Bozarth, J.  (August, 2000). Paper presentation at the American Psychological Association, Washington, D. C.

This paper contends that mental health treatment in the United States is founded upon a fallacious premise; that is, the premise that there are specific treatments for specific dysfunction. I label this premise, "The Specificity Myth". This myth has been perpetuated from the medical model and from behavioral treatment models for mental dysfunction. It is found in extreme form in the attempts to identify Empirically Validated Treatments (EVT) (Recently termed Empirically Supported Treatments (EST)).

The argument is not new from the standpoint of fundamental views of the nature of human beings. O’Hara (1993) summarizes the two most prominent views of human nature as the deterministic view and the view of humans as beings in the process of actualization. She identifies the determinsitic view as suggesting that " . . . the only valid knowledge is scientific knowledge, hence, human life is predictable, explainable and controllable" (p. 9). The view of the actualizing human being suggests " . . . a process by which the natural inner being is set free from the stunting effect of civilization to realize itself and to actualize its highest potentials" (p. 8). The deterministic view currently dominates mental health treatment with the illusion of scientific verification. This view blends with the medical model for physical illness that has dominated mental health treatment propelling the assumption that appropriate diagnosis is related to viable treatment.

This paper reviews the credibility of diagnosis, the pattern of psychotherapy outcome research, the conclusion of five decades of psychotherapy outcome research, and several intrinsic flaws in the reported evidence for Empirically Supported Treatment.

Diagnosis

Psychiatric diagnosis and empirically validated treatment are predicated upon the same assumption. Both are based upon classification with the claim that classification is central to science. Hence, both are implicitly linked to science with the implication that the foundation and process of this conclusion does not need to be examined. That is, the logic is that the assumptions are integrally related to science and not open to critique.

Diagnosis involves two assumptions. First, it is assumed that there is a relationship among certain phenomena (discovered by researchers) from which the concept of a diagnostic label can be determined. Second, it is assumed that there is a binding of the clusters identified by researchers. The validity of both of these assumptions is sorely lacking in relation to psychiatric diagnoses. For example, it was clear from my (Bozarth, 1999) personal observations in the 1950’s and my personal studies in the 1960’s that a particular diagnosis or even psychological description was more related to the diagnostician or author of the descriptive reports than to the characteristics of the "patients". It became common knowledge among hospital personnel that the diagnosis of "schizophrenia" in the 1950’s was a catch all for those who did not fit other diagnostic categories. Boyle (1990) presents a compendium of arguments that schizophrenia is a "scientific delusion". Boyle (1999) states that " . . . there is no evidence whatsoever that the original introduction of the concept of schizophrenia was accompanied by the observation of a meaningful relationship amongst the many behaviors and experiences from which the concept was inferred" (p. 80). Statistical studies of groups diagnosed as schizophrenia show no evidence of the symptoms clustering together in a meaningful way (Bentall, 1990; Slade & Cooper, 1979). Similar lack of evidence of other diagnostic concepts has been found in studies of depression (Hallett, 1990; Wiener, 1989; panic disorder (Hallam, 1989); agoraphobia (Hallam, 1983); borderline personality disorder (Kutchins & Kirk, 1997); self-defeating or masochistic personality disorder (Caplan & Gans, 1991). Boyle (1999) presents an extensive discussion of these critical points; i.e., the previous "discovery" of patterns by researchers and the existence of underlying processes, which she contends has been seriously questioned in relation to psychiatric diagnoses. Boyle concludes that:

Pattern of Psychotherapy Outcome Research

Another remarkable twist of scientific method research in psychotherapy outcome studies is the shift towards specificity research. The drive for more rigor, more precision and more focus on specific operational variables has resulted in failure to build on the outcome findings of the last four decades. This is reflected in the reviews reported later. However, the study of patterns of psychotherapy efficacy research by Stubbs and Bozarth (1994) depicts a sobering picture.

