Secular trends in blood lead concentrations of school-age children in Montevideo, Uruguay from 2009 to 2019

Abstract

Lead exposure continues to be a public health problem globally, yet very few countries perform systematic biomonitoring or surveillance of children’s blood lead levels (BLLs). Secular trends in children’s BLLs have not been well characterized outside North America and Europe. In 2009–19, we conducted a series of non-representative cross-sectional surveys in Montevideo, Uruguay, enrolling children living in areas of the city with known or suspected lead contamination. Lead was measured with atomic absorption spectrometry on fasting venous blood samples. Of the 856 children representing independent (non-sibling) observations, 759 had BLL measures. Other missing data were imputed. Using linear and logistic regression models, we estimated the covariate-adjusted year to year difference in mean BLL and the likelihood of having BLL ≥ 5 and BLL ≥ 3.5 μg/dL. At the start of the study, mean ± SD BLL was 4.8 ± 2.6 μg/dL, and at the end 1.4 ± 1.4 μg/dL. The prevalence of BLL ≥ 5 and BLL ≥ 3.5 μg/dL also differed markedly between 2009 and 2019 (30.8% vs. 2.7% and 53.8% vs. 5.8%). Similarly, where 80.8% of children had BLL ≥ 2 μg/dL in 2011, in 2019 that number was 19.3%. The estimated year to year difference in BLL was ~0.3 μg/dL. Despite this progress, pediatric lead exposure remains a problem in Montevideo. In years 2015–19, between 19 and 48% of school children had BLL ≥ 2 μg/dL, a level at which adverse neurobehavioral outcomes continue to be reported in the literature. Continued prevention and risk-reduction efforts are needed in Montevideo, including systematic surveillance of BLLs in all children.

Graphical Abstract

Introduction

Environmental lead exposure continues to be a public health problem, both in high and lower-income countries¹, placing millions of children at risk for adverse cognitive and behavioral effects even at low levels of exposure^2–5. It is abundantly clear that there is no safe level of lead, and furthermore, several studies indicate that high economic and societal costs are associated with lead exposure. For example, a staggering 824 million IQ points were estimated to be lost due to lead among people born in the United States in 1951–1980⁶. Greater variability in blood lead levels (BLL) and greater IQ deficits related to lead exposure were found in developing compared to developed countries⁷. Mild mental retardation and cardiovascular diseases resulting from lead exposure approximated 1% of the global disease burden in the year 2000, with the highest burden occurring in developing regions⁸. The economic cost attributable to lead among children < 5 years living in low- and middle-income countries (LMICs) was estimated to be $997 billion, based on a review of BLLs published after the year 2000. Regionally, these costs were $134.7 billion in Africa (4.03% of gross domestic product (GDP)), $142.4 billion in Latin America and the Caribbean ([LAC], 2.04 % of GDP), and $699.9 billion in Asia (1.88% of GDP)⁹.

The true extent of pediatric lead exposure globally is unknown, but according to a systematic review encompassing LMICs, BLLs exceed 5 μg/dL for 632 (95% confidence interval 394–780) million children¹⁰. Despite the valuable contribution of such reviews and individual scientific reports to our understanding of the extent of pediatric lead exposure, the lack of surveillance programs in many countries hampers prevention efforts and tracking of trends over time. The study of temporal trends would permit the evaluation of prevention or intervention programs. For example, following the phase-out of leaded gasoline in the U.S. (children^{11, 12}), Germany (children¹³, young adults¹⁴), Switzerland (adults)¹⁵, Czech Republic (children aged 8–10 years and adults¹⁶), and Spain (children aged 6–15 years and adults¹⁷), study of trends demonstrated clear declines in BLLs immediately afterward or decades later.

Secular trends in countries outside of Europe or North America are less well understood, in part due to lack of data. In Mexico City, studies conducted between 1988 and 2015 among children aged 1–5 years showed BLLs declining from 15.7 μg/dL in 1988 to 7.8 μg/dL in 1998 (a year after the ban of lead in gasoline), to 1.96 μg/dL in 2015. The proportion of BLL ≥ 5 μg/dL decreased from 92% (1988–1998) to 8% (2008–2015)¹⁸. In China, children’s BLLs were tracked from 1997 to 2015, revealing a downward trend after 2001 ascribed to the phase-out of leaded gasoline in 2000¹⁹. Another study in China estimated a mean annual decline of 0.3 μg/dL in children’s BLLs between 1990 and 2012²⁰.

Documenting secular trends in BLLs in countries without strong surveillance or biomonitoring programs has been problematic. Nevertheless, such efforts provide a more complete picture of the extent of lead exposure and its health burden globally. Environmental contamination is a recognized problem in LAC. Children in these regions experience several environmental exposures including lead²¹, and there is “overwhelming evidence … that current levels of exposures are hazardous and detrimental to children’s neurodevelopment”²². The present study was conducted in Montevideo, Uruguay. Lead contamination is a known problem in Montevideo (see²³) but prior to its recognition as a public health issue, it was regarded as an occupational problem, and in the case of pediatric exposure, as a medical problem touching isolated cases²⁴. Lead poisoning of the pediatric population became recognized with an “index case” in the La Teja neighborhood of Montevideo in the year 2000. Since then, lead has been detected in the blood of thousands of children in the city. Lead exposure in the general population in Uruguay has many sources, including industrial activity and waste disposal, contaminated soil and water, collection and recycling of metals including scrap and batteries, burning of wires, leaded gasoline, and lead-based paints and pigments^{25, 26}. Several legislative actions were taken to protect public health in Uruguay, including the biomonitoring of occupational exposures and lead residue management, the phase-out of leaded gasoline and lead-based paints, pipes and other products, and the disposal of lead-acid batteries that included the monitoring of lead in residential soils^24–26. Documenting secular trends in BLLs in Uruguay would advance our understanding of the evolution a key public health problem in the country and the LAC region.

Utilizing a series of cross-sectional convenience samples drawn among ~7 year old boys and girls from several areas of Montevideo, Uruguay in 2009–2019, this study had two objectives: 1) to describe the mean BLL and estimate the mean year to year difference in BLLs, while accounting for individual child and household characteristics; 2) to describe the prevalence of BLL ≥ 3.5 and ≥ 5 μg/dL and estimate the year to year difference in the likelihood of BLL ≥ 5 μg/dL (the CDC actionable level prior to 2021) and BLL ≥ 3.5 μg/dL (the current actionable level²⁷), while accounting for individual child and household characteristics. Additionally, we aimed to explore the prevalence of BLL ≥ 2 μg/dL. Calls have been made to lower the actionable level to 2 μg/dL²⁸, and this cut-point represents some of the lowest BLLs at which neurobehavioral effects have been observed^29–31.

