Please take our brief survey

Blueprints Programs = Positive Youth Development

Return to Search Results

Promising Program Seal

Adolescent Coping with Depression

Blueprints Program Rating: Promising

A 16-session group depression treatment program teaching teens how to deal with depression and ways to manage depression.

  • Paul Rohde, Ph.D.
  • Oregon Research Institute
  • 715 Franklin Blvd.
  • Eugene, OR 97403
  • USA
  • (541) 484-2123
  • (541) 484-1108
  • paulr@ori.org
  • Depression

    Program Type

    • Cognitive-Behavioral Training

    Program Setting

    • Mental Health/Treatment Center

    Continuum of Intervention

    • Selective Prevention (Elevated Risk)
    • Indicated Prevention (Early Symptoms of Problem)

    A 16-session group depression treatment program teaching teens how to deal with depression and ways to manage depression.

      Population Demographics

      Adolescent Coping with Depression targets teens ages 13-18 at risk of or with an indicated condition of depression.

      Age

      • Early Adolescence (12-14) - Middle School
      • Late Adolescence (15-18) - High School

      Gender

      • Male and Female

      Gender Specific Findings

      • Male
      • Female

      Race/Ethnicity

      • All Race/Ethnicity

      Race/Ethnicity Specific Findings

      • White

      Race/Ethnicity/Gender Details

      The studies, which use largely white samples, target all race and ethnic differences in program outcomes. However, one study finds that the group-based therapy works better for white youth than others.

      The key mediating risk factors occur at the individual level and include negative thinking, lack of pleasant activities, and poor social skills.

      • Individual
      Risk Factors
      • Family: Parent history of mental health difficulties
      Protective Factors
      • Individual: Coping Skills*, Problem solving skills, Skills for social interaction

      *Risk/Protective Factor was significantly impacted by the program.

      Adolescent Coping with Depression (CWD-A) is a therapeutic group intervention designed to reduce or prevent major depression or dysthymia (chronic depression) among adolescents, including those whose depression co-occurs with conduct disorder. Based on cognitive-behavioral therapy, the program teaches teens the skills to monitor moods, increase pleasant activities, improve communication, and resolve conflict. Adolescent groups meet with therapists over an eight-week period in 16 two-hour sessions. Groups consist of 7-14 adolescents and are conducted by a trained interventionist that has at least a master's degree in a mental health field.

      Adolescent Coping with Depression is a therapeutic group intervention designed to reduce or prevent major depression or dysthymia (chronic depression) among adolescents, including those whose depression co-occurs conduct disorder. Depressed individuals often have problems with discomfort and anxiety, irrational and negative thoughts, poor social skills, and a low rate of pleasant activities. The program aims to address each of these problematic areas.

      Based on cognitive-behavioral therapy, the program teaches skills to monitor moods, increase pleasant activities, improve communication, and resolve conflict. The group therapy sessions involve skill training to better relax, control thoughts, and resolve conflict. Therapists follow a detailed manual to ensure protocol compliance and fidelity, and adolescents and parents receive workbooks consisting of homework assignments, forms, short handouts, and readings to complement the therapy sessions.

      In the acute phase, adolescent groups with up to 10 boys and girls meet a therapist in about 16 two-hour sessions. The adolescent sessions can be supplemented with nine two-hour sessions involving a therapist and a group of parents. The intervention sessions may be followed by one to two booster sessions over the two years following the intervention. The booster sessions focus on how skills learned in the intervention can be applied to new or continuing problematic situations. However, booster sessions had very limited impact. The program has been applied to youth with depression, at risk of developing depression, and with both depression and conduct disorder.

      Adolescent Coping with Depression is based on cognitive-behavioral therapy that aims to alleviate depression by changing dysfunction thoughts and behaviors, improving social skills, and increasing participation in pleasant activities.

      • Cognitive Behavioral

      In a series of randomized control studies, Adolescent Coping with Depression was evaluated with three groups of teens: 1) adolescent recruits with major depression or dysthymia; 2) adolescents with a depressed parent and high risk for depression; and 3) adolescents with depression and conduct disorder who were referred by youth services. The depressed adolescents were assigned to an intervention therapy condition for adolescents only, an intervention therapy condition for adolescents and parents, and a waitlist control group. Adolescents in the intervention groups were also randomly assigned to conditions that differed on follow-up access to booster sessions and the number of follow-up assessments. The adolescents with depression and conduct disorder were assigned to an intervention therapy condition and a life-skills training control condition, but the intervention included an additional therapist and point system for attendance. The evaluations generally obtained preintervention measures, postintervention measures, and follow-up measures for up to 24-months.

      Skill-based cognitive-behavioral therapy consisting of approximately 16 two-hour group sessions over eight weeks appears to be an effective intervention for adolescents with comorbid major depressive disorder and conduct disorder (plus associated problems with substance abuse or dependence). The odds of recovering from a major depressive disorder for those in the program were more than twice those for participants in a comparison life-skills course. The therapy also significantly reduced self-reports and clinician ratings of depression, primarily by changing values on a mediating variable that measures negative thoughts. Adolescents of white ethnicity, with recurrent major depressive disorder, and with good coping skills benefited most from the therapy.

      However, the treatment provided no benefits beyond the comparison group at the 6-month follow-up assessment. Also, the treatment had no benefits for conduct disorder. Despite the lack of sustained benefits, the short-term gain from a faster recovery from depression helps vulnerable youth.

      The therapy also appears to help prevent the occurrence of major depression among youth at risk for developing the disorder. For adolescent children of a parent with severe depression and who have moderate symptoms of depression but not a clinical diagnosis of depression, the therapy did better than a control group in preventing major depression, at least in the short-run. However, the evidence is not as strong for this version of the program as for the comorbid version of the program.

      The intervention also aids recovery of youth with depression only. Separate group therapy for parents and booster sessions following the acute therapy had limited benefits beyond the core program. Again, however, the evidence for this version of the program is not as strong as for the comorbid version of the program.

      • Adolescent Coping with Depression appears to be effective for real-world populations such as depressed adolescents with and without conduct disorders, and it reduces depression primarily by reducing the occurrence of negative thoughts. However, it does not improve conduct disorders.
      • The group-based treatment primarily benefited adolescents of White ethnicity, with recurrent major depressive disorder, and with good coping skills.
      • The treatment provided no benefits beyond the comparison group at the 6-month follow-up assessment. Despite the lack of sustained benefits, the short-term gain from a faster recovery from depression helps vulnerable teens.

