Abstract
Preoccupation with personal physical appearance is common in society in developed countries. A term used to describe negative cognitions about one's appearance is ‘dysmorphic concern’, defined as, ‘an overconcern with an imagined or slight defect in physical appearance’ [1]. This approach side steps debate about when concern about physical appearance comes to constitute a specific disorder.
Dysmorphic concern as a disorder (termed ‘dysmorphophobia’) was included in the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) [2] as an atypical somatoform disorder. The term ‘dysmorphophobia’ has been challenged for its lack of systematic and universal definition and its indiscriminate use in describing both a symptom and a number of syndromes [3]. This imprecision, and the fact that it is not a phobia as such, led to its replacement with the label ‘body dysmorphic disorder’ (BDD) in DSM-III-R [4], a term retained in DSM-IV [5]. Implicit in this was the acceptance that there is a discrete entity characterized by overconcern with bodily appearance.
According to DSM-IV [5] diagnostic criteria, BDD describes a preoccupation with an imagined defect (or excessive concern with a slight defect) in physical appearance (i.e. ‘dysmorphic concern’). However, for a diagnosis of BDD to be made, the concern must cause significant distress or impairment in social, occupational, or other important areas of functioning. The criteria also stipulate that the bodily concern is not better accounted for by another psychiatric disorder such as overestimation of body size in anorexia nervosa.
Despite increased research interest in DSM-defined BDD, it remains clinically underrecognized, in part because many patients pursue surgical, dermatological, and other non-psychiatric treatment, and are too ashamed and embarrassed to reveal their concerns or to seek psychiatric help [6–9]. Furthermore, comorbidity between BDD and other disorders indicates that when patients consult mental health professionals, they may present with symptoms of depression, social phobia or obsessive– compulsive disorder (OCD) without mentioning symptoms of BDD. Thus, clinicians must inquire directly about symptoms to make the diagnosis [10].
Debate continues as to whether dysmorphic concern is a non-specific symptom occurring in a wide range of psychiatric disorders [4, 11, 12], or if BDD is a diagnostic entity in its own right [13–15]. It is often not clear whether comorbid disorders cause or are caused by BDD, or whether they simply coexist [10,15–25]. Also, OCD, anorexia nervosa and BDD share many symptoms (notably intrusive thoughts and rituals) [23–26], challenging the boundaries of these disorders. Despite the diagnostic controversy, dysmorphic concern can constitute a very real and distressing problem in its own right, whether as a symptom of another disorder, or as a ‘primary’ condition (i.e. BDD). Thus, it is important to have adequate ways of screening for dysmorphic concern in order to implement appropriate treatment strategies. Rosen and Reiter [14] developed a valid and reliable diagnostic instrument specifically for BDD, namely the Body Dysmorphic Disorder Examination (BDDE); this has been used subsequently in other studies [10]. Nevertheless, the interview version of the BDDE is lengthy (approximately 1 h administration time) and the self-report version may be too demanding on attention and concentration skills of some populations (e.g. psychiatric inpatients). Furthermore, it is aimed at diagnosing BDD as a disorder, rather than assessing dysmorphic concern as a symptom. This gives rise to the need for a much shorter instrument designed to screen for dysmorphic symptoms, without prejudice as to causality.
Oozthuizen et al. [1] met this need in a preliminary study carried out at a general adult psychiatric facility in Perth, Western Australia. The purpose of that study was the development of a self-report questionnaire, the Dysmorphic Concern Questionnaire (DCQ), to target dysmorphic concern as a symptom, rather than BDD as a diagnosis. The DCQ is a self-report instrument, which takes 2–5 min to complete. In a psychiatric inpatient sample [1], the DCQ exhibited good internal consistency and most of the variance could be explained by a single factor. However, that study used principal components analysis to test for factor structure of the DCQ. This method is a data reduction technique that is exploratory and assumes perfect reliability of the items (no error variance). Thus, it will always account for more variance than a more rigorous test of factor structure such as maximum likelihood factor analysis. The latter procedure is a confirmatory one that subjects the factor solution to a statistical test of goodness of fit. Moreover, Oosthuizen et al. [1] did not test for convergent validity, a form of construct validity whereby the measure in question correlates highly with another, already validated, measure of that construct.
Oosthuizen et al. [1] found dysmorphic concern to be associated with certain symptoms of depression, but not with psychotic symptoms. However, the authors did not address the relationship between dysmorphic concern and either OCD or social phobia, two disorders found to be highly associated with BDD in previous research (see above).
Thus, the current study aimed to assess more rigourously the factor structure of the DCQ, and to evaluate its convergent validity with an established measure of BDD, namely the BDDE [14]. A secondary aim was to evaluate the relationship between dysmorphic concern and depressed mood, social phobia, and OCD.
