Grosch, James W. PhD; Murphy, Lawrence R. PhD
During the past two decades, it has become increasingly clear that occupations differ in their impact on an employee's mental and physical well-being. Evidence exists for occupational differences in studies of cardiovascular disease,1,2 high blood pressure,1,3 smoking,4,5 alcohol/drug abuse,6 job stress,7,8 depression,9,10 mental health disorders,11 and suicide.12 One limitation of this research, however, has been the scarcity of large representative data sources from which comparisons across a wide variety of occupations can be made. Occasionally, occupational comparisons have been conducted relying on data that in all likelihood is not representative of the general population.11 Even when large representative samples are used, only minimal information is available concerning a person's occupation, so that factors such as tenure or a recent job change are not taken into account.9
The study presented here sought to extend research on occupational differences by examining two basic measures of well-being-depression and global health-that vary by occupation.9,13 Depression is one of the most prevalent mental health problems in the general population and has the largest medical costs of all mental health disorders in the workplace.14,15 Depression has been associated with increased absenteeism, reduced productivity, and greater safety risks in organizations.16,17 Self-evaluated global health has been found to be reasonably congruent with a physician's judgment and associated with variations in chronic illness, disability, and mortality in the elderly.18-20 The predictive value of global health remains even after controlling for health-related behavior (eg, smoking) and prior medical diagnoses from a physical examination.16,17
Together, depression and global health represent slightly different perspectives on well-being that have rarely been studied with regard to occupation because of the difficulty in studying a large cross-section of occupations with the same measures. The research described here uses data from a national health survey to rank order occupations with regard to depression and global health. Important features of the study include the measurement of depression and global health as multiple-item, continuous variables, and the collection of detailed occupational information for each respondent.
Data for this study were obtained from the National Medical Expenditure Survey (NMES), a national probability, household interview sponsored by the Agency for Health Care Policy and Research in 1987.21,22 The primary goal of the NMES was to assess the use of health services, expenditures on health care, and insurance coverage in the civilian, noninstitutionalized US population. The NMES also included questions on demographics, employment history, and health status. One component of the NMES was a self-administered questionnaire (SAQ) that respondents completed on their own and which contained the depression and global health measures, as well questions concerning health conditions (eg, high blood pressure) and habits (eg, smoking).
Data for the NMES were collected during four separate interviews with respondents, conducted at three- to four-month intervals. The SAQ was distributed between the first- and second-round interviews and returned to the interviewer after being completed by each respondent. Approximately 30,000 respondents both completed the SAQ and provided data at all four interviews. This represents an overall response rate of 72% (90% completion rate for SAQ × 80% participation rate across all four rounds) for the NMES.
NMES respondents were selected for this study only if they were between the ages of 17 and 65, were currently employed in an occupation with a valid 1980 Census code, had not changed jobs in the past year, and had completed the self-administered questionnaire. This yielded a total of 9,281 respondents that were used in the analysis. Approximately 48% of the respondents were female, with 74% being white, 17% black, and 9% Hispanic. These percentages reflect the fact that the NMES was designed to oversample disadvantaged populations. Respondents selected for this study averaged 38.4 years of age and had worked in their current job for an average of 7.8 years. Approximately 20% of the respondents were members of a labor union.
The depression scale consisted of five items patterned after the General Mental Health subscale of the Medical Outcomes Study.23 Respondents indicated on a six-point Likert scale how often during the past thirty days they had felt "nervous," "calm and peaceful," "downhearted and blue," "happy," and "so down in the dumps that nothing could cheer you up." The health scale consisted of four items, also adapted from the Medical Outcomes Study, and required respondents to rate on a four-point Likert scale the truthfulness of the following statements: "I am somewhat ill," "I'm as healthy as anybody I know," "I have been feeling bad lately," and "My health is excellent." Items in both scales were reverse-scored, so that higher values indicate a worse outcome. In other words, a higher score indicates greater depression and poorer global health.
