The practice of one human being helping another to feel better through the use of talk is as old as humanity itself. As a profession, however, psychiatry has existed for only 100 years. During that time, both the theory and the technique have evolved. Changes have been driven as often by politics and culture as by advances in knowledge. From the early days of psychoanalysis, many hoped that psychotherapy would be based in an empirical science, yet relatively little of its development has been driven by systematically acquired data. The field is maturing, however, and the coming decade holds promise for an ever-increasing place for research as a driving force.
What works for whom?
In order for psychotherapy to advance, researchers need to address the question of what works best for which patients in order to achieve which outcome (borrowing from the title of Roth and Fonagy's 1996 book1). A much-quoted finding of the psychotherapy literature is that all forms of psychotherapy have the same efficacy when compared head-to-head.2,3 This is often called the dodo bird effect, a term popularized by Luborsky and colleagues2 in reference to Alice in Wonderland, where the Mad Hatter declares, "Everybody has won and all must have prizes." Growing evidence, however, suggests that this result was, at least in part, driven by insufficient sample size (leading to low statistical power) and a failure to employ appropriate outcome measures.
When these issues are addressed, the results are more interesting. For example, in a recent randomized comparator trial of dialectical behavior therapy (DBT), transference-focused psychodynamic therapy (TFP), and supportive psychotherapy for borderline personality disorder, only TFP substantially changed the ability of patients to reflect on emotional states (a factor thought to predict ongoing symptom improvement4). On the other hand, TFP was less successful than schema-focused therapy in retaining patients and preventing dropout.5 Future trials are increasingly likely to match specific treatments to well-defined patient populations, in pursuit of certain goals.
The empirically validated treatment (EVT) model in psychotherapy, inherited from evidence-based medicine, promises to be a useful guiding force in future research.6 Practice guidelines, educational standards in psychiatry residency and psychotherapy training programs, and even insurance company reimbursement are increasingly tied to lists of treatments that claim to have been demonstrated to work. Several researchers have challenged the apparent bias of such a standard toward short-term, symptom-focused treatments (which are easier to study and thus have dominated the early EVT lists).7,8
However, as researchers learn to do more complicated trials with dynamic and other detailed and long-term treatments, these are likely to find their place on the list of EVTs. TFP, mentalization-based psychotherapy (MBT) for borderline personality disorder, and Milrod's dynamic psychotherapy for panic disorder are making strides in this direction,9-11 with others sure to follow. In line with the rest of clinical medicine, the randomized controlled trial (RCT) has emerged as the gold standard for testing the efficacy and effectiveness of psychotherapeutic treatment. The psychotherapy RCTs bring with them complications somewhat different from those in other medical fields. For example, many therapeutic schools emphasize the importance of the "patient-therapist match," which is altered by the introduction of randomization.
Many psychotherapies are intended to vary significantly from one patient to another and thus are more difficult to "manualize" than administering a medication or following a straightforward medical procedure. However, one need look only to surgical research to realize that the problem of studying individualized treatment is not a new one and does not eliminate RCTs as useful.12RCTs remain the best, and often only, way to study a difference in outcome that is attributable to treatment and not to a myriad of confounding factors such as patient self-selection, therapist selection, differences in overall level of pathology, and other biases that we may not even know how to measure. Only by randomizing a large group of patients do we reasonably assure ourselves that the groups will not differ in any significant way other than by the difference in therapy that we have set out to study.13
RCTs in psychotherapy have not always been of good quality. In the past decade, cognitive-behavioral therapy (CBT) investigators, followed with mounting enthusiasm by dynamic psychotherapy researchers,14,15 have paid careful attention to methodological difficulties. In the years ahead, improvements in RCTs and ultimately in our ability to match patients, treatments, and desired outcomes will be led by advances in each of several methodological and theoretical areas: patient description, treatment manuals, and outcome measurement.
Patient description
The DSM has subdivided psychopathology into reproducible categories; however, identifying patients who benefit most from different sorts of treatments will require that these categories be further refined and sometimes even rearranged to accord with our growing knowledge. For example, the value of psychotherapy in treating patients with depression is related to features of developmental history that are not reflected in the correct diagnostic criteria.16 Future versions of the DSM are likely to incorporate scales of psychopathology, including but not limited to severity, in a way that will facilitate their use in predicting optimal treatment.17 Temperamental and dispositional factors play a role in the usefulness of various forms of psychotherapy, yet these patterns are only beginning to be studied. For example, Blatt18 and his colleagues have shown that depressed patients with an anaclitic (depen-dent) style benefit more from psychotherapies with supportive and structured elements. In contrast, depressed patients with an introjective style, characterized by difficulties with self-definition, autonomy, and self-worth, do better with unstructured, explor-atory treatment.