Methods

Study setting

This study was conducted in Montevideo, Uruguay, between July 2009 and June 2019. According to World Bank figures, Uruguay’s classification changed from upper middle to high income country in 2012. Montevideo is home to over 3000 industries and businesses, of which ~27% are dedicated to food production, 15.5% to manufacturing or processing of metal products, machinery, and vehicles, 10.8% to chemicals, pharmaceuticals, and fuels, and 10.9% to “other” manufacturing. Industries were historically concentrated close to the “old city” and city center³². Figure 1 briefly reviews the history of lead exposure in Uruguay to provide context for this study. In 1992, the Government of Uruguay commissioned a report from the Inter-American Development Bank (IDB), which concluded that “lead contamination in Uruguay is a problem that has not been addressed in its entirety” and that “…population exposure have not been taken into account… and there are no base levels”³³. The extent of pediatric lead poisoning and its social dimensions was not recognized until the discovery of an “index case” in the working class neighborhood of La Teja^{25, 34}. Since that time, the political and health authorities have had several successes in handling the problem of lead (e.g., phase-out of leaded gasoline; regulation of lead in paint; the convocation of the Intersectoral Commission on Lead in 2001²⁴; the establishment of the Environmental Contaminants Clinic; the creation of children’s environmental health units (Environmental Pediatric Units) within the primary health-care network). There were also some missed opportunities (ex., abandoning twice (in 2001 and 2008) the legislative actions that would allow for universal BLL testing of children younger than 3 years of age as well as testing of cord blood; closure of the Environmental Contaminants Clinic in 2016).

Figure 1.

Historical context for childhood lead poisoning in Montevideo, Uruguay, and the study of secular trends in child blood lead concentrations.

Participant recruitment

Each year during the study period, except for 2014, a separate group of children attending the first grade of primary school was recruited and evaluated, thus providing 10 cross-sectional convenience samples of 26–133 children each. The selection of this age group aligns with the broader research program for SAM. Specifically, exposure to potentially neurotoxic chemicals at school entry may affect children’s cognitive trajectories and learning, which have been the focus of our work. Furthermore, preschool education is not universal in Uruguay whereas primary school attendance is compulsory, thus improving recruitment and representativeness of our sample.

The schools were selected from areas of the city with documented or suspected toxic metal contamination, particularly lead^{35, 36}. Montevideo is divided into large administrative units called Centro Comunal Zonal (CCZ), decentralized extensions of the city government responsible for providing services, processing petitions and complaints, etc. Of the 19 CCZs that make up Montevideo, four contributed 90% of the study participants. CCZ9 is in the east-central part of the city, 11 lies in the north-east quadrant, while 14 and 17 are in the city’s southwest, close to Montevideo Bay. The remaining 9% of participants hailed from other areas in or outside of the city.

From 2009 to 2013, recruitment took place in private schools; the majority were run by religious institutions for children from low-resource families. From 2015 to 2019, the study received permission from the Ministry of Education to recruit public-school children; recruitment was additionally expanded to include advertising posters in public spaces. The sole exclusion criterion for the study was a venous BLL of ≥ 45 μg/dL, a level that would have necessitated medical intervention (chelation), as outlined in the guidelines published in 2009 by the Uruguayan Ministry of Public Health (order 123/009). None of the children were excluded based on this criterion. In Uruguay, schools are classified based on the socioeconomic status of the populations they serve, from 1 (highest) to 5 (lowest). This study focused on levels 3–5 because of the established links between socioeconomic status and lead exposure. Overall, children from 50 different public and private primary schools were enrolled. Each year, the number of study participants depended on financial & human resources and timing of project activities.

Study protocol

Children and caregivers who were recruited into the study in the years 2009–13 received evaluations in the schools. Caregivers attended informational meetings during which they completed questionnaires and could seek clarification on the study. Caregivers were invited to accompany their children during a ‘clinic visit’ that consisted of anthropometric measures and biological sampling. In 2015–19, caregivers and children attended study visits at the Catholic University of Uruguay. Caregivers completed questionnaires with the help of study staff and accompanied children during the ‘clinic visit’.

Anthropometry

Children’s height was measured in triplicate to the nearest 0.1 cm using a portable stadiometer (Seca 214, Shorr Productions, Colombia, MD) by a trained pediatric nurse. Weight was measured in triplicate to the nearest 0.1 kg using a digital scale (Seca 872, Shorr Productions, Colombia, MD). Children wore light clothing like school uniforms but not shoes. The three measurements for height and weight were averaged. Standard weights of clothing were subtracted from the child’s average weight. Body mass index (BMI, kg/m²) was calculated from weight and height. In this context, anthropometric measurements provide a fuller picture of the overall health of the study children.

Sample collection and analysis

Venous blood was collected in the morning (8 to 11 am) by a phlebotomist, after an overnight fast. Lithium heparin-coated tubes (Becton Dickinson, Franklin Lakes, NJ) and 21-gauge needles with butterfly sets and tubing were used. Samples were stored on ice and then transported on the same day to the Specialized Center for Chemical Toxicology (CEQUIMTOX) at the Faculty of Chemistry at the University of the Republic of Uruguay. The laboratory has participated in the U.S. Centers for Disease Control and Prevention Lead and Multi-Element Proficiency Program (LAMP) and the Interlaboratory Program of Quality Control for Lead in Blood, Spain (PICC Pb-S). In addition to the ISO 9001 Quality Management implemented since 2004, the laboratory received the ISO IEC 17025 accreditation for BLL determination in 2014/2015. QA/QC procedures have been systematically applied to analyze blood samples, with protocols that included external controls, certified, standard, and/or fortified samples, reagent blanks, and control charts (http://www.cequimtox.fq.edu.uy/calidad/).

BLLs were measured by Atomic Absorption Spectrometry using flame (FAAS) or graphite furnace ionization (GFAAS) techniques with VARIAN SpectrAA-55B (2009–13) and Thermo ICC 3400 (2015–19). Graphite furnace was typically employed in 2009–12 to analyze small volumes (<2 mL). The laboratory continually worked to improve their limit of detection (LOD) to measure progressively lower BLLs. The LOD was defined based on 10 pool/blank samples and calculated as 3SD divided by the slope of the calibration curve, as outlined by EURACHEM³⁷. The LODs were as follows for 2009–10 (FAAS 2.5 μg/dL, GFAAS 2.0 μg/dL and later 0.8 μg/dL), 2011–13 (FAAS 1.8 μg/dL, GFAAS 1.0–0.8 μg/dL), 2015–16 (GFAAS 1.0–0.4 μg/dL), 2017–19 (GFAAS 0.4 μg/dL). For observations ≤LOD, we assigned values equivalent to $\text{LOD}/\sqrt{2}$. Each child was also assigned to an exposure category using BLL ≥ 2 (years 2011–19 only), ≥ 3.5, and ≥ 5 μg/dL.