In the article dubbed, "The Dodo Bird revisited: A qualitative study of psychotherapy research", five temporal categories characterized the evolution of psychotherapy outcome research. The title of the investigation picked up on Luborsky, Singer, and Luborsky’s (1975) review of comparative studies of psychotherapy where they concluded that there was equivalence in the effectiveness of all therapies. They used the Dodo Bird metaphor from "Alice in Wonderland" where there was a race to help the animals dry off after they had become wet with Alice’s tears. Since the animals ran in different directions, the race was just stopped. The Dodo bird was asked, "Who has won?" He finally exclaimed, "Everybody has won, and all must have prizes." Luborsky et. al. used this statement to convey the idea that all therapies should be considered equally effective. The "Dodo Bird" study reveals that common factors are likely to be the source of this equivalence. The categorical themes discovered by Stubbs and Bozarth were the following:

Category 1: Psychotherapy is no more effective than no psychotherapy (1950’s and 1960’s) (Eysenck, 1952; 1966).

Eysenck’s hypothesis that psychotherapy is no more effective than no psychotherapy stimulated considerable reaction and criticism (Bergin, 1971). Somewhat unheralded and unrealized, the research on Rogers’ hypothesis of the necessary and sufficient conditions became an important part of the responses to Eysenck. This is elaborated upon in Category 3. Other re-analyses of Eysenck’s data and other findings refuted this contention. Psychotherapy was generally found to be effective. Later studies using meta-analysis confirmed the general effectiveness of psychotherapy.

Category 2: The "core conditions" (empathic understanding, unconditional positive regard, and congruence) are necessary and sufficient for therapeutic personality change (1960’s and 1970’s).

The second category revealed that a large number of studies were related directly to Rogers’ hypothesis of the "conditions therapy theory" (Barrett-Lennard, 1998). Rogers’ hypothesis was consistently supported (Lambert, DeJulio, and Stein, 1978; Truax and Mitchell, 1971) and continued to be supported through the latter 1970’s and 1980’s (e. g., Orlinsky & Howard, 1986; Patterson, 1984) in the face of more equivocal reviews to be noted next. Truax and Mitchell (1971) presented fourteen studies (eight of which were individual therapy) consisting of 992 subjects. They identified 125 specific outcome measures favoring the hypothesis (66 of 158 were statistically significant). They report an analysis of the long-term effects of higher and lower levels of empathy, warmth, and genuineness experienced by the clients of the Wisconsin Project with hospitalized psychotics (Truax & Mitchell, 1971, p. 329). Their data over nine years indicates that patients seen by therapists low on the conditions tended not to get out of the hospital, and that clients of these same therapists did get out tended to return.

Lambert, Shapiro, and Bergin (1986) concluded in their review of the research that the attitudinal qualities: "seem to make up a significant portion of the effective ingredients of psychotherapy" (p. 202).

Orlinsky and Howard (1986) concluded their review of the research on the attitudinal conditions by stating that: "generally, 50 to 80 per cent of the substantial number of studies in this area were significantly positive, indicating that these dimensions were very consistently related to patient outcome" (p. 365).

A series of studies in Germany orchestrated by Reinhard Tausch and colleagues (1990) as well as other studies in Europe provide additional strong support for Rogers’ (1957) hypothesis of the necessary and sufficient conditions for therapeutic personality change (see Bozarth, Zimring, & Tausch, in press).

There were also studies that investigated the conditions as secondary variables that support this theme. For example, the effects of focused versus broad-spectrum behavioral therapy with problem drinkers in an effort to control their alcohol consumption was studied by Miller, Taylor and West (1980). They collected data on therapist empathy as a secondary inquiry and found that the level of therapist empathy was highly correlated (r = .82) with outcome.

Another example of the importance of relationship variables was the more recent study by the National Institute of Mental Health (NIMH) which was conducted to compare various treatments for depression (Blatt, Zuroff, Quinlan & Pilkonis, 1996). They compared the effects of the administration of a drug (imipramine), cognitive behavioral therapy, interpersonal therapy and "ward management" which served as a placebo. The placebo effect involved a therapist who spent time talking to patients about ward management. There were no significant differences between the effects of the three active treatments. The best prediction of success at the end of any of the active treatments was whether the patient perceived the therapist as empathic at the end of the second interview. Drug treatment was significantly more successful if the patient viewed the therapist as empathic after the second interview.

