Clinical assessment of bipolar depression: validity, factor structure and psychometric properties of the Korean version of the Bipolar Depression Rating Scale (BDRS)

A majority of patients with bipolar disorder experience depressive symptoms throughout approximately half of their lives [1, 2], and as a result, are often misdiagnosed with unipolar depression (depressive disorder) [3]. The failure to correctly diagnose these patients may lead to inappropriate treatment that can worsen the course of the disease [4] and, thus, a diagnostic approach for the recognition of bipolar disorder is essential. Nevertheless, it is difficult to distinguish bipolar depression from unipolar depression in real clinical practice settings. Active inquiry regarding past mania or hypomania is essential to diagnosis, but there are often no distinct prior episodes of mania to be described by the patient.

Several studies have suggested that there is a broad range of symptomatic differences between patients with unipolar depression and those with bipolar depression. Melancholic symptoms (worthlessness, marked anhedonia, and pathological guilt) and atypical depressive features (hypersomnia, hyperphagia, mood reactivity, and leaden paralysis) are more common in bipolar depression than in unipolar depression [5‐9]. It has also been found that mixed features such as irritability, mood lability, distractibility, and racing thoughts are more common in patients with bipolar depression [10, 11]. Yet, the widely used scales in clinical and research practice in bipolar depression are the Montgomery and Asberg Depression Scale (MADRS, [12]) and the 17-item Hamilton Depression Scale (HAMD, [13]), which were originally developed for use in unipolar depression. They do not capture many core symptoms of bipolar depression, including atypical and mixed features.

The Bipolar Depression Rating Scale (BDRS) is a semi-structured, observer-rated scale for clinical assessment of bipolar depression. It was developed by Berk et al. [14] based on previous phenomenological research and literature reviews that assessed differences between patients with bipolar depression and those with unipolar depression [7, 8, 15, 16]. The BDRS comprises items that evaluate the clinical features associated with the depressive phase of bipolar disorder, including atypical symptoms and mixed phenomenology [14]. This measure can be used to assess depressive symptomatology as well as to evaluate the effectiveness of therapeutic agents for the treatment of bipolar depression.

The objectives of this study were to develop a Korean version of the BDRS (K-BDRS) while maintaining its basic structure and to evaluate the reliability and validity of this measure in the Korean population.

All relevant demographic and clinical characteristics are presented in Table 1. The internal consistency of the K-BDRS was assessed using Cronbach’s α-coefficient, which was calculated as 0.866. The internal consistency of the depressive subscale of the K-BDRS (Items 1–15) was 0.893 but the mixed subscale (Items 16–20) showed a lower Cronbach’s α-coefficient of 0.756. Table 2 shows the correlations of each item with the total K-BDRS score using a criterion of 0.30 as an acceptable corrected item-total correlation [26]; all but four of the items performed adequately. This lower correlation followed an expected trend for the mixed subscale symptoms such that they correlated less strongly with the total K-BDRS score.

The inter-rater reliability was analyzed in a subsample of patients (n = 20) who were interviewed by two different raters. The inter-rater reliability was high for both the total K-BDRS score (ICC = 0.954, 95 % confidence interval [CI] = 0.884–0.982, p < 0.001) and the individual items (ICC > 0.70). The test-retest reliability was also analyzed in a subsample of patients (n = 35) who had no clinical change between two testing sessions, and was high for both the total K-BDRS score (ICC = 0.950, 95 % CI = 0.901–0.975, p < 0.001) and the individual items as well (ICC > 0.70; Table 3).

The K-BDRS was factor analyzed using the data of all 141 patients to determine the optimal number of factors that described the scale. An unweighted least squares factor analysis was followed by oblique rotations of 2–4 factors. Prior to the rotation, the eigenvalues for the first four factors were 6.67, 3.36, 1.36, and 1.14 with corresponding percentages of variance accounted for of 33.37, 16.79, 6.79, and 5.70 %. The three-factor rotation was determined to provide the most useful description of the data and these three factors were labeled as psychological depressive symptoms (Factor 1), somatic depressive symptoms (Factor 2), and mixed symptoms (Factor 3). Table 4 shows the factor loadings and Table 5 shows the correlations among the total scores of the K-BDRS and the other factor scores.

To assess the concurrent validity of the K-BDRS, the total scores and factor scores on the K-BDRS were compared with those on the HAMD, MADRS, and YMRS. The K-BDRS total scores were strongly correlated with the HAMD scores (r = 0.788, p < 0.001) and the MADRS scores (r = 0.877, p <0.001), the depression subscale scores of the K-BDRS (Items 1–15) were strongly correlated with the HAMD scores (r = 0.888, p < 0.001) and the MADRS scores (r = 0.759, p < 0.001), and the mixed subscale scores (Item 16–20) were strongly correlated with the YMRS scores (r = 0.611, p < 0.001). The correlations between the K-BDRS factor scores and the total scores on the HAMD, MADRS and YMRS are presented in Table 6.

