Evaluation of the factor structure of the Child Obsessive–Compulsive Impact Scale – Revised (COIS-R) in Scandinavia with confirmatory factor analysis
Introduction
Obsessive–compulsive disorder (OCD) in children and adolescents is a serious and chronic disorder (Rasmussen and Eisen, 1992, Ruscio et al., 2010, Stewart et al., 2004) with a population prevalence rate between 1% and 3% (Flament et al., 1988, Rapoport et al., 2000, Ruscio et al., 2010, Valleni-Basile et al., 1994). The functional impairment that OCD causes is reflected in the DSM diagnostic criteria (American Psychiatric Association, 2013). Symptoms must interfere with normal routines, occupational and academic functioning, or normal social activities or relationships in order to establish a diagnosis of OCD and differentiate it from subclinical OCD (Angold, Costello, Farmer, Burns, & Erkanli, 1999). Reasons for the importance of valid measure with normative data for OCD-related functional impairment (Piacentini, Peris, Bergman, Chang, & Jaffer, 2007) include the possibility to place such impairment within a framework that is both developmentally sensitive and takes into account potential domains and their degree of impairment in comparison to normative development among children and adolescents. For instance, children may experience temporal obsessions and compulsions that are similar to the symptoms experienced by clinical groups (Evans et al., 1997, Zohar and Bruno, 1997). Normative data on functional impairment would be important in these cases in order to interpret the symptom impairment accurately (Hollander et al., 1998, Koran et al., 1996, Riddle et al., 1990, Swedo et al., 1989, Thomsen, 1994). In addition, OCD is diverse in terms of symptom cluster and severity. These differences might facilitate the treatment as the exposure would focus specifically on the domains with the most interference (e.g., March and Mulle, 1998, Piacentini et al., 2007). Last, a standardized measure of functional impairment may be used in addition to symptom severity measures as a means of evaluating treatment effectiveness (e.g., Piacentini et al., 2011, Skarphedinsson et al., 2015, Storch et al., 2013).
Research on OCD-related functional impairment has lagged behind its importance in terms of assessment and diagnosis. Few studies have focused on OCD-related functional impairment, with most studies beginning in adult OCD (Bobes et al., 2001, Calvocoressi et al., 1995, Cooper, 1996, Hollander et al., 1998, Huppert et al., 2009, Jacoby et al., 2014, Koran et al., 1996, Sorensen et al., 2004, Vorstenbosch et al., 2012). All of these studies show that OCD is associated with severe functional impairment across many areas, such as family, social, academic and vocational functioning. Functional impairment seems to be treatment sensitive as OCD patients in remission are less functionally impaired than patients that are not in remission (Huppert et al., 2009). In addition, functional impairment is more severe in OCD patients with comorbid disorders (Huppert et al., 2009) and especially depressive symptoms (Jacoby et al., 2014).
Fewer studies have been conducted among children and adolescents with OCD (Piacentini et al., 2003, Piacentini et al., 2007, Storch et al., 2006, Valderhaug and Ivarsson, 2005). They [with one exception (Storch, Ledley, et al., 2006)] used the same methodology to show that children and adolescents with OCD suffer considerably in terms of OCD-related impairments in social-, school-, home-, and family-related functioning. These measures have shown to be positively correlated with OCD symptom severity, anxiety and depressive symptom measures (Nadeau et al., 2013, Piacentini et al., 2003, Piacentini et al., 2007, Storch et al., 2010, Valderhaug and Ivarsson, 2005) and global functional impairment (Piacentini et al., 2003, Piacentini et al., 2007, Valderhaug and Ivarsson, 2005).
The Child Obsessive Compulsive Impact Scale – Revised (COIS-R) is a measure that includes 33 items available as parent- and child-reported versions. The scale is derived from the original COIS (Piacentini et al., 2003, Piacentini and Jaffer, 1999) with 52 specific items consisting of three a priori factors (academic/school, home/family, and social functioning) based on clinical interpretation of how well items matched each domain of impairment. The revised and shortened version was created by conducting an exploratory factor analysis with the 52 items. The authors conducted exploratory factor analysis to evaluate whether there were items that would load across multiple domains and they retained all items with factor loadings greater than 0.40. This resulted in a four-factor solution for the parent-reported version (daily living skills, family, social and school) and a three-factor solution for the child-reported version (school, social, and activities). However, although both versions have an equal number of items (33), they only have 23 items in common. Among other results from the same study, all the factors and the total score were associated with OCD symptom severity and internalizing and externalizing symptom scores on the Child Behavior Checklist, although less so between the Child Behavior Checklist (CBCL) externalizing symptoms and COIS. In addition, the COIS was positively associated with a global impairment measure (Children׳s Global Assessment Scale) (Piacentini, Peris, et al., 2007).
More recent studies have indicated (using the original COIS) that the parent- and child-reports only show moderate agreement (Storch et al., 2010, Storch et al., 2014) and that they show different pattern of correlation with other measures. So although both versions correlate with OCD severity (Storch et al., 2010) depressive symptoms correlate with child-report only and family accommodation correlates with parent-rated impact in all three settings of functioning that were examined (home, school, and social) while it only correlated with the school setting in the child-report (Storch et al., 2010). This is in line with previous studies suggesting that children with parents that are engaged in accommodation will not experience decreased OCD-related functioning and therefore do not see their symptoms as impairing (Storch et al., 2007, Storch et al., 2007). In addition both versions correlate with OCD symptom insight (Storch et al., 2010). Insight has also shown to moderate the agreement between parents and children in impairment at home where children with less insight agreed much worse with their parents than children with greater insight. (Storch et al., 2014).