Category 3: Psychotherapy is for better or for worse (early 1960’s).

Therapists who were higher on the conditions were found to be related to positive outcome, while therapists lower on the conditions were related to client deterioration. As mentioned above, this was a strong argument against Eysenck’s assertions that no psychotherapy was as effective as psychotherapy.

Several reviewers pointed to the adverse effects of some therapists. Truax and Carkhuff (1967) concluded their research review with the statement that psychotherapy was "for better or for worse" (p 143). The review by Truax and Mitchell (1971) included a call for attrition in the ranks of "psychonoxious practitioners" while increasing the number of helpful counselors (p. 301). The deleterious effects of some therapists was highlighted.

Based upon a separate research review, Bergin (1971) concluded that the previous four decades of the practice of psychotherapy has had an effect that is modestly positive, adding: "However, the averaged group data on which this conclusion is based obscure the multiplicity of processes occurring in therapy, some of which are now known to be either unproductive or actually harmful" (p. 263).

Lambert, Shapiro and Bergin (1986) also found evidence to support the position that psychotherapy is for better or for worse; indicating that some therapists are detrimental as reflected in outcome data.

It is interesting that research on this rather dire finding, which suggests that therapists low on the attitudinal conditions were detrimental to their clients, virtually disappeared with the advent of the thrust for "specificity" studies in the 1980’s and 1990’s.

Category 4: The core conditions are necessary but NOT sufficient for therapeutic personality change (late 1970’s and early 1980’s).

Reviews during the middle 1970’s through the 1980’s included some that offered equivocal conclusions for Rogers’ hypothesis of the necessary and sufficient conditions. Change in the direction of research began in the middle 1970’s paralleling these equivocal reviews. The conclusions of the equivocal reviews that were supported with some critique of the designs were that (1) "more complex relationships exist among therapists, patients, and techniques" (Parloff, Waskow, & Wolf, 1978, p. 273); and that (2) the conditions have not been adequately investigated (Bozarth, 1983; Mitchell, Bozarth, & Krauft, 1977 Watson, 1984). Issues that need resolution were cited by Beutler, Crago, and Arismendi (1986) as the need to find "an acceptance of an optimal level of therapeutic skill, common methods of measurement, and the creation and control of levels of the facilitative skills" (p. 276).

Opinions predicated upon other theoretical formulations rather than upon design critique included the view that the core conditions were "nonspecific" and similar to placebo effect (Luborsky, Singer, & Luborsky, 1975; Shapiro, 1971); and that "the conditions are neither necessary nor sufficient although it seems clear that such conditions are facilitative" (Gelso & Carter, 1985, p. 220). For the most part, the data based equivocal reviews pointed to the need for more extensive examination of the complex phenomena of Rogers’ postulates and called for more rigorous methodological investigation.

There was virtually NO support for the category of the conditions being necessary but NOTsufficient. There was not one direct study that supported the assertion that the conditions are not sufficient. Nevertheless, the assertion of these reviews did affect (or perhaps served as a rationalization for) the direction of research. The research shifted from examining the attitudinal conditions to investigating "specificity". This shift was clearly NOT predicated upon previous research results.

Category 5: There are specific techniques that are uniquely effective in treating particular disorders (late 1989’s and 1990’s).

The search for the effectiveness of techniques and for specificity virtually extinguished the published studies on the Rogerian hypothesis of the necessary and sufficient conditions. On the face of it, studies in client-centered therapy and the conditions therapy theory were no longer viable inquiries in the United States.

