Volatile Hydrocarbon Exposures and Incident Coronary Heart Disease Events: Up to Ten Years of Follow-up among Deepwater Horizon Oil Spill Workers
1Department of Epidemiology, Gillings School of Global Public Health, University of North Carolina, Chapel Hill, North Carolina, USA
2Epidemiology Branch, National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina, USA
3Department of Medicine, School of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
4Stewart Exposure Assessments, LLC, Arlington, Virginia, USA
5Exposure Assessment Applications, LLC, Arlington, Virginia, USA
6Department of Epidemiology and Biostatistics, School of Public Health, West Virginia University, Morgantown, West Virginia, USA
7Department of Environmental Health and Engineering, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland, USA
8Department of Biostatistics, Fielding School of Public Health, University of California – Los Angeles, Los Angeles, California, USA
9Department of Environmental and Occupational Health, Dornsife School of Public Health, Drexel University, Philadelphia, Pennsylvania, USA
10Social & Scientific Systems, Inc, a DLH Holdings Company, Durham, North Carolina, USA
11Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, Maryland, USA
12Population Studies and Genetics Branch, National Institute on Aging, Bethesda, Maryland, USA
Abstract
Background:
During the 2010 Deepwater Horizon (DWH) disaster, response and cleanup workers were potentially exposed to toxic volatile components of crude oil. However, to our knowledge, no study has examined exposure to individual oil spill–related chemicals in relation to cardiovascular outcomes among oil spill workers.
Objectives:
Our aim was to investigate the association of several spill-related chemicals [benzene, toluene, ethylbenzene, xylene, n-hexane (BTEX-H)] and total hydrocarbons (THC) with incident coronary heart disease (CHD) events among workers enrolled in a prospective cohort.
Methods:
Cumulative exposures to THC and BTEX-H across the cleanup period were estimated via a job-exposure matrix that linked air measurement data with self-reported DWH spill work histories. We ascertained CHD events following each worker’s last day of cleanup work as the first self-reported physician-diagnosed myocardial infarction (MI) or a fatal CHD event. We estimated hazard ratios (HR) and 95% confidence intervals for the associations of exposure quintiles (Q) with risk of CHD. We applied inverse probability weights to account for bias due to confounding and loss to follow-up. We used quantile g-computation to assess the joint effect of the BTEX-H mixture.
Results:
Among 22,655 workers with no previous MI diagnoses, 509 experienced an incident CHD event through December 2019. Workers in higher quintiles of each exposure agent had increased CHD risks in comparison with the referent group (Q1) of that agent, with the strongest associations observed in Q5 (range of ). However, most associations were nonsignificant, and there was no evidence of exposure–response trends. We observed stronger associations among ever smokers, workers with education, and workers with body mass index . No apparent positive association was observed for the BTEX-H mixture.
Conclusions:
Higher exposures to volatile components of crude oil were associated with modest increases in risk of CHD among oil spill workers, although we did not observe exposure–response trends. https://doi.org/10.1289/EHP11859
Article notes
Untitled section
Footnote Group
Untitled section
Received 2022 Jul 15; Revised 2023 Apr 9; Accepted 2023 Apr 28; Collection date 2023 May.
Introduction
The 2010 Deepwater Horizon (DWH) disaster was the largest marine oil spill in U.S. history.1 An estimated barrels of crude oil were discharged into the Gulf of Mexico before the wellhead was mechanically capped on 15 July 2010.1 Shortly after the spill began, an extensive oil spill response and cleanup (OSRC) operation was launched to stop the spill and remove the crude oil from the environment. Tens of thousands of workers and volunteers participated in this operation, with most efforts completed by June 2011.2
During the OSRC, workers were exposed to a range of inhalation hazards, including volatized crude oil hydrocarbons.3 These hydrocarbons were a significant contributor to air emissions in the DWH disaster.4 Many of these components, including benzene, some alkylbenzenes, and hexane, are classified as hazardous air pollutants because of their toxicological properties.5 Studies have associated short-term ambient exposure to benzene and alkylbenzenes with acute onset of myocardial infarction (MI)6 or coronary death.7,8 Chronic benzene exposure has also been linked to persistent MI mortality in a Spanish case–control study; however, the study failed to account for important confounders, including smoking and socioeconomic status.9
Persistent respiratory effects have been observed among workers who participated in various oil spill cleanups10–13; however, associations between OSRC-related exposures and cardiovascular health were previously assessed only in workers who responded to the Hebei Spirit oil spill and the DWH disaster.14–17 In an analysis of Hebei Spirit oil spill workers, longer duration of cleanup work was associated with higher risk of self-reported angina or MI up to 10 y after the spill. In the DWH disaster, Strelitz et al.16 observed higher incidence of self-reported MI or fatal coronary heart disease (CHD) up to 5 y after the spill among OSRC workers with longer duration of work and with higher single daily maximum exposure to total (petroleum) hydrocarbons (THC), a composite measure of volatile components of the crude oil.17 Similar associations were found when nonfatal MI was examined as the outcome.15
These studies provide some evidence that exposure to crude oil chemicals may be related to increased risk of MI/coronary heart disease (CHD) over time. Recently, quantitative estimates of several cumulative oil-related exposures [THC, benzene, toluene, ethylbenzene, xylene (all isomers combined), and n-hexane (BTEX-H)] were developed from personal measurements and self-reported OSRC activities for the DWH OSRC workers,18 providing an opportunity to study health effects associated with these specific chemical exposures among the same group of workers and with several additional years of follow-up. Given that BTEX-H also arise from a single emission source in other environments (e.g., vehicular exhaust, oil/gas operations),19,20 it is useful to assess the joint effect of the exposure mixture to support interventions that target the exposure source. The objective of this study was to assess quantitative oil-related exposures, individually and as a mixture, in relation to incident CHD events among DWH OSRC workers.
Methods
Study Population
The GuLF Study (Gulf Long-Term Follow-up Study) is a prospective cohort study of the potential health effects of the DWH disaster.3 Participants included anyone y of age at enrollment who either had participated in OSRC work for at least 1 d (workers) or had completed safety training but were not hired (nonworkers). Enrollment started in March 2011 and continued through May 2013. A total of 32,608 participants were enrolled. At enrollment, all participants completed a computer-assisted telephone interview in which they provided information on sociodemographics, lifestyle, health (including physician diagnosis of MI), and a detailed history of DWH OSRC activities. Two rounds of follow-up interviews (May 2013–Apr 2016 and November 2017–July 2021) were conducted via telephone to ascertain changes in health status (including new physician diagnoses of MI) since the previous interview. We excluded from the current analysis 999 Vietnamese-only speaking participants who completed only an abbreviated enrollment interview that precluded ascertainment of MI diagnosis or estimation of oil spill-related chemical exposures.
For all analyses, we restricted the study population to the 24,375 workers; we restricted our analysis to participants who worked at least 1 d on the DWH cleanup (“workers”) because they had the opportunity to be exposed to crude oil chemicals.3 We excluded 21 workers who did not provide information on MI diagnoses in any of the interviews. We restricted our analysis to incident cases by beginning follow-up at the end of each worker’s cleanup work time and excluded 489 workers who reported an MI diagnosis before the start of follow-up. Although reports of MIs that occurred during cleanup would be informative, it is also likely that workers who had such an event might have stopped working and were not enrolled in our study; thus, including the person-time before the end of cleanup could lead to immortal time bias. Of the remaining 23,865 workers, we restricted our analysis to 23,664 workers who had complete THC and BTEX-H exposure estimates. Finally, we excluded 1,009 workers with missing covariates needed for analysis and reached a final analytical sample of 22,655 workers. Figure S1 illustrates how we derived the analytical sample from the enrolled population. Among these participants, 15,627 (69%) and 10,638 (47%) completed the first and second follow-up interviews, respectively. Response rates were over 88% in both follow-ups among those who could be reached. All participants provided informed consent prior to participating in the GuLF Study. The institutional review board of the National Institute of Environmental Health Sciences approved this study.
Exposure Assessment
Cumulative exposures to five spill-related chemicals (BTEX-H) and THC across each participant’s work period were estimated via a job-exposure matrix that linked air measurement data with detailed work histories. Measurement data came primarily from full-shift personal air samples collected on OSRC workers using organic vapor passive dosimeters during their work shifts from April 2010 to June 2011.18 These samples were analyzed for THC and BTEX-H, resulting in more than 143,000 measurements of THC (as petroleum hydrocarbons) and BTEX-H.18,21 These personal measurements were supplemented by more than direct-reading volatile organic compound (VOC) area measurements collected using multigas detectors (AreaRAE and MultiRAE) equipped with a photoionization detector lamp on 38 vessels involved in the OSRC to develop THC and BTEX-H estimates on days and vessels where there were few or no personal measurements.22,23
To estimate exposures for the full cohort, the study industrial hygienists created more than 3,000 exposure groups (EGs) based on three exposure determinants: job/activity, location, and time period.24 Job was a major determinant of exposure, but because job title [e.g., crane operator, remotely operated vehicle (ROV) technician] was not available for all workers, industrial hygienists supplemented job title information with self-reported work activities (e.g., handling oily boom, skimming, decontamination) to develop this exposure determinant. Location covered four areas of the Gulf (by increasing distance from the wellhead: hot zone/source, offshore, nearshore, land) and four Gulf coastal states (Louisiana, Mississippi, Alabama, Florida). The OSRC work period (22 April 2010–30 June 2011) was divided into seven time periods: TP1a (22 April–14 May 2010), TP1b (15 May–15 July 2010), TP2 (16 July–10 August 2010), TP3 (11 August–30 September 2010), TP4 (1 October–31 December 2010), TP5 (1 January–31 March 2011), and TP6 (1 April–30 June 2011). These time periods were developed to reflect changes in the degree of weathering of the oil (which increased over time) and in OSRC events that likely affected workers’ exposures. For instance, the mechanical capping of the wellhead on 15 July 2010, which separated TP1b and TP2, marked the stop of oil release from the damaged well and the end of oil/gas flaring and dispersant application operations. For a detailed explanation of the criteria used to define each time period, see Stewart et al.18 Each EG was a unique combination of these determinants and represented workers who, based on these determinants, were expected to have similar distributions of exposures. Using these determinants, industrial hygienists assigned air measurements to each EG and estimated exposure averages for the EGs.22,23,25–29 To handle the large number of measurements below the analytical method’s limit of detection in some EGs, a left-censored Bayesian framework was used to estimate exposure averages and other exposure statistics for the EGs.25,26
Workers were matched to the appropriate EGs based on their reported DWH OSRC work history. Many workers reported multiple work activities across the cleanup, some of which occurred on the same day. Although workers reported the start date, end date, and days worked for each job/activity, we lacked information on the specific days and the exact number of hours on these days that they performed each job/activity. Thus, we developed two daily exposure estimates for each day worked: a) the daily maximum, the value corresponding to the highest-exposed activity on a day; and b) the daily average, the average of the exposure estimates across all jobs/activities on each day. To examine the total burden of exposure received by each worker during the cleanup, two cumulative exposure metrics were created across all workdays based on their daily exposure estimates and the duration of their work: a) cumulative daily maximum, the sum of daily maximum exposure estimates across all days within a time period and then across all time periods; and b) cumulative daily average, the sum of daily average exposure estimates across all days within a time period and then across all time periods. These measures are the primary exposure estimates examined in the current analyses. We also considered as each worker’s exposure the single highest daily maximum exposure estimate (henceforth, “single daily maximum” exposure) over the entire work period in a subanalysis. In secondary analyses, we explored the health impacts among workers who had multiple unusually high daily exposures to BTEX-H or THC by comparing them with workers who exclusively had only lower daily maximum exposure estimates (as described in the section titled “Statistical Modeling” below).
Outcome Assessment
The outcome of interest was the first occurrence of a CHD event after the last day of each participant’s OSRC work, defined as either a self-reported physician-diagnosed MI or an International Classification of Diseases (ICD)–coded fatal CHD event. At enrollment and in the two follow-up interviews, participants were asked “Has a doctor ever told you that you had a heart attack, also called a myocardial infarction or ‘MI’?” Those who responded “yes” were asked to provide the month and year of, or the age at, the event. Fatal CHD events were ascertained via linkage with the National Death Index through 31 December 2019, and we included deaths attributed to ischemic heart disease (ICD-10 code I20–I25) as an underlying cause. Time at risk was measured in months from the date after each participant ended cleanup work to the first of CHD event, death from other causes, end of cohort follow-up (31 December 2019), or, for participants who were lost to follow-up (i.e., did not respond to the first or both follow-up interviews and were event-free), the date they completed the previous interview.
