Interpersonal Psychotherapy – Adolescent Skills Training

Study 1

Evaluation Methodology

Design:

Recruitment: Ten middle and high schools in central New Jersey were recruited using letters and brochures. From a total pool of 9,123 students, 2,923 (32%) obtained parental consent, provided assent, and completed the screening process. Students who scored at 16 or above on the Center for Epidemiologic Studies - Depression Scale (CES-D) were eligible for inclusion in the study. Of the 593 students with elevated CES-D scores, 271 agreed to participate in an eligibility evaluation. They were then given the Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS-PL), and the 186 who had at least two subthreshold or threshold depressive symptoms, no currently diagnosed major depression or other psychiatric disorders, no significant suicidal ideation, and no cognitive impairments were included in the study.

Assignment: Within each school, students were stratified on gender and then randomly assigned to either the intervention group (n = 95) or the group counseling active control group (n = 91), which met as often as the intervention group and allowed counselors to select therapeutic techniques.

Attrition: Assessments occurred at baseline, mid-intervention, posttest, and at 6-month follow-up. With the addition of booster sessions, however, the 6-month follow-up may also serve as a posttest. Of the 186 total students in the study, 184 (98.9%) completed the mid-intervention assessment, 183 (98.4%) provided posttest data, and 175 (94.1%) completed the 6-month follow-up.

Sample:

Across all students, the average age was around 13.5 years, about 67% of the sample was female, about 30% were a racial minority, 38% were Hispanic, and 38% were White. Family income of the sample ranged widely, but 44.3% earned more than $90,000, suggesting a middle-class sample.

Measures:

Depressive symptoms were measured using the self-reported Center for Epidemiologic Studies - Depression Scale (CES-D), with Cronbach's alphas ranging from 0.86 - 0.89. Diagnoses for depression disorders were determined using the Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS-PL). As part of the K-SADS-PL, students were also assigned a score of overall functioning using the Children's Global Assessment Scale (CGAS). This score measures functioning at home, school, and with peers, with scores over 70 indicating minor impairments in functioning (lower scores indicate greater impairments.) Reliability measures were 0.96 for depression diagnoses and 0.89 for overall functioning. Ratings for both clinical depression and overall functioning came from independent evaluators who were unaware of condition.

Analysis: For the measures of depressive symptoms and overall functioning, a 3-level hierarchical linear model was used to account for the multiple levels of clustering within the data. Individual scores over time (including baseline) were modeled at the first level, the individual intercept and slope were modeled as outcomes dependent on the group condition at the second level, and the group specific slopes were modeled at level 3 to contrast the rate of change between the two groups. Cox regression analyses were used to examine group differences in time to the onset of a depression diagnosis.

Intent-to-Treat: Analyses were conducted on an intent-to-treat basis using all randomized subjects.

Outcomes

Implementation Fidelity: Attendance in the intervention group was about 6.8 group sessions on average, out of a total of 8 group sessions. Ratings of a sample of sessions showed that 98.5% of the techniques were delivered with fidelity and were given a rating of satisfactory (49.0%) or superior (49.5%) for technique delivery. A global competency rating was given to each group leader: 8.8% of the leaders received a rating of satisfactory, 41.2% good, and 50.0% excellent.

Baseline Equivalence: Tests found no statistically significant group differences on any of 20 outcome or demographic variables at screening or baseline.

Differential Attrition: Overall attrition was quite low - less than 2% at posttest and 6% at the 6-month follow-up.

Posttest: The rate of decrease in depressive symptoms was significantly greater in the intervention group than in the control group (d = .31), and the intervention group showed significantly greater improvements in functioning (d = .31), compared to the control group. There was no significant difference in the hazard rates of the onset of depression diagnoses between the two groups.

Long-Term: Not measured.

Study 2

Evaluation Methodology

Design:

Recruitment: Students in grades 7-10 in three Catholic schools in New York City whose parent had given consent were screened for symptoms of depression. The screening used the self-report Center for Epidemiologic Studies - Depression Scale. Students with a score between 16 and 39 were eligible (30.7%); those with a score of 40 or higher were referred out for treatment. After being approached for participation, about half of the families (N = 55, 49.1%) agreed. There were no significant differences in screening depression score, age, or gender between those who did and did not participate. After an additional diagnostic screening, 14 subjects were dropped as not having at least two depression symptoms or having a current diagnosis of depression or other major diagnosis, leaving a sample of 41.

