The Italian version of the Obsessive Compulsive Inventory: Its psychometric properties on community and clinical samples

Journal of Anxiety Disorders 23 (2009) 204–211

Contents lists available at ScienceDirect

Journal of Anxiety Disorders

The Italian version of the Obsessive Compulsive Inventory: Its psychometric properties on community and clinical samples Claudio Sica a,*, Marta Ghisi b, Gianmarco Altoe` b, Luigi Rocco Chiri a, Sandro Franceschini a, Davide Coradeschi a, Gabriele Melli a a b

Department of Psychology, University of Firenze, Italy Department of General Psychology, University of Padova, Italy

A R T I C L E I N F O

A B S T R A C T

Article history: Received 18 January 2008 Received in revised form 28 June 2008 Accepted 2 July 2008

The aim of the study was to evaluate psychometric properties of the Obsessive Compulsive Inventory (OCI) on Italian community and clinical samples. The Italian version of the 42-item OCI was administered to a sample of 340 individuals belonging to the general population and to 88 patients with obsessive compulsive (OCD) or other anxiety disorders. Four different internal structures of the OCI were compared through confirmatory factor analysis (CFA): the figures for the model with six factors and 18 items (OCI-R) met the best criteria for adequacy of fit. The six scales showed on average a 10% of common variance in the community sample and 8% in the clinical sample. The OCI-R subscales showed good internal consistency and temporal stability, with the exception of washing and mental neutralizing subscales which showed a strong alpha coefficient only in the OCD sample. Psychometric data for the OCI-R were insensitive to age and sex, whereas an effect of education was found. Concurrent validity was demonstrated, since the OCI-R subscales showed a pattern of specific correlations with another conceptually related self-report measure. Moreover, although the OCI-R was positively correlated with measures of depression, anxiety, and worry, the correlations were weaker than those with the other measure of OCD symptoms. The OCI-R clearly differentiated OCD patients from non-OCD anxious patients and nonclinical controls with the exception of hoarding subscale. However, the hoarding scale discriminated OCD patients who presented hoarding symptoms from OCD counterparts without such symptoms. Thus, the OCI-R proved to be a reliable and valid measure of obsessive compulsive symptoms in the Italian context. ß 2008 Elsevier Ltd. All rights reserved.

Keyword: Obsessive compulsive disorder

1. Introduction Obsessive compulsive disorder (OCD) is a well-known form of psychopathology which is characterized by persistent, intrusive, and distressing obsessions (persistent thoughts, impulses, or images) or compulsions (repetitive, excessive behaviors or mental acts). Recent research on OCD underscores the considerable heterogeneity of this disorder. For example, in their review McKay et al. (2004) have identified as many as nine subtypes or replicable dimensions of OCD: contamination/washing, harming/checking, hoarding, symmetry/ordering, obsessionals, sexual and religious, certainty, sexual-somatic, and contamination/harming. Studies on OCD heterogeneity have several implications for researchers and clinicians, since obsessive compulsive (OC) subtypes may differ-

* Corresponding author at: Dipartimento di Psicologia, University of Firenze, Via San Niccolo`, 93, 50125, Firenze, Italy. Fax: +55 2345326. E-mail address: claudio.sica@unifi.it (C. Sica). 0887-6185/$ – see front matter ß 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.janxdis.2008.07.001

entially relate to other clinical disorders, cognitive features, neuropsychological deficits, personality traits, sociodemographic aspects, and response to treatment (e.g., Baer, 1994; Mataix-Cols et al., 2002; Minichiello, Baer, Jenike, & Holland, 1990; Sher, Frost, Kushner, Crews, & Alexander, 1989; Simpson et al., 2006; Tolin, Woods, & Abramowitz, 2003). The heterogeneity of OCD calls for instruments that reliably assess the various empirically based symptom presentations. One measure that has been purposely designed to correspond to symptom-based models of OCD and has received extensive empirical validation is the Obsessive Compulsive Inventory (OCI; Foa, Kozak, Salkovskis, Coles, & Amir, 1998). Two different versions of the OCI have been considered in the literature: (a) the original version composed by 42 items and seven rationally derived subscales (washing, checking, doubting, ordering, obsessing, hoarding, and mental neutralizing); (b) a short version containing 18 items and six subscales (washing, checking, ordering, obsessing, hoarding, and mental neutralizing), based on factor analysis of the OCI in a US clinical sample (OCI-R; Foa et al., 2002).

C. Sica et al. / Journal of Anxiety Disorders 23 (2009) 204–211

Several reports provided support for the reliability and validity of the OCI and OCI-R and showed strong convergence with established measures of OCD, moderate to high internal consistency across the various subscales, and adequate to high test– retest stability depending on the time interval (e.g., Abramowitz & Deacon, 2006; Huppert et al., 2007; Simonds, Thorpe, & Elliott, 2000). The internal structure of the OCI-R has been replicated in multiple languages including French (Zermatten, Van-der-Linden, Jermann, & Ceschi, 2006), German (Gonner, Leonhart, & Ecker, 2008), Icelandick (Smari, Olason, Eythorsdottir, & Frolunde, 2007), and Spanish (Fullana et al., 2005), whereas only one study concerned the factorial composition of the original OCI (Wu & Watson, 2003). In this study, based on a factor analysis in American undergraduate students, the authors suggested that a five-factor structure, rather than the seven rationally derived subscales, best captured the structure of symptoms measured by the OCI. In addition, Foa et al. (2002) also suggested that the proposed sevenfactor structure of the original OCI may be in question. In sum, what emerges from the literature is that the OCI-R is a robust measure of OC symptoms, whereas more studies are warranted for the OCI. On the other hand, the brevity of the OCIR subscales (three items for each of the six dimensions) is of concern for some scholars (Clark, Antony, Beck, Swinson, & Steer, 2005), even though the OCI itself contains two dimensions composed by three items (hoarding and doubting). As such, it could be worthwhile examining both the long and short version of the questionnaire within the same study. To our knowledge, no studies have been conducted to compare the OCI and the OCI-R in a cultural context outside the US. Actually, if the OCI-R should show a stronger factor structure than OCI and good psychometric characteristics, then it would be possible to conclude that OCD may not vary substantially cross-culturally in this regard, and also may provide support for the use of the subscales of the OCI-R over the OCI. On the contrary, if a different picture would emerge (i.e., no substantial differences between the two versions or stronger characteristics of the OCI over the OCI-R), then potential cultural differences that may lead to this may be discussed. The main purpose of the present study was therefore to evaluate the psychometric properties of both the OCI and the OCI-R on Italian community and clinical samples. Virtually no studies have tested the psychometric properties of the OCI and OCI-R on community samples, a surprising omission given theoretical assumptions that OC phenomena lie on a continuum from normality to psychopathology (e.g., Burns, Formea, Keortge, & Sternberger, 1995; Salkovskis & Harrison, 1984; Sterneberger & Burns, 1990). The OCI itself was devised to ‘‘provide an instrument that can be readily administered to both clinical and non-clinical populations’’ (Foa et al., 1998, p. 207). A large community sample permits study of the factor structure of the OCI, convergent and divergent validity, and examination of possible gender, age, and education effects on OCI total and subscale scores. Regarding this last issue, we hypothesized no differences in age and gender because several studies on selfreported obsessional features showed no differences across these characteristics (Kyrios, Bhar, & Wade, 1996; MacDonald & De Silva, 1999; Sterneberger & Burns, 1990). Lastly, the clinical samples permit to evaluate the criterionoriented and discriminant validity of the Italian version of the OCI. 2. Method 2.1. Participants and procedure Subjects were 340 community individuals (51% male) enrolled in four different middle-size towns of Northern Italy. All participants were Caucasian. The mean age of the sample was 33.3 (S.D. = 13;