      Analysis of mediating or targeted behaviors focused on changes over time across groups on measures of anxiety, pleasant activities, and depressogenic cognitions. Although time had significant effects (i.e., all groups improved over time), the group-by-time interactions for these mediating factors were not significant. The improved depression outcomes for the intervention groups did not correspond closely to similar improvements in the targeted behaviors (see Lewinsohn, Clarke, Hops, & Andrews, 1990)).

      In a separate study (Study 5), treatment significantly improved responses on the Automatic Thought Questionnaire, and this related strongly and significantly to two depression measures. Concerning the attenuation of the intervention effect with controls for the mediators, adding the Automatic Thoughts Questionnaire score to the model reduced the effect of the intervention on the Beck Depression Inventory by 75% and to insignificance. Adding the score reduced the effect of the intervention on the Hamilton Depression Scale but by a smaller magnitude (by 40%).

      Depression was mediated by change in depressotypic thoughts (measured by the Automatic Thoughts Questionnaire) which assesses the frequency of occurrence of negative thoughts (such as "I'm no good," "My life is a mess.").

      The results generalize to volunteer recruits ages 13-17 with major depression with and without conduct disorder who are referred by youth services. However, some evidence suggests the program works better for whites than others. The studies used samples from Eugene and Portland, Oregon but the results likely generalize to other cities, places, and states. The results do not generalize to older age groups with different sets of mental health problems.

      The Clarke et al., 1999 study has some limitations:

      • The samples sizes for the follow-up groups were quite small (only 5 in one case).
      • The results proved strong for the dichotomous measure of recovery but were mixed for multiple continuous measures of depression.
      • Attrition was substantial in the follow-up, including loss of 50% of those assigned to the booster session (hence n = 5 for this group). Analyses showed that attrition did not differ across treatment groups on nearly all measured sociodemographic and depression measures, and the design used intent-to-treat analysis, however, those with higher self-rated depression were more likely to drop out during the follow-up.
      • Reliance on volunteers means the subjects do not represent the clinical population of depressed adolescents.
      • Parental involvement in the therapy may have positively affected their reporting on adolescent outcomes.
      • The waitlist group did not receive any placebo equivalent.
      • The evidence was strongest for the acute intervention.

      Across all studies, the program evaluations used four randomized controlled trials with careful attention to fidelity, sufficient sample sizes for most comparisons, blind evaluations of outcomes plus subject self-reports, multiple site replications, analysis of mediating factors, and appropriate analyses with ITT, baseline controls, and dose-response evidence. Effect sizes reached medium levels (d about .5).

      Variations to include groups for parents, to treat conduct disorder, and to add booster sessions received less clear support and shouldn’t be recommended. Otherwise, two weaknesses may have compromised the evaluations.

      1. With numerous outcome measures of depression, there were many tests of significance relative to significant results. Focus on the acute phase treatment gave more consistent results, but only some of the outcome measures showed improvement.

      2. The follow-up for many of the studies contained no control group comparisons. The benefits for the intervention group were generally maintained over time but often could not be compared to the control group. In studies where the control group was followed, there was convergence over time. The authors argue that the treatment speeded recovery as much as led to long-term advantage.

      The highest quality study was Rohde et al., 2004 (for depressed adolescents with conduct disorder). Although there is no analysis of differential attrition, only four were lost from the treatment group and three from the comparison group, so the threat of differential attrition is minimal.

      • Blueprints: Promising

      Clarke, G. N., Hornbrook, M., Lynch, F., Polen, M., Gale, J., Beardslee, W., ... Seeley, J. (2001). A randomized trial of a group cognitive intervention for preventing depression in adolescent offspring of depressed parents. Archives of General Psychiatry, 58, 1127-1134.

      Clarke, G. N., Rohde, P., Lewinsohn, P. M., Hops, H., & Seeley, J. R. (1999). Cognitive-behavioral treatment of adolescent depression: Efficacy of acute group treatment and booster sessions. Journal of the American Academy of Child and Adolescent Psychiatry, 38, 272-279.

      Kaufman, N. K., Rohde, P., Seeley, J. R., Clarke, G. N., & Stice, E. (2005). Potential mediators of cognitive-behavioral therapy for adolescents with comorbid depression and conduct disorder. Journal of Consulting and Clinical Psychology, 73, 38-46.

      Lewinsohn, P. M., Clarke, G. N., Hops, H., & Andrews, J. (1990). Cognitive-behavioral treatment for depressed adolescents. Behavior Therapy, 21, 385-401.

      Rohde, P., Clarke, P. N., Mace, D. E., Jorgensen, J. S., & Seeley, J. R. (2004). An efficacy/effectiveness study of cognitive-behavioral treatment for adolescents with comorbid major depression and conduct disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 43, 660-668.

      Rohde, P., Seeley, J. R., Kaufman, N. K., Clarke, G. N., & Stice, E. (2006). Predicting time to recovery among depressed adolescents treated in two psychosocial group interventions. Journal of Consulting and Clinical Psychology, 74, 80-88.

      Paul Rohde, Ph.D.
      Oregon Research Institute
      1715 Franklin Blvd.
      Eugene, OR 97403
      Phone: (541) 484-2123
      Fax: (541) 484-1108
      paulr@ori.org

      Study 1

      Clarke, G. N., Rohde, P., Lewinsohn, P. M., Hops, H., & Seeley, J.R. (1999). Cognitive-behavioral treatment of adolescent depression: Efficacy of acute group treatment and booster sessions. Journal of the American Academy of Child and Adolescent Psychiatry, 38, 272-279.

      Study 2

      Rohde, P., Clarke, P. N., Mace, D. E., Jorgensen, J. S., & Seeley, J. R. (2004). An efficacy/effectiveness study of cognitive-behavioral treatment for adolescents with comorbid major depression and conduct disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 43, 660-668.

      Clarke, G.N., Rohde, P., Lewinsohn, P.M., Hops, H., & Seeley, J.R. (1999). Cognitive-behavioral treatment of adolescent depression: Efficacy of acute group treatment and booster sessions. Journal of the American Academy of Child and Adolescent Psychiatry, 38, 272-279.

      Evaluation Methodology

      The randomized control trial included an acute phase and follow-up phase. The acute phase followed the program and design of a previous 1990 study (see study 1 below) but lasted a week longer. The follow-up phase and booster sessions were unique to this study.

