Interest has grown recently in a cognitive/perceptual model of hypochondriasis. This model postulates that current life stresses can activate a previously latent cognitive schema about health and disease that is faulty, unduly alarming, or pessimistic. A confirmatory bias then leads the hypochondriacal individual to attend selectively to information supporting this schema and ignore disconfirmatory information. In this process, benign bodily sensations are mistakenly attributed to the suspected serious disease. These sensations are thereby amplified, substantiating the individual’s suspicion that he or she is seriously ill. This model helps to explain hypochondriacal patients’ striking refractoriness to standard medical reassurance; therefore, it has important implications for the strategies that physicians use to educate and reassure these patients.
Hypochondriacal patients describe a sense of imminent jeopardy and vulnerability to disease. Their sense of risk seems exaggerated, and they appear to be unable to tolerate even the unlikely possibility of developing many diseases, some of which are extremely unlikely. In addition, they are highly resistant to reassurance; appropriate information, education, and explanation only transiently diminish their disease conviction and disease fears. This sense of heightened physical risk appears to be circumscribed and limited to disease; it seems not to encompass other physical dangers and threats that may be equally or more likely, such as accidents, injuries, and criminal victimization. That is, compared with nonhypochondriacal individuals, hypochondriacal patients apparently feel more at risk of becoming sick but not more at risk of being harmed, injured, or killed in other ways.
There has been little empirical substantiation of this observation. Wells and Hackmann (1), using qualitative interviews, reported that hypochondriacal thinking involves on overestimation of the likelihood of illness and death. Hadjistavropoulos et al. (2) found that student volunteers who were rated high in health anxiety regarded themselves to be at greater risk of medical complications than students with less health anxiety. We have found in previous work (3) that patients with DSM-III-R hypochondriasis are more likely than nonhypochondriacal patients to attribute ambiguous bodily discomfort to disease.
In the current study, we hypothesized that 1) hypochondriacal medical outpatients would perceive themselves to be more susceptible to a variety of medical diseases than nonhypochondriacal medical outpatients, 2) the two groups would not differ significantly in their appraisal of risk or likely harm from nonmedical physical dangers, and 3) the sense of susceptibility to disease would be correlated with the tendency to amplify benign bodily sensations.
The study population was composed of ambulatory medical patients participating in a 4-year, naturalistic, longitudinal study of hypochondriasis. The study protocol was approved by the institutional review board, and each patient gave signed, informed consent.
Consecutive patients visiting a primary care clinic on randomly chosen days completed self-report hypochondriasis questionnaires. Those scoring above a predetermined cutoff score on this instrument, plus a random sample of all those who scored below this cutoff, were given a research battery that included a structural diagnostic interview for DSM-III-R hypochondriasis. Those patients determined to be hypochondriacal according to this interview on two separate occasions constituted the hypochondriacal study group (N=56). The random sample of all patients who scored below the cutoff on the hypochondriasis screening questionnaire constituted the comparison group (N=127). The details of this procedure and the numbers of patients accrued at each step have been published elsewhere (4). The patients’ medical records were audited to establish medical morbidity by a study physician (E.D.B.) who was blind to the other research data.
The study was conducted in the general medical clinic of Massachusetts General Hospital. The clinic serves 29,000 patients, who make more than 50,000 visits annually. It is staffed by 36 attending physicians and 65 house officers, all of whom provide primary care.
Variables and Their Measurement
Hypochondriacal symptoms were assessed with the Whiteley Index (5–7) and the Somatic Symptom Inventory (8–10). The Whiteley Index consists of 14 hypochondriacal attitudes and concerns, scored on an ordinal scale from 1 to 5, and contains three factors: disease fear, disease conviction, and bodily preoccupation. Its psychometric properties have been described elsewhere (5–7). The clinical diagnosis of DSM-III-R hypochondriasis was made with the Structured Diagnostic Interview for Hypochondriasis (11), whose reliability and convergent and concurrent validity have been described elsewhere (11). The DSM-III-R diagnosis of hypochondriasis specifically excludes hypochondriacal symptoms that are better explained by a comorbid psychiatric disorder or by major medical illness.
