Symptoms and Characteristics of Youth Hospitalized for Depression: Subthreshold Manic Symptoms Can Help Differentiate Bipolar from Unipolar Depression

Abstract

Background:

Most people who have major depressive disorder (MDD) or bipolar disorder (BD) will have their first episode of depression in adolescence. However, in the absence of significant [hypo]manic symptoms, there are no clear guidelines for distinguishing bipolar from unipolar depression, which can lead to misdiagnosis and ineffective/harmful treatment. We aimed to compare phenomenological differences among youth with MDD or BD hospitalized for an acute episode of depression.

Methods:

A retrospective electronic chart review of adolescents hospitalized in an acute care inpatient unit who had a discharge diagnosis of MDD, MDD with mixed or psychotic features (MDD+), BD-I—current episode depressed, or BD-II—current episode depressed, was performed.

Results:

Altogether, 598 patients (mean age = 15.1 ± 1.5 years, female = 71%, and White = 46%) met study inclusion criteria, i.e., BD-I: n = 39, BD-II: n = 84, MDD: n = 422, and MDD+: n = 53 patients. The admission Hamilton Depression Rating Scale (HAMD) total score was significantly higher in the BD-I (29.3 ± 9.1) and MDD+ (31.2 ± 9.3) groups versus the MDD group (24.3 ± 9.7) (p < 0.05). Although there were some group differences in the severity of individual depression symptoms, these did not line up neatly across BD and MDD groups. At admission, Young Mania Rating Scale (YMRS) total scores were significantly higher in the BD-I (14.4 ± 7.4), BD-II (13.8 ± 6.5), and MDD+ groups (14.3 ± 6.6) versus the MDD group (8.2 ± 4.6, p < 0.05). Additionally, 9 of 11 and 4 of 11 YMRS items scored significantly higher in the BD-II and BD-I groups versus the MDD group, respectively. The motor activity and hypersexuality items, in particular, were scored consistently higher in the BD groups than MDD groups.

Limitations:

All diagnoses were made based on a clinical interview and not a structured diagnostic interview, and some of the subgroup sample sizes were relatively modest, limiting the power for group comparisons.

Conclusion:

The presence of subsyndromal manic symptoms during an episode of MDD currently offers the clearest way by which to differentiate bipolar depression from unipolar depression.

Introduction

Most people with bipolar disorder (BD) will experience an episode of depression before an episode of mania or hypomania (Van Meter et al. 2016b). Additionally, people with BD spend more time (three times as many days on average) depressed than in a [hypo]manic state (Kupka et al. 2007). These factors likely contribute to the long delays many people with BD experience before getting an accurate diagnosis (Marchand et al. 2006); there is an average gap of 10 years between first seeking mental health services and being diagnosed and treated for BD (Lish et al. 1994). Furthermore, many people with BD are treated for depression in a primary care setting, where providers are less likely to have the expertise necessary to make an accurate diagnosis (Manning et al. 1997). Missed or delayed diagnosis of BD can contribute to poorer outcomes because the treatment administered is less likely to adequately manage symptoms, contributing to poor interepisode functioning and increasing risk for relapse (Stensland et al. 2008). In particular, antidepressant monotherapy may exacerbate symptoms of elevated mood (Birmaher and Brent 2007; Tondo et al. 2010).

An evidence-based approach to assessment can help to improve the accuracy and timeliness of diagnosis (Youngstrom et al. 2017). However, current diagnostic standards based on the Diagnostic and Statistical Manual (DSM-5; American Psychiatric Association 2013) do not sufficiently distinguish between unipolar and bipolar depression, which limits clinicians' ability to identify a patient with bipolar depression in the absence of significant manic symptoms (American Psychiatric Association 2013). Although most competent clinicians could reliably differentiate someone experiencing an episode of mania from someone experiencing an episode of unipolar depression, people with BD spend more time depressed than manic (or hypomanic) and are more likely to be misdiagnosed during an episode of depression than an episode of mania (or hypomania). If a young person presents with depressed mood, DSM criteria do not provide guidance for distinguishing whether the depressed mood is due to bipolar or unipolar depression.