Assignment: The study randomly assigned about two-thirds of the 41 adolescents to the intervention group (N = 27) and the remainder to school counseling (N = 14), the normal procedure for students having emotional difficulties.

Attrition: Assessments of the participants occurred at posttest (about 12.6 weeks after randomization), and 3 and 6 months after the program ended. Only one subject (2.4%) was lost in the posttest and follow-ups.

Sample: The sample was largely female (85.4%) and Hispanic (92.7%). Age ranged from 11 to 16, with an average of 13.4. Most (66%) lived in a single-parent household and half reported household income of $25,000 or less. Several of the adolescents met criteria for a current non-affective DSM-IV diagnosis, but the majority (75.6%) had only subthreshold depression symptoms with no current diagnosis.

Measures: The measure of depression symptoms came from the self-reported Center for Epidemiological Studies - Depression scale. A measure of depression diagnosis came from the interviewer-rated Schedule for Affective Disorders and Schizophrenia for School-Age Children, which was completed by a Masters-level psychologist who was blind to treatment condition. The Children's Global Assessment Scale measured overall functioning and was scored by the interviewer after the diagnostic session.

Analysis: The study used analysis of covariance with baseline outcomes treated as covariates.

Intent-to-Treat: One adolescent dropped out of treatment prior to the first group session but completed all assessments. Otherwise, the analyses imputed missing data by carrying forward the last observation.

Outcomes

Implementation Fidelity: Intervention youth attended an average of 2.0 pre-group sessions and 6.9 of 8 group sessions, but no other fidelity information was reported.

Baseline Equivalence: No significant differences between conditions emerged for any of 13 baseline demographic and clinical variables.

Differential Attrition: Attrition was minimal.

Posttest: At the posttest, adolescents in the intervention group reported significantly fewer depression symptoms on the self-reported Center for Epidemiological Studies - Depression scale than adolescents in the control group (with adjustment for baseline outcomes). The differences remained significant at the 3-month and 6-month follow-ups. The d values were large, ranging from 1.09 to 1.52.

For overall functioning, there were significant differences at posttest, 3-month follow-up, and 6-month follow-up (d values ranged from .82 to 1.21).

Over all assessments, 1 of 27 (3.7%) intervention adolescents had a depression diagnosis compared to 4 of 14 (28.6%) control adolescents (p = .08).

Long-Term: Not tested.

Study 3

Evaluation Methodology

Design: Using a randomized controlled design, researchers sought to determine the effectiveness of the IPT-AST program for reducing adolescent depression compared to a treatment as usual condition. Parents of students in grades 9 and 10 in two girls-only high schools and one boys-only high school were sent letters about the program and were provided two opportunities to refuse participation. Three hundred forty-six (31.0%) parents and 125 (11.2%) adolescents refused to participate in the screening; 4 adolescents were repeatedly absent so were not screened. Consenting students were screened in their classrooms for symptoms of depression and 237 students were found at-risk for depression and therefore eligible for the study.

One third (n=79) of the eligible students consented for an eligibility evaluation. Adolescents were excluded from the study (n=17) if they had a current diagnosis of depression, dysthymia, bipolar disorder, psychosis, panic disorder, obsessive-compulsive disorder, posttraumatic stress disorder, oppositional defiant disorder, conduct disorder, or untreated attention deficit hyperactivity disorder; adolescents were also excluded if they did not have sufficient depression symptoms (n=4). One additional student left school, resulting in 57 students being randomly assigned to the intervention (n=36) and control (n=21) condition. Each of the schools was randomized to include parental involvement during either the first or second year of the study. Therefore, within the intervention condition, 15 adolescents received IPT-AST with an additional parental involvement component.

Data were gathered at baseline, immediate posttest, and 6-month, 12-month and 18-month follow-up. After baseline, assessment completion rates for the intervention group were 97%, 94%, 94% and 89% respectively. After baseline, assessment completion rates for the control group were 100%, 86%, 67% and 57% respectively.

