- Email: [email protected]
Assessing Bodily Preoccupations is sufficient: Clinically effective screening for hypochondriasis
Journal of Psychosomatic Research 75 (2013) 526–531
Contents lists available at ScienceDirect
Journal of Psychosomatic Research
Assessing Bodily Preoccupations is sufficient: Clinically effective screening for hypochondriasis Volkmar Höfling ⁎, Florian Weck Department of Clinical Psychology and Psychotherapy, Institute of Psychology, Goethe University Frankfurt, Germany
a r t i c l e
i n f o
Article history: Received 4 August 2013 Received in revised form 11 October 2013 Accepted 21 October 2013 Keywords: Hypochondriasis Anxiety disorder Screening Sensitivity Specificity
a b s t r a c t Objective: Hypochondriasis is a persistent psychiatric disorder and is associated with increased utilisation of health care services. However, effective psychiatric consultation interventions and CBT treatments are available. In the present study, we provide evidence of clinically effective screening for hypochondriasis. We describe the clinically effective identification of patients with a high probability of suffering from hypochondriasis. This identification is achieved by means of two brief standardised screening instruments, namely the Bodily Preoccupation (BP) Scale with 3 items and the Whiteley-7 (WI-7) with 7 items. Methods: Both the BP scale and the WI-7 were examined in a sample of 228 participants (72 with hypochondriasis, 80 with anxiety disorders and 76 healthy controls) in a large psychotherapy outpatients' unit, applying the DSM-IV criteria. Cut-off values for the BP scale and the WI-7 were computed to identify patients with a high probability of suffering from hypochondriasis. Additionally, other self-report symptom severity scales were completed in order to examine discriminant and convergent validity. Data was collected from June 2010 to March 2013. Results: The BP scale and the WI-7 discriminated significantly between patients with hypochondriasis and those with an anxiety disorder (d = 2.42 and d = 2.34). Cut-off values for these two screening scales could be provided, thus identifying patients with a high probability of suffering from hypochondriasis. Conclusions: In order to reduce costs, the BP scale or the WI-7 should be applied in medical or primary care settings, to screen for patients with a high probability of hypochondriasis and to transfer them to further assessment and effective treatment. © 2013 Elsevier Inc. All rights reserved.
Introduction Severe health anxiety is the core feature of hypochondriasis, which affects a significant proportion of the population and leads to substantial medical costs [1–3]. However, there is evidence that the early identification of patients with hypochondriasis, and treatment at an early stage can save costs [4–6]. Therefore, effective and quick screening measures for hypochondriasis are needed, but should meet some specific criteria [7]. Firstly, they should be short and easy to apply in medical or psychiatric settings. Secondly, they should be reliable and valid. Thirdly, they should be able to correctly classify patients with health-related anxiety, within a group of patients with other relevant disorders. Several disorders can be considered as relevant disorders with regard to hypochondriasis. In the 1960s and 1970s, close relationships were observed between patients with depressive disorders and those with hypochondriasis, and there was discussion on whether hypochondriasis is a secondary phenomenon of depressive disorder. However, the link between hypochondriasis and depression is not as strong as once ⁎ Corresponding author at: Department of Clinical Psychology and Psychotherapy, Institute of Psychology, Goethe University Frankfurt, Varrentrappstrasse 40-42, 60486 Frankfurt, Germany. E-mail address: V.Hoefl[email protected] (V. Höfling). 0022-3999/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jpsychores.2013.10.011
believed, because many studies found a closer relationship between hypochondriasis and anxiety disorders, than between hypochondriasis and somatoform disorders [8]. Recently, the classification of hypochondriasis as an anxiety disorder was suggested because of conceptual [9] and phenomenological [10] similarities between anxiety disorders and hypochondriasis. Moreover, empirical research has demonstrated that hypochondriasis seems to be more closely connected to anxiety than to somatoform disorders [11]. Therefore, anxiety disorders should be considered as the relevant comparison group for identifying patients with hypochondriasis using a screening measure. Pilowsky [12] presented the Whiteley Index (WI) as an initial approach for the assessment of hypochondriasis. The WI reflects three dimensions of hypochondriasis, namely Bodily Preoccupation, Disease Phobia and Conviction of the Presence of Disease with Non-Response to Reassurance. The short version of the WI, the Whiteley-7 (WI-7) [13], was developed for the screening of hypochondriasis and demonstrated good classification results (sensitivity and specificity) for hypochondriasis in large primary care samples [14]. Another screening instrument is the Bodily Preoccupation (BP) Scale, which is a subscale of the Illness Attitude Scales (IAS) [15]. The BP scale demonstrated good classification results for hypochondriasis in a large general population sample [16]. However, to date, no study has been conducted of clinically diagnosed patients with hypochondriasis on the
V. Höfling, F. Weck / Journal of Psychosomatic Research 75 (2013) 526–531
one hand, and patients with a primary anxiety disorder and non-patient participants with no psychological disorder (healthy controls) on the other hand. One important outcome of such a study would be valid cut-off values for identifying patients with a high probability of hypochondriasis. Thus, the aim of the current study was to demonstrate the quality of the two brief hypochondriasis measures. Firstly, we hypothesised that, despite their brevity, the BP scale and the WI-7 would not be inferior to other instruments for hypochondriasis, regarding the ability to discriminate between patients with hypochondriasis and those with a primary anxiety disorder (Hypothesis 1). Secondly, we assumed that the BP scale and the WI-7 measure the same underlying construct (Hypothesis 2). Thirdly, diagnostically accurate prediction is possible, with the BP scale and the WI-7, through providing cut-off values to separate patients with hypochondriasis and those with an anxiety disorder, and to separate patients with hypochondriasis and healthy controls (Hypothesis 3). Finally, both instruments yield comparable relationships with other relevant measures for assessing hypochondriasis (Hypothesis 4). Method
527
clinicians who were specially trained in a two-day workshop on how to apply the SCID. In addition to the existence of a primary anxiety disorder or hypochondriasis, inclusion criteria were fluency and literacy in German, and informed consent. Exclusion criteria were a major medical illness (e.g. cancer), acute suicidality or suicidal tendencies, and a clinical diagnosis of substance addiction, schizophrenia or schizoaffective disorder, and bipolar disorders. Patients with primary anxiety disorders with comorbid hypochondriasis were also excluded. Fig. 1 provides an overview of the participation recruitment process. The entire sample of 228 participants consisted of three subsamples, with 72 participants having a primary diagnosis of hypochondriasis (hypochondriasis group), 80 with a primary diagnosis of an anxiety disorder (anxiety group) and 76 with no diagnosis (healthy control group). The healthy control group was recruited by four students from their circle of acquaintances as part of their master programme. The entire sample (56.1% female) and the subsamples were balanced with regard to gender, and the mean age was 37.5 (SD = 13.7; range = 18–75). Additionally, 37.7% of the participants with a diagnosis of hypochondriasis and 28.6% with an anxiety disorder also had another comorbid anxiety disorder, and 21.7% of the participants with hypochondriasis and 33.8% with an anxiety disorder also had another comorbid affective disorder. Institutional Board Approval was obtained for this study.
Participants Measures The study was conducted at the outpatient unit of the Department of Clinical Psychology and Psychotherapy at the University of Frankfurt, Germany (about 50 therapists and 1200 patients). In the outpatient unit, a randomised controlled trial was conducted (registered under NCT01119469) comparing the efficacy of cognitive therapy with exposure therapy. Patients with a diagnosis of hypochondriasis and those with a primary anxiety disorder were recruited in the outpatient unit for the current study. Diagnoses were determined by the Structured Clinical Interview for DSM-IV Disorders (SCID) [17] by experienced
Whiteley-7 (WI-7) The Whiteley-7 (WI-7) [13] is a short, 7-item-version of the original Whiteley Index [12]. Previous research demonstrated acceptable sensitivity (.82) and specificity (.81) coefficients of the WI-7 [14]. The German version of the WI-7 had acceptable internal consistency (α = .80) and correlated highly with the original 14-item-version (r = .94). The seven items of the WI-7 (e.g. “Do you worry a lot about your health?”) have a dichotomous response format [18,19].
Fig. 1. Participant flowchart.