Somatization was assessed with a 26-item Somatic Symptom Inventory, composed of somatic symptoms common to the Hopkins Symptom Checklist somatization subscales and the MMPI hypochondriasis subscale (8, 9). Responses are scored on an ordinal scale from 1 to 5. Its psychometric properties have been reported elsewhere (10).
The Somatosensory Amplification Scale (12, 13) is a 10-item questionnaire assessing self-reported sensitivity to normal physiological states and minor bodily sensations that are not generally regarded as symptomatic of serious disease. Its psychometric properties have been described in previous work (12, 13). It is significantly associated with DSM-III-R hypochondriasis, even after controlling for psychiatric comorbidity (13), and predicts the persistence of hypochondriacal symptoms in transiently hypochondriacal patients (7).
The estimation of risk and susceptibility to disease, health hazards, and physical harm was assessed with the Comparative Risk Questionnaire developed by Weinstein (14, 15). Respondents are asked to rate their chances of succumbing to 32 health and safety risks "compared to other men/women my age." They are instructed to omit items dealing with diseases or events that have actually occurred to them. Responses are rated on a 7-point scale on which 1=much below average, 4=average, and 7=much above average. Thus, higher scores indicate a pessimistic bias, a perception of greater risk. Although any one individual may actually be above or below average in his or her likelihood of encountering particular diseases or dangers, the comparative judgment of a large group on each item should still be average.
Previous work has revealed an optimistic bias in that most people consider themselves less likely than their peers to encounter a wide range of physical hazards, sources of harm, and health and safety risks (16–20). Optimistic biases have been reported for a wide range of medical problems, including perceived risk of heart disease (16, 21), lung and other cancers (17–19, 22), and alcoholism (19, 20). These optimistic biases are found in diverse populations and are largely unrelated to age, sex, or level of education, although they are directly correlated with past experience and perceived controllability.
Aggregate medical morbidity was assessed on the basis of patients’ medical records. All medical diagnoses were rated for seriousness according to explicit, preestablished criteria (including extent of tissue damage, clinical progression, complications, recurrence, and threat to life) (23). Each diagnosis was assigned a weight of 1–4, and then major and moderately serious diagnoses were summed as an index of aggregate morbidity. In previous work (23), this method had moderate interrater reliability and had a correlation of r=0.41 (p<0.0001) with primary physician ratings of morbidity.
First, descriptive statistics were calculated for the sociodemographic variables and aggregate medical morbidity of the hypochondriacal and nonhypochondriacal groups. Second, we used general linear modeling to compare the total Comparative Risk Questionnaire scores of the two groups, controlling for the observed significant differences between groups on social position and race. Third, the two groups were compared on the 32 individual items of the Comparative Risk Questionnaire and the total scale scores by using the t test for independent samples. Finally, intercorrelation matrixes were calculated among the set of symptom scale variables, and the total score on the Comparative Risk Questionnaire was correlated with the measure of medical morbidity. Alpha level was set at 0.05 as the criterion for statistical significance.
Fifty-six patients meeting DSM-III-R diagnostic criteria for hypochondriasis were interviewed, along with 127 nonhypochondriacal comparison patients. Seven patients were interviewed by telephone (they did not differ significantly from those interviewed in person on any of the independent variables). The mean age of the hypochondriacal patients was 54.2 years (SD=14.1), compared with 58.6 years (SD=16.4) in the nonhypochondriacal sample (F=1.72, df=1, 181, p=0.09). Forty-three (76.8%) of the hypochondriacal patients were women, compared with 87 (68.5%) of the nonhypochondriacal patients (χ2=1.30, df=1, p=0.26). There was a significantly higher proportion of African Americans in the hypochondriacal group (19.6% [N=11] versus 6.3% [N=8]) (χ2=7.92, df=2, p=0.02), and the hypochondriacal patients were of significantly lower social position (χ2=26.5, df=4, p=0.001). There was no statistically significant difference between the groups on aggregate medical morbidity (t=1.38, df=170, p=0.17).
