34. Psychosocial and Organizational Factors
Chapter Editors: Steven L. Sauter, Lawrence R. Murphy, Joseph J. Hurrell and Lennart Levi
Psychosocial and Organizational Factors
Steven L. Sauter, Joseph J. Hurrell Jr., Lawrence R. Murphy and Lennart Levi
Psychosocial Factors, Stress and Health
Social Support: An Interactive Stress Model
Person - Environment Fit
Robert D. Caplan
Hours of Work
Timothy H. Monk
Michael J. Smith
Autonomy and Control
Electronic Work Monitoring
Lawrence M. Schleifer
Role Clarity and Role Overload
Steve M. Jex
Chaya S. Piotrkowski
Job Future Ambiguity
John M. Ivancevich
Amiram D. Vinokur
Total Quality Management
Cary L. Cooper and Mike Smith
Lois E. Tetrick
Organizational Climate and Culture
Denise M. Rousseau
Performance Measures and Compensation
Richard L. Shell
Marilyn K. Gowing
Debra L. Nelson and James Campbell Quick
Type A/B Behaviour Pattern
C. David Jenkins
Suzanne C. Ouellette
John M. Schaubroeck
Locus of Control
Lawrence R. Murphy and Joseph J. Hurrell, Jr.
Ronald J. Burke
D. Wayne Corneil
Gender, Job Stress and Illness
Rosalind C. Barnett
Gwendolyn Puryear Keita
Selected Acute Physiological Outcomes
Andrew Steptoe and Tessa M. Pollard
Töres Theorell and Jeffrey V. Johnson
Bernard H. Fox
Soo-Yee Lim, Steven L. Sauter and Naomi G. Swanson
Carles Muntaner and William W. Eaton
Summary of Generic Prevention and Control Strategies
Cary L. Cooper and Sue Cartwright
Click a link below to view table in the article context.
Point to a thumbnail to see figure caption, click to see figure in article context.
Locus of control (LOC) refers to a personality trait reflecting the generalized belief that either events in life are controlled by one’s own actions (an internal LOC) or by outside influences (an external LOC). Those with an internal LOC believe that they can exert control over life events and circumstances, including the associated reinforcements, that is, those outcomes which are perceived to reward one’s behaviours and attitudes. In contrast, those with an external LOC believe they have little control over life events and circumstances, and attribute reinforcements to powerful others or to luck.
The construct of locus of control emerged from Rotter’s (1954) social learning theory. To measure LOC, Rotter (1966) developed the Internal-External (I-E) scale, which has been the instrument of choice in most research studies. However, research has questioned the unidimensionality of the I-E scale, with some authors suggesting that LOC has two dimensions (e.g., personal control and social system control), and others suggesting that LOC has three dimensions (personal efficacy, control ideology and political control). More recently developed scales to measure LOC are multidimensional, or assess LOC for specific domains, such as health or work (Hurrell and Murphy 1992).
One of the most consistent and widespread findings in the general research literature is the association between an external LOC and poor physical and mental health (Ganster and Fusilier 1989). A number of studies in occupational settings report similar findings: workers with an external LOC tended to report more burnout, job dissatisfaction, stress and lower self-esteem than those with an internal LOC (Kasl 1989). Recent evidence suggests that LOC moderates the relationship between role stressors (role ambiguity and role conflict) and symptoms of distress (Cvetanovski and Jex 1994; Spector and O’Connell 1994).
However, research linking LOC beliefs and ill health is difficult to interpret for several reasons (Kasl 1989). First, there may be conceptual overlap between the measures of health and locus of control scales. Secondly, a dispositional factor, like negative affectivity, may be present which is responsible for the relationship. For example, in the study by Spector and O’Connell (1994), LOC beliefs correlated more strongly with negative affectivity than with perceived autonomy at work, and did not correlate with physical health symptoms. Thirdly, the direction of causality is ambiguous; it is possible that the work experience may alter LOC beliefs. Finally, other studies have not found moderating effects of LOC on job stressors or health outcomes (Hurrell and Murphy 1992).
The question of how LOC moderates job stressor-health relationships has not been well researched. One proposed mechanism involves the use of more effective, problem-focused coping behaviour by those with an internal LOC. Those with an external LOC might use fewer problem-solving coping strategies because they believe that events in their lives are outside their control. There is evidence that people with an internal LOC utilize more task-centred coping behaviours and fewer emotion-centred coping behaviours than those with an external LOC (Hurrell and Murphy 1992). Other evidence indicates that in situations viewed as changeable, those with an internal LOC reported high levels of problem-solving coping and low levels of emotional suppression, whereas those with an external LOC showed the reverse pattern. It is important to bear in mind that many workplace stressors are not under the direct control of the worker, and that attempts to change uncontrollable stressors might actually increase stress symptoms (Hurrell and Murphy 1992).
A second mechanism whereby LOC could influence stressor-health relationships is via social support, another moderating factor of stress and health relationships. Fusilier, Ganster and Mays (1987) found that locus of control and social support jointly determined how workers responded to job stressors and Cummins (1989) found that social support buffered the effects of job stress, but only for those with an internal LOC and only when the support was work-related.
Although the topic of LOC is intriguing and has stimulated a great deal of research, there are serious methodological problems attaching to investigations in this area which need to be addressed. For example, the trait-like (unchanging) nature of LOC beliefs has been questioned by research which showed that people adopt a more external orientation with advancing age and after certain life experiences such as unemployment. Furthermore, LOC may be measuring worker perceptions of job control, instead of an enduring trait of the worker. Still other studies have suggested that LOC scales may not only measure beliefs about control, but also the tendency to use defensive manoeuvres, and to display anxiety or proneness to Type A behaviour (Hurrell and Murphy 1992).
Finally, there has been little research on the influence of LOC on vocational choice, and the reciprocal effects of LOC and job perceptions. Regarding the former, occupational differences in the proportion of “internals” and “externals” may be evidence that LOC influences vocational choice (Hurrell and Murphy 1992). On the other hand, such differences might reflect exposure to the job environment, just as the work environment is thought to be instrumental in the development of the Type A behaviour pattern. A final alternative is that occupational differences in LOC are be due to “drift”, that is the movement of workers into or out of certain occupations as a result of job dissatisfaction, health concerns or desire for advancement.
In summary, the research literature does not present a clear picture of the influence of LOC beliefs on job stressor or health relationships. Even where research has produced more or less consistent findings, the meaning of the relationship is obscured by confounding influences (Kasl 1989). Additional research is needed to determine the stability of the LOC construct and to identify the mechanisms or pathways through which LOC influences worker perceptions and mental and physical health. Components of the path should reflect the interaction of LOC with other traits of the worker, and the interaction of LOC beliefs with work environment factors, including reciprocal effects of the work environment and LOC beliefs. Future research should produce less ambiguous results if it incorporates measures of related individual traits (e.g., Type A behaviour or anxiety) and utilizes domain-specific measures of locus of control (e.g., work).
Coping has been defined as “efforts to reduce the negative impacts of stress on individual well-being” (Edwards 1988). Coping, like the experience of work stress itself, is a complex, dynamic process. Coping efforts are triggered by the appraisal of situations as threatening, harmful or anxiety producing (i.e., by the experience of stress). Coping is an individual difference variable that moderates the stress-outcome relationship.
Coping styles encompass trait-like combinations of thoughts, beliefs and behaviours that result from the experience of stress and may be expressed independently of the type of stressor. A coping style is a dispositional variable. Coping styles are fairly stable over time and situations and are influenced by personality traits, but are different from them. The distinction between the two is one of generality or level of abstraction. Examples of such styles, expressed in broad terms, include: monitor-blunter (Miller 1979) and repressor-sensitizer (Houston and Hodges 1970). Individual differences in personality, age, experience, gender, intellectual ability and cognitive style affect the way an individual copes with stress. Coping styles are the result of both prior experience and previous learning.
Shanan (1967) offered an early perspective on what he termed an adaptive coping style. This “response set” was characterized by four ingredients: the availability of energy directly focused on potential sources of the difficulty; a clear distinction between events internal and external to the person; confronting rather than avoiding external difficulties; and balancing external demands with needs of the self. Antonovsky (1987) similarly suggests that, to be effective, the individual person must be motivated to cope, have clarified the nature and dimensions of the problem and the reality in which it exists, and then selected the most appropriate resources for the problem at hand.
The most common typology of coping style (Lazarus and Folkman 1984) includes problem-focused coping (which includes information seeking and problem solving) and emotion-focused coping (which involves expressing emotion and regulating emotions). These two factors are sometimes complemented by a third factor, appraisal-focused coping (whose components include denial, acceptance, social comparison, redefinition and logical analysis).
Moos and Billings (1982) distinguish among the following coping styles:
Greenglass (1993) has recently proposed a coping style termed social coping, which integrates social and interpersonal factors with cognitive factors. Her research showed significant relationships between various kinds of social support and coping forms (e.g., problem-focused and emotion-focused). Women, generally possessing relatively greater interpersonal competence, were found to make greater use of social coping.
In addition, it may be possible to link another approach to coping, termed preventive coping, with a large body of previously separate writing dealing with healthy lifestyles (Roskies 1991). Wong and Reker (1984) suggest that a preventive coping style is aimed at promoting one’s well-being and reducing the likelihood of future problems. Preventive coping includes such activities as physical exercise and relaxation, as well as the development of appropriate sleeping and eating habits, and planning, time management and social support skills.
Another coping style, which has been described as a broad aspect of personality (Watson and Clark 1984), involves the concepts of negative affectivity (NA) and positive affectivity (PA). People with high NA accentuate the negative in evaluating themselves, other people and their environment in general and reflect higher levels of distress. Those with high PA focus on the positives in evaluating themselves, other people and their world in general. People with high PA report lower levels of distress.
These two dispositions can affect a person’s perceptions of the number and magnitude of potential stressors as well as his or her coping responses (i.e., one’s perceptions of the resources that one has available, as well as the actual coping strategies that are used). Thus, those with high NA will report fewer resources available and are more likely to use ineffective (defeatist) strategies (such as releasing emotions, avoidance and disengagement in coping) and less likely to use more effective strategies (such as direct action and cognitive reframing). Individuals with high PA would be more confident in their coping resources and use more productive coping strategies.
Antonovsky’s (1979; 1987) sense of coherence (SOC) concept overlaps considerably with PA. He defines SOC as a generalized view of the world as meaningful and comprehensible. This orientation allows the person to first focus on the specific situation and then to act on the problem and the emotions associated with the problem. High SOC individuals have the motivation and the cognitive resources to engage in these sorts of behaviours likely to resolve the problem. In addition, high SOC individuals are more likely to realize the importance of emotions, more likely to experience particular emotions and to regulate them, and more likely to take responsibility for their circumstances instead of blaming others or projecting their perceptions upon them. Considerable research has since supplied support for Antonovsky’s thesis.
Coping styles can be described with reference to dimensions of complexity and flexibility (Lazarus and Folkman 1984). People using a variety of strategies exhibit a complex style; those preferring a single strategy exhibit a single style. Those who use the same strategy in all situations exhibit a rigid style; those who use different strategies in the same, or different, situations exhibit a flexible style. A flexible style has been shown to be more effective than a rigid style.
Coping styles are typically measured by using self-reported questionnaires or by asking individuals, in an open-ended way, how they coped with a particular stressor. The questionnaire developed by Lazarus and Folkman (1984), the “Ways of Coping Checklist”, is the most widely used measure of problem-focused and emotion-focused coping. Dewe (1989), on the other hand, has frequently used individuals’ descriptions of their own coping initiatives in his research on coping styles.
There are a variety of practical interventions that may be implemented with regard to coping styles. Most often, intervention consists of education and training in which individuals are presented with information, sometimes coupled with self-assessment exercises that enable them to examine their own preferred coping style as well as other varieties of coping styles and their potential usefulness. Such information is typically well received by the persons to whom the intervention is directed, but the demonstrated usefulness of such information in helping them cope with real life stressors is lacking. In fact, the few studies that considered individual coping (Shinn et al. 1984; Ganster et al. 1982) have reported limited practical value in such education, particularly when a follow-up has been undertaken (Murphy 1988).
