- Email: [email protected]
Lead exposure and mortality among U.S. workers in a surveillance program: Results from 10 additional years of follow-up
Environmental Research 177 (2019) 108625
Contents lists available at ScienceDirect
Environmental Research journal homepage: www.elsevier.com/locate/envres
Lead exposure and mortality among U.S. workers in a surveillance program: Results from 10 additional years of follow-up
T
Vaughn Barry*, Kyle Steenland Department of Environmental Health, Rollins School of Public Health at Emory University, Atlanta, GA, USA
A R T I C LE I N FO
A B S T R A C T
Keywords: Lead Blood lead Mortality Lung cancer Occupation
Background: A cohort of male lead-exposed workers with past blood lead levels, previously followed for mortality over 12 years, has now been followed for an additional 10 years. This has doubled the number of deaths and allowed for examination of mortality outcomes across a wide range of blood lead levels. Objective: Evaluate association between lead exposure and 16 causes of death. Methods: The cohort included male workers from 11 U.S. states enrolled in a U.S. lead surveillance program. Maximum blood lead level for each worker was abstracted from surveillance records. Mortality was assessed using the National Death Index. We conducted internal analyses via Cox regression adjusting for age, calendar time, and race. External analyses compared cohort mortality rates with those of the U.S. population. Blood lead categories were defined as 0- < 5, 5- < 25, 25- < 40, and ≥40 μg/dL with the two lower categories combined for outcomes with < 5 deaths in the 0- < 5 group. Results: The cohort (n = 58,368) was followed for a median of 19 years and experienced 6,527 deaths. Average maximum blood lead was 25.9 μg/dL and mean year of first blood lead test was 1997. Strong associations were found between blood lead level with larynx and lung cancer mortality. For these outcomes, hazard ratios and 95% confidence intervals across blood lead categories were 1.0 (ref), 1.1 (0.4–3.2), 3.4 (1.3–9.1) for larynx and 1.0 (ref), 1.6 (1.0–2.5), 2.0 (1.3–3.1), 2.9 (1.9–4.5) for lung (trend p-values = 0.08 and < 0.01, respectively). Positive significant trends were also seen for mortality from brain cancer, chronic obstructive pulmonary disease, ischemic heart disease, and non-hodgkin's lymphoma. Findings suggested associations with chronic renal disease and rectal cancer mortality, although trends were not statistically significant. Conclusions: The additional follow up confirmed previous relationships between lead and mortality and also detected new associations.
1. Introduction Although environmental exposure to lead in the U.S. has decreased substantially since the 1970s with the elimination of lead from gasoline and paint (Pirkle et al., 1994; Tsai and Hatfield, 2011; Tsoi et al., 2016), there remain substantial numbers of US workers exposed occupationally. It is estimated that approximately 800,000 US workers in general industry and an additional 800,000 workers in construction are currently exposed to lead on the job (https://www.osha.gov/SLTC/lead/). An additional number of adults in the US have been occupationally exposed to lead in the past and remain susceptible to lead-induced adverse health effects (Staudinger and Roth, 1998). Current US Occupational Safety and Health Administration standards require workers be removed from lead exposure when their blood
lead level reaches 50 μg/dL (construction workers) or 60 μg/dL (other workers), and to not return until their blood lead drops below 40 μg/dL. However, there is concern that current standards are outdated and do not adequately protect workers’ health (Holland and Cawthon, 2016; Schwartz and Hu, 2007; Spivey, 2007) with some suggesting that workers be removed from lead exposure when blood lead levels reach 20 μg/dL (Lustberg and Silbergeld, 2002). Additional concerns exist about possible harmful health impacts of lead in adults at even lower levels. For example, studies show risks associated with mortality and chronic disease endpoints even among blood lead levels < 5 μg/dL (Lanphear et al., 2018; Navas-Acien et al., 2007; Sirivarasai et al., 2015). Health effects of lead have been observed in all organ systems over a wide range of lead levels (Abadin et al., 2007). Most general population
* Corresponding author. MPH Department of Environmental Health Rollins School of Public Health at Emory University, 1518 Clifton Road, CNR building, room 2028, Atlanta, GA, 30322, USA. E-mail address: [email protected] (V. Barry).
https://doi.org/10.1016/j.envres.2019.108625 Received 30 May 2019; Received in revised form 1 August 2019; Accepted 2 August 2019 Available online 05 August 2019 0013-9351/ © 2019 Elsevier Inc. All rights reserved.
Environmental Research 177 (2019) 108625
V. Barry and K. Steenland
workers whose highest blood lead level was either < 5 μg/dL or 5 to < 25 μg/dL were randomly selected proportionally across states within blood lead category (Chowdhury et al., 2014a). The final study cohort consisted of 58,368 workers.
studies examining blood lead effects have focused on cardiovascular morbidity and show that increasing lead levels are related to increased systolic and diastolic blood pressure as well as increased risk of hypertension, heart disease, atherosclerosis, and cardiac disease across a range of blood lead levels (Navas-Acien et al., 2007; Nawrot et al., 2002). Chronic lead exposure in adults is also associated with mortality, and in particular, cardiovascular mortality (Lustberg and Silbergeld, 2002; Lanphear et al., 2018; Menke et al., 2006; Schober et al., 2006) although results have not been consistent in demonstrating what specific blood lead levels cause increased mortality risk. Lead is designated a “probable human carcinogen” by the International Agency for Research on Cancer (IWGotEoCRt, 2006) and as a compound that is “reasonably anticipated” to be carcinogenic to humans by the National Toxicology Program (http://ntp.niehs.nih.gov/ ntp/roc/content/profiles/lead.pdf), with the sites of interest including primarily cancers of the lung and stomach, but also to some extent kidney and brain. Most studies examining the relationship between lead exposure and cancer risk have been occupational with much higher exposure levels than general population studies. However, few data exist on incidence, and mortality data are often insufficient to provide information on dose-response relationships. Mortality and cancer outcomes, in particular, often require both lengthy follow-up due to exposure lag time and sufficient numbers of cases to examine less common cancers/outcomes. We previously followed a large cohort of lead-exposed workers (n = 58,368) for mortality over 12 years and found excess lung cancer mortality among those with the highest blood lead levels (Chowdhury et al., 2014a). Cause-specific mortality was challenging to examine due to few deaths as the cohort was relatively young (average age = 51 years) and healthy (6% mortality) at the time. We now have an additional 10 years of follow-up on this cohort which has doubled the number of deaths and allowed for examination of long-term causespecific mortality outcomes across a wide range of blood lead levels. The objective of this study is to evaluate the association between lead exposure and mortality to confirm the previous association with lung cancer mortality and to detect any new associations among 16 specific causes of death.
2.2. Variables Name, date of birth, gender, date of each blood lead test, and blood lead level (in μg/dL) for each blood lead test were collected from the selected workers in each of the eleven U.S. states. We categorized each person in to one of four blood lead categories based on the highest blood lead level they had recorded in ABLES: < 5, 5 to < 25, 25 to < 40, or ≥40 μg/dL. Additional but incomplete data on race (69% missing) and social security number (74% missing) were also collected. We classified those with missing race as white since 86% of subjects with known race were white. We used name, date of birth, gender, state of residence at the time of ABLES test(s), race (when available), and social security number (when available) for matching with the National Death Index (NDI) through the end of 2017. We defined a match as any worker found in the NDI with an NDI status code of 1 (indicating a high probability of a true match). Among workers who died, we used the ICD9/10 code indicated as the primary cause of death to determine cause-specific mortality (Supplementary Table 1). We present results for 16 specific causes of death. 2.3. Statistical analysis Cox proportional hazards regression was used to calculate hazard ratios (HRs) associated with each lead category within the cohort (i.e. internal analyses) after adjusting for birth year (five-year categories) and race, using age as the time scale. Person-time at risk began at time of the first blood lead ABLES test and continued either to date of death or December 31st, 2017, whichever occurred first. HRs compared risk at each of the blood lead categories to the reference category of < 5 μg/dL or 0 to < 25 μg/dL (depending on the numbers of cases in lowest exposure category). HRs were generated using SAS. P-values to describe trends were generated for each cause of death with Cox proportional hazards regression models that used logged continuous blood lead as the exposure and adjusted for calendar time period and race with age as the time scale. The NIOSH Life Table Analysis System was then used to calculate cause-specific rates of death for the cohort and to compare rates with those of the U.S. population via standardized mortality ratios (SMRs). SMRs were adjusted for age (in five-year categories), race, and five-year calendar time period (i.e. birth year) (Schubauer-Berigan et al., 2011). Sensitivity analyses for the Cox proportional hazards regression models were conducted using imputed race. Specifically, race was imputed 5 times via simulations (using Proc MI in SAS), 5 separate regression models were run, and coefficients and standard errors were averaged (see supplemental info from (Chowdhury et al., 2014b)). Predictor variables in the imputation models included blood lead category, state, year of birth, vital status, and year of first lead test. The models correctly predicted race for 69% of workers with known race. The study was approved by the Emory University Institutional Review Board.