Diagnostic challenges related to unipolar and bipolar depression may disproportionately affect young people. Most people who have major depressive disorder (MDD) or BD will have their first episode of depression in adolescence (Lish et al. 1994; Perlis et al. 2004; Baldessarini et al. 2012), an important developmental period during which poor mental health can have a longstanding impact on the quality of life. In fact, more than half of the youth diagnosed with depression will later be diagnosed with BD (Geller et al. 2001; Smith et al. 2005; Cosgrove et al. 2013), suggesting that a substantial number of youth would benefit if there were a clear way to differentiate unipolar from bipolar depression.

Although the diagnostic criteria for MDD apply to all people with depressed mood, and unipolar depression and bipolar depression share many symptoms, there is evidence that bipolar depression has some unique characteristics that could help with the identification of BD before the onset of a [hypo]manic episode (Perris and d'Elia 1966; Bagby et al. 1996; Chen et al. 1996; Lovejoy and Steuerwald 1997; Brieger et al. 2003; Mitchell and Malhi 2004; Bowden 2005; Mendlowicz et al. 2005; Kessing et al. 2008; Oedegaard et al. 2008; Ahmed et al. 2011; Frankland et al. 2015; Leonpacher et al. 2015; Serafini et al. 2018; Patella et al. 2019). Additionally, bipolar depression tends to be more severe than unipolar depression and is more likely to require hospitalization (Kessing et al. 2008; Patella et al. 2019). This difference may be related to the fact that bipolar depression tends to be associated with an earlier age of onset and that the episodes are often more frequent (Leonpacher et al. 2015; Patella et al. 2019). Therefore, according to the kindling hypothesis, each new mood episode increases the probability of relapse and reduces the duration of the interepisode period (Post 2007).

As described, both unipolar depression and bipolar depression typically begin in adolescence, and although there has been less research on phenomenological differences between the two groups in youth, what is known largely comports with the adult literature. Specifically, adolescents with bipolar depression, compared with those with unipolar depression, report more severe depressive symptoms and are more likely to experience subsyndromal manic symptoms during depressive episodes (Karlsson 2007; Shon et al. 2013; Uchida et al. 2015; Diler et al. 2017). Compared with youth with unipolar depression, those with bipolar depression also tend to report greater levels of impairment and psychiatric comorbidity, including disruptive behavior, anxiety, and substance use disorders (Karlsson 2007; Uchida et al. 2015; Patel et al. 2020). Some of these differences, including subsyndromal manic (Fiedorowicz et al. 2011) and somatic symptoms (Bohman et al. 2012), may be apparent before the onset of a [hypo]manic episode.

Only two studies have explored the phenomenological differences at the individual symptom level in adolescents. In their retrospective review of charts from a child psychiatry department, Shon et al. found that atypical features were significantly more prevalent among youth diagnosed with BD-I, BD-II, or BD-NOS, compared with MDD. Aggressive behaviors were also more prevalent among those with a bipolar spectrum diagnosis, including BD-I, BD-II, and BD-NOS, compared with MDD (Shon et al. 2013). Diler et al. compared 30 adolescents with BD with 59 adolescents with MDD during an acute depressive episode. They found that overall depression severity was higher in youth with BD and that some specific depressive symptoms such as hopelessness and social withdrawal, as well as psychosis, mood reactivity, and cravings for sweets, were more severe among youth with BD. Additionally, youth with BD were more likely to report engaging in nonsuicidal self-injury and were more likely to meet criteria for comorbid disorders, including attention-deficit/hyperactivity disorder, disruptive behavior disorders, and anxiety disorders, than unipolar depressed youth. Interestingly, although the ratings were made during an acute depressive episode, youth with BD also reported significantly higher subsyndromal manic symptoms. The authors concluded that careful assessment of manic and depressive symptoms may help in the clinical differentiation of bipolar and unipolar depression.

