Lead-Attributable Cardiovascular Disease Burden: Global Burden of Disease Study 2023

GBD 2023 Lead Collaborators; Jeffrey D Stanaway; Sandra Spearman; Nicole K DeCleene; Vanessa Garcia; Laura Lara-Castor; Stephen S Lim; Christopher J L Murray; Gregory A Roth; Matthew A Seymour; Benjamin Stark; Michael Brauer

doi:10.1001/jama.2026.2197

. 2026 Mar 30;335(16):1418–1427. doi: 10.1001/jama.2026.2197

Lead-Attributable Cardiovascular Disease Burden

Global Burden of Disease Study 2023

GBD 2023 Lead Collaborators, Jeffrey D Stanaway ^1,^✉, Sandra Spearman ¹, Nicole K DeCleene ¹, Vanessa Garcia ¹, Laura Lara-Castor ¹, Stephen S Lim ¹, Christopher J L Murray ¹, Gregory A Roth ¹, Matthew A Seymour ¹, Benjamin Stark ¹, Michael Brauer ¹

¹Department of Health Metrics Sciences, Institute for Health Metrics and Evaluation, Seattle, Washington

Group Information: The GBD 2023 Lead Collaborators appear at the end of this article.

^✉

Corresponding Author: Jeffrey D. Stanaway, MPH, PhD, Institute for Health Metrics and Evaluation, University of Washington, Seattle, WA 98195 (stanaway@uw.edu).

Accepted for Publication: February 7, 2026.

Published Online: March 30, 2026. doi:10.1001/jama.2026.2197

GBD 2023 Lead Collaborators: Jeffrey D. Stanaway, BS, MPH, PhD; Sandra Spearman, BS, MS; Nicole K. DeCleene, BS; Vanessa Garcia, BS; Laura Lara-Castor, BS, MS, PhD; Stephen S. Lim, BA, BS, PhD; Christopher J. L. Murray, BA, BS, DPhil, MD; Gregory A. Roth, BA, MPH, MD; Matthew A. Seymour, BS, MPH; Benjamin Stark, BA, MA; Michael Brauer, BA, ScD.

Affiliations of GBD 2023 Lead Collaborators: Department of Health Metrics Sciences, Institute for Health Metrics and Evaluation, Seattle, Washington (Stanaway, Spearman, DeCleene, Garcia, Lara-Castor, Lim, Murray, Roth, Seymour, Stark, Brauer).

Author Contributions: Dr Stanaway had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Stanaway, Spearman, Lim, Brauer.

Acquisition, analysis, or interpretation of data: Stanaway, Spearman, DeCleene, Garcia, Lara-Castor, Murray, Roth, Seymour, Stark, Brauer.

Drafting of the manuscript: Stanaway, Spearman, Murray, Brauer.

Critical review of the manuscript for important intellectual content: Spearman, DeCleene, Garcia, Lara-Castor, Lim, Murray, Roth, Seymour, Stark, Brauer.

Statistical analysis: Stanaway, Spearman, DeCleene, Stark.

Administrative, technical, or material support: Lim, Murray, Roth, Seymour, Brauer.

Supervision: Stanaway, Murray, Brauer.

Conflict of Interest Disclosures: Dr Stanaway reported receipt of grants from the Novo Nordisk Foundation and Open Philanthropy outside the submitted work. No other disclosures were reported.

Funding/Support: This study was funded by the Gates Foundation.

Role of the Funder/Sponsor: The Gates Foundation had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2.

^✉

Corresponding author.

PMCID: PMC13036637 PMID: 41910985

Key Points

Question

What is the association between cumulative lead exposure and cardiovascular disease (CVD) risk and what is the resulting CVD burden attributable to lead by location, year, age, and sex?

Findings

A longitudinal analysis of 42 028 National Health and Nutrition Examination Survey participants and meta-analysis found higher bone lead levels associated with significantly increased CVD mortality risk. Globally in 2023, 3.5 million deaths and 71.5 million disability-adjusted life-years (DALYs) were attributable to lead exposure, accounting for 5.8% of all deaths and 2.6% of all DALYs.

Meaning

Cumulative lead exposure remains a major, preventable contributor to global CVD mortality.

Abstract

Importance

Despite declines, lead exposure remains a major risk factor for cardiovascular disease (CVD) mortality. Quantifying the direct effects of cumulative bone lead exposure on CVD is essential to guide prevention and policy.

Objectives

To estimate the association between bone lead levels and CVD, derive an exposure-response curve, and estimate global, regional, and national lead-attributable disease burden for 1990 to 2023.

Design, Setting, and Participants

Analysis of 42 028 adults from 9 cycles of the US National Health and Nutrition Examination Survey (NHANES) spanning 1988 through 2013, with follow-up through December 2015 capturing 1748 CVD deaths, was combined with a systematic review and meta-regression. Bone lead was estimated from blood lead, age, and cohort-specific exposure histories. Cox proportional hazards models adjusted for blood pressure and confounders estimated hazard ratios for CVD mortality. A bayesian meta-regression using the Burden of Proof framework pooled NHANES results with published data to estimate the risk curve, which was then applied to lead exposure and CVD estimates to estimate population-attributable fractions and lead-attributable CVD burden.

Exposure

Estimated bone lead levels.

Main Outcomes and Measures

Lead-attributable CVD mortality, years lived with disability, years of life lost, and disability-adjusted life-years (DALYs) by age, sex, year, and location.

Results

NHANES participants were aged 18 to 90 years (median, 46 [IQR, 32-63] years) and 51.5% of participants were female, with estimated bone lead levels ranging from 0.17 μg/g to 301 μg/g (median, 13.3 μg/g). Higher bone lead levels were associated with higher CVD risk: relative to estimated preindustrial levels of 0.027 μg/g, bone lead levels of 5 μg/g, 10 μg/g, 25 μg/g, 50 μg/g, and 100 μg/g corresponded to 7.5% (95% uncertainty interval [UI], 1.7%-14.1%), 15.8% (95% UI, 12.7%-19.1%), 41.3% (95% UI, 34.2%-49.2%), 71.3% (95% UI, 55.4%-90.0%), and 87.9% (95% UI, 61.0%-121.2%) greater CVD mortality risk, respectively. Globally in 2023, 3.5 million (95% UI, 2.6 million to 4.4 million) deaths and 71.6 million (95% UI, 52.4 million to 90.3 million) DALYs were attributable to lead exposure, accounting for 5.8% (95% UI, 4.3%-7.2%) of all deaths and 2.6% (95% UI, 1.8%-3.2%) of all DALYs. Lead was the eighth leading risk for global mortality and second leading environmental risk.