Method
Subjects were consecutive admissions to the open wards of a general psychiatric unit in Perth, Western Australia. All admitted patients were approached to participate, once they were clinically sufficiently settled. There were no specific exclusion criteria. Consenting patients were interviewed using the computerized life-time version of the Composite International Diagnostic Interview (CIDI-A), generating ICD-10 diagnoses. For the purposes of the current study, only the following sections were administered: (i) anxiety disorders; (ii) depressive disorders; (iii) manic and bipolar affective disorders; (iv) schizophrenia and other psychotic disorders; and (v) obsessive–compulsive and posttraumatic stress disorders. The eating disorders section was administered with only one participant as this was the primary reason she had been hospitalized. No other patients had been hospitalized for eating disorders, nor was there any indication of an eating disorder in these patients. Measures administered to all participants were: the Dysmorphic Concern Questionnaire, the Body Dysmorphic Disorder Examination, the Beck Depression Inventory, the Compulsion Checklist and the Fear Questionnaire.
The Dysmorphic Concern Questionnaire [1] consists of seven statements regarding the respondents' overconcern with a physical defect and their attempts to deal with the problem. The statements were derived from the dysmorphic concern literature and tap into affective, cognitive, and behavioural symptoms.
The Body Dysmorphic Disorder Examination (self-report version) is a validated standard diagnostic schedule for BDD [14]. This consists of 28 items regarding perception of appearance, rated on a sevenpoint scale with 0 indicating absence of distress concerning a particular physical feature, and 6 indicating extreme concern and/or impairment. The total score is the sum of ratings for these items. Diagnostic caseness was determined using both the total score and the cut-off score of 4 on particular items measuring level of distress and impairment.
The Beck Depression Inventory (BDI) (21-item) [27] was used with a range of possible scores from 0 (no depressive symptoms) to 63 (severe depression).
The Compulsion Checklist (CC) [28] is a 37-item inventory tapping into different compulsions and their duration, repetition and avoidance. Scores for each item range from 0 (no compulsions) to 3 (severe compulsions). Total score is the sum of all item ratings.
The Fear Questionnaire (FQ) [29] is a 15-item self-report questionnaire measuring anxiety symptoms. The 15 items are rated on a ninepoint scale ranging from 0 (would not avoid it) to 8 (always avoid it). These items make up a three-factor model describing agoraphobia, social phobia and blood/injury phobia.
Statistics
Reliability of the DCQ was evaluated using Cronbach's alpha. Construct validity was assessed in three ways: (i) the factor structure of the DCQ was evaluated to confirm a unidimensional scale using maximum likelihood factor analysis [30], it being hypothesized that only one latent variable would best represent the observed variables; (ii) the present study also compared the DCQ with the BDDE, for convergent validity [31]; and (iii) comparison of the mean level of negative body beliefs between those participants who were diagnosed with BDD versus those who were not (subjects who received a diagnosis of BDD were hypothesized to score higher on the DCQ on average than participants who did not).
Results
Sixty-five patients agreed to participate. The mean age was 35.5 years (range = 18–58) and 46% were female.
Diagnoses
Of the 65 participants, 29% had a primary diagnosis of major depression, 37% bipolar disorder, 14% schizophrenia, 3% dysthymic disorder and 9% an anxiety disorder; 55% had two or more comorbid diagnoses. Of the CIDI-A diagnoses, 85% agreed with the clinicians' diagnoses (made independently). Twenty-nine per cent of subjects met BDDE criteria for body dysmorphic disorder; the majority of these patients also met diagnostic criteria for affective disorder and social phobia (58% major depression, 32% bipolar, 58% social phobia). According to the DCQ, 71% of the total sample were considered to have significant dysmorphic concern (as defined by Oosthuizen et al. [1]).
Reliability coefficients
Cronbach's alpha for the total DCQ was 0.80, but removal of item five (‘Have you ever been told by others/doctor that you are normal in spite of you strongly believing that something is wrong with your appearance or bodily functioning?’) increased the coefficient to 0.81. Although this is not a significant increase, the interitem correlation matrix and factor analysis (see below) showed item five to be unrelated to most other items.
Construct validity
Construct validity was assessed in three ways. First, factor analysis was performed using the maximum likelihood method and specifying extraction of both one factor and two factors to assess the optimum solution [30]. The one-factor model was considered optimal. The loadings of items on the factor ranged from 0.40 (fair) to 0.87 (good) [32]. The exception was item number five with a loading of 0.37. The ?2 for the one-factor solution was 17.54 (df = 14, p < 0.20), the extracted factor accounting for 39% of the variance. With the removal of item five, the variance accounted for by the one-factor solution increased to 43% (?2 = 4.41, df = 9, p < 0.80). Item five was thus considered not to add anything to the DCQ, and was dropped from further analyses.