Matteson and Ivancevich (1987) outline a study dealing with coping styles as part of a longer programme of stress management training. Improvements in three coping skills are addressed: cognitive, interpersonal and problem solving. Coping skills are classified as problem-focused or emotion-focused. Problem-focused skills include problem solving, time management, communication and social skills, assertiveness, lifestyle changes and direct actions to change environmental demands. Emotion-focused skills are designed to relieve distress and foster emotion regulation. These include denial, expressing feelings and relaxation.
The preparation of this article was supported in part by the Faculty of Administrative Studies, York University.
During the mid-1970s public health practitioners, and in particular, epidemiologists “discovered” the concept of social support in their studies of causal relationships between stress, mortality and morbidity (Cassel 1974; Cobb 1976). In the past decade there has been an explosion in the literature relating the concept of social support to work-related stressors. By contrast, in psychology, social support as a concept had already been well integrated into clinical practice. Rogers’ (1942) client-centred therapy of unconditional positive regard is fundamentally a social support approach. Lindeman’s (1944) pioneering work on grief management identified the critical role of support in moderating the crisis of death loss. Caplin’s (1964) model of preventive community psychiatry (1964) elaborated on the importance of community and support groups.
Cassel (1976) adapted the concept of social support into public health theory as a way of explaining the differences in diseases that were thought to be stress-related. He was interested in understanding why some individuals appeared to be more resistant to stress than others. The idea of social support as a factor in disease causation was reasonable since, he noted, both people and animals who experienced stress in the company of “significant others” seemed to suffer fewer adverse consequences than those who were isolated. Cassel proposed that social support could act as a protective factor buffering an individual from the effects of stress.
Cobb (1976) expanded on the concept by noting that the mere presence of another person is not social support. He suggested that an exchange of “information” was needed. He established three categories for this exchange:
Cobb reported that those experiencing severe events without such social support were ten times more likely to come to be depressed and concluded that somehow intimate relations, or social support, was protective of the effects of stress reactions. He also proposed that social support operates throughout one’s life span, encompassing various life events such as unemployment, severe illness and bereavement. Cobb pointed out the great diversity of studies, samples, methods and outcomes as convincing evidence that social support is a common factor in modifying stress, but is, in itself, not a panacea for avoiding its effects.
According to Cobb, social support increases coping ability (environmental manipulation) and facilitates adaptation (self-change to improve the person-environment fit). He cautioned, however, that most research was focused on acute stressors and did not permit generalizations of the protective nature of social support for coping with the effects of chronic stressors or traumatic stress.
Over the intervening years since the publication of these seminal works, investigators have moved away from considering social support as a unitary concept, and have attempted to understand the components of social stress and social support.
Hirsh (1980) describes five possible elements of social support:
House felt that emotional support was the most important form of social support. In the workplace, the supportiveness of the supervisor was the most important element, followed by co-worker support. The structure and organization of the enterprise, as well as the specific jobs within it, could either enhance or inhibit potential for support. House found that greater task specialization and fragmentation of work leads to more isolated work roles and to decreased opportunities for support.
Pines’ (1983) study of burnout, which is a phenomenon discussed separately in this chapter, found that the availability of social support at work is negatively correlated with burnout. He identifies six different relevant aspects of social support which modify the burnout response. These include listening, encouragement, giving advice and, providing companionship and tangible aid.
As one may gather from the foregoing discussion in which the models proposed by several researchers have been described, while the field has attempted to specify the concept of social support, there is no clear consensus on the precise elements of the concept, although considerable overlap between models is evident.
Interaction between Stress and Social Support
Although the literature on stress and social support is quite extensive, there is still considerable debate as to the mechanisms by which stress and social support interact. A long-standing question is whether social support has a direct or indirect effect on health.
Main effect/Direct effect
Social support can have a direct or main effect by serving as a barrier to the effects of the stressor. A social support network may provide needed information or needed feedback in order to overcome the stressor. It may provide a person with the resources he or she needs to minimize the stress. An individual’s self-perception may also be influenced by group membership so as to provide self-confidence, a sense of mastery and skill and hence thereby a sense of control over the environment. This is relevant to Bandura’s (1986) theories of personal control as the mediator of stress effects. There appears to be a minimum threshold level of social contact required for good health, and increases in social support above the minimum are less important. If one considers social support as having a direct—or main—effect, then one can create an index by which to measure it (Cohen and Syme 1985; Gottlieb 1983).
Cohen and Syme (1985), however, also suggest that an alternative explanation to social support acting as a main effect is that it is the isolation, or lack of social support, which causes the ill health rather than the social support itself promoting better health. This is an unresolved issue. Gottlieb also raises the issue of what happens when the stress results in the loss of the social network itself, such as might occur during disasters, major accidents or loss of work. This effect has not yet been quantified.
The buffering hypothesis is that social support intervenes between the stressor and the stress response to reduce its effects. Buffering could change one’s perception of the stressor, thus diminishing its potency, or it could increases one’s coping skills. Social support from others may provide tangible aid in a crisis, or it may lead to suggestions that facilitate adaptive responses. Finally, social support may be the stress-modifying effect which calms the neuroendocrine system so that the person may be less reactive to the stressor.
Pines (1983) notes that the relevant aspect of social support may be in the sharing of a social reality. Gottlieb proposes that social support could offset self-recrimination and dispel notions that the individual is him or herself responsible for the problems. Interaction with a social support system can encourage the venting of fears and can assist re-establishing a meaningful social identity.
Additional Theoretical Issues
Research thus far has tended to treat social support as a static, given factor. While the issue of its change over time has been raised, little data exist on the time course of social support (Gottlieb 1983; Cohen and Syme 1985). Social support is, of course, fluid, just as the stressors that it affects. It varies as the individual passes through the stages of life. It can also change over the short-term experience of a particular stressful event (Wilcox 1981).
Such variability probably means that social support fulfils different functions during different developmental stages or during different phases of a crisis. For example at the onset of a crisis, informational support may be more essential than tangible aid. The source of support, its density and the length of time it is operative will also be in flux. The reciprocal relationship between stress and social support must be recognized. Some stressors themselves have a direct impact on available support. Death of a spouse, for example, usually reduces the extent of the network and may have serious consequences for the survivor (Goldberg et al. 1985).
Social support is not a magic bullet that reduces the impact of stress. Under certain conditions it may exacerbate or be the cause of stress. Wilcox (1981) noted that those with a denser kin network had more difficulties adjusting to divorce because their families were less likely to accept divorce as a solution to marital problems. The literature on addiction and family violence also shows possible severe negative effects of social networks. Indeed, as Pines and Aronson (1981) point out, much of professional mental health interventions are devoted to undoing destructive relationships, and to teaching interpersonal skills and to assisting people to recover from social rejection.
There are a large number of studies employing a variety of measures of the functional content of social support. These measures have a wide range of reliability and construct validity. Another methodological problem is that these analyses depend largely on the self-reports of those being studied. The responses will therefore of necessity be subjective and will cause one to wonder whether it is the actual event or level of social support that is important or whether it is the individual’s perception of support and outcomes that is more critical. If it is the perception that is critical, then it may be that some other, third variable, such as personality type, is affecting both stress and social support (Turner 1983). For example, a third factor, such as age or socio-economic status, may influence change in both social support and outcome, according to Dooley (1985). Solomon (1986) provides some evidence for this idea with a study of women who have been forced by financial constraints into involuntary interdependence on friends and kin. She found that such women opt out of these relationships as quickly as they are financially able to do so.
Thoits (1982) raises concerns about reverse causation. It may be, she points out, that certain disorders chase away friends and lead to loss of support. Studies by Peters-Golden (1982) and Maher (1982) on cancer victims and social support appear to be consistent with this proposition.
Social Support and Work Stress
Studies on the relationship between social support and work stress indicate that successful coping is related to the effective use of support systems (Cohen and Ahearn 1980). Successful coping activities have emphasized the use of both formal and informal social support in dealing with work stress. Laid-off workers, for example, are advised to actively seek support to provide informational, emotional and tangible support. There have been relatively few evaluations of the effectiveness of such interventions. It appears, however, that formal support is only effective in the short term and informal systems are necessary for longer-term coping. Attempts to provide institutional formal social support can create negative outcomes, since the anger and rage about layoff or bankruptcy, for example, may be displaced to those who provide the social support. Prolonged reliance on social support may create a sense of dependency and lowered self- esteem.
In some occupations, such as seafarers, fire-fighters or staff in remote locations such as on oil rigs, there is a consistent, long-term, highly defined social network which can be compared to a family or kin system. Given the necessity for small work groups and joint efforts, it is natural that a strong sense of social cohesion and support develops among workers. The sometimes hazardous nature of the work requires that workers develop mutual respect, trust and confidence. Strong bonds and interdependence are created when people are dependent on each other for their survival and well-being.
Further research on the nature of social support during routine periods, as well as downsizing or major organizational change, is necessary to further define this factor. For example, when an employee is promoted to a supervisory position, he or she normally must distance him or herself from the other members of the work group. Does this make a difference in the day-to-day levels of social support he or she receives or requires? Does the source of support shift to other supervisors or to the family or somewhere else? Do those in positions of responsibility or authority experience different work stressors? Do these individuals require different types, sources or functions of social support?
If the target of the group-based interventions is also changing the functions of social support or the nature of the network, does this provide a preventive effect in future stressful events?
What will be the effect of growing numbers of women in these occupations? Does their presence change the nature and functions of support for all or does each sex require different levels or types of support?
The workplace presents a unique opportunity to study the intricate web of social support. As a closed subculture, it provides a natural experimental setting for research into the role of social support, social networks and their interrelationships with acute, cumulative and traumatic stress.
Do job stressors affect men and women differently? This question has only recently been addressed in the job stress–illness literature. In fact, the word gender does not even appear in the index of the first edition of the Handbook of Stress (Goldberger and Breznitz 1982) nor does it appear in the indices of such major reference books as Job Stress and Blue Collar Work (Cooper and Smith 1985) and Job Control and Worker Health (Sauter, Hurrell and Cooper 1989). Moreover, in a 1992 review of moderator variables and interaction effects in the occupational stress literature, gender effects were not even mentioned (Holt 1992). One reason for this state of affairs lies in the history of occupational health and safety psychology, which in turn reflects the pervasive gender stereotyping in our culture. With the exception of reproductive health, when researchers have looked at physical health outcomes and physical injuries, they have generally studied men and variations in their work. When researchers have studied mental health outcomes, they have generally studied women and variations in their social roles.
As a result, the “available evidence” on the physical health impact of work has until recently been almost completely limited to men (Hall 1992). For example, attempts to identify correlates of coronary heart disease have been focused exclusively on men and on aspects of their work; researchers did not even inquire into their male subjects’ marital or parental roles (Rosenman et al. 1975). Indeed, few studies of the job stress–illness relationship in men include assessments of their marital and parental relationships (Caplan et al. 1975).
In contrast, concern about reproductive health, fertility and pregnancy focused primarily on women. Not surprisingly, “the research on reproductive effects of occupational exposures is far more extensive on females than on males” (Walsh and Kelleher 1987). With respect to psychological distress, attempts to specify the psychosocial correlates, in particular the stressors associated with balancing work and family demands, have centred heavily on women.
By reinforcing the notion of “separate spheres” for men and women, these conceptualizations and the research paradigms they generated prevented any examination of gender effects, thereby effectively controlling for the influence of gender. Extensive sex segregation in the workplace (Bergman 1986; Reskin and Hartman 1986) also acts as a control, precluding the study of gender as a moderator. If all men are employed in “men’s jobs” and all women are employed in “women’s jobs”, it would not be reasonable to ask about the moderating effect of gender on the job stress–illness relationship: job conditions and gender would be confounded. It is only when some women are employed in jobs that men occupy and when some men are employed in jobs that women occupy that the question is meaningful.
Controlling is one of three strategies for treating the effects of gender. The other two are ignoring these effects or analysing them (Hall 1991). Most investigations of health have either ignored or controlled for gender, thereby accounting for the dearth of references to gender as discussed above and for a body of research that reinforces stereotyped views about the role of gender in the job stress–illness relationship. These views portray women as essentially different from men in ways that render them less robust in the workplace, and portray men as comparatively unaffected by non-workplace experiences.
