Psychometric properties of the Japanese version of the Beck Cognitive Insight Scale: Relation of cognitive insight to clinical insight

Authors


Abstract

Aim:  Insight in schizophrenia is considered to have a multidimensional construct, and cognitive insight is thought to be an important dimension of insight: an ability to evaluate and correct one's own distorted beliefs and misinterpretations. The Beck Cognitive Insight Scale (BCIS) was developed to measure cognitive insight, and studies have shown that cognitive insight is associated with several clinical features in schizophrenia. The aim of the present study was to develop a Japanese version of the BCIS (BCIS-J) and assess the psychometric properties of this instrument.

Methods:  The BCIS-J was completed by university students (n = 183) and patients with schizophrenia (n = 30). The Japanese version of the Schedule for the Assessment of Insight was used to measure clinical insight in patients with schizophrenia, and its association with the BCIS-J was investigated.

Results:  Factor analysis in the university students indicated that the BCIS-J was composed of two factors, self-reflectiveness and self-certainty, as was seen in the original BCIS. The relation between the specific dimensions of clinical insight and each component of the BCIS-J in patients with schizophrenia indicated that overconfidence in their belief or judgment may be involved in their attitude to treatment and openness to feedback, and objectivity might be essential to attribute one's symptoms as part of mental illness.

Conclusions:  The BCIS-J is a reliable and valid instrument to measure cognitive aspects of insight and appears to complement clinical insight scales.

LACK OF INSIGHT is considered to be one of most important features in the pathogenesis, diagnosis, and treatment of schizophrenia. A contemporary model of insight regards it as a multidimensional continuum rather than a unitary ‘all or none’ phenomenon.1,2 According to this model, researchers investigate clinical aspects of insight (clinical insight3), such as awareness of mental disorder, recognition of need for treatment, and ability to relabel unusual mental events as pathological.2,4–6

Beck et al. recently proposed another aspect of insight from the cognitive point of view, namely, cognitive insight.3 Cognitive insight is conceptualized as an ability to evaluate and correct one's own distorted beliefs and misinterpretations, and a higher level of cognitive process is hypothesized to be involved in it. To measure this aspect of insight, the Beck Cognitive Insight Scale (BCIS) was developed.3 The initial study by Beck et al. found that the BCIS is composed of two components: self-reflectiveness and self-certainty.3 The former includes items measuring objectivity, reflectiveness, and openness to feedback, and the latter measures certainty about one's own beliefs or judgment. A composite index score, an estimated measure of overall cognitive insight, is calculated by subtracting the score for the self-certainty subscale from the score for the self-reflectiveness subscale.

Reliability and validity of the instruments have been shown in a mixed group of inpatients with psychosis and depression,3 a group of middle-aged and older outpatients with schizophrenia,7 and a group of patients with bipolar disorder.8 The BCIS was also applied to the normal population.9,10 Internal consistency of the BCIS was considered to be similar between clinical and non-clinical samples,8,10 but factor loading or factor structure of the BCIS in the normal population has not been reported except in a poster presentation by Warman et al. (unpubl. data, 2004).

Overall cognitive insight indicated by the composite index scale of the BCIS in patients with schizophrenia has been shown to be correlated with clinical insight measured by the Scale to Assess Unawareness of Mental Disorder (SUMD)3,11 and the Birchwood Insight Scale (IS).7 The association, however, between the individual items of the clinical insight scales and the subscales of the BCIS has not been fully explored. Beck et al. observed a correlation between the SUMD awareness of delusion and self-reflectiveness, but no other correlation was found except for the relation between the SUMD awareness of mental disorder and the composite index.3 Pedrelli et al. reported a correlation between the self-reflectiveness and the relabel subscale and the total score from the IS.7

