Comparison of clinical characteristics of bipolar and depressive disorders in Korean clinical sample of youth: a retrospective chart review
- First Online:
- Cite this article as:
- Shon, S., Joo, Y., Park, J. et al. Eur Child Adolesc Psychiatry (2014) 23: 307. doi:10.1007/s00787-013-0461-3
- 306 Views
The purpose of this study was to compare the clinical characteristics of bipolar disorder I, II (BD I and II) and not otherwise specified (BD NOS) to those of major depressive disorder (MDD) in a clinical sample of Korean children and adolescents. This study was a cross-sectional review of longitudinal observational data. Two psychiatrists retrospectively reviewed the medical records of 198 children and adolescents (age 6–18) that were diagnosed as having bipolar or depressive disorders from March 2010 to February 2012 at Department of Psychiatry of Asan Medical Center, Seoul, Korea. Every subject’s diagnoses were reviewed and confirmed. BD I, II and MDD were assessed according to the Diagnostic and Statistical Manual-IV criteria. BD NOS was defined based on the criteria for the Course and Outcome of Bipolar Youth study. Comparisons were made in demographic information, clinical characteristics, family history, and psychiatric comorbidities at baseline and during observation. Among 198 subjects, 20 (10.1 %) subjects were diagnosed as having BD I, 10 (5.1 %) as BD II, 25 (12.6 %) as BD NOS and 143 (73.7 %) as MDD. BD depression was associated with mood change while taking an antidepressant, familial bipolarity, aggressive behaviors, and atypical features. Comorbid obsessive–compulsive disorder tended to be higher in BD NOS than in MDD. Presence of psychosocial stressors was more common in MDD than in BD depression. In children and adolescents, bipolar depression is distinct from unipolar depression in family history, comorbidity, and clinical characteristics.
KeywordsAdolescentBipolar disorderChildrenClinical characteristicsDepression
Bipolar disorder (BD) is a recurrent mood illness and its symptoms are associated with significant functional impairment, and the early recognition of bipolar symptoms is important . Many bipolar disorders initially present as depression, so diagnosis only occurs after following the person over time, allowing for manifestation and identification of mania or hypomania . Although differentiating between unipolar and bipolar forms of depression poses clinical and therapeutic challenges, there has been relatively less attention paid to the phenomenology of bipolar depression and the predictive factors of bipolar development in depressive episode . In subjects with BD, misdiagnosis and delayed treatment possibly cause poorer outcome , and antidepressant treatments could be related to manic or hypomanic symptoms [5, 6].
In previous studies on the phenomenology of depression in major depressive disorder (MDD) and BD [7–9], atypical features, psychotic symptoms, psychomotor retardation, shorter depressive episodes, higher number of depressive recurrences, family history of mood disorders, comorbidity with substance abuse, and earlier age at onset were all more associated with BD than with MDD. Bipolar youth had higher rates of irritability, comorbid attention-deficit/hyperactivity (ADHD), oppositional defiant (ODD), and obsessive–compulsive disorders (OCD), agoraphobia, and alcohol abuse [10, 11]. Juvenile MDD, especially with a family history of bipolar disorder, is strongly associated with the development of later mania . Psychomotor retardation, mood congruent psychotic features and pharmacologically coincident mania also can predict diagnostic conversion to mania [13, 14]. However, except for one , these studies did not include subjects with subthreshold bipolar symptoms or BD not otherwise specified (BD NOS), which is the most common manifestation of BD in children and adolescents based on epidemiological data [15, 16] as well as clinical cohorts .
The results of a recent meta-analysis of epidemiologic studies of pediatric bipolar disorder showed that the overall prevalence was 1.8 %  and that prevalence of BD NOS was between 3.9 and 5.7 % [18, 19]. The meta-analysis also demonstrated that the prevalence of bipolar disorder among youth is highly dependent on diagnostic criteria . Diagnostic and Statistical Manual of Mental Disorders Fourth Edition (DSM-IV) criteria are known to fail to include milder but clinically significant syndromes. Thus operational diagnostic criteria such as Course and Outcome of Bipolar Youth study (COBY)  have been suggested. However, the prevalence of these operational diagnoses as well as DSM-IV diagnosis of bipolar disorder has not been studied in Asian children and adolescents .
The purpose of this study was to investigate the prevalence of BD I and II and MDD based on DSM-IV criteria, and BD NOS defined by the COBY criteria in a clinical sample of Korean children and adolescents. We also aimed to compare the clinical characteristics of BD I, II and NOS to those of MDD based on youths presenting to the same clinical setting.
