DSM-5-defined ‘mixed features’ and Benazzi's mixed depression: Which is practically useful to discriminate bipolar disorder from unipolar depression in patients with depression?

Authors


Abstract

Aims

Irritability, psychomotor agitation, and distractibility in a major depressive episode (MDE) should not be counted as manic/hypomanic symptoms of DSM-5-defined mixed features; however, this remains controversial. The practical usefulness of this definition in discriminating bipolar disorder (BP) from major depressive disorder (MDD) in patients with depression was compared with that of Benazzi's mixed depression, which includes these symptoms.

Methods

The prevalence of both definitions of mixed depression in 217 patients with MDE (57 bipolar II disorder, 35 BP not otherwise specified, and 125 MDD cases), and their operating characteristics regarding BP diagnosis were compared.

Results

The prevalence of both Benazzi's mixed depression and DSM-5-defined mixed features was significantly higher in patients with BP than it was in patients with MDD, with the latter being quite low (62.0% vs 12.8% [P < 0.0001], and 7.6% vs 0% [P < 0.0021], respectively). The area under the receiver operating curve for BP diagnosis according to the number of all manic/hypomanic symptoms was numerically larger than that according to the number of manic/hypomanic symptoms excluding the above-mentioned three symptoms (0.798; 95% confidence interval, 0.736–0.859 vs 0.722; 95% confidence interval, 0.654–0.790). The sensitivity/specificity of DSM-5-defined mixed features and Benazzi's mixed depression for BP diagnosis were 5.1%/100% and 55.1%/87.2%, respectively.

Conclusions

DSM-5-defined mixed features were too restrictive to discriminate BP from MDD in patients with depression compared with Benazzi's definition. To confirm this finding, studies that include patients with BP-I and using tools to assess manic/hypomanic symptoms during MDE are necessary.

The mixed states are clinical pictures that can be broadly defined as the simultaneous presence of depressive and manic symptoms. Although corresponding concepts have been described since classical antiquity, they were conceptualized systematically by Kraepelin.[1, 2] In conceiving the mixed states, Kraepelin started from two polarities, namely depression (weakness) or excitement of the three domains of psychic life (mood, thinking, and volition) during an affective episode. In his view, these three domains do not necessarily stay in the same polarity, and the presence of one of the three domains in opposition to the other two is sufficient for the establishment of a mixed state diagnosis (i.e., six out of eight combinations). In his eighth textbook, Kraepelin stated that the mixed states represent the major part of the course of manic–depressive illness.[3] However, beginning with the third decade of the 20th century, the interest in mixed states diminished markedly.[1, 2] Kraepelin's broad concept of mixed states was not incorporated into major psychiatric diagnostic systems, such as the DSM-IV-TR.[4] In the DSM-IV-TR, a mixed episode is defined quite narrowly as the co-occurrence of DSM-IV-TR-defined mania and depression for ≥1 week in the context of bipolar I disorder (BP-I).

In the past 30 years, a rebirth of interest in mixed states has led to the investigation of the predominantly manic side of mixed states (mixed mania), which consists of a manic episode (ME) with depressive symptoms under the threshold of a major depressive episode (MDE).[5, 6] Studies of the predominantly depressive side of mixed states (mixed depression), which consists of MDE with manic/hypomanic symptoms under the threshold of an ME or hypomanic episode (HME), started relatively later than those of mixed mania.[7-9] The definitions of mixed depression, as well as mixed mania, vary among studies.[10, 11] One of the representative definitions of mixed depression is MDE with three or more concurrent DSM-IV-defined manic/hypomanic symptoms for ≥1 week, which was proposed by an Italian psychiatrist, Franco Benazzi.[12, 13] In contrast to mixed mania, which has been studied only in patients with bipolar disorder (BP) based on its definition, mixed depression is usually studied in the context of a full spectrum of mood disorders, thus including unipolar depression. It has recently attracted attention as a bipolar-suggestive feature in MDE,[12, 14] a predictive factor of future progression from major depressive disorder (MDD) to BP,[15] and one of the most significant risk factors of suicide attempts in patients with mood disorders.[16, 17] Because BP, particularly bipolar II disorder (BP-II), tends to be misdiagnosed and to be treated as MDD, it is crucial to capture the state of mixed depression for the early diagnosis and adequate management of BP.

