There is an old saying that to a man with a hammer, the world is full of nails, which might explain why it is difficult to obtain consensus amongst researchers on any one particular subject; since researchers approach problems with their own individual perspectives, definitions and set of tools. It may be particularly so in Psychoimmunology (PI) research, where many disciplines converge into the scientific study of the interaction between psychological factors, the central nervous system and the immune system. PI promises to explore and explain the commonly held belief that our personalities and emotions influence our health. It also has the potential to deliver psychological interventions that may improve immunity, and consequently disease susceptibility, onset and progress.
PI research over the last two decades, attests to the possibility that the immune system may be mediated by psychological factors. Although early studies have been criticised for poor designs and lack of control, recent studies seem more focused and appear to be converging towards more directly useful findings. The present review examines recent research, which illuminates previously inconsistent findings, and focuses on these specific aspects of the literature: (a) the major theoretical underpinning of psychoimmunology, (b) the psychological factors which appear most salient in modulating the immune system, repressive denial, locus of control (LOC), and coping styles, and their relationship to disease, (c) psychological interventions that may delay the onset or progression of disease, and (d) following a summary, suggestions for further research and application possibilities.
The Major Theoretical Underpinning of Psychoimmunology
The theoretical model of psychoimmunology is founded upon the bidirectional relationships between psychological factors and immune system functions via the hypothalamic-pituitary-adrenocorticol axis of the sympathetic nervous system (Cohen & Herbert, 1996; Geiser, 1989; Schwartz, 1994). This model is based on Hans Selye’s General Adaptation Syndrome, which claims that while benign stress is normal and harmless, some stressful states are noxious; and that the more severe, prolonged and uncontrollable situations of psychological distress could lead to disease (Selye, 1950). Recent research indicates that changes in immune functions do not necessarily translate directly into changes in health outcomes. Other factors probably mediate the relationship, such as prior immunological and health status, the degree of pervasiveness of immune modulation, and the intensity of the psychological factors (Kiecolt-Glaser & Glaser, 1992).
Chrousos and Gold suggested that the stress-illness relationship may follow the classic inverted-U hypothesis of arousal and performance. Namely that there may exist an optimum reactivity level for hypothalamic-pituitary reactivity, with hyporeactivity leading to autoimmune inflammatory phenomena and hypereactivity leading to addiction, panic disorders, and severe chronic diseases (Chrousos & Gold, 1992). Mason had previously indicated that hypothalamic-pituitary reactivity was probably genetically determined (Mason, 1991). If these suggestions are correct, comprehensive models of psychoimmunology may need to take into account individual differences in hypothalamic-pituitary axis responsiveness and genetic predisposition to specific health problems, along with psychological modulators (Schwartz, 1994).
Due to the complexity of psychoimmunological interactions, it is not known exactly how the hypothalamus orchestrates the immune response, modulated by psychological factors. However, whatever the mechanism, eventually the body’s defense system relies on immune system cells to detect viruses, bacteria or foreign substances (antigens) within the body and engulf these through the mononuclear phagocyte system. All immune system cells originate from the bone marrow as hematopoietic stem cells, and develop into specific cell types; for example, phagocytes (e.g., neutrophils and macrophages) and lymphocytes (e.g., B cells, T cells, and natural killer (NK) cells), each with their own specific functions. An effective defense against an antigen challenge requires that three major cell groups act in concert, T cells, B cells and macrophages, although many other cellular, hormonal and biochemical reactions contribute to the immune system response (Geiser, 1989).
Most immune system cells are located in bone marrow, spleen, lymph glands and other inaccessible organs. Hence, human research is generally limited to immune system cells present in circulating white blood cells, mucous secretions and saliva. The cells most often studied are NK cells, as they kill compromised cells nonspecifically, and two antigen-specific immunoglobulins secreted by mature B cells: immunoglobulin A (IgA), present in saliva and body secretions and IgG , which makes up the bulk of circulating Ig (Geiser, 1989).
