a significant decrease in the exposure levels from drinking water, paints, foods, and air (i.e., banned leaded gasoline) [3]. Despite these attempts, chronic exposure to low-levels lead remains a public health issue [10, 11]. After a significant decrease in lead exposure, scientists focus has shifted from high-dose and clinically symptomatic to lower-dose and subclinical adverse effects, such as reproductive toxicities of relatively lower levels of lead in maternal blood, umbilical cord blood, urine, bone, and placenta during the past decades [12–16].

Although lead can cause various adverse effects for human beings, pregnant women and fetuses are more vulnerable groups to its toxicity [17, 18], even at acceptable blood lead levels (≤ 5 µg/dL) [19]. Several adverse pregnancy outcomes can be induced by lead, such as pregnancy hypertension/preeclampsia, preterm birth, low birth weight, and subsequently impaired children’s neurodevelopment [20–25]. Among them, low birth weight can be a consequence of preterm birth as well, is an important risk factor for the infants mortality and morbidity [26]. Previous studies have introduced prenatal high-level lead exposure as a risk factor for declining birth weight, but there are controversial results at relatively moderate- to low-levels lead exposure [9, 20].

By far, there has been not found definitive ‘safe’ levels for pregnant women and their fetuses yet. The current study aimed to conduct a quantitative estimation of maternal and umbilical cord blood lead levels on birth weight using systematic review and meta-analysis method based on 30 years of published studies (between 1991 and 2020). We also investigated if there are significant differences in prenatal blood lead levels according to gestational age at the delivery and newborn sex.

Methods

The present review study has followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis.

Search strategy

In the first step, the current study’s researchers selected “keywords” via Medical Subject Headings (MeSH) / Cumulative Index to Nursing and Allied Health Literature (CINAHL). For searching comprehensively online databases, including PubMed, EMBASE, Scopus, and Web of Sciences, we used the following search terms: [“blood” AND “lead OR plumbum” AND “neonat* OR newborn OR birth” AND “weight OR anthropo*”]. Literatures have been independently searched by Two reviewers (MV & LS), limiting for English language articles, published between January 1991 and December 2020 (30 years), and human studies, in November 2020.

Study selection

In the primary search, we found 5006 titles. Retrieved articles moved to EndNote software for screening duplicated reference using authors’ names, title, year, and journals’ names. After removing the duplicates references, 3424 titles remained. Then, these articles have been screened according to the titles and abstracts by the first author (MV), to determine inclusion eligibility. In addition, with random sampling, 25% of articles plus uncertain ones (7%) were re-examined by the second reviewer (LS). After these two screening steps 582 titles were excluded. If the eligibility of an articles would disagree, two reviewers discussed to solve the disagreement. Finally, form 143 full text articles 21 eligible studies [11, 20, 22, 27–44] were evaluated by the PRISMA criteria and employed for data extraction (Fig. 1).

Fig. 1.

Searching the literatures and selection procedure

Inclusion and exclusion criteria

The following inclusion criteria were selected for the present study: (1) cross-sectional or population-based cohort; (2) reported maternal and/or umbilical cord blood lead levels; (3) singleton pregnancies; (4) sufficient data of the arrhythmic mean and standard division (for birth weight, blood lead concentration, and gestational age), correlation coefficients, relative risk (RR), or 95% confidence interval (CI); (5) using a reliable method for measuring blood lead concentrations (i.e., atomic absorption spectroscopy, graphite atomic absorption spectroscopy, or inductively coupled plasma-mass spectrometry); (6) available full-text paper in English.

The study exclusion criteria were as follows: (1) experimental studies; (2) less than 20 blood samples; (3) measurement lead concentrations in other biological media (i.e., plasma, serum, bone, hair, and nail); (4) review, basic science, meta-analyses, and editorial articles.

Quality assessment

Study quality was assessed using the Newcastle-Ottawa Scale, and each study earned a score (0, 1, and 2) by the following criteria: selection of study samples, comparability, and outcome of interest. Studies quality were categorized as low, moderate, and high by total gained scores (2–3, 4–6, and 7–10, respectively) (Table 1). There was not any study in the low score category.

Table 1.