Clinical insight is known to be associated with depression in patients with psychosis,12,13 that is, patients become more depressed as insight increases. It is not clear, however, whether this is true for cognitive insight. One study found a correlation between depression measured by the Beck Depression Inventory-II (BDI-II) and cognitive insight in patients with schizophrenia or schizoaffective disorder,10 but another study did not find such a correlation.3 Pedrelli et al. found no association between depression measured by the Hamilton Rating Scale for Depression in middle-aged and older patients with schizophrenia and schizoaffective disorder.7

In the present study we developed the Japanese version of the BCIS (BCIS-J) and investigated psychometric properties of this instrument. The BCIS-J was applied to healthy university students to determine the factor structure of the BCIS-J and to evaluate the reliability of the instrument in a normal population sample. We predicted that the factor structure of the BCIS-J in the normal sample is similar to that of the original BCIS. We then evaluated the BCIS-J and the Japanese version of the Schedule for the Assessment of Insight (SAI-J)14 in patients with schizophrenia to determine the association between clinical and cognitive insight. We were particularly interested in the association between each component of the BCIS-J and the three subscales of the SAI-J, that is, adherence with treatment, awareness of illness, and relabeling of psychotic phenomena. We also investigated the relationship between the cognitive insight and psychosocial variables in patients with schizophrenia.

METHODS

Japanese version of the Beck Cognitive Insight Scale

The BCIS is a self-report instrument consisting of 15 items rated on a 4-point scale (0, do not agree at all; to 3, agree completely). With the original authors' permission, the BCIS was first translated into Japanese by one of the authors (A.K.). This Japanese version of the BCIS was back-translated into English by a bilingual psychologist who had not previously seen the original English text. The results of the back-translation were then examined and judged satisfactory by a native-English-speaking psychologist who was an expert in cognitive psychology.

Other instruments

The SAI-J was translated from the original SAI,2 and the reliability and validity of the SAI-J has been confirmed by Sakai et al.14 The SAI is a semi-structured clinical interview designed to measure clinical insight, and it consists of seven items constituting three subscales: (i) adherence to treatment; (ii) awareness of illness; and (iii) relabeling of psychotic phenomena.2 The Positive and Negative Syndrome Scale (PANSS) is a 30-item scale designed to include three subscales for different types of symptoms: positive symptoms, negative symptoms, and general psychopathology.15 Yamada et al. translated the PANSS Rating Manual into Japanese and applied this scale in Japan.16 The BDI-II is a 21-item self-report instrument developed to measure the severity of depression in adults and adolescents.17 The reliability and validity of the Japanese version of the BDI-II has been established by Kojima et al.18

Participants

One hundred and eighty-three university students (104 men and 79 women with a mean age of 18.9 ± 1.2 years) were recruited at Tohoku University. Participants were asked as part of the questionnaire if they had ever been diagnosed with a psychiatric disorder. Participants who reported a history of psychiatric disorder diagnosis were excluded from the university student sample.

Thirty patients with schizophrenia or schizophreniform disorder (15 men and 15 women) were recruited from Tohoku University Hospital (Table 1). All of the diagnoses were made according to the DSM-IV-TR criteria19 by a trained psychiatrist (K.M.) on the basis of all available information. All patients were clinically judged to be stable enough to undergo the assessment. All participants provided written informed consent and voluntarily agreed to participate. The research was approved by the Ethics Committee of Tohoku University Graduate School of Medicine and Tohoku University Hospital.

Table 1. Subject characteristics
VariablesMean ± SD
  1. BDI-II, Beck Depression Inventory-II; PANSS, Positive and Negative Syndrome Scale; SAI-J, Japanese version of the Schedule for the Assessment of Insight.