Psychological Modulation of the Immune System and Health Outcomes
Stressful Events and Perceived Stress
A major component of PI research consists of relating immune system cell counts and health measures to various psychological factors. To this end, researchers have investigated the varying effects of stress on the immune system and health. Traumatic life events, like the death of a loved one, can be assumed to be a devastating stressor for any person; on the other hand, it is likely that a statistics examination will have different effects on different students, yet both are stressors. The variability in response to stressors is due to internal processes of personality, experience, cognition and coping styles, which result in varying degrees of stress response (Cohen & Herbert, 1996).
Earlier literature from the 80s, as reported in a comprehensive review by David Geiser, did not always produce findings that were in agreement about the relationships between psychological factors and the immune system (Geiser, 1989). For example, while one study showed an increase in salivary IgA (sIgA) following an academic examination, (McClelland, Ross, & Patel, 1985) another showed a decrease in sIgA under academic stress (Jemmott et al., 1983), while a third did not find significant changes in sIgA among medical students under stress (Kiecolf-Glaser et al., 1984). It may be that earlier research did not differentiate between external stressors and the perception of stress as some of the more recent studies, reviewed next, have done.
For example, two studies by Boyce and his colleagues, examined the reactivity of children to stress and childhood respiratory illnesses. In the first study, the cardiovascular reactivity of kindergarden-age children was assessed during a series of developmentally challenging tasks; and environmental stress was evaluated with two measures of stressors in the childcare setting. The incidence of respiratory illnesses was ascertained over 6 months, using weekly respiratory tract examinations by a nurse. In the second study children were assessed for immune reactivity one week before and after kindergarten entry. Environmental stress was assessed through parent reports of family stressors; and respiratory illness incidence was measured with bi-weekly, parent-completed symptom checklists, over a 12-week period. Although environmental stress was not independently associated with respiratory illnesses in either study, both studies found that the incidence of illness was related to (a) an interaction between stress in the childcare setting, and to mean arterial pressure reactivity, and (b) an interaction between stressful life events and a measure of immune reactivity (Boyce et al., 1995). These findings suggest that the individual reaction to external stress, and not external stress per say, may be responsible for illness susceptibility.
Meta-analysis techniques were used to evaluate the findings of sixteen published studies that examined the relations of stressors and depressive symptoms to clinical recurrence of Herpes Simplex Virus (HSV). The researchers calculated average effect sizes, performed fixed effect and random effect inferential analyses, tested for heterogeneous findings, and identified potential moderating variables. The results showed that depressive symptoms, but not stressors were associated with increased risk of HSV recurrence. The analyses supports the view that in some individuals there is susceptibility to disease progression due to perceived stress and/or depressive symptoms (Zorrilla, McKay, Luborsky, & Schmidt, 1996).
If the immune system is suppressed by negative perceptions of stress, how does is it react to positive perceptions, interpreted as challenging? In order to study the reaction of perceived positive and negative daily events on the immune system’s antibody production, Stone and his colleagues administered an innocuous protein capsule, which elicited an immune system reaction, to male community volunteers, over a 12 week period. The volunteers kept a daily diary and gave daily saliva samples, which were used to assess their sIgA levels. The results showed that sIgA production was positively related to positive (desirable) events, and negatively related to negative (undesirable) events (Stone et al., 1994).
This finding is supported in a larger study of the effects of work demands on sIgA and cortisol in air traffic controllers (ATC), 158 male ATCs gave saliva samples after each of two working sessions. The nature of ATCs’ work is such that they cannot predict when a situation is likely to become critical, and when their workload is likely to increase to highly stressful levels. Zeier and his colleagues found that in marked contrast to the expected immunosuppression under stress, sIgA and salivary cortisol levels were elevated following the work sessions. Furthermore, sIgA increase was not correlated with actual and perceived workloads or with cortisol response. The ATCs showed strong professional pride and obtained high positive scores on the mood scales of an adjective checklist, measuring their emotional response to the working sessions. Zeier concluded that positive emotional engagement might be responsible for the increase in sIgA, and that sIgA might be a useful physiological measure to differentiate between positive and negative perceived stress effects (Zeier, Brauchli, & Joller, 1996). These findings are consistent with the suggestions that positive perceptions of experiences may be associated with raising immunoglobulin levels while negative ones are associated with immunosuppression.