Statistical Modeling
We used Cox proportional hazards models to estimate hazard ratios (HRs) for the first incident CHD event associated with increasing cumulative exposure to each BTEX-H chemical and THC.30 We used quintiles of exposure levels for analysis, with workers in the lowest quintile (Q1) of an exposure agent as the referent group. We also investigated exposure–response trends by examining ln-transformed continuous exposure levels. We chose time since exposure, rather than age, as the time scale. Although age was an important confounder in our study, using it as the time scale would imply that workers who entered the study at an older age were exchangeable prior to study entry with workers who entered study at a younger age, given all confounders.31 This assumption is not possible to verify with the data at hand.
We adjusted for potential confounding using inverse probability (IP) weighting.32 We selected covariates based on a directed acyclic graph (DAG) and included the minimally sufficient adjustment set and predictors of the outcome that are not descendants of the exposure in the IP-exposure weights (Figure S2).33–35 Because of high correlations among exposures (range: 0.84–0.96) (Table S1), we were not able to account for coexposures in the weights. We obtained stabilized weights by fitting a multinomial logistic regression model for each categorical exposure with respect to selected covariates. In analysis of ln-transformed continuous exposures, we generated weights for each exposure agent using a quantile binning approach.36
All covariates were self-reported or derived from information that was self-reported at enrollment and included the following: age (in years: 20–29, 30–39, 40–49, 50–59, ), sex (male; female), race (White; Black; American Indian or Alaskan Native; Asian; Native Hawaiian or Pacific Islander; other race), Hispanic ethnicity (Hispanic; non-Hispanic), cigarette smoking status [current heavy ( cigarettes/day); current light ( cigarettes/day); former; never], highest educational attainment (less than high school; high school diploma or general equivalency diploma; some college or 2-y degree; 4-y college graduate or more), body mass index [BMI, in kilograms per square meter: underweight or normal (); overweight (); obese I (); obese II ()], previous oil spill cleanup experience (yes, no), previous oil industry experience (yes, no), precleanup diabetes diagnosis (yes, no), and residential proximity to the spill (living in a coastal county directly affected by the spill or a county adjacent to the impacted counties; living in a Gulf state farther from the spill; living in a non-Gulf state). We used self-reported race and ethnicity as proxies for the downstream effects of socioeconomic disparities as well as differential life experiences resulting from structural racism that might have influenced risk of CHD. For analysis, race was collapsed into White, Black, and “other/multiracial” because of the small number of participants who were not White or Black.
To account for informative censoring due to loss to follow-up, we used IP-censoring weighting.37,38 Participants were considered censored if they a) did not complete a follow-up interview or completed the first but not the second interview and b) had not experienced a CHD event prior to being lost to follow-up. Censoring was modeled as a function of its predictors in a pooled logistic regression, and weights were stabilized by the marginal probability of censoring. Covariates in the IP-censoring weights were determined from a DAG34 and included: exposure(s) (THC for model of THC; all chemicals of BTEX-H for each model of BTEX-H), age, sex, self-reported race, Hispanic ethnicity, cigarette smoking, highest educational attainment, previous oil spill cleanup experience, and residential proximity to the spill. The finalized weights applied to the models were the product of the IP-exposure weights and the IP-censoring weights. Cox proportional hazards models with a robust variance estimator were fitted to estimate HRs and 95% confidence intervals (CIs). IP-exposure weights and IP-censoring weights were applied to the main analysis and subanalyses.
In single-agent analyses, we investigated potential effect measure modification (EMM) by cigarette smoking status (ever vs. never), because smoking could induce adverse cardiovascular effects via biological pathways similar to those of air pollutants and thus enhance the effects of air pollutants, including the crude oil chemicals examined in our study.39 We also examined EMM of the associations by educational attainment (high school or less vs. more than high school). Previous studies have shown stronger effects of ambient air pollution on CHD risk among study participants with lower educational attainment,40,41 possibly because the adverse effects of air pollutant exposure were exacerbated by detrimental lifestyle (e.g., lack of access to healthy food, participation in less leisure time physical activity) and other environmental contaminants42–44 associated with less educational attainment.45–47 In addition, we stratified analyses by obesity [nonobese (BMI ) vs. obese ()] at enrollment to see whether there was heterogeneity among associations. As a major risk factor for cardiometabolic diseases, obesity could potentially increase air pollution–related health effects because of persistent obesity-induced low-grade inflammation48 and higher particle deposition in the lung from higher breathing rates and tidal volumes among obese people.49–51 We assessed EMM by including a product term between quintile exposure and the modifier in the model and reported the p-value for the joint Wald test. In addition, we examined associations between exposures and CHD risk in the subset of participants who were age 40 y and older. We were unable to examine associations in participants younger than age 40 y because of the small number of cases in this subgroup.
We conducted a number of sensitivity analyses. First, we included self-reported precleanup hypertension in the IP-exposure weight model to see whether results differed. This covariate was not included in the main analysis because precleanup hypertension was not related to crude oil exposures and we were concerned about possible misclassification of hypertension using self-reports.52 We also accounted for several other DWH OSRC–related exposures that have been associated with CHD in other studies. To address potential confounding from exposure to higher levels of fine particulate matter (PM) with aerodynamic diameter () from controlled burning activities,53,54 we reran the analysis excluding 1,997 workers with higher burning exposures. Because ever having to stop working during the cleanup because of heat (yes, no), a proxy for high heat exposure, was associated with higher risk of nonfatal MI in a previous GuLF Study analysis,15 we included this variable in the weight model to determine whether it affected results. We also included in the model a variable for having performed any jobs during the cleanup that involved handling oily plants/wildlife or dead animal recovery (yes, no) to account for potential psychological stress from specific OSRC work that may have contributed to risk of CHD.55 To assess whether our estimates were sensitive to how age was specified in the model, we investigated restrictive cubic splines as an alternative functional form of age. In addition, we conducted analyses using an alternative definition of CHD-related deaths based on ischemic heart disease as a contributing/underlying cause rather than the underlying cause of death. We attempted to examine nonfatal MI and fatal CHD as separate outcomes; however, we could not examine fatal CHD as the outcome because of the small number of fatal cases. Workers with fatal (CHD) events that occurred after OSRC employment but before they could enroll were not identified. We explored the impact of this left truncation in a sensitivity analysis by starting the risk period at study enrollment, effectively excluding 129 preenrollment MI events. To follow up on the analysis of Strelitz et al.,17 which used an earlier, ordinal classification of the single daily maximum THC exposure and observed participants only through the first follow-up interview, we conducted a similar analysis using the newly developed quantitative single daily maximum THC exposure (categorized in quintiles) and observed participants for both the same and the extended time period (until December 2019). To explore the potential bias from inaccurate recall of the date of CHD diagnosis among participants who reported a nonfatal CHD event, we performed an exploratory analysis that coarsened the time interval in which events were identified so that CHD events were tallied every 4 months instead of every month. Last, to capture milder forms of CHD that could progress to MIs and might have been related to the exposures, we performed an analysis that included as events participants who reported a physician-diagnosed blockage in the arteries of the heart as well as all events in the main analysis.
In an exploratory analysis, we also considered the impact of short-term exposure to higher exposure levels by exploring risks associated with having exposures in the top 20th and 15th percentile of the single daily maximum exposure for a minimum of 7 or 14 d in comparison with having daily maximum exposures in the lowest quintile.
We used quantile g-computation (QGC)56 to estimate the joint effect of BTEX-H as a mixture on the risk of CHD events. This method is a special case of g-computation, which is a generalization of standardization to continuous and time-varying confounders.56 The method can be applied to any generalized linear model. We used the approach described in White et al.,57 which implements QGC in the context of a Cox proportional hazards model. Under the assumption that the effects of each exposure are linear and additive (with respect to the log-hazard) on the quantile basis, Keil et al.56 showed that QGC can be implemented using a standard Cox model with transformed exposures. In this analysis, we categorized exposure components into quintiles and assigned each quintile an integer score (). We then fit a Cox proportional hazards model of CHD against the quintile scores (treated as continuous variables), conditional on the same covariates used in single-agent IP-exposure weight models. This construction allows a weight for each BTEX-H component to represent a component’s relative contribution to positive associations (positive weights), or relative contribution to inverse associations (negative weights). Positive and negative weights sum to 1 and , respectively. Under assumptions of linearity and additivity of the effects of BTEX-H, the joint effect equals the sum of generalized linear model coefficients for all of the transformed exposures, and it is interpreted as the expected change in the log-hazard of CHD for a simultaneous one quantile increase in all of the exposures in the BTEX-H mixture. As in single-agent analyses, we applied IP-censoring weights to the mixture model. Quantile g-computation was performed using R (version 4.0.4; R Development Core Team; package “qgcomp”). To facilitate comparison with the mixture model, we additionally performed single-pollutant analyses of each BTEX-H chemical modeled the same way as that in the QGC analysis (i.e., with integer scores assigned to each quintile) to assess changes in hazard of CHD per quintile increase in exposures.
All other analyses were performed using SAS (version 9.4; SAS Institute Inc.). An alpha level of 0.05 was considered statistically significant for all analyses.
Results
In comparison with the full analytical sample (), those who completed the first () or second () follow-up interviews tended to be older, female, White, and former or never smokers (Table 1). They were also more likely to have graduated from a 4-y college and to reside outside the Gulf states. There was no substantive difference in the other characteristics. Workers in the full analytical sample were exposed to crude oil for a median 4 months (range: 1 d–15 months).
| Characteristic | Enrollmenta | 1st follow-upb | 2nd follow-upc |
|---|---|---|---|
| () | () | () | |
| n (%) | n (%) | n (%) | |
| Age at enrollment (y) | |||
| 20–29 | 4,803 (21.2) | 2,822 (18.1) | 1,782 (16.8) |
| 30–39 | 5,586 (24.7) | 3,567 (22.8) | 2,304 (21.7) |
| 40–49 | 5,585 (24.7) | 3,973 (25.4) | 2,708 (25.5) |
| 50–59 | 4,762 (21.0) | 3,703 (23.7) | 2,666 (25.1) |
| 1,919 (8.5) | 1,562 (10.0) | 1,178 (11.1) | |
| Sex | |||
| Male | 18,627 (82.2) | 12,736 (81.5) | 8,548 (80.4) |
| Female | 4,028 (17.8) | 2,891 (18.5) | 2,090 (19.7) |
| Race | |||
| White | 15,066 (66.5) | 10,550 (67.5) | 7,649 (71.9) |
| Black | 5,303 (23.4) | 3,550 (22.7) | 2,043 (19.2) |
| Asian | 202 (0.9) | 120 (0.8) | 78 (0.7) |
| American Indian or Alaskan Native | 400 (1.8) | 287 (1.8) | 180 (1.7) |
| Native Hawaiian or Pacific Islander | 47 (0.2) | 31 (0.2) | 22 (0.2) |
| Other race | 1,033 (4.6) | 654 (4.2) | 391 (3.7) |
| Multiracial | 604 (2.7) | 435 (2.8) | 275 (2.6) |
| Hispanic ethnicity | |||
| No | 21,141 (93.3) | 14,632 (93.6) | 10,023 (94.2) |
| Yes | 1,514 (6.7) | 995 (6.4) | 615 (5.8) |
| Educational attainment | |||
| Less than high school | 3,470 (15.3) | 2,210 (14.1) | 1,207 (11.4) |
| High school diploma/GED | 6,697 (29.6) | 4,358 (27.9) | 2,697 (25.4) |
| Some college/2-y degree | 6,851 (30.2) | 4,719 (30.2) | 3,238 (30.4) |
| 4-y college graduate or more | 5,637 (24.9) | 4,340 (27.8) | 3,496 (32.9) |
| Weight classification | |||
| Underweight or normal () | 6,146 (27.1) | 4,131 (26.4) | 2,746 (25.8) |
| Overweight () | 9,419 (41.6) | 6,516 (41.7) | 4,534 (42.6) |
| Obese I () | 4,633 (20.5) | 3,234 (20.7) | 2,198 (20.7) |
| Obese II () | 2,457 (10.9) | 1,746 (11.2) | 1,160 (10.9) |
| Reported precleanup diabetes diagnosis | |||
| No | 21,621 (95.4) | 14,851 (95.0) | 10,124 (95.2) |
| Yes | 1,034 (4.6) | 776 (5.0) | 514 (4.8) |
| Reported prespill hypertension diagnosis | |||
| No | 18,245 (81.9) | 12,365 (80.5) | 8,373 (80.0) |
| Yes | 4,034 (18.1) | 2,992 (19.5) | 2,092 (20.0) |
| Missing | 376 | 270 | 173 |
| Smoking status | |||
| Current heavy smoker ( cigarettes/day) | 2,266 (10.0) | 1,455 (9.3) | 845 (7.9) |
| Current light smoker ( cigarettes/day) | 4,524 (20.0) | 2,883 (18.5) | 1,717 (16.1) |
| Former smoker | 4,783 (21.1) | 3,431 (22.0) | 2,467 (23.2) |
| Never smoked | 11,082 (48.9) | 7,858 (50.3) | 5,609 (52.7) |
| Residential county proximity to Gulf of Mexicod | |||
| Direct or indirect contact | 13,339 (58.9) | 8,946 (57.3) | 5,733 (53.9) |
| Other Gulf state residence | 4,639 (20.5) | 3,138 (20.1) | 2,153 (20.2) |
| Non–Gulf state residence | 4,677 (20.6) | 3,543 (22.7) | 2,752 (25.9) |
| Previous oil spill cleanup work | |||
| No | 19,809 (87.4) | 13,524 (86.5) | 9,110 (85.6) |
| Yes | 2,846 (12.6) | 2,103 (13.5) | 1,528 (14.4) |
| Previous oil industry experience | |||
| No | 19,113 (84.4) | 13,108 (83.9) | 8,971 (84.3) |
| Yes | 3,542 (15.6) | 2,519 (16.1) | 1,667 (15.7) |
| Ever had to stop working because of heat | |||
| No | 13,383 (65.7) | 9,271 (66.4) | 6,443 (68.6) |
| Yes | 6,980 (34.3) | 4,693 (33.6) | 2,943 (31.4) |
| Missing | 2,292 | 1,663 | 1,252 |
| Worked any job involving oily wildlife, oily plants, or dead animal recovery | |||
| No | 16,287 (72.9) | 11,315 (73.5) | 8,018 (76.4) |
| Yes | 6,049 (27.1) | 4,077 (26.5) | 2,474 (23.6) |
| Missing | 319 | 235 | 146 |
During a median follow-up of 58 months (range: 1–116 months) beginning after each worker’s last date of cleanup work, 509 out of 22,655 workers experienced an incident CHD event. This group of 509 workers included 428 cases of nonfatal MI, 7 cases of nonfatal MI with a later fatal CHD event, and 74 fatal CHD events without a history of reported MI. Over 90% of cases () occurred among participants who were age 40 y or older at enrollment (Table S2). Almost 60% of the cases () occurred within 4 y of follow-up, and 21.6% of cases () were among participants followed for more than 6 y after exposure (Table S2). Among all workers in the study, half of the participants were exposed to crude oil chemicals from OSRC for months (Table 2).