A fasting venous blood sample was also collected in a serum tube containing a clot activator and separator gel. A drop of blood was removed from the tube to measure a hemoglobin concentration using a portable hemoglobinometer immediately following the draw (HemoCue, Lake Forest, CA). Quality control checks were performed daily using standard HemoCue controls (low, medium, high) provided by the manufacturer. As with anthropometry, hemoglobin levels, which reflect iron status and at low levels anemia, are reported to provide an overview of children’s overall health across study years.

Collection of family and neighborhood characteristics

Caregivers provided detailed information about maternal age, maternal education (years), employment status (yes/no), current smoking status of either parent (yes/no), family structure (child lives with both biological parents/ child lives with one biological parent/ other), and type of preschool the child had attended (private/public/none). Information regarding the socio-economic position of the household, including home ownership, number of rooms in the home, number of persons living in the house, and family possession of 12 household items such as a TV, video player, DVD player, computer, video games, radio, sound equipment, refrigerator, washer, home phone, cellular phone, and car was also obtained. A possessions score (ranging 0–5) was constructed based on items that were selected via an exploratory factor analysis from among a total of 12 household assets that were queried on a parental questionnaire. The selected items represented higher-end items, including a car and a computer. The score and its derivation are described elsewhere³⁸. A household crowding measure was calculated as the number of people living in the house divided by the number of bedrooms (crowding defined as ≥ 2 persons/room).

To represent the quality of the child’s home environment, and provide another contextual measure of the study households across the years, the Home Observation for Measurement of the Environment (HOME) Inventory was administered³⁹. It consists of observation and interview elements, with questions asked of the caregiver, and allows the calculation of a global score based on an inventory of 59 items grouped into 8 scales: parental responsibility, encouraging maturity, emotional climate, learning materials and opportunities, active stimulation, family participation, parental involvement, and physical involvement. Items are assigned scores of either 0 or 1, and the total possible global score is 59 points, with higher values suggesting a richer, more facilitating learning environment. The assessment was performed by a social worker who visited the family home during a previously scheduled visit to ensure that both the child and at least one caregiver were present.

Caregivers also indicated if they or their spouse/partner were involved in occupations known to be related to lead exposure (auto mechanic, printshop worker, metal recycling/burning, painter, etc.). Based on these responses, we created a variable to indicate any (yes/no) occupational exposure. Furthermore, caregivers indicated whether the household cultivated or consumed vegetables grown in home gardens; we created a 3-level variable representing no cultivation or consumption, consumption only, or both.

Finally, the family’s address was collected and geocoded. Aggregated 2011 census segment characteristics were downloaded from the Montevideo municipality geographic information system⁴⁰. Participants were assigned aggregated characteristics based on the census segment within which they resided. Four measures of neighborhood characteristics were obtained or calculated: 1) household distance to a polluting industry – a list compiled from the municipality’s information system and determined based on industry name or brief description to be involved in potential production or area contamination with toxic elements (ex., motor vehicle manufacturing/repair, metal product manufacturing, paint manufacturing); 2) proportion of households receiving piped water; 3) road network density within 0.5 km of the building; 4) the neighborhood disadvantage factor, calculated via confirmatory factor analysis and construct validity testing as previously described⁴¹ from 12 area-level variables related to education level (proportion of individuals within specific age groups not attending school), unemployment, proportion of inhabitants with indigenous or African ancestry, proportion of children and elderly within the area, proportion of households with high number of unmet basic needs. The neighborhood disadvantage factor is a continuous, standardized measure that has a mean score of zero and a range of −2 to +2. Each unit increase or decrease in the factor score is equivalent to 1 SD of the factor. Higher positive values indicate higher disadvantage.

Statistical analysis

Analytical sample.

Overall, 934 first graders were recruited but of those, 28 did not attend study evaluations or provided minimal information. Of 906 children who were enrolled and evaluated, 50 were siblings of other children and were removed from statistical analyses to meet the assumption of independent observations. Younger siblings or those with a large proportion of missing data were typically removed. Further 97 children were excluded due to missing observations on BLL. The analytical sample for this study consisted of 759 children with measured BLL. This sample had other missing data on variables of interest, including crowding, maternal age, and HOME Inventory score. In addition to BLL, laboratory method for BLL analysis, BLL ≥ 5 μg/dL, BLL ≥ 3.5 μg/dL, whether BLL value was ≤LOD, year of study, and the child’s sex were completely observed in the analytical sample selected for this study. Online supplemental material (see Methods) provides details on data missingness patterns and multiple imputation of missing values via chained equations.

Primary analytical approach.

Using the original, non-imputed data, sociodemographic characteristics were summarized as frequencies or mean ± SD to compare children with and without observed BLLs. Similarly, participant characteristics were summarized across years of study enrollment. Mean ± SD BLLs were computed by year of study as well as by sex.

Once data were imputed, the association between year of enrollment and BLL as well as the likelihood of having BLL ≥ 5 μg/dL and BLL ≥ 3.5 μg/dL between 2009 and 2019 was determined in unadjusted and covariate-adjusted regressions. Furthermore, the trend in year to year difference in mean BLL and the likelihood of elevated BLLs was estimated by treating year of enrollment as a continuous variable, first for the entire study period and subsequently, for years 2009–13 and 2015–19 separately. Covariates were selected based on prior literature and to account for secular trends in sociodemographic characteristics across the study period. The final set of covariates included: family possessions, HOME Inventory score, maternal age, education and employment status, household crowding, caregiver smoking status at the time of the study (yes/no), and caregiver occupational exposure to metals (yes/no). Additionally, the linear models were adjusted for whether BLL value was ≤LOD (yes/no). In prior studies, seasonality was related to differences in children’s BLLs. Based on exploratory analyses (see Supplemental Material, Methods), we did not adjust our models for study season. We tested for the presence of multicollinearity among covariates by computing a variance inflation factor (VIF) for each regression using the user-written command mivif for multiply imputed data. VIF can be interpreted as the level/factor by which the variance in the regression coefficient for a given independent variable is inflated due to its correlation with other model covariates. A VIF of 1 indicates no multicollinearity or inflation, whereas a VIF of 10 indicates a high level of multicollinearity. For this analysis, the highest VIF was 1.66.