2. Materials and methods 2.1. Data sources/study cohort The cohort is a subset of workers who were enrolled in the Adult Blood Lead Surveillance (ABLES) program which is sponsored by the National Institute for Occupational Safety and Health (NIOSH). ABLES has tracked laboratory-reported cases of elevated blood lead levels in U.S. adults since 1987. Initially, ABLES gathered data only on people with blood lead levels > 25 μg/dL but subsequently began to collect data on people with lower levels, too. Blood lead tests were conducted primarily in response to occupational exposure; during the years 2003–2004, 94% of adults with blood lead levels reported to ABLES were exposed via occupational sources (Adult blood lead epidemio, 2006). ABLES coverage increased from 4 U.S. states in 1987 to 41 states in 2012. The underlying cohort consisted of male workers from 11 U.S. states who had at least one blood lead level reported to ABLES anytime before 2005. States included California, Connecticut, Iowa, Massachusetts, Michigan, Minnesota, New Jersey, New York, Ohio, Pennsylvania, and Wisconsin and were chosen because they had the most subjects with blood lead data, and data that went back the farthest in time (due to the ABLES program expanding differently across states over time). We then excluded workers who were tested for the first time after the age of 70 years, before the age of 18 years, or who had implausible blood lead levels (> 250 μg/dL). For efficiency and cost considerations, a subset was then identified from this group of workers. All workers who ever had a blood lead level ≥25 μg/dL were included in the subset while
3. Results Cohort characteristics are described in Table 1. Among the 58,368 male workers in the cohort, 80% were white race (83% white using average of imputed race). Median year of birth was 1959 (range = 1914–1993 years). Workers were 39 years old, on average, at the time of their first blood lead test (mean year of first blood lead test was 1997) and were followed for mortality through 2017. Median follow up for the cohort was 19 years and there were a total of 6,527 deaths (11%) during the follow-up time period. Nearly half of the cohort (47%) was from either New York or California with the rest from 2
Environmental Research 177 (2019) 108625
V. Barry and K. Steenland
Table 1 Cohort demographics. Characteristics
Highest Lead Category Achieved
Number in cohort Median years of follow-up Mean age at first test in years Race White Non-white Missing/unknown Race imputeda White Non-white Median number blood lead tests in those with > 1 test Number with single tests only Mean highest blood lead level % with SSN (for matching)b Median year of first lead test Median year of birth Median year of death Number deaths
1 [0 to < 5 μg/dl]
2 [5 to < 25 μg/dl]
3 [25 to < 40 μg/dl]
4 [40+ μg/dl]
TOTAL
6,848 (12%) 13.5 40.7
18,650 (32%) 17.1 40.0
21,448 (37%) 21.1 37.9
11,422 (20%) 23.6 38.3
58,368 19.0 39.0
1,448 (21%) 252 (4%) 5,148 (75%)
2,356 (13%) 558 (3%) 15,736 (84%)
6,246 (29%) 1,673 (8%) 13,529 (63%)
4,339 (38%) 1,200 (10%) 5,883 (52%)
14,389 (25%) 3,683 (6%) 40,296 (69%)
6,026 (88%) 806 (12%) 2 6,124 (89%) 2.5 611 (9%) 2004 1962 2012 450 (7%)
15,014 (81%) 3,604 (19%) 3 12,739 (68%) 13.0 2,084 (11%) 2000 1961 2012 1,503 (8%)
17,898 (83%) 3,542 (17%) 4 7,786 (36%) 30.8 7,664 (36%) 1996 1959 2010 2,456 (11%)
9,478 1,940 6 1,940 52.0 4,883 1992 1955 2009 2,118
48,357 (83%) 9,891 (17%) 4 28,589 (49%) 25.9 15,242 (26%) 1998 1959 2010 6,527 (11%)
(83%) (17%) (17%) (43%)
(19%)
a
Race was imputed 5 times via simulations and then averaged. Three states (MA, MI, and WI) sent their own data to the National Death Index for matching and then sent us de-identified data, without SSN; hence these percentages are underestimates. b
with increasing maximum blood lead (Table 2). HRs and 95% confidence intervals (CIs) for all-cause, lung cancer, and COPD mortality across increasing blood lead categories compared to the 0–5 μg/dL category were 1.0 (0.9–1.1), 1.1 (1.0–1.2), and 1.4 (1.2–1.5) for all-cause; 1.6 (1.0–2.5), 2.0 (1.3–3.1), and 2.9 (1.9–4.5) for lung cancer; and 0.7 (0.4–1.1), 1.0 (0.6–1.5), and 1.5 (0.9–2.3) for COPD mortality with significant trend p-values for all three (all p < 0.01). Ischemic heart disease mortality risk also increased with increasing blood lead (trend p-value = 0.05) with HRs and 95% CIs across blood lead categories compared to 0 to < 5 μg/dL of 1.1 (0.8–1.4), 1.3 (1.0–1.7), and 1.5 (1.2–2.0). In internal analyses that compared workers with maximum blood lead of 25 to < 40 and ≥ 40 to those with 0–25 μg/dL (used for rarer outcomes with < 5 deaths in the 0 to < 5 μg/dL referent group), there was a significant increased risk of mortality from larynx cancer among workers with maximum blood lead ≥40 compared to those with 0–25 μg/dL (HR = 3.4, 95% CI = 1.3–9.1)) (Table 3). P-values from continuous models for malignant brain cancer, larynx cancer, and nonhodgkin's lymphoma mortality were all borderline significant (p = 0.06, 0.08, and 0.06, respectively). HRs and 95% CIs comparing
the other 9 states. Half of the cohort (49%) had only one blood lead test. Workers with the lowest blood lead level(s) were more likely to have had only one blood lead test compared to workers with higher blood lead levels (89% of workers with a lead level of 0 to < 5 μg/dL had only one lead test compared to 17% of workers with a lead level ≥40 μg/dL). Among workers with more than one blood lead test, the median number of blood lead tests per person was 4. Workers with higher maximum blood lead levels were also followed up longer than those with lower maximum blood lead levels (median of 24 vs. 14 years comparing workers with maximum lead level ≥40 μg/dL to those with maximum lead level 0 to < 5 μg/dL), more likely to be non-white race (22% vs. 15% comparing highest exposed category to lowest exposed), and more likely to die (19% vs. 7% comparing highest exposed category to lowest). Average maximum blood lead level, mean year of birth, and mortality rate did not vary across state. In internal analyses examining the cause-specific deaths that had enough cases to allow comparisons across four blood lead exposure categories (0 to < 5, 5 to < 25, 25 to < 40 and ≥ 40 μg/dL), there was significant increased risk of all-cause, lung cancer, and COPD mortality
Table 2 Hazard ratios (HRs)a and 95% confidence intervals (CIs) by lead category, with 0 to < 5 μg/dl as referent group. Cause of death
p-value for trendb
Highest Lead Category Achieved 0 to < 5 μg/dl
5 to < 25 μg/dl
25 to < 40 μg/dl
40+ μg/dl
N
HR (95% CI)
N
HR (95% CI)
N
HR (95% CI)
N
HR (95% CI)
All causes Cancer Colon Esophagus Kidney Liver Lung Stomach
450
Ref
1503
1.00 (0.90–1.11)
2456
1.09 (0.98–1.20)
2118
1.38 (1.24–1.53)
< 0.01
8 8 5 11 24 5
Ref Ref Ref Ref Ref Ref
21 15 16 30 134 6
0.77 0.57 0.94 0.86 1.61 0.33
(0.34–1.74) (0.24–1.34) (0.35–2.58) (0.43–1.72) (1.04–2.48) (0.10–1.09)
43 28 19 34 261 13
0.98 0.77 0.80 0.73 2.03 0.39
(0.45–2.10) (0.35–1.70) (0.30–2.17) (0.37–1.44) (1.34–3.10) (0.14–1.12)
38 23 14 45 268 16
1.19 0.97 0.92 1.53 2.92 0.64
(0.54–2.61) (0.43–2.20) (0.32–2.58) (0.79–2.99) (1.91–4.46) (0.22–1.82)
0.98 0.71 0.74 0.23 < 0.01 0.89
Cerebrovascular disease (i.e. stroke) COPD Ischemic heart disease
17 24 62
Ref Ref Ref
42 56 231
0.72 (0.41–1.26) 0.68 (0.42–1.10) 1.08 (0.82–1.43)
88 118 428
0.91 (0.54–1.53) 0.98 (0.63–1.52) 1.28 (0.98–1.67)
79 124 364
1.11 (0.65–1.90) 1.46 (0.94–2.28) 1.51 (1.15–1.99)
0.92 < 0.01 0.05
a b
Proportional hazards regression models adjusted for birth year (five-year categories) and race with age as timescale. From proportional hazards regression model with continuous blood lead as exposure (adjusted for calendar time period and race with age as time scale). 3
Environmental Research 177 (2019) 108625
V. Barry and K. Steenland
Table 3 Hazard ratios (HRs)a and 95% confidence intervals (CIs) by lead category, with 0 to < 25 μg/dl as referent group. Cause of death
p-value for trendb
Highest Lead Category Achieved 0 to < 25 μg/dl
25 to < 40 μg/dl
40+ μg/dl
N
HR (95% CI)
N
HR (95% CI)
N
HR (95% CI)
Cancer Brain cancer Bladder cancer Larynx cancer Non-Hodgkin's L Pancreatic cancer Rectal cancer
15 13 6 18 39 9
Ref Ref Ref Ref Ref Ref
26 25 7 20 47 17
1.67 1.60 1.05 0.84 1.07 1.61
(0.88–3.17) (0.81–3.16) (0.35–3.15) (0.44–1.62) (0.69–1.64) (0.71–3.65)
15 19 15 28 35 15
1.49 1.71 3.42 1.60 1.15 2.06
(0.71–3.12) (0.83–3.55) (1.29–9.09) (0.85–3.01) (0.72–1.85) (0.87–4.84)
0.06 0.68 0.08 0.06 0.15 0.50
Chronic renal disease
14
Ref
24
1.27 (0.65–2.49)
25
1.81 (0.91–3.57)
0.49
a b
Proportional hazards regression models adjusted for birth year (five-year categories) and race with age as timescale. From proportional hazards regression model with continuous blood lead as exposure (adjusted for calendar time period and race with age as time scale).
workers with blood lead ≥40 to those with 0–25 μg/dL for brain cancer, larynx cancer, and non-hodgkin's lymphoma were 1.5 (0.7–3.1), 3.4 (1.3–9.1), and 1.6 (0.9–3.0), respectively. Additionally, although trend p-values were not significant, there was evidence of increasing chronic renal disease mortality risk across blood lead category (HR = 1.3 (0.7–2.5) and 1.8 (0.9–3.6)) as well as rectal cancer mortality (HR = 1.6 (0.7–3.7) and 2.1 (0.9–4.8)) comparing workers with blood lead levels of 25 to < 40 and ≥ 40 compared to 0 to < 25 μg/dL. Overall SMRs comparing the cohort's risk of all-cause and causespecific mortality to the general U.S. population's risk demonstrated a decreased risk of death among the cohort with most SMRs < 1.0 (Table 4). However, there were a significant excess number of larynx, liver, and lung cancer deaths among workers who had maximum blood lead levels ≥40 μg/dL (observed/expected for larynx = 15/8.4, liver = 45/31.1, and lung = 268/242.1). Among workers with maximum blood lead level ≥40 μg/dL, there was no excess mortality compared to the external U.S. population with respect to deaths caused by other cancers, cerebrovascular disease, chronic obstructive pulmonary disease (COPD), chronic renal disease, or ischemic heart disease. Sensitivity analyses using imputed race resulted in virtually
identical results with HRs and 95% confidence intervals the same out to at least the first decimal place for nearly all mortality outcomes. The one exception was chronic renal disease mortality which showed minor differences between our applied race definition (HR = 1.3 (0.7–2.5) and HR = 1.8 (0.9–3.6) for the two highest lead categories compared to 0 to < 25 μg/dL) and the imputed race definition (HR = 1.5 (0.8–3.0) and HR = 2.1 (1.1–4.3)). 4. Discussion Similar to the findings from the earlier follow up of this cohort (Chowdhury et al., 2014a), we found significant increasing risk of allcause, lung cancer, COPD, and ischemic heart disease mortality risk among workers with increasing blood lead levels. With the additional 10 years of follow up which resulted in more deaths and thus more power, we were able to detect new associations between blood lead with mortality from malignant brain cancer, larynx cancer, and nonhodgkin's lymphoma. There was also new evidence of possible relationships between blood lead with mortality due to chronic renal disease and rectal cancer among workers with the highest blood lead levels (i.e. ≥40 μg/dL).
