Original article
Validation of a French version of the Obsessive–Compulsive Inventory-Revised in a non-clinical sampleAdaptation et validation française de l'Inventaire de compulsions et d'obsessions-révisé auprès d'un échantillon non clinique

https://doi.org/10.1016/j.erap.2005.07.003Get rights and content

Abstract

Foa et al. (2002) presented a new instrument, the Obsessive–Compulsive Inventory-Revised (OCI-R), designed to evaluate the severity of obsessive–compulsive symptoms in both clinical and non-clinical individuals. The present study investigates the psychometric properties of a French version of this scale. The OCI-R French version was completed by 583 undergraduate students. The results revealed satisfactory internal consistency as measured by Cronbach's alpha coefficients (ranging from 0.63 to 0.86). In addition, the six-factor structure found by Foa et al. (2002) was confirmed in our sample by a confirmatory factor analysis. In brief, the French version of the OCI-R seems satisfactory for measuring Obsessive–Compulsive Disorders (OCD) symptoms in non-clinical samples. Future research is, however, needed to confirm these data in a sample of OCD patients.

Résumé

Foa et al. (2002) ont proposé un nouvel instrument, l'Inventaire de compulsions et d'obsessions-révisé (ICO-R), destiné à évaluer la sévérité des symptômes obsessionnels-compulsifs à la fois dans une population clinique et non clinique. Notre étude avait pour but d'investiguer les propriétés psychométriques de la version française de cette échelle. Cinq cent quatre-vingt-trois étudiants ont ainsi rempli la version française de l'ICO-R. Les résultats montrent une consistance interne satisfaisante, avec des alphas de Cronbach s'échelonnant de 0,63 à 0,86. Par ailleurs, la structure factorielle à six facteurs, trouvée par Foa et al. (2002), a été confirmée dans notre échantillon à l'aide d'une analyse confirmatoire. En résumé, la version française de l'ICO-R semble être un outil satisfaisant pour évaluer les symptômes obsessionnels–compulsifs dans une population tout-venant. Ces données devront toutefois être confirmées par des travaux ultérieurs portant sur des échantillons cliniques.

Introduction

Several self-reporting questionnaires evaluating the severity of Obsessive–Compulsive Disorders (OCD) have been developed, such as the Maudsley Obsessive–Compulsive Inventory (MOCI, Hodgson and Rachman, 1977), the Compulsive Activity Checklist (CAC, Foa et al., 1984), the Leyton Obsessional Inventory (LOI, Cooper, 1970) and the Padua Inventory (PI, Sanavio, 1988). Although these questionnaires are commonly used to evaluate OCD, most of them were not designed to evaluate both patients and non-clinical individuals; in addition, they were constructed to capture only a subset of obsessions and compulsions.

Recently, a new instrument, the Obsessive–Compulsive Inventory (OCI, Foa et al., 1998) was developed to overcome these limitations. This new scale was validated not only with OCD patients, but also with non-psychiatric controls. Thus, the scale is intended to be administered to both clinical and non-clinical individuals. Moreover, this inventory was specifically developed to assess the heterogeneous symptoms of OCD. From this perspective, the authors chose the main symptoms of OCD, as described in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV, American Psychiatric Association, 1994), and constructed seven subscales: ‘Washing’ (eight items), ‘Checking’ (nine items), ‘Obsessing’ (eight items), ‘Mental Neutralising’ (six items), ‘Ordering’ (five items), ‘Hoarding’ (three items) and ‘Doubting’ (three items). In order to rate these 42 items, two 5-point Likert scales were constructed, one measuring the ‘frequency of symptoms’ and the other measuring the ‘associated distress’. Foa et al. (1998) reported high internal consistency for the full scale and for the subscales (range 0.59–0.96 for different populations: OCD, post-traumatic stress disorder, generalised social phobia and controls), good test–retest reliability for controls (r range = 0.68–0.90) and for OCD patients (r range = 0.77–0.97), excellent discriminant validity and satisfactory convergent validity.

The psychometric properties of the OCI have also been investigated in non-clinical samples. Thus, Simonds et al. (2000) found high internal consistency across the seven subscales, good test–retest reliability and good convergent validity in a non-clinical student sample. The seven-factor structure was, however, not confirmed with two non-clinical samples in Wu and Watson's (2003) study, who found five subscales (‘Checking’, ‘Obsessing’, ‘Washing’, ‘Ordering’ and ‘Hoarding’) rather than seven. These authors therefore proposed a revised scale with fewer items. However, the confirmatory analysis done on these five revised subscales failed to reach the level of indices necessary to indicate a good fit for the data.

