Introduction
Over the past 30 years, numerous
studies have suggested that job stress,
defined in different ways, is associated
with an increased risk of adverse health
and behavioral outComes.1-7 Earlier studies
of coronary heart disease and hypertension
defined job stress in terms of excessive
workload and pressures.8"' In
contrast, studies of job satisfaction, work
performance, and sickness absence have
focused on different work characteristics,
such as autonomy, variety, skill use, and
responsibility for completing tasks."-'5 In
1979, Karasek proposed the "job strain
model," which integrated these two approaches.'6
A number of studies have provided
empirical support for the job strain model
by suggesting that individuals in occupations
with high work demands and low
control are at an increased risk for
physical and psychological symptoms,
coronary heart disease and its concomitant
risk factors, sickness absence, job and
life dissatisfaction, and medication use.1-23
Johnson and Hall modified the job strain
model by suggesting that lack of support
at work combines with high work demands
and low control to increase strain
and adverse health outcomes.24'25
In this paper, we address three issues
raised by recent research. The first relates
to how the work environment has been
measured. Some studies have used employees'
reports of their work, which may
relate as much to personal factors as to
the work environment itself; others have
used some form of extemal assessment
that may overlook the importance of the
employees' perception of their work environment.
Second, most studies have been
restricted to men, predominantly those in
production, distribution, and service occupations.
It is therefore not clear whether
the job strain model also applies to
women or to office and professional
occupations. Third, most studies have
taken inadequate account of the role of
socioeconomic status (SES). However,
the relationship between SES, the work
environment, and health is likely to be
complex. Adjusting for SES may result in
an underestimation of the effect of the
work environment if the latter were to
partially mediate between SES and health.
Alternatively, not adjusting for SES may
result in an overestimation of the effect of
work on health. Finally, if inadequate
measures of SES are used, any adjustment
would be incomplete.
In the ongoing Whitehall II study of
British civil servants, we examine the
relationship between the psychosocial
work environment and a general measure
of health: sickness absence. Employees
and personnel managers provide ratings
of the work environment; the study
population includes both men and women
in office and professional occupations;
and the potential confounding effect of
SES is examined by adjusting for both
grade of employment and separately