528
V. Höfling, F. Weck / Journal of Psychosomatic Research 75 (2013) 526–531
Illness Attitude Scale (IAS) and Bodily Preoccupation Scale (BP) The BP is a subscale of the Illness Attitude Scale (IAS) [15] and consists of 3 items (e.g. “When you feel a sensation in your body, do you worry about it?”) with a five-point response scale. The 29 items of the IAS cover 9 subscales, along with the BP subscale, which are as follows: Worry About Illness (W), Concerns About Pain (CP), Health Habits (HH), Hypochondriacal Beliefs (HB), Thanatophobia (Th), Disease Phobia (DP), Treatment Experiences (TE) and Effects of Symptoms (ES). The German version [16] displayed good internal consistency for all items (α = .88) and acceptable internal consistency for many of the IAS subscales, e.g. for BP with α = .75. Additionally, the BP subscale was found to have high test–retest reliability with rtt = .93. Finally, the BP subscale yielded good sensitivity (.92) and specificity (.90) coefficients in previous research [16]. Scale for the Assessment of Illness Behaviour (SAIB) The SAIB evaluates illness behaviour (e.g. “When suffering from a medical complaint, I first take a look into medical dictionary to see what these symptoms might mean”) with 26 items on a four-point response scale applying five subscales [20]. For all items (α = .87) and the individual subscales (.65 ≤ α ≤ .76), acceptable to good internal consistencies were found. Furthermore, patients with a somatic syndrome score higher values on the SAIB than healthy controls. Multidimensional Inventory of Hypochondriacal Traits (MIHT) The MIHT is a 36-item self-report instrument for the assessment of hypochondriasis and health anxiety using a five-point response scale [21]. The MIHT covers cognitive, behavioural, perceptual and affective aspects of health anxiety. The original version of the MIHT and the German adaptation [22] yielded excellent reliability and validity. Furthermore, the factorial validity of the four aspects could be confirmed.
Brief Symptom Inventory (BSI) The BSI is a 53-item instrument assessing aspects of general psychopathology on a five-point response scale [29]. The measure we apply in our analysis is the BSI sum score. The German version of the BSI displayed very good internal consistency for the BSI sum score (α = .95) and acceptable to good internal consistencies for the nine subscales (.59 ≤ α ≤ .82) [30]. Statistical analysis Group differences Group differences for the measures outlined above between patients with hypochondriasis, those with a primary anxiety disorder and healthy controls were analysed with a one-way MANOVA and Scheffé post hoc tests (α = .05), to reveal which instruments are able to discriminate between patients with hypochondriasis and those with a primary anxiety disorder. Additionally, effect sizes for significant group differences were provided with 95% confidence intervals (CI). The analysis was conducted with IBM SPSS Statistics version 20. Factorial validity A confirmatory factor analysis (CFA) was carried out to demonstrate both the factorial validity of the WI-7 and the BP scale, respectively, and the closeness of their respective underlying latent factors. Therefore, the three items of the BP scale were specified as indicators for one latent factor and the seven items of the WI-7 as indicators for another latent factor. The correlation between both latent factors was expected to be very high (N .80). The data were analysed using Mplus version 6 [31] and model fit was evaluated according to accepted standards [32].
Cognitions about Body and Health (CABAH) The CABAH assesses dysfunctional cognitions with regard to the perception and interpretation of body sensations [23]. The items of five subscales were evaluated with a four-point response scale and the factorial validity of the five subscales could be supported. The internal consistency for the entire scale was α = .90 and acceptable internal consistencies (.67 ≤ α ≤ .88) were also reported for the five subscales [16]. Four of the five scales revealed significant differences between patients with hypochondriasis and a clinical control group. Thus, these 28 items of the CABAH were included in our study.
Sensitivity, specificity, and cut-offs Sensitivity, specificity, and the area under the ROC (Receiver Operating System) curve (AUC) were calculated with IBM SPSS Statistics version 20. The AUC of a scale is a global index of diagnostic accuracy and indicates the quality of a scale's diagnostic prediction (with 1.0 representing perfect diagnostic prediction). Furthermore, the Youden index (1950) was computed, because it provides an optimal cut-off value that maximises sensitivity + specificity − 1 for a measure that separates two groups [33]. Thus, two Youden indices were computed for each scale, one contrasting the hypochondriasis with the anxiety group and the other contrasting the hypochondriasis with the healthy control group.
Patient Health Questionnaire (PHQ) The PHQ-15 is a 15-item measure for assessing the severity of somatic symptoms on a three-point response scale [24]. Internal consistency proved to be good (α = .80). Additionally, the PHQ yielded acceptable coefficients for sensitivity (.78) and specificity (.71) for somatic syndromes [25]. In the German validation study, it was possible to discriminate significantly by means of the PHQ sum score between patients with a somatic syndrome and those with an evident medical syndrome [26].
Convergent and discriminant validity Pearson's correlations were computed to demonstrate the convergent and discriminant validity of the WI-7 and the BP scale by means of the same pattern of correlations between the BP scale and the WI-7, respectively and other measures. Furthermore, large correlations were expected between the BP scale and the WI-7 and specific hypochondriasis instruments such as the CABAH, the MIHT and the SAIB, while the correlations with other measures such as the BAI, the BDI-II, the BSI and the PHQ were expected to be moderate.
Beck's Depression Inventory (BDI-II) The German version of the BDI-II is a widely applied instrument for measuring symptoms of major depression, with 21 items answered on a four-point response scale, and yielded good coefficients for internal consistency (α N .89) [27].
Results
Beck's Anxiety Inventory (BAI) The German version of the BAI is a 21-item instrument assessing physiological, cognitive and somatic symptoms of anxiety on a fourpoint response scale [28]. The internal consistency was found to be good in several samples (α N .90). Furthermore, the BAI was able to discriminate significantly between patients with anxiety disorders and healthy controls.
Internal consistencies Table 1 presents the internal consistencies of all measures that were relatively good (α N .87). Especially the two screening scales for hypochondriasis showed high internal consistency, despite their small item number with α = .93 for the WI-7 and α = .93 for the BP scale. Group differences The one-way MANOVA yielded a significant Pillai's trace test (F = 21.6, df = 20, P b .001), indicating that there are substantial group differences within the applied measures. These differences were expected, especially because healthy controls on the one hand and patients with hypochondriasis and those with an anxiety disorder on the other hand, differ with respect to these instruments. However, some of the Scheffé post
V. Höfling, F. Weck / Journal of Psychosomatic Research 75 (2013) 526–531
529
Table 1 Means, standard deviations and homogeneity indicators (Cronbach's α) for all applied measures for the three subsamples (hypochondriasis group, anxiety group, control group).
WI-7 Bodily Preoccupations (BP) IAS CABAH MIHT SAIB PHQ BSI BAI BDI-II
Hypochondriasis (a)
Anxiety (b)
Mean
(SD)
Mean
(SD)
Control (c) Mean
(SD)
Group differences
5.3 9.2 64.0 34.8 116.1 1.5 10.9 39.5 19.0 13.6
(1.5) (2.5) (14.1) (11.0) (13.7) (0.4) (5.7) (28.7) (12.3) (8.4)
1.9 3.5 31.0 23.3 82.0 0.9 9.7 45.8 15.8 15.0
(2.3) (3.0) (19.9) (12.2) (25.1) (0.5) (5.8) (32.2) (12.1) (10.7)
0.2 1.2 13.2 13.3 66.3 0.5 4.0 12.0 3.0 3.7
(0.6) (1.2) (8.0) (7.3) (16.4) (0.3) (2.8) (11.1) (3.5) (4.5)
a a a a a a a a a a
NbNc NbNc NbNc NbNc NbNc NbNc =bNc =bNc =bNc =bNc
Cronbach's α
.93 .93 .96 .91 .97 .90 .87 .96 .93 .92
BAI, Beck's Anxiety Inventory; BDI-II, Beck's Depression Inventory; BSI, Brief Symptom Inventory; PHQ, Patient Health Questionnaire; CABAH, Cognitions about Body and Health; MIHT, Multidimensional Inventory of Hypochondriacal Traits; SAIB, Scale for the Assessment of Illness Behaviour; BP, Bodily Preoccupation Scale; WI-7, Whiteley-7. Hypochondriasis (n = 72), anxiety (n = 80), control (n = 76). hoc analyses also displayed significant differences between the hypochondriasis and the anxiety groups (Table 1). These two groups could be discriminated significantly (P b .001) by the MIHT (d = 1.77, 95% CI 1.40–2.14), SAIB (d = 1.77, 95% CI 1.40–2.14), CABAH (d = 1.11, 95% CI 0.78–1.44), IAS (d = 2.22, 95% CI 1.81–2.63), WI-7 (d = 2.34, 95% CI 1.93–2.75) and the BP scale (d = 2.42, 95% CI 2.01–2.83). No differences were found for the BSI, BAI, BDI-II and the PHQ. Thus, the WI-7 and the BP scale are not only very quickly administered measures with seven and three items, respectively, but they also yield the greatest discriminatory potential of all measures applied. Factorial validity The CFA model yielded good model fit (Satorra-Bentler-χ2 = 35.7, df = 34, RMSEA = .02, CFI = 1.00, WRMR = .36). The three items of the BP scale yielded large
loadings (N.88) on the underlying factor, as do the seven items of the WI-7 (N.86) on the respective underlying factor. The correlation between both factors of .89 was very large, indicating that both factors measure the same psychological variable. Sensitivity, specificity, and cut-off values The analysis contrasting the hypochondriasis and the anxiety groups resulted in the AUC = .91 for the BP scale and in the AUC = .89 for the WI-7. The cut-off value yielding the maximum Youden index for the BP scale was ≥6.5, with a sensitivity of .88 and a specificity of .83 (cf. Fig. 2b). The cut-off value yielding the maximum Youden index for the WI-7 was ≥4.8, with a sensitivity of .79 and a specificity of .83 (Fig. 2a). The analysis contrasting the hypochondriasis and the healthy control group resulted in the AUC = .99 for the BP scale and in the AUC = 1.00 for the WI-7. The cut-off value
a)
c)
b)
d)
Fig. 2. Four graphs showing sensitivity and specificity by cut-off values of the WI-7 or the BP scale, respectively. The cut-off value represents the maximised value for both sensitivity and specificity. The two graphs on the left refer to the discrimination between patients with hypochondriasis and those with anxiety disorder. The two graphs on the right refer to the discrimination between patients with hypochondriasis and healthy controls. a. WI-7 (hypochondriasis vs. anxiety disorder). b. BP (hypochondriasis vs. anxiety disorder). c. WI-7 (hypochondriasis vs. healthy controls). d. BP (hypochondriasis vs. healthy controls).
