- Email: [email protected]
Dimensions of hypochondriasis and the five-factor model of personality
Personality and Individual Dierences 29 (2000) 99±108
www.elsevier.com/locate/paid
Dimensions of hypochondriasis and the ®ve-factor model of personality Brian J. Cox a,*, Sharon C. Borger b, Gordon J.G. Asmundson c, Steven Taylor d a
Department of Psychiatry, University of Manitoba, Canada Department of Psychology, University of Manitoba, Canada c Clinical Research and Development Program, Regina Health District, Canada d Department of Psychiatry, University of British Columbia, Canada b
Received 25 February 1999; received in revised form 22 June 1999; accepted 2 August 1999
Abstract Relations between dimensions of hypochondriasis assessed by the Illness Attitudes Scales and the higher-order and lower-order dimensions of the ``Big Five'' model of personality were examined in 309 university students. Factor analysis revealed ®ve IAS dimensions, similar to those identi®ed in previous studies: Fear of Illness and Death, Treatment Experience, Symptom Eects, Disease Phobia and Conviction, and Health Habits. These ®ve IAS dimensions in turn loaded on to two higher-order factors: Health Anxiety and Health Behaviors. To help clarify the meaning of these IAS dimensions and the two higher-order factors, a series of regression analyses were conducted using the NEO-PI-R personality domains and facets. Neuroticism, and its facet of anxiety, were signi®cant predictors of several IAS dimensions and the higher-order factor of Health Anxiety. In contrast, Conscientiousness was a signi®cant predictor of one of the IAS dimensions and the second higher-order factor of Health Behaviors. The results suggest that some IAS dimensions are re¯ective of speci®c and core features of hypochondriasis while other IAS dimensions are generally nonspeci®c or too peripheral, and may in fact assess adaptive functioning. The results also suggest that previously observed negative associations between hypochondriasis and Extraversion, as well as Agreeableness, are reduced or eliminated when Neuroticism is statistically controlled. 7 2000 Elsevier Science Ltd. All rights reserved. Keywords: Hypochondriasis; Big Five model of personality; Personality domains and facets
* Corresponding author. PZ-430 PsycHealth Centre, 771 Bannatyne Avenue, Winnipeg, Manitoba, Canada R3E 3N4. Tel.: +1-204-787-5166; fax: 1-204-787-4879. 0191-8869/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 1 9 1 - 8 8 6 9 ( 9 9 ) 0 0 1 8 0 - 4
100
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
Hypochondriasis is categorized as a somatoform disorder in the DSM-IV (American Psychiatric Association, 1994) where the essential feature is de®ned as a ``preoccupation with fears of having, or the idea that one has, a serious disease based on a misinterpretation of one or more bodily signs or symptoms''. Rather than taking a categorical approach, some authors have argued that available evidence strongly suggests hypochondriasis lies on a continuum and represents an individual dierence variable (Costa & McCrae, 1985; Tyrer, Fowler-Dixon, Ferguson & Kelemen, 1990). If indeed hypochondriasis is best conceptualized as an individual dierence variable, then it becomes important to understand how it relates to well-established, higher-order and lowerorder personality domains. There has also been signi®cant interest in approaching hypochondriasis as a cognitive abnormality characterized by catastrophic misinterpretations of innocuous bodily sensations (Barsky & Klerman, 1983; Warwick & Salkovskis, 1990). Yet the question still arises as to how this type of individual dierence variable (i.e., cognitive abnormality) relates to various personality traits. To date, there has been limited investigation along these lines. In addition, the majority of studies have examined medical complaints and somatic concerns rather than hypochondriasis (illness worry) per se. Using data from normal adults in the Baltimore Longitudinal Study of Aging (BLSA), Costa and McCrae (1985) observed positive associations between somatic concerns on the Cornell Medical Index (Brodman, Erdmann & Wol, 1949) and the higher-order personality domain of neuroticism. McCrae (1991) reported that the hypochondriasis clinical scale from the MMPI (Hathaway & McKinley, 1983) was positively associated with neuroticism and negatively associated with extraversion in the BLSA. Costa and McCrae (1992b) observed a similar relationship between neuroticism, extraversion and somatic concerns on the Basic Personality Inventory (Jackson, 1989) and on the Personality Assessment Inventory (Morey, 1991). It is not known if the negative association observed between hypochondriasis and extraversion might be due to a negative emotionality component in the measurement of somatic concerns. This possibility is supported in a study of abnormal illness behavior where aective inhibition and illness disruption, rather than health worry, were negatively correlated with extraversion (Zonderman, Heft & Costa, 1985). Also, because neuroticism is a very broad personality trait that denotes a tendency to experience a wide range of negative aect (e.g., anxiety, anger, depression), it is not yet known if there are any speci®c facets within neuroticism that are particularly associated with hypochondriasis. In addition to neuroticism, a review of personality traits and the somatoform disorders (Kirmayer, Robbins & Paris, 1994) led to the hypothesis that low agreeableness and high conscientiousness might also be associated with clinical hypochondriasis. These personality traits warrant further study in relation to hypochondriacal concerns as an individual dierence variable, but there is some evidence that adaptive ``wellness behaviors'' in normal adults are positively correlated with conscientiousness (Booth-Kewley & Vickers, 1994). One of the most widely used measures of hypochondriacal concerns is the Illness Attitudes Scales (IAS; Kellner, 1986; Kellner, Abbott, Winslow & Pathak, 1987). The IAS is a 27-item, self-report inventory designed to assess fears, beliefs, and attitudes associated with hypochondriasis and abnormal illness behavior (two additional items assess types of treatment and illness experiences, but are not used in the scoring of the subscales). It was originally
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
101
designed to generate nine dierent subscales: worry about illness, concerns about pain, health habits, hypochondriacal beliefs, thanatophobia (fear of death), disease phobia, bodily preoccupation, and treatment experience. However, researchers often utilize a total IAS score. Typically, this is done when researchers seek to correlate the IAS with other variables or when selecting nonclinical individuals with high or low hypochondriacal concerns (consistent with a continuum perspective of hypochondriasis). In addition, recent factor analytic work in clinical and nonclinical samples (Ferguson & Daniel, 1995; Hadjistavropoulos & Asmundson, 1998; Hadjistavropoulos, Frombach & Asmundson, in press) has indicated that the IAS captures four and possibly ®ve dierent dimensions, rather than nine.1 The most robust factors appear to be: (1) Worry about illness or pain, (2) Symptom eects (with respect to lifestyle interference), (3) Disease phobia and conviction, and (4) Health habits. Less robust dimensions include fear of death and treatment experiences, which sometimes load on worry about illness and symptoms eects factors, respectively. Aside from psychometric considerations, Hadjistavropoulos et al. (in press) have emphasized that further work is needed to clarify the meaning of these dierent IAS dimensions. One way this goal might be furthered is by examining IAS dimensions in relation to broad and speci®c personality traits. The ®ve-factor model of personality or the ``Big Five'' has received considerable attention in personality theory and research over the past several years. The most widely used measure of the Big Five is the Revised NEO Personality Inventory (NEO-PI-R; Costa & McCrae, 1992a). This instrument has received extensive empirical support and is based on a hierarchical structure of personality. On a higher-order level the NEO-PI-R assesses the Big Five domains: Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness. For each of these domains the NEO-PI-R also assesses six corresponding, lower-order facets. The design of the present study was as follows. First, a factor analysis of the IAS was conducted in an attempt to replicate the dimensions of hypochondriasis identi®ed in recent studies. Second, each of the IAS dimensions was examined in relation to the higher-order personality domains and lower-order personality facets of the NEO-PI-R using a series of regression analyses. Third, the possible hierarchical structure of the IAS was investigated by factor analyzing the IAS dimensions (i.e., factor scores), and then examining higher-order IAS factor(s) in relation to the NEO-PI-R. It was hypothesized that the NEO-PI-R higher-order domain of neuroticism would be predictive of those IAS dimensions which were conceptually closer to the core features of hypochondriasis. Consistent with the essential feature of hypochondriasis described in the DSM-IV, it was hypothesized that neuroticism would be most strongly predictive of IAS dimensions re¯ective of fears of illness and death, and disease phobia and conviction. Similarly, it was expected neuroticism would be less predictive for those dimensions which could be seen as more peripheral (health habits and treatment experiences). There is a paucity of previous work at the facet level, but it is reasonable to hypothesize that anxiety would be the strongest predictor variable within the domain of neuroticism. In addition to neuroticism, there is preliminary evidence to suggest that the domains of agreeableness and
1
In contrast to these studies, Speckens, Spinhoven, Sloekers, Bolk and van Hemert (1996) reported a two factor solution, using a Dutch translation of the IAS. However, only 17 of the 27 IAS items met the authors' criteria of loading 0.40 on one factor and loading at least 0.20 greater than on another factor.