The narrowly defined ‘mixed episode’ of DSM-IV-TR was removed in DSM-5. Instead, this edition captured three or more symptoms of the opposite pole using a ‘with mixed features’ specifier to be applied to ME in BP-I; HME in BP-I and II; and MDE in BP-I, BP-II and MDD.[18] This revision can be appreciated, as it incorporated the significance of the broad concept of mixed states. However, it excluded irritability, psychomotor agitation, and distractibility from the manic/hypomanic symptoms of MDE with mixed features because they can occur in both mania/hypomania and depression. Some concerns regarding this stipulation have been expressed because these symptoms have been reported as being quite prevailing in patients with mixed states and have even been emphasized as the core symptoms of this condition.[19, 20]

Previously, we recorded manic/hypomanic symptoms during MDE at routine psychiatric practice, and reported that Benazzi's mixed depression, which includes all manic/hypomanic symptoms in MDE, was a powerful bipolar suggestive feature in MDE.[14] In this study, we compared the clinical relevance of DSM-5-defined mixed features to that of Benazzi's mixed depression in terms of discriminating BP from MDD in patients with depression.

Methods

Study setting and subjects

The work setting of the first author (M.T.) is an outpatient service and consultation–liaison psychiatry service in a general hospital (Kouseiren Takaoka Hospital) and that of the second author (T.O.) is an outpatient private psychiatry practice (J Clinic). These settings are generally the first- or second-line of treatment for depression, that is, non-tertiary care, in the Japanese medical system. The sample included consecutive patients suffering from MDE because of BP-II, BP not otherwise specified (BP-NOS), or MDD, as defined in the DSM-IV-TR, who first visited our institutions from September 2010 to March 2013 and who could be observed for ≥4 weeks, to enhance diagnostic reliability. Because only a limited number of patients with BP-I visited our study setting, these patients were excluded from the present study. Patients with severe mental retardation, dementia, pervasive developmental disorders, schizophrenia and related psychotic disorders, borderline personality disorder, and serious physical illnesses were also excluded, to avoid confounding the evaluation of clinical pictures.

Diagnostic procedure

Diagnoses were established by M.T. or T.O. (who have 24 and 17 years of clinical experience studying and treating mood disorders, respectively) according to the DSM-IV-TR and based on extensive clinical interviews with the patients and their significant others as much as possible; this was supplemented by all available information provided by referring clinicians and past records. A change in the diagnosis could be made through further interviews regarding history, information from other institutions, or observations of symptoms and signs during the prospective follow-up period. According to the DSM-IV-TR, patients who exhibited hypomania only during periods of treatment with antidepressants were regarded as having MDD, and not BP. Patients were diagnosed as having BP-NOS if they had experienced at least one MDE in addition to full hypomanic symptoms that lasted <4 days.

Data collection

As mentioned elsewhere, we recorded manic/hypomanic symptoms during MDE at routine psychiatric examinations. Here, manic/hypomanic symptoms were evaluated at the first visit of each patient, irrespective of ongoing antidepressant treatment, and during the prospective follow-up period thereafter when antidepressants were not prescribed. Benazzi's mixed depression was defined according to descriptions from previous studies.[12, 13] In brief, a patient was diagnosed with mixed depression if three or more DSM-IV-defined manic/hypomanic symptoms occurred during MDE, continued for ≥1 week, and were present during the interview. The definition of ‘MDE with mixed features’ conformed to the criteria of the DSM-5, with the exception of the duration of the co-existence of manic/hypomanic symptoms. Although the DSM-5's criteria require that manic/hypomanic symptoms are present during the majority of days of the MDE, we shortened the duration to ≥1 week similar to that in Benazzi's definition. According to the criteria of the DSM-5, a diagnosis of BP-II should be established in patients without prior ME/HME who meet the criteria for MDE and HME simultaneously; therefore, the number of patients who fell into this category was counted separately.

Because all data were obtained in our routine psychiatric practice, patients provided oral informed consent prior to individual examinations, and the ethical review board of each institution decided that written informed consent was unnecessary. The personal information was strictly protected, and a notice about the purpose and protocol of this study was posted at each institution in accordance with an ethical indicator about the clinical study of the Ministry of Health, Labour and Welfare of Japan.