The presence of subthreshold manic symptoms during an index episode of depression is consistent with substantial evidence of a prodrome to BD. Multiple investigations have found that both symptoms of mania and depression are common in the months (or, in some cases, years) before the onset of a mood episode that meets DSM criteria (Van Meter et al. 2016b). Defining and assessing the prodromal symptoms of schizophrenia have facilitated early identification and intervention; similarly, recognition of subthreshold manic symptoms—particularly in the context of a depressive episode—could significantly improve the detection of bipolar spectrum disorders early in the illness course and improve timely appropriate intervention (Faedda et al. 2019). However, collecting data about subthreshold or prodromal symptoms is challenging because in a majority of research studies, individuals are recruited based on specific diagnoses and, consequently, opportunities to compare symptoms and trajectories before diagnosis—or across diagnoses—are limited. The goal of the present study was to compare the symptoms reported by adolescents hospitalized for an acute episode of depression, who were diagnosed with a unipolar depressive disorder (MDD), MDD+ (individuals who were experiencing psychotic or mixed symptoms at the time of admission, but had no history of mania or hypomania), or a bipolar spectrum disorder (BD-I or BD-II; patients who were currently depressed, but reported a history of manic or hypomanic episodes), to determine whether there are symptomatic differences that could inform differential diagnosis in a clinical setting. We excluded patients experiencing a current manic or mixed episode, as we were interested in whether there are qualitative differences in symptoms experienced during depressive episodes by those with BD versus those with MDD. The primary aim of our study was to describe the symptoms of youth hospitalized for depression, comparing those with a diagnosis of MDD with those with a diagnosis of BD. We hypothesized that there would be symptomatic differences between the groups, including greater levels of subthreshold manic and psychotic symptoms in the BD group and greater overall depression severity in the BD group. The secondary aim of our study was to test whether there was symptom-level evidence in support of a mood disorder spectrum; we anticipated that the magnitude of difference in the presence and severity of symptoms would be greatest when comparing MDD with BD-I and smallest when comparing BD-II with MDD.

Methods

Design and setting

This study was conducted using data collected from the electronic medical records of adolescents who had previously been hospitalized in an acute care inpatient unit located in a suburb of New York City between 2012 and 2018. The unit is a part of a free-standing psychiatric hospital, which is adjacent to a tertiary care children's hospital. The local Institutional Review Board (IRB) approved the study as a retrospective chart review.

Sample

Patients, aged 12–17 years, were included in the study if they were hospitalized between November 1, 2012, and October 31, 2018, with a clinical discharge diagnosis of MDD, MDD with mixed or psychotic features (MDD+), BD-I—current episode depressed, or BD-II—current episode depressed. Patients experiencing a mixed or manic episode were excluded. These diagnoses were selected to address the clinical question of whether there are symptoms that distinguish between MDD and BD during a depressive episode. For patients hospitalized more than once, only the symptom rating scale data from their final admission (at our hospital) were included in the analyses, the rationale being that a later diagnosis is more likely to be stable and, consequently, the results more generalizable.

Outcomes

Primary outcomes were differences between mania and depression symptoms across groups. Secondary outcomes included differences in other demographic and clinical data.

Assessments

A diagnostic interview was conducted according to DSM by an attending psychiatrist assigned to the case upon admission. Diagnosis was made based on the consensus of team members upon admission and again before discharge, utilizing all available information. The diagnosis used to categorize patients into groups for the purposes of analyses in this study was the admission diagnosis. Therefore, all patients in the BD group would have reported a history of manic or hypomanic symptoms, in addition to their current depressed mood, whereas the MDD patients would have all denied any history of mania or hypomania at the time of admission. Additionally, patients' symptoms were rated by the attending board-certified psychiatrist (last author) at admission using both the 24-item Hamilton Depression Rating Scale [HAMD; Hamilton (1960)] and the Young Mania Rating Scale (YMRS; Young et al. 1978). The use of clinical rating scales was optional for clinicians. During the study period, there were two primary attending physicians, one of whom routinely conducted ratings and one who did not. Patients were randomly assigned to an attending physician at admission. Of 1395 patients with an MDD or BD diagnosis, 43% had complete symptom data and were included in the present analysis. In addition to diagnosis and clinical rating scales, patients' demographic data, including age, race, ethnicity, sex, length of stay, and whether they were readmitted to the same unit within 30 days of discharge or at a later date, were extracted from the electronic medical record. All procedures were approved by the IRB.

Statistical analyses

For analysis purposes, patients were separated into four groups based on their discharge diagnosis: BD-I, BD-II, MDD, and MDD+. One-way analysis of variance and chi-square analysis were used to compare the four groups on symptoms and clinical and demographic characteristics, as well as rates of readmission. Receiver operating characteristic curves were calculated for the YMRS and HAMD total scores to assess how well each scale augmented diagnostic discrimination. The primary analyses focused on symptom differences across patient groups at intake. As a sensitivity analysis, we also compared symptoms of depression and mania only among patients who had no comorbid diagnoses to gauge how much comorbidity influenced group differences. Analyses were conducted using SPSS; all hypotheses were two-tailed and alpha was set at 0.05.

