Most of the research in this area has been conducted with depression, and a systematic review by Pampallona and colleagues 9 concluded that adding psychotherapy provides significant benefits to patient outcomes. If a patient is to be referred for psychotherapy, the first questions are to whom, and for what kind of therapy? Given the principles of evidence-based medicine, it seems most obvious to look to outcome studies of particular types of psychotherapy to particular patient diagnoses.
There are many thousands of these studies to choose from. The rather uncomfortable but consistent finding, however, is that the particular type of psychotherapy makes little difference to patient outcome. It likely does not matter if the patient is referred to that rumpled psychoanalytic therapist down the street or to that bright cognitive-behavioral therapist over at the university. A recent study compared the benefits of 7 major psychotherapies for depression and concluded that none were more or less effective than the others.
When we compare effect sizes of different psychotherapies on particular patient diagnoses, we find that there are no significant differences among psychotherapy approaches. This applies whether we are looking at different treatments for adults with particular diagnoses, for youths, for couples and families, or for those with alcohol and drug abuse.
The research, however, does not support this. Psychotherapy is an effective treatment for many mental disorders, and it is a valuable adjunct treatment for most others. While most research has demonstrated robust benefits for the neuroses-type disorders involving anxiety or depression, recent research has demonstrated significant benefit for patients with psychotic disorders and even to those with some neurological disorders such as Alzheimer disease.
This article provides both a strong endorsement of referring patients for psychotherapy and a dilemma. It indicates that the research discriminating between the benefits of different types or schools of psychotherapy is weak.
It also indicates that the particular therapist is a potent factor for outcome and that ongoing measurement of therapeutic progress and therapeutic alliance offers substantial benefit. The dilemma is that at this time there is no systematic way to find out which therapists are most effective or to measure progress and alliance. The research strongly supports referring patients for psychotherapy. The dilemma of finding effective psychotherapists may be best solved by developing referring relationships with a small number of therapists who report measuring progress and alliance and by obtaining feedback from your patients to verify their effectiveness.
In their meta-analysis, Ahn and Wampold 17 examined the effects of removing components of psychotherapeutic treatments and found no evidence that removing or adding a specific ingredient altered outcomes. These conclusions are not, predictably, without their detractors.
Schools of psychotherapy can be fiercely tribal in defending the superiority of their approaches. In a recent article, the research data were interpreted as indicating that psychodynamic psychotherapy is as effective as or more effective than other forms of therapy and was met with inevitable critiques from advocates of other approaches.
A greater challenge to the equivalence conclusion comes from studies that have randomly assigned patients with a given diagnosis to receive different forms of psychotherapy. While there are many such studies, and some do find apparent superiority of one treatment over another, how these apparent differences should be interpreted is a source of dispute.
Treatments that are intended to be therapeutic, that the therapist believes in, and for which there is a compelling rationale work. There are characteristics of psychotherapy that lead to poor or damaging outcomes; others predict good outcomes.
The data indicate that some therapists are consistently better than other therapists, that therapeutic relationship factors account for much of the variability in outcome attributable to psychotherapy, and that a major way that better therapists achieve their better outcomes is through enhancing the therapeutic relationship.
Consistent with this, patients do not emphasize particular psychotherapies or methods when accounting for their improvement but instead emphasize the relationship with their therapists. Each psychotherapy includes different active ingredients that promote patient improvement. Given the complexity of human beings and their brains, it is foolhardy to suggest that there is only one way to help someone with emotional distress. What we find, however, is that these ingredients do not work very well unless the person delivering them is genuinely caring and empathic, and able to form a solid therapeutic alliance with the patient.
Similar results are found when we examine the effectiveness of antidepressant medication. The most effective psychiatrists helped their patients more using placebo than did the less effective psychiatrists who treated their patients with an antidepressant.
While the field of psychotherapy absorbs the evidence, which suggests that different approaches are equivalent, the biggest shift in psychotherapy practice in recent years has been toward the systematic measurement of patient progress and therapeutic alliance. A pervasive human foible that physicians and psychotherapists do not escape is the belief that we are all better than average. Psychotherapists tend to believe that we have good rapport and alliances with our patients and that our effectiveness with our patients is better than that of most other therapists.
For the most part, we draw these conclusions without much evidence. It turns out that collecting these data in a consistent way actually is a potent way of increasing therapeutic effectiveness. When patients are not experiencing a good alliance with the therapist, these tools allow the therapist to find out immediately and to take steps to improve it or refer the patient to another therapist.
Therefore, ongoing monitoring of patient progress and adjustment of treatment as needed are essential to [evidence-based practice in psychology] EBPP. Psychiatrist David Burns, 32 who popularized cognitive-behavioral therapy through his best-selling books eg, Feeling Good , serves as a good example for the transformation that is taking place in psychotherapy.