205

range = 16–60) and the mean years of education was 15.4 (S.D. = 3; range = 8–21). Marital status was 58.2% single, 37.9% married or cohabitating, 3% separated or divorced, and 0.9% widowed. The employment profile of the total sample was: 52.1% full-time job, 33.7% students, 3.8% part time job, 2.1% unemployed, 2.1% retired, 0.9% full time homemaker, and 5.3% other. To obtain data about the temporal stability of the OCI, a subgroup of 50 participants (40% females; mean age = 27.7 years; S.D. = 5.3) completed the questionnaires on two occasions 4 weeks apart. Participants were recruited during free-access public conferences about psychological topics of general interest. We acknowledge that such form of recruitment is unusual; however, in previous studies we obtained adequate community samples through this method (e.g., Sica & Ghisi, 2007). Clinical individuals were patients with either DSM-IV diagnosed obsessive compulsive disorder (OCD group) or any DSM-IV diagnosed anxiety disorders except OCD and simple phobia (Anxious group) as their most severe problem. Patients with secondary comorbid Axis-I or Axis-II diagnoses were included. Non-suitable patients were those with a current or past psychotic disorder, dementia, mental retardation or a current substance use disorder. In addition, anxious patients were excluded if they had a current or past obsessive compulsive disorder. OCD and anxious individuals were recruited from both two outpatient mental health clinics and 10 different private settings located in Northern and Central Italy. During the routinary assessment phase, patients were interviewed by one of the members of our research team (all Ph.D. level psychologists experienced in diagnosing psychiatric disorders) using the Structured Clinical Interview for DSM-IV (First, Spitzer, Gibbon, & Williams, 1996), to establish DSM-IV diagnoses. For OCD individuals, the most prominent obsessions and compulsions were also recorded during the interview. Although inter-rater reliability for the main diagnosis was not examined formally, each case was audio-recorded and carefully reviewed in supervisory meetings and all diagnoses were reached by rater consensus. After being assessed, suitable patients were invited to participate in the study. All individuals participated on a voluntary basis and gave their written consent before entering the study. Eligible participants were requested to complete a battery of self-report measures administered individually. The sequence of measures was rotated to control for order effects. The final sample consisted of 52 OCD patients and 36 anxious patients (all were Caucasian). In the latter group, the frequency of each principal anxiety disorder diagnosis was as follows: 36% panic disorder without agoraphobia, 25% panic disorder with agoraphobia, 28% social phobia, and 11% generalized anxiety disorder. In addition we found that 20% had a secondary comorbid Axis-I diagnosis (major depressive disorder = 17%, dysthymic disorder = 3%) and 14% had an Axis-II diagnosis (three with avoidant personality disorder, one with dependent personality disorder and one with borderline personality disorder). In the OCD group 25.4% had a secondary comorbid Axis-I diagnosis (anxiety disorders = 10%, major depressive disorder = 15.4%) and 8% had an Axis-II diagnosis (two had dependent personality disorder, one had borderline personality disorder and one had obsessive compulsive personality disorder). Table 1 provides descriptive statistics on various demographic variables for the two clinical groups as well as for a third group of 47 individuals, randomly selected from the sample of 340 subjects belonging to community (community controls, CC), for comparative purposes. The three groups were equivalent with respect to all demographic variables (all ps > 0.10). As expected, the OCD sample scored significantly higher than the anxious group (F(1,83) = 81.5, p < 0.001) on the self-report version of the YaleBrown Obsessive Compulsive Scale (Y-BOCS; Goodman et al.,

C. Sica et al. / Journal of Anxiety Disorders 23 (2009) 204–211

206

Table 1 Demographic data and levels of symptomatology across the three groups

Age Years of education % of females % of married/cohabitant % of employed % of unemployed Y-BOCS BAI

OCD (52)

AG (36)

CC (47)

x2 or F associated probability

Significant SNK post-hoc comparison (p < 0.05)

33 (9.6) 13.5 (3.6) 30 40 51 10 21.6 (5.9) 15.3 (9.5)

31.2 (8.4) 15 (4.2) 52 32 62 9 8.2 (6.2) 16.4 (12.2)

32.4 (12.8) 14.9 (3.1) 36 28 61 0 – 5.5 (6.1)

NS NS NS NS NS NS 0.001 0.001

– – – – – – OCD > AG OCD, AG > CC

Note—NS: non-significant; standard deviations in brackets; SNK: Student–Newman–Keuls; OCD: obsessive compulsive disorder group; AG: anxious group; CC: community controls; Y-BOCS: Yale–Brown Obsessive Compulsive Scale; BAI: Beck Anxiety Inventory. Cronbach’s alpha values for the Y-BOCS: OCD = 0.82; AG = 0.85.

