Psychic vulnerability and the associated risk for cancer

Abstract

BACKGROUND

Psychic vulnerability has been associated with a number of physical symptoms and diseases. This study was designed to estimate the incidence of cancer in a random sample of persons in the general Danish population in relation to a personality characteristic measured by the Test of Psychic Vulnerability.

METHODS

The authors examined the cancer incidence in a cohort of 5136 randomly sampled persons age > 25 years living in Copenhagen County, Denmark. The responses to questionnaires and the results of examinations, including the Test of Psychic Vulnerability, were collected during 1982–1984 and during 1991–1992. The observed numbers of cancers were compared with the numbers that would have been expected if the cohort members had experienced the same risk of cancer as the population of Copenhagen County. Regression analyses were performed with the Cox proportional hazards model to adjust for well-known risk factors for cancer.

RESULTS

A total of 403 cancers were observed, and 412.02 were expected, yielding a standardized incidence ratio of 0.98 (95% confidence interval [95% CI], 0.88–1.19). The authors did not observe a significant increase in the risk of site specific cancers. The risk for cancer was not influenced by the type of vulnerability in a multivariate analysis (hazards ratio, 1.16; 95% CI, 0.85–1.57).

CONCLUSIONS

The authors found no increased risk for cancer among psychically vulnerable persons compared with nonvulnerable persons; however, the results indicate that behavior and certain life-style factors may be determined by personality, which, in turn, may determine the risk for cancer. Cancer 2002;94:3299–306. © 2002 American Cancer Society.

DOI 10.1002/cncr.10601

Previous studies have considered personality as an independent risk factor for cancer. This hypothesis indicates that persons characterized by certain personalities or certain personality traits may be at increased risk for cancer. For instance, it has been claimed that a high degree of extraversion,1 a low degree of neuroticism,1 substability,2 rigidity,3 abnormal release of anger,4 repression,5 a high degree of social conformity,6 depression,7 low expression of anxiety,8, 9 reduced ability to express feelings,10 and being less competitive10 are personality traits associated with the risk for developing cancer.

It has been claimed that the mechanism whereby personality or personality traits lead to an increased risk for cancer is impaired function of the immune system,11 which would predispose the organism to the initiation and progression of neoplastic diseases.12 This hypothesis is supported by the finding of an increased cancer incidence among patients with various, well-defined immune deficiency states.13–15 There are two other theories for an association between personality and risk for cancer: One theory is that the personality predisposes to certain behaviors and exposure to life-style factors, such as smoking and dietary patterns, which influence the risk of cancer (the personality-health behavior model).16 The other theory is that a common underlying biologic cause determines both the personality and the risk for developing cancer (the constitutional risk factor model16) (Fig. 1). It is likely that behavioral exposure, e.g., tobacco smoking, may be one explanation for an association between the personality and the risk of cancer.

Figure 1. Hypothesis for an association between personality and risk for cancer. The figure illustrates three hypothesis for an association between personality and cancer. Model A was originally described by Ader et al.,11, 12 and model B and C were described by Smith and Gall.16

Kühl and Martini17 defined psychic vulnerability as a reaction readiness characterized by a low threshold for being influenced and a risk of inexpedient reactions in social interactions or in a psychosomatic direction. The psychic vulnerable person appears nervously agitated, indecisive, and apprehensive and lacks self-confidence; or he is compensatory self-assertive, may be ironic and aggressive, is sensitive to being slighted, and lacks indulgence toward other people's mistakes. The concept of psychic vulnerability, therefore, covers a wide span from extremely psychic robust to extremely psychic vulnerable, and every person has their own position within this continuum.

Psychic vulnerable persons are not characterized as neurotic, although some symptoms seem neurotic according to Kühl and Martini.17 It is possible that neuroticism, as described by Kissen and Eysenck,1 is the personality trait most closely resembling psychic vulnerability. Kühl and Martini investigated psychic vulnerability in a random sample of 10753 Danes who completed a questionnaire, which included a 12-item version of the original 22-item Test of Psychic Vulnerability17, 18 (see Fig. 2). Those authors found that respondents who were defined as psychically vulnerable could be distinguished from respondents who were nonvulnerable by a greater prevalence of divorced parents, a smaller social network, and greater use of public social services. In addition, the respondents who were defined as psychically vulnerable complained of diffuse physical symptoms, such as fatigue, lower back pain, and digestive problems; they were divorced or unmarried more frequently and were unemployed more frequently unemployed than nonvulnerable respondents.17