Table 1 presents summary statistics for the depression and global health scales. Both scales had an alpha coefficient of.80 or greater and were significantly correlated with several other health-related variables on the NMES. The global health scale was more predictive of variables related to physical well-being (eg, number of prescribed medicines, arthritis, backache), whereas the depression scale was more predictive of variables related to social functioning (eg, sharing feelings with others). In all cases, the correlations were in the expected direction, with higher scores on depression and global health having a positive association with poorer functioning on the other NMES variables. The Pearson correlation between depression and global health was .43 (P< 0.001), indicating that these two constructs, while overlapping, share less than 20% of the total variance between them.
Each respondent's employment status and occupation were assessed during the interview portion of the NMES, and these data were later merged with responses on the self-administered questionnaire. Information concerning employment status included whether or not the respondent was currently working, hours worked per week, tenure in present job, and previous employment. Occupation was assessed by asking respondents to describe their job title, type of business, and important job-related activities. This information was recorded verbatim and then used by trained NMES coders to match the respondent's current job to one of the 502 detailed occupational categories in the 1980 census, represented by a three-digit code. This code was used to place each respondent's job into one of 11 broader occupational categories also used in the 1980 census. These categories were: managerial(codes 003-037), professional (043-199), technical/administrative support(203-235), sales (243-285), clerical (303-389), service (403-469), farming and forestry (473-499), craftsmen and precision workers (503-699), machine operators (703-799), transportation (803-859), and laborers (863-889).
Data Analytic Procedure
Our approach in analyzing the NMES data was primarily descriptive, with the goal of rank-ordering occupations according to depression and global health, while controlling for the effects of several confounding variables. Each respondent's depression and global health score was first adjusted for age, tenure, race, sex, and hours worked per week, since these variables represent potential confounds. The correlation between these variables and depression and global health was consistent with that found in previous research. For example, increasing age was positively correlated with higher global health scores (r = .21), whereas being male was negatively correlated with both depression and global health scores (r = -.12 and -.07, respectively).
Once statistical adjustments had been made, respondents were then grouped according to their three-digit occupational census codes. Only occupations with at least seven respondents were retained. This minimum number was chosen to ensure more reliable occupational means and to prevent extreme responses by a single respondent from having a disproportionate impact on occupational scores for depression and global health. The selection process yielded 8,486 respondents distributed across 239 occupations. This number of respondents represents approximately 93% of all respondents who qualified for the study, indicating that the inclusion criteria resulted in minimal subject loss. Examples of the occupations that were left out of the analysis because of a small sample size include boilermakers (n = 1), nursery workers(n = 4), judges (n = 2), occupational therapists(n = 5), dressmakers (n = 3), and bookbinders(n = 4).
Occupations selected for the analysis were rank-ordered according to their adjusted means for depression and global health. The adjusted mean scores were then standardized (z scores) to facilitate direct comparisons among occupations. A z score indicates the number of standard deviation units a particular occupation falls above or below the mean for all occupations in the sample. Because of how depression and global health were scored, a negative z score indicates a healthier outcome since the adjusted mean for an occupation falls below the overall adjusted mean for all occupations in the sample. A positive z score indicates that an occupation had a higher (or worse) score on depression or global health, compared with the overall adjusted mean.
Finally, all occupations within a given occupational category (eg, clerical) were combined, and overall adjusted means and z scores for depression and global health were calculated by averaging across all occupations in that category. Throughout this analysis, only unweighted NMES data were used.
Table 2 presents a comparison of the 11 occupational categories for the depression and global health measures. Data are provided in terms of the adjusted mean score, 95% confidence interval, and meanz score for each occupational category. The final column inTable 2 presents an average of the combined z scores for depression and global health for each occupational category.
In terms of depression, managerial, technical, and professional occupations had the lowest z scores, whereas machine operators, farming/forestry, and transportation occupations had the highest. In general, occupational categories with negative z scores on depression also had negative z scores on global health. Two notable exceptions to this rule were laborers and farming/forestry workers. For laborers, a negative z score on depression was accompanied by a positivez score on global health. The reverse pattern was observed for farming/forestry workers. When z scores for depression and global health were averaged (see final column), the occupational category with the lowest score was professional, followed by managerial and technical occupations. The category with the highest score was machine operators, followed by transportation and clerical occupations.