| Minimum | Q1 cut point | Q2 cut point | Median | Q3 cut point | Q4 cut point | Maximum | |
|---|---|---|---|---|---|---|---|
| Duration of exposure (months) | 2 | 3 | 4 | 5 | 8 | 15 | |
| Cumulative daily maximum | |||||||
| THC (ppm-day) | 0.01 | 7.1 | 30.6 | 49.9 | 75.2 | 167.2 | 1,243.8 |
| Benzene (ppb-day) | 0.01 | 34.4 | 187.9 | 312.1 | 494.2 | 1,195.8 | 10,591.7 |
| Toluene (ppb-day) | 0.12 | 119.8 | 758.2 | 1,264.5 | 1,991.7 | 4,399.1 | 29,656.6 |
| Ethylbenzene (ppb-day) | 0.01 | 29.8 | 152.8 | 246.8 | 380.7 | 933.8 | 8,129.9 |
| Xylene (ppb-day) | 1.58 | 523.9 | 1,240.4 | 1,710.2 | 2,449.9 | 4,917.0 | 24,935.7 |
| n-Hexane (ppb-day) | 0.06 | 55.3 | 310.1 | 522.2 | 959.4 | 3,333.0 | 90,157.5 |
| Cumulative daily average | |||||||
| THC (ppm-day) | 0.01 | 5.5 | 19.4 | 30.5 | 44.9 | 86.9 | 761.2 |
| Benzene (ppb-day) | 0.01 | 23.1 | 115.9 | 182.6 | 270.1 | 598.8 | 7,744.1 |
| Toluene (ppb-day) | 0.12 | 85.4 | 439.5 | 717.7 | 1,092.5 | 2,219.2 | 18,067.9 |
| Ethylbenzene (ppb-day) | 0.01 | 19.2 | 99.2 | 151.1 | 217.6 | 455.9 | 8,082.1 |
| Xylene (ppb-day) | 1.58 | 430.5 | 909.0 | 1,234.0 | 1,612.5 | 2,770.0 | 24,413.1 |
| n-Hexane (ppb-day) | 0.06 | 39.4 | 181.3 | 290.3 | 504.7 | 1,507.7 | 62,717.1 |
| Single daily maximum | |||||||
| THC (ppm) | 0.01 | 0.2 | 0.6 | 0.4 | 1.2 | 2.8 | 22.4 |
| Benzene (ppb) | 0.00 | 1.3 | 3.1 | 1.9 | 8.9 | 15.1 | 62.5 |
| Toluene (ppb) | 0.04 | 4.6 | 16.2 | 7.0 | 38.3 | 66.9 | 188.0 |
| Ethylbenzene (ppb) | 0.00 | 1.2 | 2.4 | 1.4 | 6.2 | 16.0 | 137.0 |
| Xylene (ppb) | 0.51 | 10.2 | 17.1 | 11.9 | 39.8 | 75.0 | 445.3 |
| n-Hexane (ppb) | 0.06 | 2.2 | 6.0 | 3.6 | 25.3 | 68.1 | 2,441.0 |
In the IP-censoring weighted analysis, we saw modest increases in risk of CHD among those in the higher quintiles of cumulative daily maximum exposure to each agent in comparison with workers in the referent group. All exposure agents showed the strongest association in the top quintile (range of ), with the highest and significant HRs observed for THC (; 95% CI: 1.01, 2.01) and benzene (; 95% CI: 1.02, 2.02) (Table 3). Only THC showed elevated risks across all quintiles above the reference quintile. There were no apparent exposure–response trends for any exposure agents, and tests of trend were nonsignificant. Similar patterns of association were seen for cumulative daily average exposures (Table 3). The mean and range of the stabilized IP weights for each cumulative exposure are shown in Table S3. We observed similar results in analyses without IP-censoring weights (Table S4).
| Exposure | Cumulative daily maximum | Cumulative daily average | ||||
|---|---|---|---|---|---|---|
| Total cases () | Total no. () | HR (95% CI)a | Total cases () | Total no. () | HR (95% CI)a | |
| THC (ppm-daysb) | ||||||
| Q1 | 75 | 4,531 | Ref | 88 | 4,533 | Ref |
| Q2 | 111 | 4,531 | 1.29 (0.92, 1.80) | 106 | 4,529 | 1.08 (0.79, 1.47) |
| Q3 | 105 | 4,531 | 1.19 (0.85, 1.66) | 97 | 4,531 | 0.92 (0.67, 1.26) |
| Q4 | 102 | 4,531 | 1.18 (0.84, 1.66) | 104 | 4,531 | 1.00 (0.73, 1.36) |
| Q5 | 116 | 4,531 | 1.42 (1.01, 2.01) | 114 | 4,531 | 1.21 (0.88, 1.66) |
| Per ln(ppm-day) increasec | — | — | 1.05 (1.00, 1.10) | — | — | 1.05 (0.99, 1.11) |
| Benzene (ppb-daysb) | ||||||
| Q1 | 73 | 4,531 | Ref | 74 | 4,533 | Ref |
| Q2 | 113 | 4,531 | 1.29 (0.92, 1.82) | 112 | 4,529 | 1.35 (0.96, 1.91) |
| Q3 | 113 | 4,531 | 1.30 (0.92, 1.83) | 116 | 4,531 | 1.31 (0.93, 1.83) |
| Q4 | 81 | 4,531 | 0.89 (0.62, 1.28) | 81 | 4,531 | 0.91 (0.64, 1.30) |
| Q5 | 129 | 4,531 | 1.44 (1.02, 2.02) | 126 | 4,531 | 1.39 (0.99, 1.97) |
| Per ln(ppb-day) increasec | — | — | 1.04 (1.00, 1.09) | — | — | 1.03 (0.99, 1.08) |
| Per quintile increased | — | — | 1.03 (0.96, 1.10) | — | — | 1.02 (0.95, 1.09) |
| Toluene (ppb-daysb) | ||||||
| Q1 | 80 | 4,531 | Ref | 84 | 4,531 | Ref |
| Q2 | 109 | 4,531 | 1.19 (0.86, 1.66) | 106 | 4,531 | 1.10 (0.79, 1.53) |
| Q3 | 99 | 4,531 | 1.01 (0.72, 1.42) | 105 | 4,531 | 1.04 (0.75, 1.44) |
| Q4 | 92 | 4,531 | 0.90 (0.64, 1.27) | 83 | 4,531 | 0.79 (0.56, 1.11) |
| Q5 | 129 | 4,531 | 1.26 (0.91, 1.74) | 131 | 4,531 | 1.26 (0.90, 1.77) |
| Per ln(ppb-day) increasec | — | — | 1.04 (0.99, 1.09) | — | — | 1.03 (0.98, 1.08) |
| Per quintile increased | — | — | 1.01 (0.95, 1.09) | — | — | 1.02 (0.95, 1.10) |
| Ethylbenzene (ppb-daysb) | ||||||
| Q1 | 83 | 4,531 | Ref | 82 | 4,531 | Ref |
| Q2 | 104 | 4,531 | 1.02 (0.73, 1.43) | 103 | 4,531 | 1.09 (0.79, 1.52) |
| Q3 | 114 | 4,531 | 1.10 (0.79, 1.52) | 116 | 4,531 | 1.15 (0.84, 1.59) |
| Q4 | 93 | 4,531 | 0.87 (0.62, 1.21) | 95 | 4,531 | 0.95 (0.68, 1.33) |
| Q5 | 115 | 4,531 | 1.19 (0.85, 1.66) | 113 | 4,531 | 1.22 (0.87, 1.71) |
| Per ln(ppb-day) increasec | — | — | 1.04 (0.99, 1.08) | — | — | 1.04 (0.99, 1.09) |
| Per quintile increased | — | — | 1.01 (0.94, 1.08) | — | — | 1.01 (0.95, 1.09) |
| Xylene (ppb-daysb) | ||||||
| Q1 | 87 | 4,533 | Ref | 99 | 4,531 | Ref |
| Q2 | 104 | 4,529 | 1.12 (0.82, 1.52) | 97 | 4,531 | 0.90 (0.67, 1.22) |
| Q3 | 110 | 4,531 | 1.16 (0.85, 1.57) | 97 | 4,534 | 0.91 (0.68, 1.22) |
| Q4 | 90 | 4,531 | 0.87 (0.64, 1.19) | 94 | 4,528 | 0.81 (0.61, 1.10) |
| Q5 | 118 | 4,531 | 1.25 (0.91, 1.70) | 122 | 4,531 | 1.14 (0.85, 1.53) |
| Per ln(ppb-day) increasec | — | — | 1.04 (0.97, 1.11) | — | — | 1.03 (0.95, 1.11) |
| Per quintile increased | — | — | 1.02 (0.95, 1.09) | — | — | 1.01 (0.95, 1.09) |
| n-Hexane (ppb-daysb) | ||||||
| Q1 | 75 | 4,531 | Ref | 76 | 4,577 | Ref |
| Q2 | 100 | 4,531 | 1.10 (0.77, 1.56) | 108 | 4,485 | 1.19 (0.84, 1.67) |
| Q3 | 115 | 4,531 | 1.24 (0.88, 1.74) | 101 | 4,531 | 1.06 (0.75, 1.50) |
| Q4 | 102 | 4,531 | 1.00 (0.71, 1.42) | 106 | 4,531 | 1.03 (0.73, 1.46) |
| Q5 | 117 | 4,531 | 1.34 (0.94, 1.91) | 118 | 4,531 | 1.31 (0.92, 1.86) |
| Per ln(ppb-day) increasec | — | — | 1.04 (1.00, 1.08) | — | — | 1.03 (0.99, 1.08) |
| Per quintile increased | — | — | 1.04 (0.97, 1.12) | — | — | 1.04 (0.97, 1.11) |
Among ever smokers, we observed increased risk of CHD in the top quintile of cumulative daily maximum exposure to all agents, including statistically significant increases for THC (; 95% CI: 1.25, 2.97) and benzene (; 95% CI: 1.00, 2.33) (Table 4). Among never smokers, we found elevated HRs in the top quintile of cumulative daily maximum exposure to benzene, xylene, and n-hexane, although none of the associations were statistically significant. Results were generally similar when we evaluated cumulative daily average exposures, although associations among never smokers were more attenuated (Table S5). When we stratified analyses by education, associations for all agents were more pronounced among workers with high school education or less than among those with more than high school education (Table 5; Table S6). In analyses stratified by BMI, we observed stronger associations among workers who were nonobese in comparison with those who were obese (Table 6; Table S7). When we restricted the analysis to participants who were age 40 y or older, we observed generally similar associations, although risk estimates in the second and third quintiles of exposure to some agents became slightly stronger (Table S8).