This study was a cross-sectional review of longitudinal observational data. We retrospectively reviewed the medical records of 198 children and adolescents (age 6–18) who were diagnosed as having bipolar or depressive disorders from March 2010 to February 2012 at the Department of Psychiatry of Asan Medical Center, Seoul, Korea. All subjects had to meet the following inclusion criteria: (1) age between 6 and 18 years and (2) meet the diagnostic criteria for bipolar or depressive disorders according to DSM-IV criteria. Subjects were excluded from the study if they had one or more of the following exclusion criteria: (a) intelligence quotient (IQ) lower than 80 on the Korean Educational Developmental Institute’s Wechsler Intelligence Scale for Children (KEDI-WISC) ; (b) past and/or current history of schizophrenia, other psychotic, organic mental, or pervasive developmental disorder; (c) presence of seizure or other neurological disorders, and (d) sensory impairments. The study was approved by the Institutional Review Board of Asan Medical Center.
Assessment and measures
Two psychiatrists (H.W.K & S.H.S.) retrospectively and independently reviewed and confirmed psychiatric diagnoses of subjects based on the medical records. Then we obtained the following data for each subject: demographic information, clinical characteristics, family history, and psychiatric comorbidities at baseline and during observation. These data were already described in the medical record by a board-certified child psychiatrist (H.W.K.) based on interviews with both children and their parents.
Diagnosis was confirmed retrospectively based on the symptoms. Symptoms of depressive episode and screening questions for manic were listed in the chart. Bipolar I, II, and major depressive disorders were assessed according to the DSM-IV criteria . BD NOS was defined based on the criteria for the COBY study [20, 23]. The COBY study operationalized criteria as follows: (a) elated mood plus two associated B Criteria symptoms of mania or irritable mood plus three B Criteria symptoms; (b) clear change of subject’s level of functioning; (c) symptoms must be evident for a minimum of 4 h in a day on a minimum of four separate days. In addition, worsening of any pre-existing symptoms during episode is required. We classified the subjects into three groups as follows: (1) BD I and II, (2) BD NOS, and (3) MDD. DSM-IV and the DSM-5 include cyclothymic disorder as another bipolar diagnosis; however, the COBY criteria lump cyclothymic disorder together with bipolar NOS. We elected to use the COBY criteria as a research definition to have our definitions build a crosswalk to existing research. The inter-rater reliability for the mood disorders was found to be good (κ = 0.75).
Demographic characteristics included the child’s gender, age, age at onset, IQ, and socioeconomic status (high, middle, and low). Age at onset of bipolar disorder was defined as the age at onset of the first DSM-IV mood episode or an episode fulfilling the COBY’s criteria for BD NOS.
DSM-IV criteria were used to assess specific features of the mood episode. Patterns of cycling symptoms of mood episodes remain a major point of controversy [24, 25]. In this study, definitions of any mood cycling were as follows: rapid, four or more episodes per year; ultra rapid, four or more episodes per month; and ultradian, more than one episode per day in four or more days per week .
Suicidal attempts, non-suicidal self-injury (NSSI), and numbers of these behaviors were assessed retrospectively. Mood changes during use of antidepressants and precipitating psychosocial factors were also investigated. Number of suicidal attempts and NSSI were categorized as follows: (a) 0; (b) 1–5; and (c) >5 events.
We also retrospectively scored the behavioral problems (aggressive behavior, delinquent behaviors, school refusal, running away from home) during mood episodes based on the conduct disorder supplement of the Kiddie Schedule for Affective Disorders and Schizophrenia Present and Lifetime Version (K-SADS-PL) .
All psychiatric family histories were coded from subject’s medical records, retrospectively. We categorized family history as follows: (a) BD; (b) MDD; (c) other (e.g., schizophrenia, schizoaffective disorder, delusional disorder, ADHD) psychiatric disorder and (d) negative family history. We also evaluated if the subjects had first degree relative or multiple relatives with psychiatric disorder.
Comorbid psychiatric disorders were determined based on DSM–IV and internet addiction was diagnosed based on Young’s criteria  (κ = 0.63–0.77).
The Chi squared or Fisher’s exact test was used for categorical variables, univariate ANOVA for dimensional variables and Kruskal–Wallis test for ordinal variables. For comparisons of the clinical characteristics, family history and psychiatric comorbidities among the groups, adjusted p-values for multiple comparisons were used (0.05/number of comparisons). When the comparison was significant, a post hoc Bonferroni test was performed to evaluate the differences between the groups. In post hoc analysis, a significance level of 0.05 was accepted. Statistical analyses were performed using PASW (version 18.0, SPSS Inc., Chicago, Illinois, USA), and all comparisons were two-tailed.