Results

Sample characteristics

There were 1395 unique patients admitted for an acute episode of depression. Of these, 598 patients had symptom ratings from both the YMRS and HAMD and were included in the study analyses. Patients included in the study were less likely to be male (32% vs. 39%, X² = 16.88, p < 0.0005). There were differences in age between those included and those excluded from the study. Among included patients, 39 had a discharge diagnosis of BD-I, 84 had BD-II, 422 had MDD, and 53 had MDD with psychosis/mixed features. On average, the patients were 15.13 (standard deviation [SD] = 1.5) years old. Seventy-one percent were female, 46% were White, 15% were Black, 10% were Asian, and 17% were other/mixed race. Sixteen percent of patients were hospitalized more than once during the study period. Demographic characteristics were largely consistent across patient groups; youth with BD-I were more likely to be rehospitalized than the three other groups (44% vs. 21%, 12%, and 21%, p < 0.05) and were more likely to be rehospitalized in 30 days than youth with MDD (15% vs. 3%, p < 0.05). Youth with MDD were more likely to have a comorbid anxiety disorder than youth with BD-II (p < 0.05). See Table 1 for demographic and clinical characteristics.

Table 1.

Sample Demographic and Clinical Characteristics

	Full sample	BD-I	BD-II	MDD	MDD+
	598	39	84	422	53
	Mean (SD)
Age, years	15.13 (1.5)	15.64 (1.3)d	15.45 (1.4)d	15.08 (1.5)	14.62 (1.8)a,b
Length of stay (days)	7.81 (6.1)	10.08 (5.9)c	7.86 (3.3)	7.24 (5.2)a,d	10.53 (12.3)c

	n (%)
Female	427 (71)	30 (77)	64 (76)	296 (70)	37 (70)
Race
White	271 (45)	13 (33)	39 (47)	199 (47)	20 (38)
Black	91 (15)	5 (13)	18 (22)	62 (15)	6 (11)
Asian	60 (10)	4 (10)	9 (11)	38 (9)	9 (17)
Other/multiracial	104 (17)	11 (28)	13 (16)	70 (17)	10 (19)
Declined	82 (14)	6 (15)	5 (6)	53 (13)	8 (15)
Hospitalization status
Hospitalized more than once	96 (16)	17 (44)b,c,d	18 (21)a	50 (12)a	11 (21)a
Readmitted in 30 days	26 (4)	6 (15)c	5 (6)	11 (3)a	4 (7)
Comorbid disorders
Anxiety disorder	94 (16)	2 (5)	4 (5)c	78 (19)b	10 (19)
Substance use disorder	34 (6)	1 (3)	4 (5)	26 (6)	3 (6)
Disruptive behavioral disorder	10 (2)	1 (3)	2 (2)	5 (1)	2 (4)

^a Different from BD-I, p <0 .05.

^b Different from BD-II, p < 0.05.

^c Different from MDD, p < 0.05.

^d Different from MDD+ psychosis/mixed symptoms, p < 0.05.

BD, bipolar disorder; DBD, disruptive behavioral disorder (i.e., attention-deficit/hyperactivity disorder, oppositional defiant disorder, and conduct disorder); MDD, major depressive disorder; MDD+, major depressive disorder with psychosis and/or mixed features; SD, standard deviation.

Depressive symptomatology

At admission, the HAMD total score was significantly higher in the BD-I (29.28, SD = 9.1) and MDD+ (31.24, SD = 9.3) groups relative to the MDD (24.33, SD = 9.7, p < 0.05) group. The MDD+ group's total HAMD score was also higher than the BD-II group (26.45, SD = 9.8, p < 0.05). Patients with BD-I had more severe ratings for somatic energy, hypochondriasis, and paranoid symptoms versus MDD patients. BD-I patients had higher scores of hypochondriasis and paranoid symptoms versus BD-II patients. There were no statistically significant differences on individual items between the BD-I group and the MDD and MDD+ groups. BD-II patients had elevated scores for middle insomnia and somatic energy versus MDD patients and had lower scores than MDD+ patients on psychic anxiety, somatic anxiety, depersonalization, and paranoid symptoms. MDD patients had lower scores than MDD+ patients on eight items: guilt, work and interests, psychic anxiety, appetite, somatic energy, depersonalization and derealization, paranoid symptoms, and worthlessness. The HAMD total score was associated with an area under the curve (AUC) of 0.61, p < 0.0001, indicating poor diagnostic discrimination (Table 2). In the sensitivity analysis, excluding patients with comorbidities, between-group differences were consistent.

