Pharmacotherapy and Family-Focused Treatment for Adolescents With Bipolar I and II Disorders: A 2-Year Randomized Trial

The trial was conducted from August 2006 to July 2010 at the University of Colorado, the University of Pittsburgh School of Medicine, and the Cincinnati Children’s Hospital Medical Center. Referrals originated from community practitioners, inpatient and outpatient units, advertisements, and presentations or discussion forums. Inclusion criteria were age between 12 years and 18 years, 1 month; a DSM-IV-TR diagnosis of bipolar I or II disorder based on consensus ratings of separate Schedule for Affective Disorders and Schizophrenia for School-Age Children–Present and Lifetime Version (K-SADS-PL) (14, 15) interviews of the youth and at least one parent, with a manic, hypomanic, or mixed episode lasting at least 1 week or a major depressive episode lasting at least 2 weeks within the previous 3 months; a diagnosis of bipolar I or II disorder by a board-certified psychiatrist, based on a separate evaluation of the child and parent(s); symptoms of at least moderate severity (a score ≥17 on the K-SADS Mania Rating Scale [16] or a score ≥16 on the Depression Rating Scale [14, 15]) for at least 1 week of the previous month; willingness to proceed with pharmacotherapy from a study psychiatrist; and at least one parent willing to participate in family sessions. Participants were excluded if they met DSM-IV-TR criteria for current substance use or a pervasive developmental disorder or were victims of current physical or sexual abuse. Youths who were ineligible for the study were referred to appropriate clinical services.

Participants provided written informed consent after receiving a complete description of the study. The study was approved by the human subject review boards of all three institutions.

Participants were prescreened by telephone; those who appeared eligible based on parents’ reports were invited for an initial visit in which the K-SADS-PL was administered. The instrument’s mood modules were replaced with the K-SADS Mania Rating Scale and Depression Rating Scale, which rate symptoms on 6- or 7-point scales of severity and impairment. Symptom severity was rated for both the most symptomatic week in the past month (baseline rating) and the most symptomatic week in the child’s lifetime (see the data supplement that accompanies the online edition of this article). Reliabilities (intraclass r values) across the three sites (12 K-SADS-PL tapes rated by an average of 12 raters each) were 0.89 for Depression Rating Scale scores and 0.81 for Mania Rating Scale scores.

Once participants were enrolled in the trial, a data manager at the Pittsburgh site using a modification of Efron’s biased coin toss (17) randomly assigned them in a 50-50 proportion to pharmacotherapy plus family-focused treatment or pharmacotherapy plus enhanced care. The groups were balanced on study site, bipolar subtype (I or II), and index episode polarity (depressed, mixed, manic/hypomanic).

Patients in both treatment conditions were pharmacologically managed by board-certified psychiatrists who were supervised monthly by expert pharmacologists (R.A.K., M.P.D., D.A.A.). Pharmacological treatments were standardized using the algorithms of the Child Psychiatric Workgroup on Bipolar Disorder (18) as updated periodically based on literature reviews (e.g., reference 19). The treatment of adolescents in manic, hypomanic, or mixed phases generally began with a second-generation antipsychotic (quetiapine, risperidone, or aripiprazole) or a mood stabilizer (lithium or valproate). Dosages were titrated to clinical response or maximum tolerable dosage. In cases of partial response, treatment was augmented with a combination of a second-generation antipsychotic and a mood stabilizer. Patients with acute bipolar depression had their mood stabilizer and antipsychotic dosages maximized. If they did not respond, a second mood stabilizer (e.g., lamotrigine) or antipsychotic was added (further details are provided in the online data supplement).

Two pharmacotherapy supervisors (D.A.A. and M.P.D.) who were unaware of psychosocial treatment conditions classified 57%–70% of the pharmacotherapy sessions as fully adherent to the study guidelines, 17%–27% as partially adherent, and 12%–20% as nonadherent (based on 973 ratings across the three sites). Nonadherence often reflected the choices of patients or families regarding specific agents or dosages.

