Stress and illness: the structure of a belief system

Tagged:  

Research on stress conducted earlier in this century was important because it suggested that illness and disease were a function not only of pathogenic agents, but the adaptive reactions of the individual as well (Hinkle, 1973). In response, scientific and medical research began exploring the interaction between individuals, pathogenic agents, and the environment. Since this early period, however, stress has proven to be a difficult construct to assess and measure objectively (Edwards & Cooper, 1988; Harris, 1989; Hinkle, 1973; Leventhal & Tomarken, 1987; Wong, 1990; Young, 1980). A plethora of measurement approaches and confusing and contradictory definitions for stress have been proposed. Despite the problematic scientific status of the stress construct it appears that stress has become an important concept in everyday discourse. Within Western cultures the link between stress and illness has become an important social fact (Blaxter, 1983; Harris, 1989; Pollock, 1988; Young, 1980).

Several recent studies on lay beliefs about illness suggest that stress has become an important construct for explaining and interpreting illness. Pollock (1988) interviewed 114 British adults about the nature of health and illness and found that these respondents believed that stress was an important cause of illness. Respondents most strongly associated stress with illnesses such as heart attacks and "nervous breakdowns". An illness such as cancer, however, was not thought to be particularly related to stress. Blaxter (1983) interviewed 46 middle-aged British working-class women about the potential causes of a variety of illnesses. The causes mentioned by these women could be grouped into a number of basic categories. The five most frequently mentioned categories were: 1) infections, 2) heredity or familial tendencies, 3) agents in the environment (e.g., poisons, working conditions, climate), 4) drugs or the contraceptive pill, and 5) stress, and worry. Illnesses rated as particularly susceptible to stress were migraines and headaches, stomach disease, and heart disease. In this group of respondents as well, cancer was an illness not thought to be linked causally to stress. Pill and Stott (1982) studied lay beliefs about the causes of various illnesses in 41 British working class women. The concept of stress was the fourth most frequently mentioned cause; listed after "germs", lifestyle, heredity, and before environmental factors and individual susceptibility. Hunt, Browner and Jordan (1990), in a study on the illness beliefs of 12 American women diagnosed with hypoglycemia, found that stress was identified as a central causal factor in the development of this health problem in 10 of the subjects. Affleck, Tennen, Croog and Levine (1987), in a study on causal attributions and perceived control in 269 American men recovering from heart attacks, found that stress was identified most frequently as the cause for the illness.

Harris (1989) conducted open-ended interviews with 47 English speaking American adolescents and adults about potential causes of illness and injury. Stress was found to play a major role in a system of lay beliefs about the origins of disease and illness. The majority of participants in the Harris study viewed stress as an important cause in a variety of illnesses: "the body, especially in its nervous, digestive, and circulatory systems, responds to internal or external (environmental) stress by malfunctioning and thereby initiating disease processes. On the other hand, stress is linked to social situations in which relations among people are not what they should be and in which people do or feel as they ought not. Such situations are by nature potentially toxic, putting the participants at risk of mental and bodily stress" (Harris, 1989, p. 11). For the participants in the Harris study, stress was a concept sufficiently broad to "explain" a vast array of diseases and illnesses.

Popular beliefs about stress may have a profound impact on the individual's perceptions and experience of illness. Lay beliefs are used by the individual not only to interpret the nature of the threat a particular illness may pose, but to determine the type of action the individual might use to mitigate this threat. Thus, belief systems play an important role in determining whether an individual will seek medical treatment, the quality of the interaction between the individual and health professionals, and the type of compliance with the health professional's advice (Blumhagen, 1980).

The work of Affleck et al. (1987), Blaxter (1983), Harris (1989), Hunt et al. (1990), Pill and Stott (1982) and Pollock (1988) suggest that stress has come to be viewed by many individuals in Western societies as an important factor in the etiology of disease. There would also appear to be some consistency in the types of illnesses that are linked with stress (e.g., heart disease but not cancer). Important questions remain to be answered, however, about the stability and overall structure of lay beliefs regarding the relationship between stress and illness. The studies cited above did not, in most cases, set out to examine specifically beliefs about the relationship between stress and illness. The small sample sizes, the widely divergent interview strategies, or the narrow range of health problems used in these studies limit the inferences that can be drawn about the structure of beliefs on the relationship between stress and illness. Study 1 was an attempt to systematically examine lay beliefs about the relationship between stress and a diverse set of health problems.

Study 1

METHOD

Subjects

The subjects in this study were 347 undergraduates (102 males and 244 females; the gender of one subject was not given) attending a large introductory psychology class in an Ontario university. The mean age was 20.03 (SD = 2.09) for males and 19.93 (SD = 3.85) for females.

Procedure

Participants in the present study were asked to complete a health questionnaire at the endof a regularly scheduled class. Questionnaires were completed anonymously and all participants were volunteers.