Table 2.

Admission Mania and Depression Symptom Scores

	Full sample	BD-I	BD-II	MDD	MDD+
	598	39	84	422	53
Young Mania Rating Scale	Mean (SD)
Elevated mood	0.91 (1.3)	0.87 (1.3)b	1.68 (1.5)a,c,d	0.81 (1.2)b	0.53 (1.1)b
Motor activity	0.89 (1.0)	1.3 (1.1)c,d	1.33 (1.2)c,d	0.78 (0.9)a,b	0.75 (0.9)a,b
Hypersexuality	0.16 (0.5)	0.44 (0.8)c,d	0.31 (0.7)c,d	0.12 (0.4)a,b	0.06 (0.3)a,b
Decreased need for sleep	0.88 (1.2)	1.15 (1.4)	1.26 (1.4)c	0.76 (1.1)b	1.02 (1.2)
Irritable mood	1.22 (1.6)	2.62 (1.7)	2.62 (1.5)c	2.11 (1.6)b	2.15 (1.6)
Pressured speech	0.98 (1.7)	1.28 (1.9)	1.9 (2.1)c,d	0.75 (1.5)b	1.09 (1.7)b
Thought disturbance	1.25 (0.9)	1.54 (1.0)c	1.55 (1.1)c	1.13 (0.9a,b,d	1.57 (0.7)c
Thought content	1.06 (2.6)	3.49 (3.9)b,c,d	1.12 (2.5)a,c,d	0.29 (1.4)a,b,d	5.36 (3.7)a,b,c
Aggressive behavior	0.85 (1.4)	1.03 (1.4)	1.26 (1.6)c	0.73 (1.3)b	1.02 (1.5)
Appearance	0.45 (0.6)	0.51 (0.6)	0.45 (0.5)	0.44 (0.6)	0.51 (0.6)
Insight	0.23 (0.9)	0.21 (0.8)	0.31 (1.0)	0.22 (0.8)	0.21 (0.9)
YMRS total score	9.90 (5.9)	14.44 (7.4)c	13.8 (6.5)c	8.15 (4.6)a,b,d	14.26 (6.6)c

