Principal Investigators: Jane E. Gillham, Lisa H. Jaycox, Karen J. Reivich, & Martin E. P. Seligman
Level of Intervention: Selected
Target Population: Middle-school aged children (10 to 13 years) at-risk due to elevated depressive symptoms or family conflict.
References: Gillham, Reivich, Jaycox, & Seligman (1995); Jaycox, Reivich, Gillham & Seligman (1994).
Theory (Risk & Protective Factors Targeted):
The Depression Prevention Program is designed to combat cognitive distortions (e.g. pessimistic explanatory style, hostile attribution bias, poor social problem solving) and related deficits associated with depression, such as behavior problems, poor peer relations, lowered self-esteem, and poor academic achievement. Research has shown that the combination of depressive symptoms, lowered self-esteem, and conduct problems are often associated with childhood stresses such as marital or family conflict. Past research has demonstrated the efficacy of cognitive-behavioral techniques in the treatment of depression and social problem solving training for improving childrens adjustment.
Description of Intervention:
The intervention was based on cognitive-behavioral principles. Children were taught coping strategies to counteract cognitive distortions and deficiencies, specifically explanatory style. One component of the program (Cognitive Component) focused on teaching children how to interpret problem situations in more adaptive ways by identifying negative beliefs, evaluate the evidence for beliefs, and generating alternatives. This portion of the program included explanatory style training in which children are taught to identify pessimistic explanations and generate more optimistic & realistic explanations to their problems. The second component of the program (Problem Solving and Coping Component) focused on the childrens actions for solving their problems by teaching social problem solving and adaptive coping. Children were encouraged to think about their goals before acting, generate solutions, and weigh the pros and cons to their solutions. In addition, this portion of the program taught the children skills for managing parental conflict, and behavioral techniques to enhance assertiveness, negotiation, and relaxation.
Over the course of 12 weeks, treatment groups met weekly after-school for 1.5 hours. Groups consisted of 10-12 members and included in-session instruction and weekly homework assignments. Three doctoral students led the groups.
Treatment subjects were recruited by letter from all 5th and 6th grade students (approximately 900 students) in a school district outside of Philadelphia, PA. Two groups of children were identified as "at-risk" for depression: those with elevated symptomatology and those in homes with marital conflict and low family cohesion. Children from 7 elementary schools were screened using the Child Depression Inventory (CDI; Kovacs, 1985) and the Child Perception Questionnaire (Emery & OLeary, 1982), a measure of childrens perceptions of marital conflict. A risk score was created by standardizing and summing the two measures. From a pool of 262 children, 149 children with a score greater than .50 were considered "at-risk". Mean score for both groups pre-treatment on the CDI was around 10. Using similar procedures, control subjects were recruited by letter (approximately 700 students) from a second suburban school district. The final sample consisted of 69 treatment subjects (34 girls, 35 boys) and 74 no-participation control subjects (32 girls, 42 boys). Eighty-three percent of the children were Caucasian and 11% were AfricanAmerican.
The evaluation used a mixed-method, nested design with unbalanced groups. The study was unable to use true random assignment because the program was conducted district wide. One school with a higher income level was assigned to the wait-list condition so that if there was any bias in assignment it was not in favor of the intervention condition. Risk scores at each school indicated that there were between group differences on the selection criteria scores. High and low risk groups were paired prior to assignment to intervention or control groups. Experimental conditions were then randomized to one of three treatment conditions (Cognitive, Social Problem Solving, or Combined Treatment) or a control condition so that the assignment of condition to pairs was unbiased.
There were no differences between intervention and control groups on any pre-test dependent measures but differences were found for two socioeconomic indices. Families of children in the control group reported higher income levels and higher levels of education. There were no differences between the two intervention components so the groups were collapsed and compared to the control condition. No differences in attrition by condition was reported. ANCOVA and MANCOVA were used for analyses to control for initial levels of depression and behavior problems. One-tailed P-values were used for between-group analyses when the authors felt there was a clear, unidirectional prediction that the treatment group would do better than the control group. Some analyses used two-tailed tested. The authors noted that compliance for assessments of conduct problems was low. Parent ratings were not available to the same degree at each assessment which reduced the sample size.