In spite of this beginning, the situation is already changing. Witness the publication in 1987 of Gender and Stress (Barnett, Biener and Baruch 1987), the first edited volume focusing specifically on the impact of gender at all points in the stress reaction. And the second edition of the Handbook of Stress (Barnett 1992) includes a chapter on gender effects. Indeed, current studies increasingly reflect the third strategy: analysing gender effects. This strategy holds great promise, but also has pitfalls. Operationally, it involves analysing data relating to males and females and estimating both the main and the interaction effects of gender. A significant main effect tells us that after controlling for the other predictors in the model, men and women differ with respect to the level of the outcome variable. Interaction-effects analyses concern differential reactivity, that is, does the relationship between a given stressor and a health outcome differ for women and men?
The main promise of this line of inquiry is to challenge stereotyped views of women and men. The main pitfall is that conclusions about gender difference can still be drawn erroneously. Because gender is confounded with many other variables in our society, these variables have to be taken into account before conclusions about gender can be inferred. For example, samples of employed men and women will undoubtedly differ with respect to a host of work and non-work variables that could reasonably affect health outcomes. Most important among these contextual variables are occupational prestige, salary, part-time versus full-time employment, marital status, education, employment status of spouse, overall work burdens and responsibility for care of younger and older dependants. In addition, evidence suggests the existence of gender differences in several personality, cognitive, behavioural and social system variables that are related to health outcomes. These include: sensation seeking; self-efficacy (feelings of competence); external locus of control; emotion-focused versus problem-focused coping strategies; use of social resources and social support; harmful acquired risks, such as smoking and alcohol abuse; protective behaviours, such as exercise, balanced diets and preventive health regimens; early medical intervention; and social power (Walsh, Sorensen and Leonard, in press). The better one can control these contextual variables, the closer one can get to understanding the effect of gender per se on the relationships of interest, and thereby to understanding whether it is gender or other, gender-related variables that are the effective moderators.
To illustrate, in one study (Karasek 1990) job changes among white-collar workers were less likely to be associated with negative health outcomes if the changes resulted in increased job control. This finding was true for men, not women. Further analyses indicated that job control and gender were confounded. For women, one of “the less aggressive [or powerful] groups in the labour market” (Karasek 1990), white-collar job changes often involved reduced control, whereas for men, such job changes often involved increased control. Thus, power, not gender, accounted for this interaction effect. Such analyses lead us to refine the question about moderator effects. Do men and women react differentially to workplace stressors because of their inherent (i.e., biological) nature or because of their different experiences?
Although only a few studies have examined gender interaction effects, most report that when appropriate controls are utilized, the relationship between job conditions and physical or mental health outcomes is not affected by gender. (Lowe and Northcott 1988 describe one such study). In other words, there is no evidence of an inherent difference in reactivity.
Findings from a random sample of full-time employed men and women in dual-earner couples illustrates this conclusion with respect to psychological distress. In a series of cross-sectional and longitudinal analyses, a matched pairs design was used that controlled for such individual-level variables as age, education, occupational prestige and marital-role quality, and for such couple-level variables as parental status, years married and household income (Barnett et al. 1993; Barnett et al. 1995; Barnett, Brennan and Marshall 1994). Positive experiences on the job were associated with low distress; insufficient skill discretion and overload were associated with high distress; experiences in the roles of partner and parent moderated the relationship between job experiences and distress; and change over time in skill discretion and overload were each associated with change over time in psychological distress. In no case was the effect of gender significant. In other words, the magnitude of these relationships was not affected by gender.
One important exception is tokenism (see, for example, Yoder 1991). Whereas “it is clear and undeniable that there is a considerable advantage in being a member of the male minority in any female profession” (Kadushin 1976), the opposite is not true. Women who are in minority in a male work situation experience a considerable disadvantage. Such a difference is readily understandable in the context of men’s and women’s relative power and status in our culture.
Overall, studies of physical health outcomes also fail to reveal significant gender interaction effects. It appears, for example, that characteristics of work activity are stronger determinants of safety than are attributes of workers, and that women in traditionally male occupations suffer the same types of injury with approximately the same frequency as their male counterparts. Moreover, poorly designed protective equipment, not any inherent incapacity on the part of women in relation to the work, is often to blame when women in male-dominated jobs experience more injuries (Walsh, Sorensen and Leonard, 1995).
Two caveats are in order. First, no one study controls for all the gender-related covariates. Therefore, any conclusions about “gender” effects must be tentative. Secondly, because controls vary from study to study, comparisons between studies are difficult.
As increasing numbers of women enter the labour force and occupy jobs similar to those occupied by men, both the opportunity and the need for analysing the effect of gender on the job stress–illness relationship also increase. In addition, future research needs to refine the conceptualization and measurement of the stress construct to include job stressors important to women; extend interaction effects analyses to studies previously restricted to male or female samples, for example, studies of reproductive health and of stresses due to non-workplace variables; and examine the interaction effects of race and class as well as the joint interaction effects of gender x race and gender x class.
Major changes are taking place within the workforces of many of the world’s leading industrial nations, with members of ethnic minority groups making up increasingly larger proportions. However, little of the occupational stress research has focused on ethnic minority populations. The changing demographics of the world’s workforce give clear notice that these populations can no longer be ignored. This article briefly addresses some of the major issues of occupational stress in ethnic minority populations with a focus on the United States. However, much of the discussion should be generalizable to other nations of the world.
Much of the occupational stress research either excludes ethnic minorities, includes too few to allow meaningful comparisons or generalizations to be made, or does not report enough information about the sample to determine racial or ethnic participation. Many studies fail to make distinctions among ethnic minorities, treating them as one homogeneous group, thus minimizing the differences in demographic characteristics, culture, language and socio-economic status which have been documented both between and within ethnic minority groups (Olmedo and Parron 1981).
In addition to the failure to address issues of ethnicity, by far the greater part of research does not examine class or gender differences, or class by race and gender interactions. Moreover, little is known about the cross-cultural utility of many of the assessment procedures. Documentation used in such procedures is not adequately translated nor is there demonstrated equivalency between the standardized English and other language versions. Even when the reliabilities appear to indicate equivalence across ethnic or cultural groups, there is uncertainty about which symptoms in the scale are elicited in a reliable fashion, that is, whether the phenomenology of a disorder is similar across groups (Roberts, Vernon and Rhoades 1989).
Many assessment instruments inadequately assess conditions within ethnic minority populations; consequently results are often suspect. For example, many stress scales are based on models of stress as a function of undesirable change or readjustment. However, many minority individuals experience stress in large part as a function of ongoing undesirable situations such as poverty, economic marginality, inadequate housing, unemployment, crime and discrimination. These chronic stressors are not usually reflected in many of the stress scales. Models which conceptualize stress as resulting from the interplay between both chronic and acute stressors, and various internal and external mediating factors, are more appropriate for assessing stress in ethnic minority and poor populations (Watts-Jones 1990).
A major stressor affecting ethnic minorities is the prejudice and discrimination they encounter as a result of their minority status in a given society (Martin 1987; James 1994). It is a well- established fact that minority individuals experience more prejudice and discrimination as a result of their ethnic status than do members of the majority. They also perceive greater discrimination and fewer opportunities for advancement as compared with whites (Galinsky, Bond and Friedman 1993). Workers who feel discriminated against or who feel that there are fewer chances for advancement for people of their ethnic group are more likely to feel “burned out” in their jobs, care less about working hard and doing their jobs well, feel less loyal to their employers, are less satisfied with their jobs, take less initiative, feel less committed to helping their employers succeed and plan to leave their current employers sooner (Galinsky, Bond and Friedman 1993). Moreover, perceived prejudice and discrimination are positively correlated with self-reported health problems and higher blood pressure levels (James 1994).
An important focus of occupational stress research has been the relationship between social support and stress. However, there has been little attention paid to this variable with respect to ethnic minority populations. The available research tends to show conflicting results. For example, Hispanic workers who reported higher levels of social support had less job-related tension and fewer reported health problems (Gutierres, Saenz and Green 1994); ethnic minority workers with lower levels of emotional support were more likely to experience job burn-out, health symptoms, episodic job stress, chronic job stress and frustration; this relationship was strongest for women and for management as opposed to non-management personnel (Ford 1985). James (1994), however, did not find a significant relationship between social support and health outcomes in a sample of African-American workers.
Most models of job satisfaction have been derived and tested using samples of white workers. When ethnic minority groups have been included, they have tended to be African-Americans, and potential effects due to ethnicity were often masked (Tuch and Martin 1991). Research that is available on African-American employees tends to yield significantly lower scores on overall job satisfaction in comparison to whites (Weaver 1978, 1980; Staines and Quinn 1979; Tuch and Martin 1991). Examining this difference, Tuch and Martin (1991) noted that the factors determining job satisfaction were basically the same but that African-Americans were less likely to have the situations that led to job satisfaction. More specifically, extrinsic rewards increase African-Americans’ job satisfaction, but African-Americans are disadvantaged relatively to whites on these variables. On the other hand, blue-collar incumbency and urban residence decrease job satisfaction for African-Americans but African-Americans are overrepresented in these areas. Wright, King and Berg (1985) found that organizational variables (i.e., job authority, qualifications for the position and a sense that advancement within the organization is possible) were the best predictors of job satisfaction in their sample of black female managers in keeping with previous research on primarily white samples.
Ethnic minority workers are more likely than their white counterparts to be in jobs with hazardous work conditions. Bullard and Wright (1986/1987) noted this propensity and indicated that the population differences in injuries are likely to be the result of racial and ethnic disparities in income, education, type of employment and other socio-economic factors correlated with exposure to hazards. One of the most likely reasons, they noted, was that occupational injuries are highly dependent on the job and industry category of the workers and ethnic minorities tend to work in more hazardous occupations.
Foreign workers who have entered the country illegally often experience special work stress and maltreatment. They often endure substandard and unsafe working conditions and accept less than minimum wages because of fear of being reported to the immigration authorities and they have few options for better employment. Most health and safety regulations, guidelines for use, and warnings are in English and many immigrants, illegal or otherwise, may not have a good understanding of written or spoken English (Sanchez 1990).
Some areas of research have almost totally ignored ethnic minority populations. For example, hundreds of studies have examined the relationship between Type A behaviour and occupational stress. White males constitute the most frequently studied groups with ethnic minority men and women almost totally excluded. Available research—e.g., a study by Adams et al. (1986), using a sample of college freshmen, and e.g., Gamble and Matteson (1992), investigating black workers—indicates the same positive relationship between Type A behaviour and self-reported stress as that found for white samples.
Similarly, little research on issues such as job control and work demands is available for ethnic minority workers, although these are central constructs in occupational stress theory. Available research tends to show that these are important constructs for ethnic minority workers as well. For example, African-American licensed practical nurses (LPNs) report significantly less decision authority and more dead-end jobs (and hazard exposures) than do white LPNs and this difference is not a function of educational differences (Marshall and Barnett 1991); the presence of low decision latitude in the face of high demands tends to be the pattern most characteristic of jobs with low socio-economic status, which are more likely to be held by ethnic minority workers (Waitzman and Smith 1994); and middle- and upper-level white men rate their jobs consistently higher than their ethnic minority (and female) peers on six work design factors (Fernandez 1981).
Thus, it appears that many research questions remain regarding ethnic minority populations in the occupational stress and health arena as regards ethnic minority populations. These questions will not be answered until ethnic minority workers are included in study samples and in the development and validation of investigatory instruments.
Andrew Steptoe and Tessa M. Pollard
The acute physiological adjustments recorded during the performance of problem-solving or psychomotor tasks in the laboratory include: raised heart rate and blood pressure; alterations in cardiac output and peripheral vascular resistance; increased muscle tension and electrodermal (sweat gland) activity; disturbances in breathing pattern; and modifications in gastrointestinal activity and immune function. The best studied neurohormonal responses are those of the catecholamines (adrenaline and noradrenaline) and cortisol. Noradrenaline is the primary transmitter released by the nerves of the sympathetic branch of the autonomic nervous system. Adrenaline is released from the adrenal medulla following stimulation of the sympathetic nervous system, while activation of the pituitary gland by higher centres in the brain results in the release of cortisol from the adrenal cortex. These hormones support autonomic activation during stress and are responsible for other acute changes, such as stimulation of the processes that govern blood clotting, and the release of stored energy supplies from adipose tissue. It is likely that these kinds of response will also be seen during job stress, but studies in which work conditions are simulated, or in which people are tested in their normal jobs, are required to demonstrate such effects.