Hamilton Rating Scale for Depression	Mean (SD)
Depressed mood	2.53 (1.0)	2.51 (1.0)	2.52 (1)	2.5 (1.1)	2.75 (0.8)
Guilt	2.41 (1.7)	2.95 (1.4)	2.45 (1.7)	2.26 (1.7)d	3.17 (1.4)c
Suicidal thoughts	2.63 (1.0)	2.62 (1.1)	2.82 (0.8)	2.59 (1.1)	2.68 (1.1)
Initial insomnia	1.05 (1.0)	0.85 (1.1)	1.27 (0.9)	1.02 (1.1)	1.05 (1.1)
Middle insomnia	0.53 (0.7)	0.59 (0.7)	0.79 (0.8)c	0.46 (0.7)b	0.58 (0.7)
Delayed insomnia	0.32 (0.6)	0.46 (0.8)	0.4 (0.6)	0.28 (0.6)	0.32 (0.6)
Work and interests	2.54 (1.4)	2.74 (1.3)	2.68 (1.5)	2.43 (1.4)d	2.98 (1.2)c
Psychomotor retardation	0.31 (0.7)	0.36 (0.8)	0.15 (0.4)	0.34 (0.7)	0.3 (0.6)
Psychomotor agitation	0.57 (0.6)	0.69 (0.6)	0.49 (0.5)	0.58 (0.6)	0.51 (0.5)
Psychic anxiety	2.02 (1.5)	2.15 (1.6)	1.95 (1.6)d	1.94 (1.4)d	2.64 (1.4)b,c
Somatic anxiety	1.84 (1.6)	2.13 (1.6)	1.88 (1.6)d	1.72 (1.6)	2.55 (1.4)b
Appetite	0.53 (0.7)	0.54 (0.7)	0.61 (0.7)	0.48 (0.7)d	0.81 (0.8)c
Somatic energy	1.05 (0.8)	1.41 (0.7)c	1.23 (0.9)c	0.96 (0.8)a,b,d	1.28 (0.7)b
Libido	0.23 (0.5)	0.31 (0.5)	0.29 (0.6)	0.21 (0.5)	0.28 (0.6)
Hypochondriasis	0.14 (0.4)	0.38 (0.7)b,c	0.11 (0.3)a	0.11 (0.4)a	0.25 (0.6)
Loss of insight	0.11 (0.4)	0.1 (0.4)	0.19 (0.6)	0.1 (0.4)	0.08 (0.3)
Weight loss	0.18 (0.4)	0.31 (0.6)	0.19 (0.5)	0.15 (0.4)	0.28 (0.5)
Diurnal variation	0.00 (0.1)	0.36 (0.7)	0.48 (0.7)	0.36 (0.7)	0.47 (0.8)
Depersonalization and derealization	0.39 (0.7)	0.38 (0.6)	0.27 (0.5)d	0.18 (0.4)d	0.58 (0.7)b,c
Paranoid symptoms	0.24 (0.5)	0.87 (1.3)b,c	0.27 (0.5)a,d	0.24 (0.5)a,d	0.81 (1.1)b,c
Obsessional and compulsive symptoms	0.24 (0.5)	0.26 (0.6)	0.25 (0.5)	0.22 (0.5)	0.42 (0.7)
Helplessness	0.93 (1.1)	1.05 (1.1)	0.83 (1.0)	0.92 (1.1)	1.08 (1.1)
Hopelessness	2.02 (1.6)	2.38 (1.7)	1.87 (1.6)	1.99 (1.6)	2.3 (1.7)
Worthlessness	2.44 (1.8)	2.87 (1.6)	2.45 (1.8)	2.3 (1.8)d	3.19 (1.5)c
Total	25.57 (9.9)	29.28 (9.1)c	26.45 (9.8)d	24.33 (9.7)a,d	31.34 (9.3)b,c

^a Different from BD-I, p < 0.05.

^b Different from BD-II, p < 0.05.

^c Different from MDD, p < 0.05.

^d Different from MDD+ psychosis/mixed symptoms, p < 0.05.

BD, bipolar disorder; MDD, major depressive disorder; MDD+, major depressive disorder with psychosis and/or mixed features.

Mania symptomatology

At admission, there were significant differences between groups on total YMRS scores; the BD-I group (14.44, SD = 7.4), BD-II group (13.80, SD = 6.5), and MDD+ group (14.26, SD = 6.6) had significantly higher scores than the MDD group (8.15, SD = 4.6, p < 0.05). Additionally, most YMRS items (9 items of 11) scored higher in the BD-II group versus the MDD group, except for appearance and insight. There were fewer differences between BD-I and MDD groups, with only motor activity, hypersexuality, thought disturbance, and thought content being significantly more elevated for BD-I than MDD. MDD+ patients had lower scores than both BD-I and BD-II patients on motor activity and hypersexuality and lower scores than BD-II patients on elevated mood and pressured speech. Comparing MDD and MDD+ with BD-II, patients with MDD had lower scores on pressured speech. Motor activity and hypersexuality were elevated in BD-I and BD-II versus both MDD and MDD+ groups. Considering YMRS scores has utility for distinguishing unipolar from bipolar depression; the AUC for total YMRS scores in our sample was 0.72, p < 0.0001. The recommended threshold of 13 on the YMRS corresponded to a sensitivity of 0.57 and specificity of 0.78 (Table 2). In the sensitivity analysis, excluding patients with comorbidities, between-group differences were consistent, with the exception of hypersexuality (MDD+ was no longer different from BD-I or BD-II).

Discussion

Differentiating between BD and MDD can be challenging, particularly early in the illness course. We aimed to evaluate the characteristics and symptoms of adolescents hospitalized due to a depressive episode, comparing those who were diagnosed with MDD with those who were diagnosed with a BD spectrum disorder during an acute depressive episode. Consistent with our hypotheses, we found some differences, particularly in the presence of manic, somatic, and psychotic symptoms, which were more likely in youth with BD spectrum diagnoses. Unexpected results included the fact that the differences were more prominent between youth with MDD and BD-II rather than between BD-I and MDD.

