This is a blog piece by Damion Grasso, Ph.D. (Division 12 Web Editor) that explores how we communicate information about treatment effectiveness to our patients. It draws from recent articles that discuss methods for translating results from psychotherapy research into probabilistic information that aims to inform treatment consumers.
David is a middle-aged man who just one year ago was assaulted and badly beaten by a group of young adults in a subway station. Prior to the assault, David had experienced few if any emotional and behavioral problems that warranted any kind of professional attention. Now, one year after the assault, things are different for David. The quality of his work at a legal firm has diminished. His relationship with his wife is strained. His ability to care for and enjoy the company of his two daughters is compromised. He struggles with frequent nightmares, sleepless nights, and intrusive thoughts about the attack. In sharp contrast to his frequent social outings with colleagues and friends before the attack, David is now socially withdrawn, declining requests to engage in social activity. David is always on “high alert” and experiences intense bodily reactions to places, people, or situations that remind him of the attack. David avoids subways and has become avoidant of public places more generally – so much so that it prevents him from doing things he used to enjoy. David has posttraumatic stress disorder (PTSD). He has heard of it before, but knows very little about it. What he has found on the Internet is varied and mainly geared to combat veterans. He never thought this was something that could happen to him. David’s wife convinces him to seek professional help for this. He sees his primary care physician who refers him to a mental health professional in his insurance network.
Fortunately the clinical social worker David was referred to has specific training in an evidence-based model for treating PTSD – Prolonged Exposure (PE). David had never heard of PE – or any kind of evidence-based model for that matter. Has this treatment worked for others? Will this treatment work for me? Which of my problems are likely to get better with this treatment? What are the chances this treatment might not work for me? These are all questions that run through David’s head. Unfortunately, although his clinician can educate David on what PE is and how it is thought to work, she is entirely not equipped to answer these questions.
David’s clinician is not incompetent. Most clinicians would not be able to answer these questions. In fact, most in the field would have a hard time answering these questions. However, research can provide some guidance aimed at better informing the consumer about the treatment he or she is about to invest in. Clinical researchers communicate to their academic colleagues information about the effectiveness or efficacy of treatment all the time. They do so using effect sizes, like Cohen’s d, as well as things like the Absolute Risk Reduction (ARR) and the Number Needed to Treat (NNT). These metrics provide clinical utility; however, each describes average treatment outcomes relative to another treatment or a control condition. It would mean very little to David if his clinician responded to his questions by telling him that five people with PTSD need to be treated in order for one person to show statistically significant symptom reduction on a particular standardized measure. Rather, we clinicians need a better method for communicating this information in a way that is pertinent and interpretable to patients.
We can use efficacy and effectiveness research trials to inform how we talk about these treatments with our patients. Meta-analyses of treatment studies, which concatenate and summarize these data for us, can be particularly helpful. These data sources provide an opportunity to utilize this information for communicating the probability, in absolute terms, that a particular patient will respond to a particular treatment with a favorable outcome. There are several metrics we can use to communicate this information. These are reviewed in two papers that introduce and describe the Probability of Treatment Benefit (PTB) method (Beidas, Cross, & Dorsey, 2014; Lindhiem, Kolko, & Cheng, 2011). A rather simple metric that reflects treatment outcome is the probability of falling below clinical threshold at post-treatment on a standardized measure. For examining treatment response, we can use the Reliable Change Index (RCI; Jacobson & Truax, 1991). The RCI estimates the reliability of individual change scores in a sample of participants and allows us to talk about the probability of reliable improvement and the probability of reliable deterioration of symptoms while taking into consideration random error. With this we can say things like, “There is an X% chance that your symptoms will improve over six months should you receive treatment… or there is an X% chance that your symptoms will worsen over six months should you not receive treatment.” In addition, by examining median pre to post change scores of a research sample we can also calculate the expected change in severity of symptoms (either improvement or worsening) on a particular standardized measure to say things like, “Your score on this measure of depressive symptoms was an 80 out of 100. With treatment we might expect your score to drop from 80 to 30, which is in the low to moderate range.”
When there is sufficient research available we might even go a step further to provide probabilistic information that takes into consideration patient characteristics. For example, perhaps there is a body of research showing that women are more likely to respond to a particular treatment than men. We might use this information to personalize the information we provide to our patients. This is also incorporated in the PTB method (Beidas, et al., 2014; Lindhiem, et al., 2011) and described in a recent article (Grasso, Ford, & Lindhiem, 2014) that provides a step-by-step overview on how to apply the PTB method, as well as an illustration of the method using data from a study examining the effectiveness of a PTSD treatment.
By translating research findings into person-focused probabilistic statements, we can start to develop a language for talking to patients about how effective a particular treatment is given their symptom presentation and individual characteristics. Rather than assuring patients that what we intend to do is evidence-based, which means very little to patients, we can offer more specific, probabilistic statements aimed at providing information to consumers about treatment benefit… “David, given the several research studies that have looked at this treatment’s ability to reduce symptoms in people with PTSD, compared to no treatment, there is a 75% chance that the severity of your symptoms will be reduced by about half. [Note. This is an example and does not reflect actual data.] Or… David, given the several research studies that have looked at this treatment’s ability to reduce symptoms in people with PTSD, compared to no treatment – should you not receive any treatment, there is a 50% chance the severity of your symptoms will actually increase over the course of 6 months.”
Discussion Questions:
- Do you find your patients asking questions about various treatment models and their effectiveness?
- How do you respond to questions about treatment effectiveness in your practice?
- Do you see any dangers or drawbacks in providing this kind of information to patients?
References:
Beidas, R. S., Cross, W., & Dorsey, S. (2014). Show Me, Don’t Tell Me: Behavioral Rehearsal as a Training and Analogue Fidelity Tool. Cognitive and Behavioral Practice, 21(1), 1-11.
Grasso, D. J., Ford, J. D., & Lindhiem, O. (2014). A Patient-Centered Decision-Support Tool Informed by History of Interpersonal Violence: “Will This Treatment Work for Me?”. J Interpers Violence.
Jacobson, N. S., & Truax, P. (1991). Clinical significance: A statistical approach to defining meaningful change in psychotherapy research. Journal of Consulting and Clinical Psychology, 59(1), 12-19.
Lindhiem, O., Kolko, D. J., & Cheng, Y. (2011). Predicting Psychotherapy Benefit: A Probabilistic and Individualized Approach. Behavior Therapy.