Regressions were conducted without exploring sex differences. This decision was based on two considerations: 1) BLLs generally tracked very closely between the sexes; 2) in some years, the limited study samples would have yielded imprecise regression estimates.

Sensitivity analyses.

As outlined above, study participants hailed from several administrative units in the city. To disentangle the influence of time versus location on children’s BLLs, we selected the four CCZs that were consistently represented across time and had sufficient samples to allow for parametric statistical tests. Overall, 91% of study participants lived in those four CCZs. We split the data to correspond to the two periods of enrollment (2009–2013 and 2015–2019). We examined the effect of these periods (time) and administrative division (CCZ) on BLLs, performing a combined analysis of variance (ANOVA) using the general linear model (GLM). Time and CCZ represented fixed effects, while individual children were considered random effects. The sum of squares was partitioned into time, CCZ, and a time x CCZ interaction. The relative contribution of each component to the total variation was computed. A considerable variation in the sample number was observed across these four divisions, which varied from 34 in CCZ14 in 2009–2013 to 117 in CCZ09 in 2015–2019. To further examine the uncertainty of the ANOVA model associated with the unequal sample numbers, we randomly selected 34 samples (the lowest sample number in any location and time) and ran an additional ANOVA 1000 times to examine the uncertainty of the model associated with random sampling. The relative contribution of time vs. CCZ to the total variation was computed for these models.

Results

Participant characteristics.

Children included in the study were, on average, ~7 years of age, and the sample was evenly split among boys and girls (Table 1). Most of the children lived with both biological parents. Mothers were on average 33 years old, completed an average of 8 years education, and 38% were unemployed. Approximately a third of the households were crowded (>2 people per room), and at least one caregiver was a smoker in 60% of the study households. For the most part, children who were excluded from the analysis due to missing data on BLL were similar on sociodemographic characteristics to those included in the study. There were some notable differences. For example, 71% of the excluded children compared to 59% of those included in the study attended public or no preschool; more households in the excluded sample had a caregiver who was a smoker (~71% vs. ~61%) compared to those included in the study.

Table 1.

Characteristics of Montevideo children who did and did not have a blood lead test (2009–2019).

N	N	Yes BLL	N	No BLL
% Boys	759	52.2	97	50.2
Child age (mo)1	758	82.3 ± 6.1	96	83.4 ± 5.7
Child hemoglobin (g/dL)	754	13.2 ± 0.9	25	13.5 ± 1.0
Child BMI (kg/m2)	750	16.8 ± 2.7	74	17.5 ± 2.8
% Child lives with biological parents	734	59.5	84	55.9
% Child attended public or no preschool	724	59.5	83	71.1
Maternal age (y)	731	33.6 ± 7.0	85	33.7 ± 7.2
Maternal education (y)	743	8.5 ± 2.5	87	8.2 ± 2.4
% Mother unemployed	664	38.4	62	45.2
Number possessions (0–5)	720	2.7 ± 1.3	84	2.9 ± 1.3
% Crowded household2	719	32.4	84	35.7
% Either parent smokes	721	60.7	84	71.4
HOME score	640	40.9 ± 10.1	52	39.3 ± 11.1
% Households cultivate/consume garden vegetables	714	15.3	83	13.2
% Caregiver with occupational exposure to metals	756	32.1	97	37.1
House distance to metal industry (m)	751	1014 ± 892	85	1081 ± 1019
Road density within 0.5 km of home	751	15.3 ± 4.6	85	15.3 ± 4.7
% Neighborhood dwellings with municipal water	742	82.1 ± 22.0	84	84.0 ± 19.6
Neighborhood disadvantage factor	742	1.4 ± 1.1	84	1.2 ± 1.1
Year of enrollment
2009	52	91.2	5	8.8
2010	101	86.3	16	13.7
2011	26	83.7	5	16.1
2012	63	92.6	5	7.3
2013	68	95.8	3	4.2
2015	68	95.8	3	4.2
2016	108	87.8	15	12.2
2017	109	87.2	16	12.8
2018	133	83.1	27	16.9
2019	31	93.9	2	6.1

¹mo = months, y = years, m= meters, km = kilometers;

²Crowded household is defined as ≥2 persons per room.

Examining participant characteristics by year of study revealed secular differences (Table 2). For example, from 2009 to 2019, a lower proportion of children lived with biological parents and households reported fewer “luxury” possessions. In contrast, there was a higher proportion of mothers who were unemployed, of households that were crowded and had at least one caregiver who was a smoker, as well as a higher percentage of children who either did not or attend a public preschool. On the other hand, child age, hemoglobin level, and BMI were consistent over the study period. Maternal age and education were also consistent. This was also generally true for neighborhood level characteristics like household distance to a polluting industry, density of road network, and proportion of dwellings in the neighborhood connected to municipal water. Neighborhood disadvantage scores were positive, indicating higher disadvantage in study neighborhoods compared to the Montevideo average score of 0, and fell between 1.3 and 1.9 SD for most study years (with highest disadvantage score being 2 SD).

Table 2.

Characteristics of Montevideo children with a measured blood lead level, by year of study enrollment (2009–2019).