Table 4 Standardized Mortality Ratios (SMRs)a and 95% confidence intervals (CIs) by lead category (n = 58,368). Cause of Death
Highest Lead Category Achieved 1 [0 to < 5 μg/dl]
2 [5 to < 25 μg/dl]
3 [25 to < 40 μg/dl]
4 [40+ μg/dl]
TOTAL
N
SMR (95% CI)
N
SMR (95% CI)
N
SMR (95% CI)
N
SMR (95% CI)
N
SMR (95% CI)
All causes Cancer Brain Bladder Colon Esophagus Kidney Larynx Liver Lung Non-Hodgkin's Lb Pancreas Rectum Stomach
450
0.64 (0.58, 0.70)
1503
0.63 (0.60, 0.66)
2456
0.66 (0.63, 0.68)
2118
0.80 (0.77, 0.84)
6527
0.69 (0.67, 0.71)
0 1 8 8 5 1 11 24 1 4 3 5
0.00 0.20 0.56 0.91 0.80 0.45 1.13 0.37 0.13 0.32 0.74 1.16
(0.00, (0.01, (0.24, (0.39, (0.26, (0.01, (0.56, (0.24, (0.00, (0.09, (0.15, (0.38,
0.53) 1.10) 1.10) 1.78) 1.87) 2.50) 2.02) 0.56) 0.75) 0.82) 2.16) 2.71)
15 12 21 15 16 5 30 134 17 35 6 6
0.65 0.68 0.43 0.51 0.76 0.66 0.94 0.61 0.66 0.84 0.44 0.41
(0.36, (0.35, (0.26, (0.28, (0.43, (0.21, (0.64, (0.51, (0.38, (0.58, (0.16, (0.15,
1.07) 1.19) 0.65) 0.84) 1.23) 1.54) 1.35) 0.72) 1.05) 1.17) 0.96) 0.89)
26 22 43 28 19 7 34 261 20 47 17 13
0.80 0.81 0.56 0.66 0.61 0.60 0.73 0.77 0.49 0.76 0.85 0.54
(0.52, (0.51, (0.40, (0.44, (0.37, (0.24, (0.51, (0.68, (0.30, (0.56, (0.50, (0.29,
1.17) 1.22) 0.75) 0.95) 0.95) 1.23) 1.02) 0.87) 0.76) 1.02) 1.37) 0.93)
15 17 38 23 14 15 45 268 28 35 15 16
0.71 0.86 0.68 0.79 0.66 1.79 1.45 1.11 0.98 0.83 1.11 0.91
(0.40, (0.50, (0.48, (0.50, (0.36, (1.00, (1.06, (0.98, (0.65, (0.58, (0.62, (0.52,
1.18) 1.38) 0.94) 1.19) 1.10) 2.95) 1.94) 1.25) 1.42) 1.15) 1.83) 1.48)
56 52 110 74 54 28 120 687 66 121 41 40
0.67 0.75 0.56 0.67 0.68 0.94 1.01 0.79 0.65 0.77 0.80 0.66
(0.51, (0.56, (0.46, (0.53, (0.51, (0.62, (0.84, (0.74, (0.50, (0.64, (0.58, (0.47,
0.87) 0.98) 0.68) 0.85) 0.88) 1.35) 1.21) 0.86) 0.82) 0.91) 1.09) 0.90)
Cerebrovascular disease (i.e. stroke) COPDb Chronic Renal Disease Ischemic Heart Disease
17 24 2 62
0.73 0.78 0.26 0.50
(0.43, (0.50, (0.03, (0.38,
1.17) 1.16) 0.94) 0.64)
42 56 5 231
0.52 0.52 0.19 0.53
(0.37, (0.39, (0.06, (0.46,
0.70) 0.68) 0.45) 0.60)
88 118 18 428
0.62 0.70 0.41 0.61
(0.50, (0.58, (0.24, (0.55,
0.77) 0.84) 0.64) 0.67)
79 124 21 364
0.73 1.00 0.64 0.70
(0.58, (0.83, (0.40, (0.63,
0.91) 1.19) 0.98) 0.77)
226 322 46 1085
0.64 0.75 0.41 0.61
(0.56, (0.67, (0.30, (0.57,
0.73) 0.83) 0.55) 0.64)
a b
Models adjusted for age (five-year categories), race, and birth year (five-year categories). Non-Hodgkin's L: Non-Hodgkin's Lymphoma, COPD: Chronic Obstructive Pulmonary Disease. 4
Environmental Research 177 (2019) 108625
V. Barry and K. Steenland
associated outcomes related to inorganic lead (IWGotEoCRt, 2006) and other studies since then have also shown associations between lead and brain cancer (Steenland Kyle et al., 2019; Liao et al., 2016; Wu et al., 2012). There is nothing we know of that suggestively links lead to nonhodgkin's disease mortality. One study found possible increased risk of non-hodgkin lymphoma mortality among populations residing near refineries that emit chrome, lead, nickel, zinc, arsenic, cadmium, benzene, dioxins and/or furans (Ramis et al., 2012). Although trend p-values were not significant, results suggest a possible increased risk of lead on mortality due to rectal cancer and chronic renal disease. Both showed significant or borderline significant associations when comparing workers in the highest lead category (≥40 μg/dL) to the lowest (0 to < 25 μg/dL). Two studies have shown a relationship between exposure to a tetraethyl manufacturing area (which exposes workers to both organic and inorganic lead compounds) and rectal cancer (Steenland Kyle et al., 2019; Fayerweather et al., 1997). Others have shown associations between lead and increased end-stage renal disease risk or reduced kidney function (Chowdhury et al., 2014b; Fadrowski et al., 2010; Muntner et al., 2003). Incidence of chronic renal disease incidence is a better end point than mortality, and would contribute more cases to the analysis; this outcome will be examined in a separate manuscript. This study has several limitations. First, we measured exposure using blood lead level which reflects only recent exposure (i.e. previous few months) to lead and does not allow characterization of long-term or cumulative exposure. However, in a recent study using a subset of this cohort, past maximum blood lead was correlated with current (i.e. 2017–2018) bone lead (R2 = 0.27, p < 0.01) (which is a measure of cumulative exposure) and our broad exposure categories are likely to be accurate (Barry et al., 2019). Second, we do not have socioeconomic status (SES) of each worker. Like smoking, SES may confound the association between lead and mortality, especially in external analyses. It is likely that our internal analyses (which compare workers to workers) may be less susceptible to this confounding. Other mortality outcomes known to be related to SES like alcoholism, cirrhosis, injury, and transportation accidents showed no excess risk of death for the cohort workers compared to national rates suggesting that SES may not be related to blood lead level (and thus not a confounder) in this cohort. We did not have race for the majority of workers. We did impute race and found virtually identical results compared to when we assumed most workers with missing race where white. Finally, we examined mortality and not incidence so we were likely not able to detect any associations (assuming there were any) between lead with cancers or other diseases that tend to not be fatal (i.e. thyroid cancer mortality).
The increased risk of lung and larynx cancer mortality show consistency in both internal and external analyses. There are previously reported associations between lead with both lung cancer (Anttila et al., 1995; Fu and Boffetta, 1995; Lundstrom et al., 1997) and larynx cancer incidence (Bayer et al., 2016; Steenland Kyle et al., 2019). Like other studies, we can not exclude the possibility of interaction or confounding by occupational-related carcinogenic co-exposures (like arsenic or engine exhaust) (Englyst et al., 2001). However, cohort workers worked in various industries/jobs and were presumably not all exposed to the same co-exposures making it less likely that one specific co-exposure would fully explain our results. Smoking may also confound the associations found between lead and lung/larynx cancer mortality or interact with lead to increase lung and larynx cancer risk. However, if lung and larynx cancer mortality findings were due to smoking, we would have expected similar associations with esophageal and kidney cancer mortality (both of which are typically associated with smoking). Additionally, internal analyses comparing workers to workers are generally less susceptible to confounding by smoking (Kriebel et al., 2004) and internal analyses for lung and larynx cancer mortality show strong effects with blood lead level category (i.e. HRs > 2.5 comparing highest to lowest exposure category for both cancer mortalities). Finally, when a sample of this cohort (n = 211 workers) was interviewed about smoking, there was no association between maximum past blood lead level and pack-years of smoking (Barry et al., 2019). It should also be noted that alcohol is also related to larynx cancer (and has synergistic effects with smoking). We had no data on alcohol, which is potentially a confounder if high lead-exposed workers also drank more. There is another issue with both smoking and alcohol, because both of these can contribute to blood lead levels, although in small amounts (Cezard et al., 1992; Grandjean et al., 1981; Shaper et al., 1982). Hence it is possible that those with high lead exposure smoked and drank more, and that this put them in a high lead category, and this in turn led to an increased larynx (or lung) cancer risk. However, past studies examining the influence of alcohol and smoking on blood lead level show only minor increases in blood lead associated with these factors (e.g. differences of 3 μg/dL), (Grandjean et al., 1981; Shaper et al., 1982) and we think that occupational exposure dwarfs these non-occupational exposures, such that the latter are unlikely to cause workers to be in our higher lead categories. In internal comparisons, we found statistically significant positive trends in COPD and ischemic heart disease mortality with increasing blood lead level as well as increased ischemic heart disease mortality risk among workers with maximum blood lead levels ≥25 μg/dL (compared to < 5 μg/dL). However, our results show no indication of increased COPD or ischemic heart disease mortality risk in external comparisons, possibly as a result of the healthy worker effect (McMichael, 1976; Wen et al., 1983). Abundant research exists demonstrating associations between lead and increases in systolic and diastolic blood pressure and it is presumably through this mechanism that lead may increase heart disease related mortality risk (Navas-Acien et al., 2007; Nawrot et al., 2002; Gambelunghe et al., 2016; Han et al., 2018; Nash et al., 2003). Results from a randomized trial may also support a lead-ischemic heart disease mortality association. That study found that people who had experienced a previous myocardial infarction had a modestly reduced risk of adverse cardiovascular outcomes after regular chelation therapy (Lamas et al., 2013). Only one study (that we know of) demonstrates a similar lead-COPD relationship (Steenland et al., 2017). Although that study combines results from Finland, Great Britain, and the US, lead-COPD results were significant and positive in Finland (personal communication). It is unclear whether COPD findings are true associations or due to confounding by simultaneous exposure to dust. Our results point to an association between increasing lead with increased mortality risk due to malignant brain cancer and non-hodgkin's lymphoma. Our brain cancer finding agrees with IARC's 2006 determination that brain cancer was one of the more strongly
5. Conclusion We found positive trends between blood lead level and all-cause, malignant brain cancer, COPD, ischemic heart disease, larynx cancer, lung cancer, and non-hodgkin's lymphoma mortality as well as suggestions of associations with chronic renal disease and rectal cancer mortality. The cohort was large (n = 58,368), followed for a median of 19 years which allowed for deaths of more rare causes to occur, and included workers with a wide range of blood lead levels including ≥40 μg/dL (which is higher than guidelines allow) and < 5 μg/dL (similar to the general population). Future studies should confirm these associations in other populations. Funding This work was supported by the National Institute for Occupational Safety and Health award R01OH010745-04 (PI Kyle Steenland). Declarations of interest None. 5
Environmental Research 177 (2019) 108625
V. Barry and K. Steenland
Acknowledgements