Measures

Participants were given a list of 26 health problems and asked to evaluate the potential causal relationship between "stress" and each health problem on five-point Likert scales, ranging from "no relationship" to a "strong relationship". The 26 health problems were taken from a comprehensive list of common illnesses in the Older American Resources and Services Multidimensional Functional Assessment Questionnaire (OARS; Fillenbaum & Smyer, 1981).

RESULTS

Mean ratings on the impact of stress for each of the 26 health problems are presented in Table 1. Items are listed in descending order from health problems thought most related to stress to those problems thought least related to stress. The health problem rated as most related to stress was high blood pressure (mean = 4.40), while the problem rated as least related to stress was polio (mean = 1.60). Table 1 also presents mean ratings separately by gender. Two-tailed t-tests were conducted to compare male and female ratings of each item with p < .05/26 (Bonferroni correction). Females evaluated skin disorders and asthma to be significantly more related to stress than males. In general, females rated the various health problems more related to stress than males. The mean rating for all 26 health problems combined was 2.76 for females and 2.49 for males [t(344) = 3.10, p = 0.002].

A principle components analysis with varimax rotation was conducted with the 26 health ratings. This analysis will determine whether ratings for certain health problems are associated. Two factors emerged with eigenvalues greater than unity and meeting scree test criteria (Cattell, 1978). The first factor had an eigenvalue of 8.68 and accounted for 33.4 percent of the variance. The second factor had an eigenvalue of 3.13 and accounted for 12.0 percent of the variance. The ratings of five health problems (glaucoma, emphysema, liver disease, urinary tract disorders, and anemia) cross-loaded low (below 0.30) on both factors and were excluded from subsequent factor analyses. A second principle components analysis with varimax rotation was conducted with the remaining 21 ratings of health problems. Two factors emerged with eigenvalues greater than unity and meeting scree test criteria. Table 2 presents the factor loadings (above 0.35), eigenvalues, and the percentage of variance explained for each factor. The first factor accounted for 37.1 percent of the variance and had items that were the most debilitating, severe, and rare health problems listed. This factor was labelled "infrequent or severe health problems". The second factor accounted for 16.4 percent of the variance and had items related to more common or less severe health problems. The second factor was labelled "common health problems".

Separate item means were calculated for the 12 items that loaded on the first factor (infrequent or severe health problems) and the 9 items that loaded on the second factor (common health problems). The item mean for the items loading on the first factor was 2.04 (SD = 0.89) and the item mean for the items loading on the second factor was 3.67 (SD = 0.90). The item mean for the items on the first factor was significantly lower than the item mean for items on the second factor [t(346) = 31.93, p < .001].

DISCUSSION

The subjects who participated in Study 1 appear to share similar ideas about stress and illness with adults in other Western societies (e.g., Affleck et al., 1987; Blaxter, 1983; Harris, 1989; Hunt et al., 1990; Pill & Stott, 1982; Pollock, 1988). In the present study, health problems such as high blood pressure, ulcers, heart trouble, stroke, intestinal disorders, speech impairment, and skin disorders were clearly identified by young adults as particularly vulnerable to stress. On the other hand, relatively infrequent health problems, such as polio, multiple sclerosis, muscular dystrophy, cerebral palsy, tuberculosis, and cancer were rated as relatively uninfluenced by stress. Stress was conceptualized as being related to relatively uninfluenced by stress. Stress was conceptualized as being related to relatively common and less severe types of health problems (e.g., skin disorders, ulcers, high blood pressure). Severely debilitating or infrequent health problems (e.g., polio and muscular dystrophy) were not viewed as being particularly related to stress. In general, females rated the various illness items as more vulnerable to stress than did males. This result is consistent with research that has found females and males to differ in the perception and experience of health problems. Females are more likely to report health problems than males, and the severity of the reported symptoms is generally higher in females than males (Bishop, 1984; Mechanic, 1976).

Although the pattern of illnesses linked by respondents to stress was consistent with the results of several previous studies using various interview strategies (Blaxter, 1983; Harris, 1989; Pill & Stott, 1982; Pollock, 1988), it was important that the generalizability of the two factor model be examined with different populations. Study 2 examined the factor structure of the ratings of health items reported in Table 2 with a sample of normal adults. An additional problem that needed to be investigated is the possibility that some of the items loading on the first factor with the young adult sample (infrequent or sever health problems), were the result of the subject's lack of knowledge about these particular health problems. That is, items loading on the first factor may indicate the subject's unfamiliarity with the health problems, rather than reflect a belief that stress has little impact on these health problems. To test this possibility, Study 2 also included a sample of health professionals likely to be knowledgeable about the various health problems.

Study 2

METHOD

Subjects

This study included two different groups of subjects. The first group of subjects were male and female adults recruited from the passenger lounges at the Central Railway Station of a major Canadian city (normal adult sample). Respondents were asked to volunteer for a study on health. Only English-speaking Canadian residents were included in the study. The sample consisted of 82 males (mean age = 38.91, SD = 8.59) and 119 females (mean age = 41.24, SD = 9.22).