Age (years)26.73 ± 6.09
Education (years)12.60 ± 2.25
Duration of illness (months)46.13 ± 58.90
Antipsychotic drug (mg/day, chlorpromazine equivalence)655.60 ± 489.01
BDI-II total20.50 ± 13.09
SAI-J total8.93 ± 3.27
PANSS positive15.43 ± 5.52
PANSS negative18.07 ± 6.00
PANSS general36.97 ± 6.99
PANSS total70.47 ± 16.18

Procedure

University student sample

After the administration of the BCIS-J to the university student sample, factor analysis was conducted to assess the factor structure of the BCIS-J. Reliability of the BCIS-J was tested using coefficient alpha and the test–retest method. To examine the test–retest reliability of the BCIS-J, 52 subjects (27 men, 25 women with a mean age of 18.9 ± 1.8 years) from the original sample were retested 3 weeks after the initial test. The correlations among self-reflectiveness, self-certainty and composite index were investigated in order to evaluate the internal relationships in the BCIS-J.

Schizophrenia sample

The PANSS and the SAI-J were administered by a senior psychiatrist (K.M.) and a research psychologist (T.U.) with extensive training in administration of these measures, and the rating was determined by consensus of the two. The BCIS-J and the BDI-II then were completed by the participants immediately after the assessment with the PANSS and the SAI-J.

Correlation analysis with 95% confidence interval (95%CI) was conducted to determine the relation of the BCIS-J scores and the total and individual subscale scores of the SAI-J, BDI-II, the positive and negative syndrome scores and the insight subscales of the PANSS, and other demographic variables. The t-test was used to estimate the effects of sex on cognitive insight.

Statistical analysis

All statistical analyses were performed using the Japanese version of SPSS 14.0 for Windows (SPSS, Chicago, IL, USA), with the significance level set at P < 0.05 (two-tailed test).

RESULTS

University student sample

Factor analysis and reliability

The varimax-rotated principal factor method for the 15 BCIS-J items is shown in Table 2. From the results of factor analysis, the two-factor solution was determined by scree plot inspection and was found to be interpretable.

Table 2. Varimax-rotated principal factor method for the BCIS-J
ItemIIIh2
  1. BCIS-J, Japanese version of the Beck Cognitive Insight Scale.

10When people disagree with me, they are generally wrong.0.710.070.51
 2My interpretations of my experiences are definitely right.0.70−0.110.50
13I can trust my own judgment at all times.0.69−0.030.48
 9I know better than anyone else what my problems are.0.55−0.030.31
11I cannot trust other people's opinion about my experiences.0.530.080.29
 7If something feels right, it means that it is right.0.52−0.070.28
 8Even though I feel strongly that I am right, I could be wrong.−0.240.620.44
 6Some of the ideas I was certain were true turned out to be false.−0.180.590.38
 5Some of my experiences that have seemed very real may have been due to my imagination.0.060.470.23
 4I have jumped to conclusions too fast.0.010.450.21
12If somebody points out that my beliefs are wrong, I am willing to consider it.−0.270.440.26
 1At times, I have misunderstood other people's attitudes towards me.−0.010.440.19
 3Other people can understand the cause of my unusual experiences better than I can.0.040.340.12
14There is often more than one possible explanation for why people act the way they do.−0.010.310.10
15My unusual experiences may be due to my being extremely upset or stressed.0.080.200.05
%Total16.6812.1628.83
%Common5842100

Factor I consisted of six items, which had salient loadings (>0.30): 2 (definitely right), 7 (feels right is right), 9 (know problems), 10 (people are wrong), 11 (cannot trust opinion), and 13 (trust own judgment). All of these items address overconfidence about beliefs or judgments and comprise ‘self-certainty’ component as in the original BCIS. Factor II contained items 1(have misunderstood), 3 (others more objective), 4 (jumped to conclusions), 5 (due to imagination), 6 (ideas were false), 8 (could be wrong), 12 (willingness to consider), and 14 (possible explanation). Although only item 15 (due to stress) did not have a salient loading (0.20), we nevertheless included this item in the factor II in accordance with the original BCIS.3 Factor II therefore consisted of the nine statements that were termed ‘self-reflectiveness’ and which included items related to openness to feedback, recognition of having jumped to conclusions at times, and ability to acknowledge fallibility. These are the same nine items as those in the self-reflectiveness component of the original BCIS. The composite index was calculated by subtracting the self-certainty score from the self-reflectiveness score, as in the original BCIS study.3 The mean of the self-reflectiveness, self-certainty, and composite index scores for the university students was 11.53 ± 3.47, 4.24 ± 3.00, and 7.30 ± 4.70, respectively.