530
V. Höfling, F. Weck / Journal of Psychosomatic Research 75 (2013) 526–531
Table 2 Classification results for the WI-7 and the BP Scale. Predicted classification
BP
Hypochondriasis Anxiety Hypochondriasis Anxiety
WI-7
Hypochondriasis
Anxiety
63 14 65 20
9 66 7 60
72 80 72 80
BP, Bodily Preoccupation Scale; WI-7, Whiteley-7.
yielding the maximum Youden index for the BP scale was ≥3.5, with a sensitivity of .99 and a specificity of .97 (cf. Fig. 2d). The cut-off value yielding the maximum Youden index for the WI-7 was ≥2.5, with a sensitivity of .97 and a specificity of .99 (Fig. 2c). Table 2 shows the classification results with the WI-7 and the BP scale. With the WI-7, 82.2% of the cases were classified correctly, and with the BP scale, 84.9% of the cases (true positive and true negative rate). The classification quality of a screening instrument should be evaluated by the false negative rate and the false positive rate as well. For the BP scale, the false negative rate is 12.5% (9 cases) and the false positive rate is 17.5% (14 cases). For the WI-7, the false negative rate is 9.7% (7 cases) and the false positive rate is 25% (20 cases). For the BP scale and the WI-7, 50% of the cases of the false negative rate were identical, and 58.8% of the cases of the false positive rate were the same for both scales. Finally, 91.3% of the correct classified cases were the same patients for both scales. Convergent and discriminant validity Firstly, the pattern of correlations between the BP scale and the WI-7, respectively, with the other measures, were nearly identical, again indicating high convergent validity between the BP scale and the WI-7 (Table 3). Secondly, the largest correlation was found between the BP scale and the WI-7 with .87. Thirdly, other large correlations were found between the BP scale and the WI-7, and the specific instruments for hypochondriasis with correlations exceeding .7. Fourthly, there were moderate correlations (b.5) between the BDI-II and the BSI and somewhat larger correlations (.57 to .64) between the BAI and the PHQ.
Discussion The aim of this study was to provide empirical evidence of a clinically effective screening for hypochondriasis. Firstly, despite their brevity, the BP scale and the WI-7 discriminate effectively between patients with hypochondriasis and those with an anxiety disorder. In fact, these instruments perform as well as much longer hypochondriasis questionnaires (Hypothesis 1). Secondly, cut-off values for the BP scale and the WI-7 could be provided to separate patients with hypochondriasis from those with an anxiety disorder, and to separate hypochondriasis patients from healthy controls (Hypothesis 3). Thirdly, the BP scale measures practically the same underlying construct as the WI-7 (Hypothesis 2). Finally, both instruments yielded comparable relationships with other relevant measures (Hypothesis 4). Several measures (CABAH, MIHT, SAIB, WI-7, and BP scale) demonstrated a capacity to discriminate between patients with hypochondriasis and those with a primary anxiety disorder. However, only two of these measures can be regarded as appropriate for screening conditions because of their brevity, i.e. the BP scale and the WI-7. By contrast, the other measures are quite comprehensive. In line with previous findings, the BAI and the PHQ-15 failed to discriminate between patients with hypochondriasis and those with an anxiety disorder. The PHQ-15 mainly assesses somatic symptoms and was able to classify patients with somatoform syndromes [25]. However, there were no significant Table 3 Pearson's correlations between the WI-7, the BP Scale and other measures.
BP WI-7
BAI
BDI-II
BSI
PHQ
CABAH
MIHT
SAIB
BP
.61 .64
.47 .47
.48 .49
.57 .62
.75 .79
.81 .81
.81 .82
.87
BAI, Beck's Anxiety Inventory; BDI-II, Beck's Depression Inventory; BSI, Brief Symptom Inventory; PHQ, Patient Health Questionnaire; CABAH, Cognitions about Body and Health; MIHT, Multidimensional Inventory of Hypochondriacal Traits; SAIB, Scale for the Assessment of Illness Behaviour; BP, Bodily Preoccupation Scale; WI-7, Whiteley-7. All correlations were significant with P b .001.
differences in the PHQ-15 sum score with respect to the various anxiety disorders. Given that the American Psychiatric Association recommends the PHQ-15 as a “more in-depth assessment” measure for somatic symptoms, its use is not recommended for patients with hypochondriasis or illness anxiety disorder [34]. Similarly, the BAI was found to assess anxiety in particular, with regard to somatic symptoms like those of panic disorder. Furthermore, the BAI did not differentiate between various anxiety disorders and, additionally, the BAI measured depressive symptoms as well [35]. The fact that only hypochondriasis-specific measures (and not those for somatic symptoms) were able to discriminate between patients with hypochondriasis and those with an anxiety disorder, raises questions regarding the conceptualisation of hypochondriasis under the category somatoform disorder in DSM-IV, or somatic symptom and related disorders in DSM-5. Also, depressive symptoms (measured with the BDI-II) did not discriminate between patients with hypochondriasis and those with an anxiety disorder, thus suggesting that there is not a close relationship between hypochondriasis and affective disorders. In contrast, our results suggest a closer relationship to the anxiety disorders, because no differences appear between patients with an anxiety disorder and those with hypochondriasis, regarding anxiety symptoms (BAI) and somatic symptoms (PHQ-15). Moreover, it could be demonstrated in the current study, and in previous studies [36–38], that hypochondriasis is differentiable from anxiety disorders and can be regarded as a discrete disorder. For a psychiatric setting with primary care patients with anxiety disorders, the BP scale and the WI-7 are both very brief and cost-effective screening instruments. A BP score of ≥6 and a WI-7 score of ≥4 would reliably identify prospective patients with hypochondriasis. These patients could be transferred for further assessment and effective treatment if the diagnosis of hypochondriasis were confirmed. If the setting were different, for instance, if patients with hypochondriasis were identified in a sample of patients without psychological disorders, a BP score of ≥3 and a WI-7 score of ≥2 would be sufficient. The validity aspects of the BP scale and the WI-7 are almost equivalent. Both scales measure the same underlying construct, as demonstrated by the high correlation between both latent factors in the CFA. Thus, researchers or practitioners could decide whether they should apply either the WI-7 or the BP scale for their hypochondriasis screening. This would be in line with the comparable results regarding internal consistency, the AUC value and sensitivity and specificity. However, there might be disparities in the application of these two scales. The seven items of the WI-7 may be more appropriate for telephone screening, because of the dichotomous response format, whereas the three items of the BP scale could be part of an extensive assessment questionnaire at the beginning of a psychiatric consultation. With respect to the large factor loadings of the BP scale items, one could conclude that even one item would be sufficient for hypochondriasis screening. Although this idea may be appealing, it should be argued 1) that it is difficult to determine the reliability (e.g. internal consistency) of one item and 2) that the reliability decreases drastically with fewer items. However, reliable assessment is a necessary condition for valid hypochondriasis screening. Further validity aspects regarding the WI-7 and the BP scale were not addressed in this study, but are of particular importance, e.g. the ability to measure changes over time or to determine treatment outcome. In earlier investigations, the WI-7 demonstrated responsiveness in a primary care sample [39] and the BP scale for patients with hypochondriasis [40]. Limitations and strength A limitation of this study is that only patients with a primary anxiety disorder were compared to patients with hypochondriasis. Even though anxiety disorders are the most important diagnosis for the screening of hypochondriasis, because of their phenomenological similarity, other