Quality assessment of included studies (n = 21) using the Newcastle-Ottawa Scale

First author, year	Selection (Maximum 5 stars)				Comparability (Maximum 2 stars)	Outcome (Maximum 3 stars)		Quality scores
	Representativeness of the sample	Sample size	Non-respondents	Ascertainment of the exposure		Assessment of the outcome	Statistical test
Tom Greene, 1991	*	*	*	**	*	*	*	8
Pam Factor Litvak, 1991	*	*	*	**	*	*	*	8
Charles Savona-Ventura, 1994	*	*	-	**	-	*	*	6
Shally Awashti, 2002	*	*	*	**	*	*	*	8
Abdur Rahman, 2003	*	*	-	**	*	*	*	7
Pau-Chung Chen, 2006	*	*	-	**	*	*	*	7
Mehmet- Emre Atabek, 2007	*	*	-	**	*	*	*	7
Ramin Iranpour, 2007	*	*	-	**	-	*	*	6
Motao Zhu, 2010	*	*	-	**	*	*	*	7
Iman Al-Saleh, 2011	*	*	*	**	*	*	*	8
Abdur Rahman, 2012	*	*	-	**	*	*	*	7
Xin Xie, 2013	*	*	-	**	*	*	*	7
Emiko Nishioka, 2014	*	*	*	**	*	*	*	8
Yun-Chul Hong, 2014	*	*	-	**	*	*	*	7
Bayram Yuksel, 2016	*	*	-	**	-	*	*	6
Caroloine Taylor, 2016	*	*	-	**	*	*	*	7
Neda Akbari-Nassaji, 2017	*	*	-	**	-	*	*	6
Rodosthenis S. Rodosthenous, 2017	*	*	-	**	*	*	*	7
Zohreh Torabi, 2018	*	*	-	**	*	*	*	7
Per I Bank-Nielsen, 2019	*	*	*	**	*	*	*	8
Hossein Dalili, 2019	*	*	-	**	-	*	*	6

Data extraction

Full text of the selected articles (n = 24) was re-read to extract information about the first author’s name, publication year, sample size, study design, random sampling methods, statistical parameters (mean, standard division, 95%CIs, standard error, correlation coefficient, and sample size), gestational age, blood lead levels (converted to µg/dL units), blood sampling date, birth weight, the timing of baseline survey (year), the study location (country), and newborn sex. Blood lead levels, birth weight, and gestational age information were gathered separately for boys and girls if any. Studies that used duplication data of previous publications were excluded (n = 3). Finally, we include 21 papers for the meta-analysis. The risk of bias for included studies was independently assessed by two reviewers. Disagreements for selection papers were resolved by discussion among researchers.

Statistical analysis

The pooled mean of maternal and umbilical cord blood was calculated using random-effects models. Heterogeneity was evaluated by the I² statistics. The mean of maternal and umbilical cord blood was compered among variables, such as study design (cross-sectional or cohort), sample size (< 500 and ≥ 501), conducting year (< 2010 and ≥ 2010), continents of origin (America, Europe, Asia, and Africa), and Countries by Human Development Index (≥ 0.900, 0.850–0.899, 0.800–0.849, 0.750–0.799, 0.700–0.749, 0.650–0.699, and 0.600–0.649) [45].

We analyzed the correlation coefficient values for the mean lead levels in maternal blood and umbilical cord with newborns’ gender, birth weight, and gestational age. The analysis, by Fisher’s r-to-z transformation, obtained approximately normal distribution values and 95% CIs. The random-effects model was employed for the pooled mean analysis. Correlation coefficients were classified as poor (< 0.20), average (0.20–0.39), moderate (0.40–0.59), significant (0.60–0.79), and strong (> 0.80). We also calculated β through meta-regression analysis (Metareg function). Heterogeneity of values between studies was tested by I² (p < 0.05 and I² > 50% indicated the presence of heterogeneity). The publication bias for the mean maternal and umbilical cord blood lead levels was evaluated by funnel plots (Fig. 2). The asymmetry in the funnel may be influenced by over-estimation of the small sample size studies. We employed STATA-Ver 14 (Stata Corp LLC, Texas, USA) for all statistical analyses.

Fig. 2.

Funnel plot for testing asymmetry among 18 studies on maternal (A) and 12 umbilical cord (B) blood lead; standard error (SE)

Results

There were twenty-one original publications in the present meta-analysis study, including eleven cross-sectional studies and ten population-based cohort studies, with a total number of 60,825 blood lead reports. Table 2 summarizes the basic information of these studies.

Table 2.