The COIS-R has also been used in numerous treatment outcome studies of pediatric OCD (Lenhard et al., 2014, Piacentini et al., 2011, Skarphedinsson et al., 2015, Storch et al., 2013, Storch et al., 2011) and the total score has shown to predict outcome (Garcia et al., 2010, Torp et al., 2015a). Despite this wide use, only one study on the factor structure of the scale exists. There is a need for confirmatory factor analysis like the authors state in their discussion (Piacentini, Peris, et al., 2007). Moreover, it is important to address potential cultural differences and to confirm whether the domains of OCD-related functional impairment are stable across cultures (Piacentini, Peris, et al., 2007). It is also important to obtain normative data in different cultures as symptom severity rated by parents or children might differ. Several studies have shown that Scandinavian parents and their children systematically endorse fewer child and adolescent emotional and behavioral symptoms than their counterparts in the United States or the United Kingdom (Achenbach et al., 2008, Heiervang et al., 2008) although the number of internalizing disorders diagnosed is the same (Heiervang et al., 2008).
The aim of this study is to evaluate the originally proposed factor structure of the COIS-R using confirmatory factor analysis in a sample of Scandinavian OCD patients. In addition, we will evaluate the factor structure in different subsamples (countries, gender, and age). In line with previous studies showing more extensive correlation between parent-report COIS and accommodation, we also believe that the factor structure of the COIS may be different among those two groups (parents that accommodate more compared to parents that accommodate less). In addition, we expect that insight may affect the factor structure of the COIS. Thus, we evaluate the structure in these two subgroups (greater and less accommodation and insight respectively). Furthermore, our aim was to provide normative data among treatment-seeking children and adolescents with OCD. This is the first confirmatory factor analysis study of the COIS-R. Thus, no hypothesis was developed concerning the quality of the factor structure. However, in line with previous research we hypothesized that the Scandinavian youth would score significantly lower than American youth on the total score and all subscales.
Section snippets
Participants
The sample consisted of 396 children and adolescents with primary OCD according to the Diagnostic and Statistical Manual of Mental Disorders (4th edition) (American Psychiatric Association, 2000). The participants were referred to different departments of child and adolescent psychiatry within Norway, Sweden and Denmark as a part of three OCD treatment studies (Melin et al., 2012, Torp et al., 2015b, Valderhaug et al., 2007). The sample consisted of 194 boys (48.9%) and 202 girls (50.9%).
Results
The results of the factor analysis for the parent-reported version are shown in Table 1. The initial analysis (Model 1) revealed acceptable results for the CFI, TLI, and RMSEA. The results suggested potential modifications for improvement. In all cases, this was due to a strong correlation between items that should be accounted for in the modified model. The modifications with the highest modifications indices were (above 80.0) 1) Correlation between items No. 11 (Going on a date) and No. 22
Discussion
The current study is the first report on the COIS-R using confirmatory factor analysis and the first study that evaluates the factor structure of the COIS-R in a non-U.S. OCD sample. Overall, the results indicate that the factor structure suggested by Piacentini et al. (2007) does not indicate an optimal fit, especially for the child-report. Several modifications were suggested, and both parent- and child-reported versions improved by taking the modifications in consideration. We will discuss
Conclusion
Overall, our results indicate that the fit for the COIS-R original factor structure was acceptable for the parent-reported version, but not for the child-reported version. Thus, the measure would benefit significantly from revisions. More specifically, we recommend using a split-sample procedure (conduct EFA with one part and confirm with CFA in another part) in a large cross-cultural sample.
References (69)
- et al.
Impaired but undiagnosed
Journal of the American Academy of Child and Adolescent Psychiatry
(1999) - et al.
Quality of life and disability in patients with obsessive–compulsive disorder
European Psychiatry
(2001) - et al.
Obsessive compulsive disorder in adolescence: an epidemiological study
Journal of the American Academy of Child and Adolescent Psychiatry
(1988) - et al.
Predictors and moderators of treatment outcome in the pediatric obsessive compulsive treatment study (POTS I)
Journal of the American Academy of Child and Adolescent Psychiatry
(2010) - et al.
Hope and coping in family members of patients with obsessive–compulsive disorder
Journal of Anxiety Disorders
(2006) - et al.
Predictors of quality of life and functional impairment in obsessive–compulsive disorder
Comprehensive Psychiatry
(2014) - et al.
Schedule for affective disorders and schizophrenia for school-age children-present and lifetime version (K-SADS-PL): initial reliability and validity data
Journal of the American Academy of Child and Adolescent Psychiatry
(1997) - et al.
Clinical correlates of functional impairment in children and adolescents with obsessive–compulsive disorder
Journal of Obsessive–Compulsive and Related Disorders
(2013) - et al.
Controlled comparison of family cognitive behavioral therapy and psychoeducation/relaxation training for child obsessive–compulsive disorder
Journal of the American Academy of Child and Adolescent Psychiatry
(2011) - et al.
Childhood obsessive–compulsive disorder in the NIMH MECA study: parent versus child identification of cases. Methods for the epidemiology of child and adolescent mental disorders
Journal of Anxiety Disorders
(2000)