Figure 2. On the basis of responses to a questionnaire by a random sample of the Danish population (n = 10,753 participants), including this 22-item Test of Psychic Vulnerability, Kühl and Martini defined psychic vulnerability as a reaction readiness characterized by a low threshold for being influenced and a risk of inexpedient reactions in social interactions or in a psychosomatic direction.17

The Test of Psychic Vulnerability has not been used previously in investigations of risk for cancer; however, it has been used extensively to examine the correlation between this personality characteristic and various physical symptoms. It was found that psychic vulnerability was associated significantly with persistent pain after lumbar spine surgery19 and with the incidence and prevalence of upper dyspepsia20 and irritable bowel syndrome.21 Significant associations were found between psychic vulnerability and the reporting of peptic ulcer or developing peptic ulcer during the observation period.22 Persistent pain after laparoscopic cholecystectomy was significantly more frequent among psychically vulnerable persons,23 in accordance with a previous finding of a significantly decreased percentage of satisfactory outcome after cholecystectomy with increasing psychic vulnerability.24

Because psychic vulnerability has been associated with somatic symptoms and diseases, and because it has been hypothesized previously that various personality traits previously are associated with an increased risk of cancer, we used a prospective design to estimate the incidence of cancer in a random sample of 5136 persons in the general Danish population in relation to the degree of psychic vulnerability. The specific hypothesis of this study was that persons defined as psychically vulnerable are at increased risk for developing cancer.

MATERIALS AND METHODS

The initial cohort consisted of 7734 persons who were sampled randomly from the population of the western part of Copenhagen County and who were invited to participate in a health survey. The survey was conducted as a part of the international MONICA investigation.25 During the period 1982–1992, which covered the inclusion period (Table 1), the population of the county included approximately 335,000 persons age > 25 years, representing about 6% of the total Danish population.20 A total of 5812 participants (75%) accepted the invitation. We excluded 276 persons (3.6%) who were not Danish to obtain a culturally homogeneous population. Incomplete answers to any of the items constituting the psychic vulnerability score resulted in the exclusion of 272 persons (3.5%); another 24 persons (0.3%) were excluded because baseline values were missing, and 104 persons (1.3%) were excluded because a cancer other than nonmelanoma skin cancer had been diagnosed prior to the date of inclusion. Thus, the final study population was 5136 persons (88.4%).

Table 1. Sociodemographic Variables, Smoking Habits, and Alcohol Consumption in Relation to Vulnerability Score among 5136 Danes Participating in the Glostrup Population Studies, Denmark, 1982–1992

Variable	Vulnerabilitya
	Vulnerable		Nonvulnerable
	No.	%	No.	%
a Cut-off values according to gender and age are shown in Figure 1. b Significant differences in distributions between vulnerable and nonvulnerable participants (P < 0.05).
Age at entry (yrs)b
29–34	138	24.0	1101	24.1
35–44	185	32.2	1068	23.4
45–59	127	22.1	1308	28.7
≥ 60	125	21.7	1084	23.8
Genderb
Male	259	45.0	2365	51.9
Female	316	55.0	2196	48.1
Period of inclusion
1982–1984	413	71.8	3108	68.1
1991–1992	162	28.2	1453	31.9
Marital statusb
Married	344	59.8	3267	71.6
Not married	231	40.2	1294	28.4
Social classb
White collar	108	18.8	1325	29.1
Blue collar	467	81.2	3236	70.9
Tobacco smoking1b
Nonsmoker	98	17.0	1161	25.5
Former smoker	93	16.2	926	20.3
Current smoker	384	66.8	2474	54.2
Alcohol consumption (units per week)
0	107	18.6	642	14.1
1–14	348	60.5	3038	66.6
> 14	120	20.9	881	19.3
Body mass index (kg/m^2)a
Underweight (< 18.5)	21	3.7	96	2.1
Normal (18.5–25.0)	333	57.9	2549	55.9
Obese (≥ 25.0)	221	38.4	1916	42.0