Treatment children reported significantly less depressive symptoms on a composite of the CDI and Reynolds Childrens Depression Inventory (Reynolds, 1989) compared to controls (p< .05). Treatment children exhibited better classroom behavior according to codings of teacher report cards compared to controls (p<.05). There were no group differences on the Childrens Attributional Style Questionnaire (Kaslow, Tannenbaum, & Seligman, 1978), a measure of explanatory style, but treatment children were less likely to attribute negative events to stable, enduring causes (p<.05). Explanatory style mediated the impact of the treatment on depressive symptoms. There were no group differences on parental ratings of externalizing or internalizing problems on the Child Behavior Checklist (CBCL; Achenbach, 1991).
Treatment children reported significantly fewer depressive symptoms on the composite variable compared to controls (p< .05) and on a retrospective report of depression created from the CDI (p<.05). Parents of treatment children reported significantly less externalizing symptoms on the CBCL compared to controls (p<.05). There were no significant group differences on parent ratings of internalizing symptoms.
No significant interactions were found between the level of childrens perceptions of parental fighting and treatment on any measures. When groups were divided at the median on childrens perceptions of parental conflict, children in the treatment who reported high parental conflict reported significantly less depressive symptoms at post-test (p<.05) and follow up (p<.05) compared to the parallel control group. The high conflict treatment subjects also reported significantly less internalizing symptoms at post-test (p<.05) compared to high conflict controls. There were no significant treatment effects for self-reported depressive symptoms or more general internalizing symptoms between the low conflict groups. In addition, no intervention effects were found for externalizing behavior for either the low or high parent conflict at post-test. At follow-up, parents of the children in the low conflict treatment group reported significantly fewer externalizing symptoms for their children compared to controls (p<.05).
Similar analyses were conducted by creating groups according to severity of depressive symptoms (median splits). Treatment group children with high depressive symptom levels reported significantly less depressive symptoms at follow-up compared to control subjects with high symptom levels (p<.05). There were no significant differences between low symptom treatment and control subjects in terms of depressive symptoms at post-test or follow-up.
Follow-up (12, 18, and 24 months)
These results were summarized in Gillham et al. (1995). Treatment children reported significantly fewer depressive symptoms on the CDI compared to controls (p<.01). Planned comparisons revealed that the difference was significant at the 18 month (p<.01), and 24 month (p<.01) assessments. Across the follow-up period, children who participated in the intervention were also less likely to report moderate or severe depressive symptoms (score of 15 or above) on the CDI (P<.01) and this differences was significant at the 12 month (p<.01), 18 month (p<.01), and 24 month (p<.05) assessment points. The intervention continued to affect explanatory style in the participants. Overall, across the follow-up period, intervention children had a significantly more optimistic explanatory style (p<.01) than children in the control group and this difference was significant at the 12 month (p<.01), 18 month (p<.01), and 24 month (p<.01) assessment points. Explanatory style continued to mediate the effect of the intervention on depressive symptoms. Finally, while depressive symptoms increased significantly over time for both treatment and control groups (p<.01), a significant interaction of time and condition indicated that the control group evidenced a greater increase in depressive symptoms than the treatment group.
Strengths & Limitations:
The Depression Prevention Program is an intervention that targets the cognitive distortions style and deficits associated with depression by providing cognitive-behavioral skills training to individual children identified as "at-risk" for developing depression. The intervention included a unique target population that accounted for two different pathways to depressive disorders: through elevated depressive symptoms and through comorbidity with conduct problems. This attempt to address the comorbidity between internalizing and externalizing problems is rare in the prevention literature. The results of an initial evaluation (Jaycox et al., 1994) indicated that the program was successful in reducing participants self-reported depressive symptoms and that six month post-intervention, parents and teachers reported improvements in the conduct of these children. The results of an second published evaluation (Gillham et al., 1995) were consistent with the initial findings and indicated extended treatment effects including reductions in depressive symptoms over a two year period.
It is important to take a number of factors into account when interpreting these results. The quasi-experimental nature of the design limits the generalizability of the findings. Response rates to the initial screenings for subjects were low and, as a result, the study participants may not represent the general sample of depressed individuals. In addition, as the authors noted, there were problems with assessment compliance, and the analyses of conduct problems utilized a reduced sample. Only child ratings were used to measure marital conflict and the retrospective report used to assess child depressive symptoms was a new measure without reliability data. A strength of the program was that a detailed training manual was used to ensure program fidelity. The program has not been independently replicated.