The general characteristics of twenty-one included studies

Authors, year	Origin	Study Design**	Samples size of maternal blood	Samples size of cord blood	Mean Pb maternal blood (SD)	Mean Pb umbilical cord blood (SD)
Tom Greene, 1991	US	2	185	162	6.49 (1.88)	5.97 (2.09)
Pam Factor Litvak, 1991	Yugoslavia	2	907	907	11.43 (4.51)	12.91 (5.72)
Charles Savona-Ventura, 1994	Malta	1	-	82	-	14.28 (1.8)
Shally Awashti, 2002	India	2	500	-	14.34 (7.87)	-
Abdur Rahman, 2003	Pakistan	2	74	-	9.91 (4.44)	-
Pau-Chung Chen, 2006	Taiwan	2	738	-	10.10 (10.4)	-
Mehmet- Emre Atabek, 2007	Turkey	1	-	54	-	14.40 (8.9)
Ramin Iranpour, 2007	Iran	1	66	66	13.0 (2.34)	11.04 (1.79)
Motao Zhu, 2010	US	1	43,288	-	2.1 (1.48)	-
Abdur Rahman, 2012	Kuwait	2	194	194	5.77 (6.50)	10.90 (12.1)
Iman Al-Saleh, 2011	Saudi A.	1	1578	252	2.9 (1.85)	2.55 (2.59)
Xin Xie, 2013	China	1	252	1578	3.53 (1.51)	2.92 (1.58)
Yun-Chul Hong, 2014	Korea	2	2045	-	1.25 (1.49)	0.91 (1.57)
Caroloine Taylor, 2016	UK	1	4285	-	3.67 (1.47)	-
Emiko Nishioka, 2014	Japan	2	386	897	0.96 (0.61)	-
Rodosthenis S. Rodosthenous, 2017	Mexico	1	944	-	3.7 (2.7)	-
Bayram Yuksel, 2016	Turkey	2	95		4.18 (3.39)	-
Neda Akbari-Nassaji, 2017	Iran	1		147	-	0.65 (0.32)
Zohreh Torabi, 2018	Iran	2	70	613	11.01 (2.07)	9.54 (1.61)
Per I Bank-Nielsen, 2019	Denmark	1	509	-	8.60 (7.0)	-
Hossein Dalili, 2019	Iran	1	150	150	9.79 (4.31)	8.65 (3.67)

^{*1: maternal blood, 2: umbilical cord, 3 maternal blood and umbilical cord}

^{**1: cross sectional, 2: population-based cohort}

The pooled mean of maternal (18 studies, 56,266 cases) and umbilical cord (12 studies, 4559 cases) blood lead levels were 6.85 µg/dL (95% CI: 3.36–10.34) and 5.41 µg/dL (95% CI: 3.43–7.40) with a significant heterogeneity across the averages (I²_{Heterogeneity} = %72.5 and 93%, p < 0.001), respectively. The highest maternal blood lead concentration has been reported in Iran (13.0 µg/dL; Iranpour 2007) and lowest in Japan (0.96 µg/dL; Nishioka 2014). The highest umbilical cord blood lead concentration was in Turkey (14.4 µg/dL; Atabek 2007) and the lowest in S. Korea (0.91 µg/dL; Hong 2014) (Table 2).

The meta-analysis examined differences in the mean blood lead levels between the recent (2010s) and the previous decades (1990 and 2000 s). Decreasing 2.1-times (∆ = 4.43 µg/dL, I²_{heterogeneity between sub-groups} = 70.5%, p < 0.001) in the mean maternal blood lead levels and 2.8-times (∆ = 7.1 µg/dL, I²_{heterogeneity between sub-groups} = 93%, p < 0.001) in the mean umbilical cord blood lead levels revealed in the recent decade than in the last decades (Table 3). In addition, there was a noticeable higher prenatal blood lead in studies with a lower samples size (< 500) than in the relatively higher sample size studies (≥ 500). There was not enough sample to analyze studies according to the continent of origin and the human development index.

Table 3.

Comparison the pooled studies mean maternal and umbilical cord blood lead levels with different variables