The internal consistencies of the self-certainty and self-reflectiveness scores were measured by calculating Cronbach alpha coefficients. The coefficient alphas of the self-certainty and self-reflectiveness scores were 0.78 and 0.67. The stability of the scale was established by using the test–retest method. The 3-week test–retest reliability of the self-reflectiveness, self-certainty, and composite index scores were acceptable (Table 3).

Table 3. Test–retest correlations of the BCIS-J
 Mean (± SD)Reliability coefficient
FirstSecondICC
  • **

    P < 0.01.

  • BCIS-J, Japanese version of the Beck Cognitive Insight Scale; ICC, intra-class coefficient.

Self-reflectiveness11.85 ± 3.3011.92 ± 3.370.86**
Self-certainty4.27 ± 2.134.00 ± 2.130.79**
Composite index7.58 ± 3.657.92 ± 4.210.82**

The self-reflectiveness and self-certainty scores were found to significantly correlate with the composite index (r = 0.77, 95%CI: 0.70–0.82, P < 0.01; r = −0.68, 95%CI: −0.75 to −0.59, P < 0.01, respectively). No significant correlation was found between the self-reflectiveness and self-certainty scores.

Schizophrenia sample

The mean of the self-reflectiveness, self-certainty, and composite index scores for the patients with schizophrenia were 12.37 ± 4.08, 6.53 ± 3.28, and 5.83 ± 5.74, respectively.

Relation of the BCIS-J to clinical insight

The self-reflectiveness and composite index scores were significantly correlated with the SAI-J total score (Table 4) and the insight subscale of the PANSS (r = −0.39, 95%CI: −0.66 to −0.03, P < 0.05; r = −0.45, 95%CI: −0.70 to −0.11, P < 0.01, respectively).

Table 4. Pearson correlations between cognitive and clinical insight in patients
 BCIS-J
Self-reflectivenessSelf-certaintyComposite-index
r (95%CI) P r (95%CI) P r (95%CI) P
  • *

    P < 0.05,

  • **

    P < 0.01.

  • BCIS-J, Japanese version of the Beck Cognitive Insight Scale; CI, confidence interval; SAI-J, Japanese version of the Schedule for the Assessment of Insight.

SAI-J      
 Adherence to treatment0.16 (−0.21–0.50)0.39−0.38 (−0.65–−0.02)0.04*0.34 (−0.02–0.62)0.07
 Awareness of illness0.43 (0.08–0.68)0.02*−0.30 (−0.60–0.07)0.110.48 (0.14–0.72)0.01**
 Relabeling of psychotic phenomena0.41 (0.06–0.67)0.02*0.12 (−0.25–0.46)0.540.23 (−0.14–0.55)0.23
 Total score0.52 (0.20–0.74)0.00**−0.22 (−0.54–0.15)0.240.50 (0.17–0.73)0.01**

The subscales of the SAI-J were significantly correlated with the indicated BCIS-J scores: the adherence to treatment subscale with self-certainty score, the awareness of illness subscale with the self-reflectiveness and composite index, and the relabeling of psychotic phenomena subscale with self-reflectiveness score (Table 4).

Correlations of the BCIS-J with psychosocial variables

The composite index score was significantly correlated with the BDI-II (r = 0.42, 95%CI: 0.07–0.67, P < 0.05). No significant correlation was found between the BCIS-J and the other symptom measures.

The self-reflectiveness and composite index scores were significantly correlated with age (r = −0.37, 95%CI: −0.64 to −0.01, P < 0.05; r = −0.47, 95%CI: −0.71 to −0.13, P < 0.01, respectively), and the self-certainty score was significantly correlated with duration of illness (r = 0.46, 95%CI: 0.12–0.70, P < 0.05). No other effect was observed.