Analysis of variance found that the overall difference between occupational categories on the depression measure was statistically significant (F(10,238) = 2.34, P < 0.05), with occupational category accounting for 9.3% of the total variance in depression. Similarly, the overall difference between categories on the global health measure was significant (F(10,238) = 7.63,P < 0.001), with occupational category accounting for 25.1% of the total variance. Thus, occupational category appears to be more closely associated with global health than with depression.
Table presents depression and global health scores for each of the 239 occupations, listed in ascending order according to the 1980 Census codes. For each occupation, data are presented regarding the number of respondents and the occupational mean and z scores on the depression and global health measures. To identify scores that fall well beyond the overall mean, z scores below -1.0 are in boldface and underlined, whereasz scores above + 1.00 are in boldface and italics. A z score of ± 1.0 means that an occupation falls within the upper/lower 15.87% of the distribution of scores for that measure. Occupational differences in Table 3 should be interpreted cautiously since small differences may not be statistically significant. However, significant differences do exist when comparing occupations that have az score above 1.00 with those occupations that have a z score below - 1.00 (eg, file clerks vs librarians).
The data displayed in Table 3 are further broken down and organized in Tables 4 and 5.Table 4 lists the ten occupations with the highest and lowest z scores on depression and global health. With regards to depression, sawing machine operators and foreign language teachers had the highest and lowest scores, respectively. For global health, tool and die makers and business, commerce, and marketing teachers had the highest and lowest scores, respectively. Table 5 presents an alphabetical listing of occupations that had z scores for depression and global health that were either both above or equal to +1.00 or both below or equal to - 1.00. These are occupations that were consistently high or low on the depression and global health measures. There were 19 occupations with both z scores at 1.00 or above (eg, messengers, upholsterers) and 15 occupations with bothz scores at -1.00 or below (eg, counselors, educational and vocational and purchasing managers). Only three occupations had extremez scores that went in opposite directions. These were science technicians (225), supervisors, related agricultural occupations (485), and miscellaneous electrical and electronic equipment repairers (533).
The results of this study indicate that workers in different occupations and occupational groups differ significantly in terms of depression and global health. Overall, workers in professional and managerial occupations tend to have less depression and better global health than workers in machine operation or transportation occupations. However, variation exists within each category, so caution is warranted when predicting the results for a specific occupation based on category alone. For example, although workers in the category of machine operators have the highest level of depression, pressing machine operators have an adjusted depression z score less than -1.00 (Tables 2 and 3). Also, occupations that seen similar (eg, special education and kindergarten teachers) can score much differently on depression and global health (Tables 3 and 5). One potential benefit of the rankings presented in Table 3 is that they can be used to identify specific occupations in which workers may be at risk for the development of depression and/or poor health.
Given the occupational differences identified in this study, an important question remains as to why they exist. Clearly, a number of factors may contribute to and help explain the observed differences. First, occupations vary in terms of hazards in the physical environment encountered by employees. For example, the occupations of painters and dry cleaning machine operators, which were both high in depression and global health(Table 3), are also known to be at risk for chemical exposures that can produce serious health and emotional effects.24-26 Similarly, occupations in industries such as transportation, construction, manufacturing, and agriculture have been found to involve more dangerous working conditions that, over time, may take a toll on an employee's well-being.27,28
Second, since most employees have some degree of choice as to which occupation they pursue, self-selection may play a role in observed occupational differences. For example, the low depression and global health scores of firefighters, despite their hazardous work environment, may reflect a "healthy worker effect," in which employees going into a particular occupation tend to be healthier or better adjusted than employees in the general population.29,30 Conversely, some employees may gravitate or drift toward a particular occupation because it is less demanding and more tolerant of health and emotional problems they may be experiencing. In either case, self-selection results in occupational differences that reflect pre-existing personal characteristics of employees who decide to go into certain occupations.
