Development and validation of the Japanese version of the obsessive-compulsive inventory

Background

Obsessive-compulsive disorder (OCD) is a chronic psychiatric illness with a mean lifetime prevalence of 2–3% in the general population[1]. An anxiety-based disorder, OCD is characterized by persistent, intrusive, and distressing obsessions (persistent thoughts, impulses, or images) or compulsions (repetitive, excessive behaviors or mental acts)[2].

In Japan, there is a national prevalence of about 2% for OCD, as in the US[3]. In one study, researchers administered the Yale-Brown Obsessive Compulsive Scale (Y-BOCS) symptom checklist[4] to 343 Japanese OCD patients to examine whether symptom dimensions were stable across cultures[5]. They found that the OCD symptom structure has substantial transcultural stability across Western and Eastern cultures, suggesting that OCD is mediated by universal psychobiological mechanisms[5]. Although several questionnaires have been developed that evaluate the severity of OCD symptoms in the Japanese population, such as the Japanese versions of the Y-BOCS[6] and Maudsley Obsessive-Compulsive Inventory (MOCI-J)[7], none are suitable for quick, effective clinical assessments. In particular, the Y-BOCS, one of the most commonly used scales in OCD research, uses a semi-structured interview format, consisting of 10 core items that assess time spent on obsessions or compulsions, resistance, interference, distress, and control[4, 6]. The scale yields three severity scores: obsessions, compulsions, and an overall score. Furthermore, it possesses a 67-item symptom checklist for an accurate assessment of symptoms. This scale has excellent psychometric properties and is useful in research on treatment outcomes[8], but it has three notable limitations that prevent it from being effective for clinical settings. First, several studies have indicated that this instrument may not correlate well with other measures of obsessive-compulsive symptoms. For example, Goodman et al.[4] found that Y-BOCS scores did not consistently correlate with the MOCI[9]. In addition, since the Y-BOCS has moderate correlations with depression and anxiety measures[4, 10], its discriminant validity is low[11]. Second, due to its interview-based format, it can be time-consuming and expensive to administer. As an interview, it requires trained interviewers, and interviewer reliability must be established to ensure accurate results. The 10 core items do not contain information on the specific content of obsessions and compulsions. Although this information can be obtained from the YBOCS checklist, the checklist has a large number of items, which makes it much more difficult to quickly identify the nature and severity of patients’ symptoms[12]. The Y-BOCS symptom checklist includes a dichotomous list of subtypes, but it does not provide a continuous measure of OCD symptom dimensions. Therefore, Y-BOCS scores are less affected by the subtypes of OCD; for example, patients with “mixed-OCD” may have low Y-BOCS scores, even though they suffer from several OCD symptoms. Thus, the Y-BOCS is difficult to use in clinical settings[11]. A self-report version of the Y-BOCS was created by Baer[13]; this version assesses the same 15 categories as the interview version. Respondents are asked to report to what extent they have experience with items such as I have violent or horrific images in my mind or I am concerned with dirt or germs. However, Wu et al.[14] raised a number of critical points regarding the symptom checklist of the self-report Y-BOCS, including the validity of the rationally based assignment of symptoms to categories, the inadequacy of the self-report format to distinguish OCD from non-OCD samples, and issues surrounding the wording of items[14, 15].

Unlike the interview-based Y-BOCS, the MOCI[9] is a self-report measure, and a Japanese version of the MOCI (MOCI-J) has been developed[7]. It consists of 30 true/false items and has satisfactory test–retest reliability and internal consistency[7, 16]. Factor analysis has revealed four subscales: Cleaning, Checking, Slowness, and Doubting[9]. However, the MOCI has two limitations[11, 12]. First, the dichotomous true–false format makes the scale less sensitive overall, as it can only assess the severity of specific symptoms, and it may only be effective in assessing changes in severity post-treatment. Second, some of its items are not directly linked to OC symptoms. The MOCI-J does not provide an adequate assessment of obsessional rumination[17]. In addition, this scale primarily assesses compulsive rituals and overemphasizes Cleaning and Checking rituals to the exclusion of other types of neutralizing activities[11]. Thus, although the inclusion of the four subscales appears to better address the heterogeneity of OCD, the subscales capture only a subset of OCD symptoms[11][12]. Another questionnaire, the Padua Inventory[18] contains 60 items, each rated on a 0–5 scale, that describe common obsessions and behavioral compulsions. The Padua Inventory yields four factors: Impaired Mental Control, Contamination, Checking, and Loss of Control of Actions[18]. The questionnaire has adequate internal consistency, test-retest reliability, and discriminate validity[11, 18]. However, the instrument has difficulty differentiating obsession and worry[19]. Furthermore, the Padua Inventory does not include some categories of obsessions and compulsions, such as neutralizing and hoarding[12, 20].