| Exposure | Ever smokers | Never smokers | p for interactiond | ||||
|---|---|---|---|---|---|---|---|
| Total cases () | Total no. () | HR (95% CI)a | Total cases () | Total no. () | HR (95% CI)a | ||
| THC (ppm-daysb) | |||||||
| Q1 | 40 | 1,850 | Ref | 35 | 2,681 | Ref | 0.10 |
| Q2 | 78 | 2,307 | 1.81 (1.19, 2.76) | 33 | 2,224 | 0.74 (0.42, 1.28) | — |
| Q3 | 71 | 2,392 | 1.55 (1.02, 2.37) | 34 | 2,139 | 0.82 (0.48, 1.42) | — |
| Q4 | 68 | 2,472 | 1.50 (0.98, 2.30) | 34 | 2,059 | 0.88 (0.51, 1.52) | — |
| Q5 | 80 | 2,552 | 1.92 (1.25, 2.97) | 36 | 1,979 | 0.88 (0.51, 1.53) | — |
| Per ln(ppm-day) increasec | — | — | 1.05 (0.99, 1.13) | — | — | 1.05 (0.97, 1.13) | — |
| Benzene (ppb-daysb) | |||||||
| Q1 | 42 | 1,806 | Ref | 31 | 2,725 | Ref | 0.86 |
| Q2 | 73 | 2,345 | 1.34 (0.87, 2.06) | 40 | 2,186 | 1.27 (0.74, 2.19) | — |
| Q3 | 79 | 2,335 | 1.47 (0.96, 2.24) | 34 | 2,196 | 1.06 (0.61, 1.86) | — |
| Q4 | 56 | 2,462 | 0.93 (0.60, 1.46) | 25 | 2,069 | 0.88 (0.48, 1.61) | — |
| Q5 | 87 | 2,625 | 1.52 (1.00, 2.33) | 42 | 1,906 | 1.34 (0.77, 2.32) | — |
| Per ln(ppb-day) increasec | — | — | 1.04 (0.98, 1.09) | — | — | 1.05 (0.98, 1.13) | — |
| Toluene (ppb-daysb) | |||||||
| Q1 | 48 | 1,823 | Ref | 32 | 2,708 | Ref | 0.63 |
| Q2 | 69 | 2,288 | 1.17 (0.77, 1.78) | 40 | 2,243 | 1.27 (0.74, 2.18) | — |
| Q3 | 70 | 2,326 | 1.11 (0.73, 1.67) | 29 | 2,205 | 0.89 (0.51, 1.58) | — |
| Q4 | 58 | 2,516 | 0.87 (0.56, 1.33) | 34 | 2,015 | 1.02 (0.59, 1.79) | — |
| Q5 | 92 | 2,620 | 1.37 (0.92, 2.05) | 37 | 1,911 | 1.09 (0.63, 1.91) | — |
| Per ln(ppb-day) increasec | — | — | 1.05 (0.98, 1.11) | — | — | 1.05 (0.98, 1.13) | — |
| Ethylbenzene (ppb-daysb) | |||||||
| Q1 | 49 | 1,820 | Ref | 34 | 2,711 | Ref | 0.53 |
| Q2 | 71 | 2,337 | 1.07 (0.71, 1.62) | 33 | 2,194 | 0.95 (0.55, 1.64) | — |
| Q3 | 71 | 2,369 | 1.01 (0.67, 1.52) | 43 | 2,162 | 1.32 (0.79, 2.22) | — |
| Q4 | 65 | 2,447 | 0.94 (0.62, 1.43) | 28 | 2,084 | 0.78 (0.44, 1.39) | — |
| Q5 | 81 | 2,600 | 1.24 (0.82, 1.89) | 34 | 1,931 | 1.11 (0.64, 1.92) | — |
| Per ln(ppb-day) increasec | — | — | 1.04 (0.98, 1.10) | — | — | 1.03 (0.97, 1.11) | — |
| Xylene (ppb-daysb) | |||||||
| Q1 | 55 | 1,988 | Ref | 32 | 2,545 | Ref | 0.81 |
| Q2 | 67 | 2,207 | 1.15 (0.78, 1.68) | 37 | 2,322 | 1.22 (0.73, 2.04) | — |
| Q3 | 68 | 2,305 | 1.07 (0.74, 1.57) | 42 | 2,226 | 1.43 (0.86, 2.36) | — |
| Q4 | 63 | 2,477 | 0.91 (0.62, 1.34) | 27 | 2,054 | 0.87 (0.50, 1.52) | — |
| Q5 | 84 | 2,596 | 1.30 (0.89, 1.89) | 34 | 1,935 | 1.22 (0.71, 2.07) | — |
| Per ln(ppb-day) increasec | — | — | 1.05 (0.96, 1.15) | — | — | 1.05 (0.94, 1.18) | — |
| n-Hexane (ppb-daysb) | |||||||
| Q1 | 43 | 1,816 | Ref | 32 | 2,715 | Ref | 0.98 |
| Q2 | 63 | 2,281 | 1.11 (0.71, 1.73) | 37 | 2,250 | 1.11 (0.64, 1.93) | — |
| Q3 | 78 | 2,426 | 1.31 (0.85, 2.01) | 37 | 2,105 | 1.15 (0.66, 2.00) | — |
| Q4 | 71 | 2,488 | 1.12 (0.73, 1.73) | 31 | 2,043 | 0.87 (0.49, 1.56) | — |
| Q5 | 82 | 2,562 | 1.46 (0.94, 2.27) | 35 | 1,969 | 1.20 (0.68, 2.12) | — |
| Per ln(ppb-day) increasec | — | — | 1.06 (1.00, 1.12) | — | — | 1.02 (0.96, 1.08) | — |
| Exposure | High school or less | More than high school | p for interactiond | ||||
|---|---|---|---|---|---|---|---|
| Total cases () | Total no. () | HR (95% CI)a | Total cases () | Total no. () | HR (95% CI)a | ||
| THC (ppm-daysb) | |||||||
| Q1 | 25 | 1,186 | Ref | 50 | 3,345 | Ref | 0.16 |
| Q2 | 63 | 2,044 | 1.62 (0.98, 2.68) | 48 | 2,487 | 0.97 (0.62, 1.53) | — |
| Q3 | 61 | 2,167 | 1.50 (0.91, 2.48) | 44 | 2,364 | 0.90 (0.57, 1.42) | — |
| Q4 | 74 | 2,298 | 1.73 (1.06, 2.82) | 28 | 2,233 | 0.66 (0.40, 1.10) | — |
| Q5 | 70 | 2,472 | 1.72 (1.04, 2.84) | 46 | 2,059 | 1.15 (0.71, 1.88) | — |
| Per ln(ppm-day) increasec | — | — | 1.07 (0.99, 1.16) | — | — | 1.00 (0.93, 1.08) | — |
| Benzene (ppb-daysb) | |||||||
| Q1 | 24 | 1,093 | Ref | 49 | 3,438 | Ref | 0.64 |
| Q2 | 67 | 2,104 | 1.61 (0.98, 2.66) | 46 | 2,427 | 1.00 (0.63, 1.59) | — |
| Q3 | 65 | 2,124 | 1.55 (0.94, 2.56) | 48 | 2,407 | 1.04 (0.66, 1.64) | — |
| Q4 | 49 | 2,283 | 1.01 (0.60, 1.70) | 32 | 2,248 | 0.75 (0.45, 1.25) | — |
| Q5 | 88 | 2,563 | 1.84 (1.13, 3.01) | 41 | 1,968 | 1.02 (0.63, 1.65) | — |
| Per ln(ppb-day) increasec | — | — | 1.07 (1.00, 1.15) | — | — | 1.01 (0.95, 1.07) | — |
| Toluene (ppb-daysb) | |||||||
| Q1 | 26 | 1,112 | Ref | 54 | 3,419 | Ref | 0.15 |
| Q2 | 70 | 2,039 | 1.59 (0.97, 2.60) | 39 | 2,492 | 0.79 (0.49, 1.25) | — |
| Q3 | 50 | 2,077 | 1.02 (0.61, 1.71) | 49 | 2,454 | 0.94 (0.61, 1.46) | — |
| Q4 | 62 | 2,386 | 1.05 (0.64, 1.73) | 30 | 2,145 | 0.71 (0.43, 1.17) | — |
| Q5 | 85 | 2,553 | 1.54 (0.95, 2.50) | 44 | 1,978 | 0.91 (0.58, 1.44) | — |
| Per ln(ppb-day) increasec | — | — | 1.07 (0.99, 1.15) | — | — | 1.01 (0.94, 1.07) | — |
| Ethylbenzene (ppb-daysb) | |||||||
| Q1 | 27 | 1,115 | Ref | 56 | 3,416 | Ref | 0.05 |
| Q2 | 68 | 2,090 | 1.51 (0.93, 2.45) | 36 | 2,441 | 0.62 (0.39, 0.99) | — |
| Q3 | 61 | 2,166 | 1.25 (0.77, 2.04) | 53 | 2,365 | 0.98 (0.64, 1.49) | — |
| Q4 | 58 | 2,426 | 1.02 (0.62, 1.66) | 35 | 2,105 | 0.72 (0.45, 1.15) | — |
| Q5 | 79 | 2,370 | 1.67 (1.03, 2.71) | 36 | 2,161 | 0.79 (0.48, 1.32) | — |
| Per ln(ppb-day) increasec | — | — | 1.05 (0.98, 1.13) | — | — | 1.01 (0.95, 1.07) | — |
| Xylene (ppb-daysb) | |||||||
| Q1 | 35 | 1,499 | Ref | 52 | 3,034 | Ref | 0.25 |
| Q2 | 62 | 1,868 | 1.54 (1.00, 2.38) | 42 | 2,661 | 0.78 (0.50, 1.22) | — |
| Q3 | 58 | 1,989 | 1.44 (0.93, 2.22) | 52 | 2,542 | 0.93 (0.61, 1.42) | — |
| Q4 | 63 | 2,375 | 1.21 (0.79, 1.85) | 27 | 2,156 | 0.54 (0.33, 0.89) | — |
| Q5 | 75 | 2,436 | 1.57 (1.02, 2.40) | 43 | 2,095 | 1.00 (0.62, 1.61) | — |
| Per ln(ppb-day) increasec | — | — | 1.09 (0.98, 1.21) | — | — | 0.98 (0.88, 1.09) | — |
| n-Hexane (ppb-daysb) | |||||||
| Q1 | 27 | 1,080 | Ref | 48 | 3,451 | Ref | 0.17 |
| Q2 | 64 | 2,152 | 1.26 (0.77, 2.06) | 36 | 2,379 | 0.90 (0.55, 1.46) | — |
| Q3 | 57 | 2,203 | 1.11 (0.67, 1.83) | 58 | 2,328 | 1.36 (0.88, 2.11) | — |
| Q4 | 67 | 2,275 | 1.14 (0.70, 1.85) | 35 | 2,256 | 0.80 (0.49, 1.30) | — |
| Q5 | 78 | 2,457 | 1.50 (0.92, 2.45) | 39 | 2,074 | 1.16 (0.69, 1.93) | — |
| Per ln(ppb-day) increasec | — | — | 1.04 (0.98, 1.10) | — | — | 1.01 (0.95, 1.08) | — |
| Exposure | Not obese () | Obese () | p for interactiond | ||||
|---|---|---|---|---|---|---|---|
| Total cases () | Total no. () | HR (95% CI)a | Total cases () | Total no. () | HR (95% CI)a | ||
| THC (ppm-daysb) | |||||||
| Q1 | 39 | 3,279 | Ref | 36 | 1,252 | Ref | |
| Q2 | 69 | 3,125 | 1.95 (1.26, 3.02) | 42 | 1,406 | 0.83 (0.50, 1.37) | — |
| Q3 | 70 | 3,077 | 1.88 (1.22, 2.91) | 35 | 1,454 | 0.70 (0.42, 1.17) | — |
| Q4 | 50 | 3,048 | 1.41 (0.89, 2.24) | 52 | 1,483 | 1.02 (0.63, 1.66) | — |
| Q5 | 77 | 3,036 | 2.27 (1.44, 3.56) | 39 | 1,495 | 0.85 (0.51, 1.44) | — |
| Per ln(ppm-day) increasec | — | — | 1.10 (1.03, 1.17) | — | — | 1.00 (0.92, 1.08) | — |
| Benzene (ppb-daysb) | |||||||
| Q1 | 38 | 3,274 | Ref | 35 | 1,257 | Ref | 0.06 |
| Q2 | 64 | 3,082 | 1.59 (0.97, 2.61) | 49 | 1,449 | 1.07 (0.65, 1.75) | — |
| Q3 | 72 | 3,107 | 1.79 (1.10, 2.90) | 41 | 1,424 | 0.89 (0.54, 1.49) | — |
| Q4 | 57 | 3,113 | 1.36 (0.82, 2.24) | 24 | 1,418 | 0.49 (0.28, 0.86) | — |
| Q5 | 74 | 2,989 | 1.79 (1.10, 2.92) | 55 | 1,542 | 1.18 (0.72, 1.94) | — |
| Per ln(ppb-day) increasec | — | — | 1.08 (1.03, 1.14) | — | — | 0.99 (0.93, 1.06) | — |
| Toluene (ppb-daysb) | |||||||
| Q1 | 46 | 3,285 | Ref | 34 | 1,246 | Ref | 0.51 |
| Q2 | 64 | 3,107 | 1.40 (0.90, 2.17) | 45 | 1,424 | 0.99 (0.60, 1.65) | — |
| Q3 | 58 | 3,099 | 1.17 (0.75, 1.82) | 41 | 1,432 | 0.85 (0.51, 1.42) | — |
| Q4 | 58 | 3,044 | 1.16 (0.74, 1.82) | 34 | 1,487 | 0.66 (0.39, 1.12) | — |
| Q5 | 79 | 3,030 | 1.51 (0.98, 2.32) | 50 | 1,501 | 1.03 (0.62, 1.71) | — |
| Per ln(ppb-day) increasec | — | — | 1.08 (1.02, 1.14) | — | — | 1.00 (0.92, 1.07) | — |
| Ethylbenzene (ppb-daysb) | |||||||
| Q1 | 44 | 3,280 | Ref | 39 | 1,251 | Ref | 0.36 |
| Q2 | 65 | 3,121 | 1.34 (0.84, 2.14) | 39 | 1,410 | 0.75 (0.46, 1.24) | — |
| Q3 | 69 | 3,045 | 1.41 (0.89, 2.24) | 45 | 1,486 | 0.82 (0.50, 1.32) | — |
| Q4 | 57 | 3,035 | 1.17 (0.73, 1.87) | 36 | 1,496 | 0.61 (0.37, 1.01) | — |
| Q5 | 70 | 3,084 | 1.52 (0.95, 2.45) | 45 | 1,447 | 0.92 (0.56, 1.51) | — |
| Per ln(ppb-day) increasec | — | — | 1.08 (1.02, 1.14) | — | — | 0.99 (0.92, 1.06) | — |
| Xylene (ppb-daysb) | |||||||
| Q1 | 51 | 3,259 | Ref | 36 | 1,274 | Ref | 0.99 |
| Q2 | 64 | 3,130 | 1.20 (0.80, 1.80) | 40 | 1,399 | 1.03 (0.64, 1.67) | — |
| Q3 | 66 | 3,092 | 1.21 (0.81, 1.81) | 44 | 1,439 | 1.07 (0.67, 1.71) | — |
| Q4 | 54 | 3,008 | 0.96 (0.63, 1.45) | 36 | 1,523 | 0.77 (0.47, 1.26) | — |
| Q5 | 70 | 3,076 | 1.30 (0.86, 1.98) | 48 | 1,455 | 1.23 (0.77, 1.97) | — |
| Per ln(ppb-day) increasec | — | — | 1.06 (0.97, 1.16) | — | — | 1.01 (0.91, 1.13) | — |
| n-Hexane (ppb-daysb) | |||||||
| Q1 | 42 | 3,282 | Ref | 33 | 1,249 | Ref | 0.82 |
| Q2 | 60 | 3,050 | 1.25 (0.78, 2.02) | 40 | 1,481 | 0.93 (0.55, 1.57) | — |
| Q3 | 71 | 3,110 | 1.45 (0.91, 2.31) | 44 | 1,421 | 0.98 (0.59, 1.65) | — |
| Q4 | 61 | 3,084 | 1.16 (0.72, 1.87) | 41 | 1,447 | 0.81 (0.48, 1.37) | — |
| Q5 | 71 | 3,039 | 1.56 (0.96, 2.52) | 46 | 1,492 | 1.14 (0.67, 1.95) | — |
| Per ln(ppb-day) increasec | — | — | 1.06 (1.01, 1.12) | — | — | 1.01 (0.95, 1.08) | — |