Sample Characteristics: The sample was 59.7% female, 73.7% Hispanic and 70.2% living in single-parent homes; the average age was 14.51.

Measures: A total of four measures were gathered for depression and functioning. Each assessment was conducted by independent evaluators blind to the condition and consisted of the Kiddie-Schedule for Affective Disorders which provided a depression diagnosis, Children's Global Assessment Scale, Center for Epidemiologic Studies - Depression Scale, and the Children's Depression Rating Scale-Revised. These measures have been used in previous studies and described in previous sections of this write-up. An analysis of eight audiotaped assessments showed high interrater reliability (ICC > .89).

Analysis: Differences between the interventions for rates of change in depression and functioning were examined using hierarchical linear models. Mean differences between the interventions (with and without parent involvement) as well as between intervention and control groups at post-intervention and follow-up were examined using analysis of covariance with baseline scores and schools used as controls. Rates of depression diagnoses on the K-SADS were compared using Fisher's Exact Test. Effect sizes were reported as Cohen's d.

Intent-to-Treat: Data were analyzed based on an intent-to-treat analysis with imputed missing data (using the last-assessment-carried-forward method).

The paper noted that clustering within the therapy groups might bias estimates of standard errors. However, a model with random effects for group did not converge and adjustments were not made.

Outcomes

Implementation Fidelity: No information on fidelity to the program was provided. On average, intervention adolescents attended 1.94 of 2 pre-group sessions and 5.22 of 8 group sessions. Control group adolescents attended an average of 3.76 sessions.

Baseline Equivalence: The study reported no significant differences between groups at baseline on any demographic or outcome variables.

Differential Attrition: No significant differences were found between adolescents who consented and those who refused to participate based on age or gender. However, consenting adolescents had significantly higher depression scores at initial screening (p<.01).

Attrition was considerably greater in the control group. Concerning differential attrition from baseline to posttest and follow-up, the paper provided little other information, perhaps because missing data were imputed. It did say that the intervention effects were not sensitive to missing data.

Posttest: Intervention adolescents showed significantly greater rates of change than control adolescents on the two depression scales (CES-D, p=.01 and CDRS-R, p<.01) and one overall functioning scale (CGAS, p<.01). No significant differences between rates of depression diagnoses were found using the K-SADS between the two groups at immediate posttest; however, this difference was significant in the expected direction at 6-month follow-up (p<.05).

Significant mean differences were found at immediate posttest between groups as measured by the CES-D, CDRS-R and CGAS (p's<.001, d's ranged between 0.80 and 1.27).

Long-term: In the 18 months following the intervention, rates of change in symptoms were significantly different between groups on one of the two depression scales (CES-D, p=.01) and one overall functioning scale (CGAS, p=.02). No significant differences between rates of depression diagnoses were found on the K-SADS between the two groups.

Significant mean differences were found at 6-month follow-up for overall functioning on the CGAS (p=.03). No significant mean differences were found at 12- or 18-month follow up for any of the measures.

No significant differences between the two intervention conditions (with and without parent involvement) were found at any time-point.

Study 4

This study assessed the efficacy of two adolescent depression prevention programs. One was a cognitive-behavioral approach (CB) and the other was Interpersonal Psychotherapy - Adolescent Skills Training (IPT-AST). Based on previous research, the authors hypothesized that both programs would be more effective than the no-intervention control group. Also, they tested how baseline levels of depressive symptoms, gender, and sociotropy (the degree to which an individual's sense of self is dependent on relationships) moderated program effects on depression, and how changes in cognitions and coping skills (expected for CB) and improvement in personal relationships (expected for IPT-AST) mediated program effects on depression.

The programs were used as a universal prevention in high school wellness classes. In this version of IPT-AST, the materials normally covered during the two individual pre-group sessions were incorporated into the first group session and group sizes were somewhat larger than in studies with selective samples.

