Abstract
The recent rise in U.S. midlife mortality has been conceptualized as a “working-class” crisis, defined by increasing mortality among blue-collar and/or unemployed workers and the decline of manual labor; yet research on the topic overwhelmingly focuses on educational attainment as the key socioeconomic determinant of midlife mortality, especially among “despair”-related deaths. The present study addresses this gap by using data on 360,146 adults ages 25-64 from restricted-use National Health Interview Survey-Linked Mortality Files (1997-2015; average follow-up 9.87 years) to estimate associations between individuals’ occupation and employment status and alcoholic liver disease, suicide, or accidental poisoning mortality risk, net of confounders. Adults in service, manual labor, and transport occupations exhibited two-to-three times the risk of mortality from accidental poisonings compared to those in managerial/administrative positions. Notably, health professionals exhibited the highest accidental poisoning mortality risks. Relative to managerial/administrative professionals, adults not in the labor force had double the suicide risk and nearly seven times the accidental poisoning risk, net of confounders. Unemployed adults and those having never worked also had elevated risks from accidental poisoning mortality. Critically, the fact that individuals’ occupations and employment status are independently associated with midlife mortality due to deaths of despair – especially accidental poisoning – highlights the need for measures of socioeconomic status beyond educational attainment and income in understanding rising midlife mortality. Moreover, policies addressing working-aged mortality must target particular workplace contexts and the consequences of unemployment, both of which affect a large and growing segment of the working-aged U.S. population.
Keywords: Work/Occupations/Employment, Socioeconomic Status, Mortality, Deaths of Despair
Introduction
Recent research documents large and widening socioeconomic disparities in working-aged (i.e., 25-64) mortality among U.S. adults, largely driven by high and rising mortality among adults with low education or income.1-4 Alcohol-related disease, suicide, and drug poisonings – i.e., “deaths of despair” – account for an increasing proportion of mortality in this age range,1,5,6 with the latter almost entirely responsible for the reduction and stagnation in U.S. life expectancy over the past four years.7
However, the focus on education and income as the only measures of socioeconomic status (SES) important to contemporary U.S. working-age mortality overlooks the importance of employment status and occupations for understanding who is at greatest risk. While most of the aforementioned research conflates low-educated or low-income with the “working-class,” measures of ‘work’ are largely absent. Work is a key dimension of SES, determining individuals’ purpose and role in society8 and position in the social hierarchy.9 Individuals’ jobs form the basis for shared social experiences and identities10; they may also lead to shared exposures and risks of poor health, independent of other aspects of SES.
The current narrative of working-class mortality has prioritized low education as a marker of risk, alluding to the fact that low-educated, blue-collar and unemployed workers in parts of the country where manual labor and agriculture sustain(ed) the economy are most vulnerable.1 However, more education does not guarantee a ‘good’ job or good health in the U.S.,11 where declines in workplace conditions across many occupational sectors may have repercussions for mortality risk among working-age adults. Secure, steady, and well-paying jobs are increasingly scarce; many American workers now find themselves in “precarious” jobs characterized by limited financial security, perpetual employment instability, and poor workplace conditions.1,11-13 These “bad jobs” are increasingly concentrated in growing service sectors of the economy such as retail, food services, and cleaning, as well as in trucking and transportation, many of which were once considered solidly “middle-class” jobs.11
Cumulatively, uncertainty and dissatisfaction across these occupations has effectively “trapped” many workers in “dead-end” jobs, lacking a clear pathway for upward mobility.12 In turn, these sustained, negative psychosocial environments may be directly implicated in increased risks for substance abuse and self-harm.13,14 Indeed, the ‘demand’-based explanation for recent despair-associated mortality – namely, that adults turn to harmful behaviors to alleviate poor psychosocial health1 – suggests that occupations may play a key role in shaping ‘demand,’ as bad jobs lead to greater despair.13 However, occupations may also substantiate ‘supply’-based explanations. Indeed, increased access and proximity to prescription opioids, both generally and in specific occupations, may account for the dramatic surge in U.S. accidental poisoning mortality.5 For example, healthcare workers’ “access to lethal means,” such as prescription drugs, can increase risk for suicide and poisoning mortality.15-17
Thus, both having work and the kind of work one has are integral components of individuals’ health, perhaps especially for despair-related morbidity and mortality. As thousands of U.S. adults experience these avoidable deaths, there is clear need for identifying working-aged Americans at the highest risk for premature mortality due to alcoholic liver disease, suicide, and accidental poisoning. Using large-scale, nationally-representative survey data, this analysis estimates disparities in cause-specific mortality risk across individuals’ occupational groups and employment status, net of demographic and socioeconomic characteristics. Evidence of unique mortality risks associated with certain occupations is used to discuss the significance of work and work-related policy for health and mortality in the United States.
Methods
Study Data
Data come from the restricted-use National Health Interview Survey – Linked Mortality File (NHIS-LMF), which is one of a few large, nationally-representative surveys collecting data on individuals’ occupations and employment status linked with mortality records. NHIS collects detailed information, such as occupation, from one sample adult per household; thus, we focus on working-age sample adults, 25-64, to capture adults in the workforce. Analyses are limited to years 1997-2014, linked with mortality data through the end of 2015, to maintain comparability with the timeframe used in recent research,1,2,5 and because NHIS data prior to 1997 do not include more detailed employment status data. Detailed cause of death information for these years is only available through a data agreement with the National Center for Health Statistics; thus, all analyses are conducted at a Federal Restricted Data Center, with a final sample of 360,146 adults.
Main Outcomes and Measures
Detailed cause of death data are based on International Classification of Disease codes, classified into 113 leading causes. The analyses focus on deaths attributable to alcoholic liver disease, suicide, and accidental poisoning, among individuals ages 25-64 at time of survey who are followed until: (1) their death, (2) their 65th birthday, or (3) the end of 2015. 578 deaths were from alcoholic liver disease, 541 deaths from suicide, and 481 deaths from accidental poisoning, which encompasses all poisoning deaths due to drugs, medicaments, and biologicals.
Occupation is based on verbatim responses for “usual” occupation converted to an occupational code; IPUMS, which provides the harmonized NHIS data used in this analysis, has standardized codes to 1995 Standard Occupational Classifications.18 Respondents were asked to describe their employment status as currently working full-time, having worked in the past year but not in the past week, having not worked in the past year or week, or not being in the labor force (which may include individuals out of work for >1 year). Among the 46 occupation and employment status categories available, Table S1 provides a crosswalk to the 25 categories used in the analyses. The goal was to remain as detailed as possible given data constraints, while having at least ~1% of the sample in each group.