1989; Sica et al., 2004), a standard measure of the severity of OCD. The two clinical groups did not differ significantly in overall level of anxiety as measured by the Beck Anxiety Inventory (see below; p > 0.65), while the community sample reported significantly lower anxiety score than the two clinical groups (F(2,130) = 17.3, p < 0.001). 2.2. Measures 2.2.1. Translation of the OCI Standard steps outlined in the psychology literature guided the translation process used in this study (e.g., Brislin, 1986). In the first step, three independent researchers translated the questionnaire from English to Italian and then reached agreement on a common version. Idiomatic Italian at the sixth-grade level was used for this step. Moreover, the researchers reviewed the common version to ensure there were no colloquialisms, slang, or esoteric phrases that would make interpretations difficult. The shared form was then back-translated by a bilingual person with an extensive knowledge of psychological research. The back-translation proved to be nearly identical to the original one. As a final step, the OCI items of the Italian version were rated by five experts in anxiety disorders. Each expert rated the items on a 5-point scale (1 = not at all, 5 = extremely) for clarity (the extent to which the item is clearly described). The experts’ ratings indicated excellent clarity (mean across all items = 4.6; DS = 0.6) indicating that further item refinement was unnecessary. 2.2.2. Other measures of psychopathology All participants completed a background information questionnaire and the following measures: The Padua Inventory (PI; Sanavio, 1988) is a 60-item self-report instrument developed in Italy, assessing OC symptoms using a total scale and four subscales: Impaired Mental Control, Contamination, Checking, Urges and Worries. In the original study (Sanavio, 1988) using an Italian normative sample of 828 subjects of varying ages, the scales evidenced good psychometric properties. In the present study, the Urges and Worries subscale was not included because of its overlap with the construct of worry (e.g., Burns, Keortge, Formea, & Sternberger, 1996). The Beck Anxiety Inventory (BAI; Beck, Epstein, Brown, & Steer, 1988) is a 21-item self-report inventory that measures the severity of anxiety. The Italian version of the BAI was administered to 654 undergraduates, 831 community controls and 64 anxious patients. The findings indicated good psychometric properties (Sica, Coradeschi, Ghisi, & Sanavio, 2006; Sica & Ghisi, 2007). The Beck Depression Inventory-II (BDI-II; Beck, Steer, & Brown, 1996). The BDI-II is a 21-item self-report scale that assesses the severity of affective, cognitive, motivational, vegetative, and psychomotor components of depression. The Italian version of

the BDI-II was administered to 733 undergraduates, 354 community controls and 135 depressed patients and showed excellent psychometric properties (Ghisi, Flebus, Montano, Sanavio, & Sica, 2006; Sica & Ghisi, 2007). The Penn State Worry Questionnaire (PSWQ; Meyer, Miller, Metzger, & Borkovec, 1990) is a 16-item inventory designed to assess trait worry. The internal consistency of the Italian version of the PSWQ proved to be good (Morani, Pricci, & Sanavio, 1999). 3. Data analysis To test the different internal structures of the OCI, a confirmatory factor analysis (CFA) was performed on the community sample. The CFA was performed using LISREL 8.54 (Jo¨reskog & So¨rbom, SSI INC, 2003). Since the observed variables were non-normally distributed, the matrix of polychoric correlations – instead of the covariance matrix – was analyzed with the weighted least squares method (Jo¨reskog & So¨rbom, SSI INC, 2003). The goodness of fit was evaluated by the following criteria recommended by Schermelleh-Engel, Moosbrugger, and Mu¨ller (2003): chi-square (x2) Satorra–Bentler, the ratio between chisquare and degrees of freedom (x2/d.f.) 3, root mean square error of approximation (RMSEA) 0.08; non-normed fit index (NNFI) 0.95, and comparative fit index (CFI) 0.95. Product–moment correlations were computed to examine intercorrelations and temporal stability of the OCI subscales as well as various associations among the OCI scales and other variables. To test for differences of correlations within a sample, Fisher’s r to z transformation was utilized. Multivariate analysis of variance (MANOVA) and one-way ANOVA were performed to compare the OCI scales scores by sex and across the three clinical groups (OCD, anxious patients and nonclinical controls). Lastly, to ascertain the discriminative power of the OCI, receiver operating characteristic curves (ROC; Metz, 1978; Swets, 1996) were computed. The ROC analysis uses the association between sensitivity and specificity to derive an area under the curve which indicates how well a measure distinguishes between case positive (i.e., OCD individuals) and case negative (i.e., anxious patients or controls) individuals in a given sample, irrespective of the base rate. A value of 0.50 of the area under the curve indicates chance level and 1.0 indicates a perfect diagnostic tool. 4. Results 4.1. Factors structure of the OCI Consistent with many suggestions we elected to assess only distress associated with obsessions and compulsions, since the two scales set (i.e., distress and frequency) yield redundant information (e.g., Foa et al., 2002; Wu & Watson, 2003).

C. Sica et al. / Journal of Anxiety Disorders 23 (2009) 204–211

207

Table 2 Goodness of fit statistics for different models of the OCI factor structure (distress ratings) in the Italian community sample.

OCI (7 factors and 42 items) OCI-R (6 factors and 18 items) OCI (5 factors and 36 items) OCI (unifactorial and 42 items)

RMSEA

NNFI

CFI

x2 Satorra–Bentler

x2/d.f.

0.047 0.035 0.045 0.062

0.94 0.99 0.99 0.90

0.95 0.99 0.99 0.91

1399.8 167.1 979.4 1869.2

1.77 1.44 1.68 2.28

(p < 0.001) (p < 0.001) (p < 0.001) (p < 0.001)

Note—N = 340. See text for description of models and fit statistics.

