The NHANES is a continuous, cross-sectional survey conducted by the National Center for Health Statistics (NCHS). A complex, multistage probability sampling design is used to generate a representative sample of the civilian, non-institutionalized U.S. population.^10,11 Participants receive a detailed in-home interview followed by a physical examination at a mobile examination center (MEC). Data are collected continuously, but released in 2-year cycles. Data from three 2-year cycles were included in the analysis: 2007–2008, 2009–2010, and 2011–2012. The examination response rate for each cycle was 75.4%, 77.3%, and 69.5% respectively.¹² These cycles contain the most current data on each respondent’s longest held job and spirometry results.

In the 2007–2008 to 2011–2012 NHANES, 17392 persons aged 18–79 years provided interview data and attended the examinations in the MEC. Of the 17 392 that were eligible for spirometry (eligibility criteria for spirometry are available at http://www.cdc.gov/nchs/data/nhanes/nhanes_07_08/manual_mecinterview.pdf), 2511 were excluded for safety reasons or did not perform spirometry, 691 had poor quality spirometry data with grades D or F, and 533 were missing. Of the 13 657 with a complete spirometry exam and valid spirometry tests (quality grades A, B, or C), we excluded two participants without height measurements and 611 who have never been employed. Spirometry exclusion and quality details are discussed elsewhere.^13,14 The sample for the analysis of airflow obstruction included 13 044 ever-employed survey respondents aged 18–79 years.

Of 16 486 participants aged 20–79 years that were interviewed and attended the examination in the MEC, we excluded 709 who have never been employed resulting in 15 777 ever-employed respondents in the sample for the analysis of COPD. Respiratory health questions to determine COPD outcome were not asked of participants aged 18–19 years.

Study protocols were approved by NCHS’ Research Ethics Review Board and informed consent was obtained from all participants.

Body mass index (BMI) (kg/m²) was generated from height and weight measurements obtained during the physical examination. Based on BMI, respondents’ weight status was categorized as underweight (<18.5 kg/m²), normal weight (18.5–24.9 kg/m²), overweight (25.0–29.9 kg/m²), or obese (≥30 kg/m²).¹⁵

History of cigarette use was used to determine both smoking status (respondents aged 18–79 years) and cigarette pack-years (respondents aged 20–79 years). The definition of smoking status varied by age because respondents aged 18–19 years answered different smoking status questions than respondents aged 20–79 years. For respondents aged 18–19 years, never smokers were those who reported no use of tobacco or nicotine products in the past 5 days or use of tobacco or nicotine products in the past 5 days excluding cigarettes and ever smokers were those who reported using cigarettes in the past 5 days. For respondents aged 20–79 years, never smokers were those who reported smoking fewer than 100 cigarettes during their entire lifetime and ever smokers were those who reported smoking at least 100 cigarettes during their lifetime. For ever smokers, cigarette pack-years were calculated using the reported number of years smoked and the reported number of cigarettes smoked per day or the number of cigarettes they smoked per day when they last smoked cigarettes.

Respiratory health outcomes were identified using responses to questions asked in the household interview. Survey participants (aged 20–79 years) were considered to have self-reported chronic bronchitis or self-reported emphysema if they reported that a doctor or other health professional ever told them that they had chronic bronchitis and they still had chronic bronchitis or a doctor or other health professional ever told them that they had emphysema. For brevity, we refer to self-reported chronic bronchitis or self-reported emphysema as “COPD.”

Spirometry was performed following ATS/ERS recommendations using Ohio 822/827 dry-rolling seal volume spirometers.^16,17 Airflow obstruction was defined per ATS/ERS spirometry criterion (using pre-bronchodilator values) as FEV₁/FVC < lower limit of normal (LLN) (ie, the lower 5th percentile).² Those with airflow obstruction were further evaluated to identify moderate and more severe airflow obstruction when FEV₁/FVC < LLN and FEV₁ < 70% of the predicted value.² Normative reference equations developed from NHANES III data determined the predicted and LLN pulmonary function values.¹⁸ To estimate reference values for the “other” race/Hispanic origin group, we applied a correction factor of 0.88 to the non-Hispanic white reference values which has been previously published.^14,19 For the “other Hispanic” group, we applied the reference values for Mexican Americans.¹⁴

We examined the longest held job of ever-employed participants (both currently working and those not working). Respondents aged 16 years and older were asked, “Which of the following were you doing last week?” Respondents who reported that they were, “working at a job or business” or “with a job or business but not at work” were asked to provide the name of their employer, the industry/business of their employer, what their job title/occupation was, and their main work activities at their job. Those respondents, as well as respondents who reported that they were, “looking for work” or “not working at a job or business” were then asked, “Thinking of all the paid jobs or businesses you ever had, what kind of work were you doing the longest?”

Respondents who were currently employed could state that their current job was also their longest held job or they could state that another job was their longest held job. Respondents who were not employed could state what their longest held job was from all the jobs they had previously held. Respondents then reported information about their longest held job (if different from their current job): the industry/business of their employer, their job title/occupation, and their main work activities at their job. Data collection procedures were the same across all three cycles. Workers who were not currently in the U.S. Armed Forces, but who stated that their longest held job was in the “Armed Forces,” were included in our analysis.

