In most research studies, “job strain” was only measured at one point in time (Schnall, Landsbergis & Baker, 1994). Yet, it is believed that cumulative exposure to “job strain” increases risk of hypertension or CHD. Most likely it is the chronic biological arousal due to sustained “job strain” that contributes to the development of essential hypertension (Schnall, Landsbergis & Baker, 1994). If duration of exposure to “job strain” is not measured, then we cannot determine whether the stressful work environment has existed for the person for only the previous month, or for the previous 40 years. Use of current exposure as a surrogate for lifetime exposure is inaccurate, in part, since people often gain skills with time and age, may be promoted, may select out of “high strain” jobs, or their job characteristics may change even within the same job title. For example, in the Cornell study, 22% of the study participants changed “job strain” status over the course of 3 years (Landsbergis et al., 1995). Many study participants with a lengthy history of “job strain” might thus be currently classified as “non-strain” because of recent promotions or other job changes. Use of inaccurate measures of exposure to “job strain” (i.e., non-differential misclassification) can bias results towards the null hypothesis, leading to the conclusion that the effect of “job strain” on blood pressure is weaker than it truly is.
A work history interview, developed by Paul Landsbergis for the Cornell ambulatory blood pressure study, included two questions each on job demands, decision latitude and social support. These questions were asked of the study participant for each past job held. Interviews of 284 cohort study participants (who reported a total of 1,366 jobs), were completed as of 1/1/95. All subjects participating in the third round of data collection agreed to the interviews. Eligible interview subjects included 212 men enrolled in the cohort study at Time 1, 6 men enrolled at Time 2, 21 women enrolled at Time 1 and 45 women enrolled at Time 2. In addition, 100 nurses and aides, newly enrolled in the study at time 3, have completed a questionnaire version of the work history interview.
For the 284 completed interviews, internal consistency of the three two-item scales was acceptable (workload demands, a=.81; job decision latitude, a=.62; workplace social support, a=.63). In order to increase the reliability of the critical job decision latitude scale, two items were added to the questionnaire: “the freedom to decide how you do your work” was added to the decision-making authority subscale; “the chance to be creative” was added to the skill utilization subscale. Thus, the decision latitude of each past job is now measured by four items. Among the 155 subjects who have answered all 4 latitude items, scale reliability has increased to a=.83.
Work History Questionnaire
Listed below are questionnaire items used to define job demands, job decision latitude and workplace social support for each past job in the Work History Questionnaire used in the Cornell study. “On that job, did you have….
A) Psychological Job Demands
1) To work very hard
2) An excessive amount of work
B) Job Decision Latitude
1) The freedom to decide how you do your work
2) A lot of say about what happens on the job
3) The chance to be creative
4) A high level of skill
C) Workplace Social Support
1) Helpful coworkers
2) A helpful supervisor
Measures of cumulative exposure to be analyzed are based on recent research by Jeffrey Johnson and colleagues (Johnson et al., 1991; Johnson & Stewart, 1993). They used Swedish national data bases to compute average job demands, control (latitude) and support scores for each year of a person’s work history, based upon their job title, age, gender and years of employment. In order to determine the effects of total work history exposure, as well as the time course of exposure (whether earlier or later exposure affects outcome), they constructed both discrete and cumulative 5-year exposure windows. Thus, mean job characteristics scores in each of the following time periods were analyzed for association with future CVD:
In the Swedish study, exposure among men to low control jobs, within the previous 25 years, was prospectively associated with CVD mortality (Johnson et al., 1991). Higher risk was observed among blue-collar men.
Johnson JV, Hall EM, Stewart W, Fredlund P, Theorell T. Combined exposure to adverse work organization factors and cardiovascular disease: Towards a life-course perspective. In Fechter L, ed. Proceedings of the Fourth International Conference on the Combined Effects of Environmental Factors, Baltimore: Johns Hopkins University, 1991:117-121 .
Johnson JV, Stewart W. Measuring work organization exposure over the life course with a job-exposure matrix. Scand J Work Environ Health 1993;19:21-28.
Landsbergis PA, Schnall PL, Schwartz JE, Warren K, Pickering TG. Job strain, hypertension and cardiovascular disease. In Organizational Risk Factors for Job Stress, eds. SL Sauter, LR Murphy. Washington, DC: American Psychological Association 1995:97-112.
Schnall PL, Landsbergis PA, Baker D. Job strain and cardiovascular disease. Ann Rev Public Health 1994;15:381-411.
(All reference numbers are from Schnall PL, Landsbergis PA, Baker D. Job strain and cardiovascular disease. Annual Review of Public Health 1994;15:381-411.)
Self report bias is a potential problem in many “job strain” studies, since exposure has often been assessed through questionnaires completed by study participants. Self-reports may be inaccurate descriptions of job characteristics or may be biased by personality traits such as “negative affectivity”. Concerns have also been raised about the need for more objective measures of “job strain” in intervention studies.
Therefore, in 13 “job strain” studies, researchers employed an analytic technique to overcom self-report bias and obtain more objective measures of job characteristics – the imputation of average scores for a particular job title to individuals in that job title. The average job-title score, free of the individual’s subjective assessment, then predicts outcome for the individual. However, while this strategy is often presented as desirable, it developed in the U.S.A.. because of a lack of databases containing both job characteristics data and health data – a weakness of past research. Large within variance exists in job characteristics (55% of reliable variance for latitude, and 93% for demands), since job titles such as nurse, machinist, secretary or teacher are somewhat heterogenous in skill levels, autonomy, or demands (102). As a result, in the U.S. studies, mean scores of job characteristics are adjusted for demographic covariates (e.g., age, race, education, marital status, region, urban vs. rural, and self-employment status) in the HANES 1 (59) when imputed to an individual participant (102). Despite this adjustment, the imputation strategy introduces (non-differential) misclassification and a bias towards the null. Thus, positive findings using the imputation method (4, 5, 36, 59, 68, 81, 109) provide strong support for the model, while negative studies may result, in part, from loss of power. However, individual level job data and health data clearly need to be obtained in future research.