Year of enrollment
	2009	2010	2011	2012	2013	2015	2016	2017	2018	2019
N	52	101	26	63	68	68	108	109	133	31
% Boys	55.7	56.4	61.5	47.6	60.3	58.8	46.3	50.5	45.9	54.8
Child age (mo)1	82.6±6.3	82.2±7.0	79.0±4.7	80.4±6.2	79.8±6.3	82.8±5.7	84.6±5.7	83.9±5.7	82.6±5.4	78.0±4.8
Child hemoglobin (g/dL)	13.4±1.1	13.4± 1.1	13.2±0.9	12.8±0.9	13.1±1.2	13.3±0.8	13.0±0.8	13.4±0.7	13.3±0.8	13.6±0.8
Child BMI (kg/m2)	16.8±2.6	16.9±2.8	16.9±2.4	16.6±2.1	17.0±2.9	17.3±3.1	16.7±2.3	17.1±2.7	16.3±2.5	17.8±3.6
% Child lives with biological parents	74.5	61.7	75.0	60.3	61.7	57.3	55.6	48.6	62.1	56.7
% Child attended public or no preschool	25.5	44.8	45.8	41.4	35	58.2	76.7	74.5	77.3	66.7
Maternal age (y)	34.3±7.3	33.0±6.7	33.2±6.5	33.0±5.8	33.5±5.8	33.4±8.2	34.3±7.9	33.3±7.3	33.8±6.7	34.1±6.8
Maternal education (y)	9.9±3.4	8.8±2.3	7.6±1.7	8.7±2.5	9.9±2.6	8.3±2.6	7.9±2.3	8.0±2.2	8.0±2.4	7.8±2.2
% Mother unemployed	27.4	37.1	39.1	35.0	15.6	49.0	50.0	38.5	48.5	44.4
Number possessions (0–5)	3.3±1.2	3.3±1.3	3.2±1.3	3.2±1.2	3.6±1.2	3.1±1.2	2.6±1.2	2.4±1.3	2.4±1.3	2.5±1.3
% Crowded household2	12.8	29.3	16.7	19.0	21.7	26.5	41.6	41.1	43.2	36.7
% Either parent smokes	53.2	52.7	70.8	60.3	45.8	58.8	63.1	67.3	67.4	63.3
HOME score	44.2±10.7	41.8±8.6	43.5±6.3	46.3±7.2	47.8±7.1	44.6±8.1	41.0±9.6	34.8±9.1	34.7±10.9	38.2±7.3
% Household cultivates or consumes garden vegetables	19.2	15.6	17.4	24.6	8.6	14.7	9.8	8.6	23.3	15.3
% caregiver with occupational exposure to metals	25.0	24.7	30.8	25.4	30.8	33.8	40.7	22.0	44.4	35.5
House distance to metal industry (m)	988±1025	1165±1160	1110±322	1162±781	846±620	994±1131	1029±1234	925±554	951±609	1167±535
Road density within 0.5 km of home	16.7±5.1	15.4±5.0	17.2±3.0	13.8±4.0	14.5±3.7	13.7±5.2	14.5±4.9	16.5±4.9	15.8±4.3	17.4±2.6
% Neighborhood dwellings with municipal water	79.6±23.0	81.4±23.8	95.4±1.4	82.1±28.2	82.2±21.9	83.2±22.2	81.1±20.7	85.9±18.6	76.2±22.1	89.6±17.2
Neighborhood disadvantage score	0.7±0.9	1.4±1.2	2.2±0.7	1.6± 1.3	1.3± 1.2	1.4± 1.2	1.3± 1.0	1.7±1.1	1.3±1.0	1.9±1.2

¹mo = months, y = years, m= meters, km = kilometers;

²Crowded household is defined as ≥2 persons per room.

Temporal patterns in BLL.

The mean ± SD BLL of the study sample drawn in 2009 was 4.3 ± 2.9 μg/dL and progressively lower in samples drawn thereafter; in 2019, BLL was 1.2 ± 1.4 μg/dL (Table 3, Figure 2). Boys and girls were broadly similar terms of BLLs. The prevalence of BLL ≥ 5 μg/dL (from 30.8% to 3.2%), ≥ 3.5 μg/dL (from 53.8% to 6.4%), and ≥ 2 μg/dL (from 80.8% in 2011 to 19.3% in 2019) also declined over the study period. For all cut-points, the prevalence in 2019 was 5–10 – fold lower than at the beginning of the study. Due to a gap in study funding and participant enrollment in 2014, representing 18 months without study activities, no sample was drawn and no BLLs were measured. The difference in the prevalence of elevated BLLs around the enrollment gap, between 2013 and 2015, was quite large (13.2% vs. 2.9% for BLL ≥ 5 μg/dL; 42.6% vs. 5.9% for BLL ≥ 3.5 μg/dL; 88.1% vs. 19.1% for BLL ≥ 2 μg/dL).

Table 3.

Blood lead concentrations (μg/dL) of 759 children from Montevideo, Uruguay, by year of study enrollment (2009–2019).

Year of enrollment
	2009	2010	2011	2012	2013	2015	2016	2017	2018	2019
N	52	101	26	63	68	68	108	109	133	31
% GFAAS1	46.1	73.3	50.0	68.2	100	100	100	100	100	100
% <LOD2	34.6	20.8	19.2	16.8	11.8	68.7	34.3	10.2	11.4	41.9
Overall	4.3 ± 2.9	4.4 ± 2.2	4.2 ± 2.5	3.6 ± 2.0	3.4 ± 1.6	1.3 ± 1.2	1.5 ± 1.3	2.1 ± 2.5	2.3 ± 1.8	1.2 ± 1.4
Girls	5.0 ± 3.4	4.4 ± 2.5	4.9 ± 2.3	4.0 ± 2.1	3.4 ± 1.3	1.1 ± 0.9	1.4 ± 1.1	2.1 ± 2.9	2.3 ± 1.8	0.8 ± 0.6
Boys	3.9 ± 2.5	4.4 ± 2.0	3.8 ± 2.7	3.1 ± 1.8	3.4 ± 1.8	1.4 ± 1.4	1.8 ± 1.4	2.2 ± 2.2	2.2 ± 1.8	1.6 ± 1.7

¹Proportion of samples analyzed with graphite furnace atomic absorption spectrometry (either Thermo ICC 3400 [66.9% of samples] or Varian Spectra AA55B [33.1% of samples]) the remaining samples were analyzed with flame atomic absorption spectroscopy (FAAS);

²LOD values were as follows for the years 2009–10 (FAAS 2.5 μg/dL, GFAAS 2.0 μg/dL and later 0.8 μg/dL), 2011–13 (FAAS 1.8 μg/dL, GFAAS 1.0–0.8 μg/dL), 2015–16 (GFAAS 1.0–0.4 μg/dL), 2017–19 (GFAAS 0.4 μg/dL).

Figure 2.

Blood lead concentrations among children from Montevideo, Uruguay (2009–2019) by year and period of study enrollment.

Association between year of study and BLL.

Compared to 2009, mean BLLs were ~1 μg/dL lower in 2012 and 2013, and 2–3 BLL μg/dL lower in 2015–19 (Table 4). With year of study modeled as a continuous variable, the mean year to year difference in BLL was ~0.3 μg/dL over the entire study period. Separate analyses for the first and later five years of the study revealed that the greatest year to year differences (~0.5 μg/dL) occurred in 2009–13 compared to 2015–19, when the apparent rate of decline in BLLs was much lower and borderline statistically significant. Similarly, compared to 2009, the prevalence of BLL ≥ 5 μg/dL and of BLL ≥ 3.5 μg/dL was ~90% lower in years 2015–19. Nevertheless, the year to year difference in the likelihood of having a BLL above either cut point was most pronounced in the first five years of the study. Between 2011 when BLL ≥ 2.0 μg/dL and the end of the study, the year to year difference in the likelihood of having a BLL ≥ 2.0 μg/dL was 0.67 [0.61, 0.74]. Interestingly, the year to year difference for 2014–19 was 1.32 [1.08, 1.61].

Table 4.

Covariate-adjusted difference by study year and estimated year to year difference in mean BLL and the likelihood of elevated BLLs among children from Montevideo, Uruguay.

