The results of this pilot randomized effectiveness trial indicate that short-term focused dynamic psychotherapy has great promise as an intervention for depression in the community mental health setting. Patients treated by community clinicians trained in SE psychotherapy had greater improvement in symptoms of depression than patients treated in the community TAU condition. Although the effect in this small pilot study only reached statistical significance on the self-report measure of depression, the large effect sizes across multiple outcome measures indicate that dynamic psychotherapy might be an effective intervention for depression. These effects are consistent with the review by Driessen et al. (2010)
, which reported an average Cohen's d
of 0.69 for the comparison of dynamic treatment with control conditions. In comparison, a recently published fully-powered comparison of cognitive therapy to TAU for depression specifically in the community mental health system reported a Cohen's d
of .59 in favor of the cognitive therapy (Simons et al., 2010
) compared with effect sizes ranging from .83 to 2.02 for the slopes of change reported in the current pilot investigation.
An examination of the means and standard deviations for the symptom assessments across time points () indicates that there were baseline differences in depression in the treatment groups with the TAU group beginning treatment with lower depressive symptoms. The hierarchical linear models revealed an advantage in symptom reduction across treatment for the SE condition, but an examination of the means for the measures of depression indicate that the treatments wound up at similar points by the end of treatment but that SE psychotherapy had a greater slope of change. We assume that the SE condition would show a similar advantage in symptom reduction across the full range of pretreatment depression levels, especially given that the measure of general psychiatric distress (the BASIS-24 total score) also showed an advantage for SE over the TAU condition. However, because the sample size was small, a fully powered trial would be necessary to reliably estimate the symptom slopes and control for baseline levels.
Our assessment of clinically meaningful change also indicated an advantage for the dynamic psychotherapy compared with the TAU condition, with 50% of SE patients and 21% of TAU patients demonstrating reliable change on the HAMD across treatment, and 44% of SE patients and 36% of TAU patients demonstrating movement into a normative range of functioning on the HAMD. On the BASIS-24, 50% of SE patients and only 29% of TAU patients demonstrated movement into the normative range of functioning across 12 weeks of study treatment. Although these response rates may seem low for both treatment groups, these results should be understood in the context of this effectiveness trial. The patients treated at community mental health centers and included in the current study are poor and indigent with extreme stress and instability in their lives. Treatment attendance is often sporadic with many life stressors often interfering in the progress of psychotherapy. Response rates for psychotherapy reported in the best done efficacy trials still reach only 40% to 60% (DeRubeis et al., 2005
; Bielski, Ventura, & Chang, 2004
; Keller et al., 2000
) and in public sector clients, response rates for pharmacological treatments for MDD are estimated to be <30% (Rush et al., 2004
). The response rates in our pilot investigation are in line with what would be expected in this community mental health sample.
The SE treatment did not result in greater treatment retention than the TAU with both treatments averaging only 6 to 7 sessions of treatment received. Again, this attendance seems poor in comparison with the percentage of completers seen in efficacy trials. However, the number of sessions attended in this study was in line with the average number of sessions attended by patients receiving outpatient services at these community mental health centers. Our goal was to include an explicit socialization to treatment in the first session to improve treatment retention. Future research will need to evaluate the factors that contribute to treatment attrition in this population and improve techniques for motivating consumers to commit to services.
The brief treatment attendance on average also could have a significant impact on the estimate of the effects of the treatment. It is possible that either treatment intervention could demonstrate larger effects and better response rates in settings where treatment attendance was maximized. The effects reported in the current study generalize to what might be expected in settings such as this community mental health setting where attrition rates are high. Even in this setting, a larger trial would be necessary to fully examine the relation between dose of treatment and response.
Analyses of treatment adherence also demonstrate that the SE psychotherapy could be significantly discriminated from TAU on adherence to the expressive relationship–focused techniques. These results suggest that the training was effective in helping community therapists use interventions to help patients unpack their maladaptive relationship patterns. Future studies should confirm that this difference in adherence to interpretive techniques between the treatments was due to the specific training in SE therapy. There was no significant difference between treatment groups on the use of supportive techniques. This result is consistent with our experience working with therapists employed in the community mental health system. These therapists often focus treatment on a collaborative supportive working relationship but are not trained in specific techniques to address maladaptive interpersonal patterns.
Finally, exploratory analyses indicated that adherence to expressive techniques across the treatment groups was significantly related to symptom slope as measured by the BASIS-24. These results suggest the possibility that use of expressive interventions results in symptom alleviation across psychotherapy (Also see Barber, Crits-Christoph, & Luborsky, 1996
; Gaston, Thompson, Gallagher, Cournoyer, & Gagnon, 1998
; Hilsenroth, Ackerman, Blagys, Baity, & Mooney, 2003
). An examination of the individual items that predicted a decrease in depressive symptoms indicates that within dynamic psychotherapy, it is extremely important to help patients unpack their own maladaptive behaviors toward other people in their worlds and understand the repetitive nature of their patterns across relationships. The item analyses further suggest that therapists should help patients focus on describing specific relationship experiences, exploring their stereotypic ways of perceiving other peoples' responses toward them, and exploring with patients the historical origins of their relationship patterns.
