Early-Onset Bipolar Disorder: Characteristics and Outcomes in the Clinic

Abstract

Objective:

To assess patient characteristics and clinician-rated outcomes for children diagnosed with early-onset bipolar disorder in comparison to a depressive disorders cohort from a single clinic site. To assess predictors of bipolar treatment response.

Methods:

Medical records from 714 consecutive pediatric patients evaluated and treated at an academic tertiary child and adolescent psychiatry clinic between 2006 and 2012 were reviewed. Charts of bipolar children (n = 49) and children with depressive disorders (n = 58) meeting study inclusion/exclusion criteria were compared on variables assessing clinical characteristics, treatments, and outcomes. Outcomes were assessed by using pre– and post–Clinical Global Impressions (CGI)-Severity and Children's Global Assessment Scale (CGAS) scores, and a CGI-Improvement score ≤2 at final visit determined responder status. Bipolar outcome predictors were assessed by using multiple linear regression.

Results:

Clinic prevalence rates were 6.9% for early-onset bipolar disorder and 1.5% for very early-onset bipolar disorder. High rates of comorbid diagnoses, symptom severity, parental stress, and child high-risk behaviors were found in both groups. The bipolar cohort had higher rates of aggression and higher lifetime systems of care utilization. The final CGI and CGAS outcomes for unipolar depression patients differed statistically significantly from those for the bipolar cohort, reflecting better clinical status and more improvement at outcome for the depression patients. Both parent-reported Child Behavior Checklist total T-score at clinic admission and the number of lifetime systems-of-care for the child were significantly and inversely associated with improvement for the bipolar cohort.

Conclusions:

Early-onset bipolar disorder is a complex and heterogeneous psychiatric disorder. Evidence-based treatment should emphasize psychopharmacology with adjunctive family and individual psychotherapy. Strategies to improve engagement in treatment may be especially important. Given high rates of high-risk behaviors in these youth, regular mental health follow-up to assess safety is important. Additional evidence-based treatments for pediatric bipolar disorder are needed.

Introduction

Early-onset bipolar disorder is a complex heterogeneous psychiatric disorder that is identified by onset occurring before the age of 18 years. The disorder is associated with episodic mood changes and/or symptoms of irritability, symptom severity, daily functional impairment, and increased risk for adult bipolar disorder (McClellan et al. 2007; Geller et al. 2008). Though once considered a rare disorder of childhood and adolescence, the number of youth receiving a diagnosis of bipolar disorder in the United States has increased 40-fold over the past 10 years, compelling increased interest in this condition (Moreno et al. 2007). The National Comorbidity Survey Replication-Adolescent Supplement found that ∼1% of adolescents in the United States have strictly defined BP-1 (Kessler et al. 2009). Despite increased prevalence, there continues to be marked controversy about developmentally appropriate bipolar diagnostic criteria (Van Meter et al. 2016), specificity of the diagnosis (McClellan et al. 2007), and prevalence rates of early-onset bipolar disorder (McClellan et al. 2007).

Though not addressing these controversies directly, this article focuses instead on a more immediate and practical clinical question: What are the characteristics of children and adolescents who are assigned a bipolar diagnosis and what are their outcomes when treated in a child and adolescent outpatient psychiatric clinic? A better understanding of their characteristics, treatment, and clinical outcomes is important to begin to optimize interventions and service delivery in real-world settings. To date, surprisingly few studies have attempted to characterize the service delivery and treatment outcomes provided to children and adolescents with bipolar disorder in real-world clinical settings. In a study investigating the types and quantity of mental health services and medication usage provided to 85 youth diagnosed with bipolar disorder from an integrated healthcare setting, high rates of psychopharmacology, complex psychopharmacology regimens, and high rates of inpatient psychiatric hospitalizations were reported (Vande Voort et al. 2016). Insurance claims data also document high rates of inpatient psychiatric admission, emergency department utilization, and treatment with antipsychotics and mood stabilizers (Peele et al. 2004; Olfson et al. 2009; Evans-Lacko et al. 2011). Unfortunately, these studies often do not report on additional patient characteristics that may influence service delivery and treatment such as diagnostic comorbidity, comorbid symptom severity, other systems-of-care (SOC) involvement, and premature dropout from treatment. To the best of our knowledge, no extant studies report treatment outcomes for children diagnosed with bipolar disorder when treated in an outpatient psychiatric clinic setting.

Better knowledge of patient characteristics, treatment, and outcomes of children diagnosed with bipolar disorder is necessary to determine whether and how services for these children might be improved in the clinical setting. To address gaps in understanding, this study reviewed all records of children less than 19 years old who received evaluation and/or treatment for bipolar disorder from an academic tertiary child and adolescent psychiatry clinic from 2006 to 2012. We sought to: (1) characterize diagnostic comorbidity in referred children with bipolar disorder, (2) characterize symptom comorbidity in these children, (3) characterize high-risk behaviors in the sample, (4) describe the lifetime SOC that these children are involved with in the real world, (5) describe the types of treatments utilized by these children in the clinic, (6) report the number of clinic visits as a proxy for engagement in treatment, (7) report clinical outcomes in these children, and (8) investigate predictors of outcome in the clinical setting. Lastly, to put the findings for this pediatric bipolar disorder cohort in a broader context, we ascertained intake characteristics and outcomes of a depressive disorders cohort seen at the clinic over the same period with the same clinical evaluation methods.

Methods

Study setting, chart review, and data collection

The child and adolescent outpatient mental health clinic at the University of Connecticut Medical School and Health Care (UCHC) serves 5- to 18-year-old children and adolescents with early-onset behavioral and mental health disorders in central Connecticut. As an academic training site, the mission of the clinic is educational as well as clinical; the site serves as the primary child ambulatory and continuity of care training clinic for child and adolescent psychiatry fellows, general psychiatry residents, and psychology interns, and it provides third-year medical students with a participatory experience in ambulatory child and adolescent psychiatry. Clinical services offered include evaluation, psychotherapy, and psychopharmacology. The clinic has ∼2000 visits and 130 new evaluations per year and is staffed by 0.4 full-time equivalent (FTE) child psychiatry faculty, 0.6 FTE child psychology faculty, and 0.4 FTE advance-practice nursing faculty. Four child psychiatry fellows and two general adult psychiatry residents see patients 1 day per week, and four doctoral child psychology trainees see patients 2 days per week under onsite faculty supervision.