To remove the effect of age or duration of illness on cognitive insight scales, we conducted partial correlation analysis and found no noticeable difference between the result of the partial and the usual correlation analysis.

DISCUSSION

In the present study we developed the BCIS-J and examined psychometric properties of the instrument in healthy university students and patients with schizophrenia.

Factor analysis of the BCIS-J in healthy volunteers showed that the BCIS-J was composed of two factors, self-reflectiveness and self-certainty, which was the same as that in the original BCIS determined by Beck et al. in a sample of inpatients with schizophrenia, schizoaffective disorder, or mood disorder.3 The coefficient alpha of the self-reflectiveness scores was <0.70, similar to that in the original study,3 but this value is considered acceptable for the present research purpose given that the subscales are composed of fewer than 10 items.3 The present findings are consistent with the previous study by Warman et al., which confirmed that the basic factor structure and internal consistency of the BCIS was similar for the normal population and the clinical sample (Warman et al., unpubl. data, 2004). Recently Engh et al. also observed acceptable internal consistency of the BCIS in a Turkish normal sample.8 The findings supported generalizability of the two-factor components of cognitive insight to the non-clinical healthy sample as well as to the clinical sample.7 The test–retest reliability intra-class coefficients of the BCIS-J confirmed stability of cognitive insight in the normal population. The results in the university student sample indicate reliability of the BCIS-J.

Cognitive insight is considered as a cognitive process rather than an insight into illness itself, and the common cognitive process might be attributable to the cognitive insight between the non-clinical healthy sample and clinical sample. Warman and Martin, for example, demonstrated that university students who had no history of psychotic disorders but were more prone to delusions were overconfident in their judgment,9 similar to the delusional patients with psychotic disorders.10

Overall cognitive insight measured by the composite index of the BCIS-J was significantly correlated with clinical insight as measured by the total score of the SAI-J and the insight subscale of the PANSS. This is compatible with studies that found a correlation between the composite index score of the BCIS and clinical insight measured by the total score of IS,7 and the SUMD total sum of three global items.11 Overall cognitive insight appears to assess different but related aspects of insight, and the BCIS-J appears to have convergent validity.

The relation between the different aspects of clinical insight and each component of cognitive insight was examined. The negative correlation between the self-certainty and the treatment subscale of the SAI-J in the present study suggests that patients who are overconfident in their judgment less clearly realize their need for treatment. Previous studies have shown that poor clinical insight is associated with non-adherence to treatment in patients with schizophrenia.20,21 Overconfidence in belief or judgment may be involved in the attitude to treatment of patients with schizophrenia. Self-reflectiveness, however, was correlated with the awareness of illness and relabeling of psychotic phenomena subscales of the SAI-J. These results are consistent with some of the findings of previous studies that found a correlation of self-reflectiveness with the ability to relabel psychotic symptoms but not with the ability to accept mental illness.3,7 Openness to feedback and objectivity might be essential to attribute one's symptoms to mental illness. Self-reflectiveness as well as overall cognitive insight seem to be important in the awareness of mental disorder, as was found in both the present study and that by Beck et al.3

The previous study by Warman et al. observed a positive correlation between cognitive insight and depression in patients with schizophrenia,10 whereas Beck et al. found such an association only in patients with major depression but not in patients with schizophrenia.3 The present finding supports the former result, that is, cognitive insight appears to be correlated with depression in schizophrenia, as clinical insight is.12,13

A limitation of the present study was the small sample size. This precluded thorough investigation of factor structure and reliability of the BCIS-J in patients with schizophrenia. Future research should be done with a diverse and larger sample of patients to explore the additional psychometric properties of the BCIS-J.

ACKNOWLEDGMENTS

This research was supported in part by the Grant-in-aid for Young Scientists (B) 1790803, the Grand-in-aid for Scientific Research (C) 19591336 from the Ministry of Education, Culture, Sports, Science and Technology, Japan, and the Grant from Research Group for Schizophrenia, Japan.

Ancillary