Table 3 Standardized maximum likelihood estimates for the six-factor model (18 items) Washing Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item

22 38 42 7 9 10 14 23 35 28 30 33 16 25 39 6 11 34

Checking

Ordering

Obsessing

Mental neutralizing

Hoarding

R2

0.81 0.85 0.78

0.44 0.57 0.53 0.49 0.75 0.83 0.62 0.60 0.73 0.60 0.76 0.78 0.26 0.57 0.79 0.65 0.72 0.61

0.66 0.76 0.73 0.70 0.87 0.91 0.79 0.78 0.85 0.77 0.87 0.88 0.51 0.75 0.89

A CFA was performed on the community sample to test four models: (1) the original rationally derived seven factor structure (42 items): washing, checking, doubting, ordering, obsessing, hoarding, and mental neutralizing; (2) the six-factor structure of the OCI-R (18 items): washing, checking, ordering, obsessing, hoarding, and mental neutralizing; (3) the five-factor model and 36 items proposed by Wu and Watson (2003): checking, obsessing, washing, ordering, hoarding; (4) for comparison purposes, a unifactor solution in which all 42 items loaded on the same factor. As showed in Table 2, figures for models with six factors and 18 items (OCI-R) met the best criteria for adequacy of fit. Table 3 presents the standardized coefficients (factor loadings) for the six-factor model. Analyses below were performed on subscales generated from such solution. 4.2. Intercorrelations among symptom subscales and total score The product–moment correlations among each of the six subscales of the OCI-R along with the correlation between each scale and the total OCI-R score were computed both for community and OCD clinical sample (Table 4). Intercorrelation values among the six scales were low both for normal (mean r = 0.31, i.e., 10% of common

variance) and clinical (mean r = 0.28, i.e., 8% of common variance) samples; in addition all of the subscales correlated moderately to highly with the total OCI-R score. The pattern of associations differed in the two samples, in that mental neutralizing correlated higher with the remaining scales in OCD compared to controls, whereas the reverse was true (i.e., higher correlations in control group) for the washing subscale. 4.3. Internal consistency, temporal stability and descriptive statistics Means and standard deviations for total and subscale scores are shown in Table 5, along with Cronbach’s alpha coefficients calculated for each of the six subscales and the total score. All alpha coefficients but two (washing and mental neutralizing) exceeded 0.7; moreover the alpha value for the total scale was high (0.85) indicating that items converged on a common construct. Alpha coefficients were excellent for all scales when computed on OCD sample. Pearson’s r was computed on community sample to assess test– retest reliability at one-month interval (Table 5). Overall reliabilities for total and subscales scores were excellent, especially in view of the relatively long time frame for retest.

Table 4 Intercorrelations among the six subscales and between each subscale and total score for both community and OCD clinical sample

Washing Checking Ordering Obsessing Mental neutralizing Hoarding

Washing

Checking

Ordering

Obsessing

Mental neutralizing

Hoarding

Total score

–

0.40 (0.20) –

0.46 (0.23) 0.35 (0.35) –

0.30 ( 0.05) 0.32 (0.35) 0.38 (0.28) –

0.31 0.20 0.22 0.20 –

0.30 0.45 0.35 0.30 0.17 –

0.66 0.70 0.74 0.68 0.42 0.68

Note—N = 340; in parenthesis values for OCD clinical sample are showed.

(0.02) (0.49) (0.55) (0.47)

(0.11) (0.37) (0.33) (0.14) (0.29)

(0.48) (0.71) (0.78) (0.56) (0.74) (0.55)

C. Sica et al. / Journal of Anxiety Disorders 23 (2009) 204–211

208

Table 5 Alpha, test–retest reliabilities, means and standard deviations for the OCI subscales and total score

Washing Checking Ordering Obsessing Mental neutralizing Hoarding Total score

Cronbach’s alpha values

One month test–retest values

Mean

S.D.

0.60 0.76 0.77 0.80 0.61 0.77 0.85

0.81 0.76 0.95 0.96 0.99 0.85 0.93

0.9 1.3 1.9 1.6 0.3 1.7 7.8

1.5 2.0 2.3 2.4 0.93 2.1 7.6

(0.94) (0.81) (0.89) (0.89) (0.87) (0.77) (0.87)

Note—N = 340; in parenthesis values for OCD clinical sample are showed (N = 52). Table 6 Convergent and divergent validity for the OCI subscales and total score OCI scales

PI – washing

PI – checking

PI – impaired mental control

PI – total

BAI

BDI-II

PSWQ

Washing Checking Ordering Obsessing Mental neutralizing Hoarding Total score

0.63 0.34 0.38 0.25 0.20 0.34 0.53

0.43 0.57 0.41 0.26 0.21 0.31 0.56

0.39 0.43 0.44 0.52 0.28 0.44 0.65

0.54 0.50 0.49 0.48 0.36 0.44 0.71

0.19 0.31 0.25 0.42 0.22 0.34 0.45

0.22 0.27 0.26 0.35 0.25 0.27 0.41

0.28 0.28 0.30 0.35 0.10 0.28 0.43

Note—N = 340; all p-values < 0.001. PI: Padua Inventory; BAI: Beck Anxiety Inventory; BDI-II: Beck Depression Inventory-II; PSWQ: Penn State Worry Questionnaire. Bold values indicate correlations with corresponding scales.

4.4. Association of the OCI-R scores with age, gender and education The OCI-R scores did not correlate significantly with age (mean Pearson’s r = 0.06). Education, however, was significantly and negatively associated with some OCI-R subscale scores (obsessing r = 0.13, p < 0.02; total score r = 0.13, p < 0.02; mental neutralizing r = 0.12, p < 0.03; washing r = 0.11, p < 0.04). These correlations are considered small (0.10–0.29) according to Cohen’s (1988) method. A multivariate analysis of variance (MANOVA) was performed to determine gender effects for each of the six subscales of the OCI-R. Such a test was utilized in order to protect against inflated type I error. The MANOVA results were significant (Pillai’s F(6,333) = 2.2, p < 0.05); therefore, a one-way ANOVA was performed for each scale. Total score (F(1,339) = 5.7, p < 0.02), washing (F(1,339) = 5.5, p < 0.02), checking (F(1,339) = 5.3, p < 0.03) and obsessing (F(1,339) = 4.3, p < 0.04) were higher for males compared to females. To further evaluate the magnitude of differences across gender eta squared values (h2) were computed. According to Cohen (1988), h2 = 0.1 corresponds to a small effect size, h2 = 0.6 to a medium effect and h2 = 1.4 to a large effect size. The h2 values were typically around 0.01 (none exceeded 0.03) suggesting that the magnitude of the differences by gender was rather low. 4.5. Concurrent validity Concurrent validity of the OCI-R was determined by Pearson correlations with another measure of obsessive compulsive

