MacArthur SES & Health Network
MacArthur SES & Health Network


printable version

Evidence from studies using the demand-control model

The importance of work related psychosocial factors to the development of ill health and disease can be illustrated from the Whitehall studies of British civil servants. The finding of dramatic differences in mortality by grade of employment in the first Whitehall study which could not be explained by conventional risk factors alone (Marmot et al, 1984) led to the initiation of a second longitudinal study of civil servants - the Whitehall II study of 10,308 male and female civil servants. A major aim of this second study has been to investigate occupational and other social influences on health and disease in a white-collar office based population. In pursuing the work environment as providing possible explanations, we have examined characteristics of the demand-control model. One hypothesis is that the lower the grade of employment in the civil service, the lower the level of control over the job, the lower the use of skills and the higher the level of monotony. These may be related to the higher rate of cardiovascular and other diseases in lower employment grades. Our initial analyses of the psychosocial work environment confirmed the above, with men and women in lower employment grades reporting lower levels of control, less varied work and use of skills, and a slower pace of work. Overall, fewer of the lower grades expressed themselves as satisfied with their work situation (Table 1) (Marmot et al, 1991).

Table 1 - Psychosocial Work Characteristics by Grade of Employment in the Whitehall II study of British Civil Servants (age-adjusted figures)

Employment Grade+

  Sex 1 2 3
High Control (%) M 59.3 49.7 43.1
  F 51.2 45.4 47.1
Varied work (%) M 70.5 52.1 41.9
  F 71.2 55.2 40.5
Fast Pace (%) M 58.0 43.6 34.7
  F 60.9 50.3 43.7
High satisfaction (%) M 58.2 38.7 34.1
  F 57.5 42.2 40.3


  Sex 4 5 6
High Control (%) M 31.6 24.7 11.8
  F 31.2 20.1 10.2
Varied work (%) M 27.1 18.2 3.9
  F 31.7 14.0 4.7
Fast Pace (%) M 27.9 20.8 15.8
  F 31.1 29.7 18.0
High satisfaction (%) M 29.5 29.4 29.8
  F 36.6 41.6 47.7


  Sex Total Sample Test for Trend
High Control (%) M 6877 ***
  F 3341 ***
Varied work (%) M 6875 ***
  F 3356 ***
Fast Pace (%) M 6878 ***
  F 3356 ***
High satisfaction (%) M 6865 ***
  F 3337 ns
P-values: nsp>0.10 and *** P<=0.001

Grade categories+

Grade 1 - Unified Grades 1-6
Grade 2 - Unified Grade 7
Grade 3 - Senior Executive Officer } and
Grade 4 - Higher Executive Officer } professional
Grade 5 - Executive Officer } equivalents
Grade 6 - Clerical Officer / Office Support


The Whitehall II study has been studying psychosocial factors in relation to a range of health problems such as sickness absence, musculoskeletal and psychiatric disorder and CHD. In analysing these health problems, the crucial task has been to separate the effects of work from those of other influences on health. In the analyses that follow we present data from the Whitehall II study examining the contribution of the psychosocial work environment to explaining the social gradient in sickness absence and CHD, controlling for other potential confounding factors.

Psychosocial work characteristics and sickness absence

We chose to analyse sickness absence as a measure of morbidity for a number of reasons. First, in the original Whitehall study, grade of employment was associated with mortality from a range of specific causes (Marmot et al, 1984). This suggested the possibility that, in addition to searching out the determinants of specific medical diagnoses, it was appropriate to search for determinants of general susceptibility to illness. Second, we take the view that ill health is important not only because it may hasten the time of death but because it interferes with social, psychological and physical functioning during life. One way of looking at sickness absence is that it is a measure that integrates decrements in social, psychological and physical functioning. Short spells of absence are more likely to represent decrements in psychological and social functioning, long spells are more likely to represent decrements in physical functioning or ‘real illness’. Third, sickness absence is a measure of great economic importance to employers. Studies of the determinants of sickness absence may therefore be of interest not only to those whose primary interest is in the aetiology of illness, but to those interested in the health of the economy and of individual firms (Marmot and Feeney 1996).

There was a clear association between grade of employment and sickness absence. Men in the lowest grade had six times the absence rate of men in the highest grade for both short (< 7 days) and long (>= 7 days) spells of absence. Women showed a similar although slightly reduced gradient. As might be expected, the worse people rated their own health the higher the sickness absence rates; sickness absence was also related to individual characteristics such as smoking and to problems outside work including financial problems and inadequate support (North et al, 1993).

Characteristics of the psychosocial work environment were also related to sickness absence. Men and women who rated their jobs as low on control, low on variety and use of skills, reported low support at work, and a slow pace of work had higher rates of short and long spells of sickness absence compared with those who rated their jobs high on these characteristics (North et al, 1993). Psychosocial work characteristics were also associated with sickness absence for psychiatric disorder and back pain. Low variety and use of skills and low support from colleagues and supervisors were associated with higher rates of short spells for psychiatric reasons in men and women (Stansfeld et al, 1997), and low control showed the most consistent effect predicting both short and long spells of sickness absence for back pain in men and women (Hemingway et al, 1997).