variables	Cofactors	No	Maternal	I2 (P-Value)		No.	Umbilical cord	I2 (P-Value)
			Lead µg/dL (CI 95%)				Lead µg/dL (95%CI)
Sample size	< 500	10	7.18 (3.69, 10.68)		83.7 (< 0.001)	9	7.87 (3.38, 12.37)	93.4 (< 0.001)
	≥ 500	8	2.88 (1.45, 4.31)		0.0 (0.472)	3	3.09 (0.8, 6.98)	52.0 (0.125)
Years	< 2010	7	8.42 (4.14, 12.69)		69.1 (0.004)	5	10.99 (0.72, 14.63)	69.1 (0.004)
	≥ 2010	11	3.99 (1.98, 6.02)		65.9 (0.001)	7	3.89 (0.80, 6.98)	83.3 (0.001)
Continent	America	3	3.96 (1.11,6.81)		40.6 (0.186)	1	5.97 (1.87,10.07)	0.0 (0.97)
	Europe	4	4.55 (2.01,7.09)		100.0 (0.387)	3	14.16 (10.86,17.47)	100.0 (0.387)
	Asia	11	6.16 (3.10, 9.22)		81.2 (< 0.001)	8	5.06 (1.58,8.54)	84.9 (< 0.001)
Countries by Human Development Index	≥ 0.900	6	2.6 (0.834, 4.37)		54.0 (0.054)	2	3.25 ( -1.69, 8.2)	73.3 (< 0.053)
	0.850–0.899	3	6.14 (-0.214, 12.49)		40.9 (0.184)	3	9.67 (0.86, 18.48)	75.6 (0.017)
	0.800–0.849	2	4.52 (-1.37, 10.41)		0.0 (0.828)	2	13.17 (-0.88, 27.22)	85.7 (0.001)
	0.750–0.799	5	8.03 (3.71,2.36)		77.0 (0.002)	5	6.32 (1.38,11.25)	93.8 (< 0.001)
	0.700–0.749	1	14.34 (-1.09,29.77)		-	-	-	-
	0.650–0.699	-	-		-	-	-	-
	0.600–0.649	1	9.91 (1.2,18.61)		-	-	-	-

The correlation coefficient analysis showed a significant inverse association between the mean maternal blood lead level and birth weight (Fig. 3A). The Fisher Z-Transformation analysis confirmed the correlation between maternal blood lead levels and birth weight (-0.374, 95% CI: -0.382, -0.365, p < 0.01). However, the statistical analysis failed to demonstrate the same association between umbilical cord blood lead and birth weight (Fig. 3B).

Fig. 3.

Regression coefficient (beta) analysis between materna (A) and umbilical cord (B) blood lead and birthweight mean

Student’s t-test revealed a significantly lower birth weight (∆ of birth weight = 229 gr, p < 0.05) in the relatively high-level of maternal blood lead than in low-level exposure (> 5 µg/dL vs. ≤ 5 µg/dL, respectively). However, there is no significant association between umbilical cord blood lead concentrations and birth weight (Fig. 4).

Fig. 4.

Compression between low-level blood lead levels (≤ 5 ?g/dL) and the relatively higher blood lead (> 5 ?g/dL) with birth weight (Student’s t-test)

The pearsons correlation coefficient analysis showed significant asscociation between the mean maternal and umbilical cord blood lead levels with gestational age (r = 0.643 and 0.541, respectively, p < 0.01) (Fig. 5). Similarly, Fisher Z-Transformation revealed strong correlations between gestational age _mean with maternal blood lead _mean (0.763, CI 95%: 0.755–0.772, p < 0.01) and umbilical cord blood lead _mean (0.606, CI 95%: 0.576–0.635, p < 0.01).

Fig. 5.

Regression coefficient (beta) analysis between materna (A) and umbilical cord (B) blood lead and birthweight mean

The statistical analysis revealed the mean of umbilical cord blood lead level was significantly lower in boys (13.49 µg/dL, 95%CI: 13.13, 13,84) than in girls (15.12 µg/dL, 95%CI: 14.77, 15,47; I²_{heterogeneity between sub−groups}: 99%, 95% CI _{means comparison}: -0.51, -0.07, p < 0.01). Among the selected papers on maternal blood lead, a few studies presented blood lead levels and/or birth weight separately for boys and girls. Thus, the present meta-analysis could not analyze gender differences accordingly.

Discussion

The present meta-analysis study, using more than 55 thousand maternal blood lead samples (mean 4.42 µg/dL) from 18 studies, showed an inverse relationship between prenatal lead levels and birth weight, which the result was confirmed by the Fisher Z-Transformation analysis. Similarly, women with lower lead exposure had delivered newborns at higher birth weight comparison to mothers with relatively higher lead exposure (> 5 vs. ≤ 5 µg/dL, respectively). Consistent with the current study findings, some previous researches have reported an inverse relationship between birth weight and increased prenatal lead and a higher risk for decreasing birth weight in relatively elevated prenatal blood lead levels, even at the currently acceptable concentrations [9, 31, 35, 44, 46]. However, there are some studies that failed to demonstrate a significant correlation between maternal blood lead level and birth weight [13, 28, 33, 47].