symptoms, the Padua Inventory (Table 6). Total OCI-R and PI scores were highly correlated (r = 0.71). Washing, checking, obsessing and total scores of the OCI-R correlated more strongly with the corresponding washing, checking, impaired mental control, and total scales of the PI, respectively, than with the other non-corresponding subscales, indicating adequate convergent validity (all z values >1.96, p < 0.05). Moreover, these correlations remained significant after controlling for anxiety (i.e., BAI scores) and depression (i.e., BDI-II scores): washing (partial r = 0.60), total score (partial r = 0.60), checking (partial r = 0.51), and obsessing (partial r = 0.35). The correlations between the OCI-R and BAI, BDI-II, PSWQ, were small to medium in size; moreover, the OCI-R total score showed significantly lower associations with BAI, BDI-II, and PSWQ than with all PI scales (all z values >1.96, p < 0.05), suggesting therefore adequate divergent validity (Table 6). 4.6. Criterion-oriented validity To ascertain criterion-related validity mean scores on the OCI-R were compared across the three groups (OCD, anxious patients and nonclinical controls). The multivariate analysis of variance (MANOVA) was significant (Pillai’s F(12,236) = 9.3, p < 0.001). Therefore, a one-way ANOVA was performed for each scale, using Student–Newman–Keuls (SNK) for post-hoc comparisons. As shown in Table 7, the OCD sample scored significantly higher than anxious patients and nonclinical controls on every subscale

Table 7 Group comparisons on the OCI subscales and total scores OCD (52)

Washing Checking Ordering Obsessing Mental neutralizing Hoarding Total score

4.4 (4.6) 3.6 (3.5) 4.3 (3.8) 8 (3.6) 2.7 (3.9) 1.9 (2.5) 25 (13.9)

AG (36)

0.5 (1.1) 1.6 (2.5) 1.4 (2.1) 3.4 (3.3) 0.2 (0.7) 0.9 (1.7) 8 (7.4)

CC (47)

1.1 1.5 2.1 1.4 0.3 1.8 8.3

(1.5) (2.1) (2.7) (1.5) (0.8) (2.3) (9.2)

Analysis of variance outcome

Significant SNK post-hoc comparison (p < 0.05)

h2 values (OCD vs. AG)

h2 values

F(2,132) = 21.5** F(2,132) = 8.2** F(2,132) = 10** F(2,132) = 52.2** F(2,132) = 15** F(2,132) = 2.2ns F(2,132) = 34.8**

OCD > AG, OCD > AG, OCD > AG, OCD > AG, OCD > AG, – OCD > AG,

0.21 0.10 0.13 0.28 0.14 – 0.35

0.18 0.12 0.08 0.51 0.14 – 0.34

CC CC CC CC CC CC

(OCD vs. CC)

Note—Standard error in brackets; **p < 0.001; NS: non-significant; SNK: Student–Newman–Keuls; OCD: Obsessive-compulsive disorder group; AG: Anxious group; CC: community controls.

C. Sica et al. / Journal of Anxiety Disorders 23 (2009) 204–211

except hoarding. Moreover, the OCI scales did not distinguish anxious patients from controls, with the exception of obsessing subscale. To evaluate the magnitude of the significant results reported in Table 7, eta squared values (h2) were also computed by comparing the three groups in pairs. Results suggested that the magnitude of the differences was generally low to medium, both when OCDs were compared to anxious patients and when OCDs were compared to nonclinical controls. Lastly, on the basis of the most prominent obsessions and compulsions recorded during the enrollment interview, two clinicians independently classified OCD individuals according the following current symptom presentation: checking symptoms (n = 32), impaired mental control symptoms (n = 26), washing symptoms (n = 18), and hoarding symptoms (n = 13) (OCD participants could be represented in more than one symptom group if they reported more than one type of symptom, as is common in OCD). To eliminate a possible source of bias, two independent raters were blind to the aims of the study and only patients for whom both raters agreed were included; further, OCIR results were not used when making this classification. The individuals included in each symptom subgroup were then compared to the remainder of the OCD sample on all the OCI-R subscales. Results showed that those who reported washing symptoms had higher scores only on washing subscale (F(1,49) = 12.9, p < 0.001); who reported checking symptoms had higher scores on checking (F(1,49) = 14.2, p < 0.001) and total score (F(1,49) = 5.6, p < 0.02) scales; who reported impaired mental control symptoms scored higher on obsessing (F(1,49) = 5.4, p < 0.02) and washing subscales (F(1,49) = 5.7, p < 0.02). Lastly those who reported hoarding symptoms had higher scores only on hoarding subscale (F(1,49) = 11, p < 0.002). 4.7. Discriminative power The ROC analysis on OCI-R total score resulted in a large nonparametrically computed areas under the curve: 0.89 when discriminating OCDs from anxious individuals, and 0.86 when discriminating OCDs from non-clinical controls. In addition all OCIR subscales except hoarding showed values higher than 0.70 when comparing OCD participants to either anxious or non-clinical controls. Overall, these results indicate excellent discriminative power of the OCI-R and its subscales. 5. Discussion and conclusion The OCI was specifically designed to address many of the concerns levied against existing instruments, such as narrowness of symptom subtypes addressed and unsatisfactory discriminant validity. As such, it seemed important to establish which version of the questionnaire best measures OCD symptoms. With respect to internal structure of the questionnaire, a CFA performed on the community sample found evidence that the OCI consists of six subscales assessing washing, checking, ordering, obsessing, mental neutralizing, and hoarding symptoms, when applied to Italian nonclinical individuals. Such result supports the use of the subscales of the OCI-R over the OCI; more important, our study further confirmed that the structure of OC symptoms remains invariant across different cultures, since the American, French, German, Icelandick, and Spanish versions of the OCI all replicated the goodness of fit to the six factor model. Invariance of OC symptoms structure across cultures has interesting theoretical implications. In fact, it means that culture may influence the prevalence of OCD subtypes (e.g., washing, ordering, etc.) but not the essence of the disorder (see also, Sica, Novara, Sanavio,

