Study population

The Hispanic Community Health Study/Study of Latinos (HCHS/SOL) is a community-based prospective cohort study designed to evaluate risk factors and chronic diseases among the rapidly growing Hispanic/Latino population across four cities in the USA (Bronx, New York; Chicago, Illinois; Miami, Florida; and San Diego, California). HCHS/SOL included 16 415 men and women, aged 18–74 years, who self-identified as Hispanic/Latino at baseline (2008–2011). The sampling design and cohort selection have been previously described.12 13 Briefly, participants were recruited through a multistage probability sampling design. Sampling weights were established to represent the probability of selection at each stage. Baseline examinations included occupational information, sociodemographics, medical history, medication use, tobacco use, alcohol use, physical activity, dietary recall, anthropometry, blood pressure, phlebotomy and electrocardiography.

Exposure assessment

An occupational questionnaire designed to assess work environments and exposures was administered to all employed participants by a trained interviewer in either English or Spanish, based on the participant’s preference. Participants were asked ‘At the job you currently work the majority of your work hours per week, how often are you exposed to any type of organic solvents, for example styrene, trichloroethylene, toluene, or xylene?’ and ‘At the job you currently work the majority of your work hours per week, how often are you exposed to metals such as manganese, lead, or mercury?’. Response options were none of the time, 25% of the time, 50% of the time, 75% of the time, 100% of the time, occasionally and do not know. For this analysis, responses were collapsed into yes (occasional or 25%–100% of the time) or no (0%) because of data sparseness. Participants were also asked ‘In your current job(s), are you exposed to pesticides?’, with responses recorded as yes or no. Participants unfamiliar with the term ‘pesticides’ were classified as not exposed.

Outcome assessment

CVD prevalence was assessed using self-reported information from the medical history questionnaire or from a digital 12-lead ECG performed at baseline. Details regarding HCHS/SOL ECG procedures and subsequent readings performed at a centralised location (EPICARE, Wake Forest University School of Medicine, Winston-Salem, North Carolina) have been described elsewhere.14 Prevalent CVD was defined as at least one of the following: coronary heart disease, heart failure, atrial fibrillation or cerebrovascular disease. Participants who self-reported that a doctor told them they had angina, a myocardial infarction or reported undergoing coronary bypass surgery, balloon angioplasty or stent placement in coronary arteries, or had ECG evidence of major Q wave abnormalities or minor Q, QS waves with ST and T abnormalities were considered to have coronary heart disease. Atrial fibrillation was defined as either a self-reported physician diagnosis or ECG evidence of atrial fibrillation or flutter. Heart failure was based solely on self-report (a doctor told the participant he or she had heart failure). Cerebrovascular disease was defined as self-reported physician diagnosis of stroke or transient ischaemic attack.

Sociodemographic characteristics

The interviewer-administered enrolment questionnaire asked participants for sociodemographic information. Hispanic/Latino background groups were categorised as Dominican, Central American, Cuban, Mexican, Puerto Rican, South American and more than one heritage/other heritage. Educational attainment was categorised as not having a high school diploma or General Equivalency Diploma (GED), having a diploma or GED or attaining an education beyond a high school equivalent (ie, college or vocational). Health insurance was defined as current coverage through an employer, individual plan, Medicaid/Medicare, military, Indian Health Services or another provider.

Lifestyle characteristics

Alcohol intake was classified as none, low/moderate (<7 drinks/week for females; <14 drinks/week for males) or heavy (≥7 drinks/week for females; ≥14 drinks/week for males). Cigarette smoking status was categorised as never, former or current. A modified Global Physical Activity Questionnaire based on the original developed by the WHO was used to collect information on physical activity in a typical week across three domains: work, recreation and transport. Total physical activity across these domains was categorised as low, moderate or high based on established criteria.15 Dietary quality was assessed using the 2010 Alternative Health Eating Index (AHEI-2010), which was derived from up to two 24-hour dietary recalls. The index combines intakes of vegetables, fruit, whole grains, sugar-sweetened beverages and fruit juice, nuts and legumes, red/processed meats, trans fats, long-chain (n-3) fatty acids, polyunsaturated fatty acids, sodium and alcohol intakes, with higher scores representing a better quality diet.16