The Obsessive-Compulsive Inventory (OCI) is another self-report scale for measuring OC symptoms[12]. This scale has 42 items (e.g., I avoid using public toilets because I am afraid of disease or contamination), each of which is rated on a five-point Likert scale corresponding to frequency of symptoms in the past month and severity of distress (e.g., 0 = “not at all distressed” to 4 = “extremely distressed”). The full scale yields a total possible score of 168. We believe that the OCI is a more advantageous measurement than the other scales discussed for three reasons.

First, the OCI is a more comprehensive instrument than the Y-BOCS, MOCI, or Padua Inventory because it contains seven subscales, which allows it to capture the considerable heterogeneity of obsessions and compulsions[12]. These subscales include Washing (eight items), Checking (nine items), Mental Neutralizing (six items), Obsessing (eight items), Ordering (five items), Hoarding (three items), and Doubting (three items).

In addition, unlike the Y-BOCS, administration of the OCI does not require trained interviewers. Therefore, the OCI covers a wide range of OC phenomena in a format that is easy to administer and can be used to assess not only obsessions and compulsions in groups with diagnosable OCD, but also OC thoughts and behaviors in the general population[12, 21]. Foa et al.[12] reported good to excellent internal consistency for both the full scale and the subscales for patients with OCD, and found that the scale had good to excellent test-retest reliability for OCD patients across two weeks. The OCI also demonstrates excellent discriminant validity because OCD patients reported greater distress on the OCI than other-anxious controls (i.e., people with posttraumatic stress disorder or generalized social phobia). Finally, because the OCI total scores were positively associated with the total scores of the MOCI, the OCI was shown to have satisfactory convergent validity[12]. The psychometric properties of the OCI and its subscales have also been examined in a non-clinical student sample[21], which indicated a high internal consistency and good test-retest reliability for the total scale and each subscale. Simonds et al.[21] also found that the OCI had good convergent validity with the MOCI[9]. However, there has been no empirical study using the OCI in Eastern cultures.

The Obsessive–Compulsive Inventory—Revised (OCI-R[22]) is a brief (18-item) adaptation of the 42-item OCI[12]. In addition to yielding a total score, the OCI-R has six subscales: Washing, Checking, Ordering, Obsessing, Hoarding, and Neutralizing. The OCI-R has excellent psychometric properties in a mixed sample of patients with obsessive–compulsive disorder and other anxiety disorders.

Peng et al.[23] translated the OCI-R[22] into Chinese, and administered it to a non-clinical sample and patients with OCD. The study suggested that the Chinese version demonstrated good validity. Chasson et al.[24] evaluated the psychometric properties of the translation[23], and found strong evidence for its test-retest reliability and construct validity[24]. However, in a non-clinical sample, the internal consistency (Cronbach’s alpha) for Neutralizing (alpha = .34) and Checking (alpha = .65) subscales of OCI-R were inadequate[22]. Hajcak et al.[25] also found that the hoarding (alpha = .65) and neutralizing (alpha = .47) subscales of OCI-R did not have adequate internal consistency. In the study by Peng et al.[23], which included 209 Chinese undergraduates and 56 individuals with OCD, the majority of the coefficient alphas for the subscales of OCI-R were not strong (Washing = .64, Obsessing = .77, Hoarding = .66, Ordering = .63, Checking = .61, and Neutralizing = .53). Thus, the internal consistency of the subscales of OCI-R in non-clinical sample are not stable, as compared to OCI subscales which indicated a high internal consistencies (alpha = .78 to .95) in non-clinical sample[21].