Recently, Foa et al. (2002) tried to facilitate the use of the OCI by proposing some improvements. First, in light of the high correlations (above 0.90) between the ‘frequency’ and ‘distress’ scores, they decided to retain only one of these two scales. The distress scale was selected for the new version of the OCI because of its larger between-group effect size. Second, they decided to shorten the scale by reducing the number of items per subscale. They selected an equal number of items (N = 3) for each subscale on the basis of a principal-component analysis. This analysis revealed seven interpretable factors: ‘Washing’, ‘Checking/Doubting’, ‘Obsessing’, ‘Mental Neutralising’, ‘Ordering’, ‘Hoarding’ and ‘Harming’. Items that loaded on only one factor and that presented the highest factor loadings or the highest between-group effect size were selected. A principal-component analysis was performed on these selected items and revealed six rather than seven factors; consequently, the Harming subscale was eliminated from the Obsessive–Compulsive Inventory-Revised (OCI-R). This new version was found to present a stable factor structure, high internal consistency for the full scale (range 0.81–0.93) and for the subscales (range 0.65–0.90 except alpha for Mental Neutralising in controls, which reached 0.34), good to excellent test–retest reliability (r range 0.57–0.91), good discriminant validity and satisfactory convergent validity. ‘Receiver Operating Characteristic’ (ROC) analyses, conducted to measure the diagnostic power of the OCI-R, revealed that it was able to discriminate between OCD patients and patients suffering from another anxiety disorder, as well as between OCD patients and non-clinical individuals. Recently, the psychometric properties of the OCI-R were also investigated in a non-clinical college sample (Hajcak et al., 2004). This study indicates adequate test–retest reliability, solid six-factor structure, high internal consistency, and good convergent and divergent validity. In summary, the OCI and the OCI-R were shown to have good psychometric properties with both clinical and non-clinical samples.

Our study was designed to validate a French version of the OCI-R in a non-clinical sample. More specifically, our goal was to examine the internal consistency and the structure of the OCI-R in a sample of university students. Several studies have shown that non-clinical obsessions and compulsions are similar in content to clinical OCD, although they are less frequent and less intense (Freeston et al., 1991; Muris et al., 1997). From this perspective, the use of non-clinical samples in the domain of OCD has been shown to be particularly useful, allowing one to examine the different dimensions and constructs in very large samples (Wu and Watson, 2003). Thus, the purpose of this study was to validate a French translation of the OCI-R with a large non-clinical sample in order to allow French-speaking researchers to assess OCD symptoms with a reliable and accurate instrument.

Section snippets

Participants

The participants were 583 undergraduate student volunteers (301 females, 282 males), enrolled at the University of Geneva. Their mean age was 24.86 (SD = 3.85) and their mean number of years of education was 15.41 (SD = 2.25).

Measures

Participants were asked to complete the OCI-R. This self-reporting questionnaire consists of 18 items evaluating OCD symptoms. The revised version is composed of six subscales, each containing three items: ‘Washing’ (5, 11, 17), ‘Obsessing’ (6, 12, 18), ‘Hoarding’ (1, 7, 13),

Results

Mean scores for the six subscales and total scores for the present study and the original validation study are presented in Table 1. In general, the mean total score and subscores from our sample are lower than those reported by Foa et al. (2002).

In order to evaluate the reliability of the French version of the OCI-R, internal consistency was assessed with Cronbach's alpha coefficients (Cronbach, 1956). Coefficients for the OCI-R total scale (0.86), as well as the Checking subscale (0.83),

Discussion and conclusion

The aim of this study was to validate a French version of the OCI-R developed by Foa et al. (2002) in a non-clinical sample. The investigation of the psychometric properties of the French version revealed acceptable to excellent internal consistency, with Cronbach's alpha coefficients ranging from 0.63 to 0.86. Furthermore, the inter-item correlations indicated medium to large effects (Cohen, 1988) but were not high enough to postulate redundancy within the subscales. The alpha coefficients

Acknowledgements

We would like to thank Edna B. Foa for giving us the permission to translate the OCI-R. This study was supported by the Swiss National Science Foundation, grant no. 1114-067135.01/01.

References (17)

There are more references available in the full text version of this article.

Cited by (0)

View full text