A variety of methods is available to monitor these responses. Conventional psychophysiological techniques are used to assess autonomic responses to demanding tasks (Cacioppo and Tassinary 1990). Levels of stress hormones can be measured in the blood or urine, or in the case of cortisol, in the saliva. The sympathetic activity associated with challenge has also been documented by measures of noradrenaline spillover from nerve terminals, and by direct recording of sympathetic nervous activity with miniature electrodes. The parasympathetic or vagal branch of the autonomic nervous system typically responds to task performance with reduced activity, and this can, under certain circumstances, be indexed through recording heart rate variability or sinus arrhythmia. In recent years, power spectrum analysis of heart rate and blood pressure signals has revealed wave bands that are characteristically associated with sympathetic and parasympathetic activity. Measures of the power in these wavebands can be used to index autonomic balance, and have shown a shift towards the sympathetic branch at the expense of the parasympathetic branch during task performance.
Few laboratory assessments of acute physiological responses have simulated work conditions directly. However, dimensions of task demand and performance that are relevant to work have been investigated. For example, as the demands of externally paced work increase (through faster pace or more complex problem solving), there is a rise in adrenaline level, heart rate and blood pressure, a reduction in heart rate variability and an increase in muscle tension. In comparison with self-paced tasks performed at the same rate, external pacing results in greater blood pressure and heart rate increases (Steptoe et al. 1993). In general, personal control over potentially stressful stimuli reduces autonomic and neuroendocrine activation in comparison with uncontrollable situations, although the effort of maintaining control over the situation itself has its own physiological costs.
Frankenhaeuser (1991) has suggested that adrenaline levels are raised when a person is mentally aroused or performing a demanding task, and that cortisol levels are raised when an individual is distressed or unhappy. Applying these ideas to job stress, Frankenhaeuser has proposed that job demand is likely to lead to increased effort and thus to raise levels of adrenaline, while lack of job control is one of the main causes of distress at work and is therefore likely to stimulate raised cortisol levels. Studies comparing levels of these hormones in people doing their normal work with levels in the same people at leisure have shown that adrenaline is normally raised when people are at work. Effects for noradrenaline are inconsistent and may depend on the amount of physical activity that people carry out during work and leisure time. It has also been shown that adrenaline levels at work correlate positively with levels of job demand. In contrast, cortisol levels have not been shown typically to be raised in people at work, and it is yet to be demonstrated that cortisol levels vary according to the degree of job control. In the “Air Traffic Controller Health Change Study”, only a small proportion of workers produced consistent increases in cortisol as the objective workload became greater (Rose and Fogg 1993).
Thus only adrenaline among the stress hormones has been shown conclusively to rise in people at work, and to do so according to the level of demand they experience. There is evidence that levels of prolactin increase in response to stress while levels of testosterone decrease. However, studies of these hormones in people at work are very limited. Acute changes in the concentration of cholesterol in the blood have also been observed with increased workload, but the results are not consistent (Niaura, Stoney and Herbst 1992).
As far as cardiovascular variables are concerned, it has repeatedly been found that blood pressure is higher in men and women during work than either after work or during equivalent times of day spent at leisure. These effects have been observed both with self-monitored blood pressure and with automated portable (or ambulatory) monitoring instruments. Blood pressure is especially high during periods of increased work demand (Rose and Fogg 1993). It has also been found that blood pressure rises with emotional demands, for example, in studies of paramedics attending the scenes of accidents. However, it is often difficult to determine whether blood pressure fluctuations at work are due to psychological demands or to associated physical activity and changes in posture. The raised blood pressure recorded at work is especially pronounced among people reporting high job strain according to the Demand-Control model (Schnall et al. 1990).
Heart rate has not been shown to be consistently raised during work. Acute elevations of heart rate may nevertheless be elicited by disruption of work, for example with breakdown of equipment. Emergency workers such as fire-fighters exhibit extremely fast heart rates in response to alarm signals at work. On the other hand, high levels of social support at work are associated with reduced heart rates. Abnormalities of cardiac rhythm may also be elicited by stressful working conditions, but the pathological significance of such responses has not been established.
Gastrointestinal problems are commonly reported in studies of job stress (see “Gastrointestinal problems” below). Unfortunately, it is difficult to assess the physiological systems underlying gastrointestinal symptoms in the work setting. Acute mental stress has variable effects on gastric acid secretion, stimulating large increases in some individuals and reduced output in others. Shift workers have a particularly high prevalence of gastrointestinal problems, and it has been suggested that these may arise when diurnal rhythms in the central nervous system’s control of gastric acid secretion are disrupted. Anomalies of small bowel motility have been recorded using radiotelemetry in patients diagnosed with irritable bowel syndrome while they go about their everyday lives. Health complaints, including gastrointestinal symptoms, have been shown to co-vary with perceived workload, but it is not clear whether this reflects objective changes in physiological function or patterns of symptom perception and reporting.
Researchers may disagree on the meaning of the term stress. However, there is a basic agreement that perceived work-related stress may be implicated in behavioural outcomes such as absenteeism, substance abuse, sleep disturbances, smoking and caffeine use (Kahn and Byosiere 1992). Recent evidence supporting these relationships is reviewed in this chapter. Emphasis is placed upon the aetiological role of work-related stress in each of these outcomes. There are qualitative differences, along several dimensions, among these outcomes. To illustrate, in contrast to the other behavioural outcomes, which are all considered problematic to the health of those engaging in them excessively, absenteeism, while detrimental to the organization, is not necessarily harmful to those employees who are absent from work. There are, however, common problems in the research on these outcomes, as discussed in this section.
The varying definitions of work-related stress have already been mentioned above. By way of illustration, consider the different conceptualizations of stress on the one hand as events and on the other as chronic demands at the workplace. These two approaches to stress measurement have seldom been combined in a single study designed to predict the sorts of behavioural outcome considered here. The same generalization is relevant to the combined use, in the same study, of family-related and work-related stress to predict any of these outcomes. Most of the studies referred to in this chapter were based on a cross-sectional design and employees’ self-reports on the behavioural outcome in question. In most of the research that concerned behavioural outcomes of work-related stress, the joint moderating or mediating roles of predisposing personality variables, like the Type A behaviour pattern or hardiness, and situational variables like social support and control, have hardly been investigated. Seldom have antecedent variables, like objectively measured job stress, been included in the research designs of the studies reviewed here. Finally, the research covered in this article employed divergent methodologies. Because of these limitations, a frequently encountered conclusion is that the evidence for work-related stress as a precursor of a behavioural outcome is inconclusive.
Beehr (1995) considered the question of why so few studies have systematically examined the associations between work- related stress and substance abuse. He argued that such neglect may be due in part to the failure of researchers to find these associations. To this failure, one should add the well-known bias of periodicals against publishing research that reports null results. To illustrate the inconclusiveness of the evidence linking stress and substance abuse, consider two large-scale national samples of employees in the United States. The first, by French, Caplan and Van Harrison (1982), failed to find significant correlations between types of work-related stress and either smoking, drug use or on-the-job caffeine ingestion. The second, an earlier research study by Mangione and Quinn (1975), did report such associations.
The study of the behavioural outcomes of stress is further complicated because they frequently appear in pairs or triads. Different combinations of outcomes are the rule rather than the exception. The very close association of stress, smoking and caffeine is alluded to below. Yet another example concerns the comorbidity of post-traumatic stress disorder (PTSD), alcoholism and drug abuse (Kofoed, Friedman and Peck 1993). This is a basic characteristic of several behavioural outcomes considered in this article. It has led to the construction of “dual diagnosis” and “triple diagnosis” schemes and to the development of comprehensive, multifaceted treatment approaches. An example of such an approach is that in which PTSD and substance abuse are treated simultaneously (Kofoed, Friedman and Peck 1993).
The pattern represented by the appearance of several outcomes in a single individual may vary, depending on background characteristics and genetic and environmental factors. The literature on stress outcomes is only beginning to address the complex questions involved in identifying the specific pathophysiological and neurobiological disease models leading to different combinations of outcome entities.
A large body of epidemiological, clinical and pathological studies relates cigarette smoking to the development of cardiovascular heart disease and other chronic diseases. Consequently, there is a growing interest in the pathway leading from stress, including stress at work, to smoking behaviour. Stress, and the emotional responses associated with it, anxiety and irritability, are known to be attenuated by smoking. However, these effects have been shown to be short-lived (Parrott 1995). Impairments of mood and affective states tend to occur in a repetitive cycle between each cigarette smoked. This cycle provides a clear pathway leading to the addictive use of cigarettes (Parrott 1995). Smokers, therefore, obtain only a short-lived relief from adverse states of anxiety and irritability that follow the experience of stress.
The aetiology of smoking is multifactorial (like most other behavioural outcomes considered here). To illustrate, consider a recent review of smoking among nurses. Nurses, the largest professional group in health care, smoke excessively compared with the adult population (Adriaanse et al. 1991). According to their study, this is true for both male and female nurses, and is explained by work stress, lack of social support and unmet expectations that characterize nurses’ professional socialization. Nurses’ smoking is considered a special public health problem since nurses often act as role models to patients and their families.
Smokers who express high motivation to smoke have reported, in several studies, above-average stress that they had experienced before smoking, rather than below-average stress after smoking (Parrott 1995). Consequently, stress management and anxiety reduction programmes in the workplace do have the potential of influencing motivation for smoking. However, workplace-based smoking-cessation programmes do bring to the fore the conflict between health and performance. Among aviators, as an example, smoking is a health hazard in the cockpit. However, pilots who are required to abstain from smoking during and before flights may suffer cockpit performance decrements (Sommese and Patterson 1995).
Drug and Alcohol Abuse
A recurrent problem is that often researchers do not distinguish between drinking and problem-drinking behaviour (Sadava 1987). Problem-drinking is associated with adverse health or performance consequences. Its aetiology has been shown to be associated with several factors. Among them, the literature refers to prior incidents of depression, lack of supportive family environment, impulsiveness, being female, other concurrent substance abuse and stress (Sadava 1987). The distinction between the simple act of drinking alcohol and problem drinking is important because of the current controversy on the reported beneficial effects of alcohol on low density lipoprotein (LDL) cholesterol and on the incidence of heart disease. Several studies have shown a J-shaped or U-shaped relationship between alcohol ingestion and the incidence of cardiovascular heart disease (Pohorecky 1991).
The hypothesis that people ingest alcohol even in an incipiently abusive pattern to reduce stress and anxiety is no longer accepted as adequate. Contemporary approaches to alcohol abuse view it as determined by processes set forth in a multifactorial model or models (Gorman 1994). Among risk factors for alcohol abuse, recent reviews refer to the following factors: sociocultural (i.e., whether alcohol is readily available and its use tolerated, condoned or even promoted), socio-economic (i.e., the price of alcohol), environmental (alcohol advertising and licensing laws affect the consumers’ motivation to drink), interpersonal influences (such as family drinking habits), and employment-related factors, including stress at work (Gorman 1994). It follows that stress is but one of several factors in a multidimensional model that explains alcohol abuse.
The practical consequence of the multifactorial model view of alcoholism is the decrease in the emphasis on the role of stress in the diagnosis, prevention and treatment of substance abuse in the workplace. As noted by a recent review of this literature (Peyser 1992), in specific job situations, such as those illustrated below, attention to work-related stress is important in formulating preventive policies directed at substance abuse.
Despite considerable research on stress and alcohol, the mechanisms that link them are not entirely understood. The most widely accepted hypothesis is that alcohol disrupts the subject’s initial appraisal of stressful information by constraining the spread of activation of associated information previously stored in long-term memory (Petraitis, Flay and Miller 1995).
Work organizations contribute to and may induce drinking behaviour, including problem drinking, by three basic processes documented in the research literature. First, drinking, abusive or not, may be affected by the development of organizational norms with respect to drinking on the job, including the local “official” definition of problem drinking and the mechanisms for its control established by management. Secondly, some stressful working conditions, like sustained overload or machine-paced jobs or the lack of control may produce alcohol abuse as a coping strategy alleviating the stress. Thirdly, work organizations may explicitly or implicitly encourage the development of occupationally based drinking subcultures, such as those that often emerge among professional drivers of heavy vehicles (James and Ames 1993).