Conclusions and Relevance

Cumulative lead exposure remains a major, preventable contributor to global CVD mortality. Strengthened surveillance, regulation, and remediation are urgently required to reduce the lead-attributable burden.

This study assesses the association between bone lead levels and cardiovascular disease (CVD) and estimates global, regional, and national lead-attributable CVD burden for 1990 to 2023.

Introduction

Lead exposures have declined markedly in the past 50 years with the removal of lead from gasoline and restrictions on lead in paint and consumer products, and in 2021, the US Centers for Disease Control and Prevention lowered its reference value for blood lead in children from 50 μg/L to 35 μg/L, reflecting declines in blood lead levels among children aged 1 to 5 years.¹ Still, exposure remains widespread from legacy contamination and ongoing uses, notably lead-acid battery manufacture and recycling in low- and middle-income countries,^2,3 contaminated drinking water,^4,5 household dust containing lead paint and contaminated soil particles, and consumer products, including traditional cosmetics⁶ and medicines,⁷ e-cigarettes,⁸ and electronic waste.⁹

Lead’s neurodevelopmental toxicity in children is widely recognized and lead’s contribution to cardiovascular disease (CVD) in adults has gained increasing attention.¹⁰ The 2021 Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) estimated that 7% of CVD disability-adjusted life-years (DALYs) were attributable to the effects of lead exposure on blood pressure. Experimental and human studies suggest several pathways through which cumulative lead exposure can increase cardiovascular risk beyond its effect on blood pressure, including oxidative stress, reduced bioavailability of nitric oxide, endothelial dysfunction, accelerated atherogenesis, altered vascular tone, prothrombotic changes, and kidney injury.^11,12,13

Evidence from epidemiological studies indicates that lead increases CVD risk via pathways that are independent of blood pressure.¹⁰ An analysis by Lanphear et al¹⁴ of US National Health and Nutrition Examination Survey (NHANES) data found that higher blood lead levels were associated with increased all-cause, CVD, and ischemic heart disease (IHD) mortality, even after adjusting for blood pressure and a comprehensive collection of potential confounders. Ruiz-Hernandez et al¹⁵ attributed 25.4% of the decline in US CVD mortality rates between 1988 and 2004 to concurrent decreases in blood lead levels after adjusting for conventional CVD risk factors, including blood pressure but not socioeconomic markers. Building on these analyses, Larsen and Sánchez-Triana¹⁶ combined GBD 2019 exposure estimates with relative risks from the 2 aforementioned NHANES-based studies to estimate the global disease and economic burden of lead, reporting 5.5 million adult CVD deaths in 2019 attributable to lead exposure. Herein, we build on this work by pooling 9 cycles of the NHANES survey and integrating that work into the GBD 2023 study.

We performed a 3-part analysis to estimate the effect of elevated bone lead levels on CVD. First, we updated the analysis by Lanphear et al¹⁴ to estimate the association between bone lead, rather than blood lead, on CVD mortality. Second, we conducted a systematic review and applied a bayesian meta-regression model to integrate the NHANES relative risk estimates from part 1 with those from previously published studies.¹⁷ Third, we produced comprehensive blood and bone lead estimates and, using the updated relative risk curve, estimated CVD burden attributable to elevated bone lead levels for every location, year, age, and sex in the GBD 2023 study.

Methods

Exposure Estimation

Blood Lead Exposure Estimation

Blood lead levels were modeled by location (see eTable 1 in Supplement 1 for the GBD location hierarchy), year, age, and sex, based on blood lead data from the scientific literature and surveys that measured blood lead from whole blood samples (see eFigure 1 in Supplement 1 for data coverage). Data were standardized to correspond to GBD sex and age groups, to account for different reported central measures (converting geometric means and medians to arithmetic means) and to adjust data that were not representative regarding urbanicity (see eTables 2 and 3 in Supplement 1 for details).

Spatiotemporal gaussian process regression was used with 4 covariates (lag distributed income per capita, urbanicity, vehicles per capita, and leaded gasoline phaseout) to estimate population mean blood lead levels for all GBD age groups, sexes, locations, and years, with uncertainty. It was assumed that blood lead increased linearly from preindustrial levels (0.016 μg/dL) in 1920 to estimated 1980 values.¹⁸ Finally, the individual-level distribution of blood lead levels was estimated within each GBD location, age group, sex, and year using an ensemble density method fit to individual-level blood lead data sources (see the Exposure Estimation section of the eAppendix in Supplement 1 for details).¹⁹

Bone Lead Exposure Estimation

Whereas blood lead levels primarily reflect recent exposure and short-term redistribution from bone, bone lead reflects cumulative exposure over years or decades. To estimate bone lead from a single blood lead measurement, we assumed that each individual’s percentile position within the population-level distribution of blood lead levels remained stable across their lifetime (ie, it was assumed that someone with a blood lead value in the 90th percentile for the estimation year had been at the 90th percentile of the blood lead distribution for each year of their life). For each simulated individual, we then aggregated the blood lead estimates across all preceding years of life corresponding to their blood lead percentile to calculate their cumulative blood lead index and estimated bone lead as 5% (95% CI, 4.6%-5.5%) of the cumulative blood lead index.²⁰

Relative Risk Estimation

Analysis of NHANES Data

We analyzed individual-level data from 9 NHANES cycles (NHANES III, 1999-2014) linked to the National Death Index for mortality outcomes occurring through December 2015. Variables were processed to ensure consistent coding between cycles.

NHANES Bone Lead Estimation

For each NHANES participant, we estimated bone lead using the same percentile-based approach described above, incorporating cohort effects and historical population-level exposure trends, aggregating estimated blood lead levels across all preceding years of life to calculate the cumulative blood lead index, and applying the 5% (95% CI, 4.6%-5.5%) scaler²⁰ to estimate bone lead.