Evaluation Methodology

Design: The participants in this RCT study were in wellness classes in three suburban/rural high schools from predominantly working- to middle-class communities. The authors did not report how the three high schools were selected. An initial cohort was recruited in January 2004, received the program in spring 2004, and took the follow-up in October 2004. A second cohort was recruited in August 2004, received the program in fall of 2004, and completed the follow-up in April 2005.

Six-hundred students were initially recruited and potentially eligible for the program. The researchers were unable to obtain parental consent for 220 students (37%), which left an initial sample of 380 students. Participants and non-participants were not significantly different with regard to age, gender, or race. The participating students were randomly assigned to the cognitive behavior program (CB), the interpersonal psychotherapy program (IPT-AST), or the control group that received the school's typical wellness program. Because the treatment classes were limited to 15 participants and the number of classes per school was constrained by the school, the number of students in each treatment category was not equal (112 for CB, 99 for IPT-AST, 169 for control), with any students enrolled in classes larger than 15 non-randomly assigned to the control group.

A total of eight group leaders and eight coleaders led the groups which met once weekly for 90 minute sessions for eight weeks. Groups were of mixed gender and were a median size of 11 students.

Baseline questionnaires were completed the week before the program began, posttests were completed a week after the final session, and the follow-up was completed six months after the program ended.

Attrition: Consistent with an intent-to-treat approach, students who left the school or lost interest during any time during the six months were contacted by mail and telephone and urged to continue contributing data. Of the 380 students assessed at baseline, 99% completed the posttest and 84% (CB 79%, IPT-AST 85%, control 84%) completed the 6-month follow-up questionnaire.

Sample characteristics: Most of the sample subjects were freshmen in high school (94%); the average age was 14 years; and 54% of the sample was female. The sample was 79% Caucasian, 13% African American, 2% Latino, 1% Asian American, 1% Native American, 3% mixed heritage, and 1% other. The schools were in working- and middle-class neighborhoods.

Measures: The indicators necessary for this evaluation were for depressive symptoms, sociotropy, achievement orientation, cognitions, coping, quality of relationships, and curriculum knowledge. The following measures were used:

Depressive symptoms. Depressive symptoms were measured using the Children's Depression Inventory (CDI) and the Center for Epidemiological Studies Depression Scale (CES-D). The CDI is a 27-item self-report measure, but the suicide ideation measure was removed due to concerns expressed by the schools. A 26-item CDI scale was used and internal consistency of this scale at baseline was .89. The CES-D is a 20-item self-report measure of depressive symptoms, and the internal consistency of this scale at baseline was .86. The researchers also created a composite from these two measures. The composite had a high level of reliability (.93).

Sociotropy and achievement orientation: Sociotropy and achievement-orientation were measured using the Sociotropy-Achievement Scale for Children (SASC). The SASC contains 48 self-report items and yields scores on two factors: sociotropy/affiliativeness (high scores on items such as "I care a lot about what other people think of me.") and achievement orientation (high scores on items such as "I should be able to do well at anything if I try hard enough."). The alphas at baseline were .90 for sociotropy and .86 for achievement orientation.

Cognitions: Cognitions (attributional style) was measured with the Children's Attributional Style Questionnaire (CASQ-R). Lower scores reflect a more negative attributional style. The alpha at baseline for this sample was .81.

Coping: Coping was measured by the Coping Orientation to Problems Experienced Inventory (COPE). The COPE asks respondents how often they do each of 60 possible things when experiencing a stressful event. The inventory has three subscales -- rational coping, emotional coping, and avoidance coping. The alphas for this sample were .86, .94, and .70 respectively.

Quality of relationships with parents: This was measured using the 20-item, true-false Conflict Behavior Questionnaire. The respondents completed questionnaires for their mothers and fathers separately. The alphas for this sample were .87 for mothers and .88 for fathers.

Curriculum knowledge: Assessments were created by program designers based on each program's curriculum. Each consisted of seven multiple choice questions with four choices each. CB knowledge produced an alpha of .29 and IPT-AST knowledge produced an alpha of .62.

Analyses: To predict depressive symptoms based on treatment status or another categorical variable, the researchers used ANCOVA with baseline depressive symptoms as a covariate. To predict depressive symptoms using a continuous variable, linear regression was used with baseline depressive symptoms in the first step and other predictors in subsequent steps. Effect sizes were calculated using Cohen's d.