V. Höfling, F. Weck / Journal of Psychosomatic Research 75 (2013) 526–531
disorders (e.g. affective disorders or personality disorders [41]) should be considered to provide further cut-off scores for other relevant clinical populations. However, other studies demonstrated that the WI-7 works well in the general population [42] and primary care [13,14] and that the BP scale is similarly effective in the general population [16]. However, the fundamental contribution of this study is that it is the first to compare several well-established measures of hypochondriasis for hypochondriasis screening in a large diagnosed clinical sample. Furthermore, the study demonstrated the effectiveness of two very short screening instruments and their potential to reduce the costs associated with undetected health anxiety and hypochondriasis. Funding This research was supported by grant WE 4654/2 from the German Research Foundation. Conflict of interest None. References [1] Sunderland M, Newby JM, Andrews G. Health anxiety in Australia: prevalence, comorbidity, disability, and service use. Br J Psychiatry 2013;202:56–61. [2] Barrett B, Tyrer P, Tyrer H, Cooper S, Crawford MJ, Byford S. An examination of the factors that influence costs in medical patients with health anxiety. J Psychosom Res 2012;73:56–62. [3] Fink P, Ørnbøl E, Christensen KS. The outcome of health anxiety in primary care. A twoyear follow-up study on health care costs and self-rated health. PLoS ONE 2010;5 [http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0009873]. [4] Hedman E, Andersson G, Andersson E, Ljótsson B, Rück C, Asmundson GJ, et al. Internet-based cognitive-behavioural therapy for severe health-anxiety: randomised controlled trial. Br J Psychiatry 2011;198:230–6. [5] Seivewright H, Green J, Salkovskis P, Barrett B, Nur U, Tyrer P. Cognitive-behavioural therapy for health anxiety in a genitourinary medicine clinic: randomized controlled trial. Br J Psychiatry 2008;193:332–7. [6] Hedman E, Ljótsson B, Andersson E, Rück C, Andersson G, Lindefors N. Effectiveness and cost offset analysis of group CBT for hypochondriasis delivered in a psychiatric setting: an open trial. Cogn Behav Ther 2010;39:239–50. [7] Zimmermann M, Chelminski I, Young D, Dalrymple K. A clinically useful anxiety outcome scale. J Clin Psychiatry 2010;71:534–42. [8] Starcevic V. Clinical features and diagnosis of hypochondriasis. In: Starcevic V, Lipsitt DR, editors. Hypochondriasis: modern perspectives on an ancient malady. Oxford: Oxford University Press; 2001. p. 21–60. [9] Olatunji BO, Deacon BJ, Abramowitz JS. Is hypochondriasis an anxiety disorder? Brit J Psychiatry 2009;194:481–2. [10] Fink P, Ørnbøl E, Toft T, Sparle KC, Frostholm L, Olesen F. A new, empirically established hypochondriasis diagnosis. Am J Psychiatry 2004;161:1680–91. [11] Gropalis M, Bleichhardt G, Witthöft M, Hiller W. Hypochondriasis, somatoform disorders, and anxiety disorders: sociodemographic variables, general psychopathology, and naturalistic treatment effects. J Nerv Ment Dis 2012;200:406–12. [12] Pilowsky I. Dimensions of hypochondriasis. Br J Psychiatry 1967;113:89–93. [13] Fink P, Ewald H, Jensen J, Sørensen L, Engberg M, Holm M, et al. Screening for somatisation and hypochondriasis in primary care and neurological inpatients: a seven-item scale for hypochondriasis and somatisation. J Psychosom Res 1999;46:261–73. [14] Conradt M, Cavanagh M, Franklin J, Rief W. Dimensionality of the Whiteley Index: assessment of hypochondriasis in an Australian sample of primary care patients. J Psychosom Res 2006;60:137–43. [15] Kellner R. Somatisation and hypochondriasis. Praeger: New York, NY; 1986.
531
[16] Weck F, Bleichardt G, Hiller W. Screening for hypochondriasis with the Illness Attitude Scales. J Pers Assess 2010;92:260–8. [17] First MB, Spitzer RL, Gibbon M, Williams JBW. Structured clinical interview for DSM-IV axis I disorder (SCID-I). Washington, DC: American Psychiatric Press; 1997. [18] Hiller W, Rief W, Fichter MM. Dimensional and categorical approaches to hypochondriasis. Psychol Med 2002;32:707–18. [19] Hinz A, Rief W, Brähler E. Hypochondrie in der Allgemeinbevölkerung: Teststatistische Prüfung und Normierung des Whiteley-Index [Hypochondriasis in the general population: psychometric properties and norm values of the Whiteley Index]. Diagnostica 2003;49:34–42. [20] Rief W, Ihle D, Pilger F. A new approach to assess illness behaviour. J Psychosom Res 2003;54:405–14. [21] Longley SL, Watson D, Noyes Jr R. Assessment of the hypochondriasis domain: the multidimensional inventory of hypochondriacal traits (MIHT). Psychol Assess 2005;17:4–13. [22] Witthöft M, Haaf A, Rist F, Bailer J. Erfassung von Krankheitsangst mit dem MIHT [Assessment of health anxiety with the Multidimensional Inventory of Hypochondriacal Traits (MIHT)]. Diagnostica 2010;56:2–12. [23] Rief W, Hiller W, Margraf J. Cognitive aspects of hypochondriasis and the somatisation syndrome. J Abnorm Psychol 1998;107:587–95. [24] Kroenke K, Spitzer RL, Williams JB. The PHQ-15: validity of a new measure for evaluating the severity of somatic symptoms. Psychosom Med 2002;64:258–66. [25] Van Ravesteijn H, Wittkampf K, Lucassen P, van de Lisdonk E, van den Hoogen H, van Weert H, et al. Detecting somatoform disorders in primary care with the PHQ-15. Ann Fam Med 2009;7:232–8. [26] Gräfe K, Zipfel S, Herzog W, Löwe B. Screening psychischer Störungen mit dem “Gesundheitsfragebogen für Patienten (PHQ-D)”. Ergebnisse der deutschen Validierungsstudie [Screening of psychic disorders with the Patient Health Questionnaire (PHQ). Results of the German validation study]. Diagnostica 2004;50:171–81. [27] Hautzinger M, Keller F, Kühner C. Das Beck Depressions-Inventar (BDI 2) [Beck's Depression Inventory (BDI 2)]. Frankfurt am Main. Germany: Harcourt Test Services GmbH; 1996. [28] Margraf J, Ehlers A. BAI: Beck Angst-Inventar. Manual. Deutsche Bearbeitung [BAI: Beck's Anxiety Inventory. Manual. German Adaptation]. Frankfurt am Main. Germany: Harcourt Test Services GmbH; 2007. [29] Derogatis LR. Brief Symptom Inventory (BSI): administration, scoring and procedures manual. 3rd ed. Minneapolis, MN: National Computer Services; 1993. [30] Franke GH. Brief Symptom Inventory von L. R. Derogatis (Kurzform der SCL-90-R) — Deutsche Version: Manual [short version of the SCL-90-R — German version: manual]. Beltz Test GmbH: Göttingen, Germany; 2000. [31] Muthén LK, Muthén BO. Mplus user's guide. 6th ed. Los Angeles, CA: Muthén & Muthén; 2010. [32] Schermelleh-Engel K, Moosbrugger H, Müller H. Evaluating the fit of structural equation models: test of significance and descriptive goodness-of-fit measures. Meth Psychol Res Online 2003;8:23–74. [33] Youden WJ. Index for rating diagnostic tests. Cancer 1950;3:32–5. [34] American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 5th ed. Arlington: American Psychiatric Association; 2013. [35] Beck AT, Steer R, Beck JS. Types of self-reported anxiety in outpatients with DSM-IIIR anxiety disorders. Anxiety Stress Coping 1993;6:43–55. [36] Hiller W, Leibbrand R, Rief W, Fichter MM. Differentiating hypochondriasis from panic disorder. J Anxiety Disord 2005;19:29–49. [37] Weck F, Neng JMB, Richtberg S, Stangier U. Dysfunctional beliefs about symptoms and illness in patients with hypochondriasis. Psychosomatics 2012;53:148–54. [38] Weck F, Neng JMB, Richtberg S, Stangier U. The restrictive concept of good health in patients with hypochondriasis. J Anxiety Disord 2012;26:792–8. [39] Christensen KS, Bech P, Fink P. Measuring mental health by questionnaires in primary care — unidimensionality, responsiveness and compliance. Eur Psychiatr Rev 2010;3:8–12. [40] Weck F, Neng J, Stangier U. The effects of attention training on the perception of bodily sensations in patients with hypochondriasis: a randomized controlled pilot trial. Cogn Ther Res 2013;37:514–20. [41] Sakai R, Nestoriuc Y, Nolido NV, Barsky AJ. The prevalence of personality disorders in hypochondriasis. J Clin Psychiatry 2010;71:41–7. [42] Lee S, Ng KL, Ma YL, Tsang A, Kwok KP. A general population study of the Chinese Whiteley-7 index in Hong Kong. J Psychosom Res 2011;71:387–91.