Statistics

The prevalence of each manic/hypomanic symptom, DSM-5-defined mixed features, and Benazzi's mixed depression in patients with BP-II/BP-NOS and in those with MDD was compared using the χ2 or Fisher's exact probability test. To examine the diagnostic significance of irritability, psychomotor agitation, and distractibility, which were excluded from the manic/hypomanic symptoms of DSM-5-defined MDE with mixed features, two receiver–operator curves (ROC) of BP diagnosis according to the number of manic/hypomanic symptoms, namely one including all manic/hypomanic symptoms and one excluding the above-mentioned three symptoms, were generated and the areas under the curve (AUC) were compared with each other. An optimal cut-off point (giving equal weight to sensitivity/specificity) was obtained as a point on an ROC that was closest to the upper left corner of each graph. The sensitivity/specificity for BP diagnosis according to Benazzi's mixed depression and that of DSM-5-defined mixed features were calculated based on the cut-off point of ≥3 manic/hypomanic symptoms in the former and the latter ROC, respectively. Patients who met the criteria for MDE and HME simultaneously were excluded from the ROC analysis because this population should be diagnosed directly as having BP-II.

Excel Statistics 2010 for Windows (Social Survey Research Information, Tokyo, Japan) was used for analyses, and supplemented by ROCKIT 1.1B2 (downloaded from http://metz-roc.uchicago.edu/) to calculate the 95% confidence interval (CI) of the AUC. The P-values were two-tailed, and the alpha level was set at 0.05.

Results

Of the 266 patients who were diagnosed with MDE due to BP-II/BP-NOS or MDD during the study period, 217 (57 [26.3%] BP-II, 35 [16.1%] BP-NOS, and 125 [57.6%] MDD) were observed for ≥4 weeks and were evaluable. The median (range) age at first visit was 40 (16–88) years, and 122 (56.2%) patients were female. A total of 85 (39.2%) patients had been prescribed antidepressants at the first visit to our institutions.

The prevalence of manic/hypomanic symptoms, Benazzi's mixed depression, and DSM-5-defined mixed features is summarized in Table 1. Within the whole sample, psychomotor agitation, flight of ideas/racing thoughts, irritability, distractibility, and talkativeness were frequently observed (>20% of the sample), in that order, whereas inflated self-esteem and elevated mood were each observed in just one patient (0.5%), respectively. Flight of ideas/racing thoughts, irritability, distractibility, talkativeness, increased goal-directed activity, and risky activity were observed more frequently in patients with BP than in those with MDD. The proportion of the patients with BP who met the criteria of Benazzi's mixed depression was significantly higher compared with the patients with MDD who met those criteria (62.0% vs 12.8%, P < 0.0001). In contrast, only seven patients with BP met the criteria for DSM-5-defined mixed features, although this proportion was significantly higher than that observed in patients with MDD (7.6% vs 0%, P < 0.0021). Fourteen patients exhibited the co-occurrence of MDE and HME. The prevalence of manic/hypomanic symptoms in these patients was as follows: irritability, 100%; flight of ideas/racing thoughts, 100%; distractibility, 92.9%; psychomotor agitation, 85.7%; talkativeness, 71.4%; risky activity, 35.7%; increased goal-directed activity, 14.3%; decreased need for sleep, 7.1%; inflated self-esteem, 0%; and elevated mood, 0%. The rates of patients with BP who met the criteria for Benazzi's mixed depression and DSM-5-defined mixed features were significantly higher compared with the patients with MDD who met those criteria (46.7% vs 12.8%, P < 0.0001; 4.3% vs 0%, P = 0.0208; respectively), even though patients who exhibited co-occurrence of MDE and HME were excluded. Naturally, all patients with co-occurrence of MDE and HME also met the criteria of Benazzi's mixed depression, whereas 11 of them did not meet the criteria for DSM-5-defined mixed features because of its exclusion rule. For informational purposes, the dispositions of Benazzi's mixed depression, DSM-5-defined mixed features, and co-occurrence of MDE and HME within the entire sample are shown in Figure 1.