Borysenko suggested that stressful events in themselves do not necessarily relate to immunosuppression, but that the stress which appears related to illness can be defined as the perception of physical or emotional threat, coupled with the perception that available responses are inadequate to cope (Borysenko, 1995; Terr, 1995). The question which begs to be asked is which psychological modulators are responsible for the variability in the individuals perception of a stimulus as a stressful one? Three of these are discussed in the present review, the first of which is repressive denial.
When faced with stressful situations people are known to rationalise and apply coping strategies which vary from one individual to the next; and which vary from one situation to the next for the same individual (Lazarus & Folkman, 1984). Amongst the many explanations for this diversity of responses, psychodynamic theories propose that basic defense mechanisms are sometimes in place. For example, denial defense may occur when the threat cannot be attacked or avoided. While outright denial may be possible for a child whose ego may still allow this violation of reality testing, in adults the more subtle mechanism of repression may be used (Mischel, 1993). In this context, repression is seen as the self-deluding elimination from consciousness of painful memories or the rejection of thoughts or impulses that might have adverse effects on self-esteem (Weinberger, 1990).
Repressive deniers take on a particular form of self and other deception. They emphatically maintain minimal self-perceptions of negative emotions despite tendencies to respond physiologically and behaviourally in ways suggesting the presence of high levels of perceived threat (Weinberger, 1990). Such people would be characterised by a need to see themselves as well adjusted, despite the fact that they deny much of their inner emotions, and have little awareness of their needs, wishes and feelings (Shedler, Mayman, & Manis, 1993). Hence the repressive denier’s ego may be constantly at odds with reality, resulting in unacknowledged and chronic underlying stress. In order to maintain their delusive self image, they are likely to deny acknowledgment of stress, anxiety or ill health in self reports.
Scores on mental health scales are therefore ambiguous when they fall near the healthy end, as they are likely to include those of the genuinely mentally healthy, as well as those of the repressive deniers with their illusory mental health (Shedler et al., 1993). Repressive deniers can be identified using a combination of two measures: high scores (top 33%) scores on Short Form C of the Marlowe-Crowne Social Desirability Scale (MCSDS) (Reynolds, 1982), and low scores (bottom 33%) on a measure of anxiety, such as the State Trait Anxiety Inventory (STAI) (Speilberger, 1977). The MCSDS scale, was specifically designed to avoid scores from being confounded by responses that could reflect some aspect of reality rather than a desire to look good. Weinberger cited much evidence suggesting that high scorers on the MCSDS were primarily self deceivers, with repressive coping styles (Weinberger, 1990).
If as Shedler suggested, repressive deniers are at risk of ill health, what has PI research revealed about their immunological responses and illness patterns? Jammer and Schwartz (1986), proposed an endogenous opiod modulator model to account for the growing number of observations relating to repressive coping styles. They proposed an “opiod peptide” hypothesis which postulates that repressive coping styles are associated with greater central endogenous opiod system activity, consistent with the attenuated experiences of distress, affective pain appreciation and anxiety, together with accentuated reports of positive emotions documented amongst repressors (Jamner, Schwartz, & Leigh, 1988).
In a follow-up study, using the Bendig short form of the Taylor Manifest Anxiety Scales (MAS) and the MCSDS, Jammer found that out of 312 patients with stress related disorders, 79 were classified as repressives (repressive deniers, with high MCSDS and low MAS scores), and 69 as defensive high-anxious (high MCSDS and high MAS). True high and low anxious groups were also identified. The repressive denier group had significantly lower monocyte counts than the defensive high anxious group; and both had significantly lower monocyte levels than the low anxious group (Jamner et al., 1988). These studies provide evidence that the immune system of repressive deniers may be compromised by evoking stressors, which their high autonomic response and low monocyte count seems to indicate. (Hall, Anderson, & O’Grady, 1994), Jammer suggested that the defensive high anxious group were probably repressive deniers whose coping mechanism had failed, (Jamner et al., 1988) .