Contents lists available at ScienceDirect
Journal of Psychosomatic Research
Assessing Bodily Preoccupations is sufficient: Clinically effective screening for hypochondriasis Volkmar Höfling ⁎, Florian Weck Department of Clinical Psychology and Psychotherapy, Institute of Psychology, Goethe University Frankfurt, Germany
a r t i c l e
i n f o
Article history: Received 4 August 2013 Received in revised form 11 October 2013 Accepted 21 October 2013 Keywords: Hypochondriasis Anxiety disorder Screening Sensitivity Specificity
a b s t r a c t Objective: Hypochondriasis is a persistent psychiatric disorder and is associated with increased utilisation of health care services. However, effective psychiatric consultation interventions and CBT treatments are available. In the present study, we provide evidence of clinically effective screening for hypochondriasis. We describe the clinically effective identification of patients with a high probability of suffering from hypochondriasis. This identification is achieved by means of two brief standardised screening instruments, namely the Bodily Preoccupation (BP) Scale with 3 items and the Whiteley-7 (WI-7) with 7 items. Methods: Both the BP scale and the WI-7 were examined in a sample of 228 participants (72 with hypochondriasis, 80 with anxiety disorders and 76 healthy controls) in a large psychotherapy outpatients' unit, applying the DSM-IV criteria. Cut-off values for the BP scale and the WI-7 were computed to identify patients with a high probability of suffering from hypochondriasis. Additionally, other self-report symptom severity scales were completed in order to examine discriminant and convergent validity. Data was collected from June 2010 to March 2013. Results: The BP scale and the WI-7 discriminated significantly between patients with hypochondriasis and those with an anxiety disorder (d = 2.42 and d = 2.34). Cut-off values for these two screening scales could be provided, thus identifying patients with a high probability of suffering from hypochondriasis. Conclusions: In order to reduce costs, the BP scale or the WI-7 should be applied in medical or primary care settings, to screen for patients with a high probability of hypochondriasis and to transfer them to further assessment and effective treatment. © 2013 Elsevier Inc. All rights reserved.
Introduction Severe health anxiety is the core feature of hypochondriasis, which affects a significant proportion of the population and leads to substantial medical costs [1–3]. However, there is evidence that the early identification of patients with hypochondriasis, and treatment at an early stage can save costs [4–6]. Therefore, effective and quick screening measures for hypochondriasis are needed, but should meet some specific criteria [7]. Firstly, they should be short and easy to apply in medical or psychiatric settings. Secondly, they should be reliable and valid. Thirdly, they should be able to correctly classify patients with health-related anxiety, within a group of patients with other relevant disorders. Several disorders can be considered as relevant disorders with regard to hypochondriasis. In the 1960s and 1970s, close relationships were observed between patients with depressive disorders and those with hypochondriasis, and there was discussion on whether hypochondriasis is a secondary phenomenon of depressive disorder. However, the link between hypochondriasis and depression is not as strong as once ⁎ Corresponding author at: Department of Clinical Psychology and Psychotherapy, Institute of Psychology, Goethe University Frankfurt, Varrentrappstrasse 40-42, 60486 Frankfurt, Germany. E-mail address: V.Hoefl[email protected] (V. Höfling). 0022-3999/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jpsychores.2013.10.011
believed, because many studies found a closer relationship between hypochondriasis and anxiety disorders, than between hypochondriasis and somatoform disorders [8]. Recently, the classification of hypochondriasis as an anxiety disorder was suggested because of conceptual [9] and phenomenological [10] similarities between anxiety disorders and hypochondriasis. Moreover, empirical research has demonstrated that hypochondriasis seems to be more closely connected to anxiety than to somatoform disorders [11]. Therefore, anxiety disorders should be considered as the relevant comparison group for identifying patients with hypochondriasis using a screening measure. Pilowsky [12] presented the Whiteley Index (WI) as an initial approach for the assessment of hypochondriasis. The WI reflects three dimensions of hypochondriasis, namely Bodily Preoccupation, Disease Phobia and Conviction of the Presence of Disease with Non-Response to Reassurance. The short version of the WI, the Whiteley-7 (WI-7) [13], was developed for the screening of hypochondriasis and demonstrated good classification results (sensitivity and specificity) for hypochondriasis in large primary care samples [14]. Another screening instrument is the Bodily Preoccupation (BP) Scale, which is a subscale of the Illness Attitude Scales (IAS) [15]. The BP scale demonstrated good classification results for hypochondriasis in a large general population sample [16]. However, to date, no study has been conducted of clinically diagnosed patients with hypochondriasis on the
V. Höfling, F. Weck / Journal of Psychosomatic Research 75 (2013) 526–531
one hand, and patients with a primary anxiety disorder and non-patient participants with no psychological disorder (healthy controls) on the other hand. One important outcome of such a study would be valid cut-off values for identifying patients with a high probability of hypochondriasis. Thus, the aim of the current study was to demonstrate the quality of the two brief hypochondriasis measures. Firstly, we hypothesised that, despite their brevity, the BP scale and the WI-7 would not be inferior to other instruments for hypochondriasis, regarding the ability to discriminate between patients with hypochondriasis and those with a primary anxiety disorder (Hypothesis 1). Secondly, we assumed that the BP scale and the WI-7 measure the same underlying construct (Hypothesis 2). Thirdly, diagnostically accurate prediction is possible, with the BP scale and the WI-7, through providing cut-off values to separate patients with hypochondriasis and those with an anxiety disorder, and to separate patients with hypochondriasis and healthy controls (Hypothesis 3). Finally, both instruments yield comparable relationships with other relevant measures for assessing hypochondriasis (Hypothesis 4). Method
527
clinicians who were specially trained in a two-day workshop on how to apply the SCID. In addition to the existence of a primary anxiety disorder or hypochondriasis, inclusion criteria were fluency and literacy in German, and informed consent. Exclusion criteria were a major medical illness (e.g. cancer), acute suicidality or suicidal tendencies, and a clinical diagnosis of substance addiction, schizophrenia or schizoaffective disorder, and bipolar disorders. Patients with primary anxiety disorders with comorbid hypochondriasis were also excluded. Fig. 1 provides an overview of the participation recruitment process. The entire sample of 228 participants consisted of three subsamples, with 72 participants having a primary diagnosis of hypochondriasis (hypochondriasis group), 80 with a primary diagnosis of an anxiety disorder (anxiety group) and 76 with no diagnosis (healthy control group). The healthy control group was recruited by four students from their circle of acquaintances as part of their master programme. The entire sample (56.1% female) and the subsamples were balanced with regard to gender, and the mean age was 37.5 (SD = 13.7; range = 18–75). Additionally, 37.7% of the participants with a diagnosis of hypochondriasis and 28.6% with an anxiety disorder also had another comorbid anxiety disorder, and 21.7% of the participants with hypochondriasis and 33.8% with an anxiety disorder also had another comorbid affective disorder. Institutional Board Approval was obtained for this study.
Participants Measures The study was conducted at the outpatient unit of the Department of Clinical Psychology and Psychotherapy at the University of Frankfurt, Germany (about 50 therapists and 1200 patients). In the outpatient unit, a randomised controlled trial was conducted (registered under NCT01119469) comparing the efficacy of cognitive therapy with exposure therapy. Patients with a diagnosis of hypochondriasis and those with a primary anxiety disorder were recruited in the outpatient unit for the current study. Diagnoses were determined by the Structured Clinical Interview for DSM-IV Disorders (SCID) [17] by experienced
Whiteley-7 (WI-7) The Whiteley-7 (WI-7) [13] is a short, 7-item-version of the original Whiteley Index [12]. Previous research demonstrated acceptable sensitivity (.82) and specificity (.81) coefficients of the WI-7 [14]. The German version of the WI-7 had acceptable internal consistency (α = .80) and correlated highly with the original 14-item-version (r = .94). The seven items of the WI-7 (e.g. “Do you worry a lot about your health?”) have a dichotomous response format [18,19].
Fig. 1. Participant flowchart.
