It is clearly of central importance to psychoanalysis as a professional activity to sustain a vigorous program of empirical research. We need to learn more about (1) what changes actually take place in psychoanalytic therapy (the outcome question) and (2) how those changes come about, through the interactions of what factors in the patient, the therapist, the therapy, and the evolving life situation (the process question). I divide the history of psychoanalytic therapy research into four successive generations, each marked by increasing conceptual sophistication and methodological advance.
The first generation began as early as 1917, spread over half a century, and consisted of simple retrospective enumerations of percentage improved outcomes (by unspecified criteria) gathered by individuals or from treatment centers in the early psychoanalytic institutes.
The 1917 study was from the private practice of Isador Coriat in Boston and was followed by Otto Fenichel's 1930 report from the Berlin Institute, Ernest Jones's 1936 report from the London Institute, and Franz Alexander's 1937 report from the Chicago Institute. In 1941, Robert Knight at the Menninger Clinic combined all the previous reports and added 100 patients from Menninger, making an overall tabulation of 952 patients.
The overall therapeutic benefit rate was adjudged high for the neurotic and psychosomatic patients and poor for those diagnosed psychotic. Knight elaborated the pitfalls of these simple statistical summaries: the absence of consensually agreed-upon definitions and criteria, the crudeness of the nomenclature and diagnostic classification, and the failure to address issues of therapeutic skill in relation to cases of varying severity.
The most ambitious study of this genre was a report from the Central Fact-Gathering Committee of the American Psychoanalytic Association in 1967, presenting data on 10,000 initial questionnaire responses from 800 analysts, together with 3,000 completed termination questionnaires. This was followed a year later by Fred Feldman's report from the Southern California Institute. All these studies reported completely comparable improvement rates, but all of them were beset with the difficulties Knight had specified.
All this spurred second-generation research, the effort at more formal and systematic outcome study geared toward overcoming the glaring methodological simplicity that marked the first generation. Second-generation studies were of two kinds. One comprised prospectively organized, group-aggregated studies from psychoanalytic treatment centers, with specified definitions and operationalized criteria and predictions to expected outcomes. The other comprised in-depth, individually studied outcomes in a succession of patients.
The major examples of the first type come from three psychoanalytic clinics. In 1960 Peter Knapp et al., and in 1975 Jerome Sashin et al., reported on 100 and 183 patients, respectively, from the Boston Institute. Together, they found only fair prediction to analyzability, as assessed at the one-year mark, and that no effective prediction to outcomes could be made judging from the patient's characteristics at initial evaluation.
In 1985, John Weber et al. published the Columbia University Center for Psychoanalytic Training and Research studies, which reported almost 1,600 patients, 40 percent in psychoanalysis and 60 percent in psychoanalytic psychotherapy. They stated that all previous studies had been limited in at least one of the following ways: small sample size, inadequate information about outcomes, not based on terminated cases, or restricted to retrospective data. The Columbia studies had comparative data on large numbers of terminated analyses and psychotherapies conducted by the same therapists, and established criteria for therapeutic benefit distinct from criteria for an analytic process.
The most striking finding of the Columbia studies was that across every patient category, measured therapeutic benefit always substantially exceeded measured evolved analytic process. However, as with the Boston study, the outcomes, in terms of both therapeutic benefit and analyzability, were only marginally predictable from initial evaluation.
The third of these group-aggregated studies, by Joan Erle and David Goldberg (1979–84) from the New York Institute, only added to the Boston and Columbia observations that the results obtained with their sample of experienced analysts were no different than those from their own (and all earlier) studies of clinic patients treated by candidates.
Individually focused studies were being conducted along parallel lines during the same period. The first of these, in 1959–63, was conducted by Arnold Pfeffer from the New York Institute. Nine patients were intensively interviewed by a “follow-up analyst” after completion of their analyses. The chief finding, in all cases, was a rapid reactivation of the characteristic treatment transferences, even including symptom flare-ups as if in relation to the treating analysts; these subsided rapidly, at times aided by interpretations, and in a manner that indicated the new ways of conflict management that had been analytically achieved. This “Pfeffer phenomenon” was essentially replicated by two other studies, by Haskell Norman et al. and Jerome Oremland et al. (1975) in San Francisco and by Nathan Schlessinger and Fred Robbins in Chicago from 1974 to 1983.
Third-generation studies, contemporaneous with the second generation, combined both kinds of second-generation studies. These projects attempted both to assess analytic outcomes across a significant array of cases and to examine the processes through which these outcomes were achieved in the intensive longitudinal study of each case. Unlike the second-generation studies, they carefully separated outcomes at termination from functioning at a specified follow-up point, attempting to account for changes in this “post-analytic phase.”
The 1986–90 Boston Institute studies of Judy Kantrowitz et al. and the 1956–88 Menninger Foundation project of Robert Wallerstein et al. are third-generation exemplars.
Robert Wallerstein, M.D., is an emeritus training and supervising analyst with the San Francisco Psychoanalytic Institute and an emeritus professor and former chair of the Department of Psychiatry at the University of California-San Francisco School of Medicine.
The Boston studies followed twenty-two supervised analytic cases with interviews and projective tests. They also found that the majority of patients received therapeutic benefit in excess of what could be accounted for by the interpretive resolution of the analytic transferences. Again, successful outcomes could not be foretold from any of the predictor variables. The Boston studies also focused on the therapist-patient match in shaping the two-person analytic interaction. They found that with the majority of patients, the kind of match — facilitating vs. impeding — significantly affected the outcome.
The Psychotherapy Research Project of the Menninger Foundation was a thirty-year study of the treatment careers and subsequent life careers of forty-two patients, half in psychoanalysis and half in (supportive-expressive) psychoanalytic psychotherapies. The aim of the project was to study both outcome and process, specifying in detail the particular reach and limitation of therapeutic outcomes for each kind of patient treated within the indicated modality. The researchers were particularly interested in the empirical elaboration of change mechanism operative within both the uncovering (analytic-expressive) and the “ego-strengthening” (supportive) therapeutic modes.
Overall results of the project can be generalized as follows:
The treatment results in psychoanalysis and in varying mixes of expressive-supportive psychotherapies tend to converge, rather than diverge.
Across the whole spectrum of treatment courses, the treatments carried more supportive elements than originally projected. These elements accounted for substantially more of the changes achieved than had been originally anticipated.
From the study of the kinds of changes achieved by the patients, the changes themselves — separate from how they were brought about — often seemed indistinguishable in terms of being so-called structural changes in personality functioning.
The fourth and current generation of studies includes microanalytic detailed process studies, made possible by recent technological advances — audiotapes of entire treatments, combined with programmed computer searches. Programs are currently being fashioned to combine various ongoing process studies (each using its own conceptual framework and its own devised instruments) with the best established outcome measures, all (hopefully) on the same shared database of recorded and transcribed analytic hours. The goal is to realize the principle of “Patient-Treatment-Outcome Congruence” — representation, theoretically and operationally, of the patient's character functioning, the treatment process, and the achieved outcome in comparable terms.
At this point, such process and outcome therapy research, like that being attempted under the auspices of the APsaA by its Collaborative Analytic Multisite Program, can promise to have significant impact upon, and become a major component of, the evolving development of psychoanalysis as a science.