Clinic charts from a total of 714 consecutively referred and unique pediatric patients evaluated and treated in the clinic from 2006 to 2012 were retrieved. A data abstraction form was created, records were hand-searched for pertinent information, and data were entered into a clinic electronic database. Two senior faculty, one a board-certified child and adolescent psychiatrist with more than 30 years of clinical experience and one doctorate-level nurse practitioner in child mental health with more than 35 years clinical experience, reviewed 100% of medical charts (D.F.C. and G.S.P.). Inter-rater reliability on a variety of outcome measures on 10% of records ranged between 0.65 and 0.90 (fair to excellent agreement) using Kendall's tau, a nonparametric measure of the agreement between rankings (Cartwright 1957).

Inclusion criteria for the current study included: (1) patients who were newly diagnosed with bipolar disorder or had their existing bipolar diagnosis confirmed by clinic evaluation (bipolar I, bipolar II, episode mixed, manic, or depressed) or patients who were newly diagnosed with unipolar depression or had their existing depressive disorder diagnosis confirmed by a clinic evaluation (major depressive disorder, dysthymia); (2) diagnosis made by a UCHC psychiatry resident, child psychiatry fellow, or child psychiatry faculty and confirmed by a faculty supervisor; (3) patients who received evaluation and/or treatment at the UCHC child and adolescent outpatient mental health clinic; and (4) patients who had either completed treatment or prematurely ended treatment at the clinic (so a course of treatment was available to study an episode of longitudinal care). Exclusion criteria included: (1) patients with “rule-out” bipolar diagnoses and/or mood disorder not otherwise specified (NOS) (DSM-IV); (2) patients with schizophrenia, intellectual disability, and autism spectrum disorders. Of the original 714 patients, 49 bipolar children and 58 children with unipolar depressive disorders met inclusion/exclusion criteria and formed the sample for this study. These records were reviewed in depth. This study was approved by the School of Medicine Institutional Review Board.

Measures

Demographics, family data, treatment type, psychoactive medication treatment, child behavioral history, and school history were obtained by either parent or child report during psychiatric interview. We ascertained the lifetime history of involvement in nine different SOC, including child welfare, juvenile justice, in-home mental health services, foster care, prior outpatient mental health (not including treatment in our clinic), emergency department visits, inpatient psychiatry admissions, primary care system, and school-based mental health services.

Procedures

Psychiatric diagnoses adhered to DSM-IV-TR criteria and were obtained by multistage clinical evaluation, including interview of the child and parent(s) together followed by interview of the child alone. Psychiatric trainees and their case faculty supervisor interviewed all patients and parents separately and together, and they independently assigned diagnoses. Psychiatric diagnoses were assigned by using the “OR” rule. If either parent or child reported clinical symptoms meeting DSM criteria for bipolar disorder or depressive disorder, the diagnosis was assigned. Agreement on this primary diagnosis between psychiatric trainee evaluator and psychiatry faculty supervisor for these 49 cases of bipolar disorder and 58 cases of unipolar depressive disorder was 100%. Comorbid symptom severity at evaluation was ascertained by using the Child Behavior Checklist Parent-Report instrument (Achenbach and Rescorla 2001). The Child Behavior Checklist (CBCL) is a 113-item psychometrically valid measure of general psychopathology that is normed for 6- to 18-year-old children. Computer scored, the CBCL yields an internalizing, externalizing, and total T-score assessing symptoms over the 6 months before evaluation.

Participants

For the entire sample (n = 107), the average age was 14.1 ± 2.86. Bipolar children were significantly younger than children with depressive disorders (Table 1). There were 60 (56%) women and 47 (44%) men in the overall sample. The bipolar group had more men, whereas the depressive disorders group had more women (see Table 1 given next for comparison). Ethnicity was 72% Caucasian American and 28% non-Caucasian American. In 43% of families, yearly income was less than $50,000. Average parental education was 13.5 ± 2.5 years, 38% of families received Medicaid, and 55% had commercial insurance. Seven percent of families had other healthcare payment arrangements. Household structure included 50% two-parent families, 34% single-parent households, and 16% other family structures. Special education services were provided to 25% of the overall sample. Almost two-thirds (n = 66, 62%) had parents who had received psychiatric services, with no difference between the bipolar and depression cohorts. Parental legal problems and suicidality were relatively uncommon, except that almost half (47%) of the bipolar patients had a parent with a substance use disorder (vs. 29% of the depression cohort) and 13% had a parent who had attempted suicide (vs. 4% of the depression cohort) (Table 1). The bipolar patients also were more likely than unipolar depression patients to be from families with low income who were receiving Medicaid, and to have received special education services (Table 1).

Table 1.