209

Coradeschi, & Dorz, 2002; Sica, Taylor, Arrindell, & Sanavio, 2006; Wu & Shawn, 2008). As a consequence, if basic OCD manifestations are the same across cultures, the hypothesis of a biological substrate of OCD would be somewhat strengthened (for example, Cottraux & Gerard, 1998; Insel, 1992). Coherence of the OCI and its subscales is suggested by the good internal consistency and excellent temporal stability in community sample and by the excellent internal consistency in OCD sample. The two sole exceptions were washing and neutralizing subscale when computed on community sample. Modest reliability values of these two scales were reported by other studies where the OCI-R was applied to samples without OCD (Foa et al., 2002; Fullana et al., 2005; Zermatten et al., 2006). A desirability effect may be relevant in case of washing scale. In fact, whereas one item of this scale deals with a common concern (‘‘I find it difficult to touch an object when I know it has been touched by strangers’’), the other two items deal with aspects which can appear ‘‘odd’’ or bizarre (‘‘I sometimes have to wash or clean myself simply because I feel contaminated’’; ‘‘I wash my hands more often and longer than necessary’’). Our data support such interpretation: on average, the first item obtained a score (M = 0.53; S.D. = 0.88) twice as much than the two remaining ones (M = 0.24; S.D. = 0.59; M = 0.23; S.D. = 0.58, respectively). The problem with mental neutralizing is somewhat different. Because neutralizing is rare in non-OCD individuals, the diminished alpha coefficients for this subscale in our community sample may reflect a restricted range rather than structural inadequacy. As a matter of fact, range in our community sample for this subscale was 0–8, whereas for OCD was 0–12. Moreover, on average, all the three items obtained very low rating when administered to community individuals (M = 0.1; S.D. = 0.38). At the same time, it is important to note that some authors suggested a revision of mental neutralizing scale because the content of the items – all dealing with numbers and counting-complaints – may fail to capture many clinically relevant neutralization phenomena (Gonner et al., 2008; see Purdon & Clark, 2002, for a review of neutralization phenomena). In any case, excellent reliability of all the OCI-R scales in our OCD sample suggests that such features are best measured on clinical individuals compared to nonclinical ones. Results from intercorrelations confirmed that, consistent with its theoretical foundations, the OCI-R measures fairly separate dimensions both in normal and in clinical individuals, as demonstrated in other studies (e.g., Abramowitz & Deacon, 2006; Foa et al., 2002; Huppert et al., 2007). Overall, the total score and subscale scores show a pattern of specific association with relevant scales from another OC symptom-related measure. Regarding divergent validity, although the OCI is positively correlated with measures of depression and anxiety, these correlations appeared weaker than those for other measures of OCD symptoms. In addition, the OCI-R total score shows significant lower associations with BAI, BDI-II, and PSWQ than with PI scales. Finally, the correlations between the OCI and the Padua Inventory remain significant even when the effect of depressive and anxious symptomatology was controlled. In sum, although there is some shared variability, the OCI-R appears to measure OCD symptoms independently of depression and anxiety, indicating satisfactory divergent validity. The questionnaire was insensitive to age, whereas sex differences were found for some OCI subscale scores. However, evaluation of the magnitude of the effect sizes suggested that such differences had little clinical significance. Overall, these findings are consistent with our initial hypotheses and provide evidence that demographic characteristics seem not to affect the scores of the Italian version of the OCI. On the other hand, to our knowledge

210

C. Sica et al. / Journal of Anxiety Disorders 23 (2009) 204–211

this is the first study to find an effect of education on self-reported obsessions and compulsions. More educated people reported less severe OCD symptoms: it may be that individuals with higher education are more sensitive to social desirability, since many obsessional characteristics seem ‘‘odd’’ or strange. Alternatively, less educated people may be generally more superstitious than higher ones, a feature that in Italian population is well correlated with subclinical OCD manifestations (Sica, Novara, & Sanavio, 2002). Another explanation lies in the cognitive theory of obsessions: according to this model, normal intrusions develop into obsessions when intrusions are appraised as posing a threat for the individual (e.g., Salkovskis, 1985). It may be that different levels of education, by shaping less or more flexible beliefs and attitudes, affect the internal experience of obsessions and related compulsions. Of course, all these hypotheses need further study. At the present moment we conclude that OCI-R scores should be corrected for education when the questionnaire is administered to Italian individuals. The OCI-R clearly differentiated OCD patients from non-OCD anxious patients and nonclinical controls. Only the hoarding scale did not discriminate OCD patients from anxious or nonclinical controls. These findings are consistent with recent research that raises questions about whether hoarding symptoms represent a ‘‘pure’’ dimension of OCD or a separate type of mental health problem (Grisham, Brown, Liverant, & Campbell-Sills, 2005; Wu & Watson, 2003). For instance, Grisham et al. (2005) found evidence that in absence of other obsessions and compulsions, hoarding is likely distinct from OCD. Moreover, our sample contained a relatively small number of patients with hoarding symptoms, a feature that may have prevented testing the real discriminative power of the hoarding scale. Actually, when OCD patients with hoarding symptoms were compared to OCD individuals without such symptoms, the OCI-R hoarding scale did clearly differentiate between these two groups. Lastly, a recent study by Fullana et al. (2005) reported a strong convergence of the OCI-R hoarding subscale and a well-validated measure of hoarding symptoms. Therefore we cannot exclude that the OCI-R hoarding scale may be useful in detecting such invalidating subset of symptoms. For this reason, before considering whether to drop/revise the hoarding subscale, more studies are warranted. Two other sets of results confirmed the criterion-oriented validity of the OCI-R. First, patients reporting a particular subtype of OCD symptoms during the enrollment interview (e.g., washing, checking, etc.) obtained higher scores on the corresponding OCI subscale than did patients who did not mention such symptoms. Second, the ROC analysis showed that the OCI had a very good discriminant power. Some limits of the present study need to be noted. In order to evaluate different factor solutions we extracted a different subset of items from the long version of the OCI (i.e., 42 items). Such choice enabled us to compare different factor solutions and helped in establishing the use of the OCI-R over the OCI; nonetheless we cannot say whether a stand-alone version of the OCI-R would have obtained the same results in terms of psychometric properties. However, in a comparison between the carved-out and stand-alone versions of the OCI-R, no differences were found in factor structure or in other basic psychometric features (Foa et al., 2002). Results on community members were based on a sample that was relatively restricted in educational level and socio-economic status. There is no doubt that our results need to be replicated on normal samples with broader demographic characteristics. Likewise, in interpreting the outcome of the confirmatory factor analyses it is important to note that our sample might differ qualitatively from a clinical sample of OCD patients.