Although there is evidence that repressive denial is related to immune system suppression, there is less evidence that it is related to illness through immunosuppression. The relationship between repression and diseases is more apparent in the etiology of cardiovascular diseases (Adler & Matthews, 1994). For example there is evidence that individuals characterized by a positive attitude toward emotions and an ability to perceive and express them adequately used problem-focused coping and had less incidence of coronary heart disease than individuals who were uncomfortable with their emotions (Keltikangas-Jarvinen & Raikkonen, 1993). Markovitz and his colleagues found significant links between anxiety and cardiovascular function in two separate studies. In the first study, they found that after controlling for age, parental history of hypertension, obesity and diet, anxiety scores predicted changes in women’s blood pressure over a period of three years (Markovitz, Matthews, Wing, Kuller, & Meilahn, 1991). In a second study, they found that after controlling for age, obesity, glucose intolerance and other possible confounding factors, again anxiety predicted hypertension over a 20 year period (Markovitz, Matthews, Kannel, Cobb, & D’Agostino, 1993).
If truly anxious subjects are associated with hypertension, what is the likely relationship between repressive denial and cardiovascular problems? Shedler conducted three studies which confirmed that the mental health reported by repressive deniers is, in fact, illusory. In the studies repressive deniers were (a) judged psychologically distressed by clinicians, (b) shown to have verbal associations which betrayed the existence of psychological defensiveness, and (c) shown to have significantly higher cardiovascular rate pressure product (RPP) than non repressives on all tests (Shedler et al., 1993).
Since the denial of anxiety and negative affect by repressive deniers can confound research findings, there is a need for mental health scales to differentiate (a) between the true non-anxious and the repressive deniers, and (b) between repressive denial, and repressive coping style.
Once an event or situation has been appraised as stressful, individuals apply a range of coping strategies which may be emotion-focused as well as problem-focused (Lazarus & Folkman, 1984). For example, problem-focused strategies include active coping and suppression of competing activities, both are processes of taking action to overcome or circumvent the stressor. Emotion-focused strategies include positive reappraisal of the situation’s meaning, and focusing on and venting of emotions, so as to reduce negative effects. Thus emotion-focused strategies aim to cope with the distress emotions, rather than dealing with the stressor itself (Lazarus & Folkman, 1984).
Researchers report that certain coping strategies are maladaptive and result in ill health. For example, a factor analytic study of coping styles and the MMPI-2 content scales found a strong positive association between emotion-oriented coping strategies and various measures of psychopathology (Endler, Parker, & Butcher, 1993). Behavioural and mental disengagement, self-blaming, and emotional regulation were associated with severe distress, whereas seeking social support for instrumental reasons and cognitive restructuring strategies resulted in lower levels of distress (Jeney-Gammon et al., 1993).
Its not hard to conceptualise that repressive deniers, who exalt rational behaviour, in accordance with their own internalised values, are quite likely to find dysphoric feelings threatening. Hence their attempts to convince themselves that they are not upset when their physiology belies it, may in fact interfere with both their problem-focused and emotion-focused coping (Lazarus & Folkman, 1984; Weinberger, 1990), exposing them to a higher risk of ill health. Longer-term survivors of AIDS have been shown to use more problem solving strategies, and are characterised as having more flexible coping strategies than non-survivors (Temoshock, Mandel, Moulton, Solomon, & Zich, 1986).
While repressive denial may result from an attempt to avoid facing reality, and coping styles refer to different ways of dealing with events, the third variable to be examined, locus of control (LOC), may relate to concepts closer to the core of personality as it relates to the belief system of an individual.
Locus of control
Derived from social learning theory, the concept of LOC relates to an individual’s belief regarding who or what is responsible for outcomes. People with an internal LOC are characterised by their belief that what happens to them is a consequence of their own actions and is within their control. Those with an external LOC believe that what happens to them is related to external events, powerful others and chance, and thus beyond their control (Lefcourt, 1983). Research indicates that people with an internal locus of control tend to have more adaptive behaviours, take a more active interest in their health care, experience more positive psychological outcomes (less depressed and anxious), and enjoy better physical health than those with an external LOC (Oberle, 1991).