102
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
extraversion might be negatively associated, and conscientiousness positively associated, with core features of hypochondriasis. As an ancillary analysis, a more general, non-speci®c measure of anxiety proneness was also examined in relation to the NEO-PI-R, so that results could be compared to those obtained with the IAS. In other words, we sought to determine which personality factors might be associated with speci®c, health-related anxiety, and compare these to personality factors associated with non-speci®c anxiety. 1. Method 1.1. Subjects A total of 309 undergraduate students at the University of Manitoba (190 women, 119 men) participated in the study (M age =19.84 yr, SD=3.52). Their mean score on the IAS (total) was 30.99 (SD=13.28), and the internal consistency (alpha) of the IAS was 0.87. IAS item means ranged from 0.30 to 2.99 and were very similar to the nonclinical pro®le used in an IAS factor analytic paper (Hadjistavropoulos et al., in press). Alphas for the NEO-PI-R domains were all at or above 0.85. The majority of the 30 facet alphas were above 0.70 (range = 0.53± 0.93). The pro®le of facet alphas was similar to those reported in the NEO-PI-R manual (range = 0.56±0.81, Costa & McCrae, 1992a, p. 44) which were deemed acceptable for scales with only 8 items. Alpha for the STAI was 0.93. 1.2. Measures and procedures Participants were administered the IAS (Kellner, 1986; Kellner et al., 1987), the NEO-PI-R (Costa & McCrae, 1992a), and the trait form of the State-Trait Anxiety Inventory (STAI; Spielberger, 1983) as part of a psychometric study on anxiety sensitivity (Cox, Borger, Taylor, Fuentes & Ross, in press). The 240-item NEO-PI-R assesses the ®ve higher-order domains discussed earlier as well as six lower-order facets within each domain. 2. Results In an attempt to replicate previously identi®ed dimensions within the IAS, a factor analytic procedure was followed that was identical to that employed by Hadjistavropoulos and colleagues (1998, in press). Kaiser's Measure of Sampling Adequacy was 0.81, indicating that the IAS items were appropriate for principal components analysis (Tabachnick & Fidell, 1996). A principal components analysis was conducted with oblique (Oblimin) rotation. The scree plot, as well as parallel analysis using both mean and 95th percentile Eigenvalues, indicated a ®ve-factor solution. This ®ve-factor solution was very similar to factor solutions identi®ed in previous studies, contained clear, interpretable factors with few cross-loadings, and accounted for 56.4% of the variance. The content of each of these factors and their corresponding Eigenvalues were as follows: Fear of Illness and Death (7.85), Treatment Experience (2.33),
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
103
Symptom Eects (1.86), Disease Phobia and Conviction (1.62), and Health Habits (1.57). IAS factor scores were calculated using the regression method. Correlations between IAS factor scores ranged from 0.01 to 0.41. The factor analysis was also repeated separately for men and women and a very similar pattern of ®ndings emerged. To determine whether the IAS factors would in turn load onto one or more higher-order hypochondriasis factors, a principal components analysis with oblique (Oblimin) rotation was conducted using the ®ve IAS factor scores. A higher-order, two-factor solution was clearly indicated by an Eigen-value >1 criterion and by visual inspection of the scree plot, and it accounted for 57.1% of the variance. The ®rst higher-order factor (35.7% of the variance) was closer to the core features of hypochondriasis and was comprised of three IAS lower-order factors: Fear of Illness or Death, Symptom Eects, and Disease Phobia and Conviction. This factor was labeled Health Anxiety. The second higher-order factor (21.4% of the variance) was more peripheral to the core features of hypochondriasis and was comprised of two IAS lowerorder factors: Treatment Experience and Health Habits. This factor was labeled Health Behaviors. The correlation between the two higher-order factors was 0.15. Factor scores were again calculated using the regression method. Table 1 presents correlations between the IAS factor scores and the domains and facets of the NEO-PI-R. In general the largest and most consistent associations were between the neuroticism domains and its facets and the IAS factors of Fear of Illness and Death, Symptom Eects, and Disease Phobia and Conviction, and the IAS higher-order factor of Health Anxiety. A number of smaller but signi®cant negative correlations emerged between some IAS factors and the domains and facets of Extraversion and Agreeableness. Because of the number of signi®cant correlations, a series of multiple regression analyses were then conducted using each of the IAS factor scores as dependent variables and the domains and facets of the NEO-PI-R as predictor variables. In each case the following procedure was used. The ®ve NEO-PI-R higher-order domains were ®rst entered as block of predictor variables using a forced entry method. In the second stage, the lower-order facets of the signi®cant (P < 0.05) predictor domains were then entered as a block, again using a forced entry method. Tolerance values were at or above 0.35, indicating that multicollinearity was not a problem (Norusis, 1988). The ®rst regression analysis used the Fear of Illness and Death dimension as the dependent variable, and the NEO-PI-R higher-order domain of Neuroticism (b=0.51) was the only signi®cant predictor variable in the regression equation (R = 0.48, F(5,303)=18.22, P < 0.001). In the second stage of this analysis only the N facet of anxiety (b=0.50) entered as a signi®cant predictor variable in the regression equation (R = 0.55, F(6,302)=21.94, P < 0.001). The second regression analysis used the Treatment Experience dimension as the dependent variable, and the NEO-PI-R domain of Neuroticism (b=0.21) was the only signi®cant predictor variable in the regression (R = 0.22, F(5,303)=3.16, P < 0.01). In the second stage only the N facet of anxiety (b=0.18) entered as a signi®cant predictor variable (R = 0.22, F(6,302)=2.5 P < 0.05). The third regression analysis used the Symptom Eects dimension as the dependent variable, and only the NEO-PI-R domain of Neuroticism (b=0.42) was a signi®cant predictor variable (R = 0.48, F(5,303) =17.63, P < 0.001). In the second stage the facets of anxiety (b=0.15),
NEO-PI-R domains and facets
a
P < 0.05,
Fear of Illness and Death 0.47a 0.54 0.30 0.39 0.36 0.17 0.34
Treatment Experience
Symptom Eects
0.14 0.15 0.17 0.09 0.05 0.12 0.05
0.46 0.42 0.20 0.43 0.40 0.19 0.43
IAS high-order factors Disease Phobia and Conviction 0.40 0.39 0.29 0.35 0.27 0.19 0.32
Health Habits ÿ0.11 0.01 ÿ0.10 ÿ0.11 ÿ0.07 ÿ0.20 ÿ0.06
Health Anxiety 0.60 0.60 0.37 0.52 0.46 0.26 0.48
Health Behaviors 0.01 0.10 0.03 ÿ0.02 ÿ0.02 ÿ0.07 ÿ0.01
ÿ0.08 ÿ0.02 0.01 ÿ0.17 ÿ0.07 ÿ0.07 ÿ0.02 ÿ0.05
0.09 0.09 0.11 0.02 0.05 0.02 0.12 0.10
ÿ0.22 ÿ0.10 ÿ0.11 ÿ0.30 ÿ0.21 ÿ0.12 ÿ0.14 ÿ0.06
ÿ0.13 ÿ0.14 ÿ0.06 ÿ0.13 ÿ0.03 ÿ0.08 ÿ0.14 0.01
0.02 ÿ0.04 ÿ0.02 0.11 0.07 ÿ0.13 0.10 0.02
ÿ0.18 ÿ0.10 ÿ0.06 ÿ0.26 ÿ0.12 ÿ0.10 ÿ0.12 ÿ0.03
0.06 0.02 0.04 0.07 0.08 ÿ0.09 0.14 0.07
0.10 ÿ0.03 0.10 ÿ0.11 ÿ0.19 ÿ0.08 ÿ0.03 ÿ0.25 ÿ0.03 ÿ0.04 0.03 0.09 0.12 ÿ0.16 ÿ0.18 ÿ0.03 ÿ0.19 ÿ0.10
0.12 0.09 0.18 ÿ0.01 0.10 0.11 0.03 0.03 ÿ0.01 0.06 ÿ0.05 ÿ0.04 0.18 0.08 0.09 0.14 0.04 0.07
0.04 0.03 0.02 ÿ0.16 ÿ0.18 ÿ0.05 0.03 ÿ0.20 0.04 ÿ0.06 0.09 0.18 0.07 ÿ0.24 ÿ0.27 ÿ0.02 ÿ0.13 ÿ0.23
0.03 0.06 0.08 0.05 ÿ0.08 ÿ0.10 ÿ0.12 ÿ0.28 ÿ0.15 ÿ0.14 ÿ0.02 0.03 0.06 ÿ0.18 ÿ0.21 ÿ0.03 ÿ0.15 ÿ0.14
ÿ0.09 0.08 0.04 0.05 ÿ0.06 ÿ0.06 0.01 ÿ0.01 0.01 ÿ0.03 0.13 ÿ0.09 0.04 0.22 0.09 0.17 0.11 0.18
0.09 0.03 0.11 ÿ0.10 ÿ0.20 ÿ0.08 ÿ0.05 ÿ0.31 ÿ0.07 ÿ0.09 ÿ0.02 ÿ0.13 ÿ0.12 ÿ0.26 ÿ0.28 ÿ0.03 ÿ0.21 ÿ0.21
0.01 0.11 0.13 0.03 ÿ0.03 0.02 0.03 0.01 0.00 0.01 0.07 ÿ0.08 0.13 0.21 0.12 0.21 0.11 0.17