Pediatric Bipolar Disorder Versus Unipolar Depression Cohort Comparisons >0.10

Variable	Bipolar (n = 49)	Depression (n = 58)	Test	p
Intake sociodemographic variables
Age	13.2 ± 3.3	15.1 ± 2.7	t = −3.247	0.002
Male gender	30 (61.2)	17 (29.3)	X² = 10.983	0.001
Female gender	19 (38.8%)	41 (70.7%)	X² = 10.983	0.001
Income <$50k	24 (49.0%)	16 (27.6%)	X² = 5.193	0.02
Medicaid	24 (53.3%)	17 (30.4%)	X² = 7.201	0.03
Median parental education	12.9 ± 3.0	13.9 ± 2.0	t = 1.822	0.07
Household structure: two-parent	22 (44.9%)	31 (53.4%)	X² = 5.979	>0.10
Household structure: single-parent	18 (36.7%)	18 (31.1%)
Household structure: adopted/foster care	5 (10.2%)	1 (1.7%)
Household structure: other	4 (8.2%)	8 (13.8%)
Special education	19 (40%)	8 (14%)	X² = 9.753	0.008
Intake clinical variables
Comorbid ADHD	13 (26.5%)	9 (15.5%)	X² = 1.972	>0.10
Comorbid DBD	3 (6.1%)	3 (5.2%)	X² = 0.045	>0.10
Comorbid anxiety	2 (4.1%)	3 (5.2%)	X² = 0.071	>0.10
Comorbid PTSD	2 (4.1%)	1 (1.7%)	X² = 0.542	>0.10
Comorbid SUD	7 (14.3%)	10 (17.2%)	X² = 0.174	>0.10
No. of comorbid diagnoses	1.6 ± 0.9	1.5 ± 0.8	t = −0.947	>0.10
Child suicidality	5 (10.2%)	6 (10.3%)	X² = 0.001	>0.10
Child SIB	5 (10.2%)	8 (13.8%)	X² = 0.321	>0.10
Child aggression	19 (38.8%)	9 (15.5%)	X² = 7.437	0.006
CBCL total T-score	65.1 ± 8.5	63.3 ± 8.8	t = 0.937	>0.10
CBCL internalizing T-score	64.6 ± 9.1	66.4 ± 9.0	t = −0.884	>0.10
CBCL externalizing T-score	63.9 ± 10.1	60.5 ± 9.1	t = 1.603	>0.10
Parent family history: legal	4 (8.7%)	2 (3.6%)	X² = 1.198	>0.10
Parent family history: SUD	22 (46.8%)	16 (28.6%)	X² = 3.650	0.056
Parent family history: psychiatric services	30 (65.2%	33 (58.9%)	X² = 0.423	>0.10
Parent family history: completed suicide	1 (2.3%)	1 (1.8%)	X² = 0.030	>0.10
Parent family history: suicide attempt	6 (13.3%)	2 (3.6%)	X² = 3.260	0.07
Parent family history: homicide	0 (0%)	1 (1.8%)	X² = 0.812	>0.10
Service delivery and clinic treatment variables
SOC: child welfare	13 (26.5%)	11 (19.3%)	X² = 0.787	>0.10
SOC: juvenile justice	8 (16.3%)	8 (13.8%)	X² = 0.134	>0.10
SOC: in-home services	2 (4.1%)	1 (1.7%)	X² = 0.542	>0.10
SOC: foster care	6 (12.2%)	2 (3.5%)	X² = 2.882	0.09
SOC: outpatient MH services	44 (89.8%)	47 (81%)	X² = 1.603	>0.10
SOC: ED/mobile crisis team involvement	7 (14.3%)	9 (15.5%)	X² = 0.032	>0.10
SOC: inpatient MH services	13 (26.5%)	6 (10.3%)	X² = 4.765	0.03
SOC: primary care	43 987.8%)	50 (86.2%)	X² = 0.056	>0.10
SOC: school-based MH services	9 (18.4%)	9 (15.5%)	X² = 0.154	>0.10
Average no. of SOC	3.0 ± 1.3 (range: 1–5)	2.5 ± 1.1 (range: 1–4)	t = −2.170	0.03
Evaluation for second opinion only	9 (18.4%)	9 (15.5%)	X² = 1.923	>0.10
Clinic treatment: psychotherapy only	10 (20.4%)	13 (22.4%)	X² = 0.063	>0.10
Clinic treatment: psychopharmacology only	13 (26.5%)	10 (17.2%)	X² = 1.358	>0.10
Clinic treatment: combined therapy + meds	17 (34.7%)	26 (44.8%)	X² = 1.135	>0.10
Psychopharmacology monotherapy	13 (26.5%)	10 (17.2%)	X² = 1.656	>0.10
Psychotherapy monotherapy	10 (20.4%)	13 (22.4%)	X² = 0.008	>0.10
Combined therapy + medications	17 (34.7%)	26 (44.8%)	X² = 1.135	>0.10
Antipsychotics	19 (38.8%)	1 (1.7%)	X² = 23.992	<0.001
Stimulants	6 (12.2%)	2 (3.4%)	X² = 2.971	0.09
SSRI	6 (12.2%)	27 (46.6%)	X² = 14.656	<0.001
Mood stabilizers	13 (26.5%)	0 (0%)	X² = 17.516	<0.001
Alpha-agonists	5 (10.2%)	0 (0%)	X² = 6.208	0.01
Two or more meds concurrently	21 (42.9%)	7 (21.1%)	X² = 13.031	<0.001
No. of medications	1.4 ± 1.3	0.59 ± 0.6	t = −4.157	<0.001
No. of clinic visits	11.7 ± 13.1 (range: 1–73)	9.6 ± 12.0 (range 1–45)	t = −0.684	>0.10
Premature dropout	22 (44.9%)	33 (56.9%)	X² = 1.531	>0.10
Clinical outcomes
Initial CGI-S	4.70 ± 1.04	4.38 ± 0.95	t = −1.657	>0.10
Final CGI-S	3.74 ± 1.31	3.24 ± 1.19	t = −2.060	0.04
Final CGI-I	3.17 ± 1.21	2.68 ± 1.12	t = −2.008	0.05
Initial CGAS	50.7 ± 7.24	52.9 ± 6.67	t = 1.384	>0.10
Final CGAS	55.7 ± 10.38	60.1 ± 8.84	t = 2.329	0.02
Responder status^a	15 (40.5%)	26 (53.1%)	X² = 2.146	>0.10

Responder status: final CGI-I score ≤2.

ADHD, attention-deficit/hyperactivity disorder; CBCL, Child Behavior Checklist; CGAS, Children's Global Assessment Scale; CGI-I, Clinical Global Impressions-Improvement; CGI-S, Clinical Global Impressions-Severity; DBD, disruptive behavior disorders; ED, Emergency Department; MH, mental health; SIB, self-injurious behaviors; PTSD, posttraumatic stress disorder; SOC, systems-of-care; SSRI, selective serotonin reuptake inhibitor; SUD, substance use disorders.

All patients met DSM-IV-TR criteria for either bipolar disorder or unipolar depression. Unfortunately, due to limitations in electronic data base coding, bipolar subtypes could not be ascertained. The clinic prevalence rate of bipolar disorder was 6.9% (49/714). Very early-onset bipolar disorder, defined as disorder onset ≤12 years, was present in 22% (n = 11) of the sample (prevalence rate 1.5%). Comorbid psychiatric diagnoses in the bipolar cohort included attention-deficit/hyperactivity disorder (ADHD, 27%), substance abuse (14%), disruptive behavior disorders (6%), anxiety (4%), and posttraumatic stress disorder (PTSD) (4%). The depression cohort had a similar profile of comorbidities (Table 1). The average numbers of comorbid diagnoses for the bipolar (1.6 ± 0.9) and depression (1.5 ± 0.8) cohorts were almost identical. Child high-risk behaviors included suicidality (10%) (no known deaths by suicide had occurred), and nonsuicidal self-injury (10%–14%) was comparable for the cohorts. However, the bipolar cohort was more likely than the depression cohort to have a history of maladaptive aggression (39% vs. 15.5%; Table 1). High rates of comorbid symptom severity as measured by parent report on the CBCL were found for both cohorts (Table 1).