t1 presents the Comparative Risk Questionnaire scores for the patients with and without hypochondriasis. The individual items are listed in descending order of effect size for perceived risk, from arthritis, to which they felt most susceptible, to food poisoning, to which they felt least susceptible. The total score of the hypochondriacal group was significantly higher than that of the comparison group, after we adjusted for social position and race (F=7.67, df=1, 179, p=0.006). This indicates that in the aggregate, the hypochondriacal patients perceived themselves to be more likely to encounter these health threats. Their mean score was below 4, however, indicating that as a group they still evidenced an optimistic bias. In other words, although the hypochondriacal patients were significantly more pessimistic about their health than the patients without hypochondriasis, both groups saw themselves as less likely to encounter these health threats than others of their age and sex. Hypochondriacal patients perceived themselves to be at significantly greater risk than those without hypochondriasis of arthritis, ulcer, insomnia, diabetes, being overweight, pneumonia, deafness, tooth decay, being mugged, and making a suicide attempt.
The two groups did not differ significantly in their risk appraisal of cancer or several other major medical disorders such as heart attack and stroke. Since clinical impression suggests that hypochondriacal patients are often afraid of these major medical disorders, we tabulated the primary disease conviction or fear each patient presented in the Structured Diagnostic Interview for Hypochondriasis. In t2 it can be seen that cancer and heart disease were the most frequent specific disease convictions or fears expressed. In one-third of the patients, the primary disease concern was a nonspecific vague and generalized abnormal state or a highly personalized, idiosyncratic bodily change.
On the basis of clinical impressions, we suspected that the hypochondriacal patients’ sense of vulnerability would be circumscribed and limited to disease and illness and that they would not feel as much in jeopardy from other physical threats and dangers such as accidents and criminal victimization. Therefore, we looked at six items dealing with these nonmedical dangers (broken bones, being mugged, automobile accident, being a murder victim, poison ivy, and food poisoning). The total scores on these six items of the hypochondriacal and nonhypochondriacal patients did not differ significantly (t=7.44, df=180, p=0.15).
We also attempted to determine which aspects of hypochondriasis were most closely associated with risk perception. The correlation between the total Comparative Risk Questionnaire score and bodily amplification was r=0.49 (df=77, p=0.0001); the correlation with the somatization scale was r=0.32 (df=180, p=0.001); and the correlation with the measure of hypochondriacal attitudes and beliefs was r=0.28 (df=181, p=0.001). Risk perception was negatively correlated with aggregate medical morbidity, and the magnitude of the association, although statistically significant, was low (r=–0.21, df=168, p<0.005).
As interest in cognitive and behavioral interventions for hypochondriasis grows, it is increasingly important to characterize hypochondriacal cognitive distortions, misunderstandings, and misjudgments. Altering and modifying these then become treatment aims. Since hypochondriacal patients feel unusually vulnerable to disease and are unable to tolerate the possibility (even if remote) of becoming ill, this could constitute an important focus of cognitive restructuring. A better understanding of hypochondriacal cognition could also guide primary care physicians in providing these patients with medical information, reassuring them, and explaining the results of laboratory tests and procedures. For the purposes of both psychiatric treatment and medical management, then, it becomes important to learn more about how the patient with hypochondriasis thinks about health, assesses and evaluates bodily sensations, and understands disease.
The findings suggest that hypochondriasis encompasses an exaggerated sense of susceptibility to disease and that this sense of susceptibility is limited to disease and does not encompass other physical dangers, such as those posed by accidents and criminal victimization. Of the component symptoms of hypochondriasis, the tendency to amplify bodily sensations is most closely associated with this exaggerated risk appraisal. Since this is a cross-sectional association, the direction of the relationship remains unclear. A plausible interpretation is that an acute sense of the imminence of disease leads to self-monitoring, bodily hypervigilance, and amplified somatic sensations. The findings are equally compatible with the alternative interpretation, however, namely, that patients whose bodily sensations are amplified come to feel more at risk of disease because they are bombarded with so many alarming bodily sensations.