From March 2010 to February 2012, a total of 1,996 children and adolescents visited the Department of Psychiatry of Asan Medical Center. Among these 1,996 children and adolescents, 198 (9.9 %) were diagnosed as having bipolar or depressive disorders. Within those 198 subjects, 20 (10.1 %) subjects were diagnosed as having bipolar I disorder (BD I), 10 (5.1 %) as bipolar II disorder (BD II), 25 (12.6 %) as bipolar disorder NOS (BD NOS) and 143 (73.7 %) as major depressive disorder (MDD). Mean follow-up time for these subjects was 479.0 ± 596.6 days. All three subgroups did not differ significantly in the follow-up time (p = 0.146). At the time of assessment, 57.6 % of the subjects were being followed up.
The prevalence of BD I and II and MDD was 1.0, 0.5 and 7.2 %, respectively, based on DSM-IV criteria, and the prevalence of BD NOS was 1.3 % based on the COBY criteria in the clinical population. One hundred and sixty-three subjects were diagnosed with MDD at baseline. Diagnostic conversion from MDD into BD was founded in 20 subjects because of newly developed manic or hypomanic episodes. In 13 subjects among those 20, manic or hypomanic episode was related to antidepressant treatment.
Demographic information in BD I and II, NOS, and MDD
BD I and II (n = 30)
BD NOS (n = 25)
MDD (n = 143)
F or χ2
Age, years, mean (SD)
Age at onset, years, mean (SD)
FSIQ, mean (SD)
Boys, n (%)
SES, n (%)
Low (medical aid)
Clinical characteristics of depressive episode among BD I and II, NOS, and MDD
BD I and II (n = 30)
BD NOS (n = 25)
MDD (n = 143)
F or χ2
AD-concurrent mood change
Precipitating psychosocial factors
Related behavioral problems
Mood change concurrent with antidepressants was significantly prevalent in BD I and II (33.3 %) and BD NOS (20.0 %) groups than in MDD groups (3.5 %) (p < 0.001). There were no significant differences in numbers of suicidal attempts and NSSI among three groups (p = 0.253, p = 0.078, respectively).
Among related behavioral problems in depressive episode, the rates of delinquent behaviors, school refusal, and running away were not different among three groups (p = 0.796, p = 0.383, p = 0.663, respectively). Aggressive behaviors were different among three groups (p = 0.001) and significantly higher in subjects with BD I and II and BD NOS than in subjects with MDD. Precipitating psychosocial factors leading to depressive symptoms were significantly common in subjects with MDD (73.4 %) than in those with BD NOS (40.0 %) (p = 0.002). School refusal at any point in the course of mood disorder (including mania) was significantly higher in subjects with BD I and II than in subjects with MDD (p < 0.001).
Family history of BD I and II, NOS, and MDD
BD I and II (n = 30)
BD NOS (n = 25)
MDD (n = 143)
Family History of
Mania or hypomania
Other psychiatric disorders
First degree relative with
Mania or hypomania
Other psychiatric disorders
Psychiatric comorbidities of BD I and II, NOS, and MDD
BD I and II (n = 30)
BD NOS (n = 25)
MDD (n = 143)
To our knowledge, this is the only study to compare the demographic and clinical characteristics, psychiatric comorbidities, and family history of mood disorder in subjects with BD I or II, BD NOS assessed by previously validated operational criteria, and MDD in an Asian clinical sample of youths.
In this study, we demonstrated that the prevalence of BD I and II and MDD was 1.0, 0.5 and 7.2 %, respectively, based on DSM-IV criteria, and the prevalence of BD NOS was 1.3 % based on the COBY criteria in a Korean clinical sample. The prevalence of pediatric BD I and II found in this study is quite similar to those reported in other cultures , but that of BD NOS was lower compared to the previous studies [18, 19]. This discrepancy might be related to the fact that our study based on COBY criteria which is different from the definition of BD NOS of the two previous studies. Moreover, the rates reported in the meta-analysis by Van Meter et al.  were from community samples, whereas our sample consisted of youths seeking mental health treatment. The rates of bipolar spectrum disorders in other outpatient treatment seeking samples in the United States are higher, ranging from 6 to 12 % [29, 30], with a ratio of about two major depressive cases per bipolar spectrum case, versus the 3:1 ratio observed here.