NHANES Hazard Ratio Modeling

The association between estimated bone lead and time to CVD mortality was modeled using Cox proportional hazards regression (see the Analysis of NHANES Data section of the eAppendix in Supplement 1 for included International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes), adjusting for potential confounders selected based on plausibility and with reference to previous research.¹⁴ Adjustment variables included age, sex, self-reported race and ethnicity, education, income to poverty ratio, smoking status, serum cotinine, alcohol drinks per day, healthy eating index, body mass index, hemoglobin A_1c, and cholesterol, and the functional form of each variable was selected using a standardized protocol (see the Analysis of NHANES Data section of the eAppendix in Supplement 1 for details). Models were run with and without systolic blood pressure (SBP) adjustment to assess the direct and total effects of lead on CVD mortality, respectively.

Three sets of sensitivity analyses were conducted: (1) analyses using control outcomes to assess residual confounding and mediation; (2) an analysis including only participants with normotension as an alternative assessment of direct effects; and (3) analysis comparing complete-case models with models with imputed missing covariate values (see the Sensitivity Analyses section of the eAppendix in Supplement 1 for details).

Relative Risk Systematic Review and Meta-Analysis

We conducted a systematic review (PROSPERO identifier: CRD42023453813) to identify published relative risk estimates for associations between lead and CVD or IHD, building on the work of Brown et al,²¹ and searching PubMed, Embase, and Global Medicus Index through August 2023 (see the eBox in Supplement 1 for search strings and the Systematic Review section of the eAppendix in Supplement 1 for review protocol, inclusion criteria, and extraction details). We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and the Guidelines for Accurate and Transparent Health Estimates Reporting (GATHER).²² After screening 1460 articles, 1 additional study was included (eFigure 3 in Supplement 1).

Burden of Proof Model

We estimated the relative risk curve for IHD due to estimated bone lead using the Burden of Proof approach within the MR-BRT (meta-regression–bayesian, regularized, trimmed) tool^17,23 based on relative risk values from the systematic review and NHANES analysis. The model constructs a relative risk curve using splines to fit nonlinear relationships, includes automatic outlier trimming, and tests for publication bias. The tool generates a burden-of-proof risk function and evidence rating (see the MR-BRT Model section of the eAppendix in Supplement 1 for further details).¹⁷

Mediation

Bone lead affects IHD risk through direct effects and effects mediated through elevated SBP. Separately estimating these components allows for presentation of total lead-attributable IHD while avoiding double-counting the mediated component when aggregating burden across multiple risks within the GBD study. Hazard ratios from the SBP-adjusted and unadjusted NHANES models were used to decompose the mediated and direct effects, and both the direct and mediated components were included in the estimates of lead-attributable IHD (see eFigure 4 and the Mediation section of the eAppendix in Supplement 1 for details). For 11 other causes, effects were assumed to be fully mediated through SBP, as with previous versions of the GBD study: ischemic stroke, intracerebral hemorrhage, subarachnoid hemorrhage, atrial fibrillation and flutter, aortic aneurysm, lower extremity peripheral arterial disease, chronic kidney disease due to hypertension, chronic kidney disease due to type 2 diabetes, chronic kidney disease due to glomerulonephritis, hypertensive heart disease, and chronic kidney disease due to other and unspecified causes.²⁴

Attributable Burden Estimation

We estimated lead-attributable burden using the Comparative Risk Assessment framework with 4 inputs: relative risks from the Burden of Proof framework, bone lead estimates, the theoretical minimum risk exposure level (TMREL), and cause-specific burden estimates from GBD 2023.¹⁹ The TMREL is the exposure level associated with the lowest risk of adverse health outcomes and is theoretically achievable. The blood lead TMREL was set to preindustrial levels (0.16 μg/L)¹⁸ and age-specific bone lead TMRELs were estimated as 5% of the cumulative blood lead index assuming preindustrial blood lead levels.

We calculated unmediated population-attributable fractions for bone lead and IHD deaths by year, location, age, and sex, then calculated unmediated lead-attributable IHD burden as the product of the population-attributable fraction and total IHD burden. We estimated SBP-mediated lead-attributable IHD burden using the mediation factor derived by comparison of SBP-adjusted and unadjusted models. For non-IHD CVD causes, we estimated SBP-mediated burden as with previous GBD studies.¹⁹ See the Attributable Burden Estimation section of the eAppendix in Supplement 1 for details.

The University of Washington Institutional Review Board approved the GBD study (STUDY00009060). Because NHANES participants provide consent for their data to be used for future health studies, and because we used only publicly available deidentified data, no specific consent was required for this analysis.

Results

Relative Risk Estimation

NHANES Analysis

Of the 67 385 participants in the NHANES dataset, blood lead measurements were available for 57 321. After removal of participants with missing values of necessary variables, the analysis dataset included 42 028 participants with 507 000 person-years of follow-up and 8264 deaths (1748 CVD deaths) (eFigure 5 in Supplement 1). Participants were aged 18 to 90 years (median, 46 [IQR, 32-63] years), and 51.5% of participants were female (Table 1; eTable 4 in Supplement 1).

Table 1. National Health and Nutrition Examination Survey (NHANES) Participant Characteristics by Estimated Bone Lead Tertile.