The fact that manic symptoms were elevated in the two BD groups, relative to the MDD groups, may seem tautological; however, all patients included in this study were hospitalized due to a current episode of depression and the manic symptoms endorsed were of low severity in general (i.e., mean ratings <2 in almost all cases and total scores <15). This fact indicates that assessing for mild symptoms of mania may help to differentiate unipolar from bipolar depression. This finding is consistent with previous research, which has shown that bipolar depression is often characterized by mixed symptoms (Diler et al. 2017). The AUC for the YMRS in this sample was 0.72, indicating fair accuracy and suggesting that its inclusion would result in improved diagnostic accuracy.

We had hypothesized that the MDD+ group, the participants of which had psychotic and/or mixed symptoms, would be more similar to the BD groups than the MDD group. This hypothesis was supported only regarding subsyndromal manic symptoms; the MDD+ group differed from BD-I or BD-II on five manic symptoms, whereas the MDD group differed on nine individual items and had a lower overall score than the BD-I or BD-II group. We did not find the MDD+ group to be more similar to the BD groups, than to the MDD group, on depressive symptoms. The MDD+ group differed from the BD-II group of five HAMD items, whereas the MDD group differed from both the BD-II group and the BD-I group on only two items. Another unexpected finding was that the most consistent group differences were between MDD and BD-II, rather than between BD-I and MDD. Previous studies in adults have found that there are differences in the symptoms most commonly experienced by individuals with BD-I or BD-II during an episode of depression (Brugue et al. 2008; Baek et al. 2011) and that people with BD-II may have more frequent and longer lasting episodes of depression (Judd et al. 2005; Vieta and Suppes 2008). Consistent with these studies, we found that youth with BD-I were more likely to report disturbed thought content and that elevated mood was higher in youth with BD-II. Although we also found differences between BD-II and MDD in these domains, our results are likely related to the fact that the number of youth with BD-II hospitalized for depression was higher than the number with BD-I, increasing the power for these comparisons.

There were two symptoms that differentiated the overall MDD and BD groups. On the YMRS, both motor activity and hypersexuality were higher in both BD groups. Hypersexuality is clinically interesting because it is fairly specific to BD, whereas most other BD symptoms have some overlap with other disorders (Youngstrom et al. 2008). However, hypersexuality is only present in about a third of youth with BD (Van Meter et al. 2016a), thus although a specific indicator, if used alone to identify youth with BD, the false-negative rate would be high. Motor activity has the benefit of being measurable using objective methods, such as actigraphy, rather than relying on patient self-report, which is more subject to bias (Faedda et al. 2016). Additionally, actigraphy measurement of motor activity has differentiated adults with BD from those with MDD (Merikangas et al. 2019), which is an important area for further exploration in youth.

Previous research has found differences between youth with BD and youth with MDD regarding somatic symptoms (Bohman et al. 2012; Galvão et al. 2013; Diler et al. 2017), although the nature of the differences has been inconsistent. We found lower scores in the MDD group for somatic energy and hypochondriasis. However, the MDD+ group had elevated scores for somatic energy, relative to the MDD group, and the BD-I group had higher scores for hypochondriasis, relative to the BD-II group, so somatic symptoms may be a marker of severity of illness, rather than BD or MDD status. Related previous research found that youth with BD receive the highest number of as-needed pro re nata (PRN) medications for physical symptoms during a hospital stay (Saito et al. 2020). Hence, hypochondriasis and multiple somatic complaints may be related to BD.

Sleep problems are common among people with MDD and BD, but the type of sleep problem can vary. Specifically, although insomnia is prevalent in both disorders, decreased need for sleep is typically found only in BD. Interestingly, we found that decreased need for sleep and middle insomnia were both elevated in patients with BD-II relative to MDD, but that there were no other group differences on sleep (including initial and delayed insomnia). This finding suggests that sleep disturbance is a feature that could aid in differential diagnosis, particularly since BD-II or MDD may be a more commonly difficult clinical question than BD-I or MDD.

The MDD+ group had the highest scores on disordered thought content and on depersonalization and derealization and—with BD-I—on paranoia. This finding was expected, given that the group was defined, in part, by the presence of psychotic symptoms, and supports our hypothesis that youth in the MDD+ group are likely to be the most difficult to reliably differentiate from youth with BD. This hypothesis is further reinforced by the fact that the total HAMD scores for the MDD+ group were equivalent to the BD-I group and significantly higher than the BD-II and MDD groups' total scores.