Cultural differences might contribute to the lower prevalence in this study. Prevalences of bipolar and depressive disorders have been reported to be lower in Asian countries than in Western countries. In the World Mental Health Survey Initiative study which was performed in adult general populations, lower prevalence rates of bipolar spectrum disorder in Japanese (0.7 %) and Indian (0.1 %) populations were found . In the 2005, Seoul Child and Adolescent Mental Health Survey which was conducted on the Korean youth general population , the prevalences of MDD (0.52 %) and BD I and II (1.09 %) were low, concurrent with our results. These lower rates of bipolar and depressive disorders in Korea might be related to the fact that Korean parents tend to pay more attention to children’s behavior problems than mood problems . Further studies on the contribution of the cultural difference to mood disorder prevalence are needed.
We found that age of onset did not differ across the three mood groups, inconsistent with previous studies [8, 34]. Previous studies that showed lower age of onset in adolescents with BD than in those with MDD were prospective longitudinal studies with follow-up periods ranging from 3 to 7 years [11, 35]. Because our study was retrospective chart review, and mean follow-up period was relatively short (456 days), our result could underestimate diagnostic conversion of MDD to BD. Some percentage of the MDD group is likely to convert to bipolar disorder over subsequent years.
In our study, bipolar disorder was more frequent in female adolescents and depression was more common in boys. Previous studies suggest that the rates and clinical features of bipolar disorders in youth are not different between boys and girls [10, 36]. However, other studies reported that slightly BD I and BD II are more prevalent in female adolescents [37, 38], concurrent with our results. In contrast, the ratio of depression in males and females is similar in pre-pubertal children but becomes more common among females compared with males during adolescence [39, 40], discordant with our results. Higher rates of depressive disorder in boys in this study could also be related to cultural factors. Depressive girls are likely to have comorbid anxiety, eating or adjustment disorders, whereas depressed boys often have disruptive disorders . Korean parents focus more on accompanying behavioral symptoms than mood symptoms per se . Thus it is possible that boys with both depressive and behavioral problems are more often referred for treatment. The cultural influence on gender distribution in juvenile mood disorder should be investigated in further studies.
In our study, mood changes related to antidepressants and aggressive behaviors were significantly prevalent in children and adolescents with BD I, II and BD NOS than in MDD, consistent with previous reports . Although how medications and manic symptoms might be associated is complicated, mood change related to antidepressants is considered to be a risk factor for BD or perhaps enhances heightened vigilance and earlier detection of hypomania .
The National Comorbidity Survey documented that the 12-month adult population prevalence of violent behaviors was 2 %, whereas it was 16 % for adults with BD . In previous literature, subtype of BD (i.e., BD I versus II) and the polarity of current episode (depression, mixed) did not influence the severity of aggression . By contrast, Graz and colleagues  found a significantly higher rate of crimes and violent behaviors with mania than those with bipolar or unipolar depression.
In our study, atypical features tended to be more common in depressive episodes of BD I and II than in MDD. In a previous study, 24 % of subjects with atypical depression met DSM-IV criteria for BD and 78 % met broader criteria for bipolar spectrum disorder, implying atypical depression as a potential risk factor for bipolar disorder . Akiskal and Benazzi also suggested a strong link between atypical depression and bipolar disorder .
Histories of suicide attempts and non-suicidal self-injury (NSSI) were not significantly different across the three subgroups. The prevalence of suicidal behavior in BD I and II (23.3 %) and BD NOS (28.0 %) are comparable to the 32 % rate documented in previous reports using COBY criteria [48, 49]. Because we did not perform objective measures  to rate suicidal behavior and NSSI, the rate of suicidal attempt and NSSI might be underestimated.
Precipitating psychosocial stressors were more common in MDD in this study, congruent with previous studies. Prior work found that patients with bipolar disorder had fewer life events as compared with patients with unipolar disorder [51, 52], suggesting that the genetic component might be more important than the environmental factor in bipolar disorder .
In this study, family history of mania or hypomania was significantly higher in subjects with BD NOS than those with MDD, consistent with previous literatures [17, 53]. Youth with BD NOS were reported to have elevated family histories of BD, and family history of BD was associated with conversion from BD NOS to BD I or II . However, family history of mania was not associated with BD I and II in our results. This could be related to the fact that the number of subjects with BD I and II was relatively small and that family history was not obtained from a structured or semi-structured interview. In contrast, family history of depression was not different among three groups. The relationship between family history of depression and mood disorders is controversial. Familial history of MDD did not differ between adult with BD I and II and MDD . In a recent survey, family history of MDD differed significantly in adult with BD I, II and MDD . In youth, Wozniak et al.  documented that in subjects with ADHD, bipolar depression is associated with higher levels of MDD in relatives. Future research is needed to clarify the extent to which family history of depression is a risk factor for BD.