Table 1. Prevalence of manic/hypomanic symptoms, Benazzi's mixed depression, and DSM-5's mixed features during major depressive episodes
 TotalMDDBP-II/BP-NOSStatistical differencesa
(n = 217) (n = 125)(n = 92)MDD vs BP-II/BP-NOS (P-value)
n (%) 
  1. aCalculated by χ2 test or Fischer's exact probability test, as appropriate.
  2. BP-II, bipolar II disorder; BP-NOS, bipolar disorder not otherwise specified; HME, hypomanic episode; MDD, major depressive disorder; MDE, major depressive episode.
Manic/hypomanic symptoms    
Psychomotor agitation116 (53.5)61 (48.8)55 (59.8)0.1090
Flights of ideas/racing thoughts79 (36.4)25 (20.0)54 (58.7)<0.0001
Irritability78 (35.9)19 (15.2)59 (64.1)<0.0001
Distractibility55 (25.3)18 (14.4)37 (40.2)<0.0001
Talkativeness48 (22.1)19 (15.2)29 (31.5)0.0042
Increased goal-directed activity25 (11.5)9 (7.2)16 (17.9)0.0201
Risky activity18 (8.3)2 (1.6)16 (17.4)0.0001
Decreased need for sleep11 (5.1)3 (2.4)8 (8.7)0.0570
Inflated self-esteem1 (0.5)0 (0.0)1 (1.1)0.4240
Elevated mood1 (0.5)1 (0.8)0 (0.0)1.0000
Benazzi's mixed depression73 (33.6)16 (12.8)57 (62.0)<0.0001
Excluding co-occurrence of MDE and HME59 (27.2)16 (12.8)43 (46.7)<0.0001
DSM-5's mixed features7 (3.2)0 (0.0)7 (7.6)0.0021
Excluding co-occurrence of MDE and HME4 (1.8)0 (0.0)4 (4.3)0.0208
Co-occurrence of MDE and HME14 (6.5) 14 (15.2) 
Figure 1.

Dispositions of Benazzi's mixed depression, DSM-5's mixed features, and co-occurrence of major depressive episode (MDE) and hypomanic episode (HME) in the entire sample. †Among seven patients who met the criteria of MDE with mixed features, three fulfilled MDE and HME simultaneously. MDD, major depressive disorder; BP-II, bipolar II disorder; BP-NOS, bipolar disorder not otherwise specified.

The ROC for BP diagnosis (excluding cases with co-occurrence of MDE and HME) according to the number of all manic/hypomanic symptoms and that according to the number of manic/hypomanic symptoms excluding irritability, psychomotor agitation, and distractibility are shown in Figure 2. The AUC of the former was numerically larger than that of the latter (0.798, 95%CI, 0.736–0.859 vs 0.722, 95%CI, 0.654–0.790). The optimal cut-off point of the former was ≥2 manic/hypomanic symptoms and that of the latter was ≥1 manic/hypomanic symptom, and sensitivity/specificity for BP diagnosis according to these cut-off points was 88.5%/61.6% and 75.6%/59.2%, respectively. The sensitivity/specificity for BP diagnosis according to Benazzi's mixed depression and DSM-5-defined mixed features were 55.1%/87.2% and 5.1%/100%, respectively.

Figure 2.

Receiver–operator curves of bipolar disorder diagnosis according to the number of manic/hypomanic symptoms, (a) including and (b) excluding irritability, distractibility, and psychomotor agitation during a major depressive episode. The number attached to the symbols indicates the number of manic/hypomanic symptoms. The arrows indicate the optimal cut-off. †Cut-off point of Benazzi's mixed depression. ‡Cut-off point of DSM-5's mixed features. AUC, area under the curve; CI, confidence interval.

Discussion

The present study showed that the prevalence of both Benazzi's mixed depression and DSM-5-defined mixed features was significantly higher in patients with BP than it was in patients with MDD. Moreover, the prevalence of the latter was quite low. Furthermore, the prevalence of DSM-5-defined mixed features in the BP sample was even lower than that of co-occurrence of MDE and HME. This appears to be somewhat strange because the co-occurrence of MDE and HME is comparable with DSM-IV-TR-defined mixed episode, which has been criticized as being too narrowly defined,[11, 21] and DSM-5-defined mixed features were framed to capture a wider range of mixed states than the DSM-IV-TR-defined mixed episode. In contrast, Benazzi's mixed depression accounted for 62.0% of MDE caused by BP, and all of the patients with co-occurrence of MDE and HME also met the criteria of Benazzi's mixed depression, by definition. This is natural when considering the continuity within mixed states (depressive, full, and manic). A comparison between the AUC of the ROC curves for BP diagnosis (excluding cases with co-occurrence of MDE and HME) according to the number of all manic/hypomanic symptoms and that according to the number of manic/hypomanic symptoms, excluding irritability, psychomotor agitation, and distractibility, revealed that the former had a numerically superior diagnostic performance over the latter. Although DSM-5-defined mixed features showed very high specificity (100%) for BP diagnosis, their sensitivity was quite low (5.1%). Benazzi's mixed depression showed a tenfold higher sensitivity (55.1%) for BP diagnosis compared with that of DSM-5-defined mixed features, while maintaining good specificity (87.2%). These results suggest that the definition of Benazzi's mixed depression is superior to that of DSM-5-defined mixed features for the practical purpose of discriminating BP from MDD in patients with depression. The optimal cut-off points obtained from our ROC analysis show considerably lower specificity (more false positives) than the original definition of Benazzi's mixed depression and DSM-5-defined mixed features. Therefore, the original definition should be used in clinical practice.