After the middle 1980’s, the Rogerian (1997) hypothesis was investigated by only a dozen outcome studies which emphasized therapists’ empathy (Sexton & Whiston, 1994). These studies were all positive. They included a study of therapist variables that found that emotional adjustment, relationship attitudes and empathy were most predictive of effective therapists (Lafferty, Beutler, & Crago, 1989). Positive therapy outcome in several studies was linked to such constructs as "understanding and involvement" (Gaston & Marmar, 1994), "warmth and friendliness" (Gomes-Schwartz, 1978), and similar constructs (Bachelor, 1991; Gaston 1991; Windholtz, & Silbershatz, 1988). Empathy was strongly related to improvement for depressed clients who were being treated by cognitive-behavioral therapy (Burns & Nolen-Hoeksema, 1992). Despite the many positive findings it was the equivocal reviews of the research on the attitudinal conditions that proved to be part of the rationale for research directions toward "specificity" of treatment. The focus on "specificity" research replaced inquiry on Rogers’ hypotheses and on common factors in general.

Conclusions of psychotherapy outcome research

Stubbs and Bozarth (1994) concluded that: "Over four decades, the major thread in psychotherapy efficacy research is the presence of the therapist attitudes hypothesized by Rogers" (p. 120.). Concomitant to their conclusion of psychotherapy outcome research, Duncan and Moynihan (1994) independently analyzed psychotherapy outcome research. Their report titled, "Intentional utilization of the client’s frame of reference" reviewed outcome research to develop a treatment model. They conclude that the major operational variable that of intentionally utilizing the client's frame of reference. This article was associated with an explosion of psychological literature that identifies the common factors of relationship and client resources as the basis for most psychological improvement (Asay, T. P., & Lambert, M. J., 1999; Duncan, Hubble, & Miller, 1997; Hubble, Duncan, & Miller, 1999; Lambert, 1992; Miller, Duncan, & Hubble, 1997).

From 1987 to 1999, the investigations of specificity research have ironically returned full cycle to the pervasive influence of the common factors. That is, the reviews of outcome research by various reviewers including the more recent specificity research reveal that: (1) Effective psychotherapy is predicated upon the relationship of the therapist and client in combination with the inner and external resources of the client (common factors) (Hubble, Duncan, & Miller, 1999); (2) Type of therapy and technique add little to the effect of the relationship and client resources if not accompanied by common factors (Hubble et. al., 1999); and (3) Relationship variables that are most often related to effectiveness are the conditions of empathy, genuineness and unconditional positive regard (Bozarth, 1999; Patterson, 1984; Stubbs & Bozarth, 1994).

The clear message of five decades of outcome research is that it is the relationship of the client and therapist in combination with the resources of the client (extratherapeutic variables) that respectively account for 30% and 40% of the variance in successful psychotherapy. Techniques account for 15% of the success variance, comparable to 15% success rate related to placebo effect.

Intrinsic flaws of Empirically Validated Treatment

There are currently efforts in the United States and Europe to involve endorsement of specific psychotherapies by government, professional organizations, and other accrediting bodies. Those treatments to be approved are those which are "empirically validated". It is important to remember that the EVT syndrome (Now referred to as the Empirically Supported Treatment or EST by the Task Force of Division 12, Clinical Psychology of the American Psychological Association) is founded upon the belief that there are specific treatments for particular dysfunction (Task Force on Promotion and Dissemination of Psychological Procedures, 1995).

The advocates have already assumed the veracity of their claim. The task is to now to convince others; especially those who are in positions to influence policies. The postulate of specificity is accelerated through the use of manuals that delineate specific procedures.

Advocates of EST believe that this assumption is supported by "efficacious" empirical research. The arguments for EST are primarily seven points. These points are that: 1) much is already know about the effectiveness of specific treatments with specific dysfunction; 2) patient care will be improved; 3) the research will influence policy makers; 4) better training will be fostered; 5) therapy research will be encouraged; 5) it will more fair because of the professionals who have been consulted in developing the criteria; and 7) the project is intended to encourage guidelines and lists for effective treatments that can be useful to the field. It can be noted that, like the arguments for diagnosis, six of the assertions are based upon the assumption that their first argument is true. The assertion that "much is already known" is followed by the six assertions that have to do with influence, strategies and factors other than the validity of the EST stance. The only substantial argument is whether or not there are effective treatments for particular dysfunction. The bold assertion is affirmative (Barlow, 1996; Chambless, 1996). The Institute for the Study of Therapeutic Change (Web Page, talkingcure.com , 2000) succinctly responds to this assertion:

The words "efficacy" and "effectiveness" were interchangeable until recent years. The dictionary definitions are synonymous Recently, the term "efficacious" has come to identify the results of "gold standard" studies (Seligman, 1995). These are studies which have been traditionally identified as true design studies; that is, studies which are randomized, double-blind and have an adequate number of subjects and have adequate controls for therapists as well as having appropriate replications of the study. The rationale is that causation can be more accurately determined with this type of study. Efficacious studies are actually rare in the bulk of research in psychotherapy outcome. The following conclusions attend to major flaws of the assumptions and process of confirming EST’s:

Conclusion 1: There is considerable variation of the design criteria from the assumption that these are "Gold Standard" studies as implied by the advocates. The quality of the designs is no more rigorous than many of those representing the previous five decades of research.

It turns out that the "efficacious" and "gold standard" studies identified by the task force are not quite as efficacious as implied. The Task Force, in one publication, identified 36 studies of "Empirically Validated Treatments and another 32 studies of "Probably Efficacious Treatments" (Chambless et. al., 1996). The general criteria for acceptance as efficacious studies includes case study design experiments with N’s greater than 9. These experiments are part of the 36 recommended studies. The task force guidelines for the criteria of EST now defines their "Well Established Treatments" in less than rigorous terminology. (Task Force on promotion and dissemination of psychological procedures, 1995). Rather than referring to true design studies, they refer to the need for "At least two GOOD (authors’ emphasis) group design studies . . .". Such loose terminology is indicative of the deviation from their original intention to utilize "efficacious" studies as the criterion.

A thorough critique of the empirically validated treatment studies is presented in the journal of Psychotherapy Research (Bohart, O’Hara, & Leitner, 1998).

Conclusion 2: Five decades of research have been disregarded because those studies are not viewed as appropriately measuring the specific behaviors of the therapist or either because the do not fit the clusters of client dysfunction which have been reliably agreed upon, but not validly determined, by those who recommended the categories for the DSM-IV.

It is somewhat baffling how the task force conclusions could be reached after examining psychotherapy outcome research over the past five decades. It turns out that the five decades of research have been summarily disregarded for somewhat obscure reasons. The dismissal is, according to Garfield (1996) related to the idea that there are now instruments (i.e. training manuals) that identify more specific behaviors and standardize the therapy; and to the idea that there are reliable diagnoses (via the DSM-IV) to which treatment can be directed. As noted previously, it is interesting that neither the DSM-lll-R or DSM-IV actually recommend treatments for their "reliable" diagnoses. Treatments are now being determined through the EST phenomenon.

A specific example of dismissal is the renowned Smith et. al (1980) analysis of 475 studies which concludes that psychotherapy of all kinds is generally more effective than no treatment. The study is disregarded primarily on the basis that it pre-dates the Beck et. al. Manual and DSM-III (Garfield, 1996). The faux pas of dismissing the 1980 analysis has been raised anew by a meta-analysis in the November, 1997 issue of the Psychological Bulletin (Wampold, et. al.) which re-confirms the Smith et. al. study. Elliott (1997) also re-confirms these findings in his summary of meta-analysis.

Conclusion 3: The findings of five decades of psychotherapy outcome research have discovered that the client-driven/person-centered paradigm accounts for the major success variance for clients.

The most cogent conclusions of this research are:

Conclusion 4: The precise functional practice of specific treatments for a particular dysfunction is questionable.