In general, stress plays a different role in provoking drinking behaviour in different occupations, age groups, ethnic categories and other social groupings. Thus stress probably plays a predisposing role with respect to alcohol consumption among adolescents, but much less so among women, the elderly and college-age social drinkers (Pohorecky 1991).
The social stress model of substance abuse (Lindenberg, Reiskin and Gendrop 1994) suggests that the likelihood of employees’ drug abuse is influenced by the level of environmental stress, social support relevant to the experienced stress, and individual resources, particularly social competence. There are indications that drug abuse among certain minority groups (like Native American youth living on reservations: see Oetting, Edwards and Beauvais 1988) is influenced by the prevalence of acculturation stress among them. However, the same social groups are also exposed to adverse social conditions like poverty, prejudices and impoverished opportunities for economic, social and educational opportunities.
Caffeine is the most widely consumed pharmacologically active substance in the world. The evidence bearing upon its possible implications for human health, that is whether it has chronic physiological effects on habitual consumers, is as yet inconclusive (Benowitz 1990). It has long been suspected that repeated exposure to caffeine may produce tolerance to its physiological effects (James 1994). The consumption of caffeine is known to improve physical performance and endurance during prolonged activity at submaximal intensity (Nehlig and Debry 1994). Caffeine’s physiological effects are linked to the antagonism of adenosine receptors and to the increased production of plasma catecholamines (Nehlig and Debry 1994).
The study of the relationship of work-related stress on caffeine ingestion is complicated because of the significant inter-dependance of coffee consumption and smoking (Conway et al. 1981). A meta-analysis of six epidemiological studies (Swanson, Lee and Hopp 1994) has shown that about 86% of smokers consumed coffee while only 77% of the non-smokers did so. Three major mechanisms have been suggested to account for this close association: (1) a conditioning effect; (2) reciprocal interaction, that is, caffeine intake increases arousal while nicotine intake decreases it and (3) the joint effect of a third variable on both. Stress, and particularly work-related stress, is a possible third variable influencing both caffeine and nicotine intake (Swanson, Lee and Hopp 1994).
The modern era of sleep research began in the 1950s, with the discovery that sleep is a highly active state rather than a passive condition of nonresponsiveness. The most prevalent type of sleep disturbance, insomnia, may occur in a transient short-term form or in a chronic form. Stress is probably the most frequent cause of transient insomnia (Gillin and Byerley 1990). Chronic insomnia usually results from an underlying medical or psychiatric disorder. Between one-third and two-thirds of patients with chronic insomnia have a recognizable psychiatric illness (Gillin and Byerley 1990).
One of the mechanisms suggested is that the effect of stress on sleep disturbances is mediated via certain changes in the cerebral system at different levels, and changes in the biochemical body functions that disturb the 24-hour rhythms (Gillin and Byerley 1990). There is some evidence that the above linkages are moderated by personality characteristics, such as the Type A behaviour pattern (Koulack and Nesca 1992). Stress and sleep disturbances may reciprocally influence each other: stress may promote transient insomnia, which in turn causes stress and increases the risk of episodes of depression and anxiety (Partinen 1994).
Chronic stress associated with monotonous, machine-paced jobs coupled with the need for vigilance—jobs frequently found in continuous-processing manufacturing industries—may lead to sleep disturbances, subsequently causing decrements in performance (Krueger 1989). There is some evidence that there are synergetic effects among work-related stress, circadian rhythms and reduced performance (Krueger 1989). The adverse effects of sleep loss, interacting with overload and a high level of arousal, on certain important aspects of job performance have been documented in several studies of sleep deprivation among hospital doctors at the junior level (Spurgeon and Harrington 1989).
The study by Mattiason et al. (1990) provides intriguing evidence linking chronic job stress, sleep disturbances and increases in plasma cholesterol. In this study, 715 male shipyard employees exposed to the stress of unemployment were systematically compared with 261 controls before and after the economic instability stress was made apparent. It was found that among the shipyard employees exposed to job insecurity, but not among the controls, sleep disturbances were positively correlated with increases in total cholesterol. This is a naturalistic field study in which the period of uncertainty preceding actual layoffs was allowed to elapse for about a year after some employees received notices concerning the impending layoffs. Thus the stress studied was real, severe, and could be considered chronic.
Absence behaviour may be viewed as an employee coping behaviour that reflects the interaction of perceived job demands and control, on the one hand, and self-assessed health and family conditions on the other. Absenteeism has several major dimensions, including duration, spells and reasons for being absent. It was shown in a European sample that about 60% of the hours lost to absenteeism were due to illness (Ilgen 1990). To the extent that work-related stress was implicated in these illnesses, then there should be some relationship between stress on the job and that part of absenteeism classified as sick days. The literature on absenteeism covers primarily blue-collar employees, and few studies have included stress in a systematic way. (McKee, Markham and Scott 1992). Jackson and Schuler’s meta-analysis (1985) of the consequences of role stress reported an average correlation of 0.09 between role ambiguity and absence and -0.01 between role conflict and absence. As several meta-analytic studies of the literature on absenteeism show, stress is but one of many variables accounting for these phenomena, so we should not expect work-related stress and absenteeism to be strongly correlated (Beehr 1995).
The literature on absenteeism suggests that the relationship between work-related stress and absenteeism may be mediated by employee-specific characteristics. For example, the literature refers to the propensity to use avoidance coping in response to stress at work, and to being emotionally exhausted or physically fatigued (Saxton, Phillips and Blakeney 1991). To illustrate, Kristensen’s (1991) study of several thousand Danish slaughterhouse employees over a one-year period has shown that those who reported high job stress had significantly higher absence rates and that perceived health was closely associated with absenteeism due to illness.
Several studies of the relationships between stress and absenteeism provide evidence that supports the conclusion that they may be occupationally determined (Baba and Harris 1989). To illustrate, work-related stress among managers tends to be associated with the incidence of absenteeism but not with days lost attributed to illness, while this is not so with shop-floor employees (Cooper and Bramwell 1992). Occupational specificity of the stresses predisposing employees to be absent has been regarded as a major explanation of the meagre amount of absence variance explained by work-related stress across many studies (Baba and Harris 1989). Several studies have found that among blue-collar employees who work on jobs considered stressful—that is those that possess a combination of the characteristics of assembly-line type of jobs (namely, a very short cycle of operations and a piece-rate wage system)—job stress is a strong predictor of unexcused absence. (For a recent review of these studies, see McKee, Markham and Scott 1992; note that Baba and Harris 1989 do not support their conclusion that job stress is a strong predictor of unexcused absence).
The literature on stress and absenteeism provides a convincing example of a limitation noted in the introduction. The reference is to the failure of most research on stress-behavioural outcome relations to cover systematically, in the design of this research, both work and non-work stresses. It was noted that in research on absenteeism non-work stress contributed more than work-related stress to the prediction of absence, lending support to the view that absence may be non-work behaviour more than work-related behaviour (Baba and Harris 1989).
Jobs can have a substantial impact on the affective well-being of job holders. In turn, the quality of workers’ well-being on the job influences their behaviour, decision making and interactions with colleagues, and spills over into family and social life as well.
Research in many countries has pointed to the need to define the concept in terms of two separate dimensions that may be viewed as independent of each other (Watson, Clark and Tellegen 1988; Warr 1994). These dimensions may be referred to as “pleasure” and “arousal”. As illustrated in figure 1, a particular degree of pleasure or displeasure may be accompanied by high or low levels of mental arousal, and mental arousal may be either pleasurable or unpleasurable. This is indicated in terms of the three axes of well-being which are suggested for measurement: displeasure-to-pleasure, anxiety-to-comfort, and depression-to-enthusiasm.
Figure 1. Three principal axes for the measurement of affective well-being
Job-related well-being has often been measured merely along the horizontal axis, extending from “feeling bad” to “feeling good”. The measurement is usually made with reference to a scale of job satisfaction, and data are obtained by workers’ indicating their agreement or disagreement with a series of statements describing their feelings about their jobs. However, job satisfaction scales do not take into account differences in mental arousal, and are to that extent relatively insensitive. Additional forms of measurement are also needed, in terms of the other two axes in the figure.
When low scores on the horizontal axis are accompanied by raised mental arousal (upper left quadrant), low well-being is typically evidenced in the forms of anxiety and tension; however, low pleasure in association with low mental arousal (lower left) is observable as depression and associated feelings. Conversely, high job-related pleasure may be accompanied by positive feelings that are characterized either by enthusiasm and energy (3b) or by psychological relaxation and comfort (2b). This latter distinction is sometimes described in terms of motivated job satisfaction (3b) versus resigned, apathetic job satisfaction (2b).
In studying the impact of organizational and psychosocial factors on employee well-being, it is desirable to examine all three of the axes. Questionnaires are widely used for this purpose. Job satisfaction (1a to 1b) may be examined in two forms, sometimes referred to as “facet-free” and “facet-specific” job satisfaction. Facet-free, or overall, job satisfaction is an overarching set of feelings about one’s job as a whole, whereas facet-specific satisfactions are feelings about particular aspects of a job. Principal facets include pay, working conditions, one’s supervisor and the nature of the work undertaken.
These several forms of job satisfaction are positively intercorrelated, and it is sometimes appropriate merely to measure overall, facet-free satisfaction, rather than to examine separate, facet-specific satisfactions. A widely used general question is “On the whole, how satisfied are you with the work you do?”. Commonly used responses are very dissatisfied, a little dissatisfied, moderately satisfied, very satisfied and extremely satisfied, and are designated by scores from 1 to 5 respectively. In national surveys it is usual to find that about 90% of employees report themselves as satisfied to some degree, and a more sensitive measuring instrument is often desirable to yield more differentiated scores.
A multi-item approach is usually adopted, perhaps covering a range of different facets. For instance, several job satisfaction questionnaires ask about a person’s satisfaction with facets of the following kinds: the physical work conditions; the freedom to choose your own method of working; your fellow workers; the recognition you get for good work; your immediate boss; the amount of responsibility you are given; your rate of pay; your opportunity to use your abilities; relations between managers and workers; your workload; your chance of promotion; the equipment you use; the way your firm is managed; your hours of work; the amount of variety in your job; and your job security. An average satisfaction score may be calculated across all the items, responses to each item being scored from 1 to 5, for instance (see the preceding paragraph). Alternatively, separate values can be computed for “intrinsic satisfaction” items (those dealing with the content of the work itself) and “extrinsic satisfaction” items (those referring to the context of the work, such as colleagues and working conditions).
Self-report scales which measure axes two and three have often covered only one end of the possible distribution. For example, some scales of job-related anxiety ask about a worker’s feelings of tension and worry when on the job (2a), but do not in addition test for more positive forms of affect on this axis (2b). Based on studies in several settings (Watson, Clark and Tellegen 1988; Warr 1990), a possible approach is as follows.
Axes 2 and 3 may be examined by putting this question to workers: “Thinking of the past few weeks, how much of the time has your job made you feel each of the following?”, with response options of never, occasionally, some of the time, much of the time, most of the time, and all the time (scored from 1 to 6 respectively). Anxiety-to-comfort ranges across these states: tense, anxious, worried, calm, comfortable and relaxed. Depression-to-enthusiasm covers these states: depressed, gloomy, miserable, motivated, enthusiastic and optimistic. In each case, the first three items should be reverse-scored, so that a high score always reflects high well-being, and the items should be mixed randomly in the questionnaire. A total or average score can be computed for each axis.
More generally, it should be noted that affective well-being is not determined solely by a person’s current environment. Although job characteristics can have a substantial effect, well-being is also a function of some aspects of personality; people differ in their baseline well-being as well as in their reactions to particular job characteristics.
Relevant personality differences are usually described in terms of individuals’ continuing affective dispositions. The personality trait of positive affectivity (corresponding to the upper right-quadrant) is characterized by generally optimistic views of the future, emotions which tend to be positive and behaviours which are relatively extroverted. On the other hand, negative affectivity (corresponding to the upper left-hand quadrant) is a disposition to experience negative emotional states. Individuals with high negative affectivity tend in many situations to feel nervous, anxious or upset; this trait is sometimes measured by means of personality scales of neuroticism. Positive and negative affectivities are regarded as traits, that is, they are relatively constant from one situation to another, whereas a person’s well-being is viewed as an emotional state which varies in response to current activities and environmental influences.