      Design: The project recruited individuals in Eugene and Portland, Oregon, via announcements to health professionals, school counselors, television and newspaper stories, and advertisements. The population of interest was narrowly defined: Only adolescents ages 14 to 18 and with a current DSM-III-R diagnosis of major depressive disorder or dysthymia were eligible. Volunteers were excluded if they exhibited various anxiety disorders or substance abuse/dependence, mental retardation, or schizophrenia; would not agree to discontinue other therapy; or needed immediate, acute treatment. Screening interviews of 331 consenting adolescents and parents led to selection of 123 subjects who met the study requirements. Compared to excluded recruits, eligible youth were more likely to be female and older but did not differ by race, living situation, or parental education.

      For the acute phase, eligible participants were randomly assigned to one of three conditions:

      1) A 16-session cognitive-behavioral therapy group for adolescents only (n=45);

      2) An identical group for adolescents supplemented with a separate nine-session parent group (n=42); and

      3) A waitlist control group (n=36).

      Twenty-seven (22%) of the 123 randomized adolescents failed to complete the acute phase by attending fewer than 7 sessions, not completing the assessments, or violating the study protocol.

      Immediately after the acute phase, the 64 adolescents who were in the active treatments and completed the phase were randomly assigned to one of three conditions:

      1) Booster sessions and independent assessments every 4 months (n =24);

      2) Assessments every 4 months (n =16) ; and

      3) Assessments every 12 months (n =24).

      The first two conditions compared the impact of the booster sessions, while the last two conditions compared the impact of additional assessments. For the booster session condition, adolescents met 1 to 2 times with a therapist to discuss how skills learned earlier in the program could be applied to specific, problematic situations that had come up. Less than 50% of those assigned completed the booster sessions.

      The study began in 1991, the acute (or intervention) phase lasted eight weeks, and the follow-up phase lasted 24 months after the intervention.

      Sample Characteristics: Among the 96 participants completing the acute phase, females made up 70.8% of the sample. The mean age was 16.2 years, 95.8% were in school, 43.8% lived in 2-parent families, and 27.7% had 1 or 2 parents with graduate or postgraduate education. The majority (n=73) had pure major depressive disorder, 12 had pure dysthymia, and 11 had both.

      Measures:

      Diagnostic Interviews: Adolescents were interviewed at intake with the Schedule for Affective Disorders and Schizophrenia for School-Age Children – Epidemiologic version (K-SADS-E). At the posttest and follow-up assessments, participants were interviewed regarding psychiatric symptoms and disorders since the previous interview using the Longitudinal Interval Follow-up Evaluation (LIFE), which used rigorous criteria for recovery (i.e., symptom-free for at least 8 weeks). Raters were blind to the subject condition, and interrater agreement for unipolar mood disorder was .75, indicative of good to excellent agreement.

      Interviewer Ratings: Interviewers completed a 14-item version of the Hamilton Depression Rating Scale (HAM-D) for current and worst past episode of depression based on responses to the K-SADS-E. They also rated the 1) current level of functioning and 2) highest level of functioning during the past year using the DSM-III Global Assessment of Functioning scale.

      Adolescent Report Measures: Adolescents completed the Beck Depression Inventory, a 21-item self-report measure of depressive symptoms. Internal consistency (alpha) at intake was .85.

      Parent Report Measures: One or both parents completed the Child Behavior Checklist (CBCL), which assesses 118 adolescent/child behavior problems. CBCL subscales were computed for the Externalizing and Internalizing categories, as well as for depression. Reported CBCL scores were from mothers (90%) or from fathers or other guardians (10%).

      Analysis: Outcome depression measures were analyzed at the individual level with random effects regression, which estimates an intercept and slope across time points for each participant (but is augmented by data from the entire sample). Using an intent-to-treat approach, the random effects regression included all 123 participants randomly assigned to conditions and all data available for each subject. For example, subjects having data through the 24-month assessment contribute more data than those dropping out after the posttest. The models include group-by-time interactions that isolate differential change from the baseline outcomes.

      The planned comparisons (one-tailed, .05) for the acute phase involved group-by-time interactions that contrasted the two active conditions with the waitlist condition and contrasted the adolescent intervention condition with adolescent and parent intervention condition. The planned comparisons for the follow-up or maintenance phase again involved group-by-time interactions but contrasted the booster sessions condition with the annual and 4-month assessment conditions and contrasted the annual assessment condition with the 4-month assessment condition.

      Outcomes

      Baseline Equivalence and Attrition: The 3 experimental groups were statistically equivalent at baseline. Among the 123 volunteers who met the criteria for the study, 96 completed the acute phase (22% attrition). Of the 64 in the follow-up phase, 46 completed all assessments (28% attrition). Acute phase attrition was not significantly related to the experimental condition or baseline measures of gender, age, living with 1 versus 2 parents, parental education, depression diagnosis, or depression self-reports. It did vary across the two sites, but since the two sites did not differ on any of the sociodemographic variables, baseline depression measures, or recovery, the data from the two sites were pooled for the analysis.

      Attrition during the follow-up assessments was not associated with adolescent gender, age, living with both biological parents, level of parent education, study site, or diagnostic or self-reported depression at intake. Dropouts had higher scores on the Beck Depression Inventory at posttreatment, however. Completion of the booster sessions was estimated at less than 50%, largely because some subjects had recovered and were uninterested in additional treatment, some were seeing a nonstudy therapist, some were unable to schedule or attend appointments, and some moved out of the area.

      Posttest: In the acute phase, outcomes studied first included a dichotomous measure of recovery from depression (i.e., no longer meeting DSM-III-R criteria for either major depression or dysthymia for the two weeks preceding the posttreatment assessment).The results showed higher recovery rates in the intervention conditions (65% for adolescents and 69% for adolescents and parents) versus 48% in the waitlist condition. The significantly (one-tailed test, p<.05) better outcomes for the treatment groups compared to the waitlist group translate to an effect size (d) of .38 (small to medium) and an odds ratio of 2.15 (i.e., intervention groups had odds of recovery that were more than twice as high as the control group).

      One of the continuous measures (the Beck Depression Inventory) showed significantly better improvement for the intervention groups (p < .01, two-tailed), while another (Global Assessment of Functioning scale) showed significantly better improvement at p < .05, one-tailed test. Four others (the Hamilton Depression Rating Scale, CBCL depression, CBCL internalizing, CBCL externalizing) did not. No significant differences emerged from comparisons of the adolescent only with the adolescent/parent conditions. The change score effect sizes were medium for the Global Assessment of Functioning scale (d = .54) and Beck Depression Inventory (d = .61). Overall, 1 of the 12 planned comparisons reached significance with two-tailed tests.