One question we asked was how much of a contribution did work and other characteristics make to generating the social gradient in ill health as measured by sickness absence? Figure 1 shows long spells of sickness absence by grade, adjusted first for age and then for other predictors of sickness absence, including psychosocial work characteristics. This analysis suggests that about 25% of the social gradient in men and about 35% of the gradient in women is accounted for by these characteristics (North et al, 1993).

Figure 1. Long spells of sickness absence by grade. Men-Whitehall II study.


The above analysis relates to individual characteristics of the psychosocial work environment. We also examined the job strain model to see if this predicted sickness absence in the Whitehall II study. We found partial support for the model, in that men who reported high levels of work demands and low levels of control had higher rates of short spells of sickness absence than those in other jobs; the results were similar for women, but were non-significant. Although predicting short spells, the job strain model did not predict long spells of sickness absence for either men or women. In examining psychosocial aspects of the work environment, it is important to reduce the problem of confounding. Employment grade is strongly related to several work, health and personal characteristics, controlling for grade is therefore a way of controlling for many of the social and personal factors related to socio-economic status. We therefore examined the job strain model stratified by employment grade, and after adjusting for grade found there was minimal support for the job strain model in the whole study, but within the lower grades there was stronger support, with jobs characterised by high work demands and low control predicting sickness absence (North et al, 1996).


Psychosocial work characteristics and CHD


The design of the Whitehall II study is longitudinal and this has enabled us to assess the psychosocial work environment over a period of time and examine its relationship to the development of new CHD. In addition to self-reported measures, we have also been able to use independent measures of the psychosocial work environment to address the question of whether job stress is influenced by subjective perceptions or by more objective appraisals of the work, or by both. Our results show that both men and women with low control, either self-reported or independently assessed, had a higher risk of newly reported CHD during a mean follow-up period of five years (Figure 2). This association could not be explained by employment grade, negative affectivity, or classical coronary risk factors. We also examined the job strain model, but found that job demands and social supports and the interactions between work characteristics were not related to the risk of new CHD. Specific characteristics of our sample of white collar workers may have contributed to this negative finding, high job demands were more common in higher employment grades, and high job demands and high job control were positively associated, resulting in comparatively few high strain jobs (Bosma et al, 1997).

Figure 2. Five year CHD incidence by self-reported and external job control. Men and women-Whitehall II study.


In addition to Whitehall II, other studies have looked at the association of characteristics of the work environment to heart disease. The Swedish case control study of over 2,000 men and women in Stockholm (SHEEP) has investigated the role of psychosocial and other factors in the development of myocardial infarction. Men who reported high demands and low control in their job were at greater risk of developing a myocardial infarction, this relationship was more pronounced for manual workers (Hallqvist et al, 1998). An investigation of the psychosocial work environment in the ten years preceding myocardial infarction also showed that a decrease in the amount of control at work was associated with an increased risk of myocardial infarction, again this effect was stronger in manual workers and in men under 55 years of age (Theorell et al, 1998). Data from both the SHEEP and Whitehall II studies regarding loss of control and its possible effect on future risk of myocardial infarction illustrate that the increased risk does not develop rapidly after reported loss of control but increases gradually over time. Thus, there may be a possibility of preventing myocardial infarctions if individuals at risk can be identified within a specified time period.

The differing associations between aspects of the psychosocial work environment and CHD correspond to the review by Schnall and Landsbergis (1994), in which 17 out of 25 studies found significant associations between job control and cardiovascular outcome, whereas associations with job demands were significant in only eight out of 23 studies. A further review of the role of psychosocial work characteristics and CHD has recently been conducted (Hemingway and Marmot 1998). This review used a quality filter to identify the best available evidence. The filter included: i) a prospective population based design; ii) at least 500 participants (aetiological studies in healthy populations) or 100 participants (prognostic studies in CHD patient populations); iii) instruments for exposure measurement used in two or more study populations; and iv) fatal or validated non-fatal CHD as outcomes. Table 2 shows that six of the ten studies showed a positive association between aspects of the job strain model and CHD. The negative associations were possibly due to factors specific to the populations studied. For example, the studies by Reed et al, (1989) and Suadicani et a,. (1993) studied men who were relatively old, particularly at the end of the follow-up period when a large proportion had retired. The study by Hlatky et al, (1995) was not a representative sample and there may have been other selection factors in operation in determining who underwent coronary angiography. Finally, the bus drivers studied in Netterstrom and Suadicani (1993) may have more reason than other workers to deny difficulties at work. Other studies using observational techniques instead of self-reports have confirmed this possibility (Greiner et al, 1997). It is likely that these results will influence the subsequent use and development of this theoretical model in future research studies. Improved measurement of the psychosocial work environment will lead to further methodological refinement of the model, and in particular for job demands, the issue of interaction within the job strain model warrants further investigation (Hallqvist et al, 1998).

close window

 

MACSES
UCSF Home About UCSF Search UCSF UCSF Medical Center