27. Effect of disodium EDTA chelation regimen on cardiovascular events in patients with previous myocardial infarction: the TACT randomized trial. Jama 309 (12), 1241–1250. Lanphear, B.P., Rauch, S., Auinger, P., Allen, R.W., Hornung, R.W., 2018 Apr. Low-level lead exposure and mortality in US adults: a population-based cohort study. The Lancet Public health 3 (4), e177–e184. Liao, L.M., Friesen, M.C., Xiang, Y.-B., Cai, H., Koh, D.-H., Ji, B.-T., et al., 2016. Occupational lead exposure and associations with selected cancers: the shanghai men's and women's health study cohorts. Environ. Health Perspect. 124 (1), 97–103. Lundstrom, N.G., Nordberg, G., Englyst, V., Gerhardsson, L., Hagmar, L., Jin, T., et al., 1997 Feb. Cumulative lead exposure in relation to mortality and lung cancer morbidity in a cohort of primary smelter workers. Scand. J. Work Environ. Health 23 (1), 24–30. Lustberg, M., Silbergeld, E., 2002. Blood lead levels and mortality. JAMA Internal Medicine 162 (21), 2443–2449. McMichael, A.J., 1976 Mar. Standardized mortality ratios and the "healthy worker effect": scratching beneath the surface. J. Occup. Med. : Off. Publ. Ind. Med. Assoc. 18 (3), 165–168. Menke, A., Muntner, P., Batuman, V., Silbergeld, E.K., Guallar, E., 2006 Sep 26. Blood lead below 0.48 micromol/L (10 microg/dL) and mortality among US adults. Circulation 114 (13), 1388–1394. Muntner, P., He, J., Vupputuri, S., Coresh, J., Batuman, V., 2003 Mar. Blood lead and chronic kidney disease in the general United States population: results from NHANES III. Kidney Int. 63 (3), 1044–1050. Nash, D., Magder, L., Lustberg, M., Sherwin, R.W., Rubin, R.J., Kaufmann, R.B., et al., 2003 Mar 26. Blood lead, blood pressure, and hypertension in perimenopausal and postmenopausal women. Jama 289 (12), 1523–1532. Navas-Acien, A., Guallar, E., Silbergeld, E.K., Rothenberg, S.J., 2007. Lead exposure and cardiovascular disease–a systematic review. Environ. Health Perspect. 115 (3), 472–482. Nawrot, T.S., Thijs, L., Den Hond, E.M., Roels, H.A., Staessen, J.A., 2002 Feb. An epidemiological re-appraisal of the association between blood pressure and blood lead: a meta-analysis. J. Hum. Hypertens. 16 (2), 123–131. Pirkle, J.L., Brody, D.J., Gunter, E.W., Kramer, R.A., Paschal, D.C., Flegal, K.M., et al., 1994 Jul 27. The decline in blood lead levels in the United States. The national health and nutrition examination surveys (NHANES). Jama 272 (4), 284–291. Ramis, R., Diggle, P., Boldo, E., Garcia-Perez, J., Fernandez-Navarro, P., Lopez-Abente, G., 2012 Feb 6. Analysis of matched geographical areas to study potential links between environmental exposure to oil refineries and non-Hodgkin lymphoma mortality in Spain. Int. J. Health Geogr. 11, 4. Schober, S.E., Mirel, L.B., Graubard, B.I., Brody, D.J., Flegal, K.M., 2006 Oct. Blood lead levels and death from all causes, cardiovascular disease, and cancer: results from the NHANES III mortality study. Environ. Health Perspect. 114 (10), 1538–1541. Schubauer-Berigan, M.K., Hein, M.J., Raudabaugh, W.M., Ruder, A.M., Silver, S.R., Spaeth, S., et al., 2011 Dec. Update of the NIOSH life table analysis system: a personyears analysis program for the windows computing environment. Am. J. Ind. Med. 54 (12), 915–924. Schwartz, B.S., Hu, H., 2007 Mar. Adult lead exposure: time for change. Environ. Health Perspect. 115 (3), 451–454. Shaper, A.G., Pocock, S.J., Walker, M., Wale, C.J., Clayton, B., Delves, H.T., et al., 1982 Jan 30. Effects of alcohol and smoking on blood lead in middle-aged British men. Br. Med. J. 284 (6312), 299–302. Sirivarasai, J., Kaojarern, S., Chanprasertyothin, S., Panpunuan, P., Petchpoung, K., Tatsaneeyapant, A., et al., 2015. Environmental lead exposure, catalase gene, and markers of antioxidant and oxidative stress relation to hypertension: an analysis based on the EGAT study. BioMed Res. Int. 2015, 856319. Spivey, A., 2007. The weight of lead. Effects add up in adults. Environ. Health Perspect. 115 (1), A30–A36. Staudinger, K.C., Roth, V.S., 1998 Feb 15. Occupational lead poisoning. Am. Fam. Physician 57 (4), 719–726 31-2. Steenland, K., Barry, V., Anttila, A., Sallmen, M., McElvenny, D., Todd, A.C., et al., 2017 Nov. A cohort mortality study of lead-exposed workers in the USA, Finland and the UK. Occup. Environ. Med. 74 (11), 785–791. Steenland Kyle, B.V., Ahti, Anttila, Markku, Sallmen, Mueller, William, Peter, Ritchie, Damien, McElvenny, Kurt, Straif, June 2019. Cancer incidence among workers with blood lead measurements in two countries. Occup. Environ. Med (in press) as of. Tsai, P.L., Hatfield, T.H., 2011. Global benefits from the phaseout of leaded fuel. J. Environ. Health 74 (5), 8–15. Tsoi, M.F., Cheung, C.L., Cheung, T.T., Cheung, B.M., 2016 Nov. Continual decrease in blood lead level in Americans: United States national health nutrition and examination survey 1999-2014. Am. J. Med. 129 (11), 1213–1218. Wen, C.P., Tsai, S.P., Gibson, R.L., 1983 Apr. Anatomy of the healthy worker effect: a critical review. J. Occup. Med. : Off. Publ. Ind. Med. Assoc. 25 (4), 283–289. Wu, W.T., Lin, Y.J., Liou, S.H., Yang, C.Y., Cheng, K.F., Tsai, P.J., et al., 2012 Apr. Brain cancer associated with environmental lead exposure: evidence from implementation of a National Petrol-Lead Phase-Out Program (PLPOP) in Taiwan between 1979 and 2007. Environ. Int. 40, 97–101.
This work was supported by the National Institute for Occupational Safety and Health award R01OH010745-04 (PI Kyle Steenland). We would like to acknowledge the following people for their help in providing data access and matching: Paula Guro (USRDS), Michelle Goodier (NDI), Lillian Ingster (NDI), Velida Juzbasic (NDI), Kenneth D. Rosenman (Michigan State University), Joanna Kica (Michigan State University), Carrie Tomasallo (Wisconsin Division of Public Health), and Michael Flanagan (Massachusetts Dept of Labor Standards). Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.envres.2019.108625. References Abadin, H., Ashizawa, A., Stevens, Y.W., Llados, F., Diamond, G., Sage, G., et al., 2007. Agency for Toxic Substances and Disease Registry (ATSDR) Toxicological Profiles. Toxicological Profile for Lead. Agency for Toxic Substances and Disease Registry (US), Atlanta (GA). Adult blood lead epidemiology and surveillance–United States, 2003-2004. MMWR Morb. Mortal. Wkly. Rep. 55 (32), 876–879. Anttila, A., Heikkila, P., Pukkala, E., Nykyri, E., Kauppinen, T., Hernberg, S., et al., 1995 Dec. Excess lung cancer among workers exposed to lead. Scand. J. Work Environ. Health 21 (6), 460–469. Barry, V., Todd, A.C., Steenland, K., 2019 May. Bone lead associations with blood lead, kidney function and blood pressure among US, lead-exposed workers in a surveillance programme. Occup. Environ. Med. 76 (5), 349–354. Bayer, O., Camara, R., Zeissig, S.R., Ressing, M., Dietz, A., Locati, L.D., et al., 2016 Jan. Occupation and cancer of the larynx: a systematic review and meta-analysis. Eur. Arch. Oto-Rhino-Laryngol. 273 (1), 9–20 official journal of the European Federation of Oto-Rhino-Laryngological Societies (EUFOS) : affiliated with the German Society for Oto-Rhino-Laryngology - Head and Neck Surgery. Cezard, C., Demarquilly, C., Boniface, M., Haguenoer, J.M., 1992 Sep. Influence of the degree of exposure to lead on relations between alcohol consumption and the biological indices of lead exposure: epidemiological study in a lead acid battery factory. Br. J. Ind. Med. 49 (9), 645–647. Chowdhury, R., Sarnat, S.E., Darrow, L., McClellan, W., Steenland, K., 2014 Jula. Mortality among participants in a lead surveillance program. Environ. Res. 132, 100–104. Chowdhury, R., Darrow, L., McClellan, W., Sarnat, S., Steenland, K., 2014 Julb. Incident ESRD among participants in a lead surveillance program. Am. J. Kidney Dis. 64 (1), 25–31 the official journal of the National Kidney Foundation. Englyst, V., Lundstrom, N.G., Gerhardsson, L., Rylander, L., Nordberg, G., 2001 Jun 12. Lung cancer risks among lead smelter workers also exposed to arsenic. Sci. Total Environ. 273 (1–3), 77–82. Fadrowski, J.J., Navas-Acien, A., Tellez-Plaza, M., Guallar, E., Weaver, V.M., Furth, S.L., 2010. Blood lead level and kidney function in US adolescents: the third national health and nutrition examination survey. Arch. Intern. Med. 170 (1), 75–82. Fayerweather, W.E., Karns, M.E., Nuwayhid, I.A., Nelson, T.J., 1997 Jan. Case-control study of cancer risk in tetraethyl lead manufacturing. Am. J. Ind. Med. 31 (1), 28–35. Fu, H., Boffetta, P., 1995. Cancer and occupational exposure to inorganic lead compounds: a meta-analysis of published data. Occup. Environ. Med. 52 (2), 73–81. Gambelunghe, A., Sallsten, G., Borne, Y., Forsgard, N., Hedblad, B., Nilsson, P., et al., 2016 Aug. Low-level exposure to lead, blood pressure, and hypertension in a population-based cohort. Environ. Res. 149, 157–163. Grandjean, P., Olsen, N.B., Hollnagel, H., 1981. Influence of smoking and alcohol consumption on blood lead levels. Int. Arch. Occup. Environ. Health 48 (4), 391–397. Han, L., Wang, X., Han, R., Xu, M., Zhao, Y., Gao, Q., et al., 2018. Association between blood lead level and blood pressure: an occupational population-based study in Jiangsu province, China. PLoS One 13 (7), e0200289. Holland, M.G., Cawthon, D., 2016 Dec. Workplace lead exposure. J. Occup. Environ. Med. 58 (12), e371–e374. IWGotEoCRt, Humans, 2006. Evaluation of carcinogenic risk to humans, inorganic and organic lead compounds: international agency for research on cancer. IARC monographs on the evaluation of carcinogenic risks to humans. Available from: https:// www.ncbi.nlm.nih.gov/books/NBK321297/ 87. Kriebel, D., Zeka, A., Eisen, E.A., Wegman, D.H., 2004 Oct. Quantitative evaluation of the effects of uncontrolled confounding by alcohol and tobacco in occupational cancer studies. Int. J. Epidemiol. 33 (5), 1040–1045. Lamas, G.A., Goertz, C., Boineau, R., Mark, D.B., Rozema, T., Nahin, R.L., et al., 2013 Mar
6
Contents lists available at ScienceDirect
Environmental Research journal homepage: www.elsevier.com/locate/envres
Lead exposure and mortality among U.S. workers in a surveillance program: Results from 10 additional years of follow-up
T
Vaughn Barry*, Kyle Steenland Department of Environmental Health, Rollins School of Public Health at Emory University, Atlanta, GA, USA
A R T I C LE I N FO
A B S T R A C T
Keywords: Lead Blood lead Mortality Lung cancer Occupation
Background: A cohort of male lead-exposed workers with past blood lead levels, previously followed for mortality over 12 years, has now been followed for an additional 10 years. This has doubled the number of deaths and allowed for examination of mortality outcomes across a wide range of blood lead levels. Objective: Evaluate association between lead exposure and 16 causes of death. Methods: The cohort included male workers from 11 U.S. states enrolled in a U.S. lead surveillance program. Maximum blood lead level for each worker was abstracted from surveillance records. Mortality was assessed using the National Death Index. We conducted internal analyses via Cox regression adjusting for age, calendar time, and race. External analyses compared cohort mortality rates with those of the U.S. population. Blood lead categories were defined as 0- < 5, 5- < 25, 25- < 40, and ≥40 μg/dL with the two lower categories combined for outcomes with < 5 deaths in the 0- < 5 group. Results: The cohort (n = 58,368) was followed for a median of 19 years and experienced 6,527 deaths. Average maximum blood lead was 25.9 μg/dL and mean year of first blood lead test was 1997. Strong associations were found between blood lead level with larynx and lung cancer mortality. For these outcomes, hazard ratios and 95% confidence intervals across blood lead categories were 1.0 (ref), 1.1 (0.4–3.2), 3.4 (1.3–9.1) for larynx and 1.0 (ref), 1.6 (1.0–2.5), 2.0 (1.3–3.1), 2.9 (1.9–4.5) for lung (trend p-values = 0.08 and < 0.01, respectively). Positive significant trends were also seen for mortality from brain cancer, chronic obstructive pulmonary disease, ischemic heart disease, and non-hodgkin's lymphoma. Findings suggested associations with chronic renal disease and rectal cancer mortality, although trends were not statistically significant. Conclusions: The additional follow up confirmed previous relationships between lead and mortality and also detected new associations.