	β [95% CI]	OR [95% CI]
	Mean BLL1	≥ 5 μg/dL2	≥ 3.5 μg/dL2
Difference in BLL in each study year compared to 2009
2009	Ref.	Ref.	Ref.
2010	−0.42 [−1.00, 0.16]	0.70 [0.32, 1.51]	1.59 [0.78, 3.25]
2011	−0.69 [−1.50, 0.13]	0.89 [0.31, 2.55]	1.31 [0.48, 3.61]
2012	−1.33 [−1.97, −0.70]	0.49 [0.20, 1.21]	0.53 [0.25, 1.16]
2013	−1.54 [−2.17, −0.91]	0.32 [0.20, 0.83]	0.60 [0.28, 1.28]
2015	−2.34 [−2.98, −1.70]	0.05 [0.01, 0.23]	0.04 [0.01, 0.13]
2016	−3.02 [−3.60, −2.44]	0.05 [0.01, 0.23]	0.05 [0.02, 0.12]
2017	−3.09 [−3.68 −2.50]	0.06 [0.03, 0.19]	0.09 [0.04, 0.20]
2018	−2.96 [−3.54, −2.39]	0.11 [0.04, 0.36]	0.09 [0.04, 0.18]
2019	−3.18 [−3.96, −2.41]	0.04 [0.01, 0.36]	0.04 [0.01, 0.18]
Estimated year to year difference 3
Overall Estimate	−0.35 [−0.39, −0.30]	0.72 [0.67, 0.78]	0.63 [0.58, 0.68]
Years 2009–13	−0.46 [−0.60, −0.33]	0.78 [0.63, 0.95]	0.55 [0.44, 0.69]
Years 2015–19	−0.06 [−0.21, 0.08]	1.18 [0.79, 1.77]	0.96 [0.70, 1.32]

¹Adjusted for family possessions, HOME Inventory score, maternal age (y), education (y), and employment status (y/n), household crowding (y/n), caregiver smoking status (y/n), caregiver occupational exposure to metals (y/n), and whether value falls below limit of detection or not;

²Adjusted for family possessions, HOME Inventory score, maternal age (y), education (y), and employment status (y/n), household crowding (y/n), caregiver smoking status (y/n), caregiver occupational exposure to metals (y/n);

³Estimated by modeling year of enrollment (from 2009 to 2019; from 2009 to 13; and from 2015 to 2019) as a continuous variable.

Sensitivity analyses.

The analysis of variance showed that BLLs were significantly (p<0.001) related to time, CCZ, and their interaction (Table 5), but time explained ~ 83% of the variation in the observed values. In contrast, CCZ explained 12% and their interaction only 5% of the variation in BLL. The results of the 1,000 iterations based on 34 random samples across time and CCZ are presented in Supplemental Online Material (see Supplemental Figure 1).

Table 5.

Analysis of variance for the significance of the effect of time and location and their interaction on blood lead levels of children from four administrative divisions in Montevideo, Uruguay in years 2009–13 and 2015–19.

BLLs across CCZ and time		Location (administrative unit)
		CCZ91	CCZ11	CCZ14	CCZ17
		202	222	137	130
Years 2009–13		3.7±1.6	3.9±2.3	3.3±1.6	5.0±2.9
Years 2015–19		1.6±1.3	1.7±1.6	2.2±2.4	1.4±1.1
Analysis of Variance	DF	Sum Sq (SS)	Mean Sq	F value	Pr(>F)	%SS
Total	7	874.7
CCZ1	3	109.4	36.5	8.794	1.00E-05	12.5
Time2	1	722.6	722.6	174.191	<2e-16	82.6
CCZ x Time interaction	3	42.7	14.2	3.434	0.0167	4.9
Error	4	39.1	9.8	2.357	0.0523
Residuals	676	2804.2	4.1

¹Centro Comunal Zonal (CCZ) is a primary administrative unit for the Municipality of Montevideo, Uruguay;

²Time refers to two periods: 2009–13 and 2015–19.

Comparison with secular trends in other countries.

For the purposes of comparison, we plotted the trends in geometric means or median BLL values from other countries (Table 6), selected because they contained data on children and reported on more than two timepoints. The U.S. has the longest period of BLL monitoring, starting in 1976, and the most recent report encompasses 40 years¹². In several successive cycles of the National Health and Nutrition Examination Survey (NHANES), children’s BLLs declined from ~14 to <1 μg/dL. The most drastic decline occurring around the time lead was being phased out of gasoline. A report on Mexico City covers 27 years and centers on data from four birth cohorts; it also demonstrates significant declines, from ~9 to ~ 2 μg/dL¹⁸. A review of studies published on children in China over a 30-year period reflected BLL declines from ~9 μg/dL in 1991–1995 to ~3 μg/dL in 2016–2020⁴². In the Czech Republic¹⁶ and Germany¹³, the drops in BLLs were not as drastic but the starting values were not as high as elsewhere (3–4 μg/dL in Czech Republic and ~2.5 μg/dL in Germany) and the period of evaluation was shorter (12 years in Czech Republic and 8 years in Germany). Together these studies show that following an initial period of substantial declines – likely due to the phaseout of leaded gasoline – each successive drop in BLL has been more gradual.

Table 6.

Secular trends in children’s BLLs in countries around the world.

Country	Design	Year leaded gasoline phased out	Blood lead levels1
Europe & North America
Canada	Canadian Health Measures Survey Cycles 1–5 (2009– 2017) for children aged 3–5 and 6–11 years	1990
Czech Republic	Limited representative cross- sectional study of urban/suburban populations including 8–10-year-old children	2001
Germany	Repeated, bi-annual cross- sectional surveys in the State of Baden-Wuerttemberg; children aged ~10 years	1996
United States	Repeated nationally representative surveys; 1–11 year-old-children	Gradual reduction since 1973
Other parts of the world
China	Review of published studies and Monte Carlo simulation to estimate geometric mean (SD) of BLLs between 1991 and 2020; represents a population of 1.06 million children aged 1–18 years	2000
Mexico	Repeated BLL analyses in sequential birth cohorts, children aged 1–5 years	1997

¹Graphs adapted from data/tables reported in the original publications: U.S.¹², Canada⁴³, Mexico¹⁸, China⁴², Czech Republic¹⁶, Germany¹³.

Discussion

Pediatric lead exposure has been documented in most countries around the world, and it is evident that exposure to lead and elevated BLLs continue to be a public health problem globally¹. Nevertheless, very few countries have put in place the surveillance or biomonitoring programs necessary to track lead exposure trends in their vulnerable population groups, both to inform policies and to evaluate the effectiveness of public health programs. In LAC countries, systematic biomonitoring of population-level exposure to toxic elements such as lead does not exist⁴⁴. So, while the U.S. surveillance programs grapple with missed opportunities for testing⁴⁵, in other countries, the extent of pediatric lead exposure is largely unknown.