ÿ0.21 ÿ0.03
0.01 0.01
ÿ0.34 ÿ0.11
ÿ0.26 ÿ0.02
0.20 0.24
ÿ0.36 ÿ0.08
0.15 0.18
P < 0.01.
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
Neuroticism Anxiety Angry hostility Depression Self-consciousness Impulsivity Vulnerability (to stress) Extraversion Warmth Gregariousness Assertiveness Activity Excitement-seeking Positive emotions Openness to Experience Fantasy Aesthetics Feelings Actions Ideas Values Agreeableness Trust Straightforwardness Altruism Compliance Modesty Tendermindedness Conscientiousness Competence Order Dutifulness Achievementstriving Self-discipline Deliberation
IAS factors
104
Table 1 Zero-order correlations between the IAS factor scores and the domains and facets of the NEO-PI-R (N = 309)
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
105
depression (b=0.17), and vulnerability to stress (b=0.20) all entered as signi®cant predictor variables (R = 0.49, F(6,302)=15.89, P < 0.001). The fourth regression analysis used the Disease Phobia and Conviction dimension as the dependent variable and both the NEO-PI-R domains of Neuroticism (b=0.40) and Agreeableness (b=ÿ0.12) entered as signi®cant predictors (R = 0.42, F(5,303)=12.95, P < 0.001). In the second stage the anxiety facet of Neuroticism (b=0.26) and the straightforwardness facet of Agreeableness (b=ÿ0.18) entered as signi®cant predictors (R = 0.48, F(12,296)=7.30, P < 0.001). The ®fth regression analysis used the Health Habits dimension as the dependent variable and the NEO-PI-R domain of Conscientiousness (b=0.24) entered as a signi®cant predictor (R = 0.24, F(5,303)=3.58, P < 0.005). In the second stage the deliberation facet of Conscientiousness (b=0.19) entered as a signi®cant predictor (R = 0.28, F(6,302)=4.28, P < 0.001). The ®ve domains of the NEO-PI-R were entered as predictor variables in regression analyses for each of the two higher-order IAS factors. In a regression of the ®rst higher-order factor of Health Anxiety, the Neuroticism domain (b=0.60) was the only signi®cant predictor in the regression equation (R = 0.60, F(5,303)=33.85, P < 0.001). In the second stage, only the Neuroticism facets of anxiety (b=0.40) and depression (b=0.16) entered as signi®cant predictors (R = 0.63, F(6,302)=32.94, P < 0.001). In a regression of the second higher-order factor of Health Behaviors, the Conscientiousness domain (b=0.25) was the only signi®cant predictor in the regression equation (R = 0.23, F(5,303)=3.32, P < 0.01. In the second stage, only the Conscientiousness facet of order (b=0.13) entered as a signi®cant predictor (R = 0.24, F(6,302)=3.09, P < 0.01). The same regression approach was used with a measure of non-speci®c anxiety, the STAI. The NEO-PI-R domains of Neuroticism (b=0.70) and Extraversion (b=ÿ0.20) were signi®cant predictors in the regression equation (R = 0.81, F(5,311)=119.32, P < 0.001). In the second stage the Neuroticism facets of anxiety (b=0.12), depression (b=0.44), and vulnerability to stress (b=0.22), as well as the Extraversion facets of assertiveness (b=ÿ0.10) and positive emotions (b=ÿ0.16) were signi®cant predictors in the regression equation (R = 0.84, F(12,304)=62.26, P < 0.001).
3. Discussion The results of this study serve to extend previous research on the nature of hypochondriasis as measured by the IAS. Speci®cally, the present study replicated the factor structure identi®ed in recent studies (Hadjistavropoulos & Asmundson, 1998; Hadjistavropoulos et al., in press). The meaning of each of the IAS dimensions was then further elucidated through their relations with higher-order and lower-order personality traits. These results highlight the importance of conducting regression analysis in addition to correlations: several associations between IAS factors and extraversion, along with agreeableness, were not apparent when the contribution of neuroticism was statistically controlled. The higher-order domain of neuroticism and one of its six lower-order facets, anxiety, were signi®cant predictors of several IAS dimensions. However, neuroticism and anxiety were most
106
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
strongly associated with variance in the Fear of Illness and Death dimension. For example, the beta weights and strength of prediction were more than twice as great for this dimension compared to the Treatment Experience dimension. In other words, Treatment Experience (e.g., visits to a physician) was less likely to be related to neuroticism or other personality traits. Although neuroticism denotes a general tendency to experience negative aect, the anxiety facet is relatively speci®c and refers to apprehension, fearfulness, and worry proneness (Costa & McCrae, 1992a). It is interesting to compare this result to the relationship between the NEO-PI-R and a measure of nonspeci®c or ``free-¯oating'' anxiety. In contrast to the Fear of Illness and Death IAS dimension, the STAI was predicted by three dierent facets of neuroticism and two facets of extraversion. It is hardly surprising that the neuroticism facet of anxiety predicted nonspeci®c anxiety, but it was interesting to see what other personality variables emerged as predictors. In particular, there were associations with a wider range of negative aect and with interpersonal diculties, whereas this was not true for Fear of Illness and Death. These ®ndings are consistent with the observations of Cox et al. (in press) concerning anxiety sensitivity and hypochondriasis. Anxiety sensitivity is an individual dierence variable de®ned as a fear of anxiety symptoms based on the belief that these symptoms have harmful consequences (Reiss, Peterson, Gursky & McNally, 1986) and it may represent a cognitive risk factor for panic disorder (McNally, 1994). Cox et al. (in press) found that one of the NEO-PIR traits that predicted anxiety sensitivity, including a dimension of anxiety sensitivity concerning fear of somatic sensations, was the neuroticism facet of self-consciousness. Cox et al. suggested that an important cognitive dierentiation between anxiety sensitivity and hypochondriasis may be that social/interpersonal concerns are salient only in the case of anxiety sensitivity. The Symptom Eects dimension was associated with a wider range of neuroticism facets, and was predicted by anxiety, depression, and vulnerability to stress. This latter facet of neuroticism relates to perceived inability to cope with stress and individuals scoring high on this facet may become ``dependent, hopeless, or panicked when facing emergency situations'' (Costa & McCrae, 1992a, p.16). This ®nding suggests that the Symptoms Eects dimension taps into more global life functioning, and extends beyond hypochondriasis. Disease Phobia and Conviction was not only positively associated with neuroticism and anxiety, but was also negatively associated with the agreeableness domain and its facet of straightforwardness. Costa and McCrae (1992a) note that the domain of agreeableness is one of interpersonal tendencies and also note the following in regard to the facet of straightforwardness: ``Straightforward individuals, that is, those individuals with high scores on this scale are frank, sincere, and ingenuous. Low scorers on this scale are more willing to manipulate others through ¯attery, craftiness, or deception'' (p. 17). This ®nding suggests that the Disease Phobia and Conviction dimension may also sometimes assess something broader than hypochondriasis per se. The Health Habits dimension seems to stray furthest from the core features of hypochondriasis. Rather than being associated with the maladaptive traits of neuroticism, it was best predicted by the conscientiousness domain and its facet of deliberation. These traits re¯ect purposefulness, caution, and self-control. In excessive levels it is conceivable that these traits may re¯ect features associated with hypochondriasis, but only in a peripheral sense.