1. Introduction Although environmental exposure to lead in the U.S. has decreased substantially since the 1970s with the elimination of lead from gasoline and paint (Pirkle et al., 1994; Tsai and Hatfield, 2011; Tsoi et al., 2016), there remain substantial numbers of US workers exposed occupationally. It is estimated that approximately 800,000 US workers in general industry and an additional 800,000 workers in construction are currently exposed to lead on the job (https://www.osha.gov/SLTC/lead/). An additional number of adults in the US have been occupationally exposed to lead in the past and remain susceptible to lead-induced adverse health effects (Staudinger and Roth, 1998). Current US Occupational Safety and Health Administration standards require workers be removed from lead exposure when their blood
lead level reaches 50 μg/dL (construction workers) or 60 μg/dL (other workers), and to not return until their blood lead drops below 40 μg/dL. However, there is concern that current standards are outdated and do not adequately protect workers’ health (Holland and Cawthon, 2016; Schwartz and Hu, 2007; Spivey, 2007) with some suggesting that workers be removed from lead exposure when blood lead levels reach 20 μg/dL (Lustberg and Silbergeld, 2002). Additional concerns exist about possible harmful health impacts of lead in adults at even lower levels. For example, studies show risks associated with mortality and chronic disease endpoints even among blood lead levels < 5 μg/dL (Lanphear et al., 2018; Navas-Acien et al., 2007; Sirivarasai et al., 2015). Health effects of lead have been observed in all organ systems over a wide range of lead levels (Abadin et al., 2007). Most general population
* Corresponding author. MPH Department of Environmental Health Rollins School of Public Health at Emory University, 1518 Clifton Road, CNR building, room 2028, Atlanta, GA, 30322, USA. E-mail address: [email protected] (V. Barry).
https://doi.org/10.1016/j.envres.2019.108625 Received 30 May 2019; Received in revised form 1 August 2019; Accepted 2 August 2019 Available online 05 August 2019 0013-9351/ © 2019 Elsevier Inc. All rights reserved.
Environmental Research 177 (2019) 108625
V. Barry and K. Steenland
workers whose highest blood lead level was either < 5 μg/dL or 5 to < 25 μg/dL were randomly selected proportionally across states within blood lead category (Chowdhury et al., 2014a). The final study cohort consisted of 58,368 workers.
studies examining blood lead effects have focused on cardiovascular morbidity and show that increasing lead levels are related to increased systolic and diastolic blood pressure as well as increased risk of hypertension, heart disease, atherosclerosis, and cardiac disease across a range of blood lead levels (Navas-Acien et al., 2007; Nawrot et al., 2002). Chronic lead exposure in adults is also associated with mortality, and in particular, cardiovascular mortality (Lustberg and Silbergeld, 2002; Lanphear et al., 2018; Menke et al., 2006; Schober et al., 2006) although results have not been consistent in demonstrating what specific blood lead levels cause increased mortality risk. Lead is designated a “probable human carcinogen” by the International Agency for Research on Cancer (IWGotEoCRt, 2006) and as a compound that is “reasonably anticipated” to be carcinogenic to humans by the National Toxicology Program (http://ntp.niehs.nih.gov/ ntp/roc/content/profiles/lead.pdf), with the sites of interest including primarily cancers of the lung and stomach, but also to some extent kidney and brain. Most studies examining the relationship between lead exposure and cancer risk have been occupational with much higher exposure levels than general population studies. However, few data exist on incidence, and mortality data are often insufficient to provide information on dose-response relationships. Mortality and cancer outcomes, in particular, often require both lengthy follow-up due to exposure lag time and sufficient numbers of cases to examine less common cancers/outcomes. We previously followed a large cohort of lead-exposed workers (n = 58,368) for mortality over 12 years and found excess lung cancer mortality among those with the highest blood lead levels (Chowdhury et al., 2014a). Cause-specific mortality was challenging to examine due to few deaths as the cohort was relatively young (average age = 51 years) and healthy (6% mortality) at the time. We now have an additional 10 years of follow-up on this cohort which has doubled the number of deaths and allowed for examination of long-term causespecific mortality outcomes across a wide range of blood lead levels. The objective of this study is to evaluate the association between lead exposure and mortality to confirm the previous association with lung cancer mortality and to detect any new associations among 16 specific causes of death.
2.2. Variables Name, date of birth, gender, date of each blood lead test, and blood lead level (in μg/dL) for each blood lead test were collected from the selected workers in each of the eleven U.S. states. We categorized each person in to one of four blood lead categories based on the highest blood lead level they had recorded in ABLES: < 5, 5 to < 25, 25 to < 40, or ≥40 μg/dL. Additional but incomplete data on race (69% missing) and social security number (74% missing) were also collected. We classified those with missing race as white since 86% of subjects with known race were white. We used name, date of birth, gender, state of residence at the time of ABLES test(s), race (when available), and social security number (when available) for matching with the National Death Index (NDI) through the end of 2017. We defined a match as any worker found in the NDI with an NDI status code of 1 (indicating a high probability of a true match). Among workers who died, we used the ICD9/10 code indicated as the primary cause of death to determine cause-specific mortality (Supplementary Table 1). We present results for 16 specific causes of death. 2.3. Statistical analysis Cox proportional hazards regression was used to calculate hazard ratios (HRs) associated with each lead category within the cohort (i.e. internal analyses) after adjusting for birth year (five-year categories) and race, using age as the time scale. Person-time at risk began at time of the first blood lead ABLES test and continued either to date of death or December 31st, 2017, whichever occurred first. HRs compared risk at each of the blood lead categories to the reference category of < 5 μg/dL or 0 to < 25 μg/dL (depending on the numbers of cases in lowest exposure category). HRs were generated using SAS. P-values to describe trends were generated for each cause of death with Cox proportional hazards regression models that used logged continuous blood lead as the exposure and adjusted for calendar time period and race with age as the time scale. The NIOSH Life Table Analysis System was then used to calculate cause-specific rates of death for the cohort and to compare rates with those of the U.S. population via standardized mortality ratios (SMRs). SMRs were adjusted for age (in five-year categories), race, and five-year calendar time period (i.e. birth year) (Schubauer-Berigan et al., 2011). Sensitivity analyses for the Cox proportional hazards regression models were conducted using imputed race. Specifically, race was imputed 5 times via simulations (using Proc MI in SAS), 5 separate regression models were run, and coefficients and standard errors were averaged (see supplemental info from (Chowdhury et al., 2014b)). Predictor variables in the imputation models included blood lead category, state, year of birth, vital status, and year of first lead test. The models correctly predicted race for 69% of workers with known race. The study was approved by the Emory University Institutional Review Board.