Through a series of cross-sectional surveys of schoolchildren from low-average income families in Montevideo, Uruguay, we demonstrated a ~10-year trend of progressively lower BLLs among urban children. In 2009, 30.8% of children had BLL ≥ 5 μg/dL, and 53.8% had BLL ≥ 3.5 μg/dL. On the other hand, in 2019 these numbers were 2.9% and 5.9%, respectively. In regression models, year of study was a strong predictor of children’s BLL, and the estimated mean year to year difference of ~0.3 μg/dL was calculated while considering secular differences in sociodemographic characteristics of the study samples. It is noteworthy that the estimated year to year difference in mean BLL was higher (~0.5 μg/dL) in 2009–13, which falls within a 10-year period since the ban on leaded gasoline, than in 2014–15 (~0.1 μg/dL). We further confirmed in four administrative units of Montevideo that differences in children’s BLLs were largely related to time rather than location. Despite these successes, much work remains to reduce pediatric BLLs in Montevideo to those observed in other countries. For example, while the prevalence of BLL ≥ 2 μg/dL dropped overall since 2011, 19–48% of the children in our study still experienced these lead levels in 2015–19. It is important to note also that this study was conducted in school-age children. Based on how age is related to children’s BLL, with peaks around 24–36 months and declining values thereafter⁴⁶, a higher proportion of younger children in Montevideo would be expected to have elevated BLLs. As BLLs seem to decline more slowly among children born after the phase-out of leaded gasoline, the task of identifying and mitigating exposure sources for the affected populations may require extraordinary effort and resources.

Reports on BLLs over time in other countries outside of North America and Europe have been scarce⁴⁷, and BLLs have not been declining as rapidly in some countries due to unaddressed exogenous sources⁴⁷, close vigilance of BLLs in the Global South is warranted. For example, a study compiling published data from 24 provincial cities in China showed declines in children’s geometric mean BLLs after the introduction of unleaded gasoline in year 2000; the weighted mean prevalence of BLL ≥ 10 μg/dL declined from 37% to 2% from 2000 to 2015⁴⁸. The study, based on reports from different provincial cities, showed wide variability in the actual year to year difference in the prevalence of elevated BLLs. A more comprehensive study in China encompassing 30 years from 1991 to 2020, demonstrated a 69.2% reduction in BLLs (from 8.9 to 2.7 μg/dL), and a drastically lower prevalence of ≥ 10 μg/dL (absolute difference of 63%) and ≥ 5 μg/dL (absolute difference 52%). In comparison, over 10 years of study, we observed an absolute difference in the prevalence of BLLs ≥ 5 μg/dL of 28%. In another study conducted in Bombay, India, soon after the phase-out of leaded gasoline in 2021, 33.2% of children aged 0–10 years had BLLs ≥ 10 μg/dL, compared to 61.8% of children surveyed in 1999⁴⁹. In Kinshasa, Democratic Republic of Congo, the distribution of BLLs among children aged < 6 years also shifted leftward after the phase-out of leaded gasoline two years prior⁵⁰. Mean BLL in 2004 was 12.4 μg/dL (63% with BLL ≥ 10 μg/dL) in 2004 compared to 8.7 μg/dL in 2011 (41% ≥ 10 μg/dL)⁵⁰.

There is very little information on secular changes in pediatric BLLs specifically in LAC. A study of children born in multiple birth cohorts that enrolled pregnant women in Mexico City between 1987 and 2008 showed that periodic reductions and eventual elimination of lead from gasoline were accompanied by progressive reductions in both air lead concentrations and children’s BLLs (estimated annual BLL change was 7 [5, 9]%)¹⁸, which is consistent with the ~8% year to year difference over our 10-year study. The Mexican study further showed a progressively lower prevalence of BLL ≥ 5 μg/dL at each child age (1–5 years) participating in the oldest to the newest cohort, respectively¹⁸. Although we did not measure children’s BLL prior to or directly following the phase-out of leaded gasoline in Uruguay, our findings in first graders fit with an earlier report from Montevideo⁵¹, and mirror the 20-year trends observed in Mexico City, as well as conclusions from a systematic review that BLLs in populations worldwide have fallen in concert with the phase-out of leaded gasoline⁴⁷. Another analysis of global BLLs showed that year since leaded gasoline phase-out explained an estimated 30–50% of model variance⁵².

As in many LAC countries, the information on pediatric BLLs in Uruguayan children has been sporadic, mostly obtained from published scientific studies. Systematic surveillance of the pediatric population in Uruguay is not in place despite the relatively recent history of lead poisoning. In the year 2001, media attention in Uruguay focused on the La Teja neighborhood of Montevideo, where several children were found to have BLLs above 20 μg/dL. Following this incident, the Ministry of Public Health led a coordinated response, created a committee of delegates from health, environmental, labor, and non-government organizations, opened the country’s first lead contamination clinic, and directed remediation efforts; the Ministry of Housing resettled the most affected families^{53, 54}. Nevertheless, lead exposure continues to be a problem. A study published as recently as 2016 identified 44 hazardous waste sites in Uruguay – predominantly in Montevideo – as potential sources of lead exposure⁵⁵, while a study of children living in informal settlements pointed to electronics recycling and melting as the main source of lead exposure²⁶. In the city’s peripheral areas, lead exposure may also come from the reuse of doors, windows and furniture covered with leaded paint⁵⁶. It is possible that the localization of elevated BLLs to poorer neighborhoods or families living in squatter settlements contributed to the marginalization of lead exposure³⁴. Furthermore, through a series of pragmatic decisions between 2001 and 2016⁵⁴, population-wide monitoring was never implemented. Since 2010, the Environmental Pediatric Units (Unidad Pediátrica Ambiental, UPA) have conducted surveillance of high-risk children and environments, including soil testing in irregular settlements²⁶.