We observed similar results in sensitivity analyses in which we, separately, accounted for precleanup hypertension (Tables S9), modeled age as splines (Table S10), included fatal events with CHD as a contributing/underlying cause of death (Table S11), or limited the outcome to nonfatal MI (Table S12). Results were also similar when we began follow-up at the time of enrollment instead of after each worker’s last date of cleanup work (Table S13). The analysis that excluded workers who had higher exposure from controlled burning produced somewhat attenuated effect estimates, especially in the top quintile, but results were not substantively different (Table S14). When we accounted in the model for ever having to stop working during the cleanup because of heat, we observed stronger associations for several exposure agents (Table S15). There was no substantive difference in results when we accounted for workers performing jobs that involved handling oily plants/wildlife or dead animal recovery (Table S16). In comparison with previously published results from this cohort, we observed elevated HRs in all upper quintiles of single daily maximum THC exposure in relation to CHD events accrued until the first follow-up interview and similar risk estimates when follow-up was extended to December 2019 (Table S17). When we coarsened the time interval in which CHD events were identified to explore the potential impact of in accurately recalling the date of CHD diagnosis (by up to a few months), we observed minimal difference in associations (Table S18). In the sensitivity analysis where we also included as cases workers who reported being diagnosed with a blockage in the arteries of the heart, we observed somewhat attenuated associations (Table S19).
In the subanalysis exploring risk of CHD among workers who had higher daily exposures for varying numbers of days, we observed nonsignificantly elevated HRs in comparison with workers who had consistently lower daily exposures (Table S20). As we increased the thresholds for daily exposure level from top 20th to top 15th percentile of single daily maximum exposure and for exposure duration from to d, we observed stronger effects for toluene but no noticeable changes in association for the other agents.
When we assessed the joint effect of the BTEX-H mixture using quantile g-computation, we found a negligible association for a per quintile increase in the entire mixture on CHD incidence (Table 7). A one quintile increase in cumulative daily maximum exposure and cumulative daily average exposure to all chemicals was associated with an increased risk of CHD of 1.03 (95% CI: 0.96, 1.10) and 1.02 (95% CI: 0.95, 1.09), respectively. Single-agent models that examined per quintile increase in exposures showed effect estimates of similar magnitude (Table 3).
Discussion
In this study, we examined the relationship between exposure to THC and BTEX-H and risk of CHD among oil spill workers up to 10 y after the DWH disaster. We observed a modest increase in risk of CHD among workers in the top quintile of cumulative exposure to these agents. Although there was no clear evidence of exposure–response trends and effect estimates in the upper quintiles of most exposure agents were not statistically significant, the magnitude of the effect in the highest exposure quintile in the main analysis was clinically meaningful and comparable to the increase in CHD risk from secondhand smoke exposure among nonsmokers (relative risk: 1.3), as estimated in a meta-analysis.58 In subgroup analyses, effect estimates were more pronounced among ever smokers, workers who had high school education or less, and workers who were not obese. In these subgroups, effect estimates in the top quintile of exposure to some agents were approaching the increase in risk of CHD comparing individuals who smoked one cigarette per day to never smokers (relative risk: 1.48–1.57), as reported in a recent, large meta-analysis.59
Several epidemiological studies have assessed ambient levels of volatile hydrocarbons in relation to MI and CHD events.8,9,60 Unlike these air pollution studies, which assessed exposure either across several days or over a year, most workers in our study were exposed to BTEX-H for several months. The maximum exposure levels experienced by workers in our study were generally higher than those reported in these air pollution studies but were well below occupational guidelines set by the American Conference of Governmental Industrial Hygienists (Table 2).61 Differences in exposure duration, emission source, and length of follow-up limits a direct comparison of our effect estimates with those of the other studies.
Among studies that focused on a longer exposure window, Barceló et al.9 reported higher risk of MI mortality for an increase in annual average daily levels of benzene, which is consistent with our results. Our analyses addressed limitations in this prior study by accounting for covariates such as education and smoking. Moreover, the longer follow-up time in our study allowed us to observe the sustained effect many years after exposure. Self-reported MI was also associated with annual mean concentrations of benzene among residents in an industrial area of Estonia, but interpretation of the findings was limited by the cross-sectional study design.62
Studies that focused on short-term exposures have also shown modest acute effects on cardiovascular health, including positive associations of benzene exposure across several days with MI occurrence6 and emergency hospitalizations for heart failure,60 and of benzene and alkylbenzene concentrations with circulatory mortality.7,8 These latter two studies have also associated acute cardiovascular events with exposure to alkanes,8,60 but we are unaware of any study that examined the alkane n-hexane specifically. In our analysis, we were underpowered to examine acute effects of exposure because only 22 (nonfatal) MI events occurred within a month of the individuals’ last day of cleanup work, and we lacked data on preenrollment fatal CHD events. Overall, however, the results of our study add to existing evidence that exposure to BTEX-H is associated with a modest increase in risk of CHD even at levels below the occupational limits, with a potential for persistent effects years after exposure.
As closely related volatile components of the crude oil, BTEX-H are present in gasolines and used as solvents and industrial raw materials for manufacturing of consumer products.63–67 All of these hydrocarbons have been detected at varying levels in vehicular exhaust, near sites of oil/gas operations and gas stations, and in certain occupational settings.20,68–72 Because individuals are typically exposed to many or all of these chemicals simultaneously, estimating the overall mixture effect can help inform interventions that target the emission sources. To our knowledge, only one study has examined the joint effect of specific crude oil chemical groups in relation to acute cardiovascular events. Ye et al.73 investigated emergency department visits for cardiovascular diseases and same-day exposure to prespecified chemical groups and found significant associations for most hydrocarbon groups although not for the aromatic group (which contains BTEX). In our mixture analysis, we also found little evidence of a joint effect for BTEX-H. In comparison with single-agent models that examined per quintile increase in each exposure, effects in the mixture model were not noticeably stronger. The overall weak associations in the mixture analysis and its single-agent counterparts could be attributed to the apparent nonlinear relationship between the exposure and the outcome, where effects were present only above an exposure threshold (e.g., in Q5). Because other ambient pollution studies only examined exposures continuously, it is unclear whether a threshold effect existed in those other studies.
To our knowledge, our study is the first to investigate quantitative BTEX-H exposures and risk of CHD in an occupational setting. In an earlier analysis that followed the GuLF Study cohort for only 5 y, Strelitz et al.17 found a statistically significant association between an ordinal measure of single daily maximum THC exposure and hazard of a CHD event. In addition, those who worked d also had higher hazard of CHD in comparison with those who worked d.16 Longer duration of work was also associated with increased risks of self-reported MI or angina among Hebei Spirit oil spill workers up to 10 y after the spill.14 Although studies have shown a positive relationship between duration of exposure and CHD risk, no study has assessed cumulative exposure to THC or specific chemicals, which accounts for exposure duration. Taking advantage of recently developed exposure estimates and a longer duration of follow-up, we showed that cumulative exposure to THC and BTEX-H was associated with increased risk of MI, although effect estimates were lower than that associated with the ordinal single daily maximum THC exposure observed previously. The stronger associations observed for the single daily maximum THC exposure suggest that exposure to crude oil chemicals at a high intensity might have induced damage, possibly inflammatory and/or vascular, that increased workers’ risk of CHD over time. Workers with higher single daily maximum exposure, however, also tended to have higher cumulative exposures. In analyses that further assessed the role of exposure intensity in CHD risk by comparing workers who were exposed to higher daily maximum exposures for varying number of days with workers with consistently lower daily exposures, increasing the threshold of exposure intensity or exposure duration led to stronger effect estimates for toluene, but no meaningful changes in associations were observed for the other exposure agents.