Characteristics	Tertile of bone lead level
Characteristics	First (<8.8 μg/g) (n = 14 010)	Second (8.8-19.8 μg/g) (n = 14 009)	Third (≥19.9 μg/g) (n = 14 009)
NHANES cycle, No. (%)
III (1988-1994)^a	2019 (14.4)	4502 (32.1)	6799 (48.5)
1999	843 (6.0)	1225 (8.7)	974 (7.0)
2001	1437 (10.3)	1499 (10.7)	1149 (8.2)
2003	1224 (8.7)	1212 (8.7)	1053 (7.5)
2005	1537 (11.0)	1175 (8.4)	971 (6.9)
2007	1666 (11.9)	1426 (10.2)	1170 (8.4)
2009	2049 (14.6)	1369 (9.8)	968 (6.9)
2011	1931 (13.8)	1044 (7.5)	629 (4.5)
2013	1304 (9.3)	557 (4.0)	296 (2.1)
Age, median (IQR), y	34.0 (25.0-47.0)	48.0 (36.0-63.0)	59.0 (44.0-71.0)
Sex, No. (%)
Female	8318 (59.4)	7311 (52.2)	6026 (43.0)
Male	5692 (40.6)	6698 (47.8)	7983 (57.0)
Race and ethnicity (self-reported), No. (%)
Black, non-Hispanic	2660 (19.0)	2981 (21.3)	3480 (24.8)
Mexican American	2672 (19.1)	2921 (20.9)	3069 (21.9)
Other Hispanic^b	1020 (7.3)	720 (5.1)	498 (3.6)
White, non-Hispanic	6817 (48.7)	6786 (48.4)	6577 (46.9)
Other^b	841 (6.0)	601 (4.3)	385 (2.7)
Education level, No. (%)
Less than 9th grade	1239 (8.8)	1888 (13.5)	3006 (21.5)
9th-11th grade	1771 (12.6)	2141 (15.3)	2694 (19.2)
High school graduate/GED or equivalent	3460 (24.7)	3772 (26.9)	3777 (27.0)
Some college or associate-level degree	4142 (29.6)	3411 (24.3)	2578 (18.4)
College degree or above	3398 (24.3)	2797 (20.0)	1954 (13.9)
Income to poverty ratio, median (IQR)^c	2.31 (1.16-4.19)	2.31 (1.21-4.13)	1.92 (1.06-3.43)
Smoking status, No. (%)
Former	2351 (16.8)	3421 (24.4)	4208 (30.0)
Never	9494 (67.8)	7557 (53.9)	5708 (40.7)
Serum cotinine, median (IQR), ng/mL	0.0590 (0.0190-0.974)	0.123 (0.0350-15.5)	0.314 (0.0420-149)
Health Eating Index, median (IQR)^d	50.6 (40.6-60.8)	54.3 (43.7-65.0)	55.7 (45.1-66.6)
First tertile, No. (%)	5337 (38.1)	4154 (29.7)	3702 (26.4)
Second tertile, No. (%)	4803 (34.3)	4625 (33.0)	4586 (32.7)
Third tertile, No. (%)	3870 (27.6)	5230 (37.3)	5721 (40.8)
Body mass index, median (IQR)^e	27.6 (23.8-32.5)	27.4 (24.0-31.6)	26.8 (23.7-30.3)
Underweight (<18.5), No. (%)	260 (1.9)	200 (1.4)	264 (1.9)
Healthy weight (18.5-24.9), No. (%)	4404 (31.4)	4281 (30.6)	4609 (32.9)
Overweight (25.0-29.9), No. (%)	4297 (30.7)	4841 (34.6)	5382 (38.4)
Obese (≥30), No. (%)	5049 (36.0)	4687 (33.5)	3754 (26.8)
Systolic blood pressure, median (IQR), mm Hg	116 (107-126)	121 (111-134)	127 (115-142)
Diastolic blood pressure, median (IQR), mm Hg	70.0 (62.0-76.7)	72.0 (65.0-79.3)	73.0 (66.0-80.0)
History of high blood pressure, No. (%)	1583 (11.3)	2888 (20.6)	4290 (30.6)
Total cholesterol, median (IQR), mg/dL	189 (164-217)	198 (172-226)	204 (177-233)
Hemoglobin A_1c, median (IQR), %	5.30 (5.00-5.60)	5.40 (5.10-5.80)	5.50 (5.20-5.80)
Died, No. (%)	899 (6.4)	2260 (16.1)	5105 (36.4)
Death due to cardiovascular disease, No. (%)	123 (0.9)	445 (3.2)	1180 (8.4)

Open in a new tab

Abbreviation: GED, General Educational Development certification.

SI conversion: To convert total cholesterol to mmol/L, multiply by 0.0259.

^{^a}

NHANES cycle III was conducted over the years 1988 through 1994.

^{^b}

“Other Hispanic” includes those identifying as Hispanic but not Mexican; “other” includes those identifying as a non-Hispanic race not listed separately, including those reporting more than 1 race.

^{^c}

The ratio of family income to the poverty threshold.

^{^d}

The Healthy Eating Index is a measure of dietary alignment with key recommendations in the Dietary Guidelines for Americans,²⁵ ranging from 0 to 100, with higher scores corresponding to greater adherence to guidelines.

^{^e}

Calculated as weight in kilograms divided by height in meters squared.

Blood lead levels ranged from 0.7 μg/L to 418 μg/L (median, 17.5 μg/L [IQR, 10.1-30.0 μg/L]). A total of 10.1% of participants had blood lead levels of 50 μg/L or greater and 1.7% had levels of 100 μg/L or greater. A total of 3.1% of participants had blood lead levels below the lower limit of detection, most being from NHANES III, for which the lower limit of detection was higher than other cycles (eTable 5 in Supplement 1). The proportion with blood lead exceeding 50 μg/L declined from 23.8% in NHANES III to 1.6% in NHANES 2013. Estimated bone lead levels ranged from 0.17 μg/g to 301 μg/g (median, 13.3 μg/g [IQR, 6.79-24.6 μg/g]). The proportion with estimated bone lead levels greater than the equivalent to a lifetime average blood lead level of 50 μg/L declined from 80.2% in NHANES III to 23.5% in NHANES 2013. The higher prevalence of elevated estimated lifetime average vs elevated measured blood lead levels reflects higher historical exposures (eFigures 6-9 in Supplement 1).

Higher estimated bone lead levels were associated with significantly higher CVD mortality risk in all models (Figure 1). After adjusting for confounders and SBP, estimated bone lead levels of 5 μg/g, 10 μg/g, 25 μg/g, 50 μg/g, and 100 μg/g were associated with 7.5% (95% UI, 1.7%-14.1%), 15.8% (95% UI, 12.7%-19.1%), 41.3% (95% UI, 34.2%-49.2%), 71.3% (95% UI, 55.4%-90.0%), and 87.9% (95% UI, 61.0%-121.2%) greater CVD mortality risk, respectively, compared with the TMREL. Sensitivity analyses using control outcomes were consistent with adequate confounder control (ie, little residual confounding), as no significant association was observed between lead and either stroke or injury mortality risk (eFigures 10 and 11 in Supplement 1) after adjusting for confounders and SBP. Hazard ratios among participants with normotension were similar to those for all participants (eFigure 12 in Supplement 1), and estimates from models using all observations with imputed missing values were similar to those from complete-case models (eFigure 13 in Supplement 1).