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
107
The distinction between core and peripheral features of hypochondriasis contained in the IAS was further elucidated by the identi®cation of two higher-order factors: Health Anxiety and Health Behaviors. The ®rst factor of Health Anxiety appeared to capture the major elements of hypochondriasis, and was best predicted by neuroticism. In contrast, the second factor of Health Behaviors was comprised of the Treatment Experiences and Health Habits dimensions, and was best predicted by conscientiousness. In other words, rather than being indicative of psychopathology, this second IAS factor and its lower-order dimensions may often assess adaptive functioning. The present study examined hypochondriasis as an individual dierence variable comprised of concern and anxiety, rather than as a clinical somatoform disorder. One implication of this approach is that a dierent personality pro®le might prevail in clinical levels of hypochondriasis. For example, the trust facet of the agreeableness domain was signi®cantly correlated with some IAS factors in this study, although less so than neuroticism. It is possible that in clinical samples the negative association between trust and hypochondriasis could be more pronounced, and might help explain the failure to respond to medical reassurance seen in severe hypochondriasis. In conclusion, the results of this study help to clarify the dierential meaning of the dimensions of hypochondriasis underlying the IAS. Our ®ndings support the assertion of Hadjistavropoulos and Asmundson (1998) that fear or preoccupation with illness or pain is a fundamental component of hypochondriasis (the ®rst lower-order dimension in our study and also part of the ®rst higher-order factor). These results do not support the use of a total score for the IAS, and also suggest that some dimensions within the IAS (e.g., Health Habits) are too non-speci®c or peripheral to warrant their use in assessing the core features of hypochondriasis. In fact, based on the recent research ®ndings that have now accumulated, one has to wonder whether the current IAS can really be salvaged. An alternative might be to utilize some of its more valid items as part of the construction of a new measure of hypochondriasis. Further work on IAS dimensions and personality functioning in clinical samples also appears warranted.
Acknowledgements This research was supported by a grant from the Manitoba Health Research Council.
References American, Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: American Psychiatric Association. Barsky, A. J., & Klerman, G. L. (1983). Overview: hypochondriasis, bodily complaints, and somatic styles. American Journal of Psychiatry, 140, 273±283. Booth-Kewley, S., & Vickers, R. R. (1994). Associations between major domains of personality and health behavior. Journal of Personality, 62, 281±298. Brodman, K., Erdmann, A. J., & Wol, H. G. (1949). The Cornell Medical Index health questionnaire manual. New York: Cornell University.
108
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
Costa, P. T., & McCrae, R. R. (1985). Hypochondriasis, neuroticism, and aging: when are somatic complaints unfounded? American Psychologist, 40, 19±28. Costa, P. T., & McCrae, R. R. (1992a). Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) professional manual. Odessa, FL: Psychological Assessment Resources. Costa, P. T., & McCrae, R. R. (1992b). Normal personality assessment in clinical practice: the NEO Personality Inventory. Psychological Assessment, 4, 5±13. Cox B.J., Borger S.C., Taylor S., Fuentes K., Ross L.M. Anxiety sensitivity and the ®ve-factor model of personality. Behaviour Research and Therapy (in press). Ferguson, E., & Daniel, E. (1995). The Illness Attitudes Scales (IAS): a psychometric evaluation on a non-clinical population. Personality and Individual Dierences, 18, 463±469. Hadjistavropoulos, H. D., & Asmundson, G. J. G. (1998). Factor analytic investigations of the Illness Attitudes Scales: exploration in a chronic pain sample. Behaviour Research and Therapy, 36, 1185±1195. Hadjistavropoulos H.D., Frombach I., Asmundson G.J.G. Exploratory and con®rmatory factor analytic investigations of the Illness Attitudes Scales in a non-clinical sample. Behaviour Research and Therapy (in press). Hathaway, S. R., & McKinley, J. C. (1983). The Minnesota Multiphasic Personality Inventory manual. New York: Psychological Corporation. Jackson, D. N. (1989). Basic Personality Inventory manual. Port Huron, MI: Sigma Assessment Systems. Kellner, R. (1986). Somatization and hypochondriasis. New York: Praeger. Kellner, R., Abbott, P., Winslow, W. W., & Pathak, D. (1987). Fears, beliefs, and attitudes in DSM-III hypochondriasis. Journal of Nervous and Mental Disease, 175, 20±25. Kirmayer, L. J., Robbins, J. M., & Paris, J. (1994). Somatoform disorders: personality and the social matrix of somatic distress. Journal of Abnormal Psychology, 103, 125±136. McCrae, R. R. (1991). The ®ve-factor model and its assessment in clinical settings. Journal of Personality Assessment, 57, 399±414. McNally, R. M. (1994). Panic disorder: a critical analysis. New York: Guilford Press. Morey, L. (1991). Personality Assessment Inventory: professional manual. Odessa, FL: Psychological Assessment Resources. Norusis, M. J. (1988). SPSS-X Advanced statistics guide (2nd ed.). Chicago, IL: SPSS. Reiss, S., Peterson, R. A., Gursky, D. M., & McNally, R. J. (1986). Anxiety sensitivity, anxiety frequency, and the prediction of fearfulness. Behaviour Research and Therapy, 24, 1±8. Speckens, A. E. M., Spinhoven, P., Sloekers, P. P. A., Bolk, J. H., & van Hemert, A. M. (1996). A validation study of the Whitely Index, the Illness Attitudes Scales, and the Somatosensory Ampli®cation Scale in general medical and general practice patients. Journal of Psychosomatic Research, 40, 95±104. Spielberger, C. D. (1983). Manual for the state-trait anxiety inventory (Form Y). Palo Alto, CA: Consulting Psychologists Press. Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.). New York: Harper Collins. Tyrer, P., Fowler-Dixon, R., Ferguson, B., & Kelemen, A. (1990). A plea for the diagnosis of hypochondriacal personality disorder. Journal of Psychosomatic Research, 34, 637±642. Warwick, H. M. C., & Salkovskis, P. M. (1990). Hypochondriasis. Behaviour Research and Therapy, 28, 105±117. Zonderman, A. B., Heft, M. W., & Costa, P. T. (1985). Does the Illness Behavior Questionnaire measure abnormal illness behavior? Health Psychology, 4, 425±436.