2. Materials and methods 2.1. Data sources/study cohort The cohort is a subset of workers who were enrolled in the Adult Blood Lead Surveillance (ABLES) program which is sponsored by the National Institute for Occupational Safety and Health (NIOSH). ABLES has tracked laboratory-reported cases of elevated blood lead levels in U.S. adults since 1987. Initially, ABLES gathered data only on people with blood lead levels > 25 μg/dL but subsequently began to collect data on people with lower levels, too. Blood lead tests were conducted primarily in response to occupational exposure; during the years 2003–2004, 94% of adults with blood lead levels reported to ABLES were exposed via occupational sources (Adult blood lead epidemio, 2006). ABLES coverage increased from 4 U.S. states in 1987 to 41 states in 2012. The underlying cohort consisted of male workers from 11 U.S. states who had at least one blood lead level reported to ABLES anytime before 2005. States included California, Connecticut, Iowa, Massachusetts, Michigan, Minnesota, New Jersey, New York, Ohio, Pennsylvania, and Wisconsin and were chosen because they had the most subjects with blood lead data, and data that went back the farthest in time (due to the ABLES program expanding differently across states over time). We then excluded workers who were tested for the first time after the age of 70 years, before the age of 18 years, or who had implausible blood lead levels (> 250 μg/dL). For efficiency and cost considerations, a subset was then identified from this group of workers. All workers who ever had a blood lead level ≥25 μg/dL were included in the subset while
3. Results Cohort characteristics are described in Table 1. Among the 58,368 male workers in the cohort, 80% were white race (83% white using average of imputed race). Median year of birth was 1959 (range = 1914–1993 years). Workers were 39 years old, on average, at the time of their first blood lead test (mean year of first blood lead test was 1997) and were followed for mortality through 2017. Median follow up for the cohort was 19 years and there were a total of 6,527 deaths (11%) during the follow-up time period. Nearly half of the cohort (47%) was from either New York or California with the rest from 2
Environmental Research 177 (2019) 108625
V. Barry and K. Steenland
Table 1 Cohort demographics. Characteristics
Highest Lead Category Achieved
Number in cohort Median years of follow-up Mean age at first test in years Race White Non-white Missing/unknown Race imputeda White Non-white Median number blood lead tests in those with > 1 test Number with single tests only Mean highest blood lead level % with SSN (for matching)b Median year of first lead test Median year of birth Median year of death Number deaths
1 [0 to < 5 μg/dl]
2 [5 to < 25 μg/dl]
3 [25 to < 40 μg/dl]
4 [40+ μg/dl]
TOTAL
6,848 (12%) 13.5 40.7
18,650 (32%) 17.1 40.0
21,448 (37%) 21.1 37.9
11,422 (20%) 23.6 38.3
58,368 19.0 39.0
1,448 (21%) 252 (4%) 5,148 (75%)
2,356 (13%) 558 (3%) 15,736 (84%)
6,246 (29%) 1,673 (8%) 13,529 (63%)
4,339 (38%) 1,200 (10%) 5,883 (52%)
14,389 (25%) 3,683 (6%) 40,296 (69%)
6,026 (88%) 806 (12%) 2 6,124 (89%) 2.5 611 (9%) 2004 1962 2012 450 (7%)
15,014 (81%) 3,604 (19%) 3 12,739 (68%) 13.0 2,084 (11%) 2000 1961 2012 1,503 (8%)
17,898 (83%) 3,542 (17%) 4 7,786 (36%) 30.8 7,664 (36%) 1996 1959 2010 2,456 (11%)
9,478 1,940 6 1,940 52.0 4,883 1992 1955 2009 2,118
48,357 (83%) 9,891 (17%) 4 28,589 (49%) 25.9 15,242 (26%) 1998 1959 2010 6,527 (11%)
(83%) (17%) (17%) (43%)
(19%)
a
Race was imputed 5 times via simulations and then averaged. Three states (MA, MI, and WI) sent their own data to the National Death Index for matching and then sent us de-identified data, without SSN; hence these percentages are underestimates. b
with increasing maximum blood lead (Table 2). HRs and 95% confidence intervals (CIs) for all-cause, lung cancer, and COPD mortality across increasing blood lead categories compared to the 0–5 μg/dL category were 1.0 (0.9–1.1), 1.1 (1.0–1.2), and 1.4 (1.2–1.5) for all-cause; 1.6 (1.0–2.5), 2.0 (1.3–3.1), and 2.9 (1.9–4.5) for lung cancer; and 0.7 (0.4–1.1), 1.0 (0.6–1.5), and 1.5 (0.9–2.3) for COPD mortality with significant trend p-values for all three (all p < 0.01). Ischemic heart disease mortality risk also increased with increasing blood lead (trend p-value = 0.05) with HRs and 95% CIs across blood lead categories compared to 0 to < 5 μg/dL of 1.1 (0.8–1.4), 1.3 (1.0–1.7), and 1.5 (1.2–2.0). In internal analyses that compared workers with maximum blood lead of 25 to < 40 and ≥ 40 to those with 0–25 μg/dL (used for rarer outcomes with < 5 deaths in the 0 to < 5 μg/dL referent group), there was a significant increased risk of mortality from larynx cancer among workers with maximum blood lead ≥40 compared to those with 0–25 μg/dL (HR = 3.4, 95% CI = 1.3–9.1)) (Table 3). P-values from continuous models for malignant brain cancer, larynx cancer, and nonhodgkin's lymphoma mortality were all borderline significant (p = 0.06, 0.08, and 0.06, respectively). HRs and 95% CIs comparing
the other 9 states. Half of the cohort (49%) had only one blood lead test. Workers with the lowest blood lead level(s) were more likely to have had only one blood lead test compared to workers with higher blood lead levels (89% of workers with a lead level of 0 to < 5 μg/dL had only one lead test compared to 17% of workers with a lead level ≥40 μg/dL). Among workers with more than one blood lead test, the median number of blood lead tests per person was 4. Workers with higher maximum blood lead levels were also followed up longer than those with lower maximum blood lead levels (median of 24 vs. 14 years comparing workers with maximum lead level ≥40 μg/dL to those with maximum lead level 0 to < 5 μg/dL), more likely to be non-white race (22% vs. 15% comparing highest exposed category to lowest exposed), and more likely to die (19% vs. 7% comparing highest exposed category to lowest). Average maximum blood lead level, mean year of birth, and mortality rate did not vary across state. In internal analyses examining the cause-specific deaths that had enough cases to allow comparisons across four blood lead exposure categories (0 to < 5, 5 to < 25, 25 to < 40 and ≥ 40 μg/dL), there was significant increased risk of all-cause, lung cancer, and COPD mortality
Table 2 Hazard ratios (HRs)a and 95% confidence intervals (CIs) by lead category, with 0 to < 5 μg/dl as referent group. Cause of death
p-value for trendb
Highest Lead Category Achieved 0 to < 5 μg/dl
5 to < 25 μg/dl
25 to < 40 μg/dl
40+ μg/dl
N
HR (95% CI)
N
HR (95% CI)
N
HR (95% CI)
N
HR (95% CI)
All causes Cancer Colon Esophagus Kidney Liver Lung Stomach
450
Ref
1503
1.00 (0.90–1.11)
2456
1.09 (0.98–1.20)
2118
1.38 (1.24–1.53)
< 0.01
8 8 5 11 24 5
Ref Ref Ref Ref Ref Ref
21 15 16 30 134 6
0.77 0.57 0.94 0.86 1.61 0.33
(0.34–1.74) (0.24–1.34) (0.35–2.58) (0.43–1.72) (1.04–2.48) (0.10–1.09)
43 28 19 34 261 13
0.98 0.77 0.80 0.73 2.03 0.39
(0.45–2.10) (0.35–1.70) (0.30–2.17) (0.37–1.44) (1.34–3.10) (0.14–1.12)
38 23 14 45 268 16
1.19 0.97 0.92 1.53 2.92 0.64
(0.54–2.61) (0.43–2.20) (0.32–2.58) (0.79–2.99) (1.91–4.46) (0.22–1.82)
0.98 0.71 0.74 0.23 < 0.01 0.89
Cerebrovascular disease (i.e. stroke) COPD Ischemic heart disease
17 24 62
Ref Ref Ref
42 56 231
0.72 (0.41–1.26) 0.68 (0.42–1.10) 1.08 (0.82–1.43)
88 118 428
0.91 (0.54–1.53) 0.98 (0.63–1.52) 1.28 (0.98–1.67)
79 124 364
1.11 (0.65–1.90) 1.46 (0.94–2.28) 1.51 (1.15–1.99)
0.92 < 0.01 0.05
a b
Proportional hazards regression models adjusted for birth year (five-year categories) and race with age as timescale. From proportional hazards regression model with continuous blood lead as exposure (adjusted for calendar time period and race with age as time scale). 3
Environmental Research 177 (2019) 108625
V. Barry and K. Steenland
Table 3 Hazard ratios (HRs)a and 95% confidence intervals (CIs) by lead category, with 0 to < 25 μg/dl as referent group. Cause of death
p-value for trendb
Highest Lead Category Achieved 0 to < 25 μg/dl
25 to < 40 μg/dl
40+ μg/dl
N
HR (95% CI)
N
HR (95% CI)
N
HR (95% CI)
Cancer Brain cancer Bladder cancer Larynx cancer Non-Hodgkin's L Pancreatic cancer Rectal cancer
15 13 6 18 39 9
Ref Ref Ref Ref Ref Ref
26 25 7 20 47 17
1.67 1.60 1.05 0.84 1.07 1.61
(0.88–3.17) (0.81–3.16) (0.35–3.15) (0.44–1.62) (0.69–1.64) (0.71–3.65)
15 19 15 28 35 15
1.49 1.71 3.42 1.60 1.15 2.06
(0.71–3.12) (0.83–3.55) (1.29–9.09) (0.85–3.01) (0.72–1.85) (0.87–4.84)
0.06 0.68 0.08 0.06 0.15 0.50
Chronic renal disease
14
Ref
24
1.27 (0.65–2.49)
25
1.81 (0.91–3.57)
0.49
a b
Proportional hazards regression models adjusted for birth year (five-year categories) and race with age as timescale. From proportional hazards regression model with continuous blood lead as exposure (adjusted for calendar time period and race with age as time scale).
workers with blood lead ≥40 to those with 0–25 μg/dL for brain cancer, larynx cancer, and non-hodgkin's lymphoma were 1.5 (0.7–3.1), 3.4 (1.3–9.1), and 1.6 (0.9–3.0), respectively. Additionally, although trend p-values were not significant, there was evidence of increasing chronic renal disease mortality risk across blood lead category (HR = 1.3 (0.7–2.5) and 1.8 (0.9–3.6)) as well as rectal cancer mortality (HR = 1.6 (0.7–3.7) and 2.1 (0.9–4.8)) comparing workers with blood lead levels of 25 to < 40 and ≥ 40 compared to 0 to < 25 μg/dL. Overall SMRs comparing the cohort's risk of all-cause and causespecific mortality to the general U.S. population's risk demonstrated a decreased risk of death among the cohort with most SMRs < 1.0 (Table 4). However, there were a significant excess number of larynx, liver, and lung cancer deaths among workers who had maximum blood lead levels ≥40 μg/dL (observed/expected for larynx = 15/8.4, liver = 45/31.1, and lung = 268/242.1). Among workers with maximum blood lead level ≥40 μg/dL, there was no excess mortality compared to the external U.S. population with respect to deaths caused by other cancers, cerebrovascular disease, chronic obstructive pulmonary disease (COPD), chronic renal disease, or ischemic heart disease. Sensitivity analyses using imputed race resulted in virtually
identical results with HRs and 95% confidence intervals the same out to at least the first decimal place for nearly all mortality outcomes. The one exception was chronic renal disease mortality which showed minor differences between our applied race definition (HR = 1.3 (0.7–2.5) and HR = 1.8 (0.9–3.6) for the two highest lead categories compared to 0 to < 25 μg/dL) and the imputed race definition (HR = 1.5 (0.8–3.0) and HR = 2.1 (1.1–4.3)). 4. Discussion Similar to the findings from the earlier follow up of this cohort (Chowdhury et al., 2014a), we found significant increasing risk of allcause, lung cancer, COPD, and ischemic heart disease mortality risk among workers with increasing blood lead levels. With the additional 10 years of follow up which resulted in more deaths and thus more power, we were able to detect new associations between blood lead with mortality from malignant brain cancer, larynx cancer, and nonhodgkin's lymphoma. There was also new evidence of possible relationships between blood lead with mortality due to chronic renal disease and rectal cancer among workers with the highest blood lead levels (i.e. ≥40 μg/dL).