Participants in family-focused treatment (patients, parents, and when possible, siblings) were provided 21 family sessions of 50 minutes each over 9 months (12 weekly, then six biweekly, and then every 3 months) in three consecutive modules: psychoeducation, communication enhancement training, and problem-solving skills training (see the online data supplement for further descriptions). The three weekly enhanced care sessions were based on an abbreviated version of the family-focused psychoeducation manual and focused on monitoring of moods and relapse prevention planning. In both conditions, participants could receive “booster” sessions (i.e., nonprotocol individual or family crisis management) as needed throughout the study.

All clinicians underwent training in both interventions during a 2-day pretrial workshop and received monthly group teleconference supervision during the trial. Treatment fidelity in both conditions was tracked using the 13-item Therapist Competency and Adherence Scale (11, 20). Mean overall fidelity ratings (interrater reliability=0.84), which could range from 1 (nonadherent) to 7 (excellent adherence), did not differ significantly across sites, treatment conditions, or the interaction of sites with treatment conditions (the mean score for the sample was 5.4 [SD=1.3], based on 145 ratings).

Independent evaluators who were unaware of treatment assignments interviewed patients and one parent every 3 months during year 1 and every 6 months during year 2 using the Adolescent Longitudinal Interval Follow-up Evaluation (4, 21), with mania/hypomania and major depression assessed on the Psychiatric Status Rating Scales for each week of the previous 13- or 26-week interval. These weekly ratings clarified the timing of recovery (primary outcome) and recurrence events. Cross-site reliability of the 6-point Psychiatric Status Rating Scale scores was 0.74 (intraclass r) for agreement on the highest weekly depression or mania/hypomania scores (see the online data supplement for further details). Evaluators made separate ratings of the worst week of the previous month using the K-SADS Depression and Mania Rating Scale.

We compared the two treatment groups on all demographic, illness history, and medication variables listed in Table 1. Site and pretreatment variables that were imbalanced or that predicted earlier study termination were included as covariates in the statistical models. Time to recovery was calculated as the number of weeks from randomization until weekly Psychiatric Status Rating Scale scores were ≤2 for at least 8 consecutive weeks for depression, mania/hypomania, or both. Time to recurrence was computed as weeks from the point of recovery to the point when the patient had Psychiatric Status Rating Scale scores ≥5 for at least 1 week for mania/hypomania or at least 2 weeks for depression. All randomized patients were included in the at-risk sample; those who did not recover and those who left the study early were censored at the point of their final research assessment. Survival curves for time to recovery or recurrence were compared across groups using the Kaplan-Meier product-limit equation (22). Cox proportional hazards models (23) were used to control for covariates (site, baseline symptom severity and polarity, and weeks of follow-up, as well as any variables identified in the attrition analyses). With a sample size of 145, a two-sided log-rank test had 90% power to detect a 25% difference in survival proportions, assuming 20% attrition (p<0.05).

As in our first trial (11), we compared the groups on proportion of weeks well (weeks with all Psychiatric Status Rating Scale scores ≤2 divided by number of weeks followed) and proportion of weeks depressed (Psychiatric Status Rating Scale depression scores ≥5) or manic/hypomanic (Psychiatric Status Rating Scale mania or hypomania scores ≥5) in each study year. We excluded participants who had ≤6 months of total follow-up. Data were analyzed with mixed-effects models (using Proc Mixed in SAS/STAT [24]) with time (year 1 or 2) as the within-subject factor, treatment group as the between-subject factor, and a group-by-time interaction, along with baseline covariates and a subject-level random effect to account for correlations between the repeated measurements. With 75 participants per group, the design had >90% power to detect a group-by-time interaction corresponding to a change in effect size of d=0.56 (p<0.05, two-sided) from baseline to end of study.