Individuals’ birth cohort, gender, race/ethnicity, foreign-born status, residence in an urban area (50,000+ people, per Census guidelines), Census region, educational attainment, income-to-needs ratio (based on federal poverty guidelines), marital status, family type, home ownership, and health insurance are included as covariates. NHIS is a cross-sectional survey and there is potential for reverse causality in the employment/occupation-mortality relationship among people who are not working/in worse jobs due to preexisting poor health. Thus, in sensitivity analyses we include self-rated health, obesity, alcohol use, smoking, and activity/functional limitations as covariates. The results of these models provide evidence regarding whether our general pattern of results is upheld using a very inclusive set of covariates, some of which may actually be mediators of the occupation/employment-mortality association.
Statistical Analyses
Discrete time-to-event Poisson regression models are used to obtain estimates of relative mortality risk (i.e., odds ratios) for different occupations, with managerial/administrative workers as the referent group. We constructed a person-year file, wherein each individual has a record for each full or fraction of a year contributed at a specific age between 25 and 64.19 As a sensitivity check, nearly identical results were obtained with competing risks models and Cox proportional hazards, though the proportional hazards assumption was not upheld for some covariates.
All analyses are conducted using Stata 15. Multiple imputation (with 10 imputed datasets) is used to account for missing data among the covariates, with the exception of occupation/employment, where the 2% missing were dropped. Survey weights account for complex survey design and ineligibility for mortality follow-up.
Results
Table 1 provides a descriptive overview of the measures. The largest single occupational/employment category is “Not in the labor force” (15.3%), followed by “Managerial/administrative” (11.8%). 12% of adults are in service-related occupations, and 15% have jobs in manual occupations like farming, mechanics, construction, and production. About 4% of adults are unemployed. Just under half of respondents are male, the majority are non-Hispanic White (68.2%), and approximately 18% are foreign-born. Over three-quarters of respondents reside in urban areas, with most adults located in the South (36.5%). The mean age is 40.4 and 30% of respondents have a college degree or higher. Approximately one-in-ten adults report fair/poor health or any activity limitation.
Table 1:
Descriptive Statistics, National Health Interview Series-Linked Mortality Files (1997-2015; Ages 25-64)
| Occupation/Employment | Geographic factors | ||
|---|---|---|---|
| Managerial/administrative | 11.8% | Urban | 76.7% |
| Scientists/engineers | 4.0% | Census regions | |
| Health professionals | 3.2% | South | 36.5% |
| Other professional specialty | 8.4% | Northeast | 17.9% |
| Technologists | 2.7% | North Central/Midwest | 24.1% |
| Sales professionals | 3.8% | West | 21.5% |
| Other sales | 3.6% | ||
| Skilled administrative | 3.1% | Socioeconomic status | |
| Other administrative | 5.4% | Education | |
| Misc. skilled service | 1.6% | Less than HS | 13.0% |
| Protective service | 1.7% | HS or equal | 26.8% |
| Food service | 3.0% | Some college | 29.8% |
| Cleaning and building service | 2.2% | BA or higher | 30.4% |
| Health service | 1.6% | Income-to-needs ratio | |
| Personal service | 2.0% | 0-1.00 | 11.5% |
| Farming, fishing, and forestry | 1.4% | 1.01-1.99 | 16.3% |
| Mechanics and repairers | 3.1% | 2.00-3.99 | 30.7% |
| Construction and extractive trades | 4.5% | 4.00-4.99 | 13.4% |
| Production/manufacturing | 5.8% | 5.00+ | 28.0% |
| Transport | 4.1% | Married | 62.6% |
| Other manual | 0.9% | Family type | |
| Recent unemployed | 2.6% | 1 adult, no children | 15.0% |
| Long-term unemployed | 1.6% | 1+ adults, no children | 33.5% |
| Not in labor force | 15.3% | 1 adult, 1+ children | 5.1% |
| Never employed | 2.8% | 1+ adults, 1+ children | 46.5% |
| Home ownership | 66.7% | ||
| Demographic factors | Health insurance | 81.6% | |
| Age (μ) | 40.4 | ||
| Cohort | Health status and behaviors | ||
| 1932-1939 | 0.1% | Fair/poor health | 10.0% |
| 1940-1949 | 1.1% | Obese | 27.6% |
| 1950-1959 | 30.0% | Alcohol usage | |
| 1960-1969 | 34.1% | Current | 69.8% |
| 1970-1979 | 25.4% | Former | 12.7% |
| 1980-1989 | 9.3% | Never | 17.6% |
| Male | 49.3% | Smoking status | |
| Race/ethnicity | Current | 24.1% | |
| NH White | 68.2% | Former | 18.0% |
| NH Black | 12.4% | Never | 57.9% |
| Hispanic | 14.6% | Any activity limitation | 11.0% |
| Other | 4.7% | ADL limitation | 1.1% |
| Foreign-born | 17.9% | IADL limitation | 2.3% |
| Sample size | 360,146 | ||
Notes:
Analyses based on restricted-use data with non-perturbed, detailed information on cause of death. Deaths restricted to ages 25-64.
Estimates based on multiple imputation to account for missing data.
The analyses provide odds ratios from two models for each of the three causes of death (Tables 2-4). The first model includes demographic variables, while the second includes additional measures of socioeconomic status. The strongest associations between occupation/employment and alcoholic liver disease mortality risk (Table 2) are among adults not working. Relative to those in managerial/administrative occupations, adults who are recently unemployed (OR 3.20 [1.48—6.92]), long-term unemployed (OR 2.79 [1.35—5.74]), not in the labor force (OR 3.88 [2.52—5.98]), or never employed (OR 2.69 [1.29—5.62]) are at elevated mortality risk, net of demographic characteristics. Adults in food service (OR 2.56 [1.35—4.93]), cleaning and building service (OR 2.00 [1.02—3.91]), and farming, fishing and forestry (OR 2.21 [1.18—4.15]) have similarly elevated risks. Including socioeconomic measures (Model 2) fully attenuates these occupational disparities; however, we continue to observe elevated risks of alcoholic liver disease mortality for those not in the labor force (OR 1.85 [1.16—2.95]).