www.elsevier.com/locate/paid
Dimensions of hypochondriasis and the ®ve-factor model of personality Brian J. Cox a,*, Sharon C. Borger b, Gordon J.G. Asmundson c, Steven Taylor d a
Department of Psychiatry, University of Manitoba, Canada Department of Psychology, University of Manitoba, Canada c Clinical Research and Development Program, Regina Health District, Canada d Department of Psychiatry, University of British Columbia, Canada b
Received 25 February 1999; received in revised form 22 June 1999; accepted 2 August 1999
Abstract Relations between dimensions of hypochondriasis assessed by the Illness Attitudes Scales and the higher-order and lower-order dimensions of the ``Big Five'' model of personality were examined in 309 university students. Factor analysis revealed ®ve IAS dimensions, similar to those identi®ed in previous studies: Fear of Illness and Death, Treatment Experience, Symptom Eects, Disease Phobia and Conviction, and Health Habits. These ®ve IAS dimensions in turn loaded on to two higher-order factors: Health Anxiety and Health Behaviors. To help clarify the meaning of these IAS dimensions and the two higher-order factors, a series of regression analyses were conducted using the NEO-PI-R personality domains and facets. Neuroticism, and its facet of anxiety, were signi®cant predictors of several IAS dimensions and the higher-order factor of Health Anxiety. In contrast, Conscientiousness was a signi®cant predictor of one of the IAS dimensions and the second higher-order factor of Health Behaviors. The results suggest that some IAS dimensions are re¯ective of speci®c and core features of hypochondriasis while other IAS dimensions are generally nonspeci®c or too peripheral, and may in fact assess adaptive functioning. The results also suggest that previously observed negative associations between hypochondriasis and Extraversion, as well as Agreeableness, are reduced or eliminated when Neuroticism is statistically controlled. 7 2000 Elsevier Science Ltd. All rights reserved. Keywords: Hypochondriasis; Big Five model of personality; Personality domains and facets
* Corresponding author. PZ-430 PsycHealth Centre, 771 Bannatyne Avenue, Winnipeg, Manitoba, Canada R3E 3N4. Tel.: +1-204-787-5166; fax: 1-204-787-4879. 0191-8869/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 1 9 1 - 8 8 6 9 ( 9 9 ) 0 0 1 8 0 - 4
100
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
Hypochondriasis is categorized as a somatoform disorder in the DSM-IV (American Psychiatric Association, 1994) where the essential feature is de®ned as a ``preoccupation with fears of having, or the idea that one has, a serious disease based on a misinterpretation of one or more bodily signs or symptoms''. Rather than taking a categorical approach, some authors have argued that available evidence strongly suggests hypochondriasis lies on a continuum and represents an individual dierence variable (Costa & McCrae, 1985; Tyrer, Fowler-Dixon, Ferguson & Kelemen, 1990). If indeed hypochondriasis is best conceptualized as an individual dierence variable, then it becomes important to understand how it relates to well-established, higher-order and lowerorder personality domains. There has also been signi®cant interest in approaching hypochondriasis as a cognitive abnormality characterized by catastrophic misinterpretations of innocuous bodily sensations (Barsky & Klerman, 1983; Warwick & Salkovskis, 1990). Yet the question still arises as to how this type of individual dierence variable (i.e., cognitive abnormality) relates to various personality traits. To date, there has been limited investigation along these lines. In addition, the majority of studies have examined medical complaints and somatic concerns rather than hypochondriasis (illness worry) per se. Using data from normal adults in the Baltimore Longitudinal Study of Aging (BLSA), Costa and McCrae (1985) observed positive associations between somatic concerns on the Cornell Medical Index (Brodman, Erdmann & Wol, 1949) and the higher-order personality domain of neuroticism. McCrae (1991) reported that the hypochondriasis clinical scale from the MMPI (Hathaway & McKinley, 1983) was positively associated with neuroticism and negatively associated with extraversion in the BLSA. Costa and McCrae (1992b) observed a similar relationship between neuroticism, extraversion and somatic concerns on the Basic Personality Inventory (Jackson, 1989) and on the Personality Assessment Inventory (Morey, 1991). It is not known if the negative association observed between hypochondriasis and extraversion might be due to a negative emotionality component in the measurement of somatic concerns. This possibility is supported in a study of abnormal illness behavior where aective inhibition and illness disruption, rather than health worry, were negatively correlated with extraversion (Zonderman, Heft & Costa, 1985). Also, because neuroticism is a very broad personality trait that denotes a tendency to experience a wide range of negative aect (e.g., anxiety, anger, depression), it is not yet known if there are any speci®c facets within neuroticism that are particularly associated with hypochondriasis. In addition to neuroticism, a review of personality traits and the somatoform disorders (Kirmayer, Robbins & Paris, 1994) led to the hypothesis that low agreeableness and high conscientiousness might also be associated with clinical hypochondriasis. These personality traits warrant further study in relation to hypochondriacal concerns as an individual dierence variable, but there is some evidence that adaptive ``wellness behaviors'' in normal adults are positively correlated with conscientiousness (Booth-Kewley & Vickers, 1994). One of the most widely used measures of hypochondriacal concerns is the Illness Attitudes Scales (IAS; Kellner, 1986; Kellner, Abbott, Winslow & Pathak, 1987). The IAS is a 27-item, self-report inventory designed to assess fears, beliefs, and attitudes associated with hypochondriasis and abnormal illness behavior (two additional items assess types of treatment and illness experiences, but are not used in the scoring of the subscales). It was originally
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
101
designed to generate nine dierent subscales: worry about illness, concerns about pain, health habits, hypochondriacal beliefs, thanatophobia (fear of death), disease phobia, bodily preoccupation, and treatment experience. However, researchers often utilize a total IAS score. Typically, this is done when researchers seek to correlate the IAS with other variables or when selecting nonclinical individuals with high or low hypochondriacal concerns (consistent with a continuum perspective of hypochondriasis). In addition, recent factor analytic work in clinical and nonclinical samples (Ferguson & Daniel, 1995; Hadjistavropoulos & Asmundson, 1998; Hadjistavropoulos, Frombach & Asmundson, in press) has indicated that the IAS captures four and possibly ®ve dierent dimensions, rather than nine.1 The most robust factors appear to be: (1) Worry about illness or pain, (2) Symptom eects (with respect to lifestyle interference), (3) Disease phobia and conviction, and (4) Health habits. Less robust dimensions include fear of death and treatment experiences, which sometimes load on worry about illness and symptoms eects factors, respectively. Aside from psychometric considerations, Hadjistavropoulos et al. (in press) have emphasized that further work is needed to clarify the meaning of these dierent IAS dimensions. One way this goal might be furthered is by examining IAS dimensions in relation to broad and speci®c personality traits. The ®ve-factor model of personality or the ``Big Five'' has received considerable attention in personality theory and research over the past several years. The most widely used measure of the Big Five is the Revised NEO Personality Inventory (NEO-PI-R; Costa & McCrae, 1992a). This instrument has received extensive empirical support and is based on a hierarchical structure of personality. On a higher-order level the NEO-PI-R assesses the Big Five domains: Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness. For each of these domains the NEO-PI-R also assesses six corresponding, lower-order facets. The design of the present study was as follows. First, a factor analysis of the IAS was conducted in an attempt to replicate the dimensions of hypochondriasis identi®ed in recent studies. Second, each of the IAS dimensions was examined in relation to the higher-order personality domains and lower-order personality facets of the NEO-PI-R using a series of regression analyses. Third, the possible hierarchical structure of the IAS was investigated by factor analyzing the IAS dimensions (i.e., factor scores), and then examining higher-order IAS factor(s) in relation to the NEO-PI-R. It was hypothesized that the NEO-PI-R higher-order domain of neuroticism would be predictive of those IAS dimensions which were conceptually closer to the core features of hypochondriasis. Consistent with the essential feature of hypochondriasis described in the DSM-IV, it was hypothesized that neuroticism would be most strongly predictive of IAS dimensions re¯ective of fears of illness and death, and disease phobia and conviction. Similarly, it was expected neuroticism would be less predictive for those dimensions which could be seen as more peripheral (health habits and treatment experiences). There is a paucity of previous work at the facet level, but it is reasonable to hypothesize that anxiety would be the strongest predictor variable within the domain of neuroticism. In addition to neuroticism, there is preliminary evidence to suggest that the domains of agreeableness and
1
In contrast to these studies, Speckens, Spinhoven, Sloekers, Bolk and van Hemert (1996) reported a two factor solution, using a Dutch translation of the IAS. However, only 17 of the 27 IAS items met the authors' criteria of loading 0.40 on one factor and loading at least 0.20 greater than on another factor.