In analyses stratified by cigarette smoking status, we saw elevated risks in the top quintile of exposure to most agents in both ever and never smokers, although effects were generally more pronounced among ever smokers. In the analysis stratified by highest educational attainment, we found stronger effects in the subgroup with no more than high school education. This finding is consistent with two studies that found stronger associations between traffic-related air pollution and cardiovascular mortality among participants with lower educational attainment.40,41
In analyses stratified by obesity, we observed elevated HRs among workers with higher exposure to all agents in the nonobese subgroup, with generally weaker associations in the obese subgroup. One possible explanation for the weaker associations among the obese group is that variability of CHD risk is likely higher among these workers, so it is easy for the modest effect of oil exposures to be lost among this high variability. We hypothesize the risk to vary more among obese workers because their baseline risk was higher,74 so lifestyle factors, such as physical activities, would have a larger impact to alter each individual’s overall risk of CHD.75 In addition, obese workers were more likely to have other cardiovascular comorbidities,76 which would also have led to variability in CHD risk among this group. We were not able to account for all the lifestyle and health factors (e.g., diet, physical activity, cholesterol levels) that could have led to the wider range of risks in the obese group.
Two major mechanisms have been proposed that link short- and long-term exposure to air pollutants to CHD events. One mechanism involves activation of pulmonary and systemic inflammatory responses by inhaled pollutants.77,78 This inflammation and associated oxidative stress can stimulate the circulatory release of inflammatory proteins and coagulation factors, accelerating atherosclerosis and increasing the potential for thrombosis. Although an individual’s rate of atherosclerosis depends on their risk profile,79 a recent study showed rapid atherosclerotic progression among participants 40–54 y of age and without a history of cardiovascular disease during 3 y of follow-up.80 The vast majority of CHD cases in our study occurred among participants over 40 y old. Exposure to crude oil chemicals could have accelerated atherosclerosis among these workers and resulted in increased CHD events in the several years after exposure. The other mechanism involves modulation of the autonomic nervous system by pollutants trapped in the respiratory tract.77,78 A shift of the system toward the sympathetic tone elevates blood pressure, which can lead to increased risk of CHD.78 Among susceptible individuals with advanced atherosclerosis, air pollutant-induced inflammation and vasoconstriction may destabilize existing plaques and trigger an acute CHD event soon after exposure.77,78 We lacked the necessary data and sufficient cases to examine the association between exposure to crude oil chemicals and acute CHD events in this study.
Evidence supporting these possible mechanisms comes from studies showing elevated levels of oxidative stress and inflammation68,81–83 and higher risk of hypertension and electrocardiographic abnormalities84–86 among workers exposed to BTEX-H chemicals. Higher levels of oxidative stress were also observed among Hebei Spirit oil spill responders in relation to work duration.14 Among GuLF Study participants, cumulative THC exposure was associated with increased risk of hypertension 1–3 y after the spill.87 In addition, Coast Guard responders who participated in the DWH spill cleanup and reported crude oil inhalation exposure had increased risks of essential hypertension and heart palpitations 3–5 y after the spill.88 Together, current mechanistic understanding supports the plausibility of the observed cardiovascular effects of exposure to THC and BTEX-H.
A major strength of our study is the careful reconstruction of THC and BTEX-H exposures using personal air samples collected on the OSRC workers, as well as detailed work histories collected from the study participants. In addition, although most air pollution studies examined MI/CHD events immediately following either acute (days) or long-term (years) exposures, our study was unique in assessing medium-duration exposures that lasted weeks to a few months and persistent cardiovascular effects many years after cessation of exposure. Another strength of the study is the use of IP-censoring weights to account for potential informative censoring due to loss to follow-up. Associations were similar in the IP-weighted models and models without weights, suggesting that our results are robust to the potential bias from informative censoring. Last, subject-specific data on education, BMI, and smoking status allowed us to perform stratified analyses by these traits to identify groups that might be particularly vulnerable to the effects of these exposures. We are unaware of any studies of crude oil chemicals and CHD that have examined these effect measure modifiers.
One limitation of the study is potential misclassification of the outcome, as we could not obtain medical records from participants to confirm their MI diagnosis or cause of death. Previous studies have reported moderate to high accuracy of self-reported MI and of death certificate diagnosis of CHD, with sensitivities ranging from 0.78 to 0.98 and specificities varying from 0.72 to 1.0.89–98 Many of these studies have associated lower accuracy with older age.93,94,98,99 In comparison with participants examined in these validation studies, workers in our study were younger (most were y of age at enrollment), so we expect a lower degree of outcome misclassification in our population. There could be measurement error in the reported event time due to participants’ inaccurate recall of the date of MI diagnosis. However, exploratory analyses in which we coarsened the follow-up from 1 month to 4 months showed no notable changes from the main analysis results, which suggests that our analysis was robust to measurement error of at least a few months in recall time.
Second, the exposure estimates assigned to workers contain some degree of uncertainty.18 The left-censored Bayesian method used to estimate EG average exposures had low bias and imprecision25,26,100 and performed well for most EGs, but some EGs with extreme censoring or few measurements had more uncertainty.21 However, this possibility is unlikely to have substantively biased our risk estimates in analyses using categorical exposures because the vast majority of workers in these generally low-exposure EGs fell into the lowest exposure category. In addition, although many participants worked on multiple jobs/tasks during the cleanup, we lack information on the specific days and the exact number of hours on those days that they performed each job/activity, which increased uncertainty in the estimates of daily exposures. To overcome this limitation, we examined both cumulative daily maximum and cumulative daily average exposures, which produced similar results.
Third, in our analysis, we were not able to identify CHD deaths that occurred among OSRC workers between exposure and study enrollment because enrollment was contingent on survival. If more CHD deaths occurred before enrollment among DWH oil spill workers exposed to higher levels of crude oil chemicals, our results might have underestimated the true risk. However, given the relatively short time between exposure and enrollment (i.e., immortal time for the fatal outcome), the overall small number of fatal CHD cases during the entire follow-up, and the continued risk over the years of observation, we do not expect left truncation of these fatal events to meaningfully change our results. In a subanalysis, we examined nonfatal MI as the outcome, for which there was no immortal time bias, and observed similar associations. In another analysis, we explored the impact of starting follow-up at a worker’s enrollment in the cohort rather than after the last day of their cleanup work and observed slightly attenuated associations.
Last, there could be bias from unmeasured confounders or imperfect measurement of existing covariates in the models, which may explain in part the lack of observed exposure–response trends. We lacked the necessary data to adjust for certain other potential exposures experienced during the oil spill cleanup.4 In a sensitivity analysis, we accounted for one important occupational exposure that has been associated with CHD— from controlled burning by excluding workers who experienced higher exposure—and found somewhat attenuated results. The majority of workers who had higher exposure were also exposed to higher levels of THC and BTEX-H (proportion in Q4 or Q5: 71%–91% by exposure agent). It is likely that workers who were exposed to both higher burning-related and crude oil chemicals had even higher risks of CHD, but we had few cases to examine associations among this subgroup or to adjust for exposure levels. We also accounted for a measure of heat stress and one OSRC work that could have induced psychological stress among workers and observed similar or stronger associations. There could also be a bias if workers were assigned to different jobs/activities based on their health factors at the time of spill that were predictive of their future CHD risk. We adjusted for several indicators of baseline health (BMI, precleanup diabetes, precleanup hypertension, smoking) to reduce this potential bias. Last, some of our adjustment factors, such as cigarette smoking and BMI, were ascertained at enrollment as proxies for factors at the time of exposure and might have changed over time. However, we expect little change in these factors over the short span between exposure and time of their ascertainment.
Our study showed modestly increased risk of CHD among oil spill workers exposed to higher levels of THC and BTEX-H, with the strongest associations observed in the highest quintile of each exposure; however, we did not observe clear exposure–response trends. The positive associations were consistent with evidence from ambient air pollution research indicating that exposure to these agents at levels below occupational guidelines may induce adverse cardiovascular effects. To our knowledge, our study is the first to evaluate the relationship between exposure to individual crude oil chemicals and risk of CHD in the occupational setting. We further showed stronger associations in some subgroups, i.e., ever smokers, workers who had high school education or less, and workers who were not obese. Additional research is needed in other populations and settings to confirm these study findings.
Supplementary Material
Acknowledgments
The authors present this work in memory and grateful recognition of G. Heiss, Kenan Distinguished Professor of Epidemiology at the University of North Carolina at Chapel Hill Gillings School of Global Public Health, who passed away on 11 June 2022, during the completion of this manuscript, on which he played a critical role.
The authors would like to thank K. Christenbury and M. Bodkin for data management on this project.
This research was supported by the Intramural Research Program of the National Institutes of Health (NIH), National Institute of Environmental Health Sciences (NIEHS) (ZO1 ES 102945), as well as the following grants: NIH/NIEHS R01ES027027-01, NIH/NIEHS R01ES030210-01, National Science Foundation (NSF) Division of Mathematical Sciences (DMS)-1513654, NSF Information and Intelligent Systems (IIS)-1562303, and NSF DMS-1916349.
References
Untitled section
References
- 1.National Commission on the BP Deepwater Horizon Oil Spill and Offshore Drilling. 2011. Deep Water: The Gulf Oil Disaster and the Future of Offshore Drilling, Report to the President. Washington, DC: National Commission on the BP Deepwater Horizon Oil Spill and Offshore Drilling.
- 2.U.S. Coast Guard. 2011. On Scene Coordinator Report: Deepwater Horizon Oil Spill. Washington, DC: U.S. Dept. of Homeland Security, U.S. Coast Guard.
- 3.Kwok RK, Engel LS, Miller AK, Blair A, Curry MD, Jackson WB, et al. 2017. The Gulf STUDY: a prospective study of persons involved in the Deepwater Horizon oil spill response and clean-up. Environ Health Perspect 125(4):570–578, PMID: , 10.1289/EHP715.
- 4.Middlebrook AM, Murphy DM, Ahmadov R, Atlas EL, Bahreini R, Blake DR, et al. 2012. Air quality implications of the Deepwater Horizon oil spill. Proc Natl Acad Sci USA 109(50):20280–20285, PMID: , 10.1073/pnas.1110052108.
- 5.Batavia MK. 1991. Clean Air Act Amendments of 1990. J S C Med Assoc 87(2):75–77, PMID: .
- 6.Bard D, Kihal W, Schillinger C, Fermanian C, Ségala C, Glorion S, et al. 2014. Traffic-related air pollution and the onset of myocardial infarction: disclosing benzene as a trigger? A small-area case-crossover study. PLoS One 9(6):e100307, PMID: , 10.1371/journal.pone.0100307.
- 7.Ran J, Qiu H, Sun S, Tian L. 2018. Short-term effects of ambient benzene and TEX (toluene, ethylbenzene, and xylene combined) on cardiorespiratory mortality in Hong Kong. Environ Int 117:91–98, PMID: , 10.1016/j.envint.2018.04.049.
- 8.Tsai DH, Wang JL, Chuang KJ, Chan CC. 2010. Traffic-related air pollution and cardiovascular mortality in central Taiwan. Sci Total Environ 408(8):1818–1823, PMID: , 10.1016/j.scitotenv.2010.01.044.
- 9.Barceló MA, Varga D, Tobias A, Diaz J, Linares C, Saez M. 2016. Long term effects of traffic noise on mortality in the city of Barcelona, 2004–2007. Environ Res 147:193–206, PMID: , 10.1016/j.envres.2016.02.010.
- 10.Zock J-P, Rodríguez-Trigo G, Rodríguez-Rodríguez E, Espinosa A, Pozo-Rodríguez F, Gómez F, et al. 2012. Persistent respiratory symptoms in clean-up workers 5 years after the prestige oil spill. Occup Environ Med 69(7):508–513, PMID: , 10.1136/oemed-2011-100614.
- 11.Gam KB, Kwok RK, Engel LS, Curry MD, Stewart PA, Stenzel MR, et al. 2018. Lung function in oil spill response workers 1-3 years after the Deepwater Horizon disaster. Epidemiology 29(3):315–322, PMID: , 10.1097/EDE.0000000000000808.
- 12.Chen D, Lawrence KG, Pratt GC, Stenzel MR, Stewart PA, Groth CP, et al. 2022. Fine particulate matter and lung function among burning-exposed Deepwater Horizon oil spill workers. Environ Health Perspect 130(2):27001, PMID: , 10.1289/EHP8930.
- 13.Meo SA, Meo IMU, Al-Drees AM, Al-Saadi MM, Azeem MA. 2008. Lung function in subjects exposed to crude oil spill into sea water. Mar Pollut Bull 56(1):88–94, PMID: , 10.1016/j.marpolbul.2007.09.039.