Table 2:
Alcoholic Liver Disease Mortality Risk, National Health Interview Series-Linked Mortality Files (1997-2015; Ages 25-64)
| Model 1 |
Model 2 |
|||||
|---|---|---|---|---|---|---|
| Odds Ratio | 95% CI |
Odds Ratio |
95% CI | |||
| Occupation/Employment Status | ||||||
| Managerial/administrative (ref.) Scientists/engineers | 0.70 | 0.29 | 1.66 | 0.83 | 0.34 | 1.98 |
| Health professionals | 0.46 | 0.13 | 1.69 | 0.54 | 0.14 | 1.99 |
| Other professional specialty | 0.72 | 0.37 | 1.40 | 0.72 | 0.36 | 1.42 |
| Technologists | 0.73 | 0.24 | 2.21 | 0.63 | 0.21 | 1.90 |
| Sales professionals | 1.05 | 0.52 | 2.12 | 0.92 | 0.46 | 1.87 |
| Other sales | 1.55 | 0.75 | 3.16 | 1.02 | 0.50 | 2.10 |
| Skilled administrative | 1.31 | 0.54 | 3.22 | 0.97 | 0.40 | 2.38 |
| Other administrative | 0.71 | 0.34 | 1.49 | 0.50 | 0.24 | 1.07 |
| Misc. skilled service | 1.10 | 0.38 | 3.18 | 0.72 | 0.25 | 2.11 |
| Protective service | 0.76 | 0.30 | 1.90 | 0.64 | 0.26 | 1.60 |
| Food service | 2.56 | 1.34 | 4.89 | 1.16 | 0.59 | 2.29 |
| Cleaning and building service | 2.00 | 1.02 | 3.92 | 1.03 | 0.52 | 2.06 |
| Health service | 2.01 | 0.78 | 5.14 | 1.02 | 0.40 | 2.63 |
| Personal service | 0.90 | 0.31 | 2.59 | 0.49 | 0.17 | 1.44 |
| Farming, fishing, and forestry | 2.21 | 1.18 | 4.14 | 1.11 | 0.58 | 2.14 |
| Mechanics and repairers | 1.16 | 0.60 | 2.24 | 0.82 | 0.42 | 1.59 |
| Construction and extractive trades | 1.69 | 0.99 | 2.86 | 0.98 | 0.57 | 1.70 |
| Production/manufacturing | 1.53 | 0.91 | 2.59 | 0.97 | 0.57 | 1.68 |
| Transport | 1.12 | 0.59 | 2.13 | 0.69 | 0.36 | 1.33 |
| Other manual | 1.47 | 0.69 | 3.14 | 0.74 | 0.34 | 1.62 |
| Recent unemployed | 3.20 | 1.48 | 6.92 | 1.46 | 0.64 | 3.32 |
| Long-term unemployed | 2.79 | 1.35 | 5.74 | 1.10 | 0.52 | 2.32 |
| Not in labor force | 3.88 | 2.52 | 5.98 | 1.85 | 1.16 | 2.95 |
| Never employed | 2.69 | 1.29 | 5.62 | 1.08 | 0.51 | 2.30 |
| Demographic factors | ||||||
| Age | 1.08 | 1.06 | 1.10 | 1.08 | 1.06 | 1.11 |
| Cohort | ||||||
| 1932-1939 | 0.37 | 0.16 | 0.85 | 0.40 | 0.17 | 0.94 |
| 1940-1949 | 0.51 | 0.33 | 0.80 | 0.56 | 0.36 | 0.86 |
| 1950-1959 | 0.80 | 0.57 | 1.12 | 0.82 | 0.58 | 1.14 |
| 1960-1969 (ref.) | ||||||
| 1970-1979 | 0.71 | 0.45 | 1.10 | 0.65 | 0.41 | 1.02 |
| 1980-1989 | 0.44 | 0.12 | 1.57 | 0.37 | 0.10 | 1.31 |
| Male | 2.92 | 2.29 | 3.72 | 2.78 | 2.19 | 3.53 |
| Race/ethnicity | ||||||
| NH White (ref.) | ||||||
| NH Black | 1.07 | 0.77 | 1.50 | 0.74 | 0.53 | 1.04 |
| Hispanic | 2.74 | 2.12 | 3.53 | 2.09 | 1.59 | 2.75 |
| Other | 0.87 | 0.44 | 1.70 | 0.96 | 0.49 | 1.87 |
| Foreign-born | 0.65 | 0.50 | 0.86 | 0.59 | 0.44 | 0.80 |
| Geographic factors | ||||||
| Urban | 0.99 | 0.50 | 0.86 | 0.94 | 0.73 | 1.21 |
| Census regions | ||||||
| South (ref.) | ||||||
| Northeast | 0.80 | 0.59 | 1.09 | 0.82 | 0.60 | 1.12 |
| North Central/Midwest | 0.89 | 0.67 | 1.18 | 0.94 | 0.71 | 1.25 |
| West | 1.21 | 0.95 | 1.54 | 1.21 | 0.94 | 1.54 |
| Socioeconomic status | ||||||
| Education | ||||||
| Less than HS (ref.) | ||||||
| HS or equal | 0.78 | 0.59 | 1.04 | |||
| Some college | 0.79 | 0.58 | 1.08 | |||
| BA or higher | 0.59 | 0.38 | 0.91 | |||
| Income-to-needs ratio | ||||||
| 0-1.00 (ref.) | ||||||
| 1.01-1.99 | 0.94 | 0.71 | 1.26 | |||
| 2.00-3.99 | 0.63 | 0.45 | 0.88 | |||
| 4.00-4.99 | 0.76 | 0.47 | 1.24 | |||
| 5.00+ | 0.40 | 0.25 | 0.64 | |||
| Married | 0.55 | 0.41 | 0.72 | |||
| Family type | ||||||
| 1 adult, no children (ref.) | ||||||
| 1+ adults, no children | 0.87 | 0.63 | 1.20 | |||
| 1 adult, 1+ children | 1.28 | 0.98 | 1.68 | |||
| 1+ adults, 1+ children | 0.80 | 0.53 | 1.23 | |||
| Home ownership | 0.58 | 0.45 | 0.75 | |||
| Health Insurance | 0.77 | 0.59 | 0.99 | |||
| Sample size | 360,146 | |||||
| Number of deaths | 578 | |||||
Notes:
Analyses based on restricted-use data with non-perturbed, detailed information on cause of death.
Deaths restricted to ages 25-64.
Estimates based on multiple imputation to account for missing data.