The present findings suggest that the K-BDRS has good psychometric properties and may be a reliable and valid tool for measurement of the severity of depressive symptoms in Korean patients with bipolar disorder. Compared with the original BDRS (α = 0.917), the K-BDRS has a somewhat low internal consistency (α = 0.866), but the internal consistency is consistent with other translated versions from Iran (α = 0.81) [27], Turkey (α = 0.786) [28], and Spain (α = 0.870) [29].

In this study, the concurrent validity of the K-BDRS was demonstrated based on its correlations with the HAMD, MADRS, and YMRS. The strong correlation coefficients of the K-BDRS with the other scales indicate that the K-BDRS accurately assesses depressive symptoms. Additionally, the mixed subscale score of the K-BDRS (Item 16–20) exhibited a strong correlation with the YMRS which indicates that the K-BDRS accurately assesses the mixed symptoms that frequently present in patients with bipolar depression. This is an important aspect of the K-BDRS because the rating of mixed features in bipolar depression is an area of great clinical significance.

The factor analysis of the K-BDRS revealed a three-factor solution that provided the best account of the present data. Although the factor loadings are different, the three-factor solution observed in the present study resembled the three-factor structure originally reported by previous studies [14]. As expected, the depressive subscales of the K-BDRS loaded onto psychological depressive symptoms (e.g., suicidal ideation, guilt, worthlessness, and helplessness) and somatic depressive symptoms (e.g., reduced activity and concentration and appetite disturbance). However, several items in the present study differed from those of the original study by Berk et al. [14]. In that study, anhedonia and flattened affect loaded onto psychological depressive symptoms but the present study found that anhedonia and flattened affect loaded onto somatic depressive symptoms. Because these symptoms are strongly related to reduced motivation and energy, it may be rational to assume that these features can be grouped under the somatic depressive symptoms factor.

Both of the depressive symptoms factors were strongly correlated with each other but weakly correlated with the mixed symptoms factor. All of the mixed subscales of the K-BDRS loaded together under the mixed symptoms factor, with the exception of lability. Lability is common in bipolar depression [10, 11] and similarly loaded onto both the psychological depressive and mixed symptoms factors. Psychotic symptoms are also included in the mixed symptoms factors, as demonstrated by Berk et al. [14]. This indicates that psychotic symptoms have a greater association with mixed features than with depressive symptoms. Our findings that the internal structure of the K-BDRS appears to be sensitive to mixed symptoms of bipolar depression are consistent with the previous reports [30].

The development of a Korean version of the K-BDRS is significant in several respects. The results of the present study confirmed the psychometric properties of the K-BDRS, which will enable both clinicians and researchers to use this instrument in the field to assess the severity and treatment responses of depressive symptoms in patients with bipolar disorder. Most importantly, the K-BDRS includes items that rate the mixed and atypical features of bipolar disorder, which may be overlooked by conventional depressive scales such as the MADRS and HAMD. Previous reports comparing unipolar and bipolar depression with the BDRS and the MADRS showed MADRS scores were unable to distinguish unipolar from bipolar depression. However, BDRS scores were significantly marked for the mixed subscale, suggesting that presence of mixed features during a depressive episode is in favour of bipolar depression [31].

The limitations of the present study were that the numbers of patients in the subgroups for the inter-rater and test-retest reliability ratings may have been insufficient. Additionally, the present study was cross-sectional and, therefore, it is difficult to determine whether the K-BDRS would be useful for the detection of changes in illness severity following treatment.

The K-BDRS possesses good psychometric properties and is likely a valid and reliable tool for the assessment of depressive symptoms in patients with bipolar disorder. However, further studies are needed to evaluate the K-BDRS more fully. These studies should assess the applicability of the K-BDRS for various bipolar disorder subgroups as well as evaluate the scale’s utility for differentiating bipolar disorder from unipolar depression.

BDRS, the Bipolar Depression Rating Scale; CI, confidence interval; DSM-IV-TR, the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision; HAND, the 17-item Hamilton Depression Scale; ICC, intraclass correlation coefficients; K-BDRS, Korean version of the BDRS; MADRS, the Montgomery and Asberg Depression Scale; MINI, the Mini International Neuropsychiatric Interview; YMRS, the Young Mania Rating Scale