102
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
extraversion might be negatively associated, and conscientiousness positively associated, with core features of hypochondriasis. As an ancillary analysis, a more general, non-speci®c measure of anxiety proneness was also examined in relation to the NEO-PI-R, so that results could be compared to those obtained with the IAS. In other words, we sought to determine which personality factors might be associated with speci®c, health-related anxiety, and compare these to personality factors associated with non-speci®c anxiety. 1. Method 1.1. Subjects A total of 309 undergraduate students at the University of Manitoba (190 women, 119 men) participated in the study (M age =19.84 yr, SD=3.52). Their mean score on the IAS (total) was 30.99 (SD=13.28), and the internal consistency (alpha) of the IAS was 0.87. IAS item means ranged from 0.30 to 2.99 and were very similar to the nonclinical pro®le used in an IAS factor analytic paper (Hadjistavropoulos et al., in press). Alphas for the NEO-PI-R domains were all at or above 0.85. The majority of the 30 facet alphas were above 0.70 (range = 0.53± 0.93). The pro®le of facet alphas was similar to those reported in the NEO-PI-R manual (range = 0.56±0.81, Costa & McCrae, 1992a, p. 44) which were deemed acceptable for scales with only 8 items. Alpha for the STAI was 0.93. 1.2. Measures and procedures Participants were administered the IAS (Kellner, 1986; Kellner et al., 1987), the NEO-PI-R (Costa & McCrae, 1992a), and the trait form of the State-Trait Anxiety Inventory (STAI; Spielberger, 1983) as part of a psychometric study on anxiety sensitivity (Cox, Borger, Taylor, Fuentes & Ross, in press). The 240-item NEO-PI-R assesses the ®ve higher-order domains discussed earlier as well as six lower-order facets within each domain. 2. Results In an attempt to replicate previously identi®ed dimensions within the IAS, a factor analytic procedure was followed that was identical to that employed by Hadjistavropoulos and colleagues (1998, in press). Kaiser's Measure of Sampling Adequacy was 0.81, indicating that the IAS items were appropriate for principal components analysis (Tabachnick & Fidell, 1996). A principal components analysis was conducted with oblique (Oblimin) rotation. The scree plot, as well as parallel analysis using both mean and 95th percentile Eigenvalues, indicated a ®ve-factor solution. This ®ve-factor solution was very similar to factor solutions identi®ed in previous studies, contained clear, interpretable factors with few cross-loadings, and accounted for 56.4% of the variance. The content of each of these factors and their corresponding Eigenvalues were as follows: Fear of Illness and Death (7.85), Treatment Experience (2.33),
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
103
Symptom Eects (1.86), Disease Phobia and Conviction (1.62), and Health Habits (1.57). IAS factor scores were calculated using the regression method. Correlations between IAS factor scores ranged from 0.01 to 0.41. The factor analysis was also repeated separately for men and women and a very similar pattern of ®ndings emerged. To determine whether the IAS factors would in turn load onto one or more higher-order hypochondriasis factors, a principal components analysis with oblique (Oblimin) rotation was conducted using the ®ve IAS factor scores. A higher-order, two-factor solution was clearly indicated by an Eigen-value >1 criterion and by visual inspection of the scree plot, and it accounted for 57.1% of the variance. The ®rst higher-order factor (35.7% of the variance) was closer to the core features of hypochondriasis and was comprised of three IAS lower-order factors: Fear of Illness or Death, Symptom Eects, and Disease Phobia and Conviction. This factor was labeled Health Anxiety. The second higher-order factor (21.4% of the variance) was more peripheral to the core features of hypochondriasis and was comprised of two IAS lowerorder factors: Treatment Experience and Health Habits. This factor was labeled Health Behaviors. The correlation between the two higher-order factors was 0.15. Factor scores were again calculated using the regression method. Table 1 presents correlations between the IAS factor scores and the domains and facets of the NEO-PI-R. In general the largest and most consistent associations were between the neuroticism domains and its facets and the IAS factors of Fear of Illness and Death, Symptom Eects, and Disease Phobia and Conviction, and the IAS higher-order factor of Health Anxiety. A number of smaller but signi®cant negative correlations emerged between some IAS factors and the domains and facets of Extraversion and Agreeableness. Because of the number of signi®cant correlations, a series of multiple regression analyses were then conducted using each of the IAS factor scores as dependent variables and the domains and facets of the NEO-PI-R as predictor variables. In each case the following procedure was used. The ®ve NEO-PI-R higher-order domains were ®rst entered as block of predictor variables using a forced entry method. In the second stage, the lower-order facets of the signi®cant (P < 0.05) predictor domains were then entered as a block, again using a forced entry method. Tolerance values were at or above 0.35, indicating that multicollinearity was not a problem (Norusis, 1988). The ®rst regression analysis used the Fear of Illness and Death dimension as the dependent variable, and the NEO-PI-R higher-order domain of Neuroticism (b=0.51) was the only signi®cant predictor variable in the regression equation (R = 0.48, F(5,303)=18.22, P < 0.001). In the second stage of this analysis only the N facet of anxiety (b=0.50) entered as a signi®cant predictor variable in the regression equation (R = 0.55, F(6,302)=21.94, P < 0.001). The second regression analysis used the Treatment Experience dimension as the dependent variable, and the NEO-PI-R domain of Neuroticism (b=0.21) was the only signi®cant predictor variable in the regression (R = 0.22, F(5,303)=3.16, P < 0.01). In the second stage only the N facet of anxiety (b=0.18) entered as a signi®cant predictor variable (R = 0.22, F(6,302)=2.5 P < 0.05). The third regression analysis used the Symptom Eects dimension as the dependent variable, and only the NEO-PI-R domain of Neuroticism (b=0.42) was a signi®cant predictor variable (R = 0.48, F(5,303) =17.63, P < 0.001). In the second stage the facets of anxiety (b=0.15),
NEO-PI-R domains and facets
a
P < 0.05,
Fear of Illness and Death 0.47a 0.54 0.30 0.39 0.36 0.17 0.34
Treatment Experience
Symptom Eects
0.14 0.15 0.17 0.09 0.05 0.12 0.05
0.46 0.42 0.20 0.43 0.40 0.19 0.43
IAS high-order factors Disease Phobia and Conviction 0.40 0.39 0.29 0.35 0.27 0.19 0.32
Health Habits ÿ0.11 0.01 ÿ0.10 ÿ0.11 ÿ0.07 ÿ0.20 ÿ0.06
Health Anxiety 0.60 0.60 0.37 0.52 0.46 0.26 0.48
Health Behaviors 0.01 0.10 0.03 ÿ0.02 ÿ0.02 ÿ0.07 ÿ0.01
ÿ0.08 ÿ0.02 0.01 ÿ0.17 ÿ0.07 ÿ0.07 ÿ0.02 ÿ0.05
0.09 0.09 0.11 0.02 0.05 0.02 0.12 0.10
ÿ0.22 ÿ0.10 ÿ0.11 ÿ0.30 ÿ0.21 ÿ0.12 ÿ0.14 ÿ0.06
ÿ0.13 ÿ0.14 ÿ0.06 ÿ0.13 ÿ0.03 ÿ0.08 ÿ0.14 0.01
0.02 ÿ0.04 ÿ0.02 0.11 0.07 ÿ0.13 0.10 0.02
ÿ0.18 ÿ0.10 ÿ0.06 ÿ0.26 ÿ0.12 ÿ0.10 ÿ0.12 ÿ0.03
0.06 0.02 0.04 0.07 0.08 ÿ0.09 0.14 0.07
0.10 ÿ0.03 0.10 ÿ0.11 ÿ0.19 ÿ0.08 ÿ0.03 ÿ0.25 ÿ0.03 ÿ0.04 0.03 0.09 0.12 ÿ0.16 ÿ0.18 ÿ0.03 ÿ0.19 ÿ0.10
0.12 0.09 0.18 ÿ0.01 0.10 0.11 0.03 0.03 ÿ0.01 0.06 ÿ0.05 ÿ0.04 0.18 0.08 0.09 0.14 0.04 0.07
0.04 0.03 0.02 ÿ0.16 ÿ0.18 ÿ0.05 0.03 ÿ0.20 0.04 ÿ0.06 0.09 0.18 0.07 ÿ0.24 ÿ0.27 ÿ0.02 ÿ0.13 ÿ0.23
0.03 0.06 0.08 0.05 ÿ0.08 ÿ0.10 ÿ0.12 ÿ0.28 ÿ0.15 ÿ0.14 ÿ0.02 0.03 0.06 ÿ0.18 ÿ0.21 ÿ0.03 ÿ0.15 ÿ0.14
ÿ0.09 0.08 0.04 0.05 ÿ0.06 ÿ0.06 0.01 ÿ0.01 0.01 ÿ0.03 0.13 ÿ0.09 0.04 0.22 0.09 0.17 0.11 0.18
0.09 0.03 0.11 ÿ0.10 ÿ0.20 ÿ0.08 ÿ0.05 ÿ0.31 ÿ0.07 ÿ0.09 ÿ0.02 ÿ0.13 ÿ0.12 ÿ0.26 ÿ0.28 ÿ0.03 ÿ0.21 ÿ0.21
0.01 0.11 0.13 0.03 ÿ0.03 0.02 0.03 0.01 0.00 0.01 0.07 ÿ0.08 0.13 0.21 0.12 0.21 0.11 0.17