- 14.Lee M, Park MS, Cheong HK. 2020. An association between oil spill clean-up work and cardiovascular disease. Ecotoxicol Environ Saf 194:110284, PMID: , 10.1016/j.ecoenv.2020.110284.
- 15.Strelitz J, Engel LS, Kwok RK, Miller AK, Blair A, Sandler DP. 2018. Deepwater horizon oil spill exposures and nonfatal myocardial infarction in the GuLF STUDY. Environ Health 17(1):69, PMID: , 10.1186/s12940-018-0408-8.
- 16.Strelitz J, Keil AP, Richardson DB, Heiss G, Gammon MD, Kwok RK, et al. 2019. Self-reported myocardial infarction and fatal coronary heart disease among oil spill workers and community members 5 years after Deepwater Horizon. Environ Res 168:70–79, PMID: , 10.1016/j.envres.2018.09.026.
- 17.Strelitz J, Sandler DP, Keil AP, Richardson DB, Heiss G, Gammon MD, et al. 2019. Exposure to total hydrocarbons during cleanup of the Deepwater Horizon oil spill and risk of heart attack across 5 years of follow-up. Am J Epidemiol 188(5):917–927, PMID: , 10.1093/aje/kwz017.
- 18.Stewart P, Groth CP, Huynh TB, Gorman Ng M, Pratt GC, Arnold SF, et al. 2022. Assessing exposures from the Deepwater Horizon oil spill response and clean-up. Ann Work Expo Health 66(suppl 1):i3–i22, PMID: , 10.1093/annweh/wxab107.
- 19.Dehghani M, Fazlzadeh M, Sorooshian A, Tabatabaee HR, Miri M, Baghani AN, et al. 2018. Characteristics and health effects of BTEX in a hot spot for urban pollution. Ecotoxicol Environ Saf 155:133–143, PMID: , 10.1016/j.ecoenv.2018.02.065.
- 20.Gilman JB, Lerner BM, Kuster WC, de Gouw JA. 2013. Source signature of volatile organic compounds from oil and natural gas operations in northeastern Colorado. Environ Sci Technol 47(3):1297–1305, PMID: , 10.1021/es304119a.
- 21.Stenzel MR, Groth CP, Banerjee S, Ramachandran G, Kwok RK, Engel LS, et al. 2022. Exposure assessment techniques applied to the highly censored Deepwater Horizon Gulf oil spill personal measurements. Ann Work Expo Health 66(suppl 1):i56–i70, PMID: , 10.1093/annweh/wxab060.
- 22.Groth CP, Banerjee S, Ramachandran G, Stewart PA, Sandler DP, Blair A, et al. 2022. Methods for the analysis of 26 million VOC area measurements during the Deepwater Horizon oil spill clean-up. Ann Work Expo Health 66(suppl 1):i140–i155, PMID: , 10.1093/annweh/wxab038.
- 23.Ramachandran G, Groth CP, Huynh TB, Banerjee S, Stewart PA, Engel LS, et al. 2022. Using Real-time area VOC measurements to estimate total hydrocarbons exposures to workers involved in the Deepwater Horizon oil spill. Ann Work Expo Health 66(suppl 1):i156–i171, PMID: , 10.1093/annweh/wxab066.
- 24.Stenzel MR, Groth CP, Huynh TB, Ramachandran G, Banerjee S, Kwok RK, et al. 2022. Exposure group development in support of the NIEHS GuLF Study. Ann Work Expo Health 66(suppl 1):i23–i55, PMID: , 10.1093/annweh/wxab093.
- 25.Groth C, Banerjee S, Ramachandran G, Stenzel MR, Sandler DP, Blair A, et al. 2017. Bivariate Left-censored Bayesian model for predicting exposure: preliminary analysis of worker exposure during the Deepwater Horizon oil spill. Ann Work Expo Health 61(1):76–86, PMID: , 10.1093/annweh/wxw003.
- 26.Groth CP, Huynh TB, Banerjee S, Ramachandran G, Stewart PA, Quick H, et al. 2022. Linear relationships between total hydrocarbons and benzene, toluene, ethylbenzene, xylene, and n-hexane during the Deepwater Horizon response and clean-up. Ann Work Expo Health 66(suppl 1):i71–i88, PMID: , 10.1093/annweh/wxab064.
- 27.Huynh TB, Groth CP, Ramachandran G, Banerjee S, Stenzel M, Blair A, et al. 2022. Estimates of inhalation exposures among land workers during the Deepwater Horizon oil spill clean-up operations. Ann Work Expo Health 66(suppl 1):i124–i139, PMID: , 10.1093/annweh/wxab028.
- 28.Huynh TB, Groth CP, Ramachandran G, Banerjee S, Stenzel M, Blair A, et al. 2022. Estimates of inhalation exposures to oil-related components on the supporting vessels during the Deepwater Horizon oil spill. Ann Work Expo Health 66(suppl 1):i111–i123, PMID: , 10.1093/annweh/wxaa113.
- 29.Huynh TB, Groth CP, Ramachandran G, Banerjee S, Stenzel M, Quick H, et al. 2022. Estimates of occupational inhalation exposures to six oil-related compounds on the four rig vessels responding to the Deepwater Horizon oil spill. Ann Work Expo Health 66(suppl 1):i89–i110, PMID: , 10.1093/annweh/wxaa072.
- 30.Cox DR. 1972. Regression models and life-tables. J R Stat Soc Series B Stat Methodol 34(2):187–202, 10.1111/j.2517-6161.1972.tb00899.x.
- 31.Keil AP, Edwards JK. 2018. A review of time scale fundamentals in the g-formula and insidious selection bias. Curr Epidemiol Rep 5(3):205–213, PMID: .
- 32.Cole SR, Hernan MA. 2004. Adjusted survival curves with inverse probability weights. Comput Methods Programs Biomed 75(1):45–49, PMID: , 10.1016/j.cmpb.2003.10.004.
- 33.Greenland S, Pearl J, Robins JM. 1999. Causal diagrams for epidemiologic research. Epidemiology 10(1):37–48, PMID: , 10.1097/00001648-199901000-00008.
- 34.Brookhart MA, Schneeweiss S, Rothman KJ, Glynn RJ, Avorn J, Stürmer T. 2006. Variable selection for propensity score models. Am J Epidemiol 163(12):1149–1156, PMID: , 10.1093/aje/kwj149.
- 35.Textor J, van der Zander B, Gilthorpe MS, Liskiewicz M, Ellison GT. 2016. Robust causal inference using directed acyclic graphs: the R package ‘dagitty.’ Int J Epidemiol 45(6):1887–1894, PMID: , 10.1093/ije/dyw341.
- 36.Naimi AI, Moodie EE, Auger N, Kaufman JS. 2014. Constructing inverse probability weights for continuous exposures: a comparison of methods. Epidemiology 25(2):292–299, PMID: , 10.1097/EDE.0000000000000053.
- 37.Hernán MA, Hernández-Díaz S, Robins JM. 2004. A structural approach to selection bias. Epidemiology 15(5):615–625, PMID: , 10.1097/01.ede.0000135174.63482.43.
- 38.Howe CJ, Cole SR, Lau B, Napravnik S, Eron JJ Jr.. 2016. Selection bias due to loss to follow up in cohort studies. Epidemiology 27(1):91–97, PMID: , 10.1097/EDE.0000000000000409.
- 39.National Center for Chronic Disease P, Health Promotion Office on S, Health. 2014. Reports of the Surgeon General. In: The Health Consequences of Smoking—50 Years of Progress: A Report of the Surgeon General. Atlanta, GA: U.S. Centers for Disease Control and Prevention.
- 40.Ostro BD, Feng WY, Broadwin R, Malig BJ, Green RS, Lipsett MJ. 2008. The impact of components of fine particulate matter on cardiovascular mortality in susceptible subpopulations. Occup Environ Med 65(11):750–756, PMID: , 10.1136/oem.2007.036673.
- 41.Beelen R, Hoek G, van den Brandt PA, Goldbohm RA, Fischer P, Schouten LJ, et al. 2008. Long-term effects of traffic-related air pollution on mortality in a Dutch cohort (NLCS-AIR study). Environ Health Perspect 116(2):196–202, PMID: , 10.1289/ehp.10767.
- 42.Kraus WE, Powell KE, Haskell WL, Janz KF, Campbell WW, Jakicic JM, et al. 2019. Physical activity, all-cause and cardiovascular mortality, and cardiovascular disease. Med Sci Sports Exerc 51(6):1270–1281, PMID: , 10.1249/MSS.0000000000001939.
- 43.Romieu I, Téllez-Rojo MM, Lazo M, Manzano-Patiño A, Cortez-Lugo M, Julien P, et al. 2005. Omega-3 fatty acid prevents heart rate variability reductions associated with particulate matter. Am J Respir Crit Care Med 172(12):1534–1540, PMID: , 10.1164/rccm.200503-372OC.
- 44.Bhatnagar A. 2017. Environmental determinants of cardiovascular disease. Circ Res 121(2):162–180, PMID: , 10.1161/CIRCRESAHA.117.306458.
- 45.Droomers M, Schrijvers CT, Mackenbach JP. 2001. Educational level and decreases in leisure time physical activity: predictors from the longitudinal GLOBE study. J Epidemiol Community Health 55(8):562–568, PMID: , 10.1136/jech.55.8.562.
- 46.Hajat A, Diez-Roux AV, Adar SD, Auchincloss AH, Lovasi GS, O’Neill MS, et al. 2013. Air pollution and individual and neighborhood socioeconomic status: evidence from the Multi-Ethnic Study of Atherosclerosis (MESA). Environ Health Perspect 121(11–12):1325–1333, PMID: , 10.1289/ehp.1206337.
- 47.Larson NI, Story MT, Nelson MC. 2009. Neighborhood environments: disparities in access to healthy foods in the U.S. Am J Prev Med 36(1):74–81, PMID: , 10.1016/j.amepre.2008.09.025.
- 48.Lumeng CN, Saltiel AR. 2011. Inflammatory links between obesity and metabolic disease. J Clin Invest 121(6):2111–2117, PMID: , 10.1172/JCI57132.
- 49.Matos CMP, Moraes KS, França DC, Tomich GM, Farah MW, Dias RC, et al. 2012. Changes in breathing pattern and thoracoabdominal motion after bariatric surgery: a longitudinal study. Respir Physiol Neurobiol 181(2):143–147, PMID: , 10.1016/j.resp.2012.02.009.
- 50.Graham DR, Chamberlain MJ, Hutton L, King M, Morgan WK. 1990. Inhaled particle deposition and body habitus. Br J Ind Med 47(1):38–43, PMID: , 10.1136/oem.47.1.38.
- 51.Brochu P, Bouchard M, Haddad S. 2014. Physiological daily inhalation rates for health risk assessment in overweight/obese children, adults, and elderly. Risk Anal 34(3):567–582, PMID: , 10.1111/risa.12125.
- 52.Gonçalves VSS, Andrade KRC, Carvalho KMB, Silva MT, Pereira MG, Galvao TF. 2018. Accuracy of self-reported hypertension: a systematic review and meta-analysis. J Hypertens 36(5):970–978, PMID: , 10.1097/HJH.0000000000001648.
- 53.Pratt GC, Stenzel MR, Kwok RK, Groth CP, Banerjee S, Arnold SF, et al. 2022. Modeled air pollution from in situ burning and flaring of oil and gas released following the Deepwater Horizon disaster. Ann Work Expo Health 66(suppl 1):i172–i187, PMID: , 10.1093/annweh/wxaa084.
- 54.Chen D, Sandler DP, Keil AP, Heiss G, Whitsel EA, Pratt GC, et al. 2023. Fine particulate matter and incident coronary heart disease events up to 10 years of follow-up among Deepwater Horizon oil spill workers. Environ Res 217:114841, PMID: , 10.1016/j.envres.2022.114841.
- 55.Jackson CA, Sudlow CLM, Mishra GD. 2018. Psychological distress and risk of myocardial infarction and stroke in the 45 and Up Study. Circ Cardiovasc Qual Outcomes 11(9):e004500, PMID: , 10.1161/CIRCOUTCOMES.117.004500.
- 56.Keil AP, Buckley JP, O’Brien KM, Ferguson KK, Zhao S, White AJ. 2020. A quantile-based g-computation approach to addressing the effects of exposure mixtures. Environ Health Perspect 128(4):47004, PMID: , 10.1289/EHP5838.