The population for this study was recruited in two separate investigations that took place during 1982–1984 and 1991–1992 within the framework of the Glostrup Population Studies. Participants who took part in clinical and paraclinical examinations also completed questionnaires about their personal status and health, including the Test of Psychic Vulnerability. The participants did not complete further psychologic tests. In the Test of Psychic Vulnerability, each question is answered yes or no and is thereby assigned a score of 1 or 0, respectively. This test is based on a test from the Psychological Services of the Military in Denmark and was developed for use in surveys done by the Danish Council of Social Research in the 1970s. The test underwent the Rasch item analysis.26 A random sample of 10,753 Danes filled in the Test of Psychic Vulnerability due to their participation in one of the surveys done by the Danish Council of Social Research. Based on supplementary information collected from the physician of 1300 participants, the material was analyzed further, and it was concluded that the Test of Psychic Vulnerability was a valid measure of psychic vulnerability. Additional analyses of reinterviews of 3000 participants 1 year after the original interview showed that the test results were stable to such an extent that the Test of Psychic Vulnerability consequently was considered a reliable instrument.17 The method of data collection was identical in the two surveys.

Data on all participants in the study population were linked to the Civil Registration System for verification of the personal identification number and for information on vital statistics and migration. The Civil Registration System was established in Denmark on April 1, 1968; since that date, all Danish residents have been assigned a 10-digit personal identification number that incorporates information on gender and date of birth and permits accurate linkage of information among registries. Subsequently, the data on the study cohort were linked to the Danish Cancer Registry, which began reporting cancer incidence on a nationwide scale in 1943. Each record in this registry includes the personal identification number, date of diagnosis of the tumor, and information on the tumor. Tumors are coded according to a modified Danish version of the International Classification of Diseases, Seventh Revision,27, 28 and, since 1978, all tumors have been coded according to the International Classification of Diseases for Oncology.29 Comprehensive evaluation has shown that the Danish Cancer Registry is almost (99%) complete and valid.30

Members of the cohort were followed for cancer occurrence from the day of the health examination to the date of emigration, a cancer diagnosis (other than nonmelanoma skin cancer), death, or December 31, 1996, whichever came first. The observed numbers of cancers were compared with the expected numbers of cancers on the basis of the age specific, gender specific, and calendar year specific incidence rates of first primary tumors (other than nonmelanoma skin cancer) in the population of Copenhagen County. Multiplication of the person-years of observation by the incidence rate yielded the number of cancers that would be expected had the cohort members experienced the same risk of cancer as the prevailing risk in the population of Copenhagen County. Tests of significance and 95% confidence intervals (95%CI) for the standardized incidence ratios (SIRs), taken as the ratio of the observed to expected numbers of cancers, were calculated. Miettinen exact confidence limits were used when the observed number of end points was small; otherwise, an accurate asymptotic approximation was used.31

For the regression analyses, the Cox proportional hazards model32 was used with the SAS statistical package with the phreg procedure (version 6.12; SAS Institute, Inc., Cary, NC). Strata were defined by classifying the observation period by gender, age (ages 0–34 years, 35–44 years, 45–59 years, and ≥ 60 years), alcohol consumption (0 units per week, 1–14 units per week, or > 14 units per week), smoking habit (nonsmoker, former smoker, or current smoker), social class (white collar or blue collar), marital status (married or not married), and body mass index (normal, underweight, or obese).

The psychic vulnerability score was calculated as the overall sum of the 12 items that constituted the Test of Psychic Vulnerability (possible score, 0–12) and that were included in the analyses by dichotomization of the dimension (see Fig. 2). Arbitrary cut-off points were chosen on the basis of the results of empiric studies showing that approximately 10% of the population is characterized as psychoneurotic.17 To meet the assumption of the normal distribution, the study population was divided according to gender and age (see Fig. 2). We analyzed the data further in the following ways: 1) by creating four groups of psychic vulnerability, 2) by including the psychic vulnerability score as a continuous variable, and 3) by analyses that included all 22 items on the original Test of Psychic Vulnerability to obtain the psychic vulnerability score. The results reported in the current article originated from the dichotomized scoring on the 12-item test, because none of the further analyses changed the original results.

RESULTS

The 5136 participants in the study population contributed a total of 53,623 person-years of follow-up, with a mean follow-up of 10.4 years (range, 0–14 years). We observed a significant difference in distribution with regard to age at entry, gender, marital status, social class, tobacco smoking, and body mass index between participants who were classified as vulnerable and participants who were classified as nonvulnerable (Table 1).

A total of 403 malignancies were observed, with 412.02 expected, yielding an SIR of 0.98 (95% CI, 0.88–1.19) (Table 2). Site specific analyses revealed no significant deviations from unity with respect to SIRs; however, women had a significantly decreased risk for digestive malignancies but a significantly increased risk for multiple myeloma (SIR, 4.00; 95% CI, 1.29–9.33). These results were based on 12 women and 5 women, respectively (data not shown).

