Extensive research attests to the value of psychotherapy in an array of psychiatric disorders. However, when it comes to understanding how psychotherapy works, we must reluctantly acknowledge that we do not know as much as we’d like to know. Psychotherapists who ask their former patients to identify what was most helpful to them often find the answer disconcerting. What the therapist thinks was a magnificent interpretation based on a keen understanding of the patient’s psychopathology may be completely forgotten. The patient’s fondest memory may be the time that the therapist told a joke. If one asks a psychotherapist to identify the factors that produced improvement in a former patient, one will probably hear a response that reflects the personal theoretical biases of the therapist (and assures the maintenance of the therapist’s self-esteem). Hence, we are largely in the dark when we attempt to pinpoint the therapeutic action of psychotherapy. We have many theories but little data.
In this regard, the study reported by Høglend and his colleagues in this issue of the Journal arrives at an auspicious moment. The investigators provide meaningful data on one of the longstanding controversies in dynamic therapy, namely, the role that transference interpretation plays in the therapeutic action of psychotherapy. One point of view has been that a focus on the conflicts and themes that arise in the therapeutic relationship will illuminate the nature of problems in the patient’s relationships outside of therapy (1). An alternative perspective, especially in brief psychotherapy, is that too much attention to the transference may make patients inordinately anxious. In light of this concern, an alternative approach is to examine extratransference relationships and interpret patterns, conflict, and fantasies as they emerge in those contexts.
The study reported by Høglend et al. is a randomized controlled trial of dynamic psychotherapy designed to determine the impact of a moderate level of transference interpretations (1–3 per session) in a once-weekly psychotherapy for a duration of 1 year. One hundred patients were randomly assigned to either a group using interpretation of the transference or a group that did not use such interventions. The authors include brief vignettes from the therapies so the reader can gain some understanding of the types of interventions considered to be transference interpretations. They wisely avoid the “allegiance effect,” so common in psychotherapy research, by cross-training therapists in each of the therapies used and arranging for the same therapists to conduct both treatments. The results were somewhat surprising; while there were no overall differences in outcome between the two treatment cells, the subgroup of patients with impaired object relations benefited more from the therapy using transference interpretation than from the alternative treatment.
The conventional wisdom in predicting psychotherapy outcome has been that “the rich get richer.” In other words, those patients with greater psychological resources and more mutually gratifying relationships are able to form a solid therapeutic alliance with the therapist and gain greater benefit from the therapy, although reviews of the literature on the relationship of transference interpretation to outcome in time-limited therapy (2, 3) do not confirm a positive correlation between the two. We generally associate lower scores on quality of object relations with personality disorders, and indeed the SCID-II (4) diagnosed 60% of the subjects within the low quality of object relations subsample in both groups as having one or more axis II conditions. By contrast, only 20% of those measured as having high-quality object relations had personality disorders.
If we turn to randomized controlled trials of axis II conditions, particularly those with compromised object relations, we find perplexing data again. A dynamic psychotherapy that specifically eschews transference interpretation, mentalization-based therapy, has been shown to be efficacious in the treatment of borderline personality disorder (5, 6). On the other hand, transference-focused therapy also appears to be successful with such patients (7). How do we understand these findings?
Transference interpretation, of course, does not occur in a vacuum. Timing is of great importance. Our process research using audiotapes of dynamic psychotherapy with borderline personality disorder patients (8) demonstrated that transference interpretations tended to have greater impact on the therapeutic alliance—both positive and negative—than other interventions. In other words, it is a high-risk, high-gain intervention. A climate of empathy, support, and validation may be necessary for the patient to accept a transference interpretation. A surgeon needs anesthesia to operate. A psychotherapist may need to create a holding environment before interpreting.
Another explanation of the contradictions in the literature is that different patients respond to different elements of the therapeutic action. Blatt (9), for example, has found that one subtype of patients, those he terms introjective, are more responsive to interpretation than a second subtype, the anaclitic, a group that derives greater therapeutic gain from the therapeutic relationship itself. In the naturalistic setting in which most psychotherapy is practiced, the astute clinician continually monitors the impact of interventions and modifies the approach accordingly.
In the study by Høglend et al., the study design has shortcomings that must be taken into account. Axis I disorders were not rigorously diagnosed using standard research interviews. The effects of depression, for example, on the outcome cannot be evaluated with precision. The experienced therapists used in the study may have been biased in favor of transference work and secretly felt that the patients deprived of it were getting less than optimal treatment. Dynamic therapists often overvalue transference interventions, and this bias could influence the therapist’s investment in the two different treatments used in the study. Similarly, while the investigators attempted to “blind” the raters who were listening to audiotapes, the content of these tapes might well indicate to which group the patient belonged.
The Høglend et al. report raises provocative questions and will require replication with some of the above concerns in mind. However, the findings indirectly support the idea that attention to the therapeutic alliance may be crucial. Research consistently demonstrates that the strength of the therapeutic alliance may be a critical ingredient in outcome (10). Patients with poor object relations may have difficulty seeing the therapist as a trusting, helpful figure. By addressing the therapeutic relationship directly and interpreting the distortions that appear in the patient’s transferences, the therapeutic alliance may be improved so that the patient can continue to collaborate with the therapist in a constructive way.
Finally, the notion that “the rich get richer” may be part of therapeutic mythology. Moreover, therapeutic action is not a monolithic entity. There is probably an array of active ingredients in the therapeutic action of psychotherapy, and it requires a sensitive practitioner to tailor the approach to the patient (11). After all, we should adjust the treatment to the patient, not the patient to the treatment.