      The dose-response analysis examined the influence of treatment attendance and recovery. About 48% of the waitlist group recovered, 57% of the intervention group attending 9-12 sessions recovered, and 69% of those attending 13-16 sessions recovered. The linear association indicates a significant trend toward greater recovery at higher levels of attendance.

      Long-Term: Results for assessments at 12 and 24 months again examined recovery in binary form and six continuous measures of depression. Recovery in binary form was measured as 8 weeks or more of minimal or absent depression symptoms.

      For the analysis of recovery, the 12-month follow-up showed that 100% (5/5) of depressed adolescents randomly assigned to the booster condition had recovered, versus 50% (6/12) of the participants in the two assessment-only conditions. The difference was significant. At the 24-month assessment, however, rates had converged, with 100% (5/5) of the booster-only condition recovered versus 90% (9/10) of the participants in the two assessment-only conditions. The difference between the two different periods of follow-up indicates that, despite ending up with similar rates of recovery, those in the booster condition recovered at a significantly faster rate than the assessment-only conditions.

      Subjects having recovered in the acute phase could experience a recurrence in the follow-up phase. Of the 46 adolescents in the 2 active treatments who had recovered at posttreatment, 12-month recurrence rates were 14% (2/14) in the annual assessment conditions, 0% (0/11) in the frequent assessment condition, and 27% (4/15) in the booster condition. These differences were not significant. Also, differences across planned comparisons at 24 months were not significant.

      For the continuous measures, the group-by-time interactions indicated significant benefits of the booster session relative to the two assessment-only conditions for one measure — the CBCL externalizing. The group-by-time interactions indicated significant benefits of the frequent assessment relative to the annual assessment only for the CBCL depression and CBCL internalizing outcomes. Overall, 3 of the 12 planned comparisons reached significance.

      Rohde, P., Clarke, P.N., Mace, D.E., Jorgensen, J.S., & Seeley, J.R. (2004). An efficacy/effectiveness study of cognitive-behavioral treatment for adolescents with comorbid major depression and conduct disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 43, 660-668.

      This study applied the Coping with Depression course to a different population than previous studies – depressed adolescents with comorbid conduct disorders. Whereas previous studies of depression excluded subjects with comorbid conditions, this study focuses on youth from a real-world setting and with multiple problems. The study also differs from previous ones in that subjects could continue therapy outside the experimental study. This means the study combines components of both efficacy (in random assignment) and effectiveness (in the possible influence of other real-world activities). The study examined gender differences in more detail than previous ones. Otherwise, the course work for those in the treatment group was the same as in previous studies.

      Evaluation Methodology

      Design: Individuals aged 13 to 17 were referred to the study by staff from the Department of Youth Services of Lake County, Oregon. All were under the supervision of an intake, probation, or parole officer but not incarcerated. After screening for both major depressive disorder and conduct disorder, 93 adolescents were randomly assigned to one of two conditions: the Coping with Depression therapy (n = 45) or a life-skills control (n = 48).

      The two treatments were made similar in the number of sessions, interventionist contact time, and non-specific therapeutic factors. Mixed gender groups of approximately 10 adolescents were treated in 16 2-hour sessions conducted over an 8-week period. The Adolescent Coping with Depression course was modified to address conduct disorder but still used cognitive-behavioral skill-based training. The life-skills course emphasized activities such as filling out a job application and renting an apartment, and it offered academic tutoring. It attempted to educate participants on basic life skills in a supportive and non-judgmental manner.

      The study began between 1998 and 2001. Posttest assessment occurred after completion of the 8-week intervention, and 6-month and 12-month assessments followed.

      Sample: The subjects were referred from the Department of Youth Services (n = 281), provided consent (n = 205), completed the intake evaluation (n = 187), and met the inclusion criteria (n =98). Subjects had to be able to converse in English and expect to stay in the county over the next 12 months. Those charged with first-degree assault, robbery, homicide, or rape were excluded (and sent to special programs), and those with psychotic symptoms were excluded. The 93 of the 98 who agreed to continue and were randomly assigned then completed the posttest, the 6-month follow-up, and the 12-month follow-up.

      The assigned sample had a mean age of 15.1 years, consisted of 48% female and 81% white, 15% resided with both parents, 75% were attending school, 18% had repeated a grade, and 15% had a parent with a bachelor’s degree or higher. The mean duration of major depressive disorder was 81.3 weeks, the mean duration of conduct disorder was 185.3 weeks, and 40% had attempted suicide.

      Measures:

      Psychiatric Outcomes: Adolescents and a parent were interviewed at intake to assess adolescent past and current episodes of DSM-IV disorders using the Schedule for Affective Disorders and Schizophrenia for School-Age Children-Epidemiologic version (K-SADS-E). At the follow-up assessments, participants were interviewed regarding psychiatric symptoms and disorders since the previous interview with the Longitudinal Interval Follow-up Evaluation (LIFE), which used rigorous criteria for recovery (i.e., symptom-free for at least 8 weeks). Recovery from conduct disorder required 6 months of nearly absent symptoms. Interrater agreement for a 10% sample was .88 for depression and 1.00 for conduct disorder.

      Dimensional Measures: Interviewers completed a 17-item version of the Hamilton Depression Rating Scale, and adolescents completed the 21-item Beck Depression Inventory.

      Conduct Disorder: Parent or adult informants completed the Child Behavior Checklist (CBCL), and the externalizing subscale was computed to measure disruptive behavior.

      Psychosocial functioning: Interviewers rated current functioning using the Children’s Global Adjustment scale and adolescents completed the Social Adjustment Scale – Self Report for Youth.

      Nonresearch treatment: Subjects were permitted to receive external mental health treatment during the course of the study. Based on parent/adult reports, three variables related to such treatment were created: pharmacotherapy, residential treatment, and number of hours of nonresearch outpatient treatment.

      Criminal records: Participants granted researchers access to criminal arrest records for the 12 months preceding and 12 months after treatment.

      Analysis: The individual-level analysis used contingency tables and logistic regression for dichotomous outcomes, and random effects regression for dimensional measures. Random effects regression was done for the full assignment sample of 93 in an intent-to-treat analysis and for the subsample with complete data on the follow-up assessments. Because attrition was low (6%) and both analyses produced the same pattern of results, the paper presents only the results for the complete follow-up sample. The analyses included group-by-time interaction terms that control for baseline differences in the outcomes.