Lastly, further research is needed to address sensitivity of the specific subscales to treatment effects when applied to Italian individuals. In conclusion, the present study replicated and extended previous findings with the OCI-R and supported its use in Italy. Considering the amount of data now available, it is possible to conclude that the OCI-R is a very strong and useful measure of OCD symptoms in different cultural contexts. Acknowledgement The authors are grateful to Gail Steketee for advice on revision of the manuscript. References Abramowitz, J. S., & Deacon, B. J. (2006). Psychometric properties and construct validity of the Obsessive-Compulsive Inventory-Revised: replication and extension with a clinical sample. Journal of Anxiety Disorders, 20, 1016–1035. Baer, L. (1994). Factor analysis of symptom subtypes of obsessive compulsive disorder and their relation to personality and tic disorders. Journal of Clinical Psychiatry, 55, 18–23. Beck, A. T., Epstein, N., Brown, G., & Steer, R. A. (1988). An inventory for measuring clinical anxiety: psychometric properties. Journal of Consulting and Clinical Psychology, 56, 893–897. Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Beck Depression Inventory. Second Edition Manual. San Antonio, TX: The Psychological Corporation Harcourt Brace & Company. Brislin, R. W. (1986). The wording and translation of research instruments. In: W. J. Lonner & J. W. Berry (Eds.), Field methods in cross-cultural research. Beverly Hills, CA: Sage. Burns, G. L., Formea, G. M., Keortge, S., & Sternberger, L. G. (1995). The utilization of non-patient samples in study of obsessive-compulsive disorder. Behavior Research and Therapy, 33, 133–144. Burns, G. L., Keortge, S., Formea, G. M., & Sternberger, L. G. (1996). Revision of the Padua Inventory of obsessive-compulsive disorder symptoms: distinctions between worry, obsessions, and compulsions. Behavior Research and Therapy, 34, 163–173. Clark, D. A., Antony, M. M., Beck, A. T., Swinson, R. P., & Steer, R. A. (2005). Screening for obsessive and compulsive symptoms: validation of the Clark-Beck obsessivecompulsive inventory. Psychological Assessment, 17, 132–143. Cohen, J. (1988). Statistical power analyses for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Cottraux, J., & Gerard, D. (1998). Neuroimaging and neuroanatomical issues in obsessive-compulsive disorder. In: R. P. Swinson, M. M. Antony, S. Rachman, & M. A. Richter (Eds.), Obsessive-compulsive disorder: theory, research, and treatment (pp. 154–180). New York, NY: Guilford. First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. W. (1996). Structured Clinical Interview for DSM-IV-Patient Edition (SCID-I/P). New York: Biometrics Research Department, New York State Psychiatric Institute. Foa, E. B., Huppert, J. D., Leiberg, S., Langner, R., Kichic, R., Hajcak, G., et al. (2002). The Obsessive-Compulsive Inventory: development and validation of a short version. Psychological Assessment, 14, 485–496. Foa, E. B., Kozak, M. J., Salkovskis, P. M., Coles, M. E., & Amir, N. (1998). The validation of a new obsessive-compulsive disorder scale: the Obsessive-Compulsive Inventory. Psychological Assessment, 10, 206–214. Fullana, M. A., Tortella-Feliu, M., Caseras, X., Andion, O., Torrubia, R., & Mataix-Cols, D. (2005). Psychometric properties of the Spanish version of the Obsessive-Compulsive Inventory-Revised in a non-clinical sample. Journal of Anxiety Disorders, 19, 893–903. Ghisi, M., Flebus, G. B., Montano, A., Sanavio, E., & Sica, C. (2006). Beck Depression Inventory-II. BDI-II. Manuale. Firenze: O.S. Organizzazioni Speciali. Gonner, S., Leonhart, R., & Ecker, W. (2008). The obsessive-compulsive inventoryrevised (OCI-R): validation of the German version in a sample of patients with OCD, anxiety disorders, and depressive disorders. Journal of Anxiety Disorders, 22, 734– 749. Goodman, W. K., Price, L. H., Rasmussen, S. A., Mazure, C., Fleischmann, R. L., Hill, C. L., et al. (1989). The Yale-Brown Obsessive Compulsive Scale. I. Development and reliability. Archives of General Psychiatry, 46, 1011–1066. Grisham, J., Brown, T., Liverant, G., & Campbell-Sills, L. (2005). The distinctiveness of compulsive hoarding from obsessive-compulsive disorder. Journal of Anxiety Disorders, 19, 767–779. Huppert, J. D., Walther, M. R., Hajcak, G., Yadin, E., Foa, E. B., Simpson, H. B., et al. (2007). The OCI-R: validation of the subscales in a clinical sample. Journal of Anxiety Disorders, 21, 394–406. Insel, T. R. (1992). Toward a neuroanatomy of obsessive-compulsive disorder. Archives of General Psychiatry, 49, 739–744. Jo¨reskog, K., & So¨rbom, D. (2003). LISREL 8 (Version 8.54). Chicago: Scientific Software International.