Table 4 Standardized Mortality Ratios (SMRs)a and 95% confidence intervals (CIs) by lead category (n = 58,368). Cause of Death
Highest Lead Category Achieved 1 [0 to < 5 μg/dl]
2 [5 to < 25 μg/dl]
3 [25 to < 40 μg/dl]
4 [40+ μg/dl]
TOTAL
N
SMR (95% CI)
N
SMR (95% CI)
N
SMR (95% CI)
N
SMR (95% CI)
N
SMR (95% CI)
All causes Cancer Brain Bladder Colon Esophagus Kidney Larynx Liver Lung Non-Hodgkin's Lb Pancreas Rectum Stomach
450
0.64 (0.58, 0.70)
1503
0.63 (0.60, 0.66)
2456
0.66 (0.63, 0.68)
2118
0.80 (0.77, 0.84)
6527
0.69 (0.67, 0.71)
0 1 8 8 5 1 11 24 1 4 3 5
0.00 0.20 0.56 0.91 0.80 0.45 1.13 0.37 0.13 0.32 0.74 1.16
(0.00, (0.01, (0.24, (0.39, (0.26, (0.01, (0.56, (0.24, (0.00, (0.09, (0.15, (0.38,
0.53) 1.10) 1.10) 1.78) 1.87) 2.50) 2.02) 0.56) 0.75) 0.82) 2.16) 2.71)
15 12 21 15 16 5 30 134 17 35 6 6
0.65 0.68 0.43 0.51 0.76 0.66 0.94 0.61 0.66 0.84 0.44 0.41
(0.36, (0.35, (0.26, (0.28, (0.43, (0.21, (0.64, (0.51, (0.38, (0.58, (0.16, (0.15,
1.07) 1.19) 0.65) 0.84) 1.23) 1.54) 1.35) 0.72) 1.05) 1.17) 0.96) 0.89)
26 22 43 28 19 7 34 261 20 47 17 13
0.80 0.81 0.56 0.66 0.61 0.60 0.73 0.77 0.49 0.76 0.85 0.54
(0.52, (0.51, (0.40, (0.44, (0.37, (0.24, (0.51, (0.68, (0.30, (0.56, (0.50, (0.29,
1.17) 1.22) 0.75) 0.95) 0.95) 1.23) 1.02) 0.87) 0.76) 1.02) 1.37) 0.93)
15 17 38 23 14 15 45 268 28 35 15 16
0.71 0.86 0.68 0.79 0.66 1.79 1.45 1.11 0.98 0.83 1.11 0.91
(0.40, (0.50, (0.48, (0.50, (0.36, (1.00, (1.06, (0.98, (0.65, (0.58, (0.62, (0.52,
1.18) 1.38) 0.94) 1.19) 1.10) 2.95) 1.94) 1.25) 1.42) 1.15) 1.83) 1.48)
56 52 110 74 54 28 120 687 66 121 41 40
0.67 0.75 0.56 0.67 0.68 0.94 1.01 0.79 0.65 0.77 0.80 0.66
(0.51, (0.56, (0.46, (0.53, (0.51, (0.62, (0.84, (0.74, (0.50, (0.64, (0.58, (0.47,
0.87) 0.98) 0.68) 0.85) 0.88) 1.35) 1.21) 0.86) 0.82) 0.91) 1.09) 0.90)
Cerebrovascular disease (i.e. stroke) COPDb Chronic Renal Disease Ischemic Heart Disease
17 24 2 62
0.73 0.78 0.26 0.50
(0.43, (0.50, (0.03, (0.38,
1.17) 1.16) 0.94) 0.64)
42 56 5 231
0.52 0.52 0.19 0.53
(0.37, (0.39, (0.06, (0.46,
0.70) 0.68) 0.45) 0.60)
88 118 18 428
0.62 0.70 0.41 0.61
(0.50, (0.58, (0.24, (0.55,
0.77) 0.84) 0.64) 0.67)
79 124 21 364
0.73 1.00 0.64 0.70
(0.58, (0.83, (0.40, (0.63,
0.91) 1.19) 0.98) 0.77)
226 322 46 1085
0.64 0.75 0.41 0.61
(0.56, (0.67, (0.30, (0.57,
0.73) 0.83) 0.55) 0.64)
a b
Models adjusted for age (five-year categories), race, and birth year (five-year categories). Non-Hodgkin's L: Non-Hodgkin's Lymphoma, COPD: Chronic Obstructive Pulmonary Disease. 4
Environmental Research 177 (2019) 108625
V. Barry and K. Steenland
associated outcomes related to inorganic lead (IWGotEoCRt, 2006) and other studies since then have also shown associations between lead and brain cancer (Steenland Kyle et al., 2019; Liao et al., 2016; Wu et al., 2012). There is nothing we know of that suggestively links lead to nonhodgkin's disease mortality. One study found possible increased risk of non-hodgkin lymphoma mortality among populations residing near refineries that emit chrome, lead, nickel, zinc, arsenic, cadmium, benzene, dioxins and/or furans (Ramis et al., 2012). Although trend p-values were not significant, results suggest a possible increased risk of lead on mortality due to rectal cancer and chronic renal disease. Both showed significant or borderline significant associations when comparing workers in the highest lead category (≥40 μg/dL) to the lowest (0 to < 25 μg/dL). Two studies have shown a relationship between exposure to a tetraethyl manufacturing area (which exposes workers to both organic and inorganic lead compounds) and rectal cancer (Steenland Kyle et al., 2019; Fayerweather et al., 1997). Others have shown associations between lead and increased end-stage renal disease risk or reduced kidney function (Chowdhury et al., 2014b; Fadrowski et al., 2010; Muntner et al., 2003). Incidence of chronic renal disease incidence is a better end point than mortality, and would contribute more cases to the analysis; this outcome will be examined in a separate manuscript. This study has several limitations. First, we measured exposure using blood lead level which reflects only recent exposure (i.e. previous few months) to lead and does not allow characterization of long-term or cumulative exposure. However, in a recent study using a subset of this cohort, past maximum blood lead was correlated with current (i.e. 2017–2018) bone lead (R2 = 0.27, p < 0.01) (which is a measure of cumulative exposure) and our broad exposure categories are likely to be accurate (Barry et al., 2019). Second, we do not have socioeconomic status (SES) of each worker. Like smoking, SES may confound the association between lead and mortality, especially in external analyses. It is likely that our internal analyses (which compare workers to workers) may be less susceptible to this confounding. Other mortality outcomes known to be related to SES like alcoholism, cirrhosis, injury, and transportation accidents showed no excess risk of death for the cohort workers compared to national rates suggesting that SES may not be related to blood lead level (and thus not a confounder) in this cohort. We did not have race for the majority of workers. We did impute race and found virtually identical results compared to when we assumed most workers with missing race where white. Finally, we examined mortality and not incidence so we were likely not able to detect any associations (assuming there were any) between lead with cancers or other diseases that tend to not be fatal (i.e. thyroid cancer mortality).
The increased risk of lung and larynx cancer mortality show consistency in both internal and external analyses. There are previously reported associations between lead with both lung cancer (Anttila et al., 1995; Fu and Boffetta, 1995; Lundstrom et al., 1997) and larynx cancer incidence (Bayer et al., 2016; Steenland Kyle et al., 2019). Like other studies, we can not exclude the possibility of interaction or confounding by occupational-related carcinogenic co-exposures (like arsenic or engine exhaust) (Englyst et al., 2001). However, cohort workers worked in various industries/jobs and were presumably not all exposed to the same co-exposures making it less likely that one specific co-exposure would fully explain our results. Smoking may also confound the associations found between lead and lung/larynx cancer mortality or interact with lead to increase lung and larynx cancer risk. However, if lung and larynx cancer mortality findings were due to smoking, we would have expected similar associations with esophageal and kidney cancer mortality (both of which are typically associated with smoking). Additionally, internal analyses comparing workers to workers are generally less susceptible to confounding by smoking (Kriebel et al., 2004) and internal analyses for lung and larynx cancer mortality show strong effects with blood lead level category (i.e. HRs > 2.5 comparing highest to lowest exposure category for both cancer mortalities). Finally, when a sample of this cohort (n = 211 workers) was interviewed about smoking, there was no association between maximum past blood lead level and pack-years of smoking (Barry et al., 2019). It should also be noted that alcohol is also related to larynx cancer (and has synergistic effects with smoking). We had no data on alcohol, which is potentially a confounder if high lead-exposed workers also drank more. There is another issue with both smoking and alcohol, because both of these can contribute to blood lead levels, although in small amounts (Cezard et al., 1992; Grandjean et al., 1981; Shaper et al., 1982). Hence it is possible that those with high lead exposure smoked and drank more, and that this put them in a high lead category, and this in turn led to an increased larynx (or lung) cancer risk. However, past studies examining the influence of alcohol and smoking on blood lead level show only minor increases in blood lead associated with these factors (e.g. differences of 3 μg/dL), (Grandjean et al., 1981; Shaper et al., 1982) and we think that occupational exposure dwarfs these non-occupational exposures, such that the latter are unlikely to cause workers to be in our higher lead categories. In internal comparisons, we found statistically significant positive trends in COPD and ischemic heart disease mortality with increasing blood lead level as well as increased ischemic heart disease mortality risk among workers with maximum blood lead levels ≥25 μg/dL (compared to < 5 μg/dL). However, our results show no indication of increased COPD or ischemic heart disease mortality risk in external comparisons, possibly as a result of the healthy worker effect (McMichael, 1976; Wen et al., 1983). Abundant research exists demonstrating associations between lead and increases in systolic and diastolic blood pressure and it is presumably through this mechanism that lead may increase heart disease related mortality risk (Navas-Acien et al., 2007; Nawrot et al., 2002; Gambelunghe et al., 2016; Han et al., 2018; Nash et al., 2003). Results from a randomized trial may also support a lead-ischemic heart disease mortality association. That study found that people who had experienced a previous myocardial infarction had a modestly reduced risk of adverse cardiovascular outcomes after regular chelation therapy (Lamas et al., 2013). Only one study (that we know of) demonstrates a similar lead-COPD relationship (Steenland et al., 2017). Although that study combines results from Finland, Great Britain, and the US, lead-COPD results were significant and positive in Finland (personal communication). It is unclear whether COPD findings are true associations or due to confounding by simultaneous exposure to dust. Our results point to an association between increasing lead with increased mortality risk due to malignant brain cancer and non-hodgkin's lymphoma. Our brain cancer finding agrees with IARC's 2006 determination that brain cancer was one of the more strongly
5. Conclusion We found positive trends between blood lead level and all-cause, malignant brain cancer, COPD, ischemic heart disease, larynx cancer, lung cancer, and non-hodgkin's lymphoma mortality as well as suggestions of associations with chronic renal disease and rectal cancer mortality. The cohort was large (n = 58,368), followed for a median of 19 years which allowed for deaths of more rare causes to occur, and included workers with a wide range of blood lead levels including ≥40 μg/dL (which is higher than guidelines allow) and < 5 μg/dL (similar to the general population). Future studies should confirm these associations in other populations. Funding This work was supported by the National Institute for Occupational Safety and Health award R01OH010745-04 (PI Kyle Steenland). Declarations of interest None. 5