Beginning in 2001, children in Montevideo who had BLLs above 20 μg/dL received public health and/or medical assistance⁵³, and a follow-up study of ~387 of these children reported a decrease in the BLLs following the medical interventions⁵⁷. The Municipality of Montevideo carried out environmental testing of lead content in soil from several informal settlements, and the reports showed a correlation of lead levels in soil with the traffic intensity in areas all over Montevideo⁵³. Finally, in November 2003, the state-owned petroleum refinery ANCAP (Administración Nacional de Combustibles, Alcoholes y Portland) began the production of lead-free gasoline, which was marketed starting in January 2004⁵³. We did not measure children’s BLLs around the time of lead phase-out from gasoline in 2004. We also have no information on air or soil lead concentrations from this period, and therefore, cannot make direct attributions to any change in policy or the environment. Nevertheless, a systematic review of the literature clearly indicates that children’s BLLs in several countries, including Uruguay, are related to the level of lead in gasoline and that declines in BLLs can be attributed to the phase-out of leaded gasoline⁴⁷ (see also Table 6). This supposition is further bolstered by the fact that BLLs in Montevideo appear to have declined despite worsening levels of exhaust emissions. Data from several monitoring stations in Montevideo indicate an increase in levels of total particulates in suspension (TPS) between 2004 and 2008⁵⁸, and there was an increase in motor vehicle sales in Uruguay, from an annual average of ~17,000 in 2005 to ~51,000 in 2015⁵⁹. Moreover, in prior studies conducted in 2009–2013, we saw little evidence of an association between lead levels in water and lead in children’s blood or urine^{60, 61}, possibly due to the strict controls enacted by the public water authority.

Our findings should be viewed in the context of the socioeconomic trends preceding and concurrent to the study. Poverty rates in Montevideo ranged 21.3–28.9% in 2002–7, but began to decline thereafter, first to 19% in 2008, to 10.5% in 2013, and to 7.3% in 2017⁶². Unemployment rates in Uruguay also changed, decreasing from 12.8% in 2004 to 6.5% in 2013 but increasing again to 8.8% in 2019; unemployment was ~1.5–2 times higher in all periods among women than men⁶³. These trends suggest a general economic recovery following the recession in the first decade of the 21^st century.

Limitations and Strengths

Several limitations of the study should be recognized as precluding firm conclusions regarding causality. First, our sample is not representative of all children in Montevideo. By design, our study focused on areas of the city with known or suspected lead contamination. We also elected to conduct recruitment among schools serving families on the lower spectrum of the income distribution, suggesting that the study children were at higher risk of lead exposure and its effects than children from other locations and sectors of society. Our sampling consistently focused on areas peripheral to the city center; as shown in our previous study, these areas also have higher levels of neighborhood disadvantage⁴¹. A second limitation is the varying sampling frame, which we recognize could potentially bias our findings; the neighborhoods from which children were sampled repeated across the study but not necessarily from year to year. We see, however, that the density of the road network around children’s homes was similar across study years, as was the distance, around one kilometer, of study households to metal-related industries for which we had location data. Furthermore, our additional analyses based on children recruited across the study years from four large administrative units in Montevideo suggest that time rather than location explained the bulk (~20-fold more) of the variability in BLLs. Our sampling strategy was also not dissimilar to other published studies. For example, in the Czech Republic¹⁶ and China¹⁹, samples were drawn from different geographical areas of the country. Third, we see a large drop in the prevalence of BLL above each cut point we tested, occurring around years 2013–15. It may be impossible to attribute this drop to any one factor and may in fact be explained by a combination of several factors. For one, this period overlaps with a gap in study recruitment spanning about 18 months. Furthermore, in 2015, we received the long-awaited permission to work with public schools, which expanded our ability to recruit children from a broader range of family circumstances and somewhat increase the representativeness of our sample. The trends in BLLs also co-occurred with differences in sociodemographic characteristics of the samples, but we were able to account for a range of individual-level characteristics in our regression models. Finally, we only utilized individual or household-level covariates in model adjustment. Yet, it is possible that differences or changes in neighborhood-level characteristics played some role in the lowering children’s BLLs. However, because many neighborhood characteristics we report are based on the 2011 population census or data gathered at a single time, they cannot capture changes during the study period.

The study also had some important strengths. First, it was conducted over a ~10-year period, allowing us to note clear differences over time, despite the variability in BLLs from year to year. Second, by focusing on first graders (age at school entry is 6–7 years in Uruguay), we avoided issues of age-related variability in children’s BLLs. Thus, we were able to consistently present secular trends in one segment of the Uruguayan population. Children’s BLLs were measured by a certified laboratory according to well established methods and although the limit of detection changed during the study to keep up with the change in population BLLs, we took this into account by including an indicator variable in the regressions specifying whether each value was ≤ or > the appropriate LOD. While we are not fully able to explain the difference in BLLs between 2013 and 2015, we do not think it has to do with the laboratory techniques as the LODs were very similar in those years (0.8–1.0 μg/dL). Critically, the change in LOD did not affect our ability to consistently identify children with BLL ≥ 5, ≥ 3.5, or ≥ 2 μg/dL (after 2011). It is also common for LOD to change across long-term surveys as laboratory instrumentation and methods change. For example, from the 1999–2000 to 2009–2010 survey cycles of the U.S. NHANES, LODs changed three times⁶⁴. While the study samples differed over 10 years, several key characteristics were remarkably consistent, including child age and nutritional status (hemoglobin level, BMI), maternal age and education, as well as the proportion of households within the family’s neighborhood served by municipal water. Another strength of this study was our ability to include a wide range of individual and family characteristics as covariates in regression models.

Conclusions

Through cross-sectional surveys conducted in 2009–19 in Montevideo, Uruguay, we saw a trend of progressively lower BLLs among school-age children. The prevalence of BLL ≥ 5 and BLL ≥ 3.5 μg/dL became markedly lower in 2019 compared to 2009 (3.2% vs. 30.8% and 6.4% vs. 53.8%, respectively). Most of the progress in lowering BLLs appears to have occurred in 2009–13, within 10 years of the leaded gasoline ban implemented in 2004. Nevertheless, pediatric lead exposure remains a problem in Montevideo, with 19–48% of school-age children testing at BLL ≥ 2 μg/dL in 2015–19, a level at which a range of adverse health outcomes have been observed in other studies. These findings are concerning given the smaller differences in BLLs in the later years of the study. Continued prevention and risk-reduction efforts are needed, based on current evidence of the most common sources and salient predictors of exposure in this population. Furthermore, systematic surveillance of BLLs in 0–18-year-olds is needed across the socioeconomic strata of the city for continued monitoring of trends and progress.

Highlights.

• Secular trends in blood lead levels (BLLs) of ~7-year-old children in Uruguay.

• Ten cross-sectional samples were drawn in 2009–2019 among low-average income families.

• Mean BLL was significantly lower in 2019 (1.4 ± 1.4 μg/dL) than 2009 (4.8 ± 2.6 μg/dL).

• The prevalence of BLL ≥3.5 μg/dL was 53.8% in 2009 and 5.8% in 2019.

• In 2015–2019, 19–45% of children had BLLs that are potentially neurotoxic (≥2 μg/dL).