Next, to examine more fine-grained patterns of symptom changes over the 2 years, we fitted analogous mixed models with mean scores for depression or mania/hypomania (on the Psychiatric Status Rating Scale or the K-SADS Depression or Mania Rating Scale), calculated for each 3-month study epoch in months 0–24, as the outcomes. Finally, exploratory general linear mixed models (24) were used to examine whether the groups differed in patterns of medication use (mood stabilizers, second-generation antipsychotics, and antidepressants) over time.

Participants were 145 adolescents with bipolar I disorder (N=77) or bipolar II disorder (N=68) (mean age, 15.6 years, SD=1.4). A total of 37 patients (25.5%) entered in a bipolar I or II depressed episode, 72 (49.7%) in a manic or hypomanic episode, and 36 (24.8%) in a mixed or subthreshold mixed (e.g., hypomanic for ≥1 week and depressed for <2 weeks) episode. None of the participants met the DSM-IV-TR course specifier for rapid cycling. The 72 participants in family-focused treatment did not differ significantly from the 73 in enhanced care on any of the variables listed in Table 1. The 145 participants did not differ significantly in sex, age, or race/ethnicity from the 168 screened individuals who were ineligible for the study (Figure 1). (Site differences are explored in the online data supplement.)

Of the 145 participants, 123 (84.8%) had at least one follow-up interview. This proportion did not vary significantly across treatment condition, site, sex, bipolar subtype, or index episode polarity. However, participants who terminated early had higher baseline K-SADS Mania Rating Scale scores than those who were followed for 2 years (F=5.57, df=1, 143, p=0.02). Participants who lived with both biological parents stayed in the study longer (mean=92.2 weeks, SD=28.3) than those who lived with one biological parent or in another living situation (mean=77.0 weeks, SD=35.9) (F=6.07, df=1, 121, p=0.02). Study site, living situation, baseline K-SADS Mania and Depression Rating Scale scores, and number of weeks of follow-up were included as covariates in the statistical models.

During the 2-year study, 87 of the 123 (70.7%) patients for whom follow-up data were available met the 8-week recovery criteria from the index episode; the Kaplan-Meier estimate of cumulative probability of recovery was 87.1% (median time to recovery, 38 weeks, 95% CI=33–46). The 2-year Kaplan-Meier estimate of recovery in family-focused treatment was 85.2% (median time to recovery, 41 weeks, 95% CI=26–54) and in enhanced care, 88.7% (median=36 weeks, 95% CI=25–48; hazard ratio=1.21). In a Cox model that included all covariates, higher baseline K-SADS Depression Rating Scale scores (χ²=11.37, df=1, p<0.001) and lower Mania Rating Scale scores (χ²=6.63, df=1, p=0.01) were associated with longer time to recovery; the two treatment arms, however, did not differ in recovery time. (See the online data supplement for survival analyses using a briefer period [≥4 weeks] to define recovery.)

The family-focused treatment and enhanced care groups also did not differ in weeks to recovery from baseline depressive symptoms or from baseline manic/hypomanic symptoms. Furthermore, there was no interaction between treatment group and baseline illness polarity in any time-to-event analysis. Within the mixed (N=36) subgroup, the median time to recovery from the index episode was 42 weeks (95% CI not calculable) for patients in family-focused treatment and 40 weeks (95% CI=23–62) for patients in enhanced care, a nonsignificant difference.

Among the 87 patients who recovered, 50 (57.5%) had a recurrence during the 2-year study. The Kaplan-Meier estimate of cumulative probability of recurrence was 65.1%, with a median time to recurrence of 37 weeks (95% CI=18–55). Of the 50 participants with recurrences, eight had manic, 11 had hypomanic, three had mixed, and 28 had depressive recurrences. Of 39 participants in family-focused treatment who recovered, 25 had recurrences (Kaplan-Meier probability estimate, 74.4%; median time to recurrence, 27 weeks, 95% CI=16–55). Of 48 in enhanced care who recovered, 25 had recurrences (Kaplan-Meier estimate, 58.8%; median time to recurrence, 37 weeks, 95% CI not calculable). In a Cox model that included all demographic and baseline illness scores as covariates, treatment condition was not associated with time to any recurrence, time to depression recurrence, or time to manic or hypomanic recurrence.

Adolescents who lived with both biological parents had a longer time to manic recurrence than those living in another arrangement (χ²= 3.69, df=1, p=0.05). There were no interactions of treatment group, site, and living situation on time to recovery or recurrence.

Across sites, patients in family-focused treatment attended a mean of 15.4 (SD=7.3) of 21 expected protocol therapy sessions and a mean of 2.0 (SD=5.5) extra (nonprotocol) sessions. In enhanced care, patients attended a mean of 2.5 (SD=1.1) of three expected protocol sessions and 0.8 (SD=2.3) nonprotocol sessions. The number of nonprotocol sessions did not differ across treatment conditions or sites. Inclusion of a therapy contact ratio (number of psychosocial contacts to number of expected sessions in each treatment) in the survival models did not alter the null effects of treatment condition on time to recovery or recurrence.

There were no treatment group differences in the percentage of weeks free of mood symptoms (Psychiatric Status Rating Scale scores ≤2) or the percentage of weeks with acute mood symptoms (Psychiatric Status Rating Scale scores ≥5) across study years 1 and 2 (Table 2). There were also no main effects of treatment group or treatment group-by-time (year 1, year 2) interactions on percentage of weeks with depressive symptoms. However, adolescents in family-focused therapy showed a greater increase from year 1 to year 2 in proportion of weeks without mania/hypomania symptoms compared with those in enhanced care (F=4.02, df=1, 87, p=0.048).

Patients in family-focused treatment did not differ in mean Psychiatric Status Rating Scale scores for depression across the 3-month study epochs. However, patients in family-focused treatment showed greater improvements in mean Psychiatric Status Rating Scale scores for mania/hypomania across 3-month intervals than did patients in enhanced care (F=1.98, df=8, 742, p=0.046) (Figure 2). This treatment-by-time interaction remained significant when covarying site, baseline K-SADS Mania Rating Scale score, and living situation. There was no treatment-by-baseline K-SADS Mania Rating Scale score interaction on Psychiatric Status Rating Scale scores across the 3-month epochs, and there were no treatment effects or treatment-by-time interactions on K-SADS Depression or Mania Rating Scale scores across the study epochs.

Patients with comorbid anxiety disorders (N=57; Table 1) had earlier mood recurrences (median=19 weeks, 95% CI=13–42) than patients without anxiety disorders (median=67 weeks, 95% CI not calculable) (χ²=6.62, df=1, p=0.01). Patients with comorbid attention deficit hyperactivity disorder (ADHD) (N=48) had earlier mood recurrences (median=17 weeks, 95% CI=11–25) than patients without ADHD (N=97; median=53 weeks, 95% CI not calculable) (χ²=4.39, df=1, p=0.04). In a Cox model that included treatment group, site, living situation, and baseline symptom scores, only concurrent anxiety disorders remained strongly associated with time to mood recurrence (χ²=7.65, df=1, p=0.006).

The number of pharmacotherapy visits during the trial (mean=11.6, SD=6.9) did not differ between treatment groups and did not predict time to recovery or recurrence. Participants in the two groups did not differ in the mean number of medications prescribed at baseline (family-focused treatment: mean=2.0, SD=1.0; enhanced care: mean=1.6, SD=0.9) or at the 12-month or 24-month follow-ups (Table 3). The two groups also did not differ in the likelihood of taking lithium (compared with other mood stabilizers), second-generation antipsychotics (considered individually or as a class), or adjunctive antidepressants, psychostimulants, or anxiolytics at any point in the trial, or in having a mood stabilizer or antidepressant added or discontinued (Table 3). When dichotomous variables indicating the presence or absence of each class of medications were included in Cox survival models, no effects emerged for psychosocial treatments on time to recovery or recurrence.