528
V. Höfling, F. Weck / Journal of Psychosomatic Research 75 (2013) 526–531
Illness Attitude Scale (IAS) and Bodily Preoccupation Scale (BP) The BP is a subscale of the Illness Attitude Scale (IAS) [15] and consists of 3 items (e.g. “When you feel a sensation in your body, do you worry about it?”) with a five-point response scale. The 29 items of the IAS cover 9 subscales, along with the BP subscale, which are as follows: Worry About Illness (W), Concerns About Pain (CP), Health Habits (HH), Hypochondriacal Beliefs (HB), Thanatophobia (Th), Disease Phobia (DP), Treatment Experiences (TE) and Effects of Symptoms (ES). The German version [16] displayed good internal consistency for all items (α = .88) and acceptable internal consistency for many of the IAS subscales, e.g. for BP with α = .75. Additionally, the BP subscale was found to have high test–retest reliability with rtt = .93. Finally, the BP subscale yielded good sensitivity (.92) and specificity (.90) coefficients in previous research [16]. Scale for the Assessment of Illness Behaviour (SAIB) The SAIB evaluates illness behaviour (e.g. “When suffering from a medical complaint, I first take a look into medical dictionary to see what these symptoms might mean”) with 26 items on a four-point response scale applying five subscales [20]. For all items (α = .87) and the individual subscales (.65 ≤ α ≤ .76), acceptable to good internal consistencies were found. Furthermore, patients with a somatic syndrome score higher values on the SAIB than healthy controls. Multidimensional Inventory of Hypochondriacal Traits (MIHT) The MIHT is a 36-item self-report instrument for the assessment of hypochondriasis and health anxiety using a five-point response scale [21]. The MIHT covers cognitive, behavioural, perceptual and affective aspects of health anxiety. The original version of the MIHT and the German adaptation [22] yielded excellent reliability and validity. Furthermore, the factorial validity of the four aspects could be confirmed.
Brief Symptom Inventory (BSI) The BSI is a 53-item instrument assessing aspects of general psychopathology on a five-point response scale [29]. The measure we apply in our analysis is the BSI sum score. The German version of the BSI displayed very good internal consistency for the BSI sum score (α = .95) and acceptable to good internal consistencies for the nine subscales (.59 ≤ α ≤ .82) [30]. Statistical analysis Group differences Group differences for the measures outlined above between patients with hypochondriasis, those with a primary anxiety disorder and healthy controls were analysed with a one-way MANOVA and Scheffé post hoc tests (α = .05), to reveal which instruments are able to discriminate between patients with hypochondriasis and those with a primary anxiety disorder. Additionally, effect sizes for significant group differences were provided with 95% confidence intervals (CI). The analysis was conducted with IBM SPSS Statistics version 20. Factorial validity A confirmatory factor analysis (CFA) was carried out to demonstrate both the factorial validity of the WI-7 and the BP scale, respectively, and the closeness of their respective underlying latent factors. Therefore, the three items of the BP scale were specified as indicators for one latent factor and the seven items of the WI-7 as indicators for another latent factor. The correlation between both latent factors was expected to be very high (N .80). The data were analysed using Mplus version 6 [31] and model fit was evaluated according to accepted standards [32].
Cognitions about Body and Health (CABAH) The CABAH assesses dysfunctional cognitions with regard to the perception and interpretation of body sensations [23]. The items of five subscales were evaluated with a four-point response scale and the factorial validity of the five subscales could be supported. The internal consistency for the entire scale was α = .90 and acceptable internal consistencies (.67 ≤ α ≤ .88) were also reported for the five subscales [16]. Four of the five scales revealed significant differences between patients with hypochondriasis and a clinical control group. Thus, these 28 items of the CABAH were included in our study.
Sensitivity, specificity, and cut-offs Sensitivity, specificity, and the area under the ROC (Receiver Operating System) curve (AUC) were calculated with IBM SPSS Statistics version 20. The AUC of a scale is a global index of diagnostic accuracy and indicates the quality of a scale's diagnostic prediction (with 1.0 representing perfect diagnostic prediction). Furthermore, the Youden index (1950) was computed, because it provides an optimal cut-off value that maximises sensitivity + specificity − 1 for a measure that separates two groups [33]. Thus, two Youden indices were computed for each scale, one contrasting the hypochondriasis with the anxiety group and the other contrasting the hypochondriasis with the healthy control group.
Patient Health Questionnaire (PHQ) The PHQ-15 is a 15-item measure for assessing the severity of somatic symptoms on a three-point response scale [24]. Internal consistency proved to be good (α = .80). Additionally, the PHQ yielded acceptable coefficients for sensitivity (.78) and specificity (.71) for somatic syndromes [25]. In the German validation study, it was possible to discriminate significantly by means of the PHQ sum score between patients with a somatic syndrome and those with an evident medical syndrome [26].
Convergent and discriminant validity Pearson's correlations were computed to demonstrate the convergent and discriminant validity of the WI-7 and the BP scale by means of the same pattern of correlations between the BP scale and the WI-7, respectively and other measures. Furthermore, large correlations were expected between the BP scale and the WI-7 and specific hypochondriasis instruments such as the CABAH, the MIHT and the SAIB, while the correlations with other measures such as the BAI, the BDI-II, the BSI and the PHQ were expected to be moderate.
Beck's Depression Inventory (BDI-II) The German version of the BDI-II is a widely applied instrument for measuring symptoms of major depression, with 21 items answered on a four-point response scale, and yielded good coefficients for internal consistency (α N .89) [27].
Results
Beck's Anxiety Inventory (BAI) The German version of the BAI is a 21-item instrument assessing physiological, cognitive and somatic symptoms of anxiety on a fourpoint response scale [28]. The internal consistency was found to be good in several samples (α N .90). Furthermore, the BAI was able to discriminate significantly between patients with anxiety disorders and healthy controls.
Internal consistencies Table 1 presents the internal consistencies of all measures that were relatively good (α N .87). Especially the two screening scales for hypochondriasis showed high internal consistency, despite their small item number with α = .93 for the WI-7 and α = .93 for the BP scale. Group differences The one-way MANOVA yielded a significant Pillai's trace test (F = 21.6, df = 20, P b .001), indicating that there are substantial group differences within the applied measures. These differences were expected, especially because healthy controls on the one hand and patients with hypochondriasis and those with an anxiety disorder on the other hand, differ with respect to these instruments. However, some of the Scheffé post
V. Höfling, F. Weck / Journal of Psychosomatic Research 75 (2013) 526–531
529
Table 1 Means, standard deviations and homogeneity indicators (Cronbach's α) for all applied measures for the three subsamples (hypochondriasis group, anxiety group, control group).
WI-7 Bodily Preoccupations (BP) IAS CABAH MIHT SAIB PHQ BSI BAI BDI-II
Hypochondriasis (a)
Anxiety (b)
Mean
(SD)
Mean
(SD)
Control (c) Mean
(SD)
Group differences
5.3 9.2 64.0 34.8 116.1 1.5 10.9 39.5 19.0 13.6
(1.5) (2.5) (14.1) (11.0) (13.7) (0.4) (5.7) (28.7) (12.3) (8.4)
1.9 3.5 31.0 23.3 82.0 0.9 9.7 45.8 15.8 15.0
(2.3) (3.0) (19.9) (12.2) (25.1) (0.5) (5.8) (32.2) (12.1) (10.7)
0.2 1.2 13.2 13.3 66.3 0.5 4.0 12.0 3.0 3.7
(0.6) (1.2) (8.0) (7.3) (16.4) (0.3) (2.8) (11.1) (3.5) (4.5)
a a a a a a a a a a
NbNc NbNc NbNc NbNc NbNc NbNc =bNc =bNc =bNc =bNc
Cronbach's α
.93 .93 .96 .91 .97 .90 .87 .96 .93 .92
BAI, Beck's Anxiety Inventory; BDI-II, Beck's Depression Inventory; BSI, Brief Symptom Inventory; PHQ, Patient Health Questionnaire; CABAH, Cognitions about Body and Health; MIHT, Multidimensional Inventory of Hypochondriacal Traits; SAIB, Scale for the Assessment of Illness Behaviour; BP, Bodily Preoccupation Scale; WI-7, Whiteley-7. Hypochondriasis (n = 72), anxiety (n = 80), control (n = 76). hoc analyses also displayed significant differences between the hypochondriasis and the anxiety groups (Table 1). These two groups could be discriminated significantly (P b .001) by the MIHT (d = 1.77, 95% CI 1.40–2.14), SAIB (d = 1.77, 95% CI 1.40–2.14), CABAH (d = 1.11, 95% CI 0.78–1.44), IAS (d = 2.22, 95% CI 1.81–2.63), WI-7 (d = 2.34, 95% CI 1.93–2.75) and the BP scale (d = 2.42, 95% CI 2.01–2.83). No differences were found for the BSI, BAI, BDI-II and the PHQ. Thus, the WI-7 and the BP scale are not only very quickly administered measures with seven and three items, respectively, but they also yield the greatest discriminatory potential of all measures applied. Factorial validity The CFA model yielded good model fit (Satorra-Bentler-χ2 = 35.7, df = 34, RMSEA = .02, CFI = 1.00, WRMR = .36). The three items of the BP scale yielded large
loadings (N.88) on the underlying factor, as do the seven items of the WI-7 (N.86) on the respective underlying factor. The correlation between both factors of .89 was very large, indicating that both factors measure the same psychological variable. Sensitivity, specificity, and cut-off values The analysis contrasting the hypochondriasis and the anxiety groups resulted in the AUC = .91 for the BP scale and in the AUC = .89 for the WI-7. The cut-off value yielding the maximum Youden index for the BP scale was ≥6.5, with a sensitivity of .88 and a specificity of .83 (cf. Fig. 2b). The cut-off value yielding the maximum Youden index for the WI-7 was ≥4.8, with a sensitivity of .79 and a specificity of .83 (Fig. 2a). The analysis contrasting the hypochondriasis and the healthy control group resulted in the AUC = .99 for the BP scale and in the AUC = 1.00 for the WI-7. The cut-off value
a)
c)
b)
d)
Fig. 2. Four graphs showing sensitivity and specificity by cut-off values of the WI-7 or the BP scale, respectively. The cut-off value represents the maximised value for both sensitivity and specificity. The two graphs on the left refer to the discrimination between patients with hypochondriasis and those with anxiety disorder. The two graphs on the right refer to the discrimination between patients with hypochondriasis and healthy controls. a. WI-7 (hypochondriasis vs. anxiety disorder). b. BP (hypochondriasis vs. anxiety disorder). c. WI-7 (hypochondriasis vs. healthy controls). d. BP (hypochondriasis vs. healthy controls).