The low prevalence of DSM-5-defined mixed features observed in the present study clearly stemmed from its stipulation of excluding irritability, distractibility, and psychomotor agitation from the manic/hypomanic symptoms to be assessed. As shown here, irritability, distractibility, and psychomotor agitation have been reported constantly as prevailing manic/hypomanic symptoms during MDE, together with flight of ideas/racing thoughts.[13, 21] Benazzi[22] and Judd et al.[23] reported that irritability/anger in unipolar MDE was associated with a higher rate of BP in relatives. Although an opposing result has been reported,[24] irritability/anger may independently suggest bipolarity in MDE. In the present study, psychomotor agitation was the manic/hypomanic symptom that was observed most frequently in both patients with BP and those with MDD. Although its prevalence in the patients with BP was numerically higher than that recorded in patients with MDD, no statistical significance was found regarding this difference. As reported previously,[25, 26] psychomotor agitation by itself may not be enough to identify a specific depression subtype. Because mixed states are realized as the simultaneous presence of depressive and manic symptoms, there exists a view to consider an overlap of core manic symptoms (e.g., elevated mood or increased self-esteem) and core depressive symptoms (e.g., depressive mood, loss of interests, feelings of guilt, or suicidal ideation), a prototype of mixed states. However, as shown in the present study, the observation of elevated mood or increased self-esteem during the majority of days of MDE is rare in clinical practice. Rather, the mixed states may be better understood as a spectrum of states that predominantly show poorly directed hyperactivities in mood, thinking, and volition (i.e., irritability, psychomotor agitation, distractibility, and flight of ideas/racing thoughts), backed by the gradation from typical depressive symptoms to typical manic symptoms. Bertschy et al.[27] conducted a principal component analysis of symptoms and signs covering the full spectrum of major mood disorders, from pure (non-mixed) depression to pure (non-mixed) mania and including all subtypes of mixed states (depressive, full, and manic). Those authors obtained the following three components: a ‘mania/depression’ component that contrasted typical depressive symptoms on one pole to typical manic symptoms of euphoria and activation on the other pole; a ‘dysphoria’ component that had strong positive loading for irritability and was also associated with distractibility and motor agitation; and a third component that included sleep disturbances. They showed that scores for the dysphoria component were highest among patients with full mixed states, and decreased towards both patients with pure depression and those with pure mania.

We should mention a precaution in applying Benazzi's concept. Although Benazzi's mixed depression is suggestive of bipolarity, it doesn't directly mean that a patient with Benazzi's mixed depression is suffering from BP. Many patients with MDD showed Benazzi's mixed depression in the present study. To avoid overdiagnosis of BP, strict evaluation of the past (or future) ME/HME is essential.

The limitations of our study should be considered. First, reflecting the typical clinical practice, we did not use any tool to assess manic/hypomanic symptoms during MDE. Although the evaluations of manic/hypomanic symptoms during MDE were made by trained psychiatrists, and the prevalence of manic/hypomanic symptoms in our study was generally comparable with that reported by previous studies,[13, 21] mild hypomanic symptoms during MDE may have been missed. The use of adequate tools to assess manic/hypomanic symptoms during MDE would have maximized the reliability of the results of this study.[13] Second, we did not include patients with BP-I because of our study setting. Studies that include patients with BP-I require descriptions of the entire picture of mixed states in mood disorders.

In conclusion, the results of this study suggest that the definition of DSM-5's mixed features is too restrictive to discriminate BP from MDD in patients with depression compared with that of Benazzi's mixed depression. To confirm this finding, studies that include patients with BP-I and using tools to assess manic/hypomanic symptoms during MDE are necessary.

Acknowledgments

No financial support was provided for this study. Dr Takeshima has served as an advisory or a speaker for the following companies: Eli Lilly, Otsuka, GlaxoSmithKline, Meiji Seika Pharma. Dr Oka has served as an advisory or a speaker for the following companies: Eli Lilly, Otsuka, GlaxoSmithKline, Meiji Seika Pharma, and Yoshitomi. We thank Mrs Kanae Nagae and Miss Yuka Yamagishi for their secretarial support.

Ancillary