|
Those who use research-based treatment practices or recognize the most basic of research methods would voice familiarity with the concept—randomized control trial (RCT). It is the strength behind the dominant, most reliable treatment research methods used in the last three decades called the technology model of psychotherapy research. Most folks know it by the term evidence-based treatment. When you see it you know you’re getting quality.
Basically, RCT is a scientific procedure in which subjects are randomly assigned within an experiment. In good addiction treatment research, subjects are indiscriminately assigned to a group that receives a certain treatment (treatment group), or a group that does not receive the treatment (non-treatment group). When you get enough subjects for both groups, you can then compare the differences between the groups in terms of outcome. The strength of the RCT is that subjects have an equal chance of being assigned to either group. What does that do? It improves the chances that your independent variable (e.g., a treatment administered to one group, and not the other) will be considered the difference or the cause of any differences between the groups (groups in this case are the dependent variables). In research parlance, RCT improves the reliability of your data and decreases error; and who doesn’t want that?
On the whole, the technology model of psychotherapy research and RCT is behind all the empirically-based treatments being touted these days. Without it, we have to resort to educated guesses. Not good. However, cracks in this gold standard have begun to appear.
Problems on the horizon
It always happens. Just when you think you have a good thing in hand (the technology model) something comes along to challenge your theory. Or to put it another way, ugly facts will dispute beautiful theories. Case in point: for decades researchers assumed that by following the technology model, we would eventually discover some consistent, reliable and detailed addiction counseling techniques for us to use in our therapy sessions, but that has not happened.
The central problem seems to be that the technology model may well be showing its limitations. According to Morgenstern and McKay (2007), we cannot extract better information from it than we already have. That’s why we only get general information about addiction treatment outcomes. We cannot seem to find specific mechanisms within various therapies that constantly succeed across a spectrum of clients. Essentially, we cannot identify what it is about treatment specifics that make them work (see examples below). In addition, the technology model is giving us conflicting results. Many research studies often give varying results from the same treatment. I often see this when I review the research literature for this column. And ta’ boot, sometimes the treatment works with one client and is a total bust on another. So, what’s up?
To answer that question we need to ask a few questions:
• What are treatment specifics?
• Or in research terms, what are efficacy, moderator and mediator effects?
Treatment specifics—efficacy and moderator effects
Efficacy, or the efficacy research method, compares a specific treatment (e.g., Motivational Interviewing) against a control group under tight research conditions to determine if the treatment is indeed causing a desired outcome (Wigner & Solberg, 2001). Generally, all the treatments examined in the Morgenstern and McKay (2007) study indicated some level of efficacy. The treatments reviewed included Motivational Interviewing (MI), Cognitive Behavioral Therapy (CBT), Behavioral couples treatment (convincing evidence for this) and 12-Step treatment (TST).
Moderators are effects such as sex, age, race or the strength of a reward that can affect the direction and/or strength of a relation between a independent variable on a dependent one. Yes, believe it or not, all such moderator effects influence treatment outcome. (Begin to see the problem here?) (Baron & Kenny, 1986). Morgenstern and McKay (2007) reported mixed outcomes across all therapies reviewed.
Treatment specifics—mediator effects
Mediators describe how, not when, certain internal variables beside the independent variable impact the dependent variable (Baron & Kenny, 1986). For example, in MI there are supposedly vital mediators that make change happen. One particular mediator is the client’s commitment language, or the client using more change language. Using such language, as suggested by the MI model, should bring about more readiness for change. However, eight studies cited by Morgenstern and McKay (2007) found little evidence for such an effect on readiness to change. Other studies they reviewed indicated essentially mixed results using MI with patients who had low readiness to change.
But what about other therapies? Well, consider that the main tools of CBT are the acquisition of coping tools or skills (e.g., correcting inaccurate self thoughts, dealing with cravings). Those things are suppose to mediate outcome, right? Morgenstern and McKay’s analysis found no support for the idea that CBT works via skill acquisition. In many of the studies reviewed, the authors found skill increase was unrelated to outcome.
Okay, what about behavioral couples treatment? While no full test of mediator effects exists for this approach, it looks like behavioral contracting had the best efficacy evidence.
Finally, what about TST? The same investigators looked at three mediators specific to 12-Step programs—spirituality, endorsement of abstinence and commitment to Alcoholics Anonymous practices. Only the commitment to AA practices mediator made a difference, but not the belief in powerlessness.
Having found all that, the authors are quick to point out that they are not willing to say there are no specific effects to be had from these therapies. The problem may not be with the mediators, but in the technology model itself.
It’s the model, Silly
It is beginning to look like technology applied research is giving us mixed results. Why? According to our authors, the technology model has problems. Two big knotty issues have become evident. One is that clients are diverse and do not uniformly respond to general treatment approaches. Such a criticism has been leveled at various treatments for years, but the point is that our clients are different. They vary in the way they think, feel, and certainly in terms of their culture and individual history. The one knotty problem is how then can one therapy account or treat such variably? Moreover, treatment is dynamic or changeable. Anyone who has conducted therapy will attest to the fact that treatment not only changes from session to session, but will sometimes change significantly from one minute of a session to another. The current state of the technology model has a difficult time accounting for these changes and the many client differences you and I see all the time. Even with the best RCT design, all these factors mentioned cannot be tracked well and we end up with mixed results.*
Knotty problem number two is the relationship between specific and non-specific factors. It is assumed that all one needs is a bare minimum level of non-specific factors or what we call “engagement” (rapport) between a client and counselor for the specific therapy factors (e.g., mediator factors mentioned above) to kick in. In the first place, this is a simplistic assumption because attaining a relationship with a client is a complex, challenging and dynamic process. The relationship itself fluctuates over time and even within a session. So, what chance does a treatment specific technique have in such a variable environment?
In summary, the problem, to some degree, seems to be in the way we research our treatments. That recognition calls for change.
Next time
Given the importance of this subject, this column needs to be completed in two sections. For now, know that there are problems with our research design gold standards (Technology Model of Psychotherapy Research). However, hope is also on the horizon. We address those promising possibilities next time in Part II.
*We are not going to get into the new statistical models floating around out here like Structured Equation Modeling.
Michael J. Taleff, PhD, CSAC, MAC, has written numerous articles, several books, teaches at the college level, and conducts trainings and workshops (e.g., the latest treatment practices, Critical Thinking and Advanced Ethics). He can be contacted at
This e-mail address is being protected from spambots. You need JavaScript enabled to view it
, or
This e-mail address is being protected from spambots. You need JavaScript enabled to view it
.
References
Baron, R.M. & Kenny, D.A. (1986). The moderator-mediator variable distinction in social psycholoical research: Conceptual, strategic, and statistical considersations. Journal of Personality and Social Psychology, 51, 6, 1173–1182.
Morgenstern, J. & McKay, J.R. (2007). Rethinking the paradigms that inform behavioral treatment research for substance use disorders. Addiction, 102, 1377–1389.
Wigner, D.E., & Solberg, K.B. (2001). Tracking mental health outcomes: A therapist’s guide to measuring client progress, analyzing data, and improving your practice. New York: John Wiley & Sons
This article is published in Counselor, The Magazine for Addiction Professionals, December 2009, v.10, n.6, pp.16-17.
|