In summary, the OCI is quick, easy to administer, and can assess a range of obsessional and compulsive behaviors. In addition, the scale has a strong evidence for its internal consistency, test-retest reliability, and construct validity in both clinical and non-clinical samples. This makes it a useful supplement to existing measures that are capable of assessing both clinical and non-clinical OC phenomena. The OCI could be especially useful in providing a comprehensive assessment of the severity of a range of OC phenomena in non-clinical samples, which could then serve as analogues for those with clinical OCD. For these reasons, a comprehensive yet brief self-report measure of OCD symptoms, such as the OCI, would be useful for assessing OCD in Japan.

The purpose of this study was to validate a Japanese version of the OCI (OCI-J) in non-clinical and clinical samples. To achieve this, we examined its convergent validity and test-retest reliability. We also examined the scale’s ability to discriminate between individuals with OCD and individuals with another anxiety disorders by comparing the scores of three key groups: OCD patients, healthy controls (non-clinical participants), and anxiety controls with panic disorder. We predicted that OCD participants would have much higher scores on the OCI than would the participants in the other two groups.

Purpose

The aims of Study 1, which used a non-clinical sample, were to investigate the internal consistency of the OCI-J; confirm the convergent validity of the OCI-J by determining the correlation with another measure of OC symptoms (MOCI-J) and other measures of psychopathology, such as symptoms of depression and anxiety; and determine the test-retest reliability of the OCI-J.

Methods

The data of 150 Japanese undergraduates (68.3% female; age range: 18–50 years, M = 20.60, SD = 3.94) were used to estimate the convergent validity of the OCI-J. In addition, 118 non-clinical Japanese undergraduates (56.3% male; age range: 18–27 years; M = 19.62, SD = 1.12) participated in establishing the test-retest reliability; these participants were different from those who took part in testing the convergent validity.

Information about this study was provided through handouts and oral presentations given in lecture rooms at the university sites. Interested students were asked to contact the researchers, and those who did received and completed the questionnaire. This study was approved by the ethics committee of, the Graduate School of Medicine, Chiba University (receipt number, 1122).

Measurements

Mean, standard deviation, and internal consistency

Correlations between variables

Test-retest reliability

The test-retest interval was 14 days. Reliability estimates were calculated using Pearson’s r between scores on the first and second tests. Overall, the pre-post score correlation of the total score of the OCI-J was 0.76 (p < .001). The results for the subscales were as follows: Washing, 0.81 (p < .001); Checking, 0.80 (p < .001); Doubting, 0.77 (p < .001); Ordering, 0.72 (p < .001); Obsessing, 0.75 (p < .001); Hoarding, 0.70 (p < .001); and Neutralizing, 0.75 (p < .001). These values indicated adequate test-retest reliability.

Study 2

The purpose of our second study was to investigate whether the OCI-J could successfully discriminate between OCD patients, healthy controls, and anxiety controls (patients with panic disorder).

Participants and procedure

The clinical participants were recruited from eight outpatient mental health clinics in Japan. Participants were excluded if they were under age 18 or over age 60.

Potential OCD participants (N = 48) were diagnosed by a psychiatrist using the OCD criteria of the Diagnostic and Statistical Manual of Mental Disorders, 4th edition, Text Revision (DSM-IV-TR, American Psychiatric Association[34]). Individuals were excluded if they had a comorbid disorder on Axis I or II of the DSM-IV-TR (e.g., other anxiety disorder, major depression, mental retardation, personality disorder, psychotic disorder, dementia, mental retardation, or substance use disorder). In addition, the Japanese version of the self-report Y-BOCS[35] was administered for the participants with OCD; individuals were included if they scored over 16 points (the suggested cut-off for OCD). To confirm that potential OCD participants had no panic disorder or depression symptoms, the Japanese version of the Self-Report PDSS (PDSS-SR-J)[36] and Japanese version of the Beck Depression Inventory, Version II (BDI-II)[37] were administered; individuals were excluded if they scored over the cut-off point (minimal symptoms vs. mild symptoms) of 8 for the PDSS-SR-J[36] and 14 for the BDI-II[38]. The final OCD group consisted of 35 individuals (OCD group; male = 12, female = 23; age range: 20–48 years, M = 30.11, SD = 3.44).