Outcome measures

Based on considerations of cost, patient diagnostic complexity, and patient and clinician time, effort, and workflow, the child and adolescent outpatient clinic uses two broad-based clinician-reported outcome measures assessed at each patient visit. The Clinical Global Impressions-Severity (CGI-S) Scale (Guy 1976) is a brief clinician-rated global measure of transdiagnostic symptom severity. The scale is one item asking the clinician based on their total clinical experience to rate how mentally ill the child has been over the past week by using a 7-point Likert scale from normal to among the most severely ill. The CGI-S is rated by the clinician using all available information and rating relative to baseline (first clinical contact) at each visit. The CGI-Improvement (CGI-I) Scale (Guy 1976) is a clinician-rated scale of symptom change (improvement/deterioration) on a 7-point Likert scale from very much improved to very much worse. The CGI-I is rated at every visit, beginning with the second session relative to baseline (first clinical contact). The CGI scales are extensively used in clinical treatment trials and in naturalistic effectiveness studies (Kwan and Rickwood 2015). The CGI scales are reliable, valid, and sensitive to change, and they are identified as suitable for routine clinical use due to their brevity, reliability, and ease of administration (Busner and Targum 2007; Berk et al. 2008; Forkmann et al. 2011). A third outcome measure utilized is the Children's Global Assessment Scale (CGAS) (Shaffer et al. 1983), which is a unidimensional clinician-completed rating scale of global daily functioning and impairment assessed on a 100-point scale. The CGAS is reliable, valid, and sensitive to change (Gold et al. 2009; Lundh et al. 2010). It is completed at baseline (evaluation) and at treatment termination or at the last evaluable visit. The CGAS is well established in clinical settings and widely used (Lundh et al. 2010).

Rater training and initial calibration for reliability

Trainees and staff undergo an annual 4-hour group training in application and scoring of the CGI and CGAS. Training consists of lectures, readings (Shaffer et al. 1983; Busner and Targum 2007), and scoring of two case vignettes led by a senior faculty member. Trainee and faculty scores are compared with scores of a senior faculty member with more than 30 years of clinical experience with these measures (D.F.C.). After scoring a first (practice) clinical vignette, raters present their rating scores to the group, which then discusses rating similarities and differences compared with each other and with the senior rater. They then rate a second (accuracy) vignette. Raters are scored as accurate for rating if they score ±1 point on the CGI and ±5 points on the CGAS compared with the senior rater (Lundh et al. 2010). Over the past 8 years, 83 independent raters have rated 16 written vignettes on the CGI and CGAS. Clinician agreement on the second vignette (accuracy rating) with a senior rater is 92% (CGI-S), 81% (CGI-I), and 84% (CGAS). For the purposes of this study, two senior faculty reviewed all charts. Trainee CGI and CGAS ratings were accepted on review by senior staff. If disagreement on ratings arose, the senior staff rating was accepted.

Data analysis

Descriptive data are reported as percentiles, means, and standard deviations. Responder status was defined a priori as a last evaluable clinic visit CGI-I score ≤2 (improved or very much improved). Change on the CGI-S and CGAS from intake to the final clinic visit was tested with paired-sample t-tests. Bivariate tests of associations between potential predictors of outcome and CGI improvement ratings at the last evaluable clinic visit were conducted with independent-samples t-tests or one-way analysis of variance with post hoc Scheffe tests. Predictors of outcome were analyzed on a multivariate basis by multiple linear regression. Comparisons between the bipolar and depression cohorts utilized chi-square tests for dichotomous variables and t-tests for continuous variables. The alpha coefficient for statistical significance was set at p ≤ 0.05.

Results

High rates of lifetime outpatient mental health services were found for our sample, and about one in four bipolar children already had experienced a psychiatric inpatient admission (number of admissions: 1–11). Although most bipolar and depressive disorders children were identified by the primary care system, 12% to 14% had no identifiable primary care provider. The average number of lifetime SOC for our bipolar sample was 3.0 ± 1.3 (range 1–5). In comparison with the depressive disorders group, bipolar children had significantly higher lifetime SOC service utilization, particularly involving higher likelihood of having received inpatient psychiatric services (see Table 1 and Fig. 1).

FIG. 1.

History of lifetime participation in nine systems-of-care for 50 bipolar children and adolescents. *p ≤ 0.05; ^t p ≤ 0.10. MH, mental health.

Bipolar patients received a variety of treatments in the clinic, including evaluation for second opinion only (18%), psychotherapy only (20%), psychopharmacology only (27%), and combined therapy and medications (35%). Of the 41 patients who received clinic treatment (excluding those who were evaluated for second opinion only), 75% received psychopharmacological treatments and 68% received some type of psychotherapy. Types of medications used included second-generation antipsychotics (52%), first-generation antipsychotics (2%), mood stabilizers (lithium, lamotrigine, lithium, oxcarbamazepine) in 36%, antidepressants (32%), stimulants and atomoxetine (20%), and alpha-agonists (guanfacine, clonidine) in 12%. Of those bipolar patients receiving prescribed medications, 43% were taking ≥2 medications concurrently, with an average number of psychiatric medications of 1.4 ± 1.3 (range 1–4). Although the bipolar and depression cohorts did not differ in terms of general type of treatment received, the bipolar cohort was more likely to have received multiple medications of all types except that they were only marginally more likely to have received stimulants (Table 1).

We used the number of clinic visits as a proxy for engagement in treatment (Shim et al. 2017). The average number of clinic visits was 11.7 ± 13.1 (range 1–73 visits). Premature treatment dropout occurred in 45% of the sample. The bipolar and depression cohorts were comparable on both the number of clinic visits and the likelihood of premature dropout from treatment.

Clinical outcomes

For outcomes analysis, we excluded patients who were seen in the clinic only for a single second opinion evaluation (n = 9 in each cohort). Patients seen for a single evaluation were slightly higher functioning than children who received treatment at the clinic (average initial CGI-S score equal to 4.38 ± 1.41 vs. 4.77 ± 0.96, respectively, and average initial CGAS score equal to 53.1 ± 8.40 vs. 50.7 ± 7.24, respectively [higher CGAS scores indicate better daily functioning]).