Table 4:
Accidental Poisoning Mortality Risk, National Health Interview Series-Linked Mortality Files (1997-2015; Ages 25-64)
| Model 1 |
Model 2 |
|||||
|---|---|---|---|---|---|---|
|
Odds Ratio |
95% CI |
Odds Ratio |
95% CI | |||
| Occupation/Employment Status | ||||||
| Managerial/administrative (ref.) | ||||||
| Scientists/engineers | 1.23 | 0.33 | 4.68 | 1.54 | 0.41 | 5.80 |
| Health professionals | 2.77 | 0.97 | 7.93 | 3.35 | 1.18 | 9.47 |
| Other professional specialty | 1.26 | 0.53 | 3.00 | 1.48 | 0.63 | 3.49 |
| Technologists | 4.00 | 1.80 | 8.89 | 3.27 | 1.47 | 7.29 |
| Sales professionals | 1.98 | 0.87 | 4.49 | 1.70 | 0.75 | 3.84 |
| Other sales | 1.73 | 0.69 | 4.37 | 1.17 | 0.46 | 2.96 |
| Skilled administrative | 2.31 | 0.92 | 5.77 | 1.51 | 0.61 | 3.76 |
| Other administrative | 2.44 | 1.18 | 5.03 | 1.62 | 0.79 | 3.34 |
| Misc. skilled service | 1.90 | 0.43 | 8.42 | 1.19 | 0.27 | 5.34 |
| Protective service | 2.72 | 0.97 | 7.65 | 2.26 | 0.80 | 6.37 |
| Food service | 6.67 | 3.37 | 13.20 | 3.25 | 1.61 | 6.57 |
| Cleaning and building service | 6.08 | 2.76 | 13.41 | 3.10 | 1.38 | 6.94 |
| Health service | 3.95 | 1.44 | 10.79 | 2.09 | 0.76 | 5.75 |
| Personal service | 1.00 | 0.29 | 3.46 | 0.58 | 0.17 | 2.00 |
| Farming, fishing, and forestry | 4.52 | 1.86 | 10.98 | 2.54 | 1.03 | 6.31 |
| Mechanics and repairers | 3.24 | 1.46 | 7.15 | 2.22 | 1.00 | 4.96 |
| Construction and extractive trades | 4.87 | 2.36 | 10.04 | 3.00 | 1.41 | 6.35 |
| Production/manufacturing | 3.14 | 1.62 | 6.06 | 1.87 | 0.95 | 3.70 |
| Transport | 3.95 | 1.83 | 8.52 | 2.38 | 1.08 | 5.25 |
| Other manual | 3.59 | 1.15 | 11.15 | 1.87 | 0.58 | 6.00 |
| Recent unemployed | 5.74 | 2.73 | 12.07 | 3.14 | 1.47 | 6.72 |
| Long-term unemployed | 6.44 | 2.69 | 15.41 | 3.05 | 1.25 | 7.46 |
| Not in labor force | 12.71 | 7.46 | 21.67 | 6.71 | 3.81 | 11.81 |
| Never employed | 11.93 | 5.76 | 24.70 | 4.73 | 2.19 | 10.24 |
| Demographic factors | ||||||
| Age | 1.02 | 1.00 | 1.04 | 1.02 | 1.00 | 1.04 |
| Cohort | ||||||
| 1932-1939 | 0.09 | 0.02 | 0.39 | 0.08 | 0.02 | 0.37 |
| 1940-1949 | 0.19 | 0.10 | 0.36 | 0.18 | 0.10 | 0.34 |
| 1950-1959 | 0.49 | 0.34 | 0.71 | 0.47 | 0.33 | 0.68 |
| 1960-1969 (ref.) | ||||||
| 1970-1979 | 0.89 | 0.65 | 1.22 | 0.86 | 0.63 | 1.19 |
| 1980-1989 | 0.99 | 0.49 | 1.97 | 0.88 | 0.44 | 1.76 |
| Male | 1.86 | 1.42 | 2.45 | 1.52 | 1.16 | 2.00 |
| Race/ethnicity | ||||||
| NH White (ref.) | ||||||
| NH Black | 0.63 | 0.44 | 0.91 | 0.46 | 0.32 | 0.66 |
| Hispanic | 1.09 | 0.77 | 1.55 | 0.87 | 0.59 | 1.28 |
| Other | 0.68 | 0.26 | 1.82 | 0.78 | 0.29 | 2.06 |
| Foreign-born | 0.31 | 0.20 | 0.49 | 0.34 | 0.21 | 0.53 |
| Geographic factors | ||||||
| Urban | 1.32 | 0.98 | 1.79 | 1.34 | 0.99 | 1.81 |
| Census regions | ||||||
| South (ref.) | ||||||
| Northeast | 0.93 | 0.67 | 1.29 | 0.94 | 0.68 | 1.31 |
| North Central/Midwest | 0.84 | 0.61 | 1.17 | 0.86 | 0.62 | 1.20 |
| West | 1.15 | 0.86 | 1.55 | 1.17 | 0.86 | 1.58 |
| Socioeconomic status | ||||||
| Education | ||||||
| Less than HS (ref.) | ||||||
| HS or equal | 0.78 | 0.56 | 1.09 | |||
| Some college | 0.64 | 0.45 | 0.91 | |||
| BA or higher | 0.29 | 0.18 | 0.48 | |||
| Income-to-needs ratio | ||||||
| 0-1.00 (ref.) | ||||||
| 1.01-1.99 | 0.90 | 0.60 | 1.37 | |||
| 2.00-3.99 | 0.77 | 0.50 | 1.17 | |||
| 4.00-4.99 | 0.36 | 0.18 | 0.74 | |||
| 5.00+ | 0.67 | 0.38 | 1.18 | |||
| Married | 0.55 | 0.41 | 0.75 | |||
| Family type | ||||||
| 1 adult, no children (ref.) | ||||||
| 1+ adults, no children | 1.20 | 0.84 | 1.69 | |||
| 1 adult, 1+ children | 1.36 | 0.98 | 1.88 | |||
| 1+ adults, 1+ children | 0.66 | 0.41 | 1.06 | |||
| Home ownership | 0.84 | 0.64 | 1.09 | |||
| Health Insurance | 1.16 | 0.87 | 1.54 | |||
| Sample size | 360,146 | |||||
| Number of deaths | 481 | |||||
Notes:
Analyses based on restricted-use data with non-perturbed, detailed information on cause of death.
Deaths restricted to ages 25-64.
Estimates based on multiple imputation to account for missing data.
Similarly, results for suicide mortality risk in Table 3 show no clear occupational differences in risk, even in Model 1. However, across both models unemployment and not being in the labor force are strongly associated with increased suicide mortality risk. There is almost four times greater suicide risk associated with long-term unemployment (3.91 [2.04—7.48]), and two times greater risk associated with not being in the labor force (2.28 [1.49—3.49]), with some attenuation in these estimates in Model 2.
