Introduction
The scarcity of research on OCD symptom presentation in China might be due to the lack of validated Chinese instruments to quantify OCD subtype severity. The Padua Inventory has been translated and validated for a Chinese population (Zhong et al., Reference Zhong, Qin, Cai, Tan, Wang and Xu2006), but it does not measure the many common subtypes of OCD (e.g. ordering, hoarding). A research team translated and tested a Mandarin version of the Obsessive-Compulsive Inventory- Revised (OCI-R; Foa et al., Reference Foa, Huppert, Leiberg, Langner, Kichic and Hajcak2002) in a non-clinical sample, as well as in patients with OCD (Peng, Yang, Miao, Jing and Chan, Reference Peng, Yang, Miao, Jing and Chan2011). The Peng et al. (Reference Peng, Yang, Miao, Jing and Chan2011) study suggested that the Chinese variant of the OCI-R demonstrated good validity, but the study only presented limited data regarding discriminant validity by demonstrating a statistically significant higher score on the measure in OCD patients compared to the healthy control participants. However, the study would have been strengthened by comparing OCI-R scores in a third clinical group (e.g. non-OCD anxious controls), as well as by incorporating measures of related psychopathology (e.g. anxiety and depression).
Building on preliminary findings from the Peng et al. (Reference Peng, Yang, Miao, Jing and Chan2011) study, the current investigation presents data on reliability and validity of a Mandarin translation of the OCI-R, using a student and clinical sample (with two groups, OCD and non-OCD clinical controls).
Method
Participants
The college student sample (CS group) consisted of 1950 student volunteers (63.2% female; M age = 20.56, SD = 1.17; M = 13.46 years of education; SD = .74) from Weifang Medical University, Shandong Province. The clinical sample was divided into two groups. The first group included patients with diagnosed (DSM-IV; American Psychiatric Association, 2000) obsessive compulsive disorder (OCD group; 40% female; 46% employed; 28% married/cohabitating; M age = 25.90, SD = 9.42; M = 11.36 years of education; SD = 2.34) as their most severe problem, and the second group included patients with any DSM-IV diagnosed anxiety disorder, except OCD as their most severe problem (anxiety group; 66% female; 48% employed; 40% married/cohabitating; M age = 35.20, SD = 12.16; M = 10.90 years of education; SD = 3.15). The OCD group consisted of 50 patients (29 males, 20 females, 1 unspecified), and the anxiety group consisted of 50 patients (15 males, 33 females, 2 unspecified) without current OCD, but with one or more of the following: panic disorder (39%), posttraumatic stress disorder (6%), generalized anxiety disorder (22%), social phobia (28%), and anxiety disorder not otherwise specified (6%).
Measures
The Obsessive-Compulsive Inventory-Revised was translated and back-translated by study authors (GSC and JPW) and their students. A revised back-translated version was approved by OCI-R developers (Edna Foa and staff, personal communication). We opted for the OCI-R because it eliminates several problems with the full-length OCI, including redundancy between subscales, scoring confusion due to irregular subscale lengths, and unwieldiness for routine clinical use (Foa et al., Reference Foa, Huppert, Leiberg, Langner, Kichic and Hajcak2002). The Chinese version of the Padua Inventory (PI; Zhong et al., Reference Zhong, Qin, Cai, Tan, Wang and Xu2006) was used to evaluate convergent validity. The Chinese version of the Center for Epidemiologic Studies Depression Scale (CESD; Zhang et al., Reference Zhang, Wu, Fang, Li, Han and Chen2010) and Chinese version of the Self-Rating Anxiety Scale (SAS; Liu, Oda, Peng and Asai, Reference Liu, Oda, Peng and Asai1997) were included for discriminant validity.
Procedure
CS data were collected at Weifang Medical University with approval from an ethics committee. Four weeks later, 229 completed a re-administration for evaluating test–retest reliability. The OCD and Anxiety groups were recruited from mental health centres at two hospitals in Shangdong and Beijing. Student and clinical participants completed the same questionnaires.
Missing data. Participants who had more than 20% missing values on the scale were removed from further analyses (n = 33). For the remaining participants, the missing values were substituted with the mean score of the scale.
Results
Descriptive
Age was significantly different in all three groups. However, we conducted a second set of analyses in which we controlled for age, and findings did not differ. As a result, for simplicity, analyses without controlling for age are reported below. The OCD group scored highly on the PI (M = 60.06, SD = 32.95), SAS (M = 42.06, SD = 9.16), and CESD (M = 29.30, SD = 11.20). The anxiety group also scored highly on the PI (M = 48.79, SD = 31.29), SAS (M = 45.88, SD = 9.38), and CESD (M = 26.35, SD = 12.04). As a comparison, students scored a mean of 29.42 (SD = 22.26) on the PI, 32.50 (SD = 6.42) on the SAS, and 12.14 (SD = 8.68) on the CESD.
Factor structure
To test the stability of the factor structure of the CH-OCI, we conducted CFA with the data from college students, as the sample size for patients was too limited. The goodness-of-fit indices for the CH-OCI indicated that the college student data fit well to the 6-factor structure model, χ2 (120, N = 1950) = 871.738, p < .001, CFI = .940, RMSEA = .057, RMR = .037. See Table 1 for factor loadings and squared multiple correlations (SMC). For a comparison, the one-factor model, in which 18 items loaded on the one factor, was also tested. The goodness-of-fit indices indicated poor fit, χ2 (135, N = 1950) = 2897.074, p < .001, CFI = .779, RMSEA = .102, RMR = .065.