For the 41 bipolar disorder patients who received treatment, the mean final CGI-S was 3.74 ± 1.31 (lower scores indicate less severity). A CGI-S score paired-samples t-test was completed, t = 4.71, df = 38, p < 0.001 (two cases with missing data) indicating a statistically significant improvement in the CGI-S as rated by clinicians. The final mean CGAS score for the bipolar cohort was 55.7 ± 10.38. A CGAS paired-samples t-test was completed, t = −3.55, d = 39, p = 0.001 (one case with missing data) indicating improvement in daily impairment with clinic treatment. The mean CGI-I at final visit for the bipolar was 3.16 ± 1.19 (minimally improved). Using an a priori definition of responder as a clinician-rated CGI-I score ≤2 (improved or very much improved) at the last evaluable clinic visit, 40.5% (n = 15 of 37 bipolar disorder patients [four cases with missing data]) were responders to treatment. The final CGI and CGAS outcomes for unipolar depression patients differed statistically significantly from those for the bipolar cohort, reflecting better clinical status and more improvement at outcome for the depression patients (Table 1). However, the depression patients were not more likely than the bipolar patients to be treatment responders (Table 1).

Predictors of outcome in the bipolar disorder cohort

Bivariate correlations between the outcome measures (discharge CGI-I; discharge CGAS) and potential predictors of outcome were calculated for: (1) age at admission, (2) CBCL total T-score, (3) initial CGI-S score, (4) initial CGAS score, (5) discharge CGAS score, (6) total number of SOC, and (7) total number of clinic visits completed (Table 1). Some bivariate relationships were found. CGI-I Score at final visit correlated strongly with the final CGAS score, the CBCL total T-score completed at admission, and the number of clinic appointments completed. The final CGAS score was significantly related to the number of clinic appointments completed. Lifetime number of SOC was unrelated to any of the other study variables.

Considering demographic and diagnostic variables that might be associated with outcome, on a bivariate basis, gender was not associated with CGI improvement (t[41] = −0.79, p = 0.43; M[SD] = 3.33 ± 1.33 for girls vs. 3.04 ± 1.09 for boys) at discharge. Minority ethnocultural background was not associated with CGI improvement (t[41] = 0.17, p = 0.87; M[SD] = 3.22 ± 0.97 for minority ethnicity youth vs. 3.15 ± 1.26 for non-Hispanic White children) at discharge. Living in a two-parent family was not associated with CGI improvement (t[40] = −0.13, p = 0.89; M[SD] = 3.10 ± 1.18 for children in two-parent families vs. 3.14 ± 1.19 for children not in two-parent families). ADHD comorbidity was not associated with CGI improvement (t[41] = 1.09, p = 0.28; M[SD] = 2.78 ± 0.97 for children diagnosed with ADHD vs. 3.26 ± 1.24 for children not diagnosed with ADHD) at discharge.

Considering treatment variables that might be associated with outcome, on a bivariate basis type of treatment was not associated with CGI improvement (F[2,34] = 1.23, p = 0.30; M[SD] = 3.31 ± 1.25 for medication only vs. 2.43 ± 0.98 for psychotherapy only vs. M[SD] = 3.00 ± 1.23 for medication plus psychotherapy) at discharge. Treatment with mood stabilizer medication was not associated with CGI-I (t[35] = −1.74, p = 0.09; M[SD] = 3.43 ± 1.4 for mood stabilizer medication vs. 2.74 ± 1.01 for children not receiving mood stabilizer medication) at discharge. Treatment with atypical antipsychotic medication was not associated with CGI-I (t[35] = 0.54, p = 0.59; M[SD] = 2.89 ± 1.1 atypical medication vs. 3.11 ± 1.3 for children not receiving atypical medication) at discharge. Treatment with two or more medications was not associated with CGI improvement (t[35] = 0.000, p = 1.00; M[SD] = 3.00 ± 1.2 for children receiving two or more medications vs. 3.00 ± 1.24 for one or no medications) at discharge.

Based on evidence from prior studies of relationships between age, dose of treatment, initial psychopathology severity, and extensiveness of system-of-care involvement with child psychiatric treatment outcome, a linear regression analysis, including those independent variables, was conducted with CGI-I as the outcome indicator (n = 25). Although SOC was not significant in the bivariate analyses (Table 2), we thought it important to include in the regression because prior research has suggested its importance for childhood development (de Voursney and Huang 2016; Black et al. 2017). Table 3 presents the results of the regression analysis, which resulted in a multiple R = 0.561 and R² = 0.315. Age and the number of clinic visits completed were unrelated to improvement, but both the CBCL total T-score and the number of lifetime SOC for the child were statistically significantly and inversely associated with improvement (higher CGI-I scores indicating less improvement).

Table 2.

Correlations of Predictor and Outcome Variables in the Bipolar Disorder Cohort (n = 25)

	CGI-I score at discharge	Age at admission	CBCL total T-score	Initial CGI-S score	Initial CGAS score	Discharge CGAS score	Total SOC	No. of clinic appointments
CGI-I score at discharge	1	−0.111	0.369^a	−0.006	−0.75	−0.705^b	0.119	−0.309^a
Age at admission		1	−0.175	−0.111	0.294^a	0.381^b	0.094	0.056
CBCL total T-score			1	0.247	−0.032	−0.508^b	−0.017	−0.051
Initial CGI-S score				1	−0.747^b	−0.335^a	0.167	0.003
Initial CGAS score					1	0.427^b	−0.203	0.172
Discharge CGAS score						1	−0.109	0.308^a
Total SOC							1	−0.013

p ≤ 0.05.

p ≤ 0.01.

CGAS, Children's Global Assessment Scale; CGI, Clinical Global Impressions.

Table 3.

Linear Regression Testing Predictors of Clinical Global Impressions Improvement (n = 25)

Variable	Unstandardized B	Std. error	Standardized beta	t	p
Age at admission	0.123	0.109	0.234	1.130	0.272
No. of clinic appointments	−0.010	0.032	−0.064	−0.332	0.743
CBCL total T-score	0.067	0.032	0.416	2.106	0.048
Total number SOC (lifetime)	0.474	0.220	0.436	2.156	0.043

CBCL, Child Behavior Checklist; SOC, systems-of-care.

Discussion

To facilitate a better understanding of children and adolescents who are assigned a bipolar diagnosis by clinicians, and to describe what happens to them in the clinic after diagnosis, we assessed the characteristics, service delivery history, treatment, outcomes, and predictors of clinician-rated outcomes in a small sample of bipolar children and adolescents from a tertiary academic mental health clinic. To better appreciate the data in perspective, we also compared the bipolar group with a depressive disorders group evaluated from the same clinic over the same period. In the discussion that follows, we focus on the bipolar cohort's characteristics and outcomes, and we include findings from the depression cohort to provide a broader perspective on this child psychiatry clinic's patients and outcomes.

