Recently, a middle-aged woman came into my office presenting with a relatively common mix of mild anxiety and more severe depression. She recounted a history of chronic depression, several courses of psychotherapy of varying lengths and multiple trials of medication with little or no success.
After the usual review and the beginnings of assessment, including ongoing self-report measures of severity of depression, I began treatment. Since the woman was athletic and had enjoyed a number of sports, such as golf, I immediately recommended that she increase her aerobic exercise before initiating my usual cognitive behavioral strategies. She came in the next week and had barely taken her seat before proclaiming, "My God, I'm feeling so much better … I feel like I'm getting my life back … I wish I had started seeing you years ago!" Several additional weeks showed further improvement on her measures of depression.
Well, I felt very good about this, as I suspect all clinicians have at one time or another after a similar experience. But was her improvement really my doing, and was it important for me to attempt to find out?
All of us in clinical practice make hypotheses regarding reasons for change or the lack of it. Our hypotheses or educated guesses in the context of our clinical experience then may lead us to attribute change or lack of it to specific factors. I did have a good relationship with the patient (we shared a passion for the New England Patriots) and there is pretty good evidence that aerobic exercise can lift depression, but not necessarily a depression so deep or so chronic, and not so quickly.
But would all of us, patients and therapists alike, benefit if we could actually identify the reasons why treatment worked or didn't work? In the case described above, I suggested to the patient that she stop the aerobic exercise for a bit so that together we could attempt to determine its impact on her depression. The rationale we worked out was that if the intensive aerobic exercise was having such a dramatic impact on her depression, we would want to tailor it in a way that would be sustainable. After a hiatus of several weeks, we reintroduced the exercise.
As the patient and I examined her data, it became clear that removing the exercise and reintroducing it had no effect on the rate of her continuing improvement. The patient and I then began to analyze if any other major changes had occurred that might have contributed to her improvement. We discovered that her relationship with her husband had suffered considerably during the years she was depressed but that, with renewed hope, he joined her for exercise sessions at the health club and then they would go out for dinner. They both enjoyed this activity so much that they substituted another joint activity when exercise temporarily stopped. This allowed us to focus on their relationship in subsequent sessions as well as to focus on additional cognitive and behavioral strategies since she was at risk for relapse, as are all patients suffering major depressive episodes.
This approach to ascertaining causes of change, often referred to as an idiographic approach (Barlow, Nock, and Hersen, 2009; Kazdin, 2010), involves systematically altering some aspects of the treatment strategy while observing effects on the repeated measures, or sometimes simply observing the effects of naturalistic changes in the environment, particularly if they are occurring repeatedly.
There are a number of idiographic strategies for determining if a treatment or its components are having an impact, or if initiating a new therapeutic strategy leads to increased rates of improvement over past strategies, or even for comparing several strategies early in treatment to decide which one to go with.
But was it ethical to remove a treatment component that was correlated with improvement? Were the benefits of discovering what seemed to be truly important in lifting this patient's depression worth the risk that some deterioration could have occurred?
The broadly held view among both practitioners and scientists is that if these strategies are implemented solely for the clinical benefit of the patient and are undertaken in a collaborative fashion, then it is simply part of good practice. This would be contrasted with manipulating treatment strategies to increase knowledge about the effects of treatment more generally, where the patient's well-being might not be the primary goal. In the latter case, informed consent on the goals of the project would be required, as would other steps involved in ethical clinical research.
David H. Barlow, PhD, is professor of psychology and psychiatry at Boston University, and founder and director emeritus of the Center for Anxiety and Related Disorders.