- 57.White AJ, O’Brien KM, Niehoff NM, Jackson BP, Karagas MR, Weinberg CR, et al. 2020. Toenail metal concentrations and age at menopause: a prospective study. Environ Epidemiol 4(4):e0104, PMID: , 10.1097/EE9.0000000000000104.
- 58.He J, Vupputuri S, Allen K, Prerost MR, Hughes J, Whelton PK. 1999. Passive smoking and the risk of coronary heart disease–a meta-analysis of epidemiologic studies. N Engl J Med 340(12):920–926, PMID: , 10.1056/NEJM199903253401204.
- 59.Hackshaw A, Morris JK, Boniface S, Tang JL, Milenković D. 2018. Low cigarette consumption and risk of coronary heart disease and stroke: meta-analysis of 141 cohort studies in 55 study reports. BMJ 360:j5855, PMID: , 10.1136/bmj.j5855.
- 60.Ran J, Qiu H, Sun S, Yang A, Tian L. 2018. Are ambient volatile organic compounds environmental stressors for heart failure? Environ Pollut 242(pt B):1810–1816, PMID: , 10.1016/j.envpol.2018.07.086.
- 61.American Conference of Governmental Industrial Hygienists. 2012. Appendix B. ACGIH Threshold Limit Values (TLVs) and Biological Exposure Indices (BEIs). In: 2012 TLVs and BEIs. 7th ed. Cincinnati, OH: American Conference of Governmental Industrial Hygienists.
- 62.Orru H, Idavain J, Pindus M, Orru Kati, Kesanurm K, Lang A, et al. 2018. Residents’ self-reported health effects and annoyance in relation to air pollution exposure in an industrial area in Eastern-Estonia. Int J Environ Res Public Health 15(2):252, PMID: , 10.3390/ijerph15020252.
- 63.ATSDR (Agency for Toxic Substances and Disease Registry). 2007. Toxicological Profile for Benzene. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service, Agency for Toxic Substances and Disease Registry.
- 64.ATSDR. 2010. Toxicological Profile for Ethylbenzene. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service, Agency for Toxic Substances and Disease Registry.
- 65.ATSDR. 1999. Toxicological Profile for n-Hexane. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service, Agency for Toxic Substances and Disease Registry.
- 66.ATSDR. 2000. Toxicological Profile for Toluene. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service, Agency for Toxic Substances and Disease Registry.
- 67.ATSDR. 2007. Toxicological Profile for Xylene. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service, Agency for Toxic Substances and Disease Registry.
- 68.Xiong F, Li Q, Zhou B, Huang J, Liang G, Zhang L, et al. 2016. Oxidative stress and genotoxicity of long-term occupational exposure to low levels of BTEX in gas station workers. Int J Environ Res Public Health 13(12):1212, PMID: , 10.3390/ijerph13121212.
- 69.Heibati B, Pollitt KJG, Karimi A, Yazdani Charati J, Ducatman A, Shokrzadeh M, et al. 2017. BTEX exposure assessment and quantitative risk assessment among petroleum product distributors. Ecotoxicol Environ Saf 144:445–449, PMID: , 10.1016/j.ecoenv.2017.06.055.
- 70.Fujita EM, Campbell DE, Zielinska B, Arnott WP, Chow JC. 2011. Concentrations of air toxics in motor vehicle-dominated environments. Res Rep Health Eff Inst 156:3–77, PMID: .
- 71.Wilson MP, Hammond SK, Nicas M, Hubbard AE. 2007. Worker exposure to volatile organic compounds in the vehicle repair industry. J Occup Environ Hyg 4(5):301–310, PMID: , 10.1080/15459620701258021.
- 72.Bogadi-Sare A, Zavalic M, Trosić I, Turk R, Kontosić I, Jelcić I. 2000. Study of some immunological parameters in workers occupationally exposed to benzene. Int Arch Occup Environ Health 73(6):397–400, PMID: , 10.1007/s004200000126.
- 73.Ye D, Klein M, Chang HH, Sarnat JA, Mulholland JA, Edgerton ES, et al. 2017. Estimating acute cardiorespiratory effects of ambient volatile organic compounds. Epidemiology 28(2):197–206, PMID: , 10.1097/EDE.0000000000000607.
- 74.Bogers RP, Bemelmans WJE, Hoogenveen RT, Boshuizen HC, Woodward M, Knekt P, et al. 2007. Association of overweight with increased risk of coronary heart disease partly independent of blood pressure and cholesterol levels: a meta-analysis of 21 cohort studies including more than 300,000 persons. Arch Intern Med 167(16):1720–1728, PMID: , 10.1001/archinte.167.16.1720.
- 75.Koolhaas CM, Dhana K, Schoufour JD, Ikram MA, Kavousi M, Franco OH. 2017. Impact of physical activity on the association of overweight and obesity with cardiovascular disease: The Rotterdam Study. Eur J Prev Cardiol 24(9):934–941, PMID: , 10.1177/2047487317693952.
- 76.Powell-Wiley TM, Poirier P, Burke LE, Després J-P, Gordon-Larsen P, Lavie CJ, et al. 2021. Obesity and cardiovascular disease: a scientific statement from the American Heart Association. Circulation 143(21):e984–e1010, PMID: , 10.1161/CIR.0000000000000973.
- 77.Brook RD, Franklin B, Cascio W, Hong Y, Howard G, Lipsett M, et al. 2004. Air pollution and cardiovascular disease: a statement for healthcare professionals from the Expert Panel on Population and Prevention Science of the American Heart Association. Circulation 109(21):2655–2671, PMID: , 10.1161/01.CIR.0000128587.30041.C8.
- 78.U.S. EPA (U.S. Environmental Protection Agency). 2020. Integrated Science Assessment for Particulate Matter. Washington, DC: U.S. Environmental Protection Agency.
- 79.Rose G. 1982. Incubation period of coronary heart disease. Br Med J (Clin Res Ed) 284(6329):1600–1601, PMID: , 10.1136/bmj.284.6329.1600.
- 80.López-Melgar B, Fernández-Friera L, Oliva B, García-Ruiz JM, Sánchez-Cabo F, Bueno H, et al. 2020. Short-term progression of multiterritorial subclinical atherosclerosis. J Am Coll Cardiol 75(14):1617–1627, PMID: , 10.1016/j.jacc.2020.02.026.
- 81.Moro AM, Charão M, Brucker N, Bulcão R, Freitas F, Guerreiro G, et al. 2010. Effects of low-level exposure to xenobiotics present in paints on oxidative stress in workers. Sci Total Environ 408(20):4461–4467, PMID: , 10.1016/j.scitotenv.2010.06.058.
- 82.Uzma N, Kumar BS, Hazari MA. 2010. Exposure to benzene induces oxidative stress, alters the immune response and expression of p53 in gasoline filling workers. Am J Ind Med 53(12):1264–1270, PMID: , 10.1002/ajim.20901.
- 83.Chang FK, Mao IF, Chen ML, Cheng SF. 2011. Urinary 8-hydroxydeoxyguanosine as a biomarker of oxidative DNA damage in workers exposed to ethylbenzene. Ann Occup Hyg 55(5):519–525, PMID: , 10.1093/annhyg/mer010.
- 84.Chang T-Y, Wang V-S, Hwang B-F, Yen H-Y, Lai J-S, Liu C-S, et al. 2009. Effects of co-exposure to noise and mixture of organic solvents on blood pressure. J Occup Health 51(4):332–339, PMID: , 10.1539/joh.l8121.
- 85.Kotseva K, Popov T. 1998. Study of the cardiovascular effects of occupational exposure to organic solvents. Int Arch Occup Environ Health 71 Suppl:S87–S91, PMID: .
- 86.Attarchi M, Golabadi M, Labbafinejad Y, Mohammadi S. 2013. Combined effects of exposure to occupational noise and mixed organic solvents on blood pressure in car manufacturing company workers. Am J Ind Med 56(2):243–251, PMID: , 10.1002/ajim.22086.
- 87.Kwok RK, Jackson WB, Curry MD, Stewart PA, McGrath JA, Stenzel M, et al. 2022. Association of Deepwater Horizon oil spill response and cleanup work with risk of developing hypertension. JAMA Netw Open 5(2):e220108, PMID: , 10.1001/jamanetworkopen.2022.0108.
- 88.Denic-Roberts H, Rowley N, Haigney MC, Christenbury K, Barrett J, Thomas DL, et al. 2022. Acute and longer-term cardiovascular conditions in the Deepwater Horizon Oil Spill Coast Guard Cohort. Environ Int 158:106937, PMID: , 10.1016/j.envint.2021.106937.
- 89.Barr ELM, Tonkin AM, Welborn TA, Shaw JE. 2009. Validity of self-reported cardiovascular disease events in comparison to medical record adjudication and a statewide hospital morbidity database: the AusDiab study. Intern Med J 39(1):49–53, PMID: , 10.1111/j.1445-5994.2008.01864.x.
- 90.Eliassen B-M, Melhus M, Tell GS, Borch KB, Braaten T, Broderstad AR, et al. 2016. Validity of self-reported myocardial infarction and stroke in regions with Sami and Norwegian populations: the SAMINOR 1 Survey and the CVDNOR project. BMJ Open 6(11):e012717, PMID: , 10.1136/bmjopen-2016-012717.
- 91.Fourrier-Réglat A, Cuong HM, Lassalle R, Depont F, Robinson P, Droz-Perroteau C, et al. 2010. Concordance between prescriber- and patient-reported previous medical history and NSAID indication in the CADEUS cohort. Pharmacoepidemiol Drug Saf 19(5):474–481, PMID: , 10.1002/pds.1951.
- 92.Machón M, Arriola L, Larrañaga N, Amiano P, Moreno-Iribas C, Agudo A, et al. 2013. Validity of self-reported prevalent cases of stroke and acute myocardial infarction in the Spanish cohort of the EPIC study. J Epidemiol Community Health 67(1):71–75, PMID: , 10.1136/jech-2011-200104.
- 93.Okura Y, Urban LH, Mahoney DW, Jacobsen SJ, Rodeheffer RJ. 2004. Agreement between self-report questionnaires and medical record data was substantial for diabetes, hypertension, myocardial infarction and stroke but not for heart failure. J Clin Epidemiol 57(10):1096–1103, PMID: , 10.1016/j.jclinepi.2004.04.005.
- 94.Yamagishi K, Ikeda A, Iso H, Inoue M, Tsugane S, Group JS, et al. 2009. Self-reported stroke and myocardial infarction had adequate sensitivity in a population-based prospective study JPHC (Japan Public Health Center)-based Prospective Study. J Clin Epidemiol 62(6):667–673, PMID: , 10.1016/j.jclinepi.2008.07.016.
- 95.Coady SA, Sorlie PD, Cooper LS, Folsom AR, Rosamond WD, Conwill DE. 2001. Validation of death certificate diagnosis for coronary heart disease: the Atherosclerosis Risk in Communities (ARIC) Study. J Clin Epidemiol 54(1):40–50, PMID: , 10.1016/s0895-4356(00)00272-9.
- 96.Folsom AR, Gomez-Marin O, Gillum RF, Kottke TE, Lohman W, Jacobs DR Jr.. 1987. Out-of-hospital coronary death in an urban population–validation of death certificate diagnosis. The Minnesota Heart Survey. Am J Epidemiol 125(6):1012–1018, PMID: , 10.1093/oxfordjournals.aje.a114617.
- 97.Goraya TY, Jacobsen SJ, Belau PG, Weston SA, Kottke TE, Roger VL. 2000. Validation of death certificate diagnosis of out-of-hospital coronary heart disease deaths in Olmsted County, Minnesota. Mayo Clin Proc 75(7):681–687, PMID: , 10.4065/75.7.681.
- 98.Lloyd-Jones DM, Martin DO, Larson MG, Levy D. 1998. Accuracy of death certificates for coding coronary heart disease as the cause of death. Ann Intern Med 129(12):1020–1026, PMID: , 10.7326/0003-4819-129-12-199812150-00005.
- 99.Olubowale OT, Safford MM, Brown TM, Durant RW, Howard VJ, Gamboa C, et al. 2017. Comparison of expert adjudicated coronary heart disease and cardiovascular disease mortality with the national death index: results from the REasons for Geographic And Racial Differences in Stroke (REGARDS) study. J Am Heart Assoc 6(5):e004966, PMID: , 10.1161/JAHA.116.004966.
- 100.Huynh T, Quick H, Ramachandran G, Banerjee S, Stenzel M, Sandler DP, et al. 2016. A comparison of the β-substitution method and a Bayesian method for analyzing left-censored data. Ann Occup Hyg 60(1):56–73, PMID: , 10.1093/annhyg/mev049.