Measures of well-being necessarily identify both the trait (the affective disposition) and the state (current affect). This fact should be borne in mind in examining people’s well-being score on an individual basis, but it is not a substantial problem in studies of the average findings for a group of employees. In longitudinal investigations of group scores, observed changes in well-being can be attributed directly to changes in the environment, since every person’s baseline well-being is held constant across the occasions of measurement; and in cross-sectional group studies an average affective disposition is recorded as a background influence in all cases.
Note also that affective well-being may be viewed at two levels. The more focused perspective relates to a specific domain, such as an occupational setting: this may be a question of “job-related” well-being (as discussed here) and is measured through scales which directly concern feelings when a person is at work. However, more wide-ranging, “context-free” or “general,” well-being is sometimes of interest, and measurement of that wider construct requires a less specific focus. The same three axes should be examined in both cases, and more general scales are available for life satisfaction or general distress (axis 1), context-free anxiety (axis 2) and context-free depression (axis 3).
When a human being or an animal is subjected to a psychological stress situation, there is a general response involving psychological as well as somatic (bodily) responses. This is a general alarm response, or general activation or wake-up call, which affects all physiological responses, including the musculoskeletal system, the vegetative system (the autonomic system), the hormones and also the immune system.
Since the 1960s, we have been learning how the brain, and through it, psychological factors, regulates and influences all physiological processes, whether directly or indirectly. Previously it was held that large and essential parts of our physiology were regulated “unconsciously,” or not by brain processes at all. The nerves that regulate the gut, glands and the cardiovascular system were “autonomic”, or independent of the central nervous system (CNS); similarly, the hormones and the immune system were beyond central nervous control. However, the autonomic nervous system is regulated by the limbic structures of the brain, and may be brought under direct instrumental control through classical and instrumental learning procedures. The fact that the central nervous system controls endocrinological processes is also well established.
The last development to undercut the view that the CNS was isolated from many physiological processes was the evolution of psychoimmunology. It has now been demonstrated that the interaction of the brain (and psychological processes), may influence immune processes, either via the endocrine system or by direct innervation of lymphoid tissue. The white blood cells themselves may also be influenced directly by signal molecules from nervous tissue. Depressed lymphocyte function has been demonstrated to follow bereavement (Bartrop et al. 1977), and conditioning of the immune-suppressive response in animals (Cohen et al. 1979) and psychological processes were shown to have effects bearing on animal survival (Riley 1981); these discoveries were milestones in the development of psychoimmunology.
It is now well established that psychological stress produces changes in the level of antibodies in the blood, and in the level of many of the white blood cells. A brief stress period of 30 minutes may produce significant increases in lymphocytes and natural killer (NK) cells. Following more long-lasting stress situations, changes are also found in the other components of the immune system. Changes have been reported in the counts of almost all types of white blood cell and in the levels of immunoglobulins and their complements; the changes also affect important elements of the total immune response and the “immune cascade” as well. These changes are complex and seem to be bidirectional. Both increases and decreases have been reported. The changes seem to depend not only on the stress-inducing situation, but on also what type of coping and defence mechanisms the individual is using to handle this situation. This is particularly clear when the effects of real long-lasting stress situations are studied, for instance those associated with the job or with difficult life situations (“life stressors”). Highly specific relationships between coping and defence styles and several subsets of immune cells (number of lympho-, leuko- and monocytes; total T cells and NK cells) have been described (Olff et al. 1993).
The search for immune parameters as markers for long-lasting, sustained stress has not been all that successful. Since the relationships between immunoglobulins and stress factors have been demonstrated to be so complex, there is, understandably, no simple marker available. Such relationships as have been found are sometimes positive, sometimes negative. As far as psycho-logical profiles are concerned, to some extent the correlation matrix with one and the same psychological battery shows different patterns, varying from one occupational group to another (Endresen et al. 1991). Within each group, the patterns seem stable over long periods of time, up to three years. It is not known whether there are genetic factors that influence the highly specific relationships between coping styles and immune responses; if so, the manifestions of these factors must be highly dependent on interaction with life stressors. Also, it is not known whether it is possible to follow an individual’s stress level over a long period, given that the individual’s coping, defence and immune response style is known. This type of research is being pursued with highly selected personnel, for instance astronauts.
There may be a major flaw in the basic argument that immunoglobulins can be used as valid health risk markers. The original hypothesis was that low levels of circulating immunoglobulins might signal a low resistance and low immune competence. However, low values may not signal low resistance: they may only signal that this particular individual has not been challenged by infectious agents for a while—in fact, they may signal an extraordinary degree of health. The low values sometimes reported from returning astronauts and Antarctic personnel may not be a signal of stress, but only of the low levels of bacterial and viral challenge in the environment they have left.
There are many anecdotes in the clinical literature suggesting that psychological stress or critical life events can have an impact on the course of serious and non-serious illness. In the opinion of some, placebos and “alternative medicine” may exert their effects through psychoimmunological mechanisms. There are claims that reduced (and sometimes increased) immune competence should lead to increased susceptibility to infections in animals and in humans, and to inflammatory states like rheumatoid arthritis as well. It has been demonstrated convincingly that psychological stress affects the immune response to various types of inoculations. Students under examination stress report more symptoms of infectious illness in this period, which coincides with poorer cellular immune control (Glaser et al. 1992). There are also some claims that psychotherapy, in particular cognitive stress-management training, together with physical training, may affect the antibody response to viral infection.
There are also some positive findings with regard to cancer development, but only a few. The controversy over the claimed relationship between personality and cancer susceptibility has not been solved. Replications should be extended to include measures of immune responses to other factors, including lifestyle factors, which may be related to psychology, but the cancer effect may be a direct consequence of the lifestyle.
There is ample evidence that acute stress alters immune functions in human subjects and that chronic stress may also affect these functions. But to what extent are these changes valid and useful indicators of job stress? To what extent are immune changes—if they occur—a real health risk factor? There is no consensus in the field as of the time of this writing (1995).
Sound clinical trials and sound epidemiological research are required to advance in this field. But this type of research requires more funds than are available to the researchers. This work also requires an understanding of the psychology of stress, which is not always available to immunologists, and a profound understanding of how the immune system operates, which is not always available to psychologists.
Töres Theorell and Jeffrey V. Johnson
The scientific evidence suggesting that exposure to job stress increases the risk for cardiovascular disease increased substantially beginning in the mid-1980s (Gardell 1981; Karasek and Theorell 1990; Johnson and Johansson 1991). Cardiovascular disease (CVD) remains the number one cause of death in economically developed societies, and contributes to increasing medical care costs. Diseases of the cardiovascular system include coronary heart disease (CHD), hypertensive disease, cerebrovascular disease and other disorders of the heart and circulatory system.
Most manifestations of coronary heart disease are caused partly by narrowing of the coronary arteries due to atherosclerosis. Coronary atherosclerosis is known to be influenced by a number of individual factors including: family history, dietary intake of saturated fat, high blood pressure, cigarette smoking and physical exercise. Except for heredity, all these factors could be influenced by the work environment. A poor work environment may decrease the willingness to stop smoking and adopt a healthy lifestyle. Thus, an adverse work environment could influence coronary heart disease via its effects on the classical risk factors.
There are also direct effects of stressful work environments on neurohormonal elevations as well as on heart metabolism. A combination of physiological mechanisms, shown to be related to stressful work activities, may increase the risk of myocardial infarction. The elevation of energy-mobilizing hormones, which increase during periods of excessive stress, may make the heart more vulnerable to the actual death of the muscle tissue. Conversely, energy-restoring and repairing hormones which protect the heart muscle from the adverse effects of energy-mobilizing hormones, decrease during periods of stress. During emotional (and physical) stress the heart beats faster and harder over an extended period of time, leading to excessive oxygen consumption in the heart muscle and the increased possibility of a heart attack. Stress may also disturb the cardiac rhythm of the heart. A disturbance associated with a fast heart rhythm is called tachyarrhythmia. When the heart rate is so fast that the heartbeat becomes inefficient a life-threatening ventricular fibrillation may result.
Early epidemiological studies of psychosocial working conditions associated with CVD suggested that high levels of work demands increased CHD risk. For example a prospective study of Belgian bank employees found that those in a privately owned bank had a significantly higher incidence of myocardial infarction than workers in public banks, even after adjustment was made for biomedical risk factors (Komitzer et al. 1982). This study indicated a possible relationship between work demands (which were higher in the private banks) and risk of myocardial infarction. Early studies also indicated a higher incidence of myocardial infarction among lower level employees in large companies (Pell and d’Alonzo 1963). This raised the possibility that psychosocial stress may not primarily be a problem for people with a high degree of responsibility, as had been assumed previously.
Since the early 1980s, many epidemiological studies have examined the specific hypothesis suggested by the Demand/ Control model developed by Karasek and others (Karasek and Theorell 1990; Johnson and Johansson 1991). This model states that job strain results from work organizations that combine high- performance demands with low levels of control over how the work is to be done. According to the model, work control can be understood as “job decision latitude”, or the task-related decision-making authority permitted by a given job or work organization. This model predicts that those workers who are exposed to high demand and low control over an extended period of time will have a higher risk of neurohormonal arousal which may result in adverse pathophysiological effects on the CVD system—which could eventually lead to increased risk of atherosclerotic heart disease and myocardial infarction.
Between 1981 and 1993, the majority of the 36 studies that examined the effects of high demands and low control on cardiovascular disease found significant and positive associations. These studies employed a variety of research designs and were performed in Sweden, Japan, the United States, Finland and Australia. A variety of outcomes was examined including CHD morbidity and mortality, as well as CHD risk factors including blood pressure, cigarette smoking, left ventricular mass index and CHD symptoms. Several recent review papers summarize these studies (Kristensen 1989; Baker et al. 1992; Schnall, Landsbergis and Baker 1994; Theorell and Karasek 1996). These reviewers note that the epidemiological quality of these studies is high and, moreover, that the stronger study designs have generally found greater support for the Demand/Control models. In general the adjustment for standard risk factors for cardiovascular disease does not eliminate nor significantly reduce the magnitude of the association between the high demand/low control combination and the risk of cardiovascular disease.
It is important to note, however, that the methodology in these studies varied considerably. The most important distinction is that some studies used the respondent’s own descriptions of their work situations, whereas others used an ‘average score’ method based on aggregating the responses of a nationally representative sample of workers within their respective job title groups. Studies utilizing self-reported work descriptions showed higher relative risks (2.0–4.0 versus 1.3–2.0). Psychological job demands were shown to be relatively more important in studies utilizing self-reported data than in studies utilizing aggregated data. The work control variables were more consistently found to be associated with excess CVD risk regardless of which exposure method was used.
Recently, work-related social support has been added to the demand-control formulation and workers with high demands, low control and low support, have been shown to have over a twofold risk for CVD morbidity and mortality compared to those with low demands, high control and high support (Johnson and Hall 1994). Currently efforts are being made to examine sustained exposure to demands, control and support over the course of the “psychosocial work career”. Descriptions of all the occupations during the whole work career are obtained for the participants and occupational scores are used for a calculation of the total lifetime exposure. The “total job control exposure” in relation to cardiovascular mortality incidence in working Swedes was studied and even after adjustment was made for age, smoking habits, exercise, ethnicity, education and social class, low total job control exposure was associated with a nearly twofold risk of dying a cardiovascular death over a 14-year follow-up period (Johnson et al. 1996).
A model similar to the Demand/Control model has been developed and tested by Siegrist and co-workers 1990 that uses “effort” and “social reward” as the crucial dimensions, the hypothesis being that high effort without social reward leads to increasing risk of cardiovascular disease. In a study of industrial workers it was shown that combinations of high effort and lack of reward predicted increased myocardial infarction risk independently of biomedical risk factors.
Other aspects of work organization, such as shift work, have also been shown to be associated with CVD risk. Constant rotation between night and day work has been found to be associated with increased risk of developing a myocardial infarction (Kristensen 1989; Theorell 1992).
Future research in this area particularly needs to focus on specifying the relationship between work stress exposure and CVD risk across different class, gender and ethnic groups.
For many years, psychological stress has been assumed to contribute to the development of peptic ulcer disease (which involves ulcerating lesions in the stomach or duodenum). Researchers and health care providers have proposed more recently that stress might also be related to other gastrointestinal disorders such as non-ulcer dyspepsia (associated with symptoms of upper abdominal pain, discomfort and nausea persisting in the absence of any identifiable organic cause) and irritable bowel syndrome (defined as altered bowel habits plus abdominal pain in the absence of abnormal physical findings). In this article, the question is examined whether there is strong empirical evidence to suggest that psychological stress is a predisposing factor in the aetiology or exacerbation of these three gastrointestinal disorders.