Almost 30 years ago, the American Psychiatric Association convened a Commission on Psychotherapies to review and integrate the research data that were available at that time. To analyze such complex interplay of multiple factors we need the type of advanced psychotherapy process research that we will describe later.
Recently, researchers leaning on complexity theory have used computational models to, for example, formalize ideographic theories of functional analysis for panic disorder and to test a perceptual control theory account of psychological change 21 , Many psychotherapy researchers have experienced that the RCT design is difficult to use for patients with more complex psychopathology As an illustrative example, let us consider patients with concurrent borderline personality disorder BPD and substance use disorder SUD , which was the target group for a pragmatic RCT conducted by one the authors.
Patients with such severe psychopathology are often traumatized, and lacking trust and hope that treatment providers can help them. It often takes a long time to build a secure relationship with such individuals and to strengthen their motivation to dare begin psychotherapy and work with their problems.
At that moment, randomization is detrimental to the patients who end up in the control condition and do not receive active treatment. Our study of mentalization-based treatment MBT vs. In our view, the just-mentioned small sample sizes are an indication that psychotherapy with fragile patients should be studied with research designs other than a typical RCT. Our example was dual diagnosis, but the same argument would be valid for disorders such as severe personality disorders e.
A variant of RCT that might be appropriate is the one comparing two bona fide psychotherapy methods so that none of the patients received inactive therapy. But such a comparative RCT would be extremely expensive if it were to meet design needs for sufficient statistical power and adequate treatment duration probably more than 1 year.
Also, when there is no existing treatment with established empirical support for a particular pathology, the first step of gathering evidence for a novel treatment would be to test it against a control condition such as a placebo treatment. A more productive strategy might be to declare another gold standard research design for patients with complex psychopathology, for example cohort studies with repeated measurement of outcome and process variables and comparison to a benchmark Further complications are the large heterogeneity within diagnostic categories and growing evidence that a general psychopathology factor might contribute to all mental disorders In our study of MBT vs.
That was good news for the patients, but it was bad news for our research study and for the RCT design! Since RCT designs leave a knowledge gap about the change mechanisms in psychotherapy, this gap has to be filled by research focusing directly on this question.
Partly this can be done by using the experimental component designs with dismantling or constructive strategies Haynes and O'Brien indicate the following four requirements for causal inference:. The variables must covary. The hypothesized causal variable must precede the outcome variable. Realistic alternative explanations for the observed covariance must be reasonably excluded. There must be a plausible explanation for the hypothesized causal relation A well-conducted RCT meets all of these requirements, which is the reason it is considered the gold standard for causal inference.
We would argue that psychotherapy process research is increasingly able to meet most of these requirements, and methodological developments to meet them all are well under way. Since such research can be done on naturalistic data as well as on RCT data, it is less subject to the limitations of RCTs that we have outlined above.
Process research also has the advantage of giving much more specific information to the therapist about what to do during the therapy sessions compared to a difference between two group means regarding two treatments or comparing one treatment against a control condition. A common misunderstanding is that process research is only meaningful after having established efficacy for a treatment package using RCT.
If this was true, the usefulness of process research would be much more limited than we propose. Temporality 2 has traditionally been problematic in process designs in which a process was simply related to pre-post outcome, but with session-by-session measurements of process and outcome time-lagged associations can be analyzed, ensuring that this requirement is met The most difficult requirement for causal inference is ruling out alternative explanations 3 , or the risk for third-variable confounding.
Most researchers are aware that potential confounders can be measured and included as covariates in multivariate statistical models, but what to do about confounders that are not measured, perhaps even not known? This is where the power of randomization comes in, but there are statistical ways of achieving this when randomization is not feasible.
Modern cross-lagged panel models enable the separation of average differences in processes and outcomes between patients from over-time fluctuations within patients, which ensures that confounders that are stable over time—even ones that are unobserved—can be ruled out This means that these designs are approaching the RCT in terms of the potential for causal inference, while not being marked by the disadvantages of these.
Although these designs protect against confounders that are stable over time, there are a number of potential confounders that vary over time, e. Recent developments in statistical methods in other scientific fields most prominently in economics, sociology, and biology may, under certain assumptions, enable even such confounding to be ruled out. For instance, instrumental variable regression has been used for causal inference in econometrics for a long time, probably since the 's This method requires fairly strong assumptions, and it is difficult in practice to find instruments that satisfy these.