Means, correlations, Cronbach's alphas, and factor analyses data for CH-OCI
Notes: OCD = OCD patient sample; AG = anxiety group; CS = college student sample; CH-OCI = Chinese Obsessive-Compulsive Inventory-Revised; PI Cont = Padua Inventory Contamination subscale; PI Check = Padua Inventory Checking subscale; PI Control = Padua Inventory Mental Control subscale; PI Impulse = Padua Inventory Impulses subscale; SAS = Zung Self-Rating Anxiety Scale; CESD = Center for Epidemiologic Studies Depression Scale; α = Cronbach's Alpha for the OCD group; SMC = Squared Multiple Correlation *All p-values < .05. a n = 50, b n = 50, c n = 1950, d n = 718, e n = 1232, fall df1 = 2, df2 = 2047, g n = 1950 (college students).
Correlations
For the CS group, correlations between each of the subscales and the total scale of the CH-OCI were large (between .69 and .80), but inter-correlations among the subscales were only moderate (between .35 and .65), reflecting a small overlap between them. Correlations of CH-OCI with the PI, SAS, and CESD highlight patterns of convergent and discriminant validity (see Table 1).
Criterion-related validity
There are significant differences across the three groups on the CH-OCI total and all subscales (see Table 1 for group means). Compared with the college student group, the anxiety group scored higher on the CH-OCI total (t = 8.57, p < .001) and its subscales (washing: t = 1.27, p < .001; checking: t = 1.10, p < .001; ordering: t = 1.44, p < .001; obsessing: t = 2.90, p < .001; neutralizing: t = 1.46, p < .001), except the hoarding subscale (t = 0.56, p = .092), and patients with OCD scored higher on the CH-OCI total (t = 14.89, p < .001) and its subscales (washing: t = 1.89, p < .001; checking: t = 3.40, p < .001; ordering: t = 2.34, p < .001; obsessing: t = 5.20, p < .001; neutralizing: t = 1.18, p < .001), except the hoarding subscale (t = 0.44, p = .230). Moreover, compared with the anxiety group, the OCD group scored higher on the CH-OCI total (t = 6.32, p = .002), checking subscale (t = 2.30, p < .001) and obsessing subscale (t = 2.30, p < .001).
Test-retest reliability (4-week interval)
Results suggests adequate stability for the total score (r = .63). However, for the CH-OCI subscales, the coefficients were smaller (washing: r = .46; obsessing: r = .60; hoarding: r = .56; ordering: r = .64; checking: r = .45; neutralizing: r = .31).
Internal reliability
Presented in Table 1, the coefficient alpha for the entire scale was very high (α = .84 for the OCD group), and the alphas of the subscales were adequate (α = .75 and above for the OCD group), except for the hoarding subscale (α = .45 for the OCD group).
Discussion
The current study evaluated the psychometric properties of a Chinese version of the OCI-R (CH-OCI) and found strong evidence for its reliability and validity. Evidence for construct validity of the CH-OCI was supported by its factor structure, as well as overall patterns of convergent and discriminant validity. However, further research is needed to support convergent and discriminant validity in this measure, as the magnitude of the correlation between SAS and CH-OCI was not substantially larger than the correlation between CESD and CH-OCI, as would be expected. Similarly, a larger correlation between CH-OCI and PI would be expected. Nonetheless, patterns of correlations suggest convergent and discriminant validity (i.e., r CH-OCI*PI > r CH-OCI*SAS > r CH-OCI*CESD).
With respect to criterion-related validity, the CH-OCI was useful for discriminating those with OCD from those with other forms of anxiety, as well as discriminating patients with anxiety and OCD from non-clinical samples. One notable exception is the hoarding subscale of the CH-OCI, which demonstrated poor discrimination between OCD patients and anxiety patients, as well as between OCD patients and student participants. The measure also demonstrated good test-retest stability for the total score of the CH-OCI. However, test-retest reliability for the subscales of CH-OCI were less robust, which might be a result of the subscales containing fewer items. In addition, the total score for the CH-OCI demonstrated good internal consistency, and Cronbach's alphas were adequate for all subscales, except for the hoarding subscale. Thus, the hoarding subscale should be interpreted cautiously, especially when also considering the evidence of its inability to discriminate groups.
Our study provides further evidence of the measure's discriminant and criterion-related validity, thus highlighting its utility for discriminating the OCD construct from alternative forms of distress, such as anxiety and depression. A limitation of the current study is the lack of establishing clinical diagnoses in the patient sample using a standardized semi-structured interview. In addition, further validation of the CH-OCI using a clinical sample that is characterized by a reliable and valid diagnostic interview would help allay this concern. Future research should evaluate the factor analyses using data from patients with OCD instead of student samples. Nonetheless, evidence from the current investigation supports the use of the CH-OCI in clinical and research endeavours in the service of expanding our understanding of, and treatment for, OCD phenomena in China.
Acknowledgements
The authors would like to thank Edna Foa and colleagues for their generous assistance in translating the Obsessive-Compulsive Inventory-Revised to Mandarin.
Comments
No Comments have been published for this article.