The second group of subjects were male and female nurses and physiotherapists recruited from the cafeteria of a large hospital in a major Canadian city (health profession sample). Respondents were also asked to volunteer for a study on health. The sample consisted of 31 males (mean age = 35.32, SD = 11.13) and 169 females (mean age = 33.28, SD = 8.85).

Procedure

After agreeing to participate in the study, subjects were asked to complete the health questionnaire used in Study 1. Subjects completed this measure individually. The investigators attempted to make the testing atmosphere as quiet and relaxed as possible.

Measures

Participants were given a list of 26 health problems (described in Study 1) and asked to evaluate the potential causal relationship between "stress" and each health problem on five-point Likert scales (same procedure as Study 1).

RESULTS

A principal components analysis with varimax rotation was conducted with the ratings of the 21 health problems listed in Table 2 for the normal adult sample and the health profession sample. In each sample two factors emerged with eigenvalues greater than unity and meeting scree test criteria. Table 3 presents the factor loadings (above 0.35), eigenvalues, and the percentage of variance explained for each factor for the two samples. The first factor in both samples had items that were either the debilitating, severe, or rare health problems. This factor was labelled as an infrequent or severe health problem factor. The second factor in both samples had items related to more common or less severe health problems and was labelled as a common health problem factor.

Congruence coefficients (Armenakis, Field & Wilmouth, 1977) were calculated to compare the two factor solution of the health items from the undergraduate, normal adult, and health professional samples. These congruence coefficients are presented in Table 4 for all possible comparisons of samples. Significance levels (p < .001) for the coefficients were determined from values presented by Cattell (1978). For all comparisons among the factor structures of the three samples, congruence coefficients along the upper to lower diagonal were very high and significant. The reverse was true of congruence coefficients along the lower to upper diagonal. These results suggest that the factor structure of the three samples were virtually identical.

Separate item means were calculated for the 12 items that loaded on the first factor (infrequent or severe health problems) and the 9 items that loaded on the second factor (common health problems) for the normal adult sample. The item mean for the items loading on the first factor was 1.99 (SD = 0.79), while the item mean for the items loading on the second factor was 3.87 (SD = 0.63). These two item means were significantly different [t(200) = 32.32, p < .001]. Results for the health profession sample were virtually identical. The item mean for the items loading on the first factor was 1.99 (SD = 0.77), while the item mean for the items loading on the second factor was 3.82 (SD = 0.73). These two item means were also significantly different [t(199) = 31.32, p < .001].

GENERAL DISCUSSION

The normal adults who participated in Study 2 shared virtually identical ideas with the undergraduates used in Study 1 about the relationship between stress and particular types of health problems. Individuals in the health profession sample, with special knowledge about various health problems, held a set of beliefs about the relationship between stress and illness identical to the undergraduate and normal adult samples. Infrequent and/or severe health problems, such as polio, multiple sclerosis, muscular dystrophy, cerebral palsy, tuberculosis, and cancer were rated by all three groups as relatively uninfluenced by stress. Relatively common or less debilitating health problems, such as high blood pressure, ulcers, heart trouble, stroke, intestinal disorders, and skin disorders were clearly identified by all three groups as particularly related to stress.

The strong association between stress and relatively common and everyday health problems is a particularly interesting feature about the beliefs held by participants in the present study. Stress is believed to find expression in the everyday world of skin irritations, stomach problems, and high blood pressure. Infrequent or severe disorders, such as cancer or muscular dystrophy are perceived to be unrelated to stress. The work of Blaxter (1983), Harris (1989), Pill and Stott (1982), and Pollack (1988) suggests that the more severe or debilitating health problems may be perceived to be related to other "causes" like "heredity" or "germs". Future research might attempt to disentangle beliefs about the potential effects of causes such as stress, germs, and heredity on specific types of health problems. It might also be useful to examine how beliefs about stress influence other attitudes and beliefs. Are there particular types of activities or situations perceived to be more likely to create stress? Are these activities associated with particular types of health problems? The relationship between beliefs about stress and personality variables might also prove to be a fruitful topic for future research. In particular, research might focus on personality variables like neuroticism or alexithymia--constructs found to play an important role in the experience and perception of illness (Costa & McCrae, 1987; Taylor, 1984; Taylor, Bagby & Parker, 1991).

Health professionals should be aware of lay conceptions and meanings regarding the concept of stress. This word is frequently used by health professionals without regard for the powerful connotations it may hold for the individual (Young, 1980). Researchers earlier in this century focussed their attention on stress because it seemed to suggest, in part, that perceptions and beliefs about experiences played a significant role in manifestations of illness and disease. It is somewhat ironic that while subsequent research has found the stress concept to be a problematic construct to investigate, stress now appears to be integral to lay perceptions and beliefs about a number of health problems.