C. Sica et al. / Journal of Anxiety Disorders 23 (2009) 204–211 Kyrios, M., Bhar, S., & Wade, D. (1996). The assessment of obsessive-compulsive phenomena: psychometric and normative data on the Padua Inventory from an Australian non clinical student sample. Behaviour Research and Therapy, 34, 85–95. Macdonald, A. M., & Silva, P. (1999). The assessment of obsessionality using the Padua Inventory: its validity in a British non clinical sample. Personality and Individual Differences, 27, 1027–1046. Mataix-Cols, D., Rauch, S. L., Baer, L., Eisen, J. L., Shera, D. M., Goodman, W. K., et al. (2002). Symptom stability in adult obsessive-compulsive disorder: data from a naturalistic two-year follow-up study. American Journal of Psychiatry, 159, 263–268. McKay, D., Abramowitz, J. S., Calamari, J. E., Kyrios, M., Radomsky, A., Sookman, D., et al. (2004). A critical evaluation of obsessive-compulsive disorder subtypes: symptoms versus mechanisms. Clinical Psychology Review, 24, 283–313. Metz, C. E. (1978). Basic principles of ROC analysis. Seminars in Nuclear Medicine, 8, 283–298. Meyer, T. J., Miller, M. L., Metzger, R. L., & Borkovec, T. D. (1990). Development and validation of the Penn State Worry Questionnaire. Behaviour Research and Therapy, 28, 487–495. Minichiello, W. E., Baer, L., Jenike, M. A., & Holland, A. (1990). Age of onset of major subtypes of obsessive-compulsive disorder. Journal of Anxiety Disorder, 4, 147–150. Morani, S., Pricci, D., & Sanavio, E. (1999). Penn State Worry Questionnaire e Worry Domains Questionnaire. Presentazione delle versioni italiane ed analisi della fedelta`. Psicoterapia Cognitiva e Comportamentale, 5, 195–209. Purdon, C., & Clark, D. A. (2002). The need to control thoughts. In: R. O. Frost & G. Steketee (Eds.), Cognitive approaches to obsessions and compulsions: theory, assessment, and treatment (pp. 371–384). Oxford: Elsevier Press. Salkovskis, P. M. (1985). Obsessional-compulsive problems: a cognitive-behavioral analysis. Behaviour Research and Therapy, 23, 571–583. Salkovskis, P. M., & Harrison, J. (1984). Abnormal and normal obsessions—a replication. Behaviour Research and Therapy, 22, 549–552. Sanavio, E. (1988). Obsessions and compulsions: the Padua Inventory. Behaviour Research and Therapy, 26, 169–177. Schermelleh-Engel, K., Moosbrugger, H., & Mu¨ller, H. (2003). Evaluating the fit of structural equation models: tests of significance and descriptive goodness-of-fit measures. Methods of Psychological Research Online, 8(2), 23–74. Sher, K. J., Frost, R. O., Kushner, M., Crews, T., & Alexander, J. (1989). Memory deficits in sub-clinical checkers: replication and extension in a clinical sample. Behaviour Research and Therapy, 27, 65–69. Sica, C., Coradeschi, D., Ghisi, M., & Sanavio, E. (2006). Beck Anxiety Inventory – BAI. Manuale. Firenze: O.S. Organizzazioni Speciali. Sica, C., Coradeschi, D., Sanavio, E., Dorz, S., Manchisi, D., & Novara, C. (2004). A study of the psychometric properties of the Obsessive Beliefs Inventory and Interpretations

211

of Intrusions Inventory on clinical Italian individuals. Journal of Anxiety Disorders, 18, 291–307. Sica, C., & Ghisi, M. (2007). The Italian Versions of the Beck Anxiety Inventory and the Beck Depression Inventory-II: psychometric properties and discriminant power. In: M. A. Lange (Ed.), Leading-edge psychological tests and testing (pp. 27–50). Nova Science Publishers. Sica, C., Novara, C., & Sanavio, E. (2002a). Culture and psychopathology: superstition and obsessive-compulsive cognitions and symptoms in a non-clinical Italian sample. Personality and Individual Differences, 32, 1001–1012. Sica, C., Novara, C., Sanavio, E., Coradeschi, D., & Dorz, S. (2002b). OCD cognitions across cultures. In: R. O. Frost & G. Steketee (Eds.), Cognitive approaches to obsessions and compulsions: theory, assessment, and treatment (pp. 371–384). Oxford: Elsevier Press. Sica, C., Taylor, S., Arrindell, W. A., & Sanavio, E. (2006b). A cross-cultural test of the cognitive theory of obsessions and compulsions: a comparison of Greek, Italian, and American individuals. A preliminary study. Cognitive Therapy and Research, 30, 585–597. Simonds, L. M., Thorpe, S. J., & Elliott, S. A. (2000). The Obsessive Compulsive Inventory: psychometric properties in a nonclinical student sample. Behavioural and Cognitive Psychotherapy, 28, 153–159. Simpson, H. B., Rosen, W., Huppert, J. D., Lin, S. H., Foa, E. B., & Liebowitz, M. R. (2006). Are there reliable neuropsychological deficits in OCD? Journal of Psychiatric Research, 40, 247–257. Smari, J., Olason, D. T., Eythorsdottir, A., & Frolunde, M. B. (2007). Psychometric properties of the Obsessive Compulsive Inventory-Revised among Icelandic college students. Scandinavian Journal of Psychology, 48, 127–133. Sterneberger, L., & Burns, G. (1990). Obsessive compulsive disorder: symptoms and diagnosis in a college sample. Behavior Therapy, 22, 569–576. Swets, J. A. (1996). Signal detection theory and ROC analysis in psychology and diagnostics: collected papers. Hillsdale, NJ, England: Lawrence Erlbaum Associates. Tolin, D. F., Woods, C. M., & Abramowitz, J. S. (2003). Relationships between obsessive beliefs and obsessive-compulsive symptoms. Cognitive Therapy and Research, 27, 657–669. Wu, K. D., & Shawn, A. C. (2008). Further investigation of the Obsessive beliefs Questionnaire: factor structure and specificity of relations with OCD symptoms. Journal of Anxiety Disorders, 22, 824–836. Wu, K. D., & Watson, D. (2003). Further investigation of the obsessive-compulsive inventory: psychometric analysis in two non-clinical samples. Journal of Anxiety Disorders, 17, 305–319. Zermatten, A., Van der Linden, M., Jermann, F., & Ceschi, G. (2006). Validation of a French version of the Obsessive-Compulsive Inventory-Revised in a non-clinical sample. European Review of Applied Psychology, 56, 151–155.