Third, occupations vary in terms of the psychosocial conditions experienced by employees. One popular model of job stress, for instance, proposes that work environments low in employee control and high in work demands produce the worst health outcomes, particularly as measured by heart-related disorders (eg, high-blood pressure, myocardial infarction, etc).31,32 According to this theory, occupations such as messengers, kitchen workers, and postal clerks score poorly on global health (Table 3) because they typically involve little autonomy and a high workload. Additional psychosocial conditions that affect employee well-being, and which may vary across occupations, include co-worker support33 and the amount of role conflict and ambiguity in the workplace.34 Unfortunately, in the study presented here, psychosocial conditions were not directly assessed on the NMES and can only be inferred after the fact from what is generally known about each occupation.
Fourth, occupations may differ in factors that have an impact on depression and health but which are not directly part of the psychosocial conditions in the workplace. Two such factors, related to socioeconomic status, are education and personal income, which are widely acknowledged to be predictive of health and emotional functioning.35-37 In the present study, level of education was correlated-.07 and -.18 with depression and global health, respectively. The corresponding correlations for personal income were -.07 and -.10. Education and personal income were not controlled for in Tables 2 through 5 because they did not remain significant predictors of depression and global health once we accounted for occupation. Additional analyses, adjusting for the influence of education and income, were conducted but did not produce occupational rankings much different from those presented in this article. For example, of the ten occupations inTable 4 with the highest z scores for depression, nine retain that status after adjustment for education and personal income. The one change was for miscellaneous precision workers, n.e.c., which shifted from having the 10th highest score to the 11th highest. Similarly, when global health was examined, nine of the ten occupations listed in Table 5 as having the highest scores continued to do so after adjustment for education and personal income (miscellaneous textile machine operators changed from the 8th highest score to the 17th highest). These findings indicate that adjustment for education and personal income may produce small changes in occupational rankings, but large changes are rare. However, both education and personal income remain important variables that may influence health-related behaviors (eg, diet, smoking) and how employees respond to health problems they experience. Additional extraneous factors affecting levels of depression and global health include the availability and cost of health insurance and sick leave policy, which can vary greatly across different occupations.
In summary, the underlying reasons for occupational differences in depression and global health are complex and require data that go beyond that available in the NMES. Indeed, a strong need exists for nationally representative data on working conditions and job characteristics associated with different occupations.38 To date, the only national effort to monitor working conditions and explore quality-of-work-life issues was the Quality of Employment Survey, which was sponsored by the Department of Labor in 1969, 1973, and 1977. More extensive and up-to-date data are needed before we can begin to systematically examine the different causes of occupational differences in well-being.
Once the causes are better understood, effective interventions will be easier to identify. For example, if psychosocial work conditions (eg, lack of control) play an important role in occupational differences, then interventions might focus on changing specific features of the job or work environment (eg, greater employee participation in decision making). On the other hand, if occupational differences are due more to physical conditions or self-selection factors, then interventions might focus on better ergonomic design of equipment and training.
The data from this study reinforce the notion that mental and physical health are closely interrelated and predictive of several outcome variables, including high blood pressure, backache, and days missed from work because of illness. The fact that two occupational categories inTable 2 (forestry and laborers) had widely differentz scores for depression and global health is an anomalous finding in need of further investigation. This result may be due to the small number of occupations in each category or reflect the fact that some occupations have qualities that promote one type of health at the expense of the other. In either case, depression and global health tend to change together, with global health being more closely linked to occupation.
In terms of previous research, studies on occupation and well-being have adopted different methodologies and have reported results not entirely consistent with those of this study. For example, one study compared 30 occupations in terms of admission rates for mental health disorders in the state of Tennessee.11 Unfortunately, census codes were unavailable, and the sample was probably not representative of the general population. Of the 22 occupations with the highest admission rates, six were related to hospital/health care professions (eg, practical nurses, health technologists). In the study presented here, none of the occupations associated with health care had z scores above 1.00. However, one occupation that had high admission rates for mental health disorders was dishwasher, which is similar to the occupation of kitchen worker that scored high on both depression and global health (Table 3). In another study,9 depression was measured across 105 occupational categories through the use of the Diagnostic Interview Schedule (DIS), a structured interview designed to assess whether an individual is clinically depressed. A few of the occupations that ranked high on depression also have z scores above 1.0 in Table 3 (eg, tool and die makers, waiters/waitress assistants, maids, and housemen); however, others, such as lawyers, typists, and social workers, do not. Using the same dataset, another study reported that occupational categories with the highest prevalence of depression included machine operators, professionals, and clerical workers.6 Unfortunately, these prevalence rates were not adjusted for either tenure or sex, two variables that may significantly affect the rankings. Together, these findings indicate that occupational rankings of well-being may depend on how well-being is measured, as well as the method used to categorize and compare occupations.