Similar to previous early-onset bipolar research, high rates of comorbid diagnoses and symptoms were found. The most common comorbid diagnosis in our bipolar sample was ADHD in 27%, which is similar to rates in some pediatric bipolar studies (35%) (Keenan-Miller et al. 2012) but lower than that reported for other early-onset bipolar treatment studies (93%) (Vitiello et al. 2012). Lack of the availability of gold-standard structured diagnostic interviewing in our clinic may reduce clinician recognition of comorbid diagnoses. In addition, clinicians may have placed diagnostic emphasis on the episodic nature of mood and irritability changes in making the diagnosis of bipolar disorder, as advocated by the American Academy of Child and Adolescent Psychiatry (AACAP) (McClellan et al. 2007). This is similar to the concept of a narrow phenotype of severe mood dysregulation emphasizing episodic rather than nonepisodic symptoms, which is believed to be more associated with bipolar disorder than with ADHD (Leibenluft 2011). Rates of ADHD and other comorbid diagnoses in the bipolar group did not significantly differ when compared with the depressive disorders group.

High-risk behaviors were common in our bipolar sample, including suicidality, nonsuicidal self-injury, and substance use. However, no deaths were reported. High rates of these behaviors were also common in the families of bipolar children (by parental report), including substance use, arrest, suicide, and suicide attempts in first-degree relatives. These results are similar to findings from previous studies showing high rates of substance abuse in bipolar adolescents (Wilens et al. 2014) and suicidal ideation and attempts in pediatric bipolar disorder (Hauser et al. 2013). For perspective, rates of child suicidality and self-injurious behaviors did not differ between groups. Across both groups, these high-risk behaviors were prevalent (10% to 13%). There was a trend for parents of bipolar children to report more stress (e.g., substance abuse, suicide attempts) than parents of depressive disorder children.

High rates of parent-reported aggressive behavior occurred in our bipolar sample. Indeed, reported aggression was significantly higher in the bipolar than in the depressive disorders group. Aggression is associated with impairment in family and interpersonal functioning in bipolar youth (Keenan-Miller et al. 2012), and it is more common in early-onset than adult bipolar disorder (Safer et al. 2012). High levels of aggression in children and adolescents with bipolar disorder may be an important predictor of variation in pharmacological treatment response, predicting response to atypical antipsychotics over divalproex (West et al. 2011; Masi et al. 2015).

Children in our bipolar sample had high rates of lifetime SOC involvement. Compared with the depressive disorders group, the bipolar group had significantly higher lifetime SOC involvement, which was largely accounted for by high rates of inpatient psychiatric hospitalization and a trend for higher foster care use in the bipolar sample. To the best of our knowledge, SOC utilization has not previously been reported for bipolar children and adolescents from a real-world clinical setting. Our sample consisted of high lifetime users of multiple services. Special education services, child welfare, prior mental health services including ambulatory and inpatient psychiatry, and emergency psychiatric care utilization were frequent for our sample. Of note, bipolar children who were involved in a higher number of systems (lifetime) had poorer outcomes, even after accounting for the strong association between admission internalizing and externalizing problem severity and poorer outcome. Thus, along with the high degree of affective and behavioral dysregulation that is characteristic of children with bipolar disorder (Geller et al. 2012; Dvir et al. 2014), involvement in multiple SOC may reflect an additional level of adversity affecting these children and their families, which may require more intensive cross-system collaborative care than is typically provided in child psychiatric outpatient services. Addressing the needs of both the bipolar group and the depressive disorder group, the need for such collaborative case coordination is underscored by the finding that slightly more than 1 in 10 of bipolar and depressed children had no identifiable primary care provider. The absence of coordination of care across multiple systems could have serious adverse consequences for the health and development of these children.

Children in the bipolar group received significantly more antipsychotics, mood stabilizers, alpha-agonists, and combined pharmacotherapy, and significantly less selective serotonin reuptake inhibitors (SSRIs) than the depressed group. Consistent with best practices (Pfeifer et al. 2010; Geller et al. 2012; Kowatch et al. 2013), treatment in the bipolar group emphasized evidence-based psychopharmacology, with many children also receiving adjunctive psychotherapy. Currently, the FDA approves six medications for the treatment of bipolar disorder in adolescents: aripiprazole, asenapine, olanzapine, risperidone, quetiapine, and lithium. A recent randomized, controlled trial supports the clinical use of lithium in the acute treatment of bipolar disorder in children aged 7–17 years old (Findling et al. 2015). However, no medications are currently approved for very early-onset bipolar disorder in children aged <12 years. We also found that 43% of our bipolar sample received psychopharmacological regimens with the use of two or more concurrent medications compared with only 21% in the depressed sample. Currently, there are no randomized clinical trials supporting combined pharmacotherapy in pediatric bipolar disorders. However, high rates of diagnostic comorbidity in pediatric bipolar disorders may necessitate rational combined pharmacotherapy regimens in this population (Kowatch et al. 2013).

A growing literature suggests that psychosocial interventions are important to provide families with an understanding of symptoms, course of illness, and treatment of early-onset bipolar spectrum disorders (Fristad and MacPherson 2014). Recent randomized clinical trials support both child- and family-focused cognitive-behavioral therapy for pediatric bipolar disorder (West et al. 2014) and dialectical behavior therapy for adolescents with bipolar disorder (Goldstein et al. 2015). Combined pharmacotherapy and psychotherapy treatment may be most important for the bipolar child with severe illness or high rates of comorbid disorders (McClellan et al. 2007).

However, in our sample, clinical improvement as measured by CGI-I scores did not vary by type of treatment provided. We did not find any difference in outcome based on whether bipolar children received medication monotherapy, psychotherapy monotherapy, or combined treatment with both medications and psychotherapy. The reasons for this are somewhat unclear, but our results could be confounded by small sample size (a methodological issue) or by family treatment choice (a clinical issue). Further, in regression analysis, two predictors of outcome in the bipolar group were found, including the total number of lifetime SOC and the parent-reported total T-score at initial evaluation. More severely impaired bipolar youth with a lifetime history of high SOC utilization may need more extensive or intensive interventions than were provided in the current clinic utilizing only three different treatment types. Further research with a larger sample size and more stringent controls is needed to clarify this issue.