Of course this pilot investigation was only able to explore covariation of adherence to expressive techniques assessed at Session 3 with symptom slope assessed from baseline to month-3 assessment. It is possible that early symptom change, which is often highly associated with final outcome, led to better adherence. Thus, this and other “third variables” might explain the obtained correlation between adherence and outcome. However, previous investigations (Barber, Connolly, Crits-Christoph, Gladis, & Siqueland, 2000
) suggest that process-outcome correlations are not simply due to early symptom improvement. Future research using fully powered samples will be needed to unpack the temporal course of the relation between intervention adherence and symptom course across psychotherapy.
Although this pilot study did not allow for an independent structured interview to obtain formal diagnoses for study entry, all patients included in the study first met the self-report QIDS cutoff that has been shown to represent a score of 14 or above on the HAMD 17-item score and received at least a 14 on the independent HAMD evaluation. In addition, the majority of patients (74%) received a diagnosis of MDD from the intake clinician at the community mental health center. We included an additional 10% of patients who received a diagnosis of depressive disorder not otherwise specified because this diagnosis is often given in the community mental health setting in cases that report significant depressive symptomatology but are unclear about the length of time that symptoms have been experienced. In addition, our sample consisted of an additional 16% of patients who were not given a depression diagnosis by the community intake clinician but rather were given primary diagnoses including posttraumatic stress disorder, adjustment disorder, and dysthymia. Because these were not standardized diagnostic assessments, the reliability of the clinical diagnoses cannot be calculated. Since the majority of patients were diagnosed at intake with MDD and the baseline HAMD scores averaged 21, these results appear likely to generalize to the treatment of moderate-to-severe depression, and most likely to MDD, as diagnosed in the community mental health setting.
This pilot investigation has multiple limitations. First, effect sizes calculated using small samples may be relatively less reliable estimates of the true effect sizes. Although there was a statistically significant treatment by time effect for symptom change on the BASIS-24 depression score and treatment effect sizes were large across multiple outcome measures, it is possible that the small sample size led to an unreliable estimate of the effect. Another limitation was that the first author of the manuscript was also the clinical supervisor for the SE psychotherapy condition. Ideally, the expert clinical supervisors would be independent of the investigative team. Otherwise, it is possible that researcher allegiance could influence the expectations and confidence of the therapy providers which alone may influence treatment outcome over and above the effects of specific interventions. Finally, the HAMD ratings were conducted by trained bachelor's level research assistants who were not necessarily blind to time in treatment or treatment condition. Although the research assistants were not aware of the specific research hypotheses, it is possible that their ratings were influenced by the allegiance of the research team. It should be noted that the treatment effect sizes evident on the HAMD ratings were consistent with the effect sizes of the self-report depression measure.
Another limitation is that therapists were recruited separately for the two interventions. Therapists interested in further training were recruited for the SE training phase while therapists for the TAU were not recruited until we were ready to randomize patients. It is possible that other systematic differences between the therapist groups accounted for the effects demonstrated here. However, the fact that the treatments could be discriminated by blind adherence judges and the fact that adherence to expressive techniques predicted symptom course suggest that the effects demonstrated in this project were a result of the dynamic interventions. Further research with large samples of patients and therapists would be necessary to confirm the treatment effects.
We also provided monetary reimbursement to both patients and therapists. It is possible that payment to patients influenced their attendance at the psychotherapy sessions, although attendance for this study was similar to what is typical of outpatients at this setting. It is also possible that payment to therapists influenced study results. Although we attempted to balance payments to therapists across treatment groups, therapists in the SE group were paid honorariums for study patients treated and supervision sessions attended across both the training and randomization phases, whereas TAU therapists received payments across the randomization phase only. It is possible that differences in total compensation resulted in differences in allegiance that could influence study effects.
Finally, our adherence ratings were limited by a focus on supportive and expressive techniques only. As we adapted this treatment for use in the community mental health system, we included additional components to help this treatment succeed in this setting. This small pilot investigation was not able to parse out the effectiveness of these additional treatment components but rather evaluated only the effects of the package as a whole. In addition, our adherence ratings based on audio recordings may have not captured the subtle effects of interventions.
Despite the pilot study status of the current investigation, the results indicate that short-term dynamic psychotherapies should be further evaluated as treatments for MDD. These results indicate that a modified version of SE psychotherapy may be an especially important intervention in the treatment of MDD in the community mental health setting. Our survey of therapists working in the community indicated that relationship-focused techniques were consistent with therapists' theories of depression causality and treatment effectiveness, and these adherence results further support that therapists can learn to effectively implement the interventions focused on helping patients learn about their maladaptive relationship patterns.