      Outcomes:

      Baseline Equivalence and Attrition: The two treatment groups differed at baseline on gender, which the investigators then included as a covariate, but not on the outcome variables or other measured determinants. No formal attrition analysis of group differences in attrition was presented, but, as noted above, attrition was low and the pattern of results proved similar when including and not including drop-outs in the analysis.

      Posttest: Major depressive disorder recovery rates at posttreatment were significantly greater in the treatment condition (39%) than in the life-skills control condition (19%) and the odds ratio equaled 2.66. Recovery rates for conduct disorder were not significantly different across the groups. Interactions by gender were not significant.

      For the five dimensional measures, group-by-time interactions were significant for three of the five outcomes – the Beck Depression Index, the Hamilton Depression Rating Scale, and the Social Adjustment Scale – Self-Report. The tables report the variance explained by group-by-time interactions; for the significant relationships, the variance explained ranged from .047 to .064 (equivalent to correlations ranging from .22 to .25). No measures of mediators were available for analysis.

      Longterm: Differences in major depressive disorder recovery rates were not significant at 6 and 12 months. Survival analysis and Cox proportional hazard models showed no difference on the 12-month recovery function. With only one exception among the 10 tests, dimensional measures did not differ significantly at 6 and 12 months. Among the 25 adolescents who recovered at posttest, the reoccurrence was higher for the treatment group (25%) than the life-skills group (11%). The authors interpret the convergence as the result of the late effectiveness of the life-skills condition rather than the loss of gains from the therapeutic treatment.

      A dose-response analysis based on attendance at the sessions showed no main effect or group-by-time interaction effect on recovery rates for major depressive disorder or conduct disorder. Also, no group-by-time interaction coefficients were significant for secondary outcomes such as suicide attempts, substance abuse/dependence, anxiety disorders, and residential treatment.

      Limitations:

      • In comparing cognitive-behavioral therapy to a life-skills course, the study did not have a no-treatment control group.
      • The treatment results were inconsistent. For depression, they showed benefits compared to a life-skills treatment only at posttreatment, but not at follow-up and not for conduct disorder.
      • Therapists were nested within conditions, making it difficult to separate condition from therapist effects.
      • The sample comes from a relatively narrow population that was largely white and referred from a county department of juvenile corrections.
      • The treatment groups included 89 adolescents who were not part of the experimental assessment, had high rates of affective disorder, and may have influenced the nature of the interactions in the group.

      Clarke, G.N., Hornbrook, M., Lynch, F., Polen, M., Gale, J., Beardslee, W., ... Seeley, J. (2001). A randomized trial of a group cognitive intervention for preventing depression in adolescent offspring of depressed parents. Archives of General Psychiatry, 58, 1127-1134.

      Much as in previous studies, this study examined the effects of group-based cognitive-behavioral therapy. However, it used a sample of adolescents at risk for depression – because their parents had been diagnosed as depressed – but not meeting the criteria for a depressive disorder. The intervention aimed to prevent progression to future episodes of major depression rather than reduce depression among those with severe conditions. It focused on families that were not seeking care for the youth, and thus involved more in the way of outreach than studies of depressed youth seeking treatment.

      Evaluation Methodology

      Design: This randomized control trial sampled from adult members of an HMO centered around Portland, Oregon, during the period from 1994 to 1996. The members needed to be ages 30 to 65, have covered dependents ages 13-16, and live reasonably close to the study center. A search of pharmacy records and medical records of HMO members meeting these criteria identified 5904 potentially depressed adults. A review of medical charts then identified 3935 of these adults as having a depression diagnosis or symptoms, and 2995 were judged by their physicians to be appropriate for the study. The adolescent children of the 2083 adults willing to participate were asked to also participate. A total of 744 families with 966 youth agreed to an interview and 481 families with 551 adolescents completed the interview. To qualify for the study, parents could either be in a current episode of major depression and/or dysthymia, or have had an episode in the past 12 months. With 79 parents not meeting this criterion, 472 youth remained and were classified as having high-severity depression, medium-severity depression, or low-severity depression. Only the medium-severity depressed youth (n =123), who did not meet the standard of an affective disorder but had some depression problems, were used in the study. Of the 123 youth, 29 declined to participate. Those participating and not participating did not differ on parent’s age or sex, youth sex, and most youth mental health measures. However, the participants had higher baseline scores on the Center for Epidemiological Studies Depression scale and were roughly a year younger than nonparticipants.

      The remaining 94 youth were randomly assigned to the experimental condition (n = 45) and the usual-care condition (n = 49). Assessments of parents and adolescents occurred at intake, after treatment, and at 12 and 24 months after the end of treatment.

      The intervention consisted of 15 one-hour sessions for groups of 6-10 adolescents led by a therapist who was trained in the cognitive-behavioral approach. Using homework assignments to supplement the sessions, adolescents were taught cognitive restructuring techniques to identify and challenge irrational, unrealistic, or overly negative thoughts. Youth attended an average of 9.5 treatment sessions and completed homework assignments at an average of 46% of the sessions they attended. Sessions were taped and rated for adherence to the study protocol; compliance was 96%. Youth in both conditions could initiate or continue any nonstudy mental health care services provided by the HMO.

      Measures:

      The Achenbach Child Behavior Checklist (CBCL) consists of parent ratings of several social competence items and 113 youth behavior and emotional problem items. Scores were reported for an extracted depression subscale created to match DSM-III-R criteria for major depression and for the standard CBCL externalizing and internalizing symptom subscales.

      The Schedule of Affective Disorders and Schizophrenia for School-Age Children – Epidemiological Version (K-SADS-E25) was administered to adolescents to obtain DSM-III-R diagnoses. Raters were blind to group membership of the adolescents. Interrater reliability was excellent for mood disorders (0.87 for current diagnoses, 1.00 for past diagnoses at baseline, and 0.90 for diagnoses made at follow-up assessments), and good for all other nonaffective disorders (0.74, 0.67, and 0.90, respectively).

      The Center for Epidemiological Studies D19 is a self-report measure of the frequency of 20 depressive symptoms during the past week. Parents and adolescents completed self-rated versions during separate administrations.

      Interviewers completed a 14-item “extracted” version of the Hamilton Depression Rating Scale (HAM-D). The item scores were extracted from corresponding K-SADS-E depression symptom ratings. Reliability for the extracted scale ranged from r = 0.87 to r = 0.94 in adults; alpha = 0.83 for adolescents.