Table 3:
Suicide Mortality Risk, National Health Interview Series-Linked Mortality Files (1997-2015; Ages 25-64)
| Model 1 |
Model 2 |
|||||
|---|---|---|---|---|---|---|
| Odds Ratio | 95% CI |
Odds Ratio |
95% CI | |||
| Occupation/Employment Status | ||||||
| Managerial/administrative (ref.) | ||||||
| Scientists/engineers | 0.62 | 0.32 | 1.20 | 0.63 | 0.33 | 1.22 |
| Health professionals | 0.82 | 0.39 | 1.71 | 0.83 | 0.39 | 1.74 |
| Other professional specialty | 0.65 | 0.37 | 1.15 | 0.63 | 0.35 | 1.13 |
| Technologists | 1.09 | 0.54 | 2.19 | 1.02 | 0.51 | 2.05 |
| Sales professionals | 1.24 | 0.69 | 2.21 | 1.20 | 0.67 | 2.16 |
| Other sales | 1.32 | 0.71 | 2.47 | 1.20 | 0.64 | 2.25 |
| Skilled administrative | 1.13 | 0.51 | 2.54 | 1.02 | 0.46 | 2.29 |
| Other administrative | 1.29 | 0.65 | 2.56 | 1.17 | 0.59 | 2.30 |
| Misc. skilled service | 0.76 | 0.25 | 2.36 | 0.69 | 0.22 | 2.13 |
| Protective service | 1.07 | 0.43 | 2.64 | 1.05 | 0.42 | 2.64 |
| Food service | 1.20 | 0.58 | 2.49 | 0.97 | 0.46 | 2.05 |
| Cleaning and building service | 1.33 | 0.58 | 3.00 | 1.18 | 0.51 | 2.70 |
| Health service | 0.14 | 0.02 | 1.04 | 0.12 | 0.02 | 0.89 |
| Personal service | 1.00 | 0.40 | 2.47 | 0.84 | 0.34 | 2.12 |
| Farming, fishing, and forestry | 1.10 | 0.52 | 2.32 | 0.98 | 0.46 | 2.07 |
| Mechanics and repairers | 1.02 | 0.56 | 1.87 | 0.97 | 0.52 | 1.80 |
| Construction and extractive trades | 1.40 | 0.83 | 2.35 | 1.27 | 0.73 | 2.20 |
| Production/manufacturing | 1.20 | 0.70 | 2.04 | 1.14 | 0.66 | 1.96 |
| Transport | 1.19 | 0.62 | 2.26 | 1.11 | 0.58 | 2.14 |
| Other manual | 1.56 | 0.76 | 3.21 | 1.37 | 0.65 | 2.89 |
| Recent unemployed | 1.69 | 0.89 | 3.23 | 1.34 | 0.68 | 2.63 |
| Long-term unemployed | 3.91 | 2.04 | 7.48 | 2.91 | 1.46 | 5.81 |
| Not in labor force | 2.28 | 1.49 | 3.49 | 1.92 | 1.23 | 3.00 |
| Never employed | 1.45 | 0.66 | 3.18 | 1.18 | 0.52 | 2.63 |
| Demographic factors | ||||||
| Age | 1.00 | 0.98 | 1.02 | 1.00 | 0.98 | 1.02 |
| Cohort | ||||||
| 1932-1939 | 0.61 | 0.20 | 1.89 | 0.63 | 0.20 | 1.93 |
| 1940-1949 | 0.65 | 0.42 | 1.01 | 0.64 | 0.41 | 1.00 |
| 1950-1959 | 0.86 | 0.64 | 1.15 | 0.84 | 0.63 | 1.13 |
| 1960-1969 (ref.) | ||||||
| 1970-1979 | 1.05 | 0.76 | 1.46 | 0.98 | 0.70 | 1.36 |
| 1980-1989 | 0.93 | 0.50 | 1.76 | 0.79 | 0.42 | 1.51 |
| Male | 3.62 | 2.81 | 4.66 | 3.30 | 2.57 | 4.25 |
| Race/ethnicity | ||||||
| NH White (ref.) | ||||||
| NH Black | 0.62 | 0.42 | 0.92 | 0.57 | 0.38 | 0.85 |
| Hispanic | 0.88 | 0.60 | 1.27 | 0.88 | 0.60 | 1.28 |
| Other | 2.01 | 1.30 | 3.09 | 2.02 | 1.32 | 3.10 |
| Foreign-born | 0.71 | 0.52 | 0.99 | 0.75 | 0.54 | 1.05 |
| Geographic factors | ||||||
| Urban | 0.94 | 0.74 | 1.21 | 0.90 | 0.70 | 1.15 |
| Census regions | ||||||
| South (ref.) | ||||||
| Northeast | 0.67 | 0.49 | 0.92 | 0.67 | 0.49 | 0.93 |
| North Central/Midwest | 0.88 | 0.68 | 1.14 | 0.89 | 0.69 | 1.16 |
| West | 1.09 | 0.83 | 1.42 | 1.08 | 0.83 | 1.41 |
| Socioeconomic status | ||||||
| Education | ||||||
| Less than HS (ref.) | ||||||
| HS or equal | 1.11 | 0.78 | 1.56 | |||
| Some college | 1.26 | 0.90 | 1.77 | |||
| BA or higher | 1.13 | 0.74 | 1.72 | |||
| Income-to-needs ratio | ||||||
| 0-1.00 (ref.) | ||||||
| 1.01-1.99 | 0.84 | 0.56 | 1.25 | |||
| 2.00-3.99 | 0.80 | 0.55 | 1.18 | |||
| 4.00-4.99 | 0.88 | 0.54 | 1.45 | |||
| 5.00+ | 0.74 | 0.47 | 1.17 | |||
| Married | 0.65 | 0.49 | 0.87 | |||
| Family type | ||||||
| 1 adult, no children (ref.) | ||||||
| 1+ adults, no children | 1.15 | 0.83 | 1.59 | |||
| 1 adult, 1+ children | 1.16 | 0.89 | 1.53 | |||
| 1+ adults, 1+ children | 0.58 | 0.35 | 0.97 | |||
| Home ownership | 0.98 | 0.76 | 1.25 | |||
| Health Insurance | 0.84 | 0.62 | 1.14 | |||
| Sample size | 360,146 | |||||
| Number of deaths | 541 | |||||
Notes:
Analyses based on restricted-use data with non-perturbed, detailed information on cause of death.