MR-BRT Model and Relative Risk Curves

We found no significant evidence of publication bias (eFigure 14 in Supplement 1). The model returned a risk score of 0.145 with a 3-star rating (moderate strength of evidence), and no tested bias covariates were significantly associated with the observed relative risks. Relative to the global mean TMREL of 0.027 μg/g, bone lead levels of 5 μg/g, 10 μg/g, 25 μg/g, 50 μg/g, and 100 μg/g were associated with 7.8% (95% UI, 6.9%-8.6%), 15.9% (95% UI, 14.1%-17.7%), 40.9% (95% UI, 36.0%-45.9%), 68.7% (95% UI, 59.9%-78.0%), and 80.3% (95% UI, 69.7%-91.5%) greater CVD mortality risk, respectively (eFigure 14 in Supplement 1). The mediation factor was 0.059 (95% UI, 0.052-0.068).

Exposure Estimates

Estimated mean age-standardized blood lead levels declined globally from 121.9 μg/L in 1990 to 55.0 μg/L in 2023, with consistent declines in all superregions except sub-Saharan Africa and Central Europe, Eastern Europe, and Central Asia, where estimated levels increased through the 1990s before declining. In 2023, estimated blood lead levels were highest in sub-Saharan Africa (79.4 μg/L), South Asia (76.4 μg/L), and North Africa and the Middle East (66.6 μg/L) and were lowest in the high-income superregion (18.5 μg/L). Trends in estimated bone lead levels lagged those of blood lead, and estimated global mean bone lead levels increased from 14.8 μg/g in 1990 to a peak of 16.0 μg/g in 2001, and declined to 12.8 μg/g in 2023. In 2023, the highest estimated bone lead levels occurred in South Asia (18.0 μg/g), sub-Saharan Africa (14.2 μg/g), and Southeast Asia, East Asia, and Oceania (13.7 μg/g) (Figure 2).

Figure 2. — Estimated population-level mean blood lead level (A) and bone lead level (B) trends by superregion and globally for the period from 1990 through 2023. Bone lead–attributable mortality estimates (C) are age-standardized rates per 100 000 and include both direct effects of elevated bone lead on ischemic heart disease and indirect effects that are mediated through the effect of lead on systolic blood pressure on all associated outcomes.

Both blood and bone lead estimates were higher in older age groups, in earlier cohorts, and among males. In 2023, estimated mean global age-standardized levels were 57.6 μg/L vs 52.4 μg/L for blood and 13.9 μg/g vs 11.7 μg/g for bone among males and females, respectively. In 2023, estimated blood lead levels ranged from 50.3 μg/L among those younger than 1 year to 73.7 μg/L among those aged 95 years or older, down from 96.8 μg/L and 168.7 μg/L, respectively, in 1990. Estimated global mean bone lead levels among children younger than 1 year declined from 0.35 μg/g in 1990 to 0.18 μg/g in 2023. We estimated higher age-specific bone lead levels in earlier birth cohorts up to those born in the 1930s, with slightly lower levels in pre-1930 birth cohorts (eFigure 15 in Supplement 1). Among those aged 95 years or older, estimated bone lead levels increased from 28.4 μg/g in 1990 (1890s birth cohort) to 38.2 μg/g in 2023 (1920s birth cohort). Among those aged 70 to 74 years, estimated bone lead increased from 29.1 μg/g in 1990 (1920s birth cohort) to 35.2 μg/g in 2007 (1930s birth cohort), then declined to 33.8 μg/g in 2023 (1950s birth cohort), illustrating the nonmonotonic trend created by the 1930s peak exposure cohort (eFigure 15 in Supplement 1).

Attributable Burden

Globally in 2023, we estimated that 3.5 million (95% UI, 2.6 million to 4.4 million) deaths and 71.6 million (95% UI, 52.4 million to 90.3 million) DALYs were attributable to elevated bone lead levels, corresponding to 5.8% (95% UI, 4.3%-7.2%) of deaths and 2.6% (95% UI, 1.8%-3.2%) of DALYs. This places lead as the eighth leading risk for deaths in the GBD and the second leading environmental risk after particulate matter air pollution. Globally, 29.6% (95% UI, 26.3%-32.3%) of IHD was estimated to be attributable to bone lead (eFigure 16 in Supplement 1), with the highest population-attributable fractions in South Asia (34.2%; 95% UI, 30.5%-36.8%) and Southeast Asia, East Asia, and Oceania (32.4%; 95% UI, 28.9%-35.0%) and the lowest in the high-income superregion (19.1%; 95% UI, 16.7%-21.5%) (Figure 3).

Figure 3. — Population-attributable fractions for bone lead and ischemic heart disease estimated using the Comparative Risk Assessment framework based on relative risks from the Burden of Proof meta-analysis,¹⁹ Global Burden of Disease Study 2023 bone lead estimates, and theoretical minimum risk exposure level equal to preindustrial lead levels. Data are displayed using an equirectangular projection. Colors represent geographic boundaries and are not weighted by population size.

The highest estimated age-standardized bone lead–attributable DALY rates in 2023 were in North Africa and the Middle East (1500 [95% UI, 1160-1860] DALYs per 100 000), South Asia (1340 [95% UI, 942-1710] DALYs per 100 000), and the Central Europe, Eastern Europe, and Central Asia superregion (1060 [95% UI, 896-1230] DALYs per 100 000); the lowest were in the high-income superregion (223 [95% UI, 176-270] DALYs per 100 000). The largest estimated absolute burden occurred in the Southeast Asia, East Asia, and Oceania superregion (23.9 million [95% UI, 16.5 million to 31.3 million] DALYs) and the South Asia superregion (20.5 million [95% UI, 14.8 million to 26.2 million] DALYs), accounting for 62.0% of the total estimated global bone lead–attributable DALYs (Table 2; eTable 6 in Supplement 1).

Table 2. Lead-Attributable Death and DALY Rates and Counts by GBD Superregion^a.