Acculturation characteristics

Residential duration in the mainland USA and the Short Acculturation Scale for Hispanics (SASH) language subscale were used as measures of acculturation. The SASH language subscale is based on five-point Likert-type scale questions; the average score was calculated with higher scores indicating a higher degree of language acculturation.17

Occupational characteristics

Full-time employment was defined as working more than 35 hours per week in one job or more than one job; part-time employment was defined as working 35 or fewer hours per week. Primary occupations were categorised into five groups: non-skilled workers (typically does not require a high school degree and involves manual labour), service workers (provides personal services to individuals), skilled workers (requires specific skills that typically requires technical training or certifications), professional/technical workers (requires a professional degree) and other. Participants were asked whether their current job was the same as their longest held job, and if so, how many years they worked at that job.

CVD risk factors

Participants were instructed to bring all medications (prescription and non-prescription) taken in the past month with them to the examination where they were scanned and documented. Medications were inventoried and classified using a Master Drug Data Base (Medispan MDDB). Three blood pressure measurements were performed using an automated sphygmomanometer (Omron model HEM-907 XL, Omron Healthcare, Bannockburn, Illinois, USA), with the average of the second and third readings calculated. Hypertension was defined as a systolic or diastolic blood pressure greater than or equal to 140/90 mm Hg, or the use of antihypertensive medications. Body weight and height were measured by trained research technicians. Body mass index (BMI) was calculated as weight in kilograms divided by height in metres squared. Obesity was defined as a BMI of 30.0 kg/m² or greater.

Total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, glucose and glycated haemoglobin were measured from fasting blood samples at the HCHS/SOL Central Laboratory. Laboratory assay methodologies are further described in the HCHS/SOL Manual.18 Hypercholesterolaemia was defined as a total cholesterol ≥240 mg/dL, LDL cholesterol ≥160 mg/dL, HDL cholesterol <40 mg/dL or antihyperlipidaemic medication use. Diabetes mellitus was defined as a fasting plasma glucose ≥126 mg/dL, 2-hour postload plasma glucose ≥200 mg/dL, glycated haemoglobin ≥6.5% or antidiabetic medication use.

Statistical analyses

The prevalence of occupational exposures, CVD and CVD subtypes were calculated and compared across categorical variables using the Pearson χ² test. Poisson regression models were used to investigate the relationships between occupational exposures and CVD prevalence. Confounders were identified using a directed acyclic graph based on subject matter knowledge and the relevant epidemiological literature and were sequentially entered into models according to their hypothesised strengths of association with exposure status and CVD (online supplementary figure 1).19 Adjusted model 1 included basic sociodemographic characteristics considered to be most strongly related to both exposure and CVD: age, gender, field centre and Hispanic/Latino background. Adjusted model 2 further adjusted for lifestyle and behavioural factors that were additionally considered to confound exposure–CVD associations, although not as strongly sociodemographic characteristics. These included employment status, current health insurance status, cigarette use, alcohol use level, years of residential duration in the USA, physical activity level and the AHEI-2010 and the SASH language subscale. Adjusted model 3 included all variables in adjusted model 2 and additionally included hypertension, hypercholesterolaemia, BMI and diabetes mellitus as these were hypothesised to be potential intermediates on the causal pathways that could potentially explain exposure–CVD relationships. In all models, restricted cubic splines with four equally spaced knots at the 5th, 35th, 65th and 95th percentiles were used for continuous variables (age, years of residential duration in the USA, AHEI-2010, SASH language subscale and BMI) to allow for non-linearity. P values less than 0.05 were considered statistically significant. All analyses were conducted in Stata V.14.2. The Stata svy prefix was used to incorporate the HCHS/SOL sampling weights and account for the complex sampling design.

Sensitivity analyses

Since solvent, metal and pesticide exposures in the participants’ current job may not accurately reflect chronic exposures, we performed a sensitivity analysis restricting to participants who indicated their current held job was their longest held. The models were identical to those used in our main analyses, with the addition of years worked as a covariate.