      Interviewers rated severity of impairment using the Global Assessment of Functioning scale (GAF). The scores range from 1 to 90 and include behavioral examples that serve as anchor points. Scores below 40 reflect major impairment.

      Since youth in both conditions were free to continue or initiate use of any other health services, the study used computerized HMO data systems to obtain data on extra experimental health services such as inpatient and outpatient services, prescriptions, and emergency department visits. Research staff interviewed subjects monthly regarding health care services obtained from non-HMO sources.

      Analysis: For continuous variables, the analysis used random effects regression with intercepts and slopes for each participant treated as random variables. The approach allows for use of all 94 cases and an intent-to-treat sample, despite missing data on follow-up assessments. For the dichotomous outcome of experiencing a major depressive episode, survival analysis was used. Treatment-by-time interactions tested for differences in changes across the two groups and number of sessions-by-time interactions tested for dose-response within the treatment group, where time was measured at intake, posttest, 12-month follow-up, and 24-month follow-up.

      Outcomes:

      Baseline Equivalence and Attrition: The two experimental conditions did not differ at baseline on rates of current and past psychiatric disorders or measures of key demographic, depression severity, functioning or other psychosocial characteristics.

      Attrition was modest and did not appear to affect the findings. Of the 94 participants, 2 did not participate in any of the follow-up assessments. Four, 9, and 16 did not participate in the posttreatment, the 12-month interview, and the 24-month interview, respectively. The authors state: “We found few baseline or treatment interaction differences between participating subjects and those unavailable for follow-up at any follow-up point on any of the key demographic, major affective, or psychopathological measures. None of the few differences were consistent across time, suggesting that there was no systematic bias in dropout.” They also replicated the analysis for only the participants with complete data but found that the pattern of results did not change.

      Depression: Significant group-by-time effects favored the experimental group for the Center for Epidemiological Studies Depression Scale and the Hamilton Depression Rating Scale but not for the Child Behavior Checklist for depression. In supplementary cross-sectional comparisons, the Center for Epidemiological Studies Depression Scale differed across groups at posttest and the 12-month follow-up but not at the 24-month follow-up. The Hamilton Depression Rating Scale did not differ significantly across the two groups for any of the cross-sectional comparisons. The time trend was also significantly better for the experimental group on the subset of suicide items from the depression section of the K-SADS-E interview.

      Survival analysis showed that, at the 12-month follow-up, 9% of the experimental group experienced major depression compared to 29% of the usual-care control group. However, the preventive effects of the treatment faded over time, with smaller but significant differences remaining at 18 months and 24 months. For the 9 youths in the experimental group who developed an episode of major depression, the median time of onset was 14.0 months; for the 12 control youths who developed an episode, the median time was 4.0 months. The difference translates into 33 fewer depressed days in the year after intake for the intervention group and represents a significant delay in onset. The differences in the timing of depression onset and days depressed were the only measures of effect size presented.

      The analysis found no significant dose effects based on the number of intervention sessions attended. It did not examine changes in mediators that might account for the intervention effect.

      Nonaffective Outcomes: The results showed no significant differences for any of the nonaffective continuous mental health outcome measures such as the CBCL internalizing and externalizing scales and the number of current diagnoses. The Global Assessment of Functioning Scale showed a significant treatment-by-time effect but no significant cross-sectional difference across groups within each assessment period.

      Limitations:

      • The benefits of the experimental treatment appear to fade over time and show for about half the outcome measures examined.
      • The sample of adolescents who had parents with HMO medical records of depression and themselves had subdiagnostic depression is difficult to replicate.
      • Although little attrition occurred after assignment, the enrolled subjects assigned to the two groups made up only a small portion of the initial pool of potential subjects and tended to be older and more often female.

      Lewinsohn, P.M., Clarke, G.N., Hops, H., & Andrews, J. (1990). Cognitive-behavioral treatment for depressed adolescents. Behavior Therapy, 21, 385-401.

      Compared to the 1999 study described above, this study had a smaller sample and no follow-up booster sessions. It used 14 rather than 16 sessions of treatment over 7 rather than 8 weeks. Otherwise, it was similar in design and outcome to the previous study and provided additional evidence of the effectiveness of group-based cognitive-behavioral therapy for adolescents with depression.

      Evaluation Methodology

      Design: Adolescent individuals with depression were recruited via letters and announcements. After screening, a total of 114 adolescents were interviewed and 69 selected as meeting the inclusion criteria (ages 14-18, having a depressive disorder, and in grades 9-12) and exclusion criteria (having other disorders or needing immediate treatment). Ten withdrew before treatment but after assignment (5 from the wait list, 3 from one treatment condition, and 2 from the other treatment condition).The withdrawals were not studied, leaving 59 subjects for analysis.

      Subjects were randomly assigned to three conditions. The two treatment conditions consisted of 14 2-hour sessions of skill-based, group cognitive-behavioral therapy, one with the adolescent only (n = 21) and one with the adolescent and a parallel parent group (n = 19). A third waitlist group served as the control (n = 19).

      Assessments were made at pretreatment (intake), at posttreatment (after the seven-week course), and at 1, 6, 12, and 24 months after posttreatment. However, given attrition, 10 of 21 and 13 of 19 remained in the two treatment groups at 24 months.

      Sample: The 59 subjects assigned to the three groups differed little on demographic characteristics at baseline. The mean age was about 16, about 61% were female, and 58% lived with one parent. The control group had significantly more siblings.

      Measures: The study used most of the same measures reported above.

      Diagnostic Interviews:: Adolescents were interviewed at intake with the Schedule for Affective Disorders and Schizophrenia for School-Age Children – Epidemiologic version (K-SADS-E). This schedule was used with the DSM-III to make a summary clinical judgment about the presence of depression. Interrater agreement for a sample of the ratings at intake was .83.

      Beck Depression Inventory:: Adolescents completed a 21-item self-report measure of depressive symptoms. Other studies show this inventory has high concurrent validity, internal consistency, and test-retest reliability.

      Center for Epidemiological Studies Depression Scale:: Adolescents completed a 7-item abbreviated form of the full 20-item self-report measure of depressive symptoms over the past week. The alpha coefficient equaled .78.