Table 2. Observed and Expected Numbers of Site Specific Malignancies and Standardized Incidence Ratios among 5136 Danes Participating in the Glostrup Population Studies, Denmark, 1982–1992

Site	Obs	Exp	SIR	95% CI
Obs: observed; Exp: expected; SIR: standardized incidence ratios; 95%CI: confidence interval. a Including International Classification of Diseases, Seventh revision (ICD-7) codes 140–205. b Including ICD-7 codes 170, 172, 175, and 177. c Including ICD-7 codes 155.0, 171, 191, 201, 200, 202, and 204. d Including ICD-7 codes 150–159 (except 155.0). e Including ICD-7 codes 160–164. f Including ICD-7 codes 140–148, 173–174, 176, 178–179, 180–181, 190, 192–197, 198–199, 203, and 205.
All sites1a	403	412.02	0.98	0.88–1.08
Hormone-related organsb	79	91.10	0.87	0.69–1.08
Virus-related and immune-related malignanciesc	109	105.31	1.04	0.85–1.25
Digestive organs (excluding liver)d	55	65.96	0.83	0.63–1.09
Respiratory organse	73	59.30	1.23	0.96–1.55
Other sitesf	87	88.83	0.98	0.78–1.21

Age-adjusted and gender-adjusted analyses did not reveal an increased risk for all cancers in relation to psychic vulnerability (hazard ratio, 1.21; 95% CI, 0.89–1.63) (Table 3). The multivariate analyses, however, changed the result only slightly (hazard ratio, 1.16; 95% CI, 0.85–1.57). A significantly increased risk for cancer, as expected, was observed among participants with increasing age. Current smokers had a significantly increased risk of cancer (hazard ratio, 1.39; 95% CI, 1.05–1.83) as well as persons who drank more than 14 units of alcohol a week (hazard ratio, 1.50; 95% CI, 1.05–2.15).

Table 3. Age-Adjusted, Gender-Adjusted, and Multivariate-Adjusted Hazard Ratios and Their 95% Confidence Intervals for Cancer Risk by Baseline Variable among 5136 Danes Participating in the Glostrup Population Studies, Denmark, 1982–1992

Variable	Persons with cancer	Age-adjusted and gender-adjusted HR	95%CI	Multivariate adjusted HRa	95% CI
HR: hazard ratio; 95%CI: 95% confidence interval. a Adjusted for gender, age, alcohol consumption, smoking habit, social class, marital status, and body mass index. b Cut-off points according to sex and age are shown in Figure 1.
Age at entry (yrs)
29–34	16	—	—	1.00	—
35–44	64	—	—	4.22	2.43–7.33
45–59	140	—	—	9.29	5.50–15.70
≥ 60	183	—	—	15.37	9.14–25.85
Gender
Female	202	—	—	1.00	—
Male	201	—	—	0.86	0.69–1.07
Marital status
Married	303	1.00		1.00	—
Not married	100	1.07	0.85–1.34	1.02	0.81–1.29
Social class
White collar	100	1.00	—	1.00	—
Blue collar	303	1.03	0.82–1.29	1.01	0.80–1.27
Tobacco smoking
Nonsmokers	69	1.00	—	1.00	—
Former smokers	89	1.28	0.93–1.77	1.26	0.91–1.74
Current smokers	245	1.47	1.12–1.93	1.39	1.05–1.83
Alcohol consumption (units per week)
0	56	1.00	—	1.00	—
1–14	250	1.18	0.88–1.59	1.16	0.86–1.56
> 14	97	1.58	1.11–2.25	1.50	1.05–2.15
Body mass index (kg/m^2)
Normal (18.5–25.0)	212	1.00	—	1.00	—
Underweight (< 18.5)	11	1.70	0.93–3.13	1.58	0.86–2.92
Obese (≥ 25.0)	180	0.87	0.71–1.07	0.89	0.73–1.10
Vulnerabilityb
Nonvulnerable	355	1.00	—	1.00	—
Vulnerable	48	1.21	0.89–1.63	1.16	0.85–1.57