Deaths restricted to ages 25-64.
Estimates based on multiple imputation to account for missing data.
Finally, Table 4 includes estimates of accidental poisoning mortality risk, which are far more pronounced across occupational/employment groups. Adults employed in service sector jobs like food service (OR 6.67 [3.37—13.30]) and cleaning and building service (OR 6.08 [2.76—13.41]) are at highest relative risk in Model 1, with elevated risks among those in other administrative, health service, farming, fishing, and forestry, mechanics and repairs, construction and extractive, production/manufacturing, transport, and other manual labor occupations. There is attenuation in these risks after the inclusion of socioeconomic measures in Model 2, but odds ratios remain elevated for food service (OR 3.25 [1.61—6.57]), cleaning and building service (OR 3.10 [1.38—6.94]), farming, fishing, and forestry (OR 2.54 [1.13—6.31]), mechanics and repairs (OR 2.22 [1.00—4.96]), construction and extractive trades (OR 3.00 [1.41—6.35]), and transport occupations (OR 2.38 [1.08—5.25). Notably, health professionals (OR 3.35 [1.18—9.47]) and technologists (OR 3.27 [1.47—7.29]) have the highest risks of accidental poisoning mortality among the occupational categories.
Elevated accidental poisoning mortality risks associated with not working are robust across both models as well. Adults who are recently unemployed (OR 5.74 [2.73—12.07]), long-term unemployed (OR 6.44 [2.69—15.41]), not in the labor force (OR 12.71 [7.46—21.67]), or have never been employed (OR 11.93 [5.76—24.70]) have higher relative risks than their employed, managerial/administrative counterparts. Including socioeconomic factors attenuate these differences, but even in Model 2 there are high relative risks among recently unemployed (OR 3.14 [1.47—6.72]) and long-term unemployed (OR 3.05 [1.25—7.46) adults, as well those not in the labor force (OR 6.71 [3.81—11.81]) or never employed (OR 4.73 [2.19—10.24]).
To provide a better sense of the magnitude of these occupational/employment differences, Figure 1 shows predicted mortality rates corresponding with accidental poisoning mortality risk in Model 2. The range in estimated rates is considerable; the referent managerial/professional administrative group has an estimated 2.88 deaths per 100,000 individuals compared to approximately 9.00 deaths among unemployed adults, 13.63 deaths among never employed adults, and 19.32 deaths among those not in the labor force. Estimated rates for transport, manual labor, service, and health-related professions ranged from approximately 7 to 10 deaths per 100,000.
Figures 1:
Accidental Poisoning Mortality
As discussed prior, we estimated a far more stringent model inclusive of individuals’ health behaviors and conditions (Table S2 in Supplement). Including health and behavioral covariates attenuates the occupation/employment-mortality associations observed for all three causes of death. Suicide mortality risk continues to be higher for those not in the labor force or long-term unemployed, but there are no clear occupation/employment-mortality associations are for alcoholic liver disease mortality. However, for accidental poisoning deaths, adults in service, manual labor, transport, and health-related occupations – as well as those who are not in the labor force – still exhibit two-to-four times the accidental poisoning mortality risk relative to adults in managerial/administrative positions.
Discussion
Amid rising U.S. working-aged mortality from alcoholic liver disease, suicide, and accidental poisoning, the present study examined occupation and employment status disparities in cause-specific mortality risk. Results provide empirical support for the association between traditionally “blue-collar” manual labor and manufacturing jobs and elevated premature mortality risk – especially from accidental poisoning. Moreover, the analyses underscore the significance of not working as a key risk factor for premature mortality from despair-related causes, even after accounting for measures of health. There are strong associations between recent job loss and risk of suicide mortality; conversely, exit from the labor force appears to exhibit a stronger association with accidental poisoning mortality risk. Indeed, the impact of drug and opioid abuse and overdoses is speculated to have contributed to declining labor force participation in the U.S.20,21 The present results echo these concerns, highlighting the strong association between a lack of employment and accidental poisoning mortality risk among working-aged adults.
However, the analysis of detailed occupations further emphasizes the much broader set of occupation/employment groups at elevated risk for accidental poisoning deaths. Adults in service sector and transport occupations – accounting for a large, and growing, proportion of the U.S. labor force11-13,22 – exhibit mortality risks on par with those employed in manual labor positions. Those in the health sector are particularly vulnerable as well, serving as an important reminder that even higher status occupations may offer little protection against harmful aspects of one’s workplace, such as high-stress environments and access to harmful substances.15 Critically, these results challenge the assertion that accidental poisoning deaths are solely concentrated among low-educated, “blue-collar” adults disproportionately impacted by transformations in the economy.
Nevertheless, previous research documents that many manual labor occupations often characterized as “precarious” jobs may contribute to poor quality of life among workers. Workers in these occupations encounter high social and economic insecurity and instability due to outsourcing and decreasing union power in advocating for workers’ rights.1,11-14 The resulting status loss and “despair” that this uncertainty brings about is hypothesized to underlie increases in harmful, and often lethal, behaviors.1,13,14 Conversely, a number of high-risk occupations are in higher demand, such as service jobs in the health care and food industries and various types of transportation 22 However, workers in these jobs often lack long-term employment contracts, are not unionized, and possess few opportunities for advancement.11-13 These psychosocially poor working conditions might also translate into increased mortality risk from alcoholic liver disease and accidental poisonings. While we found stronger evidence for the latter, it is worth noting that alcoholic liver disease can take decades to manifest;23 we may need more years of data to observe such deaths, the etiology of which may be more cumulative and suggestive of sustained exposure to poor workplace conditions.
Limitations
While the occupational categories used are more detailed than many survey-based studies of mortality, data constraints prohibit even greater specification of occupations. Second, we could not assess temporal trends given the moderate number of deaths available. Results from analyses comparing poisoning risk before and after 2011 were inconclusive, especially as the structure of NHIS data results in a greater accumulation of deaths in more recent years; ideally, we could examine whether certain occupations have seen increasing risks over time. Third, employment status and occupation are obtained at time of survey, and may change during the follow-up. Most individuals tend to stay in a primary occupational category throughout their lives;24 however, job mobility, and its impact on health, is important to consider.25 Fourth, the accidental poisoning category reflects a diverse set of deaths. While we know that the majority of deaths are drug overdoses,26 there have been fluctuations in the proportion of deaths attributable to prescription (e.g., OxyContin) versus illegal drugs (e.g., fentanyl and heroin),27 and these different etiologies are worth further investigation. Finally, comprehensive data on individuals’ workplace experiences, exposures, and rewards would allow for a better measure of “precariousness,” which can reveal how psychosocial, physiological, and material sources of risk influence mortality.