      Development of Abbreviated Measures for Adolescent Target Behaviors: In addition to self-report depression scales, adolescents complete instruments selected to assess psychosocial constructs known to be associated with depression in individuals: depressogenic cognitions, pleasant events, anxiety, social skills, and conflict resolution skills. These characteristics were targeted by the treatment.

      Parent Report Measures: One or both parents completed the Child Behavior Checklist (CBCL), which assessed 118 adolescent/child behavior problems. Subscales were computed for the externalizing and internalizing categories, as well as for a depression scale. Also, an issues checklist identifies disagreement between adolescents and parents.

      Analysis: For the dichotomous diagnostic measure of having depressive symptoms, the individual-level analysis compared groups with chi-square and ANOVA statistics. For the reported measures from adolescents and parents, group-by­-gender and group-by-time interactions were examined in MANOVA with repeated measures. However, neither the gender main effects nor gender interactions reached significance. The results therefore focused on the group-by-time interactions.

      By excluding 10 dropouts after the initial assignment and 36 dropouts after the posttest, the analysis did not use an intent-to-treat approach. The attrition analysis reported below was used to justify the approach.

      The study did not include a dose-response analysis. Nor did it report effect sizes.

      Outcomes:

      Baseline Equivalence and Attrition: The study mentions but did not analyze 10 subjects who dropped out after being assigned to a condition but before any treatments. About 75% of the intervention group eligible for follow-up completed the 6-month assessment and 50% completed the 24-month assessment. The analysis used only the subjects with completed data rather than an intent-to-treat sample. However, the authors note that a MANOVA comparing the posttreatment scores of those who participated in the follow-up assessment with those who did not revealed no significant differences on either adolescent or parent measures. Also, group-by-time (pretreatment versus 6-month follow-up) repeated measures MANOVAs showed a significant effect only for time, with both groups showing improvements on both adolescent and parent measures.

      Posttest: Based on the K-SADS-E interview classification of depression, the difference in the effect of treatment versus waitlist was significant. At the end of treatment, 52% of the adolescent and parent group and 57% of the adolescent-only group met diagnostic criteria for depression, while 95% of the waitlist group met the criteria. The difference between the two treatment groups was not significant.

      For the reported measures, the MANOVA indicated a significant interaction between group and time such that there were no significant differences between groups at baseline but there were significant differences at posttest. Planned comparisons at posttest showed significantly lower depression for the intervention groups versus the waitlist group on the Beck Depression Inventory and Center for Epidemiological Studies Depression scale but not for any of the other 5 measures obtained from the adolescent or parent. Comparing the adolescent-only with the adolescent-and-parent treatment showed significantly lower scores for the latter group on some parent-reported measures but not on adolescent-reported measures.

      Analysis of mediating or targeted behaviors focused on changes over time across groups on measures of anxiety, pleasant activities, and depressogenic cognitions. Although time had significant effects (i.e., all groups improved over time), the group-by-time interactions for these mediating factors were not significant. The improved depression outcomes for the intervention groups did not correspond closely to similar improvements in the targeted behaviors.

      Longterm: Follow-up data were not available for the control group, which was offered enrollment in the therapy after the posttreatment interview. The analysis examined those remaining in the two treatment groups after attrition to see if the improvements were maintained and if they differed by treatment type. None of the adolescent measures of depression showed significant time effects, group effects, or group-by-time effects, meaning gains were maintained. For most of the parent measures of depression, significant time effects showed better ratings in the follow-up, and significant group-by-time effects showed better improvement in the adolescent-only group. The advantages of the adolescent-and-parent group at posttest thus disappeared by 6 months and both groups did equally well. Overall, then, posttest improvements were maintained in follow-up, and the adolescent-only condition enjoyed additional gains in some measures.

      Over the full 24-month period, diagnoses and depression scores for the intervention groups continued to decline and remained at very low levels. With 100% depressed at intake, the percentages fell to less than 60% at posttest, about 30% at 1 month, less than 20% at 6 months, less than 10% at 12 months, and about 15% at 24 months.

      Limitations:

      • The small sample size reduces the power of the tests, particularly for the follow-up assessments.
      • Analysis was not intent to treat.
      • Gains were maintained at follow-up, but the lack of data on the control group prevents precise comparisons of long-term benefits of the treatment.
      • The lack of effects on mediators means the source of the benefits of the intervention remained unidentified.

      Kaufman, N.K., Rohde, P., Seeley, J.R., Clarke, G.N., & Stice, E. (2005). Potential mediators of cognitive-behavioral therapy for adolescents with comorbid depression and conduct disorder. Journal of Consulting and Clinical Psychology, 73, 38-46.

      This study examined mediators intervening between exposure to cognitive-behavioral therapy in the Coping with Depression program and depression. The key mediators studied were improved social skills, increased engagement in pleasant activities, the use of relaxation techniques, identification of negative irrational thoughts and creation of positive counter-thoughts, and improved problem-solving or conflict resolution skills. In other words, the study tested the theoretical mechanisms that were posited to underlie the program and identified the components that do most to reduce depression. It also aimed to examine the possibility that the treatment benefits stem from non-specific therapeutic factors such as placebo-expectancy and therapist alliance. The data were the same as used in Study 4.

      Evaluation Methodology

      Design: Same as study 4. However, the study focuses only on depression at intake and posttest assessments and the effects of only the acute phase of the intervention. Since the program had no long-term effects compared to the control condition of life-skills training and had no effects on conduct disorder, these analyses excluded follow-up data and conduct disorder measures.

      Sample: Same as Study 4.

      Measures: In addition to the measures of depression used previously, the study includes measures of specific and non-specific therapeutic factors that may mediate the observed intervention effects.

      Cognitions: A 30-item Automatic Thoughts Questionnaire assesses the frequency of occurrence of negative thoughts (alpha = .98), and a 9-item Dysfunctional Attitudes scale assesses the degree of agreement with attitudes and beliefs sometimes held by depressed individuals (alpha = .80).

      Relaxation: Four items from the Pleasant Activities Schedule relate to being relaxed, thinking about something good in the future, having peace and quiet, and sleeping soundly at night (alpha = .77).

      Social Skills: Seven items from the Pleasant Activities Schedule measure participation in activities requiring use of social skills such as meeting someone new of the same sex, having a lively talk or discussion, and having people show interest in what I’ve said (alpha .75).

      Pleasant Activities: The remaining dichotomous 41 items from the Pleasant Activities Schedule were summed to measure engagement (over the past month) in activities that were known to correspond to improved mood (alpha = .90).