In this study, children and adolescents with BD NOS experienced high rates of comorbid obsessive–compulsive disorder than those with MDD. Youth with BD I and II tended to experience high rates of comorbid panic disorder compared to those with BD NOS and MDD. These results are fairly consistent with Wozniak et al.’s study , which reported higher rate of comorbid agoraphobia, obsessive–compulsive disorder (OCD), and alcohol abuse, compared to children in the bipolar group than those in the unipolar depressive group.
Comorbid ADHD with BD I and II and BD NOS was lower than we expected. Systematic studies of children and adolescents with a diagnosis of bipolar disorder show that rates of ADHD often range from 60 to 90 % [56–58]. However, these studies were mostly conducted in Western populations. To the best of our knowledge, except for our study, only one study from India identified the rate of comorbid ADHD in Asian bipolar youth and the rate was similarly low (4 %) . In addition, in our study, comorbid diagnosis was not based on structured interviews, contributing to lower comorbid rates of ADHD. Moreover, studies of ADHD samples often find low rates of bipolar disorder, and two reviews recently concluded that the association between the disorders is small to moderate, and evidence of an etiological linkage between ADHD and bipolar disorder is likely to be attributable to shared underlying processes [60, 61]. Further research is needed to address the relationship between ADHD and BD.
In our study, no subjects had comorbid substance use disorders. Cultural influences should be considered when interpretating this finding. Although alcohol and tobacco use disorders are highly prevalent in Korean adults and the rate of adolescents who experienced alcohol or tobacco is higher [62, 63], no epidemiological studies have determined the prevalence of alcohol and tobacco use disorders based on adolescents’ and parents’ interviews. The rates of drug use (0.7–1.9 %) were known to be lower and those of internet addiction (10.7–11.9 %) higher compared to Western countries [62, 63]. Moreover, parents who visited mental health clinics reported more problematic internet use than alcohol and tobacco-related behaviors, possibly because Koreans are traditionally generous about drinking.
Some limitations should be considered in interpreting our findings. First, our study is a retrospective chart review. Clinical diagnoses and chart reviews tend to have strong specificity—symptoms and events recorded are unlikely to be false positives. However, sensitivity is much higher when using research interviews and prospective methods, particularly for hypomania and for events subject to social desirability biases. To illuminate the predictive factors of bipolar development in depressive episode, long-term prospective follow-up study is needed. Second, the diagnoses of mood disorder and comorbid psychiatric disorders were not based on structured interviews. Clinical diagnoses tend to be specific, with fewer false positives, but they may be less sensitive to comorbid conditions and often identify fewer total diagnoses than would result from using structured or semi-structured interviews with the same cases . Family history was also not obtained from structured interviews. Third, follow-up duration was relatively short compared to previous studies . Fourth, because participants were seeking treatment, rates of disorder are influenced by factors such as stigma, severity of presentation, and attitudes towards treatment . This issue is less likely to influence comparisons across mood groups within the same setting, which were the primary analyses here; but it would still be valuable to examine rates of mood disorder from the general population. Fifth, we were not able to examine cyclothymic disorder separately from bipolar NOS. This would be helpful in future studies; however, the collapsing of cyclothymic disorder and bipolar NOS into a single category yielded an operational definition more directly comparable to that used in prior research  as well as much of clinical practice . Sixth, the sample size of BD I and II and BD NOS groups was relatively small. Because they yield very small cell sizes when subdivided into several categories, we could not compare the clinical characteristics of BD I and II. Seventh, pubertal state, which can be related to the age of onset of bipolar or depressive disorder, was not assessed. Eighth, this study was conducted in a single institution and possibly subjected to selection bias. Lastly, we did not assess potential confounding variables, such as type of drug and drug compliance.
Our results suggest that bipolar depression is distinct from unipolar depression in family history, comorbidity, and clinical characteristics in Korean children and adolescents. Long-term prospective follow-up study is needed.
There was no financial support for this study.
Conflict of Interest
Dr. Eric Youngstrom has received travel support from Bristol-Myers Squibb and consulted with Lundbeck; he receives grant funding from the NIH. Other authors have no conflict of interest.