The differentiation of youth experiencing a bipolar versus unipolar episode of depression is important given differences in treatment guidelines (Birmaher and Brent 2007; Birmaher et al. 2007; McClellan et al. 2007; Lauxen Peruzzolo et al. 2013; Fristad and MacPherson 2014) and prognosis (Birmaher et al. 2004; Leverich et al. 2007; Patton et al. 2014; Frank et al. 2015). For example, although selective serotonin reuptake inhibitor (SSRI) antidepressant medications are commonly prescribed for adolescents with depression, for those with bipolar depression or who are at risk for BD, these medications can be contraindicated due to the association with affective switching (Kapur et al. 2020). When prescribed, SSRIs are typically accompanied by a mood stabilizer for those on the bipolar spectrum (McClellan et al. 2007), although research on the pharmacologic treatment of BD in youth is generally sparse (Lauxen Peruzzolo et al. 2013). On a related point, although both youth with unipolar depression and youth with bipolar depression are likely to relapse, risk is generally higher among those with bipolar depression and the long-term prognosis is generally worse (Rao et al. 2010; Birmaher et al. 2014; Schubert et al. 2017).

Limitations

Results of this study need to be interpreted with consideration of its limitations. First, the generalizability of our findings is limited by the fact that all diagnoses were made based on a clinical interview and not a structured diagnostic interview (although rating scales contributed to diagnostic conceptualization and improving the accuracy of clinical diagnosis) (Youngstrom et al. 2017). It is also important to note that because the patients in this study were all youth, their diagnoses may shift; those with a diagnosis of MDD, in particular, may later meet criteria for a bipolar spectrum disorder, thus introducing some error into our diagnostic groups. Furthermore, although consecutively admitted patients were included, we limited our sample to patients who were rated on both the YMRS and the HAMD. Although this fact strengthens confidence in our findings, it also means that not all youth with MDD or BD were represented in our dataset. Second, some of the subgroup sample sizes were relatively modest, limiting the power for group comparisons. Third, we did not have the exact medication status for each patient at the time of admission when the symptom assessment was made. Nevertheless, medications are generally prescribed in response to symptoms and not the reverse. As such, the medication information would not likely be informative in the pursuit of differential diagnosis. Finally, although the hospital serves a broad area, there are other psychiatric facilities, so the data on rehospitalization are imperfect. Nevertheless, this investigation provides valuable information about the presentation of unipolar and bipolar depression in adolescents hospitalized during an acute episode of depression.

Conclusions

Although there are differences between patients experiencing an episode of unipolar depression and those experiencing bipolar depression, the presentations tend to be quite similar during an acute episode. Interestingly, we did not find the four groups to be on a clear spectrum; symptom differences were greatest between BD-II and MDD for manic symptoms and between MDD+ and MDD for depressive symptoms. Across groups, differences in depressive symptoms were subtle when comparing the unipolar and bipolar groups, but the group differences in subsyndromal manic symptoms highlight the importance of conducting a structured assessment of both depressive and manic symptoms (ideally clinician rated, but patient or caregiver rated would also be beneficial) when young people experience an episode of depression. The presence of subsyndromal manic symptoms during an episode of MDD currently offers the clearest way by which to differentiate bipolar from unipolar depression in adolescents.

Disclosures

Dr. Ahmad, Dr. Dulin, and Dr. Van Meter have no conflicts to disclose. Dr. Correll has been a consultant and/or advisor to or has received honoraria from Acadia, Alkermes, Allergan, Angelini, Axsome, Gedeon Richter, Gerson Lehrman Group, Indivior, Intra-Cellular Therapies, Janssen/J&J, LB Pharma, Lundbeck, MedAvante-ProPhase, Medscape, Merck, Mylan, Neurocrine, Noven, Otsuka, Pfizer, Recordati, Rovi, Servier, Sumitomo Dainippon, Sunovion, Supernus, Takeda, and Teva. He provided expert testimony for Janssen and Otsuka. He served on the Data Safety Monitoring Board for Lundbeck, Rovi, Supernus, and Teva. He is also a stock option holder of LB Pharma. Dr. Saito received a one-time honorarium from Dainippon Sumitomo Pharmaceutical Company for a lecture in 2019.