Gastric and Duodenal Ulcer
There is clear evidence that humans who are exposed to acute stress in the context of severe physical trauma are prone to the development of ulcers. It is less obvious, however, whether life stressors per se (such as job demotion or the death of a close relative) precipitate or exacerbate ulcers. Lay people and health care practitioners alike commonly associate ulcers and stress, perhaps as a consequence of Alexander’s (1950) early psychoanalytic perspective on the topic. Alexander proposed that ulcer-prone persons suffered dependency conflicts in their relationships with others; coupled with a constitutional tendency toward chronic hypersecretion of gastric acid, dependency conflicts were believed to lead to ulcer formation. The psychoanalytic perspective has not received strong empirical support. Ulcer patients do not appear to display greater dependency conflicts than comparison groups, though ulcer patients do exhibit higher levels of anxiety, submissiveness and depression (Whitehead and Schuster 1985). The level of neuroticism characterizing some ulcer patients tends to be slight, however, and few could be considered as exhibiting psychopathological signs. In any case, studies of emotional disorder in ulcer patients have generally involved those persons who seek medical attention for their disorder; these individuals may not be representative of all ulcer patients.
The association between stress and ulcers follows from the assumption that certain persons are genetically predisposed to hypersecrete gastric acid, especially during stressful episodes. Indeed, about two thirds of duodenal ulcer patients show elevated pepsinogen levels; elevated levels of pepsinogen are also associated with peptic ulcer disease. Brady and associates’ (1958) studies of “executive” monkeys lent initial support to the idea that a stressful lifestyle or vocation may contribute to the pathogenesis of gastrointestinal disease. They found that monkeys required to perform a lever press task to avoid painful electric shocks (the presumed “executives”, which controlled the stressor) developed more gastric ulcers than comparison monkeys that passively received the same number and intensity of shocks. The analogy to the hard-driving businessman was very cogent for a time. Unfortunately, their results were confounded with anxiety; anxious monkeys were more likely to be assigned to the “executive” role in Brady’s laboratory because they learned the lever press task quickly. Efforts to replicate their results, using random assignment of subjects to conditions, have failed. Indeed, evidence shows that animals who lack control over environmental stressors develop ulcers (Weiss 1971). Human ulcer patients also tend to be shy and inhibited, which runs counter to the stereotype of the ulcer-prone hard-driving businessman. Finally, animal models are of limited utility because they focus on the development of gastric ulcers, while most ulcers in humans occur in the duodenum. Laboratory animals rarely develop duodenal ulcers in response to stress.
Experimental studies of the physiological reactions of ulcer patients versus normal subjects to laboratory stressors do not uniformly show excessive reactions in the patients. The premise that stress leads to increased acid secretion which, in turn, leads to ulceration, is problematic when one realizes that psychological stress usually produces a response from the sympathetic nervous system. The sympathetic nervous system inhibits, rather than enhances, the gastric secretion that is mediated via the splanchnic nerve. Besides hypersecretion other factors in the aetiology of ulcer have been proposed, namely, rapid gastric emptying, inadequate secretion of bicarbonate and mucus, and infection. Stress could potentially affect these processes though evidence is lacking.
Ulcers have been reported to be more common during wartime, but methodological problems in these studies necessitate caution. A study of air traffic controllers is sometimes cited as evidence supporting the role of psychological stress for the development of ulcers (Cobb and Rose 1973). Although air traffic controllers were significantly more likely than a control group of pilots to report symptoms typical of ulcer, the incidence of confirmed ulcer among the air traffic controllers was not elevated above the base rate of ulcer occurrence in the general population.
Studies of acute life events also present a confusing picture of the relationship between stress and ulcer (Piper and Tennant 1993). Many investigations have been conducted, though most of these studies employed small samples and were cross-sectional or retrospective in design. The majority of studies did not find that ulcer patients incurred more acute life events than community controls or patients with conditions in which stress is not implicated, such as gallstones or renal stones. However, ulcer patients reported more chronic stressors involving personal threat or goal frustration prior to the onset or recrudescence of ulcer. In two prospective studies, reports of subjects being under stress or having family problems at baseline levels predicted subsequent development of ulcers. Unfortunately, both prospective studies used single-item scales to measure stress. Other research has shown that slow healing of ulcers or relapse was associated with higher stress levels, but the stress indices used in these studies were unvalidated and may have been confounded with personality factors.
In summary, evidence for the role of stress in ulcer causation and exacerbation is limited. Large-scale population-based prospective studies of the occurrence of life events are needed which use validated measures of acute and chronic stress and objective indicators of ulcer. At this point, evidence for an association between psychological stress and ulcer is weak.
Irritable Bowel Syndrome
Irritable bowel syndrome (IBS) has been considered a stress- related disorder in the past, in part because the physiological mechanism of the syndrome is unknown and because a large proportion of IBS sufferers report that stress caused a change in their bowel habits. As in the ulcer literature, it is difficult to evaluate the value of retrospective accounts of stressors and symptoms among IBS patients. In an effort to explain their discomfort, ill persons may mistakenly associate symptoms with stressful life events. Two recent prospective studies shed more light on the subject, and both found a limited role for stressful events in the occurrence of IBS symptoms. Whitehead et al. (1992) had a sample of community residents suffering from IBS symptoms report life events and IBS symptoms at three-month intervals. Only about 10% of the variance in bowel symptoms among these residents could be attributed to stress. Suls, Wan and Blanchard (1994) had IBS patients keep diary records of stressors and symptoms for 21 successive days. They found no consistent evidence that daily stressors increased the incidence or severity of IBS symptomatology. Life stress appears to have little effect on acute changes in IBS.
The symptoms of non-ulcer dyspepsia (NUD) include bloating and fullness, belching, borborygmi, nausea and heartburn. In one retrospective study, NUD patients reported more acute life events and more highly threatening chronic difficulties compared to healthy community members, but other investigations failed to find a relationship between life stress and functional dyspepsia. NUD cases also show high levels of psychopathology, notably anxiety disorders. In the absence of prospective studies of life stress, few conclusions can be made (Bass 1986; Whitehead 1992).
Despite considerable empirical attention, no verdict has yet been reached on the relationship between stress and the development of ulcers. Contemporary gastroenterologists have focused mainly on heritable pepsinogen levels, inadequate secretion of bicarbonate and mucus, and Heliobacter pylori infection as causes of ulcer. If life stress plays a role in these processes, its contribution is probably weak. Though fewer studies address the role of stress in IBS and NUD, evidence for a connection to stress is also weak here. For all three disorders, there is evidence that anxiety is higher among patients compared to the general population, at least among those persons who refer themselves for medical care (Whitehead 1992). Whether this is a precursor or a consequence of gastrointestinal disease has not been definitively determined, although the latter opinion seems to be more likely to be true. In current practice, ulcer patients receive pharmacological treatment, and psychotherapy is rarely recommended. Anti-anxiety drugs are commonly prescribed to IBS and NUD patients, probably because the physiological origins of these disorders are still unknown. Stress management has been employed with IBS patients with some success (Blanchard et al. 1992) although this patient group also responds to placebo treatments quite readily. Finally, patients experiencing ulcer, IBS or NUD may well be frustrated by assumptions from family members, friends and practitioners alike that their condition was produced by stress.
Stress, the physical and/or psychological departure from a person’s stable equilibrium, can result from a large number of stressors, those stimuli that produce stress. For a good general view of stress and the most common job stressors, Levi’s discussion in this chapter of job stress theories is recommended.
In addressing the question of whether job stress can and does affect the epidemiology of cancer, we face limitations: a search of the literature located only one study on actual job stress and cancer in urban bus drivers (Michaels and Zoloth 1991) (and there are only few studies in which the question is considered more generally). We cannot accept the findings of that study, because the authors did not take into account either the effects of high density exhaust fumes or smoking. Further, one cannot carry over the findings from other diseases to cancer because the disease mechanisms are so vastly different.
Nevertheless, it is possible to describe what is known about the connections between more general life stressors and cancer, and further, one might reasonably apply those findings to the job situation. We differentiate relationships of stress to two outcomes: cancer incidence and cancer prognosis. The term incidence evidently means the occurrence of cancer. However, incidence is established either by the doctor’s clinical diagnosis or at autopsy. Since tumour growth is slow—1 to 20 years may elapse from the malignant mutation of one cell to the detection of the tumour mass—incidence studies include both initiation and growth. The second question, whether stress can affect prognosis, can be answered only in studies of cancer patients after diagnosis.
We distinguish cohort studies from case-control studies. This discussion focuses on cohort studies, where a factor of interest, in this case stress, is measured on a cohort of healthy persons, and cancer incidence or mortality is determined after a number of years. For several reasons, little emphasis is given to case-control studies, those which compare reports of stress, either current or before diagnosis, in cancer patients (cases) and persons without cancer (controls). First, one can never be sure that the control group is well-matched to the case group with respect to other factors that can influence the comparison. Secondly, cancer can and does produce physical, psychological and attitudinal changes, mostly negative, that can bias conclusions. Thirdly, these changes are known to result in an increase in the number of reports of stressful events (or of their severity) compared to reports by controls, thus leading to biased conclusions that patients experienced more, or more severe, stressful events than did controls (Watson and Pennebaker 1989).
Stress and Cancer Incidence
Most studies on stress and cancer incidence have been of the case-control sort, and we find a wild mix of results. Because, in varying degrees, these studies have failed to control contaminating factors, we don’t know which ones to trust, and they are ignored here. Among cohort studies, the number of studies showing that persons under greater stress did not experience more cancer than those under lesser stress exceeded by a large margin the number showing the reverse (Fox 1995). The results for several stressed groups are given.
Stress and cancer prognosis
This topic is of lesser interest because so few people of working age get cancer. Nevertheless, it ought to be mentioned that while survival differences have been found in some studies with regard to reported pre-diagnosis stress, other studies have shown no differences. One should, in judging these findings, recall the parallel ones showing that not only cancer patients, but also those with other ills, report more past stressful events than well people to a substantial degree because of the psychological changes brought about by the disease itself and, further, by the knowledge that one has the disease. With respect to prognosis, several studies have shown increased survival among those with good social support as against those with less social support. Perhaps more social support produces less stress, and vice versa. As regards both incidence and prognosis, however, the extant studies are at best only suggestive (Fox 1995).
It might be instructive to see what effects stress has had in experiments with animals. The results among well-conducted studies are much clearer, but not decisive. It was found that stressed animals with viral tumours show faster tumour growth and die earlier than unstressed animals. But the reverse is true of non-viral tumours, that is, those produced in the laboratory by chemical carcinogens. For these, stressed animals have fewer tumours and longer survival after the start of cancer than unstressed animals (Justice 1985). In industrial nations, however, only 3 to 4% of human malignancies are viral. All the rest are due to chemical or physical stimuli—smoking, x rays, industrial chemicals, nuclear radiation (e.g., that due to radon), excessive sunlight and so on. Thus, if one were to extrapolate from the findings for animals, one would conclude that stress is beneficial both to cancer incidence and survival. For a number of reasons one should not draw such an inference (Justice 1985; Fox 1981). Results with animals can be used to generate hypotheses relating to data describing humans, but cannot be the basis for conclusions about them.
In view of the variety of stressors that has been examined in the literature—long-term, short-term, more severe, less severe, of many types—and the preponderance of results suggesting little or no effect on later cancer incidence, it is reasonable to suggest that the same results apply in the work situation. As for cancer prognosis, too few studies have been done to draw any conclusions, even tentative ones, about stressors. It is, however, possible that strong social support may decrease incidence a little, and perhaps increase survival.
There is growing evidence in the occupational health literature that psychosocial work factors may influence the development of musculoskeletal problems, including both low back and upper extremity disorders (Bongers et al. 1993). Psychosocial work factors are defined as aspects of the work environment (such as work roles, work pressure, relationships at work) that can contribute to the experience of stress in individuals (Lim and Carayon 1994; ILO 1986). This paper provides a synopsis of the evidence and underlying mechanisms linking psychosocial work factors and musculoskeletal problems with the emphasis on studies of upper extremity disorders among office workers. Directions for future research are also discussed.