      Problem Solving: The mean intensity rating by adolescents of parent-adolescent discussions concerning 19 potential conflict events over the past two weeks served as a proxy measure of problem-solving or conflict-resolution skills (alpha = .87).

      Working Alliance: To measure one non-specific therapeutic factor, a 12-item Working Alliance Inventory used items from the adolescent about confidence in the instructor’s ability to help and trust between instructor and client (alpha = .89).

      Group Cohesion: To measure another nonspecific therapeutic factor, the adolescents rated group cohesiveness using 14 items from the Group Cohesiveness Questionnaire (alpha = .89).

      Analysis: Regression-based path models at the individual level examined 1) the effect of the treatment on the mediators, 2) the effect of the mediators on the outcome depression measures, and 3) attenuation of the relationship between the intervention and depression with controls for the mediators. Estimates come from MPlus, which uses maximum likelihood methods to include missing data. The analysis controls for baseline characteristics with inclusion of group-by-time interactions, but it did not present a dose-response analysis.

      Outcomes:

      As in study 4, this study demonstrated a relationship between the intervention and two posttest outcomes – the Beck Depression Inventory (d = .48) and the Hamilton Depression Rating Scale (d = .44). Since the previous study showed that the intervention did better than the control condition (i.e., the life-skills class) in the posttest but not in the follow-up assessments, the analysis examines posttest data only.

      Concerning the influence of treatment on the mediators, the treatment group showed significantly greater improvement on one specific therapeutic factor – the Automatic Thought Questionnaire measure (d = .50) – and one non-specific therapeutic factor – the Working Alliance Inventory. The two groups did not differ on measures of behavior or on long-standing beliefs or assumptions associated with depression.

      Concerning the influence of the mediators on the outcome, cognitions measured by the Automatic Thoughts Questionnaire related strongly and significantly to the Beck Depression Inventory (beta = .71) and the Hamilton Depression Rating Scale (beta = .36). However, the Working Alliance Index did not have a significant relationship with depression scores.

      Concerning the attenuation of the intervention effect with controls for the mediators, adding the Automatic Thoughts Questionnaire score to the model reduced the effect of the intervention on the Beck Depression Inventory by 75% and to insignificance. Adding the score reduced the effect of the intervention on the Hamilton Depression Scale but by a smaller magnitude (by 40%).

      Limitations:

      • The mediating variables were measured at intake and posttest – the same time depression outcomes were measured. The relationships assume rather than demonstrate that the mediators cause the outcome. Rather than causally related, the mediators and outcomes may be part of broad improvements in mental health from intake to posttest.
      • The measures relied on self-reports of skill acquisition and depression rather than direct assessment.
      • Only one of the several potential mediators improved under the intervention.
      • Rates of attendance at the sessions were low.

      Rohde, P., Seeley, J.R., Kaufman, N.K., Clarke, G.N., & Stice, E. (2006). Predicting time to recovery among depressed adolescents treated in two psychosocial group interventions. Journal of Consulting and Clinical Psychology, 74, 80-88.

      This study used the same data as studies 4 and 5, but differs from the previous one by focusing on the time to recovery from major depressive disorder. In addition to comparing outcomes for the cognitive-behavioral therapy intervention versus a life-skills class, the study examined how sociodemographic and other characteristics predict recovery and moderate the influence of the intervention. The moderation analysis allowed the investigators to identify subgroups that benefitted most from the treatment.

      Evaluation Methodology

      Design: Same as study 4. However, the study did not require comorbidity with conduct disorder, and therefore could use 114 subjects (rather than the 93 in studies 4 and 5). These subjects were randomly assigned to two conditions: the Coping with Depression treatment (n = 56) and the life-skills course (n = 58). Assessments were done at intake, immediately after completion of the intervention, at 6 months posttreatment, and at 12 months posttreatment.

      Sample: The sample had a mean age of 15.2 years (SD = 1.4) and was 48% female and 52% male. The sample included White (71%), African-American (1%), Hispanic (1%), Asian-American (1%), and Native-American (1%) individuals, as well as participants endorsing “other” or “mixed ethnicity” (25%). Fifteen percent of adolescents lived with both biological parents, 21% lived with a biological parent and stepparent, 23% lived with biological mother only, and the remainder lived with other relatives or in other arrangements.

      Measures: Diagnosis of major depressive disorder and its absence followed procedures used in previous studies. For those having recovered, posttreatment diagnostic data identified the week of recovery.

      Other variables fall into several groups:

      • demographic (sex, age, race/ethnicity),
      • depression psychopathology (age of onset, number of prior episodes of major depressive disorder, depression severity, suicide ideation),
      • broader psychological factors (current ADHD, current substance abuse or dependence, current anxiety disorder, functional impairment, parent reports of problem behavior),
      • cognitive-behavioral-specific psychosocial factors (negative thoughts, dysfunctional attitudes, hopelessness, frequency of pleasant events), and
      • psychological resiliency factors (social adjustment, family cohesion, coping skills).

      Analysis: Using an intent-to-treat analysis at the individual level, the study treated time to recovery as the dependent variable (and therefore could include censored subjects who did not complete all assessments). The Cox proportional hazard model used measures of the determinants at intake, thus maintaining proper temporal order for causal inferences. Treatment was measured dichotomously throughout rather than as a continuum in a dose-response model. Hazard ratios were used to report effect sizes.

      Outcomes

      Two variables made a unique contribution to predicting recovery time in the multivariate model: parent report of total problem behaviors and suicide ideation. Treatment condition interacted with 3 of the 19 examined factors: race/ethnicity, number of previous episodes of major depressive disorder, and coping skills. In other words, the intervention speeded recovery for whites but not non-whites; for adolescents with two or more episodes but not for those with experiencing their first episode at intake; and for adolescents with positive coping skills but not for those with poor coping skills. To illustrate, the hazard ratio or risk of recovering of the treatment group to the control group equaled 1.76 for whites and 1.04 for non-whites. It equaled 2.67 for those with two or more previous episodes and .94 for others, and it equaled 1.79 for those with positive coping skills and 1.04 for others.

      Limitations:

      • The timing of recovery was based on assessments that required up to 6 months of retrospective recall.
      • High correlations among the predictors, including both main and interaction terms, produce many bivariate influences on recovery time but few multivariate influences; the high correlations may make the results sensitive to the particular sample.
      • The sample includes a high proportion of depressed adolescents with conduct disorders, who may differ in treatment response from adolescents with depression alone.