Environmental Research 177 (2019) 108625
V. Barry and K. Steenland
Acknowledgements
27. Effect of disodium EDTA chelation regimen on cardiovascular events in patients with previous myocardial infarction: the TACT randomized trial. Jama 309 (12), 1241–1250. Lanphear, B.P., Rauch, S., Auinger, P., Allen, R.W., Hornung, R.W., 2018 Apr. Low-level lead exposure and mortality in US adults: a population-based cohort study. The Lancet Public health 3 (4), e177–e184. Liao, L.M., Friesen, M.C., Xiang, Y.-B., Cai, H., Koh, D.-H., Ji, B.-T., et al., 2016. Occupational lead exposure and associations with selected cancers: the shanghai men's and women's health study cohorts. Environ. Health Perspect. 124 (1), 97–103. Lundstrom, N.G., Nordberg, G., Englyst, V., Gerhardsson, L., Hagmar, L., Jin, T., et al., 1997 Feb. Cumulative lead exposure in relation to mortality and lung cancer morbidity in a cohort of primary smelter workers. Scand. J. Work Environ. Health 23 (1), 24–30. Lustberg, M., Silbergeld, E., 2002. Blood lead levels and mortality. JAMA Internal Medicine 162 (21), 2443–2449. McMichael, A.J., 1976 Mar. Standardized mortality ratios and the "healthy worker effect": scratching beneath the surface. J. Occup. Med. : Off. Publ. Ind. Med. Assoc. 18 (3), 165–168. Menke, A., Muntner, P., Batuman, V., Silbergeld, E.K., Guallar, E., 2006 Sep 26. Blood lead below 0.48 micromol/L (10 microg/dL) and mortality among US adults. Circulation 114 (13), 1388–1394. Muntner, P., He, J., Vupputuri, S., Coresh, J., Batuman, V., 2003 Mar. Blood lead and chronic kidney disease in the general United States population: results from NHANES III. Kidney Int. 63 (3), 1044–1050. Nash, D., Magder, L., Lustberg, M., Sherwin, R.W., Rubin, R.J., Kaufmann, R.B., et al., 2003 Mar 26. Blood lead, blood pressure, and hypertension in perimenopausal and postmenopausal women. Jama 289 (12), 1523–1532. Navas-Acien, A., Guallar, E., Silbergeld, E.K., Rothenberg, S.J., 2007. Lead exposure and cardiovascular disease–a systematic review. Environ. Health Perspect. 115 (3), 472–482. Nawrot, T.S., Thijs, L., Den Hond, E.M., Roels, H.A., Staessen, J.A., 2002 Feb. An epidemiological re-appraisal of the association between blood pressure and blood lead: a meta-analysis. J. Hum. Hypertens. 16 (2), 123–131. Pirkle, J.L., Brody, D.J., Gunter, E.W., Kramer, R.A., Paschal, D.C., Flegal, K.M., et al., 1994 Jul 27. The decline in blood lead levels in the United States. The national health and nutrition examination surveys (NHANES). Jama 272 (4), 284–291. Ramis, R., Diggle, P., Boldo, E., Garcia-Perez, J., Fernandez-Navarro, P., Lopez-Abente, G., 2012 Feb 6. Analysis of matched geographical areas to study potential links between environmental exposure to oil refineries and non-Hodgkin lymphoma mortality in Spain. Int. J. Health Geogr. 11, 4. Schober, S.E., Mirel, L.B., Graubard, B.I., Brody, D.J., Flegal, K.M., 2006 Oct. Blood lead levels and death from all causes, cardiovascular disease, and cancer: results from the NHANES III mortality study. Environ. Health Perspect. 114 (10), 1538–1541. Schubauer-Berigan, M.K., Hein, M.J., Raudabaugh, W.M., Ruder, A.M., Silver, S.R., Spaeth, S., et al., 2011 Dec. Update of the NIOSH life table analysis system: a personyears analysis program for the windows computing environment. Am. J. Ind. Med. 54 (12), 915–924. Schwartz, B.S., Hu, H., 2007 Mar. Adult lead exposure: time for change. Environ. Health Perspect. 115 (3), 451–454. Shaper, A.G., Pocock, S.J., Walker, M., Wale, C.J., Clayton, B., Delves, H.T., et al., 1982 Jan 30. Effects of alcohol and smoking on blood lead in middle-aged British men. Br. Med. J. 284 (6312), 299–302. Sirivarasai, J., Kaojarern, S., Chanprasertyothin, S., Panpunuan, P., Petchpoung, K., Tatsaneeyapant, A., et al., 2015. Environmental lead exposure, catalase gene, and markers of antioxidant and oxidative stress relation to hypertension: an analysis based on the EGAT study. BioMed Res. Int. 2015, 856319. Spivey, A., 2007. The weight of lead. Effects add up in adults. Environ. Health Perspect. 115 (1), A30–A36. Staudinger, K.C., Roth, V.S., 1998 Feb 15. Occupational lead poisoning. Am. Fam. Physician 57 (4), 719–726 31-2. Steenland, K., Barry, V., Anttila, A., Sallmen, M., McElvenny, D., Todd, A.C., et al., 2017 Nov. A cohort mortality study of lead-exposed workers in the USA, Finland and the UK. Occup. Environ. Med. 74 (11), 785–791. Steenland Kyle, B.V., Ahti, Anttila, Markku, Sallmen, Mueller, William, Peter, Ritchie, Damien, McElvenny, Kurt, Straif, June 2019. Cancer incidence among workers with blood lead measurements in two countries. Occup. Environ. Med (in press) as of. Tsai, P.L., Hatfield, T.H., 2011. Global benefits from the phaseout of leaded fuel. J. Environ. Health 74 (5), 8–15. Tsoi, M.F., Cheung, C.L., Cheung, T.T., Cheung, B.M., 2016 Nov. Continual decrease in blood lead level in Americans: United States national health nutrition and examination survey 1999-2014. Am. J. Med. 129 (11), 1213–1218. Wen, C.P., Tsai, S.P., Gibson, R.L., 1983 Apr. Anatomy of the healthy worker effect: a critical review. J. Occup. Med. : Off. Publ. Ind. Med. Assoc. 25 (4), 283–289. Wu, W.T., Lin, Y.J., Liou, S.H., Yang, C.Y., Cheng, K.F., Tsai, P.J., et al., 2012 Apr. Brain cancer associated with environmental lead exposure: evidence from implementation of a National Petrol-Lead Phase-Out Program (PLPOP) in Taiwan between 1979 and 2007. Environ. Int. 40, 97–101.
This work was supported by the National Institute for Occupational Safety and Health award R01OH010745-04 (PI Kyle Steenland). We would like to acknowledge the following people for their help in providing data access and matching: Paula Guro (USRDS), Michelle Goodier (NDI), Lillian Ingster (NDI), Velida Juzbasic (NDI), Kenneth D. Rosenman (Michigan State University), Joanna Kica (Michigan State University), Carrie Tomasallo (Wisconsin Division of Public Health), and Michael Flanagan (Massachusetts Dept of Labor Standards). Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.envres.2019.108625. References Abadin, H., Ashizawa, A., Stevens, Y.W., Llados, F., Diamond, G., Sage, G., et al., 2007. Agency for Toxic Substances and Disease Registry (ATSDR) Toxicological Profiles. Toxicological Profile for Lead. Agency for Toxic Substances and Disease Registry (US), Atlanta (GA). Adult blood lead epidemiology and surveillance–United States, 2003-2004. MMWR Morb. Mortal. Wkly. Rep. 55 (32), 876–879. Anttila, A., Heikkila, P., Pukkala, E., Nykyri, E., Kauppinen, T., Hernberg, S., et al., 1995 Dec. Excess lung cancer among workers exposed to lead. Scand. J. Work Environ. Health 21 (6), 460–469. Barry, V., Todd, A.C., Steenland, K., 2019 May. Bone lead associations with blood lead, kidney function and blood pressure among US, lead-exposed workers in a surveillance programme. Occup. Environ. Med. 76 (5), 349–354. Bayer, O., Camara, R., Zeissig, S.R., Ressing, M., Dietz, A., Locati, L.D., et al., 2016 Jan. Occupation and cancer of the larynx: a systematic review and meta-analysis. Eur. Arch. Oto-Rhino-Laryngol. 273 (1), 9–20 official journal of the European Federation of Oto-Rhino-Laryngological Societies (EUFOS) : affiliated with the German Society for Oto-Rhino-Laryngology - Head and Neck Surgery. Cezard, C., Demarquilly, C., Boniface, M., Haguenoer, J.M., 1992 Sep. Influence of the degree of exposure to lead on relations between alcohol consumption and the biological indices of lead exposure: epidemiological study in a lead acid battery factory. Br. J. Ind. Med. 49 (9), 645–647. Chowdhury, R., Sarnat, S.E., Darrow, L., McClellan, W., Steenland, K., 2014 Jula. Mortality among participants in a lead surveillance program. Environ. Res. 132, 100–104. Chowdhury, R., Darrow, L., McClellan, W., Sarnat, S., Steenland, K., 2014 Julb. Incident ESRD among participants in a lead surveillance program. Am. J. Kidney Dis. 64 (1), 25–31 the official journal of the National Kidney Foundation. Englyst, V., Lundstrom, N.G., Gerhardsson, L., Rylander, L., Nordberg, G., 2001 Jun 12. Lung cancer risks among lead smelter workers also exposed to arsenic. Sci. Total Environ. 273 (1–3), 77–82. Fadrowski, J.J., Navas-Acien, A., Tellez-Plaza, M., Guallar, E., Weaver, V.M., Furth, S.L., 2010. Blood lead level and kidney function in US adolescents: the third national health and nutrition examination survey. Arch. Intern. Med. 170 (1), 75–82. Fayerweather, W.E., Karns, M.E., Nuwayhid, I.A., Nelson, T.J., 1997 Jan. Case-control study of cancer risk in tetraethyl lead manufacturing. Am. J. Ind. Med. 31 (1), 28–35. Fu, H., Boffetta, P., 1995. Cancer and occupational exposure to inorganic lead compounds: a meta-analysis of published data. Occup. Environ. Med. 52 (2), 73–81. Gambelunghe, A., Sallsten, G., Borne, Y., Forsgard, N., Hedblad, B., Nilsson, P., et al., 2016 Aug. Low-level exposure to lead, blood pressure, and hypertension in a population-based cohort. Environ. Res. 149, 157–163. Grandjean, P., Olsen, N.B., Hollnagel, H., 1981. Influence of smoking and alcohol consumption on blood lead levels. Int. Arch. Occup. Environ. Health 48 (4), 391–397. Han, L., Wang, X., Han, R., Xu, M., Zhao, Y., Gao, Q., et al., 2018. Association between blood lead level and blood pressure: an occupational population-based study in Jiangsu province, China. PLoS One 13 (7), e0200289. Holland, M.G., Cawthon, D., 2016 Dec. Workplace lead exposure. J. Occup. Environ. Med. 58 (12), e371–e374. IWGotEoCRt, Humans, 2006. Evaluation of carcinogenic risk to humans, inorganic and organic lead compounds: international agency for research on cancer. IARC monographs on the evaluation of carcinogenic risks to humans. Available from: https:// www.ncbi.nlm.nih.gov/books/NBK321297/ 87. Kriebel, D., Zeka, A., Eisen, E.A., Wegman, D.H., 2004 Oct. Quantitative evaluation of the effects of uncontrolled confounding by alcohol and tobacco in occupational cancer studies. Int. J. Epidemiol. 33 (5), 1040–1045. Lamas, G.A., Goertz, C., Boineau, R., Mark, D.B., Rozema, T., Nahin, R.L., et al., 2013 Mar
6