ÿ0.21 ÿ0.03
0.01 0.01
ÿ0.34 ÿ0.11
ÿ0.26 ÿ0.02
0.20 0.24
ÿ0.36 ÿ0.08
0.15 0.18
P < 0.01.
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
Neuroticism Anxiety Angry hostility Depression Self-consciousness Impulsivity Vulnerability (to stress) Extraversion Warmth Gregariousness Assertiveness Activity Excitement-seeking Positive emotions Openness to Experience Fantasy Aesthetics Feelings Actions Ideas Values Agreeableness Trust Straightforwardness Altruism Compliance Modesty Tendermindedness Conscientiousness Competence Order Dutifulness Achievementstriving Self-discipline Deliberation
IAS factors
104
Table 1 Zero-order correlations between the IAS factor scores and the domains and facets of the NEO-PI-R (N = 309)
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
105
depression (b=0.17), and vulnerability to stress (b=0.20) all entered as signi®cant predictor variables (R = 0.49, F(6,302)=15.89, P < 0.001). The fourth regression analysis used the Disease Phobia and Conviction dimension as the dependent variable and both the NEO-PI-R domains of Neuroticism (b=0.40) and Agreeableness (b=ÿ0.12) entered as signi®cant predictors (R = 0.42, F(5,303)=12.95, P < 0.001). In the second stage the anxiety facet of Neuroticism (b=0.26) and the straightforwardness facet of Agreeableness (b=ÿ0.18) entered as signi®cant predictors (R = 0.48, F(12,296)=7.30, P < 0.001). The ®fth regression analysis used the Health Habits dimension as the dependent variable and the NEO-PI-R domain of Conscientiousness (b=0.24) entered as a signi®cant predictor (R = 0.24, F(5,303)=3.58, P < 0.005). In the second stage the deliberation facet of Conscientiousness (b=0.19) entered as a signi®cant predictor (R = 0.28, F(6,302)=4.28, P < 0.001). The ®ve domains of the NEO-PI-R were entered as predictor variables in regression analyses for each of the two higher-order IAS factors. In a regression of the ®rst higher-order factor of Health Anxiety, the Neuroticism domain (b=0.60) was the only signi®cant predictor in the regression equation (R = 0.60, F(5,303)=33.85, P < 0.001). In the second stage, only the Neuroticism facets of anxiety (b=0.40) and depression (b=0.16) entered as signi®cant predictors (R = 0.63, F(6,302)=32.94, P < 0.001). In a regression of the second higher-order factor of Health Behaviors, the Conscientiousness domain (b=0.25) was the only signi®cant predictor in the regression equation (R = 0.23, F(5,303)=3.32, P < 0.01. In the second stage, only the Conscientiousness facet of order (b=0.13) entered as a signi®cant predictor (R = 0.24, F(6,302)=3.09, P < 0.01). The same regression approach was used with a measure of non-speci®c anxiety, the STAI. The NEO-PI-R domains of Neuroticism (b=0.70) and Extraversion (b=ÿ0.20) were signi®cant predictors in the regression equation (R = 0.81, F(5,311)=119.32, P < 0.001). In the second stage the Neuroticism facets of anxiety (b=0.12), depression (b=0.44), and vulnerability to stress (b=0.22), as well as the Extraversion facets of assertiveness (b=ÿ0.10) and positive emotions (b=ÿ0.16) were signi®cant predictors in the regression equation (R = 0.84, F(12,304)=62.26, P < 0.001).
3. Discussion The results of this study serve to extend previous research on the nature of hypochondriasis as measured by the IAS. Speci®cally, the present study replicated the factor structure identi®ed in recent studies (Hadjistavropoulos & Asmundson, 1998; Hadjistavropoulos et al., in press). The meaning of each of the IAS dimensions was then further elucidated through their relations with higher-order and lower-order personality traits. These results highlight the importance of conducting regression analysis in addition to correlations: several associations between IAS factors and extraversion, along with agreeableness, were not apparent when the contribution of neuroticism was statistically controlled. The higher-order domain of neuroticism and one of its six lower-order facets, anxiety, were signi®cant predictors of several IAS dimensions. However, neuroticism and anxiety were most
106
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
strongly associated with variance in the Fear of Illness and Death dimension. For example, the beta weights and strength of prediction were more than twice as great for this dimension compared to the Treatment Experience dimension. In other words, Treatment Experience (e.g., visits to a physician) was less likely to be related to neuroticism or other personality traits. Although neuroticism denotes a general tendency to experience negative aect, the anxiety facet is relatively speci®c and refers to apprehension, fearfulness, and worry proneness (Costa & McCrae, 1992a). It is interesting to compare this result to the relationship between the NEO-PI-R and a measure of nonspeci®c or ``free-¯oating'' anxiety. In contrast to the Fear of Illness and Death IAS dimension, the STAI was predicted by three dierent facets of neuroticism and two facets of extraversion. It is hardly surprising that the neuroticism facet of anxiety predicted nonspeci®c anxiety, but it was interesting to see what other personality variables emerged as predictors. In particular, there were associations with a wider range of negative aect and with interpersonal diculties, whereas this was not true for Fear of Illness and Death. These ®ndings are consistent with the observations of Cox et al. (in press) concerning anxiety sensitivity and hypochondriasis. Anxiety sensitivity is an individual dierence variable de®ned as a fear of anxiety symptoms based on the belief that these symptoms have harmful consequences (Reiss, Peterson, Gursky & McNally, 1986) and it may represent a cognitive risk factor for panic disorder (McNally, 1994). Cox et al. (in press) found that one of the NEO-PIR traits that predicted anxiety sensitivity, including a dimension of anxiety sensitivity concerning fear of somatic sensations, was the neuroticism facet of self-consciousness. Cox et al. suggested that an important cognitive dierentiation between anxiety sensitivity and hypochondriasis may be that social/interpersonal concerns are salient only in the case of anxiety sensitivity. The Symptom Eects dimension was associated with a wider range of neuroticism facets, and was predicted by anxiety, depression, and vulnerability to stress. This latter facet of neuroticism relates to perceived inability to cope with stress and individuals scoring high on this facet may become ``dependent, hopeless, or panicked when facing emergency situations'' (Costa & McCrae, 1992a, p.16). This ®nding suggests that the Symptoms Eects dimension taps into more global life functioning, and extends beyond hypochondriasis. Disease Phobia and Conviction was not only positively associated with neuroticism and anxiety, but was also negatively associated with the agreeableness domain and its facet of straightforwardness. Costa and McCrae (1992a) note that the domain of agreeableness is one of interpersonal tendencies and also note the following in regard to the facet of straightforwardness: ``Straightforward individuals, that is, those individuals with high scores on this scale are frank, sincere, and ingenuous. Low scorers on this scale are more willing to manipulate others through ¯attery, craftiness, or deception'' (p. 17). This ®nding suggests that the Disease Phobia and Conviction dimension may also sometimes assess something broader than hypochondriasis per se. The Health Habits dimension seems to stray furthest from the core features of hypochondriasis. Rather than being associated with the maladaptive traits of neuroticism, it was best predicted by the conscientiousness domain and its facet of deliberation. These traits re¯ect purposefulness, caution, and self-control. In excessive levels it is conceivable that these traits may re¯ect features associated with hypochondriasis, but only in a peripheral sense.