Using these data, the Centers for Disease Control and Prevention’s National Institute for Occupation Safety and Health (NIOSH) generated four-digit industry and occupation codes for each respondent’s current and/or longest held job using the U.S. Census Bureau’s 2002 version of its Occupation and Industry coding system.²⁰ The coding procedures remained the same across all three cycles. Four-digit industry and occupational coded NHANES data are not publicly available and were accessed via the NCHS’ Research Data Center.²¹

For this analysis, we combined 264 detailed, four-digit industry codes into 44 industry groups and 501 four-digit occupations into 57 occupation groups. An epidemiologist (LK) and Certified Industrial Hygienist (BD) first combined industry codes and occupation codes (a priori and blinded to the subject’s airflow obstruction and COPD status) having a similar potential for respiratory hazards associated with COPD based on consensus and in consultation with the literature.²² After applying theses initial industry and occupation groups to the data, industry and occupation groups were further combined with additional input from sociologist (SF) to have enough observations in each group for statistical analysis (Supplement Tables S1 and S2).

Statistical analyses were performed using SAS® 9.4 (SAS Institute Inc., Cary, NC) complex survey procedures to adjust for differential probabilities of selection and the complex sampling design. Age-specific and age-standardized prevalence of airflow obstruction and COPD with corresponding 95% confidence intervals (CIs) were calculated for the ever-employed civilian, non-institutionalized U.S. population using PROC SURVEYREG for demographic and health characteristics. We used the standard age distribution of the 2000 U.S. Census Population (age groups 18–39, 40–59, and 60–79 or 20–39, 40–59, and 60–79) for standardization.²³ NHANES examination sampling weights were used to obtain prevalence estimates representative of the civilian, non-institutionalized U.S. population for airflow obstruction outcomes. Interview sampling weights were used to obtain prevalence estimates for COPD except when COPD prevalence was analyzed by BMI status and examination sampling weights were used. Variance estimates were computed using the Taylor series linearization approximation method. We calculated relative standard errors (RSE) ([the standard error of the estimate/estimate]*100), and identified estimates with an RSE > 30% which are potentially unreliable and should be interpreted with caution. Estimates with a RSE > 36% are not presented.

We used multivariable logistic regression models to calculate prevalence odds ratios (PORs) and 95% CIs for airflow obstruction by industry and occupation. PORs were adjusted for age, gender, race/Hispanic origin, and smoking status (never/ever). The reference industry group was all other ever-employed U.S. adults who were not in the industry group of interest. The reference occupation group was all other ever-employed U.S. adults who were not in the occupation group of interest. We did not present prevalence estimates and PORs for moderate and more severe airflow obstruction and COPD by industry and occupation because many of these estimates had potentially unreliable RSEs.

Self-reported doctor-diagnosed chronic bronchitis and self-reported emphysema were used to define COPD and clinical data were not available to validate COPD. The NHANES is a cross-sectional survey, so it is not possible to determine causality of airflow obstruction and COPD. Similarly, we cannot determine the occupational or non-occupational causes of airflow obstruction since occupational exposure was not included in this analysis. However, analysis of NHANES 1988–1994 data indicate tobacco smoking, occupational exposure, and socioeconomic status contribute to the increased risk of spirometry-defined obstructive lung disease among some workers.^8,27 Because airflow obstruction generally results from long-term exposure, we used longest held industry and occupation data to include the job which potentially contributed to disease. However, relevant occupational exposures associated with the development of airflow obstruction may have occurred in an occupation other than the longest held. We did not determine the type of airflow obstruction (eg, asthma, chronic bronchitis, emphysema) among those with spirometry-defined airflow obstruction and some occupational exposures are attributed to different types of obstructive disease. Another limitation is that those with COPD may or may not have spirometry-defined airflow obstruction.

Despite combining data from three NHANES survey cycles, confidence intervals for some prevalence estimates and PORs provided a wide range of possible values for the parameter and should be examined when interpreting data. Due to small cell sizes, we were not able to report gender-specific or smoking-specific prevalence estimates and we had to group together some industries and occupations that we had interest in analyzing at the detailed, four-digit level. For example, we were interested in the prevalence of airflow obstruction among workers in specific service industries, such as barber shops, beauty salons, and nail salons, as these workers may be exposed to products and chemicals that can cause lung irritation and respiratory diseases. However, due to small cell sizes, these industries were grouped with other service industries in the barber shops, beauty and nail salons, dry-cleaning and laundry services, funeral homes, other personal services, and labor unions industry group so we could not calculate the airflow obstruction prevalence in these specific service industries.

The strengths of this study were that its data come from a recent, nationally representative data set. NHANES ensures rigorous quality standards are followed for data collection, which is especially important for tests like spirometry, where data quality can impact interpretation of disease classifications. In addition, we were able to examine 44 industry groups and 57 occupation groups, which had not been done in prior research on this topic using NHANES data.

This study provides the distribution of ever workers with airflow obstruction by detailed industry and occupation group. This study observed elevated PORs for specific industries including mining, manufacturing motor vehicles and motor vehicle equipment, construction, and services to buildings, landscaping and waste management services, industries that had previously been shown to be at risk for airflow obstruction. However, we identified workers in the occupation group including installation, maintenance, and repair (power-line, telecommunications line, vending machine, locksmiths, manufactured building, signal and track switch, commercial divers, riggers) to have an elevated prevalence of airflow obstruction. To our knowledge, this occupation group has not been previously studied or identified as having increased risk for airflow obstruction. Additional research is needed to understand the association between these groups and an increased risk for airflow obstruction and also what exposures may be causing the increased risk.