Potential anxiety control participants (N = 36) were diagnosed by a psychiatrist using the DSM-IV-TR criteria for panic disorder. The individuals were excluded if they had a comorbid Axis I or II disorder, but not if they had a comorbid anxiety disorder other than OCD (e.g., a patient with panic disorder and social anxiety disorder). In addition, the PDSS-SR-J[36] was administered; individuals were included if they scored over 8 points. To confirm that they had no OCD or depression symptoms, the self-report Y-BOCS[35] and BDI-II[37] were administered; potential participants were excluded if they scored over the cut-off point of 16 on the self-report Y-BOCS and 14 on the BDI-II. Twenty-two individuals met the above criteria (anxiety control group; male = 9, female = 13; age range: 22–53 years, M = 35.60, SD = 4.50). The frequency of each principal anxiety disorder was as follows: panic disorder without agoraphobia, 50%; panic disorder with agoraphobia, 32%; panic disorder without agoraphobia and comorbid general anxiety disorder, 9%; panic disorder with agoraphobia and comorbid social anxiety disorder, 5%; and panic disorder with agoraphobia and comorbid post-traumatic stress disorders, 5%.

Non-clinical participants were Japanese university students. Potential participants (N = 40) were excluded if they reported a clinical history of mental disorder, head injury, central nervous system diseases, or substance abuse. To confirm that these individuals had no OCD, panic, or depression symptoms, the self-report Y-BOCS, PDSS-SR-J and BDI-II[37] were administered; participants were excluded if they scored over the cut-off point of 16 on the Y-BOCS, 8 on the PDSS-SR-J, and 14 on the BDI-II. Thirty-seven non-clinical Japanese individuals met the inclusion criteria for the healthy control group (male = 17, female = 20; age range: 20–52 years; M = 25.61, SD = 5.44).

This study was approved by the ethics committee of, the Graduate School of Medicine, Chiba University (receipt number, 1122). All individuals participated on a voluntary basis and gave their written consent before entering the study. Eligible participants completed the OCI-J individually.

Measurements

In this study, the Self-Report Y-BOCS, PDSS, and BDI-II were administered to confirm whether potential participants met the inclusion or exclusion criteria (see the Participants and procedure section).

Limitations and future research

This study has a number of limitations that must be addressed. First, we did not investigate the relationship between the OCI-J and other self-report measures of OCD, such as the Japanese version of Padua Inventory[48, 49], or clinician-administered scales, such as the Y-BOCS[6]. These comparisons would be helpful to fully establish concurrent validity. Nevertheless, we believe that our use of the MOCI-J was suitable for this study because the MOCI-J was used to assess the concurrent validity of the OCI[21] and the Chinese version of the OCI-R[23] in previous studies.

The small clinical sample size and lack of a standardized semi-structured interview, such as the SCID-I/P[50], to confirm clinical diagnoses in the clinical sample, must be noted as a limitation. Furthermore, we did not investigate the factorial validity of the OCI due to the small sample size. The original study reporting the development of the OCI[12] suggested that the OCI has a seven-factor structure, while Sica et al.[27] proposed a five-factor structure based on the factor analysis of a non-clinical sample. Thus, the factor structure of the OCI appears to vary by study. There has been no research on the factorial validity of the OCI using a clinical sample in an Eastern culture. Further studies using clinical samples are required to confirm if the factor structure of the OCI is consistent across Western and Eastern cultures or if the OCI-J has a different factor structure that is more reflective of the Japanese culture.

Nonetheless, the present study is a first step in confirming the validity of the OCI-J. We plan to study the psychometric properties of the OCI-J further, including performing factor analysis using Japanese clinical samples. In addition, we hope that our results will contribute to the development of a short version of the OCI (i.e., the OCI-R) for use in Japanese populations.