Addationaly, the present study univariate-regression analysis showed a 22 gr birth weight decreasing for each 1 µg/dL increasing in maternal blood lead level. A Japanese cohort study on 20,000 blood samples, with a very low-level blood lead (median 0.63 µg/dL and range 0.16–7.4 µg/dL), showed a 5.4 g decrease in mean birth weight for each 0.1 µg/dL increase in maternal blood lead, after adjusting for covariates [23]. Similarly, Perkins (2014) and Rabito (2014) studies have reported up to 43 gr decrease in mean birth weight for each 0.1 µg/dL increased blood lead level, adjusting for potential confounders [24, 25]. In other world, a dose-response relationships between maternal blood lead levels and the risk of small for gestational age was demonstrated in a meta-analysis study [26]. Thus, according to the pervious researches and the current study results, the association between prenatal lead exposure and decreased birth weight is dose-depended and may occure, even at the very low levels exposure. or a significant difference in birth weight when higher lead exposure (2-times) compared with the low-exposure pregnant women [48]. However, a study did not find a dose response relationship between birth weight and blood lead levels (at median blood lead of 3.40 µg/dL) [9].

In addition, the present study’s statistical analysis did not find significant correlations between umbilical cord blood lead and birth weight. This may cause by significantly lower sample size on the umbilical cord than in the maternal blood (12-times), which may decrease the statistical power to detect differences. On the other hand, there is another study that has failed to find a significant association between umbilical cord blood lead and birth weight [28].

The present study findings suggested a positive correlation between the mean maternal and umbilical cord blood lead levels with gestational age. The increased blood lead mostly takes place in the second half of pregnancy by mobilization saved bones’ lead into the bloodstream, which gradually increases blood lead concentrations until delivery [49–52]. Thus, mothers who deliver later can be retrieved more lead from bones. However, there are studies that did not find significant associations between the maternal and/or umbilical cord blood lead levels with the gestational age [47, 51], or even reported an inverse relationship between blood lead and gestational (i.e., preterm birth) [25]. In short, during gestation lead may increase by releasing saved lead from bone, instead, elevated blood lead may cause early delivery (shorter gestational age).

There was a significant difference between boys and girls according to umbilical cord blood lead levels in the current study analysis. However, we did not find the same correlation between maternal blood lead and/or gender difference for birth weight. Only a few studies have examined gender influencing prenatal blood lead concentrations, or more rarely, investigated prenatal blood lead effects on birth outcomes [53–55]. Our previous research [24], consistent with Chen et al. study (2006) [31], has reported a significant difference between boys and girls according to maternal blood lead effects on newborn anthropometric characteristics. On the other hand, some studies have revealed a similar pattern of prenatal lead exposure and newborn size in boys and girls [23, 26, 27]. The underlying mechanism(s) and pathophysiology of gender effects on blood lead concentrations and birth weight on the exposure subjects need to be investigated more in future studies, such as the roll of the human placenta function on lead toxicity specifying by gender [55].

The current analysis showed a two-times decrease in prenatal blood lead levels during the past three decades. Similarly, several studies in the USA have revealed a continuous decline in blood lead levels in different races and age groups during recent years [3, 53, 56]. Additionally, a large review study selected 388 relevant original articles published between 1996 and 2013, reported a significant decline in blood lead levels among Chinese (mean 11.2 µg/dL before 2002 and 5.0 µg/dL in 2013 [54]. These significant decreases in the blood lead levels during the last decades demonstrated partial success in the lead elimination agenda, such as banded leaded gasoline in many countries.

Conclussion

The present meta-analysis study results suggested an inverse relationship between maternal blood lead levels and birth weight. Thus, increased blood lead during pregnancy could be a potential risk factor for fetal growth. Although many previous research have supported the present study’s findings, regarding the effect size, underlying mechanisms, and the trimester with higher risk for birth weight, the pieces of evidences remained inconsistent. Because no “safe” prenatal lead concentrations identified yet, pregnant women should avoid lead exposure, as much as possible.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Authors’ contributions

All authors of the present paper contributed to the study conception and design. Material preparation, data collection and analysis were performed by Mohsen Vigeh and Leyla Sahebi. The first draft of the manuscript was written by Mohsen Vigeh. All authors read and approved the final manuscript.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data Availability

All authors of the present work agree to share data, materials, software application, if needed.

Declarations

Statements and declarations

The authors declare that there are no conflicts of interest in this study.

Ethics approval and consent to participate

The present research design and procedure was approved by the Ethical Committee, Tehran University of Medical Sciences, Tehran, Iran. According to nature of the study, systematic review, obtaining informed consent from participants did not indicate.

Consent for publication

Authors of the present paper agree to transfer the article copyright to the Publisher.

Competing interests

The authors of the present study disclose that there is not any directly or indirectly interest related to the work submitted for publication.