The strengths of this study include its large representative sample of employees, and the availability of work-related information (eg, tenure, hours worked per week) for each employee, which is not always collected in other national health surveys. In addition, occupation and well-being were measured at different times so that the respondents were not aware of how the information they provided would be analyzed. Limitations include the study's cross-sectional design, which prevents any conclusion concerning cause and effect, and the relatively small number of employees in some occupations. Clearly, research of a longitudinal nature is needed to better understand the dynamics of how work and non-work variables interact to affect employee depression and global health.
In conclusion, the results from this study are perhaps most consistent with research that has continued to show that many types of blue-collar occupations are at risk for poor health when compared with professional or managerial occupations.10,13,39 However, we currently have little systematic data to explain why these occupational differences exist. Additional research is needed that will either incorporate already existing databases on occupational characteristics(eg, Dictionary of Occupational Titles) or collect new data on a national basis that will provide insight into variables that will help explain occupational differences in depression and global health.
1. Caplan RD, Cobb S, French JRP Jr, Harrison RV, Pinneau SR Jr. Job Demands and Worker Health. Washington, DC: US Department of Health, Education, and Welfare; 1975.
2. Murphy LR. Job dimensions with severe disability due to cardiovascular disease. J Clin Epidemiol. 1991;44:155-166.
3. Leigh JP. A ranking of occupations based on the blood pressures of incumbents in the National Health and Nutrition Examination Survey I. J Occup Med. 1991;33:853-861.
4. Sterling TD, Weinkam JJ. Smoking characteristics by type of employment. J Occup Med. 1976;18:743-754.
5. Nelson DE, Emont SL, Brackbill RM, Cameron LL, Peddicord J, Fiore MC. Cigarette smoking prevalence by occupation in the United States.J Occup Med. 1994;36:516-525.
6. Roberts RE, Lee ES. Occupation and the prevalence of major depression, alcohol, and drug abuse in the United States.Environ Res. 1993;61:266-278.
7. Karasek RA. Job demands, job decision latitude, and mental strain: implications for job redesign. Adm Sci Q. 1979;24:285-308.
8. Ganster DC, Schaubroeck J. Work stress and employee health. J Manage. 1991;17:235-271.
9. Eaton WW, Anthony JC, Mandel W, Garrison R. Occupations and the prevalence of major depressive disorder. J Occup Med. 1990;32:1079-1087.
10. Vezina M Gingras S. Travail et santé mentale: les groupes à risque. Can J Public Health. 1996;87:135-140.
11. Colligan MJ, Smith MJ, Hurrell JJ Jr. Occupational incidence rates of mental health disorders. J Human Stress. 1977;3:34-39.
12. Boxer PA, Burnett C, Swanson N. Suicide and occupation: a review of the literature. J Occup Med. 1995;37:442-452.
13. Dryson E, Metcalf P, Baker J, Scragg R. Health status and occupation: use of a health status index to measure the health of occupations. Occup Med. 1995;45:35-39.
14. Conti DJ, Burton WN. The economic impact of depression in a workplace. J Occup Med. 1994;36:983-988.
15. Rice DP, Miller LS. The economic burden of affective disorders. In: Advances in Health Economics and Health Services Research, vol. 14. Greenwich, CT: JAI Press Inc; 1993:37-53.
16. Greenberg PE, Stiglin LE, Findelstein SN, Berndt ER. The economic burden of depression in 1990. J Clin Psychiatry. 1993;54:405-418.
17. Martin JK, Blum TC, Beach SRH, Roman PM. Subclinical depression and performance at work. Soc Psychiatry Psychiatr Epidemiol. 1996;31:3-9.
18. Goldstein MS, Siegel JM, Boyer R. Predicting changes in perceived health status. Am J Public Health. 1984;74:611-613.