Conclusion
Despite increased awareness of the health consequences of work,14,28 most attempts at improving workers’ health – such as wellness programs and rewards – are proximate-level solutions to occupational disparities in health and mortality. These initiatives largely target physical health,29 ignoring its more distal determinants. Successful programs should be a staple of employer-led initiatives to improve employees’ health. However, we contend that a more fundamental shift is necessary, with a focus on structural-level policies that hold workplaces accountable for the “socialization or externalization of the private costs of operating an unhealthy workplace”13 and recognize the importance of stable and secure work for overall health and wellbeing.
For instance, some international counterparts of the U.S. have stringent policies requiring employers to demonstrate that layoffs are truly necessary for maintaining fiscal solvency, and to then provide workers with sufficient severance, recognizing that employees bear the most direct effects of job loss.30 Sweden provides employees with income support and retraining opportunities following large-scale downsizing, such as those associated with job automation.31 Conversely, the U.S. provides no such legal protections to mitigate poor job availability; in fact, certain actions, such as the Supreme Court upholding “Right to Work” laws, only further increase the precariousness associated with many “working-class” occupations by reducing the power and effectiveness of organized labor.32 Given the decline of many “blue-collar” and manual labor jobs, chronic unemployment is an especially important target for effective population health policies. Though resource-intensive, the re-training and subsequent re-employment of working-age adults – especially in emerging industries with a perpetual skill shortage33 – is critical to ensuring their active membership in the labor force and guaranteeing better health and longevity.
However, our results suggest that reducing unemployment is by no means a panacea for achieving lower mortality among working-aged adults; re-employing adults into bad jobs may expose them to comparable levels of risk for poor heath and mortality as their non-working counterparts. This is especially salient for accidental poisoning mortality, as specific occupational groupings – food service workers, cleaning professionals, transport workers, health professionals, and technologists – are at especially high risk. Our study cannot ascertain whether the mechanisms that lead to these elevated risks are the same or different for both the unemployed and those employed in bad jobs; per the despair narrative, we may assume they are similar in that the loss of security and certainty associated with both not working and having a bad job – and being unable “construct a rational life plan or career narrative”12 – produce the same negative psychosocial outcomes.
The precariousness of many blue-collar jobs has increased due to declining job stability and benefits, loss of institutional protections, limited autonomy, and fewer opportunities for upward mobility.11-13,34 These workers’ elevated risks of despair-related death might be a consequence of increased stress and subsequent negative coping behaviors.13,14 Accordingly, an appropriate policy-based response would counteract the increasing precarious nature of these jobs, ideally in the form of large-scale reforms like a higher minimum wage, a mandate for essential employer-provided benefits, and laws that protect and empower collective bargaining, all of which would help to maintain a standard of living and respectability for all occupations. Further, more targeted interventions, such as granting workers greater control over their work schedule,35 can reduce the instability and anxiety associated with these occupations as well.
Conversely, there are occupation-specific risks that such broad policies may overlook; if health professionals are at especially high risk of accidental poisonings because of greater access to prescription drugs, educational programs and screening policies should be implemented to develop greater awareness of these lethal and potentially unrecognized hazards.36 The precariousness of these jobs may have less to do with workplace-induced ‘demand’ for drugs – as may be the case among disenfranchised blue-collar workers – and may instead stem from a greater ‘supply’ of or access to drugs, resulting in comparable levels of risk. Indeed, no occupation has a “monopoly” on precariousness,14 as there are many ways that jobs have harmful (and beneficial) attributes. A better understanding of the unique challenges encountered by both the unemployed and precariously-employed – which account for a large, and growing, proportion of the U.S. population – is essential for reducing mortality from etiologically complex and multifactorial causes of death like alcoholic liver disease, suicide, and accidental poisoning.
To close, policymakers must look beyond educational attainment and income as socioeconomic determinants of U.S. mortality, especially given the declining availability of traditionally good jobs.11 Increasing educational attainment and more egalitarian income policies are important policy concerns; however, neither offers full protection against unemployment, or is a guarantee of a stable, secure, and safe work environment. Working-aged adults who are not working – and thus not deriving the material and psychosocial rewards associated with employment – are a particularly high-risk population whose elevated vulnerability amidst the ongoing drug and opioid epidemic warrants further attention. Even among those currently employed, a bad and unrewarding job presents its own set of risks. Certain working-class sectors such as service, transport, and some manual jobs are growing and will continue to require millions of new workers; failing to address the harmful environments associated with these occupations may lead to an increasing proportion of the working-age population ‘at risk’ for worsening health and premature mortality.
Supplementary Material
Research Highlights.
We examine the association between work and “despair”-associated mortality risk.
Service, manual labor, and transport workers have high accidental poisoning risk.
Accidental poisoning risk is comparable among health professionals.
Non-working adults have consistently high “despair”-associated mortality risk.
Workplace context and unemployment are key risk factors for working-aged mortality.
Acknowledgements
We would like to acknowledge our funders, as this research was supported by the Population Research Training grant (T32 HD007168) and the Population Research Infrastructure Program (P2C HD050924) awarded to the Carolina Population Center at The University of North Carolina at Chapel Hill by the Eunice Kennedy Shriver National Institute of Child Health and Human Development. We are grateful to the National Center for Health Statistics (NCHS) and Minnesota Population Center for making the public- and restricted-use data available for this paper. Finally, we would like to thank our colleagues Richard G. Rogers, Dan A. Powers, Elizabeth Lawrence, Andrea M. Tilstra, Nathan T. Dollar, David B. Braudt, and Samuel H. Fishman for their feedback and statistical expertise.