Although hypochondriacal patients feel a greater sense of risk than nonhypochondriacal patients, they nonetheless maintain an optimistic bias and as a group perceive themselves to be less at risk than others of their age and sex. In a sense, then, it is the nonhypochondriacal patients who are manifesting a cognitive distortion, since they are more inappropriately and unrealistically optimistic and dismissive of the medical risks they face. This difference in perceived risk between hypochondriacal and nonhypochondriacal patients is not attributable to differing medical experiences in the two groups: they did not differ in aggregate medical morbidity; patients omitted items referring to a disease that they had; and total risk scores were negatively correlated with medical morbidity ratings. The inverse relationship between perceived risk and medical morbidity is surprising. Perhaps for many hypochondriacal patients the experience of illness turns out to be less fearsome in reality than they had anticipated. In prospective work (4), we have found that the remission of DSM-III-R hypochondriasis over a 5-year period was positively associated with the incidence of major medical illness during that period.
It is noteworthy that hypochondriacal patients did not feel significantly more at risk of cancer, heart disease, or stroke than nonhypochondriacal patients, although the first two were among the most commonly voiced hypochondriacal disease convictions or disease fears during the diagnostic interview. Hypochondriacal patients did perceive themselves as at risk of cancer and heart disease, but since these concerns were also very common in the nonhypochondriacal patients as well, the two groups did not differ statistically in this regard. Several factors may be at work here. First, the hypochondriacal patients encountered in medical settings may differ phenomenologically from those seen in psychiatric settings; therefore, their specific disease fears and perceived vulnerabilities may differ. For example, hypochondriacal patients with panic anxiety may be more likely to be seen in psychiatric settings and to fear sudden, acute medical catastrophes, but hypochondriacal patients without intense anxiety may be more likely to remain exclusively in medical settings and to fear more prolonged, chronic, and degenerative illnesses. Second, the differences between the results of the risk questionnaire and the diagnostic interview may be partly attributable to the differing responses obtained when people are free to volunteer their concerns in an open-ended fashion, compared with being asked to endorse specific questionnaire items.
This study has a number of limitations. First, the subjects were drawn from the outpatient clinic of an urban, academic medical center, and it is unclear to what degree the results can be generalized beyond this population. Second, although the Comparative Risk Questionnaire has been used in previous studies, further evidence of its psychometric properties is desirable, such as a components analysis to establish its factor structure and determine whether medical and nonmedical risk items constitute two distinct factors. Third, there are many possible confounding factors that may affect the relationship between hypochondriasis and risk. Psychiatric comorbidity is one important such confound. The rates of anxiety and depressive disorders are elevated in hypochondriasis (24), and the differential appraisal of risk found here could be attributed to these comorbid disorders rather than hypochondriasis per se. Finally, these findings should be regarded cautiously because the magnitude of the correlations found is low to moderate.
Nonetheless, the findings reported here are compatible with the observations that hypochondriasis entails a high degree of perceived risk from disease but not from nonmedical threats and that risk appraisal is closely associated with the amplification of bodily sensations. If these findings can be confirmed in future work, they have important implications for the ways in which primary care physicians reassure hypochondriacal patients and for the cognitive behavior treatment of hypochondriasis.
Received Sept. 15, 1999; revisions received Jan. 31 and Aug. 22, 2000; accepted Sept. 6, 2000. From the Department of Psychiatry, Brigham and Women’s Hospital; the Department of Psychiatry, Harvard Medical School, Boston; and the Psychiatry Service, Massachusetts General Hospital, Boston. Address reprint requests to Dr. Barsky, Department of Psychiatry, Brigham and Women’s Hospital, 75 Francis St., Boston, MA 02115. Supported by NIMH research grant MH-40487.