In a prospective five year follow-up study of women’s health, Lawler and Schmied (1992) used measures of LOC, stress levels, hardiness and cardiac reactivity for 32 women who had participated in a previous experiment, and compared them with measures of illness in the preceding 12 months. They found that the psychological variable most predictive of good health was internal LOC. The result of multiple regression analyses revealed that increases in stress for women with external LOC was related to subsequent illness-frequency, but not for women with internal LOC. In stepwise regression, the interaction of (LOC x stress) significantly predicted illness frequency, while LOC alone predicted illness severity. They also found that the physiological variable that most consistently predicted good health was resting systolic blood pressure (Lawler & Schmied, 1992).
A longitudinal study by Wallagen and her colleagues investigated the relationship between internal health locus of control (IHLC) and physical functioning over a 6-year period, in 356 adults with a mean age of 72. They found that for the men in the sample, IHLC had an interactive effect with mean function scores at baseline. There was no relationship between IHLC and functioning at follow-up, for men with mean function scores at baseline, however IHLC had an increasingly positive relationship with functioning at follow-up, as baseline function scores decreased. On the other hand for women in the sample, IHLC had a strong direct relationship with functioning at follow-up. These findings indicate that men who had poorer initial function, but maintained their internality achieved better health outcomes (Wallhagen, Strawbridge, & Kaplan, 1994). The measure of IHLC in this study consisted of only two items from Wallston’s six item Form B of the Multidimensional HLC scale (Wallston, Wallston, & De Vellis, 1978), and is open to criticism for being a poor measure of LOC.
Reynaert and colleagues conducted a more thorough experiment involving 34 hospitalised patients meeting the DSM III-R criteria for major depression, and 18 healthy controls, matched for mean age and sex. Both groups underwent a series of psychological, physiological and haematological laboratory tests, for immune major organ functions. Amongst the psychological measures, three LOC scales were used, Levenson’s LOC (Levenson, 1981), Rotter’s general one-dimensional LOC scale (Rotter, 1966) and Wallston and Wallston Multidimensional Health Locus of Control Scale (Wallston et al., 1978). The resulting data, analysed using t-tests and Discriminant Function Analysis, revealed that the major depression group had a significantly lower NK cell count than the control group. However, depressives with internal LOC had nearly the same NK cell count as the control group, whereas depressives with external LOC had extremely lowered NK cell counts (Reynaert et al., 1995). The study shows clearly that amongst depressed patients, those who have decreased cellular immunity are those who have a conviction that they have lost control.
Reynaert suggests that earlier studies investigating depression and immunosuppression should perhaps be reassessed in the light of this new paradigm. The question that needs to be asked, is whether the immunosuppressive effects observed in earlier studies are due to depression or due to the subphenomenon of helplessness, related to loss of control (Reynaert et al., 1995)? In any case these recent studies not only illuminate previous findings but provide a strong and convincing case for the effect of LOC on immunity and disease outcome.
Psychological Interventions that may Modulate Disease Onset and Progress.
The promise of psychoimmunology to deliver psychological interventions to delay disease onset and progression is already being realised (Gawler, 1995). Practitioners, Medical Centres, Hospitals and many universities already offer services like biofeedback, hypnosis, guided imagery, relaxation training and other psychological interventions to patients with cancer, HIV, AIDS, asthma and cardiovascular disease (Lazar, 1996).