Construct validity was also examined by assessing the convergence of the DCQ with the BDDE. Hierarchical multiple regression analysis was used not only to test the validity of the DCQ, but also for hypothesized relations between DCQ and BDI, FQ (social phobia) and CC scores. The dependent variable was total score on the DCQ (with the omission of item five) and predictors entered into the equation were scores on the BDDE, BDI, FQ (social phobia score only) and CC. Zero-order correlation between variables were all statistically significant at the p < 0.0001 level.
The BDDE significantly predicted dysmorphic concern on the DCQ, accounting for 48% of the variance (F = 58.2, df = 1,63, p < 0.001). Though the BDI, FQ (social phobia), and CC were significantly positively correlated with the DCQ, adding all three to the regression equation only added 3% to the amount of variance accounted for. Furthermore, F-values were significant each time a variable was entered into the equation, but t-values indicated that only the BDDE added to predictive power in the final equation.
The third assessment of construct validity was a test of mean DCQ differences between BDD diagnostic groups. A one-way ANOVA confirmed that those individuals who were diagnosed with BDD according to the BDDE (n = 19) scored more highly on the DCQ (mean = 8.80) than did non-BDD sufferers (n = 46, mean = 4.50, F = 19.27, df = 1,63, p < 0.001). Moreover, Levene's test for homogeneity of variance was 0.36 (p < 0.05), indicating that DCQ variance was greater between BDD groups than within groups. Thus, BDD sufferers had a significantly higher degree of negative body beliefs on average than did non-BDD participants.
Relationships involving body dysmorphic disorder
Relationships involving BDD were evaluated using a second multiple regression equation. The BDDE score was the dependent variable and the independent variables were entered in the following order: BDI, CC, FQ (social phobia) and DCQ. Consistent with the theoretical rationale that negative body beliefs may be regarded as symptomatic of one or more of the disorders, the BDI, CC and FQ (social phobia) were regarded as antecedent to the DCQ.
Both the DCQ and the BDI accounted for significant unique variance in the BDDE. The BDI explained 31% of the variance in the BDDE when no other variables were included in the equation. The inclusion of the CC and FQ (social phobia) accounted for an additional 5% of the BDDE variance. The DCQ, however, accounted for an extra 20% of the variance in BDDE when the other variables were in the equation. The combination of all four variables accounted for a total of 56% of the variance in the BDDE. The DCQ was the strongest predictor of BDDE score, while the BDI was the only other significant predictor (t = 2.19, p < 0.001). Thus, negative body beliefs and depression contribute significantly to the variability in the BDDE. Moreover, negative beliefs about one's body are uniquely related to the BDDE beyond the negative cognitions associated with depression and it is these negative body beliefs that have the stronger relationship with BDD.
Discussion
The principle objectives of the current study were to validate the Dysmorphic Concern Questionnaire [1] in a psychiatric inpatient population and investigate relationships between dysmorphic concern and BDD, depression, social phobia and OCD.
Overall, the results suggest that the DCQ is a reliable and valid instrument that is sensitive to dysmorphic concern. There was evidence of good internal consistency, suggesting that all items (with the exception of item five) measure the same construct.
The large amount of variance in the DCQ that was accounted for by the BDDE, after controlling for other variables, suggests that the two measures are assessing the same construct. Furthermore, participants who received a diagnosis of BDD scored higher on the DCQ than did those who did not. Body dysmorphic disorder symptomatology was best defined by the presence of negative body beliefs as measured by the DCQ, rather than negative beliefs associated with depression, OCD and social phobia.
These results have important theoretical implications. As noted, there was a significant and positive relationship between all of these measures, but the BDI, FQ (social phobia) and CC did not predict DCQ score in the regression analysis beyond the predictive ability of the BDDE. This by no means discounts the relationship between BDD and those other disorders, but offers clarity to those relationships. Thus, negative body beliefs in BDD appear to be of a specific form not fully encompassed by the negative cognitions of a more general nature associated with depression.
Nevertheless, it is important to note that the relationship between BDD and negative body beliefs may not remain when examined in the context of other comorbid disorders. Interestingly, BDD comorbidity results indicated that BDD, as diagnosed by the BDDE, was not present in any patients who did not meet diagnostic criteria for at least one other psychiatric disorder according to the CIDI-A. This also held true for dysmorphic concern. However, one might expect high levels of comorbidity in a clinical sample such as ours, and the DCQ remains to be validated in a non-clinical sample.