An impressive array of studies from 1985 to 1995 had linked workplace psychosocial factors to upper extremity musculoskeletal problems in the office work environment (see Moon and Sauter 1996 for an extensive review). In the United States, this relationship was first suggested in an exploratory research by the National Institute for Occupational Safety and Health (NIOSH) (Smith et al. 1981). Results of this research indicated that video display unit (VDU) operators who reported less autonomy and role clarity and greater work pressure and management control over their work processes also reported more musculoskeletal problems than their counterparts who did not work with VDUs (Smith et al. 1981).
Recent studies employing more powerful inferential statistical techniques point more strongly to an effect of psychosocial work factors on upper extremity musculoskeletal disorders among office workers. For example, Lim and Carayon (1994) used structural analysis methods to examine the relationship between psychosocial work factors and upper extremity musculoskeletal discomfort in a sample of 129 office workers. Results showed that psychosocial factors such as work pressure, task control and production quotas were important predictors of upper extremity musculoskeletal discomfort, especially in the neck and shoulder regions. Demographic factors (age, gender, tenure with employer, hours of computer use per day) and other confounding factors (self-reports of medical conditions, hobbies and keyboard use outside work) were controlled for in the study and were not related to any of these problems.
Confirmatory findings were reported by Hales et al. (1994) in a NIOSH study of musculoskeletal disorders in 533 tele-communication workers from 3 different metropolitan cities. Two types of musculoskeletal outcomes were investigated: (1) upper extremity musculoskeletal symptoms determined by questionnaire alone; and (2) potential work-related upper extremity musculoskeletal disorders which were determined by physical examination in addition to the questionnaire. Using regression techniques, the study found that factors such as work pressure and little decision-making opportunity were associated both with intensified musculoskeletal symptoms and also with increased physical evidence of disease. Similar relationships have been observed in the industrial environment, but mainly for back pain (Bongers et al. 1993).
Researchers have suggested a variety of mechanisms underlying the relationship between psychosocial factors and musculoskeletal problems (Sauter and Swanson 1996; Smith and Carayon 1996; Lim 1994; Bongers et al. 1993). These mechanisms can be classified into four categories:
It has been demonstrated that individuals subject to stressful psychosocial working conditions also exhibit increased autonomic arousal (e.g., increased catecholomine secretion, increased heart rate and blood pressure, increased muscle tension etc.) (Frankenhaeuser and Gardell 1976). This is a normal and adaptive psychophysiological response which prepares the individual for action. However, prolonged exposure to stress may have a deleterious effect on musculoskeletal function as well as on health in general. For example, stress-related muscle tension may increase the static loading of muscles, thereby accelerating muscle fatigue and associated discomfort (Westgaard and Bjorklund 1987; Grandjean 1986).
Individuals who are under stress may alter their work behaviour in a way that increases musculoskeletal strain. For example, psychological stress may result in greater application of force than necessary during typing or other manual tasks, leading to increased wear and tear on the musculoskeletal system.
Psychosocial factors may influence the physical (ergonomic) demands of the job directly. For example, an increase in time pressure is likely to lead to an increase in work pace (i.e., increased repetition) and increased strain. Alternatively, workers who are given more control over their tasks may be able to adjust their tasks in ways that lead to reduced repetitiveness (Lim and Carayon 1994).
Sauter and Swanson (1996) suggest that the relationship between biomechanical stressors (e.g., ergonomic factors) and the development of musculoskeletal problems is mediated by perceptual processes which are influenced by workplace psychosocial factors. For example, symptoms might become more evident in dull, routine jobs than in more engrossing tasks which more fully occupy the attention of the worker (Pennebaker and Hall 1982).
Additional research is needed to assess the relative importance of each of these mechanisms and their possible interactions. Further, our understanding of causal relationships between psychosocial work factors and musculoskeletal disorders would benefit from: (1) increased use of longitudinal study designs; (2) improved methods for assessing and disentangling psychosocial and physical exposures; and (3) improved measurement of musculoskeletal outcomes.
Still, the current evidence linking psychosocial factors and musculoskeletal disorders is impressive and suggests that psychosocial interventions probably play an important role in preventing musculoskeletal problems in the workplace. In this regard, several publications (NIOSH 1988; ILO 1986) provide directions for optimizing the psychosocial environment at work. As suggested by Bongers et al. (1993), special attention should be given to providing a supportive work environment, manageable workloads and increased worker autonomy. Positive effects of such variables were evident in a case study by Westin (1990) of the Federal Express Corporation. According to Westin, a programme of work reorganization to provide an “employee-supportive” work environment, improve communications and reduce work and time pressures was associated with minimal evidence of musculoskeletal health problems.
Carles Muntaner and William W. Eaton
Mental illness is one of the chronic outcomes of work stress that inflicts a major social and economic burden on communities (Jenkins and Coney 1992; Miller and Kelman 1992). Two disciplines, psychiatric epidemiology and mental health sociology (Aneshensel, Rutter and Lachenbruch 1991), have studied the effects of psychosocial and organizational factors of work on mental illness. These studies can be classified according to four different theoretical and methodological approaches: (1) studies of only a single occupation; (2) studies of broad occupational categories as indicators of social stratification; (3) comparative studies of occupational categories; and (4) studies of specific psychosocial and organizational risk factors. We review each of these approaches and discuss their implications for research and prevention.
Studies of a Single Occupation
There are numerous studies in which the focus has been a single occupation. Depression has been the focus of interest in recent studies of secretaries (Garrison and Eaton 1992), professionals and managers (Phelan et al. 1991; Bromet et al. 1990), computer workers (Mino et al. 1993), fire-fighters (Guidotti 1992), teachers (Schonfeld 1992), and “maquiladoras” (Guendelman and Silberg 1993). Alcoholism and drug abuse and dependence have been recently related to mortality among bus drivers (Michaels and Zoloth 1991) and to managerial and professional occupations (Bromet et al. 1990). Symptoms of anxiety and depression which are indicative of psychiatric disorder have been found among garment workers, nurses, teachers, social workers, offshore oil industry workers and young physicians (Brisson, Vezina and Vinet 1992; Fith-Cozens 1987; Fletcher 1988; McGrath, Reid and Boore 1989; Parkes 1992). The lack of a comparison group makes it difficult to determine the significance of this type of study.
Studies of Broad Occupational Categories as Indicators of Social Stratification
The use of occupations as indicators of social stratification has a long tradition in mental health research (Liberatos, Link and Kelsey 1988). Workers in unskilled manual jobs and lower-grade civil servants have shown high prevalence rates of minor psychiatric disorders in England (Rodgers 1991; Stansfeld and Marmot 1992). Alcoholism has been found to be prevalent among blue-collar workers in Sweden (Ojesjo 1980) and even more prevalent among managers in Japan (Kawakami et al. 1992). Failure to differentiate conceptually between effects of occupations per se from “lifestyle” factors associated with occupational strata is a serious weakness of this type of study. It is also true that occupation is an indicator of social stratification in a sense different from social class, that is, as the latter implies control over productive assets (Kohn et al. 1990; Muntaner et al. 1994). However, there have not been empirical studies of mental illness using this conceptualization.
Comparative Studies of Occupational Categories
Census categories for occupations constitute a readily available source of information that allows one to explore associations between occupations and mental illness (Eaton et al. 1990). Epidemiological Catchment Area (ECA) study analyses of comprehensive occupational categories have yielded findings of a high prevalence of depression for professional, administrative support and household services occupations (Roberts and Lee 1993). In another major epidemiological study, the Alameda county study, high rates of depression were found among workers in blue-collar occupations (Kaplan et al. 1991). High 12-month prevalence rates of alcohol dependence among workers in the Unites States have been found in craft occupations (15.6%) and labourers (15.2%) among men, and in farming, forestry and fishing occupations (7.5%) and unskilled service occupations (7.2%) among women (Harford et al. 1992). ECA rates of alcohol abuse and dependence yielded high prevalence among transportation, craft and labourer occupations (Roberts and Lee 1993). Workers in the service sector, drivers and unskilled workers showed high rates of alcoholism in a study of the Swedish population (Agren and Romelsjo 1992). Twelve-month prevalence of drug abuse or dependence in the ECA study was higher among farming (6%), craft (4.7%), and operator, transportation and labourer (3.3%) occupations (Roberts and Lee 1993). The ECA analysis of combined prevalence for all psychoactive substance abuse or dependence syndromes (Anthony et al. 1992) yielded higher prevalence rates for construction labourers, carpenters, construction trades as a whole, waiters, waitresses and transportation and moving occupations. In another ECA analysis (Muntaner et al. 1991), as compared to managerial occupations, greater risk of schizophrenia was found among private household workers, while artists and construction trades were found at higher risk of schizophrenia (delusions and hallucinations), according to criterion A of the Diagnostic and Statistics Manual of Mental Disorders (DSM-III) (APA 1980).
Several ECA studies have been conducted with more specific occupational categories. In addition to specifying occupational environments more closely, they adjust for sociodemographic factors which might have led to spurious results in uncontrolled studies. High 12-month prevalence rates of major depression (above the 3 to 5% found in the general population (Robins and Regier 1990), have been reported for data entry keyers and computer equipment operators (13%) and typists, lawyers, special education teachers and counsellors (10%) (Eaton et al. 1990). After adjustment for sociodemographic factors, lawyers, teachers and counsellors had significantly elevated rates when compared to the employed population (Eaton et al. 1990). In a detailed analysis of 104 occupations, construction labourers, skilled construction trades, heavy truck drivers and material movers showed high rates of alcohol abuse or dependence (Mandell et al. 1992).
Comparative studies of occupational categories suffer from the same flaws as social stratification studies. Thus, a problem with occupational categories is that specific risk factors are bound to be missed. In addition, “lifestyle” factors associated with occupational categories remain a potent explanation for results.
Studies of Specific Psychosocial and Organizational Risk Factors
Most studies of work stress and mental illness have been conducted with scales from Karasek’s Demand/Control model (Karasek and Theorell 1990) or with measures derived from the Dictionary of Occupational Titles (DOT) (Cain and Treiman 1981). In spite of the methodological and theoretical differences underlying these systems, they measure similar psychosocial dimensions (control, substantive complexity and job demands) (Muntaner et al. 1993). Job demands have been associated with major depressive disorder among male power-plant workers (Bromet 1988). Occupations involving lack of direction, control or planning have been shown to mediate the relation between socioeconomic status and depression (Link et al. 1993). However, in one study the relationship between low control and depression was not found (Guendelman and Silberg 1993). The number of negative work-related effects, lack of intrinsic job rewards and organizational stressors such as role conflict and ambiguity have also been associated with major depression (Phelan et al. 1991). Heavy alcohol drinking and alcohol-related problems have been linked to working overtime and to lack of intrinsic job rewards among men and to job insecurity among women in Japan (Kawakami et al. 1993), and to high demands and low control among males in the United States (Bromet 1988). Also among US males, high psychological or physical demands and low control were predictive of alcohol abuse or dependence (Crum et al. 1995). In another ECA analysis, high physical demands and low skill discretion were predictive of drug dependence (Muntaner et al. 1995). Physical demands and job hazards were predictors of schizophrenia or delusions or hallucinations in three US studies (Muntaner et al. 1991; Link et al. 1986; Muntaner et al. 1993). Physical demands have also been associated with psychiatric disease in the Swedish population (Lundberg 1991). These investigations have the potential for prevention because specific, potentially malleable risk factors are the focus of study.
Implications for Research and Prevention
Future studies might benefit from studying the demographic and sociological characteristics of workers in order to sharpen their focus on the occupations proper (Mandell et al. 1992). When occupation is considered an indicator of social stratification, adjustment for non-work stressors should be attempted. The effects of chronic exposure to lack of democracy in the workplace need to be investigated (Johnson and Johansson 1991). A major initiative for the prevention of work-related psychological disorders has emphasized improving working conditions, services, research and surveillance (Keita and Sauter 1992; Sauter, Murphy and Hurrell 1990).
While some researchers maintain that job redesign can improve both productivity and workers’ health (Karasek and Theorell 1990), others have argued that a firm’s profit maximization goals and workers’ mental health are in conflict (Phelan et al. 1991; Muntaner and O’Campo 1993; Ralph 1983).