Few studies report outcomes for bipolar youth in the clinical setting. Using the clinician-rated CGI-S, CGI-I, and CGAS paired samples testing suggested a significant treatment effect in our bipolar sample. Without random assignment and a control condition, this finding is only suggestive and in need of replication with more rigorous methodology. Nevertheless, it is important to report treatment effects from real-world settings to better understand treatment optimization for these children. However, using an a priori definition of response as a last-visit CGI-I ≤2, less than half of our sample met response criteria. This suggests the need to develop more effective treatments for pediatric bipolar disorder that can easily translate into the clinical setting.

Although CGI-S and CGAS scores did not significantly differ at initial evaluation between the two groups of bipolar and depressive disorders children, our comparison data suggest that in our clinic children with depressive disorders may be more treatment responsive than children with bipolar disorders. We found a significant difference across groups in final CGI-S, CGI-I, and CGAS scores favoring the depressed group. Although not reaching statistical significance, more depressed children achieved responder status (53%) than did bipolar children (41%). However, this result is only suggestive given limitations in our study methodology and requires replication.

High rates of premature dropout from treatment were found for both the bipolar and depression groups. In the bipolar group, bivariate analyses number of clinic appointments kept was associated with improved bipolar outcomes on the CGI-I and CGAS scales. To the extent that number of clinic visits is a proxy for treatment engagement (Shimet al. 2017), strategies to address high dropout rates from clinical treatment may improve outcomes.

With measurement-based practice and electronic health record capability, many clinics can do similar analyses such as this, perhaps as part of a quality improvement project. Describing the characteristics, symptom severity, comorbidity, specific treatments, and outcomes of patients referred to one's clinic is useful in facilitating knowledge about the real-world practice of clinical child and adolescent psychiatry.

Limitations

The study has several limitations. The sample size was small and was drawn from a single academic child clinical psychiatry site. Thus, the findings may not generalize to other groups of bipolar and depressive disorders youth. As noted earlier, without randomization or a control condition, outcome findings remain suggestive and not definitive. We did not measure treatment outcome by using a more specific rating scale for pediatric bipolar disorder such as the Young Mania Rating Scale or a validated scale for depression severity. Because of the pressures and limitations of clinic work, a more global clinician-rated outcome measure was used. Due to technical difficulties, bipolar subtypes are not reported and narrow-band CBCL T-scores are not available. Nevertheless, results from our single-site sample do suggest that evidence-based treatment demonstrates effectiveness for some bipolar children, although less than for children with depressive disorders, in the real world of the clinic.

Clinical Significance

Early-onset bipolar disorder is a complex and heterogeneous psychiatric disorder with high rates of diagnostic and symptom comorbidity. Individual clinical evaluation should be comprehensive. Evidence-based treatment should emphasize psychopharmacology with adjunctive family and individual psychotherapy. Bipolar youth with more complex and/or severe disorder may require rational combined psychopharmacology regimens. Communication across multiple SOC may be necessary to better integrate care for children and adolescents with early-onset bipolar disorders. Strategies to improve engagement in treatment may be especially important. Given high rates of high-risk behaviors in these youth, regular mental health follow-up to assess safety is important.

Conclusion

Further research to identify additional evidence-based treatments for pediatric bipolar disorder is needed.

Authors' Contribution

Statistical analyses: D.F.C. and J.D.F.

Footnotes

Disclosures

D.F.C. is a consultant to Shire, Rhodes, Supernus, and Arbor Pharmaceuticals and the OAK Group of Wellesley, MA. He receives grant support from Shire Pharmaceuticals and NIMH. He receives royalties from WW Norton Co. J.D.F., V.L.S., and A.D. have no disclosures to report.

References

Achenbach

, Rescorla

: Manual for the ASEBA School-Age Forms & Profiles. Burlington, VT, ASEBA, 2001.

Berk

, Ng

, Dodd

, Callaly

, Campbell

, Bernardo

, Trauer

: The validity of the CGI severity and improvement scales as measures of clinical effectiveness suitable for routine clinical use. J Eval Clin Pract, 14:979–983, 2008.

Black

, Walker

, Fernald

, Andersen

, DiGirolamo

, Lu

, McCoy

, Fink

, Shawar

, Shiffman

, Devercelli

, Wodon

, Vargas-Baron

, Grantham-McGregor

; Lancet Early Childhood Development Series Steering Committee: Early childhood development coming of age: Science through the life course. Lancet, 389:77–90, 2017.

Busner

, Targum

: The clinical global impressions scale: Applying a research tool in clinical practice. Psychiatry (Edgmont), 4:28–37, 2007.

Cartwright

: A note concerning Kendall's tau. Psychol Bull, 54:423–425, 1957.

de Voursney

, Huang

: Meeting the mental health needs of children and youth through integrated care: A systems and policy perspective. Psychol Serv, 13:77–91, 2016.

Dvir

, Ford

, Hill

, Frazier

: Childhood maltreatment, emotional dysregulation, and psychiatric comorbidities. Harv Rev Psychiatry, 22:149–161, 2014.

Evans-Lacko

, dosReis

, Kastelic

, Riley

: Mental health service use before and after diagnosis of early-onset bipolar disorder. J Behav Health Serv Res, 38:398–413, 2011.

Findling

, Robb

, McNamara

, Pavuluri

, Kafantaris

, Scheffer

, Frazier

, Rynn

, DelBello

, Kowatch

, Rowles

, Lingler

, Martz

, Anand

, Clemons

, Taylor-Zapata

: Lithium in the acute treatment of bipolar I disorder: A double-blind, placebo-controlled study. Pediatrics, 136:885–894, 2015.

10.

Forkmann

, Scherer

, Boecker

, Pawelzik

, Jostes

, Gauggel

: The clinical global impression scale and the influence of patient or staff perspective on outcome. BMC Psychiatry, 11:83, 2011.

11.

Fristad

, MacPherson

: Evidence-based psychosocial treatments for child and adolescent bipolar spectrum disorders. J Clin Child Adolesc Psychol, 43:339–355, 2014.

12.

Geller

, Luby

, Joshi

, Wagner

, Emslie

, Walkup

, Axelson

, Bolhofner

, Robb

, Wolf

, Riddle

, Birmaher

, Nusrat

, Ryan

, Vitiello

, Tillman

, Lavori

: A randomized controlled trial of risperidone, lithium, or divalproex sodium for initial treatment of bipolar I disorder, manic or mixed phase, in children and adolescents. Arch Gen Psychiatry, 69:515–528, 2012.

