In a sample of chronically depressed adolescents who had already failed to respond to an adequate trial of SSRI treatment, nearly 40% achieved remission after 6 months of randomly assigned treatment in the TORDIA study. Initial treatment assignment did not affect the rates of remission, but greater clinical severity of depression predicted both failure to remit and, among responders, greater likelihood of relapse. By week 6, the rate of decrease of depressive symptoms among those who remitted had already begun to diverge from the rate among those who did not remit.
The rate of remission in the TORDIA study at 24 weeks was lower than the reported rate of 60% in the Treatment for Adolescents with Depression Study (TADS), despite similar lengths of treatment (21
). Treatment-resistant depressed samples show lower rates of remission with each subsequent treatment step (22
). The TORDIA study sample had greater chronicity and higher levels of suicidal ideation than the TADS sample, both of which have been shown to predict a less robust treatment response (23
Our findings point to the importance of the early trajectory of treatment response in determining remission after 6 months. Initial response at 12 weeks predicted a greater than threefold increased likelihood of remission. Moreover, the clinical course of those who eventually remitted already began to diverge by 6 weeks of treatment, with individuals who remitted showing a rate of decline in symptoms that was nearly twice the rate found among those who did not remit after 6 months, which is consistent with a previous report (25
In addition, among nonresponders, augmentation with either a mood stabilizer or psychotherapy offered during the first 12 weeks of treatment resulted in eventual remission. These augmentation findings parallel those in controlled trials of adolescents and adults (22
). The same interventions offered during the second 12 weeks of treatment did not affect the remission rates among nonresponders.
We found no enduring effects of initial treatment assignment to CBT with respect to remission, protection from relapse, and protection from adverse events. Other studies have also shown that the effects of various initial treatments converge over time (21
), and while one study found protective effects of CBT against adverse events, another did not (31
). A greater number of sessions of CBT might be required to prevent relapse (27
). The open treatment provided in the TORDIA study, such as offering CBT to nonresponders who initially did not receive it, also contributed to the convergence of outcomes among the different treatment arms.
Counter to hypotheses and some reports on depressed adults, we did not find that treatment with venlafaxine was superior to treatment with SSRIs in achieving remission, although the TORDIA study was not powered to detect risk differences of 5%–10% (34
). We did find a nearly significant difference for a switch from an SSRI to venlafaxine in a third treatment step, which showed a relatively better remission rate than vice versa.
In addition to clinical severity, we also identified other clinical variables that predicted a failure to remit, namely family conflict, drug and alcohol use, and anxiety disorder, consistent with our initial predictors of response and other studies (4
). These clinical variables, assessed after acute treatment, frame treatment strategies that could substantially increase the rate of remission.
There are several limitations to be considered when evaluating the results of this study. Many participants were in open treatment by week 24, making it difficult to evaluate the relationship between initial treatment assignment and outcome, especially because open treatment was not assigned randomly and there were small numbers of each type of open treatment. In addition, there was no placebo-comparison group, making it impossible to compare the outcomes obtained with study interventions with the natural history of the disorder. While up to 80% of depressed youth eventually remit, the expected rate is lower among those with more chronic depression (38
). Finally, approximately 20% of participants did not complete the week-24 assessment or had unknown medication status, and these individuals had a lower rate of initial response and most likely would have been less likely to remit as well had they been retained.
These findings suggest that current clinical guidelines, which recommend pursuing a given treatment strategy for at least 8–12 weeks, may need to be revisited. Instead, our data support more vigorous intervention earlier in the course of treatment for nonresponding patients. Clinicians may consider incorporating potentially promising strategies such as augmentation of antidepressants with psychotherapy or mood stabilizers, recognizing that clinical trials have only been conducted among adults for these treatments. In addition, when indicated, broadening treatment targets to include comorbid anxiety, alcohol and substance use, and family conflict may be important in achieving remission and preventing relapse. Further research is recommended to identify both promising intervention strategies and the optimal time for their implementation when a depressed adolescent is not responding to current treatment.