A recent review by Edelman and Kidman (1997) from Sydney University of Technology indicated that the idea that psychological interventions might affect the course of cancer is highly contentious, and that few studies have been done in this area, and of those that exist, the findings are also inconsistent (Edelman & Kidman, 1997). In marked contrast, Meyer and Mark (1995) indicated that there were over 100 studies of psychosocial interventions with cancer patients published in peer reviewed journals. They conducted a meta-analysis of randomised experiments and concluded that the findings were significant enough, and that psychosocial interventions had been proven effective enough to conclude that further research on treatment effectiveness was not warranted (Meyer & Mark, 1995). William Redd (1995), of the Memorial Sloan-Kettering Cancer Center, agreed with Meyer’s contribution to the validation of psychosocial interventions, but suggested that there was still a need to identify the effective components of interventions. Redd (1995) advocated that instead of bigger or more replication studies, a more fruitful approach might be to conduct more focused analyses of specific biological changes, associated with particular psychological conditions.
There is clear value in Redd’s (1995) suggestion, for example, T cells and NK cells have consistently showed that they are clearly associated with behavioural interventions. A study assessing the enhancement of relaxation and social contact in 45 geriatric residence found that there was enhanced NK cell activity associated with relaxation (Kiecolt-Glaser et al., 1985). This is of particular interest because of the antitumor and antiviral functions associated with natural killer cells. In another (18 month) study of immunological responses of breast cancer patients to behavioural interventions, researchers provided stage 1 breast cancer patients with relaxation, guided imagery and biofeedback training. Thirteen subjects were randomly assigned to two groups, an immediate treatment and a delayed treatment (control group). Pathology reports were obtained to confirm diagnoses, tissue type, and immune measures. Training consisted of a set protocol including Jacobsonian relaxation training, guided imagery with the freedom to create their own images around the development of health-promoting processes in their bodies. EMG biofeedback session were used twice weekly until a frontal muscle EMG level of < 1.75 v/RMS was achieved for 200 seconds over two sessions. The results showed significant effects on T cell populations, including NK cells, and minimal effects on antibodies (Gruber et al., 1993), thereby replicating a previous study on metastatic cancer patients by the same research centre (Gruber, Hall, Hersh, & Dubois, 1988) (cited in Gruber et al. 1993). By conducting research which associates particular interventions with specific immune system changes, a clearer picture of the validity of interventions in given circumstances can be ascertained, ensuring that interventions are effectively and parsimoniously applied.
Despite two decades of research, it is difficult to draw cause and effect conclusions on the relationship between the mind, the immune system and disease. This is not to say that cause and effect relationships do not exist, but that it is harder to demonstrate, given that (a) true experimental designs cannot be used in this area of research, (b) many factors, biological, genetic and psychological, interact to produce health outcomes, and (c) definitions of stress and other psychological variables vary between studies, as do the method of measurement and measuring instruments, often resulting in apparent contradictions in the findings. Nonetheless, it is becoming clearer that certain psychological factors, three of which are reviewed here, are related to immune changes and ill health.
As discussed in this review, repressive denial has been shown to confound mental health scales and researchers should consider controlling for its effects. Reynaert’s suggestion that the immunosuppressive effects observed in earlier studies of depression may be due to the subphenomenon of helplessness, related to loss of control, and not depression itself (Reynaert et al., 1995), is particularly valid in the light of other findings in the LOC studies reviewed, which identify LOC as a predictor of health outcome. What is missing in the literature are studies investigating the links between repressive denial, immune function and cardiovascular function, and further studies in that area may illuminate the consequences of repressive denial. There is also a shortage of studies investigating coping styles, immune functions and health outcomes.
Studies may focus on the effectiveness of intervention methods which improve the adaptive coping styles of individuals, and its effects on immunity and health. LOC is derived from a person’s belief system, and those with internal LOC have been shown to use more positive and adaptive coping strategies. It has also been shown to be the best predictor of health, hence more research is required into interventions that offer the most effective ways of promoting an internal LOC.
The literature discussed in this review leads to the suggestion that psychosocial and behavioural interventions currently being applied are effective in alleviating emotional pain and in positively affecting the immune system. Future developments and applications of interventions may in the long run prove to be the best way of providing the missing causal relationships to disease modulation (Cohen & Herbert, 1996), while in the mean time providing a valuable clinical adjunct to the medical model. This may require an effort from institutions to involve practitioners in research, and use the huge volume of day to day interventions in meaningful research programs.