13.

Geller

, Tillman

, Bolhofner

, Zimerman

: Child bipolar I disorder: Prospective continuity with adult bipolar I disorder; characteristics of second and third episodes; predictors of 8-year outcome. Arch Gen Psychiatry, 65:1125–1133, 2008.

14.

Gold

, Buonopane

, Caggiano

, Picciotto

, Vogeli

, Kanner

, Babcock

, Li

, Jellinek

, Drubner

, Sklar

, Murphy

: Assessing outcomes in child psychiatry. Am J Manag Care, 15:210–216, 2009.

15.

Goldstein

, Fersch-Podrat

, Rivera

, Axelson

, Merranko

, Yu

, Brent

, Birmaher

: Dialectical behavior therapy for adolescents with bipolar disorder: Results from a pilot randomized trial. J Child Adolesc Psychopharmacol, 25:140–149, 2015.

16.

Guy

: Assessment Manual for Psychopharmacology. Washington, DC, U.S. Department of Health, Education, and Welfare, 1976.

17.

Hauser

, Galling

, Correll

: Suicidal ideation and suicide attempts in children and adolescents with bipolar disorder: A systematic review of prevalence and incidence rates, correlates, and targeted interventions. Bipolar Disord, 15:507–523, 2013.

18.

Keenan-Miller

, Peris

, Axelson

, Kowatch

, Miklowitz

: Family functioning, social impairment, and symptoms among adolescents with bipolar disorder. J Am Acad Child Adolesc Psychiatry, 51:1085–1094, 2012.

19.

Kessler

, Avenevoli

, Green

, Gruber

, Guyer

, He

, Jin

, Kaufman

, Sampson

, Zaslavsky

: National comorbidity survey replication adolescent supplement (NCS-A): III. Concordance of DSM-IV/CIDI diagnoses with clinical reassessments. J Am Acad Child Adolesc Psychiatry, 48:386–399, 2009.

20.

Kowatch

, Youngstrom

, Horwitz

, Demeter

, Fristad

, Birmaher

, Axelson

, Ryan

, Frazier

, Arnold

, Young

, Gill

, Findling

: Prescription of psychiatric medications and polypharmacy in the LAMS cohort. Psychiatr Serv, 64:1026–1034, 2013.

21.

Kwan

, Rickwood

: A systematic review of mental health outcome measures for young people aged 12 to 25 years. BMC Psychiatry, 15:279–298, 2015.

22.

Leibenluft

: Severe mood dysregulation, irritability, and the diagnostic boundaries of bipolar disorder in youths. Am J Psychiatry, 168:129–142, 2011.

23.

Lundh

, Kowalski

, Sundberg

, Gumpert

, Landen

: Children's Global Assessment Scale (CGAS) in a naturalistic clinical setting: Inter-rater reliability and comparison with expert ratings. Psychiatry Res, 177:206–210, 2010.

24.

Masi

, Milone

, Stawinoga

, Veltri

, Pisano

: Efficacy and safety of risperidone and quetiapine in adolescents with bipolar II disorder comorbid with conduct disorder. J Clin Psychopharmacol, 35:587–590, 2015.

25.

McClellan

, Kowatch

, Findling

: Practice parameter for the assessment and treatment of children and adolescents with bipolar disorder. J Am Acad Child Adolesc Psychiatry, 46:107–125, 2007.

26.

Moreno

, Laje

, Blanco

, Jiang

, Schmidt

, Olfson

: National trends in the outpatient diagnosis and treatment of bipolar disorder in youth. Arch Gen Psychiatry, 64:1032–1039, 2007.

27.

Olfson

, Crystal

, Gerhard

, Huang

, Carlson

: Mental health treatment received by youths in the year before and after a new diagnosis of bipolar disorder. Psychiatr Serv, 60:1098–1106, 2009.

28.

Peele

, Axelson

, Xu

, Malley

: Use of medical and behavioral health services by adolescents with bipolar disorder. Psychiatr Serv, 55:1392–1396, 2004.

29.

Pfeifer

, Kowatch

, DelBello

: Pharmacotherapy of bipolar disorder in children and adolescents: Recent progress. CNS Drugs, 24:575–593, 2010.

30.

Safer

, Magno Zito

, Safer

: Age-grouped differences in bipolar mania. Compr Psychiatry, 53:1110–1117, 2012.

31.

Shaffer

, Gould

, Brasic

, Ambrosini

, Fisher

, Bird

, Aluwahlia

: A children's global assessment scale (CGAS). Arch Gen Psychiatry, 40:1228–1231, 1983.

32.

Shim

, Compton

, Zhang

, Roberts

, Rust

, Druss

: Predictors of mental health treatment seeking and engagement in a community mental health center. Community Ment Health J, 53:510–514, 2017.

33.

Vande Voort

, Singh

, Bernardi

, Wall

, Swintak

, Schak

, Jensen

: Treatments and services provided to children diagnosed with bipolar disorder. Child Psychiatry Hum Dev, 47:494–502, 2016.

34.

Van Meter

, Burke

, Kowatch

, Findling

, Youngstrom

: Ten-year updated meta-analysis of the clinical characteristics of pediatric mania and hypomania. Bipolar Disord, 18:19–32, 2016.

35.

Vitiello

, Riddle

, Yenokyan

, Axelson

, Wagner

, Joshi

, Walkup

, Luby

, Birmaher

, Ryan

, Emslie

, Robb

, Tillman

: Treatment moderators and predictors of outcome in the Treatment of Early Age Mania (TEAM) study. J Am Acad Child Adolesc Psychiatry, 51:867–878, 2012.

36.

West

, Weinstein

, Celio

, Henry

, Pavuluri

: Co-morbid disruptive behavior disorder and aggression predict functional outcomes and differential response to risperidone versus divalproex in pharmacotherapy for pediatric bipolar disorder. J Child Adolesc Psychopharmacol, 21:545–553, 2011.

37.

West

, Weinstein

, Peters

, Katz

, Henry

, Cruz

, Pavuluri

: Child- and family-focused cognitive-behavioral therapy for pediatric bipolar disorder: A randomized clinical trial. J Am Acad Child Adolesc Psychiatry, 53:1168–1178, 1178.e1161, 2014.

38.

Wilens

, Yule

, Martelon

, Zulauf

, Faraone

: Parental history of substance use disorders (SUD) and SUD in offspring: A controlled family study of bipolar disorder. Am J Addict, 23:440–446, 2014.