530
V. Höfling, F. Weck / Journal of Psychosomatic Research 75 (2013) 526–531
Table 2 Classification results for the WI-7 and the BP Scale. Predicted classification
BP
Hypochondriasis Anxiety Hypochondriasis Anxiety
WI-7
Hypochondriasis
Anxiety
63 14 65 20
9 66 7 60
72 80 72 80
BP, Bodily Preoccupation Scale; WI-7, Whiteley-7.
yielding the maximum Youden index for the BP scale was ≥3.5, with a sensitivity of .99 and a specificity of .97 (cf. Fig. 2d). The cut-off value yielding the maximum Youden index for the WI-7 was ≥2.5, with a sensitivity of .97 and a specificity of .99 (Fig. 2c). Table 2 shows the classification results with the WI-7 and the BP scale. With the WI-7, 82.2% of the cases were classified correctly, and with the BP scale, 84.9% of the cases (true positive and true negative rate). The classification quality of a screening instrument should be evaluated by the false negative rate and the false positive rate as well. For the BP scale, the false negative rate is 12.5% (9 cases) and the false positive rate is 17.5% (14 cases). For the WI-7, the false negative rate is 9.7% (7 cases) and the false positive rate is 25% (20 cases). For the BP scale and the WI-7, 50% of the cases of the false negative rate were identical, and 58.8% of the cases of the false positive rate were the same for both scales. Finally, 91.3% of the correct classified cases were the same patients for both scales. Convergent and discriminant validity Firstly, the pattern of correlations between the BP scale and the WI-7, respectively, with the other measures, were nearly identical, again indicating high convergent validity between the BP scale and the WI-7 (Table 3). Secondly, the largest correlation was found between the BP scale and the WI-7 with .87. Thirdly, other large correlations were found between the BP scale and the WI-7, and the specific instruments for hypochondriasis with correlations exceeding .7. Fourthly, there were moderate correlations (b.5) between the BDI-II and the BSI and somewhat larger correlations (.57 to .64) between the BAI and the PHQ.
Discussion The aim of this study was to provide empirical evidence of a clinically effective screening for hypochondriasis. Firstly, despite their brevity, the BP scale and the WI-7 discriminate effectively between patients with hypochondriasis and those with an anxiety disorder. In fact, these instruments perform as well as much longer hypochondriasis questionnaires (Hypothesis 1). Secondly, cut-off values for the BP scale and the WI-7 could be provided to separate patients with hypochondriasis from those with an anxiety disorder, and to separate hypochondriasis patients from healthy controls (Hypothesis 3). Thirdly, the BP scale measures practically the same underlying construct as the WI-7 (Hypothesis 2). Finally, both instruments yielded comparable relationships with other relevant measures (Hypothesis 4). Several measures (CABAH, MIHT, SAIB, WI-7, and BP scale) demonstrated a capacity to discriminate between patients with hypochondriasis and those with a primary anxiety disorder. However, only two of these measures can be regarded as appropriate for screening conditions because of their brevity, i.e. the BP scale and the WI-7. By contrast, the other measures are quite comprehensive. In line with previous findings, the BAI and the PHQ-15 failed to discriminate between patients with hypochondriasis and those with an anxiety disorder. The PHQ-15 mainly assesses somatic symptoms and was able to classify patients with somatoform syndromes [25]. However, there were no significant Table 3 Pearson's correlations between the WI-7, the BP Scale and other measures.
BP WI-7
BAI
BDI-II
BSI
PHQ
CABAH
MIHT
SAIB
BP
.61 .64
.47 .47
.48 .49
.57 .62
.75 .79
.81 .81
.81 .82
.87
BAI, Beck's Anxiety Inventory; BDI-II, Beck's Depression Inventory; BSI, Brief Symptom Inventory; PHQ, Patient Health Questionnaire; CABAH, Cognitions about Body and Health; MIHT, Multidimensional Inventory of Hypochondriacal Traits; SAIB, Scale for the Assessment of Illness Behaviour; BP, Bodily Preoccupation Scale; WI-7, Whiteley-7. All correlations were significant with P b .001.
differences in the PHQ-15 sum score with respect to the various anxiety disorders. Given that the American Psychiatric Association recommends the PHQ-15 as a “more in-depth assessment” measure for somatic symptoms, its use is not recommended for patients with hypochondriasis or illness anxiety disorder [34]. Similarly, the BAI was found to assess anxiety in particular, with regard to somatic symptoms like those of panic disorder. Furthermore, the BAI did not differentiate between various anxiety disorders and, additionally, the BAI measured depressive symptoms as well [35]. The fact that only hypochondriasis-specific measures (and not those for somatic symptoms) were able to discriminate between patients with hypochondriasis and those with an anxiety disorder, raises questions regarding the conceptualisation of hypochondriasis under the category somatoform disorder in DSM-IV, or somatic symptom and related disorders in DSM-5. Also, depressive symptoms (measured with the BDI-II) did not discriminate between patients with hypochondriasis and those with an anxiety disorder, thus suggesting that there is not a close relationship between hypochondriasis and affective disorders. In contrast, our results suggest a closer relationship to the anxiety disorders, because no differences appear between patients with an anxiety disorder and those with hypochondriasis, regarding anxiety symptoms (BAI) and somatic symptoms (PHQ-15). Moreover, it could be demonstrated in the current study, and in previous studies [36–38], that hypochondriasis is differentiable from anxiety disorders and can be regarded as a discrete disorder. For a psychiatric setting with primary care patients with anxiety disorders, the BP scale and the WI-7 are both very brief and cost-effective screening instruments. A BP score of ≥6 and a WI-7 score of ≥4 would reliably identify prospective patients with hypochondriasis. These patients could be transferred for further assessment and effective treatment if the diagnosis of hypochondriasis were confirmed. If the setting were different, for instance, if patients with hypochondriasis were identified in a sample of patients without psychological disorders, a BP score of ≥3 and a WI-7 score of ≥2 would be sufficient. The validity aspects of the BP scale and the WI-7 are almost equivalent. Both scales measure the same underlying construct, as demonstrated by the high correlation between both latent factors in the CFA. Thus, researchers or practitioners could decide whether they should apply either the WI-7 or the BP scale for their hypochondriasis screening. This would be in line with the comparable results regarding internal consistency, the AUC value and sensitivity and specificity. However, there might be disparities in the application of these two scales. The seven items of the WI-7 may be more appropriate for telephone screening, because of the dichotomous response format, whereas the three items of the BP scale could be part of an extensive assessment questionnaire at the beginning of a psychiatric consultation. With respect to the large factor loadings of the BP scale items, one could conclude that even one item would be sufficient for hypochondriasis screening. Although this idea may be appealing, it should be argued 1) that it is difficult to determine the reliability (e.g. internal consistency) of one item and 2) that the reliability decreases drastically with fewer items. However, reliable assessment is a necessary condition for valid hypochondriasis screening. Further validity aspects regarding the WI-7 and the BP scale were not addressed in this study, but are of particular importance, e.g. the ability to measure changes over time or to determine treatment outcome. In earlier investigations, the WI-7 demonstrated responsiveness in a primary care sample [39] and the BP scale for patients with hypochondriasis [40]. Limitations and strength A limitation of this study is that only patients with a primary anxiety disorder were compared to patients with hypochondriasis. Even though anxiety disorders are the most important diagnosis for the screening of hypochondriasis, because of their phenomenological similarity, other