Location	Age-standardized rate per 100 000 (95% UI)		All-age count, No. (95% UI), in thousands
Location	1990	2023	1990	2023
Deaths
Global	53.1 (39.6-68.2)	39.0 (28.7-48.7)	1840 (1360-2370)	3490 (2570-4360)
Central Europe, Eastern Europe, and Central Asia	79.2 (65.2-92.2)	59.0 (49.7-68.0)	323 (266-376)	394 (333-453)
High income	33.2 (27.2-39.7)	12.1 (9.5-14.9)	399 (328-478)	334 (256-417)
Latin America and Caribbean	45.0 (35.1-54.8)	31.6 (24.6-37.9)	86.5 (67.5-105)	200 (156-240)
North Africa and Middle East	106 (79.2-133)	81.4 (61.8-100)	138 (103-171)	313 (238-385)
South Asia	59.6 (42-77.2)	63.5 (43.5-81.9)	294 (209-374)	864 (601-1110)
Southeast Asia, East Asia, and Oceania	61.4 (35.1-90.9)	43.0 (29.9-55.4)	548 (308-815)	1220 (851-1570)
Sub-Saharan Africa	29.1 (17.9-43.4)	34.0 (19.4-48.9)	52.4 (31.7-78.1)	161 (91.7-232)
DALYs
Global	1080 (796-1400)	787 (576-993)	41 500 (30 600-53 900)	71 600 (52 400-90 300)
Central Europe, Eastern Europe, and Central Asia	1450 (1200-1700)	1060 (896-1230)	6500 (5390-7620)	7150 (6060-8310)
High income	613 (506-728)	223 (176-270)	7350 (6070-8720)	5350 (4190-6520)
Latin America and Caribbean	893 (691-1100)	601 (470-727)	1930 (1500-2390)	3920 (3080-4750)
North Africa and Middle East	2090 (1570-2590)	1500 (1160-1860)	3310 (2480-4090)	6860 (5380-8480)
South Asia	1360 (968-1730)	1340 (942-1710)	8050 (5690-10 200)	20 500 (14 800-26 200)
Southeast Asia, East Asia, and Oceania	1220 (711-1820)	800 (552-1050)	13 100 (7630-19 300)	23 900 (16 500-31 300)
Sub-Saharan Africa	611 (366-913)	717 (409-1050)	1300 (769-1970)	3920 (2220-5730)

Open in a new tab

Abbreviations: DALY, disability-adjusted life-year; GBD, Global Burden of Disease Study; UI, uncertainty interval.

^{^a}

Deaths and DALY estimates include both direct effects of elevated bone lead on ischemic heart disease and indirect effects that are mediated through the effect of lead on systolic blood pressure on all associated outcomes.

Estimated global bone lead–attributable deaths increased by 89.6% from 1.84 million (95% UI, 1.36 million to 2.37 million) in 1990 to 3.49 million (95% UI, 2.57 million to 4.36 million) in 2023, while estimated global bone lead–attributable DALYs increased by 72.5%, from 41.5 million (95% UI, 30.6 million to 53.9 million) to 71.6 million (95% UI, 52.4 million to 90.3 million). This increase was driven by population growth and aging, as estimated global age-standardized mortality rates declined 26.7%, from 53.1 (95% UI, 39.6-68.2) deaths per 100 000 in 1990 to 39.0 (95% UI, 28.7-48.7) deaths per 100 000 in 2023, and age-standardized DALY rates declined 26.9%, from 1080 DALYs (95% UI, 796-1400) per 100 000 to 787 (95% UI, 576-993) DALYs per 100 000. Estimated age-standardized DALY rates declined in every superregion except sub-Saharan Africa, where rates increased by 17.3% (Figure 2).

Discussion

We analyzed associations between bone lead levels and CVD mortality among NHANES participants, combined results with previously published estimates using a bayesian meta-regression, and incorporated the resulting risk curve with novel exposure estimates for the GBD 2023. We estimated IHD burden attributable to the direct effects of lead and the effects of IHD and other causes that are mediated through SBP. Including the direct effects on IHD more than doubled the lead-attributable burden estimates compared with GBD 2021, elevating lead from 18th to 8th among leading risks for deaths and from 13th to 10th among leading risks for DALYs.

The systematic review found 1 study that matched the inclusion criteria with SBP adjustment and bone lead exposure. This study measured bone lead in the tibia and patella with K-shell x-ray fluorescence spectroscopy, whereas NHANES bone lead was estimated from blood lead.²⁶ Despite this difference, the mean mortality relative risks from the 2 studies were similar at comparable exposure ranges.

Larsen and Sánchez-Triana¹⁶ used GBD 2019 lead estimates and relative risk curves from published NHANES-based analyses to estimate 5.5 million (95% CI, 2.3 million to 8.3 million) lead-attributable CVD deaths globally in 2019. The current study’s estimate for the same year was lower, although not significantly different, primarily due to the different outcomes included in the 2 studies: while Larsen and Sánchez-Triana estimated direct lead effects for all CVD causes, our estimates include direct and mediated effects of IHD and only SBP-mediated effects for other causes. Stroke was the largest cause for which we did not include direct effects. The sensitivity analysis found no significant association between lead and stroke mortality after adjusting for confounders and blood pressure, supporting the exclusion of stroke as a direct outcome.

The 3-star evidence score for the bone lead–CVD association corresponds to moderate evidence,¹⁷ reflecting the moderately large effect size and modest between-study heterogeneity balanced against the small number of studies. While the evidence is reasonably compelling, more research is needed to (1) estimate the relative risk curve more precisely, (2) assess generalizability, (3) better define risks associated with extreme exposure levels poorly represented in NHANES, and (4) understand variation across study designs or analysis methods.

Limitations

This study has some limitations. First, the greatest challenge to estimating lead-attributable CVD accurately is limited data on bone lead and its association with CVD. This study estimated bone lead to partially address this challenge; while this approach requires fewer assumptions than using blood lead directly, it still assumes individual-level exposures from population trends, and the lack of paired blood and bone lead measurements precluded rigorous validation. Second, this study’s method assumes a constant relationship between cumulative blood lead exposure and bone lead accumulation and does not account for individual variation in bone turnover rates that may affect this relationship. Third, the relative risk estimates are entirely derived from US data, and the generalizability of these estimates to different populations requires further study. Fourth, data were lacking to model associations by detailed CVD causes, and pathways beyond elevated SBP likely exist through which non-IHD CVD and non-CVD outcomes (eg, kidney disease, dementia) are affected by lead. To the extent that such pathways exist, the lead-attributable burden was likely underestimated.