DISCUSSION

The results of this study, which was based on prospective, population-based data, do not support the hypothesis that psychically vulnerable persons are at increased risk for cancer, even when the results are adjusted for some well-known risk factors for cancer. Defining personality is a complex matter due to the fact that no theory of personality is accepted universally. Previous studies have been characterized by diversity in construct and measure definition; therefore, investigations of an association between personality and risk for cancer have focused on various aspects of personality. In 1962, Kissen and Eysenck observed a higher extraversion score and a lower neuroticism score on the first version of the Eysenck Personality Inventory33 among 116 male patients with carcinoma of the lung compared with the scores from 123 control participants.1 Likewise, in Virginia, Abse and colleagues observed that 31 young persons with carcinoma of the lung were more rigid and overly controlled compared with 28 control participants, as assessed from interviews.3 In Germany, Blohmke and colleagues used the C. B. Bahnson questionnaire to question 419 men with carcinoma of the lung, 419 healthy men, and 162 men with nonmalignant lung diseases. Those authors found that the patients with lung carcinoma were less nervous and sensitive and showed a greater social conformity compared with the other groups.6 Faragher and Cooper examined 171 women with breast carcinoma and 1992 healthy women with the Bortner Type-A Behavior Inventory in a study from England. The women with breast carcinoma tended to be less competitive, less rushed, less ambitious, less able to express their feelings, less eager to get things done, slower in doing things, less able to take things one at a time, and had fewer interests outside home and work compared with the group of healthy control women.10

The observations made in the current study confirm those from our recent study of similar design,34 in which we investigated the effect of personality, as measured with the Eysenck Personality Inventory,33 on the incidence of cancer among 1052 participants in a Danish health survey during 1976–1977 (participation rate, 88%) who were followed for 20 years. Our data provided no support for the hypothesized association between personality and the risk for cancer. The results are also in line with those from several recently published, population-based, prospectively designed studies, which showed consistently that psychologic factors do not influence the risk for cancer. In a Dutch nested case–control study of personality factors and breast carcinoma development among 9705 women,35 the only indication that personality traits affected an increase in the risk for developing carcinoma was a borderline association with antiemotionality. The participation rate of the study was 34%, however, which may suggest that the study population was a selected one. We previously investigated the effect of a psychologic stressor (a child with cancer) on the incidence of cancer among 11,231 parents in a nationwide, population-based Danish study but could not confirm a hypothesized association.36 Likewise, we found no indication of an increased risk for cancer in a nationwide cohort study of 89,491 individuals in whom depression was diagnosed at a psychiatric department between 1969 and 1993.37

The current investigation of psychic vulnerability and risk for cancer has certain strengths. The cohort was population-based, and the study had a prospective design with complete follow-up; the internal comparisons in the multivariate analyses reduced the possibility of confounding. Furthermore, compared with other prospective cohort studies, we achieved a participation rate of 75%. Studies of the association between psyche and cancer often are based on unvalidated psychometric scales. We addressed this problem by using a Rasch item analysis,26 which enabled us to make comparisons between two random persons independent of which items were used in the comparison. Furthermore, we were able to make comparisons between two random persons independent of the other persons under study. Item analysis thereby postulates that the response patterns, defined by answers to different questions or items, are homogeneous across items and individuals.38

This study also had limitations. Part of the study population entered the investigation in 1992; because follow-up was terminated in 1996, these persons have not been followed for more than 4 years. The response rate to the health surveys was 75%, leaving a group of nonparticipants who, theoretically, may have differed from the participating group in health habits, personality, life-style factors, etc. It is a shortcoming, however, that the Test of Psychic Vulnerability has been used only among Danes and has never been chosen before for a study of the incidence of cancer. In the final multivariate model, we included risk factors that, initially, were not associated significantly with an increased risk for cancer. However, we do not believe that these minor limitations invalidate the results.

In summary, this large, prospective, population-based study provided no support for the hypothesis that personality, as measured by the Test of Psychic Vulnerability, is an independent risk factor for cancer. Based on the observations of this study, we were able to elucidate the theories of an association between personality and risk of cancer, particularly regarding the personality-health behavior model (Fig. 1). We did observe that persons who were categorized as psychically vulnerable had different health-related behavior compared with those who were categorized as nonvulnerable, i.e., more current smokers were observed among psychically vulnerable persons. Furthermore, an increased risk of cancer was observed among current smokers and persons who regularly consumed alcohol. If an association between personality and risk for cancer does exist, then we believe it is likely that personality influences health practices, as suggested in the personality-health behavior model.

Acknowledgements

The authors thank Ms. Andrea Meersohn, Svend Larsen, and Abdi Hersi at the Center for Preventive Medicine, Glostrup, for their skillful computer assistance and Lars H. Thomassen for his helpful assistance with the data processing.