V. Höfling, F. Weck / Journal of Psychosomatic Research 75 (2013) 526–531
disorders (e.g. affective disorders or personality disorders [41]) should be considered to provide further cut-off scores for other relevant clinical populations. However, other studies demonstrated that the WI-7 works well in the general population [42] and primary care [13,14] and that the BP scale is similarly effective in the general population [16]. However, the fundamental contribution of this study is that it is the first to compare several well-established measures of hypochondriasis for hypochondriasis screening in a large diagnosed clinical sample. Furthermore, the study demonstrated the effectiveness of two very short screening instruments and their potential to reduce the costs associated with undetected health anxiety and hypochondriasis. Funding This research was supported by grant WE 4654/2 from the German Research Foundation. Conflict of interest None. References [1] Sunderland M, Newby JM, Andrews G. Health anxiety in Australia: prevalence, comorbidity, disability, and service use. Br J Psychiatry 2013;202:56–61. [2] Barrett B, Tyrer P, Tyrer H, Cooper S, Crawford MJ, Byford S. An examination of the factors that influence costs in medical patients with health anxiety. J Psychosom Res 2012;73:56–62. [3] Fink P, Ørnbøl E, Christensen KS. The outcome of health anxiety in primary care. A twoyear follow-up study on health care costs and self-rated health. PLoS ONE 2010;5 [http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0009873]. [4] Hedman E, Andersson G, Andersson E, Ljótsson B, Rück C, Asmundson GJ, et al. Internet-based cognitive-behavioural therapy for severe health-anxiety: randomised controlled trial. Br J Psychiatry 2011;198:230–6. [5] Seivewright H, Green J, Salkovskis P, Barrett B, Nur U, Tyrer P. Cognitive-behavioural therapy for health anxiety in a genitourinary medicine clinic: randomized controlled trial. Br J Psychiatry 2008;193:332–7. [6] Hedman E, Ljótsson B, Andersson E, Rück C, Andersson G, Lindefors N. Effectiveness and cost offset analysis of group CBT for hypochondriasis delivered in a psychiatric setting: an open trial. Cogn Behav Ther 2010;39:239–50. [7] Zimmermann M, Chelminski I, Young D, Dalrymple K. A clinically useful anxiety outcome scale. J Clin Psychiatry 2010;71:534–42. [8] Starcevic V. Clinical features and diagnosis of hypochondriasis. In: Starcevic V, Lipsitt DR, editors. Hypochondriasis: modern perspectives on an ancient malady. Oxford: Oxford University Press; 2001. p. 21–60. [9] Olatunji BO, Deacon BJ, Abramowitz JS. Is hypochondriasis an anxiety disorder? Brit J Psychiatry 2009;194:481–2. [10] Fink P, Ørnbøl E, Toft T, Sparle KC, Frostholm L, Olesen F. A new, empirically established hypochondriasis diagnosis. Am J Psychiatry 2004;161:1680–91. [11] Gropalis M, Bleichhardt G, Witthöft M, Hiller W. Hypochondriasis, somatoform disorders, and anxiety disorders: sociodemographic variables, general psychopathology, and naturalistic treatment effects. J Nerv Ment Dis 2012;200:406–12. [12] Pilowsky I. Dimensions of hypochondriasis. Br J Psychiatry 1967;113:89–93. [13] Fink P, Ewald H, Jensen J, Sørensen L, Engberg M, Holm M, et al. Screening for somatisation and hypochondriasis in primary care and neurological inpatients: a seven-item scale for hypochondriasis and somatisation. J Psychosom Res 1999;46:261–73. [14] Conradt M, Cavanagh M, Franklin J, Rief W. Dimensionality of the Whiteley Index: assessment of hypochondriasis in an Australian sample of primary care patients. J Psychosom Res 2006;60:137–43. [15] Kellner R. Somatisation and hypochondriasis. Praeger: New York, NY; 1986.
531
[16] Weck F, Bleichardt G, Hiller W. Screening for hypochondriasis with the Illness Attitude Scales. J Pers Assess 2010;92:260–8. [17] First MB, Spitzer RL, Gibbon M, Williams JBW. Structured clinical interview for DSM-IV axis I disorder (SCID-I). Washington, DC: American Psychiatric Press; 1997. [18] Hiller W, Rief W, Fichter MM. Dimensional and categorical approaches to hypochondriasis. Psychol Med 2002;32:707–18. [19] Hinz A, Rief W, Brähler E. Hypochondrie in der Allgemeinbevölkerung: Teststatistische Prüfung und Normierung des Whiteley-Index [Hypochondriasis in the general population: psychometric properties and norm values of the Whiteley Index]. Diagnostica 2003;49:34–42. [20] Rief W, Ihle D, Pilger F. A new approach to assess illness behaviour. J Psychosom Res 2003;54:405–14. [21] Longley SL, Watson D, Noyes Jr R. Assessment of the hypochondriasis domain: the multidimensional inventory of hypochondriacal traits (MIHT). Psychol Assess 2005;17:4–13. [22] Witthöft M, Haaf A, Rist F, Bailer J. Erfassung von Krankheitsangst mit dem MIHT [Assessment of health anxiety with the Multidimensional Inventory of Hypochondriacal Traits (MIHT)]. Diagnostica 2010;56:2–12. [23] Rief W, Hiller W, Margraf J. Cognitive aspects of hypochondriasis and the somatisation syndrome. J Abnorm Psychol 1998;107:587–95. [24] Kroenke K, Spitzer RL, Williams JB. The PHQ-15: validity of a new measure for evaluating the severity of somatic symptoms. Psychosom Med 2002;64:258–66. [25] Van Ravesteijn H, Wittkampf K, Lucassen P, van de Lisdonk E, van den Hoogen H, van Weert H, et al. Detecting somatoform disorders in primary care with the PHQ-15. Ann Fam Med 2009;7:232–8. [26] Gräfe K, Zipfel S, Herzog W, Löwe B. Screening psychischer Störungen mit dem “Gesundheitsfragebogen für Patienten (PHQ-D)”. Ergebnisse der deutschen Validierungsstudie [Screening of psychic disorders with the Patient Health Questionnaire (PHQ). Results of the German validation study]. Diagnostica 2004;50:171–81. [27] Hautzinger M, Keller F, Kühner C. Das Beck Depressions-Inventar (BDI 2) [Beck's Depression Inventory (BDI 2)]. Frankfurt am Main. Germany: Harcourt Test Services GmbH; 1996. [28] Margraf J, Ehlers A. BAI: Beck Angst-Inventar. Manual. Deutsche Bearbeitung [BAI: Beck's Anxiety Inventory. Manual. German Adaptation]. Frankfurt am Main. Germany: Harcourt Test Services GmbH; 2007. [29] Derogatis LR. Brief Symptom Inventory (BSI): administration, scoring and procedures manual. 3rd ed. Minneapolis, MN: National Computer Services; 1993. [30] Franke GH. Brief Symptom Inventory von L. R. Derogatis (Kurzform der SCL-90-R) — Deutsche Version: Manual [short version of the SCL-90-R — German version: manual]. Beltz Test GmbH: Göttingen, Germany; 2000. [31] Muthén LK, Muthén BO. Mplus user's guide. 6th ed. Los Angeles, CA: Muthén & Muthén; 2010. [32] Schermelleh-Engel K, Moosbrugger H, Müller H. Evaluating the fit of structural equation models: test of significance and descriptive goodness-of-fit measures. Meth Psychol Res Online 2003;8:23–74. [33] Youden WJ. Index for rating diagnostic tests. Cancer 1950;3:32–5. [34] American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 5th ed. Arlington: American Psychiatric Association; 2013. [35] Beck AT, Steer R, Beck JS. Types of self-reported anxiety in outpatients with DSM-IIIR anxiety disorders. Anxiety Stress Coping 1993;6:43–55. [36] Hiller W, Leibbrand R, Rief W, Fichter MM. Differentiating hypochondriasis from panic disorder. J Anxiety Disord 2005;19:29–49. [37] Weck F, Neng JMB, Richtberg S, Stangier U. Dysfunctional beliefs about symptoms and illness in patients with hypochondriasis. Psychosomatics 2012;53:148–54. [38] Weck F, Neng JMB, Richtberg S, Stangier U. The restrictive concept of good health in patients with hypochondriasis. J Anxiety Disord 2012;26:792–8. [39] Christensen KS, Bech P, Fink P. Measuring mental health by questionnaires in primary care — unidimensionality, responsiveness and compliance. Eur Psychiatr Rev 2010;3:8–12. [40] Weck F, Neng J, Stangier U. The effects of attention training on the perception of bodily sensations in patients with hypochondriasis: a randomized controlled pilot trial. Cogn Ther Res 2013;37:514–20. [41] Sakai R, Nestoriuc Y, Nolido NV, Barsky AJ. The prevalence of personality disorders in hypochondriasis. J Clin Psychiatry 2010;71:41–7. [42] Lee S, Ng KL, Ma YL, Tsang A, Kwok KP. A general population study of the Chinese Whiteley-7 index in Hong Kong. J Psychosom Res 2011;71:387–91.