Conclusions

Lead regulation, including removal of lead from gasoline and paint, is reflected in nearly universal blood lead declines in recent decades¹⁹; however, given cumulative lifetime exposure, current CVD burden is likely largely the result of historical exposure. This study’s estimates suggest that lead exposure peaked between 1970 and 1990, making those born in the 1930s and 1940s the highest lifetime exposure cohort. We expect lead-attributable CVD burden to decline as lower-exposure cohorts age into higher CVD risk, although demographic changes may counteract this, as populations decline in regions with the greatest declines in lead exposure (eg, high-income countries) and growth continues in sub-Saharan Africa, where lead levels have improved little.²⁷ Evidence from chelation trials^28,29 and before-and-after studies of the removal of lead from race car gasoline³⁰ suggests that interventions later in life may reverse at least some of the CVD risk accrued through cumulative lead exposure. However, interventions to reduce lead exposure across the lifespan, as with leaded gasoline phaseout, are likely to be the most effective and scalable approaches to reducing future lead-attributable CVD. As each person’s risk continues to increase with each day of accumulating lead exposure, there is urgency to accelerate lead exposure reductions to avert future burden. Broad-reaching and equitable improvements are therefore only possible with sustained lead-reduction efforts, and those efforts should focus most heavily on low- and middle-income settings, where exposures remain high.²⁸

Supplement 1.

eAppendix. Supplemental methods

eBox. Systematic review search strings

eTable 1. GBD location hierarchy

eTable 2. MR-BRT crosswalk adjustment factors for lead exposure (mean)

eTable 3. MR-BRT crosswalk adjustment factors for lead exposure (urbanicity)

eTable 4. NHANES participant characteristics by estimated blood lead tertile

eTable 5. Limits of detection by NHANES cycle

eTable 6. Death and DALY rates and counts by GBD region

eFigure 1. Exposure model data coverage by region and year

eFigure 2. Directed acyclic graph (DAG) showing associations with exposure, outcome, mediator and adjustment variables

eFigure 3. PRISMA diagram for blood/bone lead-CVD/IHD systematic review

eFigure 4. Direct and indirect pathways between elevated lead levels and IHD risk, with overlapping burden from all sources of elevated blood pressured and lead-attributable elevated blood pressure

eFigure 5. Flowchart of NHANES data exclusions

eFigure 6. Mean blood and estimated bone lead concentrations by age and NHANES cycle, 1988–2014

eFigure 7. Associations between measured blood and estimated bone lead levels amongst NHANES participants, by NHANES cycle

eFigure 8. Mean blood lead levels by age and NHANES cycle

eFigure 9. Mean bone lead levels by age and NHANES cycle

eFigure 10. Associations between estimated bone lead levels and stroke mortality

eFigure 11. Associations between estimated bone lead levels and injury mortality

eFigure 12. Associations between estimated bone lead levels and cardiovascular disease mortality including normotensive sensitivity analysis

eFigure 13. Sensitivity analysis of the effect of missing data on associations between estimated bone lead levels and cardiovascular disease mortality

eFigure 14. Relative risk curve and funnel plot from the meta-regression of bone lead and cardiovascular disease (CVD) mortality

eFigure 15. Blood and bone levels by age and birth year

eFigure 16. Global age-specific population attributable fractions for bone lead and ischemic heart disease by year

jama-e262197-s001.pdf^{(2.7MB, pdf)}

Supplement 2.