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
107
The distinction between core and peripheral features of hypochondriasis contained in the IAS was further elucidated by the identi®cation of two higher-order factors: Health Anxiety and Health Behaviors. The ®rst factor of Health Anxiety appeared to capture the major elements of hypochondriasis, and was best predicted by neuroticism. In contrast, the second factor of Health Behaviors was comprised of the Treatment Experiences and Health Habits dimensions, and was best predicted by conscientiousness. In other words, rather than being indicative of psychopathology, this second IAS factor and its lower-order dimensions may often assess adaptive functioning. The present study examined hypochondriasis as an individual dierence variable comprised of concern and anxiety, rather than as a clinical somatoform disorder. One implication of this approach is that a dierent personality pro®le might prevail in clinical levels of hypochondriasis. For example, the trust facet of the agreeableness domain was signi®cantly correlated with some IAS factors in this study, although less so than neuroticism. It is possible that in clinical samples the negative association between trust and hypochondriasis could be more pronounced, and might help explain the failure to respond to medical reassurance seen in severe hypochondriasis. In conclusion, the results of this study help to clarify the dierential meaning of the dimensions of hypochondriasis underlying the IAS. Our ®ndings support the assertion of Hadjistavropoulos and Asmundson (1998) that fear or preoccupation with illness or pain is a fundamental component of hypochondriasis (the ®rst lower-order dimension in our study and also part of the ®rst higher-order factor). These results do not support the use of a total score for the IAS, and also suggest that some dimensions within the IAS (e.g., Health Habits) are too non-speci®c or peripheral to warrant their use in assessing the core features of hypochondriasis. In fact, based on the recent research ®ndings that have now accumulated, one has to wonder whether the current IAS can really be salvaged. An alternative might be to utilize some of its more valid items as part of the construction of a new measure of hypochondriasis. Further work on IAS dimensions and personality functioning in clinical samples also appears warranted.
Acknowledgements This research was supported by a grant from the Manitoba Health Research Council.
References American, Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: American Psychiatric Association. Barsky, A. J., & Klerman, G. L. (1983). Overview: hypochondriasis, bodily complaints, and somatic styles. American Journal of Psychiatry, 140, 273±283. Booth-Kewley, S., & Vickers, R. R. (1994). Associations between major domains of personality and health behavior. Journal of Personality, 62, 281±298. Brodman, K., Erdmann, A. J., & Wol, H. G. (1949). The Cornell Medical Index health questionnaire manual. New York: Cornell University.
108
B.J. Cox et al. / Personality and Individual Dierences 29 (2000) 99±108
Costa, P. T., & McCrae, R. R. (1985). Hypochondriasis, neuroticism, and aging: when are somatic complaints unfounded? American Psychologist, 40, 19±28. Costa, P. T., & McCrae, R. R. (1992a). Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) professional manual. Odessa, FL: Psychological Assessment Resources. Costa, P. T., & McCrae, R. R. (1992b). Normal personality assessment in clinical practice: the NEO Personality Inventory. Psychological Assessment, 4, 5±13. Cox B.J., Borger S.C., Taylor S., Fuentes K., Ross L.M. Anxiety sensitivity and the ®ve-factor model of personality. Behaviour Research and Therapy (in press). Ferguson, E., & Daniel, E. (1995). The Illness Attitudes Scales (IAS): a psychometric evaluation on a non-clinical population. Personality and Individual Dierences, 18, 463±469. Hadjistavropoulos, H. D., & Asmundson, G. J. G. (1998). Factor analytic investigations of the Illness Attitudes Scales: exploration in a chronic pain sample. Behaviour Research and Therapy, 36, 1185±1195. Hadjistavropoulos H.D., Frombach I., Asmundson G.J.G. Exploratory and con®rmatory factor analytic investigations of the Illness Attitudes Scales in a non-clinical sample. Behaviour Research and Therapy (in press). Hathaway, S. R., & McKinley, J. C. (1983). The Minnesota Multiphasic Personality Inventory manual. New York: Psychological Corporation. Jackson, D. N. (1989). Basic Personality Inventory manual. Port Huron, MI: Sigma Assessment Systems. Kellner, R. (1986). Somatization and hypochondriasis. New York: Praeger. Kellner, R., Abbott, P., Winslow, W. W., & Pathak, D. (1987). Fears, beliefs, and attitudes in DSM-III hypochondriasis. Journal of Nervous and Mental Disease, 175, 20±25. Kirmayer, L. J., Robbins, J. M., & Paris, J. (1994). Somatoform disorders: personality and the social matrix of somatic distress. Journal of Abnormal Psychology, 103, 125±136. McCrae, R. R. (1991). The ®ve-factor model and its assessment in clinical settings. Journal of Personality Assessment, 57, 399±414. McNally, R. M. (1994). Panic disorder: a critical analysis. New York: Guilford Press. Morey, L. (1991). Personality Assessment Inventory: professional manual. Odessa, FL: Psychological Assessment Resources. Norusis, M. J. (1988). SPSS-X Advanced statistics guide (2nd ed.). Chicago, IL: SPSS. Reiss, S., Peterson, R. A., Gursky, D. M., & McNally, R. J. (1986). Anxiety sensitivity, anxiety frequency, and the prediction of fearfulness. Behaviour Research and Therapy, 24, 1±8. Speckens, A. E. M., Spinhoven, P., Sloekers, P. P. A., Bolk, J. H., & van Hemert, A. M. (1996). A validation study of the Whitely Index, the Illness Attitudes Scales, and the Somatosensory Ampli®cation Scale in general medical and general practice patients. Journal of Psychosomatic Research, 40, 95±104. Spielberger, C. D. (1983). Manual for the state-trait anxiety inventory (Form Y). Palo Alto, CA: Consulting Psychologists Press. Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.). New York: Harper Collins. Tyrer, P., Fowler-Dixon, R., Ferguson, B., & Kelemen, A. (1990). A plea for the diagnosis of hypochondriacal personality disorder. Journal of Psychosomatic Research, 34, 637±642. Warwick, H. M. C., & Salkovskis, P. M. (1990). Hypochondriasis. Behaviour Research and Therapy, 28, 105±117. Zonderman, A. B., Heft, M. W., & Costa, P. T. (1985). Does the Illness Behavior Questionnaire measure abnormal illness behavior? Health Psychology, 4, 425±436.