However, recent developments enable researchers to relax those assumptions, and research is under way testing the performance of these methods In the study, we used a microanalytic sequential process design, showing that within a session, therapist interventions directed at exploring mental processes were connected with a subsequent higher patient level of mentalization. In other words, therapist interventions guiding the patient to explore mental processes lead to increased patient mentalizing, supporting this theoretically proposed change mechanism in MBT In this study, covariation [criterion 1 ] is estimated using regression models of patient mentalizing on therapist interventions, correct temporality [criterion 2 ] is ensured by using therapist interventions that immediately precede patient mentalizing statements, alternative explanations [criterion 3 ] are partly ruled out by the design and statistical model, e.
Finally, the study hypotheses are based on mentalization theory [criterion 4 ]. Such rigorous process research can be based on psychotherapy sessions from RCTs or naturalistic studies. An advantage of using data from RCTs is that they have more well-defined patient samples and treatment methods; hence, it is easier to know which generalizations of the findings are adequate.
However, as mentioned, less strict designs often have the advantage of increased ecological validity. As we have shown, the RCT is the strongest research methodology for testing treatment efficacy 9 , However, the more complex the treatments and the patients' psychopathologies are, the more difficult to use the RCT design gets.
As psychotherapy methods almost invariably are complex treatment packages, the RCT design also misses the critical question of interest, namely: what specific treatment principles and interventions are helpful for the patient, or in other words what are the crucial change mechanisms?
Based on these arguments we suggest some modifications of the EBM model concerning psychotherapy for mental disorders. We propose that the following three designs should have an equal place at the top of the hierarchy for research evidence concerning psychotherapy:. For short-term psychotherapy for patients with circumscribed, less severe psychiatric disorders, RCT is still the research design of choice. For long-term psychotherapy for patients with severe or complex psychopathology, cohort study with repeated measurements and comparison to a benchmark is the research design of choice.
In order to investigate the crucial therapeutic principles and change mechanisms in psychotherapy for particular mental disorders, process research using stringent strategies to establish causal connections and appropriate statistical analytic methods is the research design of choice. BP wrote most of the introduction, the part about RCTs, and the discussion. FF wrote most of the part about process research. We have sent the developing manuscript back and forth to each other and made revisions.
The submitted version is approved by both. Both authors have contributed to the manuscript. No external funding for this article. Open access publication fees are paid by Stockholm University Library. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Djulbegovic B, Guyatt GH. Progress in evidence-based medicine: a quarter century on. Pace et al. In addition to these advances in data coding and analysis, another important development in psychotherapy research is the development of additional ways to collect data from clients in the first place.
Given the proliferation of social scientists collecting data online, this development has the strong potential to contribute to our understanding of psychotherapy.
Further, ethical issues such as fair pay and ensuring confidentiality should be considered. Psychotherapy researchers have the monumental task of working to improve the effectiveness of the interventions that we offer to clients who seek out treatment. Several important questions about treatment efficacy, causes of change, external application of research, and the dynamic nature of psychotherapy remain.
Fortunately, psychotherapy researchers now have a number of important technological, methodological, and statistical advances that can help them in answering these unresolved questions. We have come a long way through our scientific studies as a field and it will be exciting to see where the next age of psychotherapy research will bring us. Tompkins, K. The future of psychotherapy research: Where are we going and how can we get there?
Psychotherapy Bulletin, 54 1 , Altenstein, D. Interpersonal microprocesses predict cognitive-emotional processing and the therapeutic alliance in psychotherapy for depression.
Journal of Counseling Psychology, 60, Baldwin, S. Open access meta-analysis for psychotherapy research. Journal of Counseling Psychology, 63 , Therapist effects. Lambert Ed. Hoboken, NJ: Wiley. Beutler, L. Bridging scientist and practitioner perspectives in clinical psychology. American Psychologist, 50 , Campbell, D. Experimental and quasi-experimental designs for research and teaching. Castonguay, L. Practice-oriented research.
Cohen, L. Use of psychotherapy research by professional psychologists. Cooper, Z. Cognitive and Behavioral Practice, 18 , Psychotherapy, 51, Glass, G. Primary, secondary, and meta-analysis of research. Educational Researcher, 5 , Hansen, N. The psychotherapy dose-response effect and its implication for treatment delivery services. Clinical Psychology: Science and Practice, 9 , Hirschberg J.
Advances in natural language processing. Science, , Horton, J. The online laboratory. Experimental Economics, 14 , Inthout, J. Obtaining evidence by a single well-powered trial or several modestly powered trials. Statistical Methods in Medical Research, 25 , King, D. Kosinski, M. Facebook as a research tool for the social sciences: Opportunities, challenges, ethical considerations, and practical guidelines.
American Psychologist, 70 ,
0コメント