Footnotes
Conflict of Interest
None.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Case A, Deaton A. Rising morbidity and mortality in midlife among white non-Hispanic Americans in the 21st century. Proceedings of the National Academy of Sciences. 2015. December 8;112(49):15078–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Chetty R, Stepner M, Abraham S, Lin S, Scuderi B, Turner N, Bergeron A, Cutler D. The association between income and life expectancy in the United States, 2001-2014. Jama. 2016. April 26;315(16):1750–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Montez JK, Zajacova A. Explaining the widening education gap in mortality among US white women. Journal of health and social behavior. 2013. June;54(2): 166–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Sasson I Trends in life expectancy and lifespan variation by educational attainment: United States, 1990–2010. Demography. 2016. April 1;53(2):269–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Masters RK, Tilstra AM, Simon DH. Explaining recent mortality trends among younger and middle-aged White Americans. International journal of epidemiology. 2017. July 19;47(1): 81–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Woolf SH, Chapman DA, Buchanich JM, Bobby KJ, Zimmerman EB, Blackburn SM. Changes in midlife death rates across racial and ethnic groups in the United States: systematic analysis of vital statistics. Bmj. 2018. August 15;362:k3096. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Xu JQ, Murphy SL, Kochanek KD, Arias E. Mortality in the United States, 2018. NCHS Data Brief. 2020. January 30:355. [PubMed] [Google Scholar]
- 8.Durkheim E The Division of Labor in Society. London: Collier-Macmillan; 1964. [Google Scholar]
- 9.Marx K The Marx-Engels Reader. New York: WW Norton & Co; 1978. [Google Scholar]
- 10.Weber M Economy and Society: an Outline of Interpretive Sociology. Berkeley: University of California Press; 1968. [Google Scholar]
- 11.Kalleberg AL. Good Jobs, Bad Jobs: the Rise of Polarized and Precarious Employment Systems in the United States, 1970s to 2000s. New York, NY: Russell Sage Foundation; 2011. [Google Scholar]
- 12.Kalleberg AL, Vallas SP, editors. Probing precarious work: Theory, research, and politics In Precarious work 2017. (pp. 1–30). Emerald Publishing Limited. [Google Scholar]
- 13.Pfeffer J Dying for a Paycheck: How Modern Management Harms Employee Health and Company Performance - and What We Can Do About It. New York, NY: HarperCollins; 2018. [Google Scholar]
- 14.Tompa E, Scott-Marshall H, Dolinschi R, Trevithick S, Bhattacharyya S. Precarious employment experiences and their health consequences: towards a theoretical framework. Work. 2007. January 1;28(3):209–24. [PubMed] [Google Scholar]
- 15.Agerbo E, Gunnell D, Bonde JP, Mortensen PB, Nordentoft M. Suicide and occupation: the impact of socio-economic, demographic and psychiatric differences. Psychological medicine. 2007. August;37(8):1131–40. [DOI] [PubMed] [Google Scholar]
- 16.Coggon D, Harris EC, Brown T, Rice S, Palmer KT. Occupation and mortality related to alcohol, drugs and sexual habits. Occupational medicine. 2010. August 1;60(5):348–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Stack S Occupation and suicide. Social Science Quarterly. 2001. June;82(2):384–96. [Google Scholar]
- 18.Blewett LA, Rivera Drew JA, Griffin R, King ML, Williams K. IPUMS health surveys: National health interview survey, Version 6.2. Minneapolis: University of Minnesota; 2016. [Google Scholar]
- 19.Keyes KM, Rutherford C, Popham F, Martins SS, Gray L. How healthy are survey respondents compared with the general population?: using survey-linked death records to compare mortality outcomes. Epidemiology. 2018. March;29(2):299–307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Krueger AB. Where have all the workers gone? An inquiry into the decline of the US labor force participation rate. Brookings papers on economic activity. 2017;2017(2):1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Ruhm CJ. Deaths of despair or drug problems?. National Bureau of Economic Research; 2018. January 5. [Google Scholar]
- 22.Fastest Growing Occupations: Occupational Outlook Handbook: U.S. Bureau of Labor Statistics, https://www.bls.gov/ooh/fastest-growing.htm. Accessed August 12, 2019. [Google Scholar]
- 23.Osna NA, Donohue TM Jr, Kharbanda KK. Alcoholic liver disease: Pathogenesis and current management. Alcohol research: current reviews. 2017;38(2):147. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Rytina S Is occupational mobility declining in the US?. Social Forces. 2000. June 1;78(4): 1227–76. [Google Scholar]
- 25.Moore DE, Hayward MD. Occupational careers and mortality of elderly men. Demography. 1990. February 1 ;27(1):31–53. [PubMed] [Google Scholar]
- 26.Rudd RA, Seth P, David F, Scholl L. Increases in drug and opioid-involved overdose deaths—United States, 2010–2015. Morbidity and mortality weekly report. 2016. December 30;65(50 & 51): 1445–52. [DOI] [PubMed] [Google Scholar]
- 27.Kiang MV, Basu S, Chen J, Alexander MJ. Assessment of changes in the geographical distribution of opioid-related mortality across the United States by opioid type, 1999-2016. JAMA network open. 2019. February 1;2(2):e190040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Benach J, Vives A, Amable M, Vanroelen C, Tarafa G, Muntaner C. Precarious employment: understanding an emerging social determinant of health. Annual review of public health. 2014. March 18;35. [DOI] [PubMed] [Google Scholar]
- 29.Goetzel RZ, Henke RM, Tabrizi M, Pelletier KR, Loeppke R, Ballard DW, Grossmeier J, Anderson DR, Yach D, Kelly RK, Serxner S. Do workplace health promotion (wellness) programs work?. Journal of Occupational and Environmental Medicine. 2014. September 1;56(9):927–34. [DOI] [PubMed] [Google Scholar]
- 30.Bender S, Dustmann C, Margolis D, Meghir C. (2002). Worker displacement in France and Germany In Losing Work, Moving on: International Perspectives on Worker Displacement 2002 (pp. 375–470). Kalamazoo, MI: W.E. Upjohn Institute for Employment Research. [Google Scholar]
- 31.Manyika J, Lund S, Chui M, et al. Jobs lost, jobs gained: Workforce transitions in a time of automation. McKinsey Global Institute. 2017. [Google Scholar]
- 32.Wallace D, Wallace R. Right-to-Work Laws and the Crumbling of American Public Health. Springer International; 2019. [Google Scholar]
- 33.Cappelli PH. Skill gaps, skill shortages, and skill mismatches: Evidence and arguments for the United States. ILR Review. 2015. March;68(2):251–90. [Google Scholar]
- 34.Standing G A precariat charter: From denizens to citizens. A&C Black; 2014. April 10. [Google Scholar]
- 35.Kowalski-McGraw M, Green-McKenzie J, Pandalai SP, Schulte PA. Characterizing the interrelationships of prescription opioid and benzodiazepine drugs with worker health and workplace hazards. Journal of occupational and environmental medicine. 2017. November;59(11):1114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Merlo LJ, Gold MS. Prescription opioid abuse and dependence among physicians: hypotheses and treatment. Harvard review of psychiatry. 2008. January 1;16(3): 181–94. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.