Data Sharing Statement

jama-e262197-s002.pdf^{(16.2KB, pdf)}

References

1.Centers for Disease Control and Prevention . CDC updates blood lead reference value. Published January 31, 2025. Accessed October 14, 2025. https://www.cdc.gov/lead-prevention/php/news-features/updates-blood-lead-reference-value.html
2.World Health Organization . Lead poisoning. Accessed October 13, 2025. https://www.who.int/news-room/fact-sheets/detail/lead-poisoning-and-health
3.Rees N, Fuller R. The Toxic Truth: Children’s Exposure to Lead Pollution Undermines a Generation of Future Potential. UNICEF and Pure Earth. Accessed October 13, 2025. https://www.unicef.org/media/73246/file/The-toxic-truth-children%E2%80%99s-exposure-to-lead-pollution-2020.pdf
4.Hanna-Attisha M, LaChance J, Sadler RC, Champney Schnepp A. Elevated blood lead levels in children associated with the Flint drinking water crisis: a spatial analysis of risk and public health response. Am J Public Health. 2016;106(2):283-290. doi: 10.2105/AJPH.2015.303003 [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Pieper KJ, Tang M, Edwards MA. Flint water crisis caused by interrupted corrosion control: investigating “ground zero” home. Environ Sci Technol. 2017;51(4):2007-2014. doi: 10.1021/acs.est.6b04034 [DOI] [PubMed] [Google Scholar]
6.Hore P, Sedlar S, Ehrlich J. Lead poisoning in a mother and her four children using a traditional eye cosmetic—New York City, 2012-2023. MMWR Morb Mortal Wkly Rep. 2024;73(30):667-671. doi: 10.15585/mmwr.mm7330a2 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Saper RB, Phillips RS, Sehgal A, et al. Lead, mercury, and arsenic in US- and Indian-manufactured Ayurvedic medicines sold via the internet. JAMA. 2008;300(8):915-923. doi: 10.1001/jama.300.8.915 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Olmedo P, Goessler W, Tanda S, et al. Metal concentrations in e-cigarette liquid and aerosol samples: the contribution of metallic coils. Environ Health Perspect. 2018;126(2):027010. doi: 10.1289/EHP2175 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Grant K, Goldizen FC, Sly PD, et al. Health consequences of exposure to e-waste: a systematic review. Lancet Glob Health. 2013;1(6):e350-e361. doi: 10.1016/S2214-109X(13)70101-3 [DOI] [PubMed] [Google Scholar]
10.Lamas GA, Bhatnagar A, Jones MR, et al. ; American Heart Association Council on Epidemiology and Prevention, Council on Cardiovascular and Stroke Nursing, Council on Lifestyle and Cardiometabolic Health, Council on Peripheral Vascular Disease, and Council on the Kidney in Cardiovascular Disease . Contaminant metals as cardiovascular risk factors: a scientific statement from the American Heart Association. J Am Heart Assoc. 2023;12(13):e029852. doi: 10.1161/JAHA.123.029852 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Vaziri ND, Ding Y, Ni Z. Nitric oxide synthase expression in the course of lead-induced hypertension. Hypertension. 1999;34(4 pt 1):558-562. doi: 10.1161/01.HYP.34.4.558 [DOI] [PubMed] [Google Scholar]
12.Navas-Acien A, Guallar E, Silbergeld EK, Rothenberg SJ. Lead exposure and cardiovascular disease—a systematic review. Environ Health Perspect. 2007;115(3):472-482. doi: 10.1289/ehp.9785 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Chowdhury R, Ramond A, O’Keeffe LM, et al. Environmental toxic metal contaminants and risk of cardiovascular disease: systematic review and meta-analysis. BMJ. 2018;362:k3310. doi: 10.1136/bmj.k3310 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Lanphear BP, Rauch S, Auinger P, Allen RW, Hornung RW. Low-level lead exposure and mortality in US adults: a population-based cohort study. Lancet Public Health. 2018;3(4):e177-e184. doi: 10.1016/S2468-2667(18)30025-2 [DOI] [PubMed] [Google Scholar]
15.Ruiz-Hernandez A, Navas-Acien A, Pastor-Barriuso R, et al. Declining exposures to lead and cadmium contribute to explaining the reduction of cardiovascular mortality in the US population, 1988-2004. Int J Epidemiol. 2017;46(6):1903-1912. doi: 10.1093/ije/dyx176 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Larsen B, Sánchez-Triana E. Global health burden and cost of lead exposure in children and adults: a health impact and economic modelling analysis. Lancet Planet Health. 2023;7(10):e831-e840. doi: 10.1016/S2542-5196(23)00166-3 [DOI] [PubMed] [Google Scholar]
17.Zheng P, Afshin A, Biryukov S, et al. The Burden of Proof studies: assessing the evidence of risk. Nat Med. 2022;28(10):2038-2044. doi: 10.1038/s41591-022-01973-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Lead levels in preindustrial humans. N Engl J Med. 1992;326(19):1293-1294. doi: 10.1056/NEJM199205073261916 [DOI] [PubMed] [Google Scholar]
19.GBD 2023 Disease and Injury and Risk Factor Collaborators . Burden of 375 diseases and injuries, risk-attributable burden of 88 risk factors, and healthy life expectancy in 204 countries and territories, including 660 subnational locations, 1990-2023: a systematic analysis for the Global Burden of Disease Study 2023. Lancet. 2025;406(10513):1873-1922. doi: 10.1016/S0140-6736(25)01637-X [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Hu H, Shih R, Rothenberg S, Schwartz BS. The epidemiology of lead toxicity in adults: measuring dose and consideration of other methodologic issues. Environ Health Perspect. 2007;115(3):455-462. doi: 10.1289/ehp.9783 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Brown L, Lynch M, Belova A, Klein R, Chiger A. Developing a health impact model for adult lead exposure and cardiovascular disease mortality. Environ Health Perspect. 2020;128(9):97005. doi: 10.1289/EHP6552 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Stevens GA, Alkema L, Black RE, et al. ; GATHER Working Group . Guidelines for Accurate and Transparent Health Estimates Reporting: the GATHER statement. Lancet. 2016;388(10062):e19-e23. doi: 10.1016/S0140-6736(16)30388-9 [DOI] [PubMed] [Google Scholar]
23.Zheng P, Barber R, Sorensen RJD, Murray CJL, Aravkin AY. Trimmed constrained mixed effects models: formulations and algorithms. J Comput Graph Stat. 2021;30(3):544-556. doi: 10.1080/10618600.2020.1868303 [DOI] [Google Scholar]
24.Brauer M, Roth GA, Aravkin AY, et al. ; GBD 2021 Risk Factors Collaborators . Global burden and strength of evidence for 88 risk factors in 204 countries and 811 subnational locations, 1990-2021: a systematic analysis for the Global Burden of Disease Study 2021. Lancet. 2024;403(10440):2162-2203. doi: 10.1016/S0140-6736(24)00933-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Shams-White MM, Pannucci TE, Lerman JL, et al. Healthy Eating Index–2020: review and update process to reflect the Dietary Guidelines for Americans, 2020-2025. J Acad Nutr Diet. 2023;123(9):1280-1288. doi: 10.1016/j.jand.2023.05.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Lieberman-Cribbin W, Domingo-Relloso A, Navas-Acien A, et al. Epigenetic biomarkers of lead exposure and cardiovascular disease: prospective evidence in the Strong Heart Study. J Am Heart Assoc. 2022;11(23):e026934. doi: 10.1161/JAHA.122.026934 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Vollset SE, Ababneh HS, Abate YH, et al. ; GBD 2021 Forecasting Collaborators . Burden of disease scenarios for 204 countries and territories, 2022-2050: a forecasting analysis for the Global Burden of Disease Study 2021. Lancet. 2024;403(10440):2204-2256. doi: 10.1016/S0140-6736(24)00685-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Lamas GA, Anstrom KJ, Navas-Acien A, et al. ; TACT2 Investigators . Edetate disodium–based chelation for patients with a previous myocardial infarction and diabetes: TACT2 randomized clinical trial. JAMA. 2024;332(10):794-803. doi: 10.1001/jama.2024.11463 [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Ujueta F, Arenas IA, Escolar E, et al. The effect of EDTA-based chelation on patients with diabetes and peripheral artery disease in the Trial to Assess Chelation Therapy (TACT). J Diabetes Complications. 2019;33(7):490-494. doi: 10.1016/j.jdiacomp.2019.04.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Hollingsworth A, Rudik I. The effect of leaded gasoline on elderly mortality: evidence from regulatory exemptions. Am Econ J Econ Policy. 2021;13(3):345-373. doi: 10.1257/pol.20190654 [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials