Abstract
Objective:
Childhood-onset depression is associated with increased risk of recurrent depression and high morbidity extending into adolescence and adulthood. This multi-site randomized controlled trial evaluated two active psychosocial treatments for childhood depression: Family Focused Treatment for Childhood Depression (FFT-CD); and individual supportive psychotherapy (IP). Aims were to describe effects through 52 weeks post-randomization on measures of depression, functioning, non-depressive symptoms, and harm events.
Methods:
Children meeting criteria for depressive disorders (N = 134) were randomly assigned to 15 sessions of FFT-CD or IP and evaluated at mid-treatment for depressive symptoms and fully at roughly 16-weeks (after acute-treatment), 32-weeks, and 52-weeks/one-year. See: clinicaltrials.gov: NCT01159041.
Results:
Analyses using generalized linear mixed models confirmed the previously-reported FFT-CD advantage on rates of acute depression response (≥50% Children’s Depression Rating Scale reduction). Improvements in depression and other outcomes were most rapid during the acute-treatment period, and leveled off between week-16 and 52, with a corresponding attenuation of observed group differences, although both groups showed improved depression and functioning over 52-weeks. Survival analyses indicated that most children recovered from their index depressive episodes by week-52: estimated 76% FFT-CD, 77% IP. However, by the week-52 assessment one FFT-CD child and six IP children had suffered recurrent depressive episodes. Four children attempted suicide, all in the IP group. Other indicators of possible harm were relatively evenly distributed across groups.
Conclusions:
Results indicate a quicker depression response in FFT-CD and hint at greater protection from recurrence and suicide attempts. However, outcomes were similar for both active treatments by week-52/one-year. Although community care received after acute-treatment may have influenced results, findings suggest the value of a more extended/chronic disease model that includes monitoring and guidance regarding optimal interventions when signs of depression-risk emerge.
Keywords: Depression, treatment trials, psychotherapy, outcome, family therapy
Introduction
Depression is common, estimated to affect 350,000,000 people and to be the leading cause of total years lost to disability worldwide (WHO, 2012). Although relatively rare in childhood, depressive disorders become more prevalent during adolescence, and are associated with disruptions in development and impaired functioning (Avenevoli, Swendsen, He, Burstein, & Merikangas, 2015; Luby, Barch, Whalen, Tillman, & Freedland, 2018). The limited data on depression in children indicate that most recover over time, but recurrence after recovery is common and estimated to occur in up to 72% of children over 15 years of follow-up (Kovacs, Obrosky, & George, 2016). Adult outcomes of children with depressive disorders tend to be poor, with elevated rates of suicide attempts, substance abuse, and conduct problems (Kovacs et al., 2016; Weissman et al., 1999).
Despite the general preference for psychosocial treatment over medications for children (Tarnowski, Simonian, Bekeny, & Park, 1992), few studies have examined the comparative efficacy of psychosocial treatments for childhood depressive disorders (Luby et al., 2018). There are some promising treatments for children with depressive disorders evaluated only in open trials (Kovacs et al., 2016), and randomized controlled trials (RCTs) evaluating psychosocial interventions for children selected for elevated depressive symptoms (for review, Weisz, McCarty, & Valeri, 2006). However, results from sub-clinical samples may not generalize to samples with depressive disorders.
Four RCTs examined treatments for depressive-disorders in school-age children. One small treatment-development RCT found significant reductions in depressive symptoms and higher remission rates at post-treatment with family-based interpersonal psychotherapy, relative to child-centered supportive therapy (Dietz, Weinberg, Brent, & Mufson, 2015). Another found that children with depression or bipolar disorders receiving multifamily psychoeducation groups plus usual care (UC) had reduced mood symptoms compared to waitlist controls plus usual-care over 12-months (Fristad, Verducci, Walters, & Young, 2009). A third found an advantage of brief behavioral therapy, compared to assisted referral, among children and adolescents with depressive or anxiety disorders (Weersing, Brent, et al., 2017). A pilot trial with a mixed child/adolescent sample found that individual family-education, as a monotherapy or combined with omega-3 fatty acids, had similar impacts on depressive severity trajectories to placebo alone over 12-weeks (Fristad, Vesco, Young et al., 2016).
This paper reports on the first large multisite randomized controlled trial comparing two active psychosocial treatments for depressive disorders in children (ages 7–14). We compared a developmentally-informed family-focused treatment targeting childhood depression (FFT-CD) to individual supportive psychotherapy (IP). Rooted in family and cognitive-behavioral models of change, FFT-CD aimed to strengthen family processes to support recovery. Building on our previously published findings indicating that FFT-CD children showed significantly higher depression-response rates than IP children after acute-treatment (primary RCT outcome), this secondary report examines longer term outcomes over 52-weeks/one-year. Specific aims are to describe: patterns of depression response, recovery, and recurrence; functioning and non-affective symptom outcomes; and harm events. Based on evidence from adolescent depression trials that change occurs in a piecewise fashion with the most rapid improvement observed by treatment end-point, followed by stabilization and/or recurrence during follow-up (Asarnow et al., 2009; Goodyer et al., 2017; Vitiello et al., 2011), we predicted a significant time effect, with a stronger FFT-CD advantage on depression response following acute treatment, and lower recurrence risk among FFT-CD children.
Method
Previous reports on this single-blind RCT describe the design, methods, and acute-treatment results (Tompson, Langer, Hughes, & Asarnow, 2017; Tompson, Sugar, Langer, & Asarnow, 2017); an overview is provided below. Institutional Review Boards at both sites and a Data Safety and Monitoring Board approved and oversaw study procedures. Informed consent/assent was obtained for all participants. See: clinicaltrials.gov: NCT01159041.
Participants
Children were recruited (N = 134) between February 2011 and January 2014 from two sites in large metropolitan areas in the western and northeastern U.S. Inclusion criteria: current diagnosis of major depression (MDD), dysthymic disorder (DD), or Depressive Disorder-Not Otherwise Specified (DD-Nos); age 7–14; if taking medication, stable dose (≥ 3 months); living with participating parent. Exclusion criteria: symptoms/problems that would interfere with participation in treatment or assessments (e.g., psychotic disorder). These criteria represent expansions of the original age range (7–12) and diagnosis (MDD/DD) to enhance recruitment/feasibility.
Trial design
Eligible participants were randomly assigned to FFT-CD or IP (1:1 ratio) using a computerized algorithm with a random mix of blocks lengths (2 and 4). Children were stratified by site, gender, baseline depression diagnosis (MDD/DD versus DD-Nos) and presence/absence of antidepressant medication treatment. Treatment allocation was concealed from assessment and enrollment staff. Both intervention conditions offered comparable treatment exposure: 15 sessions over ≤ 22 weeks; the same therapists; and up to 3 booster sessions of the randomized intervention (FFT-CD or IP) following the end of acute-treatment.
FFT-CD.
FFT-CD aimed to enhance developmentally appropriate family interpersonal processes that can promote recovery, buffer children from the adverse impacts of stress and distress, and enhance protective processes (Tompson, Langer, et al., 2017). Sessions emphasized: 1) strengthening family relationships and fostering supportive parent–child interactions (e.g. “family thanks notes” to increase supportive communication, active listening, communication, pleasant activities); and 2) boosting skills for effective coping and emotion regulation (e.g. problem-solving, helpful thoughts, behavioral activation). Through psychoeducation, role-plays, and practice/homework, families were assisted in identifying helpful/upward interaction spirals and unhelpful/downward spirals and using skills to improve mood.
IP.
The study IP used a manualized supportive approach (adapted from Cohen & Mannarino, 1996). This individual treatment focused on helping children better understand their feelings through empathic listening and a safe supportive relationship. An initial parent session and brief monthly supportive meetings were allowed.
Therapist adherence and competence were monitored on randomly-selected sessions, using a 7-point scale (7 = highest adherence/competence). Results indicated high competence (FFT-CD: Mean 6.43, SD 1.10; IP: Mean 7.00, SD .00) and adherence (FFT: Mean 6.34, SD 1.44; IP: Mean 6.57, SD 1.38), with strong differentiation between treatments (low IP adherence ratings for FFT-CD sessions, Mean 1.03, SD .25; low FFT-CD adherence for IP sessions, Mean 1.07, SD .26). High IP competence ratings for FFT-CD sessions (Mean 6.95, SD .17) indicated strong therapist warmth and genuineness across conditions.
Assessments/outcomes
Full assessments were scheduled at baseline, at end of acute-treatment/week-16, week-32, and week-52 (February 2011-January 2015). Assessors had graduate education in a mental health field, were naive to intervention assignment, and received regular supervision. Inter-rater reliability was strong for all interviewer measures: Intraclass Correlation Coefficients .77–.94; Kappa Depression Diagnosis .91 (Tompson, Sugar et al., 2017).
Depression.
As reported previously (Tompson, Langer et al., 2017), the primary study outcome was depression response at post-treatment, defined as ≥ 50% reduction from the baseline score on the interviewer-rated Children’s Depression Severity-Revised Scale (CDRS-R; Poznanski, Hartmut, Grossman, & Freeman, 1985), based on combined child and parent report. Other depression outcomes included: depression remission, defined as CDRS-R ≤ 28; depression severity measured by the total CDRS-R Score; and self-reported depressive symptoms on the Children’s Depression Inventory (CDI; Kovacs, 1992) completed by children (CDI-C) and parents (CDI-P) mid-treatment at weeks 5 and 10. Depression recovery and recurrence after recovery were assessed with the Schedule for Affective Disorders and Schizophrenia for School-Aged Children (K-SADS-PL, Kaufman et al., 1997) administered to children and parents. A life chart was used to define dates of recovery and recurrent depressive episodes following recovery from the index episode. Recovery was defined based on the date when children no longer met depressive disorder criteria, and recurrence as the time when children met criteria for a new depressive episode (MDD or DD) following recovery. When the exact date was uncertain, we identified the time interval when symptoms emerged and dated recovery/recurrence at the midpoint (Kovacs et al., 2016).
Functioning.
The interviewer-rated Children’s Global Adjustment Scale (C-GAS; Shaffer et al., 1983) and child-reported Social Adjustment Scale for Children - Self-Report (Weissman, Orvaschel, & Padian, 1980) measured functioning.
Non-depressive symptoms.
Child-reported anxiety symptoms were assessed using the Multidimensional Anxiety Scale for Children (MASC; March, Parker, Sullivan, Stallings, & Conners, 1997). Parent-reported internalizing, externalizing, and total problems were assessed on the Child Behavior Checklist (CBCL; Achenbach, 1991).
Possible harm events.
Possible harm events were assessed using IRB adverse and clinical incident reports plus the K-SADS-PL supplemented by CDRS-R and our versions of the Columbia Suicide Severity Rating Scale (C-SSRS) and the Suicide and Self-Harm History Interview (SSI), which provided information on dates and self-harm classification (suicide attempt or non-suicidal self-injury (NSSI), Asarnow, Hughes, Babeva, & Sugar, 2017).
Non-study services.
The parent-reported Child and Adolescent Services Assessment(CASA; Ascher, Farmer, Burns, & Angold, 1994) assessed ED visits, hospitalizations, and non-study services. Medication treatment rates from CASA parent-report are used for consistency across study time-points and are slightly lower than those based on multiple data-sources used for randomization reported previously (Tompson et al., 2017).
Statistical analysis
This report describes pre-planned analyses of outcomes over the 52-week study observation period, supplementing our prior publication describing acute-treatment effects. The goal of this secondary exploratory report is description of the outcome trajectory patterns over the 52-week follow-up. We used generalized linear mixed models (GLMMs) (logistic link for binary outcomes; identity link for continuous outcomes) with subject level random effects to account for correlations induced by the repeated measures data structure. For binary outcomes that were defined post-baseline (depression response, remission), there were three repeated measurements corresponding to planned visits at 16, 32, and 52 weeks. For continuous outcomes, baseline was included as a time-point, yielding four repeated measurements (6 for the CDI-C/CDI-P). Models used actual time from entry (continuous measure) rather than treating the planned visit times as categorical. They allowed for a change in slope at the end of acute treatment, and included the interactions of treatment with each of these two time components. This resulted in a piecewise linear fit within each of the treatment arms. For each of the continuous outcomes, we first performed an omnibus two-degree of freedom test for an overall difference in the treatment trajectories (i.e. we tested jointly whether the two group by time interaction terms were significant). To further characterize the longitudinal pattern of treatment effects, when the omnibus test reached p<.05, post-hoc contrasts examined each component of the trajectory separately within and between groups. Specifically, we assessed change within each of the acute and follow-up periods, along with change in slope from the acute to follow-up period. We present estimated slope coefficients (parameterized as weekly rates of change) for each treatment arm, along with effect sizes using Cohen’s f2, the standard metric for regression and related modeling types, with f2 = .02, .15 and .35 corresponding to small, medium and large effects respectively (Cohen, 1988).
Analyses that consider the full longitudinal outcome trajectories have embedded in them the original acute treatment comparisons. However, because the models are using additional time points, the standard error estimates of the key parameters may be slightly different and imposing the piece-wise linear structure may affect the fit at the end of acute treatment. We, therefore, felt it was worthwhile to affirm that the model augmentation did not in any way affect the major findings from the previously published report of acute outcomes (Tompson et al., 2017).
To assess whether longer-term trajectory patterns could have been affected by post-intervention treatment-seeking or individual provider characteristics, we conducted sensitivity analyses including therapist effects (both main effects and interactions with treatment group) and treatment usage variables (receipt of study-treatment booster sessions, medications, non-study therapy after acute-treatment as time-varying covariates) in the longitudinal models. As this did not change the observed pattern of treatment effects and usage of non-study adjunctive treatments was similar in the FFT-CD and IP groups (see Table S1 in the Supporting Information), we restrict the manuscript presentation to the primary analyses.
Survival analyses examined between group differences in time to recovery and recurrence post-recovery. Based on our prediction that treatment effects would be strongest acutely, we employed Wilcoxon tests which are more sensitive than the traditional log-rank procedure when the ratio of hazards is higher earlier in the follow-up window. GLMMs and survival analyses automatically account for missing data and censoring respectively, producing unbiased estimates assuming observations are missing at random or the censoring mechanism is non-informative.
Our basic analytic approach called for considering as covariates variables that (a) were used for stratification, or (b) were significantly associated with loss to follow-up. (The latter was necessary as our GLMM approach will produce biased parameter estimates unless observations are missing at random.) There were four stratification variables used in this study: site, gender, baseline depression diagnosis (MDD or DD versus DD-NOS), and presence/absence of anti-depression medication treatment. We included as covariates baseline factors associated with incomplete longitudinal data, as determined by logistic regression. These were: single vs dual parent status; baseline antidepressant medication; and DD-NOS versus MDD or DD. Following current standards, no other covariates were included in the primary models (Kraemer, 2015).
Statistical analyses were conducted using SAS 9.4 (SAS Institute Inc., 2017). We note that the primary RCT outcome was 16-week depression response and the study was designed to have 80% power to detect differential effects acute (Tompson et al., 2017); the trajectory analyses in this report, describing changes in outcome over the full 52 weeks, were considered exploratory. Additionally, the initial omnibus tests in the GLMMs provide some protection against multiple comparisons for the post-hoc contrasts; thus we report unadjusted p-values, although we note that some of the observed effects would survive a formal correction for multiple comparisons.
Results
Among the 134 children enrolled, 119 (88.8%) had at least one post-baseline evaluation: 116 (86.6%) post-treatment/16-week; 97 (72.4%) 32-week; 100 (74.6%) 52-week (Figure 1). The baseline sample had a mean age of 10.84 (SD 2.09), was 56% female and racially and ethnically diverse (49% endorsed minority race or Hispanic/Latino ethnicity). (Table S2). At baseline, 89 children (66%) met MDD criteria, 24 (18%) met DD criteria, 7 (5%) met double depression (MDD plus DD) criteria, and 14 (11%) met DD-NOS criteria. Comorbidity was common: disruptive behavior disorders, n = 56 (42%); anxiety disorders, n = 50 (37%). Treatment groups did not differ on acute treatment dose (Mean #acute treatment sessions FFT-CD, 11.25, SD 4.97; IP 11.78, SD 4.64, t(132) = −.59, p = .55). Booster sessions delivered in the child’s randomized arm were received by 28 children (20.9%): 10 FFT-CD 14.9%; 18 IP 26.9%, χ2 = 2.89, p = .09 (Table S1 for detail by visit).
Figure 1. Participant Flow.
*Defined as ≥11 sessions
**Range 1–3
Ancillary/non-study treatment
Based on CASA parent-report, during the trial 25.6% of children received medication treatment (22.4%,15/67 FFT-CD; 28.8%, 19/66 IP); 12.0% received antidepressant treatment (10.45%, 7/67 FFT-CD; 13.6%, 9/66 IP). After acute treatment, 29.0% (31/107) received non-study psychotherapy (26.5% (13/49) FFT-CD, 31% (18/58) IP, χ2(1) = .26, p = .61, Table S1 for detail by visit).
Depression outcomes
Tables 1 and S3–4 show raw frequencies for binary outcomes and means for continuous outcomes respectively. Table 2 provides results of GLMM models. Although we show the best estimates of the (weekly) rates of change in the outcomes for each treatment arm and study phase for descriptive purposes, inferential comparisons between groups are only valid in the presence of a significant differential trajectory finding (group by time interaction, weeks 16–52, for binary outcomes; omnibus test for continuous outcomes). Figure 2 illustrates these model estimates of trajectories over time for the CDRS-R and CDI-C as representative examples. Note that in each case the FFT-CD group appears to improve more rapidly than the IP group over the acute treatment period; post-treatment rates of change are less rapid in both arms, although the leveling off is more pronounced in the FFT-CD group, leading to an attenuation of the estimated treatment effects over follow-up.
Table 1.
Binary Depression Outcomes (Raw counts)
| Week-16 | Week-32 | Week-52 | |||||
|---|---|---|---|---|---|---|---|
| N | Freq (%) | N | Freq (%) | N | Freq (%) | ||
| Response: CDRS-R ≥ 50% Reduction | FFT | 54 | 43 (79.63) | 45 | 37 (82.22) | 46 | 31 (67.39) |
| IP | 62 | 37 (59.68) | 52 | 36 (69.23) | 54 | 41 (75.93) | |
| Remission: CDRS-R ≤ 28 | FFT | 54 | 29 (53.70) | 45 | 21 (46.67) | 46 | 24 (52.17) |
| IP | 62 | 22 (35.48) | 52 | 22 (42.31) | 54 | 33 (61.11) | |
Abbreviations: CDRS-R = Children’s Depression Severity-Revised Scale.
Table 2.
Rates of change in the FFT and IP groups over each study period and corresponding group differences, adjusted for site, gender, baseline-diagnostic-status, baseline-medication, family-composition.
| Acute Treatment Period Baseline to Week-16 |
Follow-up Period 16–52 Weeks |
Change from Acute to Follow-up | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Slope | F | p-value | f2 | Slope | F | p-value | f2 | Δ Slope | F | p-value | f2 | |
| CDRS-R Response | ||||||||||||
| FFT | −.021 | 2.56 | .112 | .013 | ||||||||
| IP | .025 | 4.26 | .041 | .022 | ||||||||
| FFT-IP | −.046 | 6.58 | .012 | .034 | ||||||||
| CDRS-R Remission | ||||||||||||
| FFT | .0003 | 0.00 | .982 | .000 | ||||||||
| IP | .028 | 7.11 | .009 | .037 | ||||||||
| FFT-IP | −.027 | 3.28 | .072 | .017 | ||||||||
| CDRS-R | ||||||||||||
| Trajectory-Test | F=2.30, p-value=.103, f2=.015 | |||||||||||
| FFT | −1.41 | 208.95 | <.0001 | .676 | −.029 | 0.49 | .485 | .002 | 1.38 | 117.61 | <.0001 | 0.381 |
| IP | −1.17 | 164.88 | <.0001 | .534 | −.140 | 14.93 | .0001 | .048 | 1.03 | 79.07 | <.0001 | 0.256 |
| CDI-C | ||||||||||||
| Trajectory-Test | F=2.28, p-value=.104, f2=.009 | |||||||||||
| FFT | −.482 | 64.53 | <.0001 | .128 | −.020 | 0.70 | .404 | .001 | .462 | 37.26 | <.0001 | .074 |
| IP | −.311 | 29.71 | <.0001 | .059 | −.070 | 9.19 | .003 | .018 | .241 | 11.25 | .0009 | .022 |
| CDI-P | ||||||||||||
| Trajectory Test | F=2.44, p-value=.089, f2=.010 | |||||||||||
| FFT | −.557 | 114.63 | <.0001 | .226 | −.027 | 1.62 | .205 | .003 | .530 | 65.60 | <.0001 | .129 |
| IP | −.498 | 100.78 | <.0001 | .199 | −.090 | 19.39 | <.0001 | .038 | .408 | 42.13 | <.0001 | .083 |
| CGAS | ||||||||||||
| Trajectory Test | F=0.18, p-value=.836, f2=.001 | |||||||||||
| FFT | .670 | 53.78 | <.0001 | .183 | .121 | 9.11 | .003 | .031 | −.549 | 21.04 | <.0001 | .072 |
| IP | .658 | 55.87 | <.0001 | .190 | .098 | 6.47 | .012 | .023 | −.560 | 23.60 | <.0001 | .080 |
| SAS-SR | ||||||||||||
| Trajectory Test | F=2.36, p-value=.096, f2=.016 | |||||||||||
| FFT | −.372 | 21.89 | <.0001 | .075 | .019 | 0.33 | .568 | .001 | .391 | 14.40 | .0002 | .050 |
| IP | −.220 | 8.35 | .005 | .029 | −.082 | 6.18 | .014 | .021 | .139 | 1.94 | .165 | .007 |
| MASC | ||||||||||||
| Trajectory Test | F = 3.21, p-value=.042, f2=.022 | |||||||||||
| FFT | −.101 | 0.59 | .444 | .002 | −.131 | 5.50 | .020 | .019 | −.030 | 0.01 | .859 | 0.00003 |
| IP | −.297 | 5.62 | .019 | .018 | −.223 | 17.24 | <.0001 | .059 | .074 | 0.20 | .652 | 0.001 |
| FFT-IP | .196 | 1.17 | .280 | .004 | .092 | 1.42 | .235 | .005 | −.104 | 0.19 | .660 | 0.001 |
| CBCL Internalizing | ||||||||||||
| Trajectory Test | F=1.56, p-value=.212, f2=.011 | |||||||||||
| FFT | −.642 | 74.97 | <.0001 | .267 | −.047 | 2.22 | .138 | .008 | .594 | 37.50 | <.0001 | .133 |
| IP | −.485 | 46.76 | <.0001 | .166 | −.119 | 15.20 | .0001 | .054 | .366 | 15.48 | .0001 | .055 |
| CBCL Externalizing | ||||||||||||
| Trajectory Test | F=2.32, p-value = .101, f2=.017 | |||||||||||
| FFT | −.434 | 37.15 | <.0001 | .132 | −.021 | 0.47 | .492 | .002 | .413 | 19.62 | <.0001 | .070 |
| IP | −.224 | 10.80 | .002 | .038 | −.082 | 7.89 | .006 | .028 | .142 | 2.51 | .115 | .009 |
| CBCL Total Problems | ||||||||||||
| Trajectory Test | F=2.80, p-value=.063, f2=.020 | |||||||||||
| FFT | −.580 | 63.02 | <.0001 | .224 | −.036 | 1.31 | .254 | .005 | .544 | 32.35 | <.0001 | .115 |
| IP | −.358 | 26.11 | <.0001 | .093 | −.123 | 16.78 | <.0001 | .060 | .234 | 6.51 | .012 | .023 |
Note: Binary outcomes were evaluable only post-baseline. Slopes represent change in log odds (binary) or number of points increased (+) or decreased (−) (continuous) per week. Between-group differences (computed as FFT-CD – IP) were included for continuous measures only if the omnibus tests for differential trajectories were significant; negative slope differences indicate an FFT-CD advantage, except for C-GAS where higher scores reflect better functioning. Slope differences between acute and follow-up periods, were computed as (Follow-up – Acute); positive values correspond to a leveling off of change during the follow-up period. Effect sizes are shown as Cohen’s f2, the standard metric for regression/related models; f2 = .02, .15 and .35 correspond to small, medium, and large effects respectively. Degrees of freedom: numerator = 2 for trajectory tests and 1 for all other tests; denominator differed due to missing data (CDRS-R Binary 192; Total 309; MASC 295; SAS 290; C-GAS 294; CBCL 282; and larger number of assessment points for the CDI-C 506; and CDI-P 507).
Figure 2. Mixed Model Analyses: CDRS-R and CDI-C from Baseline to Week-52.1.
1Plots show the estimated marginal means for each treatment group over time, based on the mixed models adjusted for site, gender, baseline diagnostic status, baseline medication usage and family composition.
Consistent with our previous report (Tompson et al., 2017), mixed effects logistic regression analyses confirmed a significant intervention effect on the primary binary outcome, depression response at post-treatment (week-16 group difference contrast: F1,192 = 7.09, p = .009).3 There was a statistically significant treatment by time interaction over the follow-up (difference in slopes −0.046, F1,192 = 6.58, p = .012), corresponding to an attenuation of the acute-treatment effect, whereby the IP group continued to improve slightly (slope 0.025, F1,192 = 4.26, p =.041) while the FFT-CD group if anything showed slight (non-significant) deterioration (slope −0.021, F1,192 = 2.56, p = .112). A similar but weaker pattern emerged for CDRS-R-remission (slopes FFT-CD 0.0003, IP 0.028, difference −0.027). Note that for the binary outcomes, the slopes correspond to (weekly) changes in log-odds of response or remission and can be exponentiated to obtain the corresponding odds ratios. For depression response, we have an odds ratio of 0.979 in the FFT-CD group and 1.025 in the IP group, meaning the odds of depression response go down by 2.1% per week in the former and up by 2.5% in the latter.
For the CDRS Total Score, there was a bigger estimated initial improvement in the FFT-CD group than the IP group (depression scores decreased by 1.41 vs 1.17 points per week respectively over the acute period), followed by an attenuation of that effect as the IP group continues to improve, albeit more slowly, while the FFT-CD group levels off (slopes −.029 vs −.140 over weeks 16–52). However, the omnibus test for differential trajectories is not significant. (See Table 2 and Figure 2.) The pattern of findings was similar for the CDI-C and CDI-P.4
Thus, on all three of the continuous depression measures we saw highly significant improvements in both study arms over the acute treatment period (Block 1 of Table 2 results; all p-values < .0001). There was evidence for an acute differential treatment effect favoring FFT-CD on depression response and parallel though non-significant patterns of estimates for the other binary and continuous outcomes. For all continuous depression measures the rate of improvement leveled off significantly in both treatment arms (block 3 of Table 2 results; all p-values < .0009). In all depression measures there was a suggestion that the IP group continued to improve significantly over follow-up while the FFT-CD group did not (Block 2 of Table 2 results). We note that the cross-group improvements over the acute period and leveling off over follow-up were so significant that they would survive a correction for the full number of component tests provided in Table 2, clearly indicating that both treatments performed well at reducing youth depression and that effects did not continue at the same rate post-intervention.
Analyses limited to children ages 7–12, and those limited to children with MDD and/or DD, indicated the same pattern and magnitude of effects for the CDRS-R and CDI-C variables, with somewhat higher p values due to the smaller sample size (Table S5a,b).
Survival analyses showed similar rates of recovery over time for FFT-CD and IP groups (Wilcoxon χ2 (1) = .083, p = .77). Based on the survival curves, we estimate one-year rates of recovery from the index episode of 76% for FFT-CD and 77% for IP children, with median times to recovery of 81 and 88 days in FFT-CD and IP respectively. Among children who recovered, survival analyses examining time to recurrence suggest a possible FFT-CD advantage (Wilcoxon χ2 (1) =3.34, p = .068). One FFT-CD and 6 IP children suffered recurrences with estimated rates of 3% and 14% respectively.
Functioning and non-depressive symptoms
The omnibus tests for differential treatment effects were non-significant for secondary functioning (CGAS, SAS-SR) and non-depressive symptom measures (CBCL Internalizing, Externalizing, Total Problems). However, most of these measures showed the same basic pattern of effects as the depression measures, with significant initial improvements in both groups (estimates slightly larger but not significantly so in the FFT-CD group), followed by leveling off over the follow-up period (estimates slightly but not significantly favoring the IP group). The exception was the MASC which did show evidence of an overall differential treatment effect, but did not have the same leveling off pattern as the other measures and the group differences for the individual study phases were not significant. (Tables 2, S3).
Harm events
There were no reported deaths. Five suicide attempts (SAs, including interrupted attempts) were identified in 4 children (1 hanging, 2 overdose, 1 cutting, 1 method unknown), all in the IP group (1 during treatment, 4 during weeks 16–52). NSSI was reported in 15 children: 7 FFT-CD (15 episodes in 6 children during treatment, 1 episode in 1 child during weeks 16–52); 8 IP (12 episodes in 6 children during treatment; 9 episodes in 5 children during weeks 16–52). Mental health-related ED visits and hospitalizations were similar across treatment arms: 4 FFT-CD children with 7 total ED visits (2 children with 3 total visits during treatment, 2 children with 4 total visits weeks 16–52); 4 IP children with 5 total ED visits (2 children, 2 total visits, during treatment-period; 2 children with 3 total visits weeks 16–52); 3 FFT-CD children with 5 total hospitalizations (2 children with 2 total hospitalizations during treatment; 1 child with 3 total hospitalizations during weeks 16–52); 2 IP children with 2 total hospitalizations (1 during treatment, 1 weeks 16–52).
Discussion
This study reports results of the first large multi-site RCT to our knowledge to evaluate two active psychosocial treatment strategies for children with depressive disorders. Expanding on our prior report (Tompson et al., 2017) of an initial advantage for FFT-CD on depression response, we found that, after acute-treatment, as children returned to usual-care, symptom trajectories leveled off in both groups and treatment effects attenuated, resulting in similar depression levels by 52-weeks/one-year.
Because the study did not evaluate extended FFT-CD versus IP treatment over 52-weeks, with families free to pursue treatments as needed after acute-treatment, the early FFT-CD advantage may have been weakened by the lack of clinical control after week-16. Consistent with the possibility that differential treatment use after acute treatment may have contributed to the trajectory towards equivalence across treatment arms over follow-up, IP children were somewhat more likely to receive booster sessions of their study treatment during the follow-up period; although patterns of ancillary non-study medication and psychosocial treatments were similar in both groups. While sensitivity analyses adjusting for use of study booster sessions and ancillary non-study treatments didn’t alter the longitudinal findings, we cannot rule out the possibility that differences in non-study care impacted results. Nevertheless, our full trial results support the need for conservative interpretation of the early FFT-CD advantage, and readers should note the potential for slippage of differential treatment benefits over time.
Most children recovered within a year (estimated rates 76%−77%); and both treatments had large effect sizes on the CDRS-R at week-16. These rates are high and comparable to those reported for clinically referred samples in observational studies (Kovacs et al., 2016), and in adolescent depression treatment trials (Birmaher et al., 2000; Curry et al., 2011; Goodyer et al., 2017; Vitiello et al., 2011). However, with time some children suffered recurrent episodes, all but one in the IP group.
Supporting the safety of both treatments, there were no deaths and relatively few harm events. We observed five SAs, all in IP children. Consistent with our hypothesis that a family-focused approach could mobilize protective strengths, and similar to research showing benefits of treatments with strong family components for reducing SA-risk (Asarnow, Hughes, et al., 2017; Ougrin, Tranah, Stahl, Moran, & Asarnow, 2015) the greater family involvement in FFT-CD may have had an advantage for SA prevention.
The emergence of recurrent episodes over time, suggests the need for a more extended/chronic treatment model. These results are consistent with those from the largest longitudinal evaluation of depression in childhood which indicated that up to 72% of children who recovered from their first major depressive episodes suffered recurrent episodes (median between-episode interval 3–5 years, (Kovacs et al., 2016). This suggests the importance of more extensive prevention efforts, perhaps with interventions designed specifically for relapse/recurrence prevention (Kennard et al., 2008) or with monitoring strategies such as check-ups and “caring contacts” that remind parents and children that treatment is attainable and can help them to address emerging symptoms and stress. Incorporating monitoring and outreach strategies within primary care or school services could improve access and reach.
Our results point to the limitations of examining only single time points or simple acute-treatment interaction effects and the importance of examining trajectories over time. Focusing on the acute-period missed the observation that the initial treatment advantage didn’t last, whereas focusing on 52-week outcomes would neglect the earlier FFT-CD benefits. Examination of trajectories, as done here, allows discovery of potential group differences in pathways to longer term outcomes. Knowing the change trajectories with one treatment versus another allows an evidence-based approach to the timing of treatments including acute and post-acute treatment strategies.
One potential explanation for the equivalence of FFT-CD and IP outcomes at 52-weeks, is that both are effective treatments, particularly when delivered by the same therapists rated as high in nonspecific-factors such as therapist warmth and genuineness. Further, while IP youths had depression improvements similar to FFT-CD youths at 52-weeks, this is also the furthest point from original randomization and the likelihood of non-study influences is greatest, and our results do indicate that FFT-CD children showed a more rapid depression response as indicated by the significant FFT-CD advantage at 16-weeks.
The absence of a no-treatment condition was a study limitation, leaving uncertainty regarding whether both treatments were effective or whether time alone would yield similar healing. For ethical and scientific reasons, we used an active treatment comparator modeled after UC and matched to FFT-CD for key features (e.g. # sessions offered). Other limitations were: the exclusive focus on psychosocial treatment; the short follow-up; and the absence of differential treatment effects on measures of functioning and other symptoms. Because the trial was powered for acute treatment effects, we were likely underpowered for some trajectory analyses. Nevertheless, our findings of a leveling off between weeks 16–52 and an attenuation of the acute treatment effect for depression, call for caution in interpreting the initial FFT-CD advantage. Longer follow-up might have revealed stronger benefits (consistent with results on FFT for bipolar illness, Miklowitz et al., 2014) and is needed for evaluation of longer term risks including depression-recurrence, SAs, and health-risk behaviors (Bai et al., 2018; Kovacs et al., 2016; Weissman et al., 1999). Study treatments were delivered within a rigorous efficacy trial with extensive therapist training and clinical monitoring, likely strengthening both safety and benefits. Future research is needed to evaluate effectiveness under less controlled practice conditions.
In conclusion, our full trial results indicate an earlier depression response with FFT-CD, with a hint of possible protection from recurrence and SAs. However, our data suggest general equivalence of these two active treatments at 52-weeks. While this trajectory towards equivalence may have been affected by differential treatment use and other factors after the end of acute treatment, our results are consistent with findings from large adolescent depression trials that have found initial differential effects of psychosocial treatments but similar outcomes at longer follow-ups (Weisz et al., 2006). Multiple treatments lead to recovery, as do time and natural healing for many children. Attention to mechanisms contributing to recovery, identifying which children are most likely to benefit from treatments targeting different mechanisms, considering both psychosocial and medication treatment options, and understanding trajectories of improvement may help us to personalize and sequence treatments to provide optimal acute and longer-term care. While most children in FFT-CD and IP recovered, with time recurrent episodes emerged, and the literature suggests that depressive disorders will have longer term adverse impact for many children. The challenge is to develop the science to optimize clinical decision-making, a critical challenge given links between early depression and suicide and increasing rates of suicide deaths in the United States and other nations.
Supplementary Material
Table S1. Use of booster sessions and non-study medication and psychotherapy treatments during the trial.
Table S2. Sample characteristics at baseline.
Table S3. Means and standard deviations for continuous outcomes baseline to 52-weeks for each treatment group.
Table S4. Mean and standard deviations for Child Depression Inventory Child (CDI-C) and Parent (CDI-P) from baseline to 52-weeks for each treatment group.
Table S5a. Rates of change in the FFT and IP groups over each of the study periods and corresponding group differences for Age ≤12, adjusted for site, gender, baseline-diagnostic-status, baseline-medication, and family-composition.
Table S5b. Rates of change in the FFT and IP groups over each of the study periods and corresponding group differences for Syndromal Only, adjusted for site, gender, baseline-medication, and family-composition.
Key points.
Childhood-onset depressive disorders are associated with substantial morbidity and mortality.
This is the first large multi-site RCT to evaluate two active psychosocial treatment strategies for children ages 7–14 with depressive disorders.
Results indicate a quicker depression response in family focused treatment, compared to individual psychotherapy after roughly 16 weeks of acute treatment, and a hint of greater protection from recurrence and suicide attempts. However, outcomes were similar for both treatments by week-52/one-year.
Although community care received after acute-treatment may have influenced results, findings suggest the value of an extended/chronic disease model that includes monitoring and clinical guidance regarding optimal interventions when signs of depression-risk emerge.
Acknowledgements
Supported by National Institute of Mental Health (NIMH) grants MH082856 and MH082861. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or other funding agencies. J.R.A. received grant/other support from NIMH, Substance Abuse and Mental Health Services Administration, American Foundation for Suicide Prevention, American Psychological Association, Society of Clinical Child and Adolescent Psychology, Association for Child and Adolescent Mental Health, and Klingenstein Third-Generation Foundation. She consulted/presented on depression and suicidal/self-harm behavior, served on Data Safety and Monitoring and non-commercial Advisory Boards/Expert Panels. C.S. received research support from NIH through multiple divisions; Health Resources and Services Administration; and the Veterans Administration; served on technical expert panels for the Centers for Medicare and Medicaid Services, and Data Safety and Monitoring Boards for academic institutions and Kaiser Permanente. M.T. received research support from the NIMH, the Patient Centered Outcomes Research Institute, and the Smith Family Foundation; book royalties from Guilford Press and BVT Publishing; and honoraria from the American Psychological Association. The authors thank the children, families, and many colleagues who made this project possible including Judith Cohen who provided the IP training, Thomas Belin who provided statistical consultation, and the project DSMB: Donald Guthrie, Gabrielle Carlson, and Sherryl Goodman. The authors have declared that they have no competing or potential conflicts of interest.
Footnotes
Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article:
Conflict of interest statement: No conflicts declared.
References
- Achenbach TM (1991). Manual for the child behavior checklist and 1991 profile. Burlington: University of Vermont. [Google Scholar]
- Asarnow J, Hughes JL, Babeva KN, & Sugar CA (2017). Cognitive-behavioral family treatment for suicide attempt prevention: A randomized controlled trial. Journal of the American Academy of Child and Adolescent Psychiatry, 56(6), 506–514. doi: 10.1016/j.jaac.2017.03.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Asarnow J, Jaycox LH, Tang L, Duan N, LaBorde AP, Zeledon LR, … Wells KB (2009). Long-term benefits of short-term quality improvement interventions for depressed youths in primary care. American Journal of Psychiatry, 166(9), 1002–1010. doi: 10.1176/appi.ajp.2009.08121909 [DOI] [PubMed] [Google Scholar]
- Ascher BH, Farmer EMZ, Burns BJ, & Angold A (1994). The childhood and adolescent services assessment (CASA): Description and psychometrics. Journal of Emotional and Behavioral Disorders 4(1), 12–20. doi: 10.1177/106342669600400102 [DOI] [Google Scholar]
- Avenevoli S, Swendsen J, He JP, Burstein M, & Merikangas KR (2015). Major depression in the national comorbidity survey–adolescent supplement: prevalence, correlates, and treatment. Journal of the American Academy of Child and Adolescent Psychiatry, 54(1), 37–44.e2. doi: 10.1016/j.jaac.2014.10.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bai S, Zeledon LR, LaBorde AP, D’Amico E, Shoptaw S, Anderson M, … Asarnow J (2018). Reducing health risk behaviors and improving depression in adolescents: A randomized controlled trial in primary care clinics. Journal of Pediatric Psychology. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Birmaher B, Brent DA, Kolko D, Baugher M, Bridge J, Holder D, … Ulloa RE (2000). Clinical outcome after short-term psychotherapy for adolescents with major depressive disorder. Arch Gen Psychiatry, 57(1), 29–36. [DOI] [PubMed] [Google Scholar]
- Cohen J, & Mannarino AP (1996). Child-centered therapy treatment manual. Pittsburgh, PA: MCP Hahnemann University School of Medicine. [Google Scholar]
- Cohen J (1988). Statistical power analysis for the behavioral sciences (2nd Edition) Hillsdale, N.J.: L. Erlbaum Associates. [Google Scholar]
- Curry J, Silva S, Rohde P, Ginsburg G, Kratochvil C, Simons A, … March J (2011). Recovery and recurrence following treatment for adolescent major depression. Archives of General Psychiatry, 68(3), 263–270. doi: 10.1001/archgenpsychiatry.2010.150 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dietz LJ, Weinberg RJ, Brent DA, & Mufson L (2015). Family-based interpersonal psychotherapy for depressed preadolescents: Examining efficacy and potential treatment mechanisms. Journal of the American Academy of Child and Adolescent Psychiatry, 54(3), 191–199. doi: 10.1016/j.jaac.2014.12.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fristad MA, Verducci JS, Walters K, & Young ME (2009). Impact of multifamily psychoeducational psychotherapy in treating children aged 8 to 12 years with mood disorders. Arch Gen Psychiatry, 66(9), 1013–1021. doi: 10.1001/archgenpsychiatry.2009.112 [DOI] [PubMed] [Google Scholar]
- Goodyer IM, Dubicka B, Wilkinson P, Kelvin R, Roberts C, Byford S, … Harrington R (2008). A randomised controlled trial of cognitive behaviour therapy in adolescents with major depression treated by selective serotonin reuptake inhibitors. The ADAPT trial. Health Technology Assessment (Winchester, England), 12(14), 3–60. doi: 10.3310/hta12140 [DOI] [PubMed] [Google Scholar]
- Goodyer IM, Reynolds S, Barrett B, Byford S, Dubicka B, Hill J, … Fonagy P (2017). Cognitive behavioural therapy and short-term psychoanalytical psychotherapy versus a brief psychosocial intervention in adolescents with unipolar major depressive disorder (IMPACT): A multicentre, pragmatic, observer-blind, randomised controlled superiori. The Lancet Psychiatry, 4(2), 109–119. doi: 10.1016/S2215-0366(16)30378-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kaufman J, Birmaher B, Brent D, Rao U, Flynn C, Moreci P, … Ryan N (1997). Schedule for affective disorders and schizophrenia for school-age children-present and lifetime version (K-SADS-PL): Initial reliability and validity data. Journal of the American Academy of Child and Adolescent Psychiatry, 36(7), 980–988. doi: 10.1097/00004583-199707000-00021 [DOI] [PubMed] [Google Scholar]
- Kennard BD, Emslie GJ, Mayes TL, Nightingale-Teresi J, Nakonezny PA, Hughes JL, … Jarrett RB (2008). Cognitive-behavioral therapy to prevent relapse in pediatric responders to pharmacotherapy for major depressive disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 47(12), 1395–1404. doi: 10.1097/CHI.0b013e31818914a1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kovacs M (1992). Children’s Depression Inventory. New York: Multi-Health Systems. [Google Scholar]
- Kovacs M, Obrosky S, & George C (2016). The course of major depressive disorder from childhood to young adulthood: Recovery and recurrence in a longitudinal observational study. Journal of Affective Disorders, 203, 374–381. doi: 10.1016/j.jad.2016.05.042 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kraemer HC “A Source of False Findings in Published Research Studies: Adjusting for Covariates”, JAMA Psychiatry, 2015, vol 72(10), 961–962. [DOI] [PubMed] [Google Scholar]
- Luby JL, Barch DM, Whalen D, Tillman R, & Freedland KE (2018). A randomized controlled trial of parent-child psychotherapy targeting emotion development for early childhood depression. American Journal of Psychiatry, (24), appi.ajp.2018.1. doi: 10.1176/appi.ajp.2018.18030321 [DOI] [PMC free article] [PubMed] [Google Scholar]
- March JS, Parker JDA, Sullivan K, Stallings P, & Conners CK (1997). The multidimensional anxiety scale for children (MASC): Factor structure, reliability, and validity. Journal of the American Academy of Child and Adolescent Psychiatry, 36(4), 554–565. doi: 10.1097/00004583-199704000-00019 [DOI] [PubMed] [Google Scholar]
- Miklowitz DJ, Schneck CD, George EL, Taylor DO, Sugar CA, Birmaher B, … Axelson DA (2014). Pharmacotherapy and family-focused treatment for adolescents with bipolar I and II disorders: A 2-year randomized trial. American Journal of Psychiatry, 171(6), 658–667. doi: 10.1176/appi.ajp.2014.13081130 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ougrin D, Tranah T, Stahl D, Moran P, & Asarnow JR (2015). Therapeutic interventions for suicide attempts and self-harm in adolescents: Systematic review and meta-analysis. Journal of the American Academy of Child and Adolescent Psychiatry, 54(2), 97–107.e2. doi: 10.1016/j.jaac.2014.10.009 [DOI] [PubMed] [Google Scholar]
- Poznanski E, Hartmut MB, Grossman J, & Freeman LN (1985). Diagnostic criteria in childhood depression. American Journal of Psychiatry, 142(10), 1168–1173. [DOI] [PubMed] [Google Scholar]
- SAS Institute Inc. (2017). Statistical Analysis System 9.4 Software. Cary, NC: SAS Institute Inc. [Google Scholar]
- Shaffer D, Gould MS, Brasic J, Ambrosini P, Fisher P, Bird H, & Aluwahlia S (1983). A children ‘s global assessment scale. Archives of General Psychiatry, 40, 5–8. [DOI] [PubMed] [Google Scholar]
- Tarnowski K, Simonian S, Bekeny P, & Park A (1992). Acceptability of interventions for childhood depression. Behavior Modification, 16(1), 103–117. doi: 10.1177/01454455920161005 [DOI] [PubMed] [Google Scholar]
- March J, Silva S, Curry J, Wells K, Fairbank J, … Bartoi M (2009). The treatment for adolescents with depression study (TADS): Outcomes over 1 year of naturalistic follow-up. The American Journal of Psychiatry, 166, 1141–1149. doi: 10.1176/appi.ajp.2009.08111620 [DOI] [PubMed] [Google Scholar]
- Tompson M, Langer D, Hughes J, & Asarnow J (2017). Family-focused treatment for childhood depression: Model and case illustrations. Cognitive and Behavioral Practice, 24(3). doi: 10.1016/j.cbpra.2016.06.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tompson M, Sugar CA, Langer DA, & Asarnow JR (2017). A randomized clinical trial comparing family-focused treatment and individual supportive therapy for depression in childhood and early adolescence. Journal of the American Academy of Child and Adolescent Psychiatry, 56(6), 515–523. doi: 10.1016/j.jaac.2017.03.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Vitiello B, Emslie G, Clarke G, Wagner KD, Asarnow JR, Keller MB, … Brent DA (2011). Long-term outcome of adolescent depression initially resistant to selective serotonin reuptake inhibitor treatment: A follow-up study of the TORDIA sample. Journal of Clinical Psychiatry, 72(3), 388–396. doi: 10.4088/JCP.09m05885blu [DOI] [PMC free article] [PubMed] [Google Scholar]
- Weersing VR, Brent DA, Rozenman MS, Gonzalez A, Jeffreys M, Dickerson JF, … Iyengar S (2017). Brief behavioral therapy for pediatric anxiety and depression in primary care. JAMA Psychiatry, 74(6), 571–578. doi: 10.1001/jamapsychiatry.2017.0429 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Weissman MM, Orvaschel H, & Padian N (1980). Children’s symptom and social functioning self-report scales. Comparison of mothers’ and children’s reports. The Journal of Nervous and Mental Disease, 168(12), 736–740. [DOI] [PubMed] [Google Scholar]
- Weissman MM, Wolk S, Goldstein RB, Moreau D, Adams P, Greenwald S, … Wickramaratne P (1999). Depressed adolescents grown up. Journal of the American Medical Association, 281(18), 1707–1713. doi: 10.1001/jama.281.18.1707 [DOI] [PubMed] [Google Scholar]
- Weisz JR, McCarty CA, & Valeri SM (2006). Effects of psychotherapy for depression in children and adolescents: A meta-analysis. Psychological Bulletin, 132(1), 132–149. doi: 10.1037/0033-2909.132.1.132 [DOI] [PMC free article] [PubMed] [Google Scholar]
- WHO. (2012). Depression, a global public health concern. WHO Department of Mental Health and Substance Abuse, 1–8. doi: 10.1007/978-3-642-11688-9_20 [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Use of booster sessions and non-study medication and psychotherapy treatments during the trial.
Table S2. Sample characteristics at baseline.
Table S3. Means and standard deviations for continuous outcomes baseline to 52-weeks for each treatment group.
Table S4. Mean and standard deviations for Child Depression Inventory Child (CDI-C) and Parent (CDI-P) from baseline to 52-weeks for each treatment group.
Table S5a. Rates of change in the FFT and IP groups over each of the study periods and corresponding group differences for Age ≤12, adjusted for site, gender, baseline-diagnostic-status, baseline-medication, and family-composition.
Table S5b. Rates of change in the FFT and IP groups over each of the study periods and corresponding group differences for Syndromal Only, adjusted for site, gender, baseline-medication, and family-composition.