19. Idler EL, Angel RJ. Self-rated health and mortality in the NHANES-I epidemiologic follow-up study. Am J Public Health. 1990;80:446-452.
20. Mossey JM, Shapiro E. Self-rated health: a predictor of mortality among the elderly. Am J Public Health. 1982;72:800-808.
21. Agency for Health Care Policy and Research.Supplement Data File Documentation: 1987 National Medical Expenditure Survey: Public Use Tape 9-Household Survey Health Status Questionnaires and Access to Care. Rockville, MD: Public Health Service; 1991.
22. Cohen S, DiGaetano R, Waksberg J. Sample Design of the 1987 Household Survey (AHCPR Pub. No. 91-0037). National Medical Expenditure Survey Methods 3. Agency for Health Care Policy and Research. Rockville, MD: Public Health Service; 1991.
23. Ware JE, Sherbourne CD. The MOS 36-item short-form health survey (SF-36): conceptual framework and item selection. Med Care. 1992;30:473-483.
24. Baker EL. A review of recent research on health effects of human occupational exposure to organic solvents. J Occup Med. 1994;36:1079-1092.
25. Grosch JW, Neale AV, Demers RY. Neurobehavioral and health-related deficits in solvent-exposed painters. Am J Ind Med. 1996;30:623-632.
26. Tuttle TC, Wood GD, Grether CB, Johnson BL, Xintaras C. A Behavioral and Neurological Evalution of Dry Cleaners Exposed to Perchloroethylene. [USDEH (NIOSH) Publication No. 77-214.] Cincinnati, OH: NIOSH Publication Office; 1977.
27. Bell CA, Stout NA, Bender TR, Conroy CS, Crouse WE, Myers JR. Fatal occupational injuries in the United States, 1980 through 1985. JAMA. 1990;263:3047-3050.
28. Levy BS, Wegman DH. Occupational Health: Recognizing and Preventing Work-related Disease, 3rd ed. Boston: Little, Brown and Company; 1995.
29. Choi BCK. Definition, sources, magnitude, effect modifiers, and strategies of reduction of the health worker effect. J Occup Med. 1992;34:979-988.
30. Howe GR, Chiarelli AM, Lindsay JP. Component and modifiers of the healthy worker effect: evidence from three occupational cohorts and implications for industrial compensation. Am J Epidemiol. 1988;128:1364-1375.
31. Karasek RA, Baker E, Marxer F, Ahlbom A, Theorell T. Job decision latitude, job demands, and cardiovascular disease: a prospective study of Swedish men. Am J Public Health. 1981;71:694-705.
32. Schnall PL, Pieper C, Schwartz JE, et al. The relationship between 'job strain,' workplace diastolic blood pressure, and left ventricular mass index. JAMA. 1990;263:1929-1935.
33. Landsbergis PA, Schnall PL, Dietz D, Friedman R, Pickering T. The patterning of psychological attributes and distress by "job strain" and social support in a sample of working men. J Behav Med. 1992;15:379-405.
34. Jackson SE, Schuler RS. A meta-analysis and conceptual critique of research on role ambiguity and role conflict in work settings.Organ Behav Hum Decis Process. 1985;36:16-28.
35. Syme SL, Berkman LF. Social class, susceptibility and sickness. Am J Epidemiol. 1976;104:1-8.
36. Matthews, KA, Kelsey SF, Meilahn EN, Kuller LH, Wing RR. Educational attainment and behavioral and biologic risk factors for coronary heart disease in middle-aged women. Am J Epidemiol. 1989;129:1132-1144.
37. McDonough P, Duncan GJ, Williams D, House J. Income dynamics and adult mortality in the United States, 1972 through 1989.Am J Public Health. 1997;87:1476-1483.
38. Sauter SL, Murphy LR, Hurrell JJ Jr. Prevention of work-related psychological disorders. Am Psychol. 1990;45:1146-1158.
39. Buring JE, Evans DA, Fiore M, Rosner B, Hennekens CH. Occupation and risk of death from coronary heart disease. JAMA. 1987;258:791-792.
© Williams & Wilkins 1998. All Rights Reserved.