Outcomes

Fidelity: To avoid cross-contamination, group leaders administered only one of the two interventions. Group leaders were graduate students or PhDs in clinical psychology, and coleaders were graduate students or undergraduate honors students. Detailed manuals were used, group leaders and coleaders participated in training workshops, and the leaders met weekly with clinical experts in the respective programs.

Baseline equivalence: Students in the three conditions did not differ significantly in age, sex, or race (p-values were not reported).

Differential attrition: Baseline depression scores were not significantly associated with treatment status, attrition status, or their interaction. However, posttest depression scores were significantly associated with the treatment status x attrition status interaction. Specifically, within the control group, participants who did not complete the 6-month follow-up had higher posttest depression scores than those who did complete the 6-month follow-up. This pattern did not hold for the intervention groups.

Three different methods were used to estimate the effects of missing data on the results. None of the analyses yielded significant differences at follow-up on any index of depressive symptoms.

Posttest and 6-month follow-up: Results can be organized by main effects of the intervention, moderators, mediators, and effects among high- and low-risk groups.

Main effects: The effects of the programs were evaluated using three different measures of depressive symptoms -- the CDI, the CES-D, and the composite. Controlling for baseline CDI scores, the CB group had significantly lower CDI scores at posttest (ES=.40), while the IPT-AST group did not have significantly lower CDI scores at posttest. Controlling for baseline CES-D scores, neither the CB group nor the IPT-AST group had significantly lower CES-D scores at posttest. Controlling for composite depression scores, both the CB treatment and the IPT-AST treatment showed significant effects on composite depression scores (ES=.37 and ES=.26, respectively). No program effects were sustained at the 6-month follow-up.

Moderators: The researchers proposed a number of potential moderators of the effect of the programs on depressive symptoms, including level of baseline depressive symptoms, gender, sociotropy, and achievement orientation. Baseline CDI scores significantly moderated the effect of both CB (beta=-.12, p<.05) and IPT-AST (beta= -.13, p<.05), but this effect disappeared at 6-month follow-up. However, there was no moderating effect of baseline CDI scores at 6-month follow-up. When the CES-D measure of depression was used, the researchers found no moderating effects of baseline depressive symptoms.

There were no main effects or moderation effects for gender using CDI or CES-D at posttest or at 6-month follow-up.

Sociotropy moderated the effect of IPT-AST for the CES-D measure (beta=-.77, p<.01). In other words, the IPT-AST program was more effective for those who scored higher on measures of affiliativeness than for those who scored lower on measures of affiliativeness. This relationship was also found in the control group, however. Instances when sociotropy was not a significant moderator included the CB program, either program when CDI was used to measure depressive symptoms, or any 6-month follow-up for any depressive measure or program.

Achievement orientation was found to moderate the effect of IPT-AST for the CDI measure (beta=-.20, p<.01), but not found to moderate the effect of the CB program. In other words, the IPT-AST program was more effective for those who were more achievement oriented than for those who were less achievement oriented. Instances when achievement orientation was not a significant moderator include the CB program, either program when CES-D was used to measure depressive symptoms, or any 6-month follow-up for any depressive measure or program.

Mediators: Neither programs' effects were significantly mediated by cognition/attributional style. Neither programs' effects were significantly mediated by rational and active coping, emotion-based coping, or avoidance coping. Neither programs' effects were significantly mediated by perceived conflict with mothers or fathers. The CB program effects were not significantly mediated by knowledge about the program, but IPT-AST program effects were significantly mediated by knowledge of the IPT-AST program.

High-Risk Subgroup: This subgroup was defined as those who scored in the "top 25th percentile" in the composite depression measure. Both CB and IPT-AST programs showed significant effects at posttest (ES=.89 and ES=.84, respectively) that were much larger than the effects for the sample as a whole.

Low-Risk Subgroup: This subgroup was defined as those who scored in the "bottom 75th percentile" in the composite depression measure. Neither CB nor IPT-AST programs showed significant effects at posttest.