Research has generally found that both antidepressants and psychotherapy offer similar efficacy in the short-term, but that after treatment discontinuation, results are better with psychotherapy. For example, one meta-analysis found that psychotherapy patients were half as likely to suffer depressive relapse in the long-term, at least one year after acute treatment was discontinued (DeMaat S et al, Psychother Res 2006;16(5):562–572). And another meta-analysis found that patients receiving psychotherapy had lower depression scores than patients on antidepressants in the long-term, with this difference increasing with the length of follow-up (Imel ZE et al, J Affect Disord 2008;110(3):197–206).
However, nearly all prior meta-analyses have suffered from a major problem—the included trials mostly used tricyclics and MAOIs. This could give an “unfair” advantage to psychotherapy, because the heavy side effect burden of these older medications may cause early dropouts or may in other ways mask their antidepressant effects. Given that 90% of patients on antidepressants are taking second-generation antidepressants (SGAs)—SSRIs, SNRIs, or other newer medications such as bupropion (Wellbutrin)—it seems dicey to assume that findings comparing psychotherapy to older medications would yield similar results to trials comparing therapy to SGAs (Olfson M et al, Arch Gen Psychiatry 2009;66(8):848–856).
Only one meta-analysis specifically examined how psychotherapy fared versus SGAs, SSRIs in particular, and it found a very small but statistically significant advantage for SSRIs (Cuijpers P et al, J Clin Psychiatry 2008;69(11):1675–1685). However, two problems remain: 1) several relevant trials have been published subsequent to the Cuijpers et al meta-analysis, and 2) the quality of the treatments used in clinical trials requires close consideration. Surprisingly, treatment in psychotherapy studies is sometimes handicapped by researchers, limiting its effectiveness. Indeed, research has consistently found that non-bona fide therapies are less effective than legitimate treatments (Spielmans GI et al, Clin Psychol Rev 2007;27(5):642–654; Wampold BE et al, J Affect Disord 2002;68(2–3):159–165). What makes for a
1. Therapists must receive specialist training (eg, psychologists, psychiatrists, social workers, or supervised psychology graduate students)
2. Therapy is delivered face to face and individualized to the patient (as opposed to something like a prerecorded relaxation protocol)
3. The treatment relies on a theory of psychotherapy, as opposed to open discussion without any theoretically-guided intervention. Thus, any review of studies comparing SGAs to psychotherapy must account for whether psychotherapy was bona fide; this was not done in the Cuijpers et al meta-analysis.
With these issues in mind, my research team conducted a meta-analysis of trials that compared SGAs to psychotherapy. We located 15 high quality trials, including six not included in the Cuijpers review, with a total of 19 comparisons between psychotherapy and SGAs. Trials ranged from six to 26 weeks in duration. Psychotherapies in these studies included cognitive and/or behavioral (these made up the majority), psychodynamic, interpersonal, and supportive. Drugs included fluoxetine Prozac), paroxetine (Paxil), venlafaxine (Effexor), citalopram (Celexa), nefazodone (Serzone), sertraline (Zoloft), and fluvoxamine (Luvox) (Spielmans GI et al, J Nerv Ment Dis 2011;199(3):142–149).
Overall, at the end of the acute phase of trials, our results were nearly the same as Cuijpers et al (2008): there was a statistically significant but very small advantage for SGAs—an effect size of 0.19. You might recall that for effect size, conventional wisdom holds that an effect size greater than 0.20 is small, greater than 0.50 is medium, and greater than 0.80 is large.
However, the advantage for SGAs was driven by their notable superiority over non-bona fide therapies: an effect size of 0.58. Patients receiving non-bona fide psychotherapy were also significantly more likely to discontinue treatment. When considering only bona fide therapies, SGAs and therapy showed no significant difference.
The non-bona fide therapies (five comparisons) were labeled as such because they were provided by non-mental health specialists (nurses, general practice physicians, or therapists with unclear training). The researchers apparently used such therapists out of convenience or because they were attempting to train non-specialists to provide psychotherapy.
The longer term follow-up results painted a somewhat different picture. Patients receiving bona fide therapy had a small but statistically significant advantage on depression rating scales over patients who took SGAs: an effect size of 0.26.
There was no indication that any particular variety of psychotherapy fared much better or worse compared to SGAs, but there were very few trials of non-CBT therapies, so further research might find some sort of difference. However, a comprehensive meta-analysis of psychotherapy clinical trials found that there was no clinically significant difference between various forms of psychotherapy for depression (Cuijpers P et al, J Consult Clin Psychol 2008;76(6):909–922).
There are some limitations to our meta-analysis. In eight of the 15 studies, patients met with study staff for at least half of the weeks of the study, which is clearly not a realistic portrayal of clinical practice. Also, since there were no placebo arms in the studies, efficacy versus placebo was not established, but since SGAs and psychotherapy are both well-established treatments, this is probably not a big problem.
SGA dosages were typically flexible whereas psychotherapy dosage was usually fixed at once per week, though some studies provided less than 10 sessions of therapy, which does not represent clinical therapy practice and could have handicapped psychotherapy in the research setting. The relationship between depression severity and outcome was often not clearly described in the trials, so there is little guidance from the literature regarding whether SGAs or psychotherapy are preferable for severe depression. Nevertheless, it appears that when we limited our analysis to only evidence-based therapies, therapy and antidepressants came out about neck and neck in the short term, with therapy probably nudging ahead of medication in the long term.
Regardless of whether you provide psychotherapy or an SGA (or both), some other ideas can help improve outcomes. A meta-analysis of 26 studies across a variety of mental health issues found that patient preferences appear to impact outcome. Clinical trial patients who receive their preferred treatment had a 58% chance of having a better outcome than patients who received their nonpreferred intervention. Further, patients receiving their preferred treatment were about half as likely to drop out of treatment (Swift JK et al, J Clin Psychol 2009;65(4):368–381). We should carefully solicit and consider patients’ ideas regarding which treatment they would prefer.
Systematically tracking outcomes is also useful. Patients who show the quickest treatment response often have the best outcomes, while patients who show little improvement within the first few weeks typically continue to struggle; this is applicable to both antidepressants and psychotherapy (see Machado-Vieira R et al, J Clin Psychiatry 2008;69(6):946–958; Szegedi A et al, J Clin Psychiatry 2009;70(3):344–353; Tadic A et al, J Affect Disord 2010;120(1–3):86–93). Thus, regularly tracking rating scale scores can help identify patients whose progress is lagging and might benefit from discussing a change in treatments.
Research has yet to identify which treatments might be most appropriate if treatment switching occurs, so we’re left to rely on clinical judgment and collaborative decisionmaking with our patients. For more on using rating scales in practice, please see the October 2008 issue of TCPR.