How does the concept of EVT relate to efficacious treatment? We asked this question to a number of therapists. Here is one response:

The myth of EST is further compromised in the violation of the fundamental premise. The primary premise is that there are certain procedures that will ameliorate or diminish particular dysfunction. It is so certain that this is the case that the procedures are identified via treatment manuals. This is pretty good because what it means is that anyone who is reasonably intelligent can follow the procedures and the result will be positive. In the 1960’s,we did this with behavior modification procedures that were integrated into hospitals and schools for the mentally retarded and mentally ill. Ward attendants, many with less than high school education, could follow these procedures with reported successful results as long as the procedures were followed. Many of these procedures are still in the institutions in spite of serious questions about the validity of the results. But, if it works as asserted, let’s do it. However, somehow we find that it requires a doctoral level psychologist to apply the technical manual. Why is this? If the procedure is, in fact, the point of the whole thing, then why do we need the clinical psychologist as a treatment phenomenon? We don’t! If the premise of specificity (that there are particular treatments for particular dysfunction) is correct, the specific treatment is the thing. But now we have an intervening variable present. That is, we need a competent clinician in case there are clinical judgments to be made when the procedure is not working. This means that we must be ready to change the procedure at any given moment and that it is deemed by someone to not be working. The procedure, which is the thing, must be open to be tailor made to the particular client. Among other things, the common factors have entered into the realm of the EST’s.

The fact that five decades of psychotherapy outcome research has been ignored in the search for specificity as highlighted by the EST proposals is further compounded by the murkiness of "good" research designs which support procedures that can not be trusted without the murkiness of clinical judgment.

Summary

Our examination of the credibility of diagnosis, the pattern of psychotherapy outcome research, the conclusion of five decades of psychotherapy outcome research, and the intrinsic flaws in the reported evidence for Empirically Supported Treatment suggest a radical conclusion.

This conclusion is that the foundation of the mental health system in the United States is founded upon a myth; that is, the myth that there are specific treatments for particular dysfunction. This conclusion calls for a radical re-structuring of the mental health system to accentuate the variables related to success. These are the common factor variables of therapist/client relationship and emphasis of client resources and client frame of reference.

Comments by Sam Evans:

I do not agree with the notion that a diagnosis is given for the purpose of classifying and treating.  I tend to find a diagnosis somewhat helpful when communicating with other professionals regarding mutual clients.  I also find that in this day and age, many people need a diagnosis for no other reason than to faciliate payment by their insurance company.  If a psychotherapist submitted an invoice for payment by an insurance company, and stated that the reason for the visit was "problem actualizing" or "stunted personal growth", I doubt if the bean counters at the imsurance company would understand.  As our culture is used to looking at a medical model and is comfortable with this model, using a diagnosis to facilitate communication only makes it easier for the client to receive the help that is needed.  In an ideal world where clients did not need insurance, did not have problems with the legal system, judges, probation, parole, prison, and professionals did not need a concise way to sum up their behaviors, perhaps diagnosis would not be helpful, but for now, I really need it in my practice.  I really think that most psychotherapists, who are not academics or interns, really do not  rely on matching a diagnosis to a specific unproven treatment.  There are too many variations of psychotherapy based on a huge number of theoretical orientations and I find that a good therapist relies upon what has worked the best for him/her and their clients.  We are all different in our approaches to psychotherapy, and for each different approach we use different tools.  Personally I find empathy, genuineness, and unconditional positive regard the most helpful in a purely therapeutic relationship, however, there are psychotherapists who provide a host of services that cannot be strictly therapeutic in form.  For example, we are often limited in our practice by laws that mandate specific psycho-educational presentations that shall not be altered by the presenter.  Certain laws, such as Family Violence law in Georgia, go into the smallest detail of how the legislature has commanded service providers to present a model that was adopted by the state.  So, again, in an ideal world, I think Dr. Bozarth is on to something, however, we are practicing in a professional culture that is full of entangling alliances and struggling to survive.  We must not only "help" our clients, but we must understand and conduct business with corparate entities, politicians, government bureaucrats, and a host of well meaning and not so well meaning people that touch the lives of our clients.  A drastic overhaul of our mental health system is not practical or possible, however with quality continuing education for professionals, small gradual change for the better may be more likely.
 
 
 
 

References

American Psychiatric Association. (1994). DSM-IV. Washington D. C.:

American Psychiatric Association.

Asay, T. P. , & Lambert, M. J. (1999) . The empirical case for the common factors