This month a study of the treatment of young people diagnosed with schizophrenia attracted a good deal of attention. The headlines on my various internet feeds read “Talk therapy helps schizophrenia” and hailed a new treatment that includes listening to young people’s concerns and helping them find their way back to school or work. I often wonder what others make of my profession. What happened to render the notion that talking to people about their experiences and helping them find jobs or go back to school is something novel?
I was a psychiatrist who participated in the Recovery After an Initial Schizophrenia Episode Early Treatment Program (RAISE ETP). Although I welcomed the headlines, the reports left me with mixed feelings. While I celebrate the attention given to this study, I did not find the news reports to be entirely accurate. Several issues were raised, so to speak, that I want to address in this post:
- What is RAISE? How was it funded? What was the experimental design? What was its stated goal?
- What are the assumptions and hypotheses that drove this study?
- What results have and have not yet been released in published reports?
- What does RAISE teach us about science, spin, and politics?
- What are the implications for a post-RAISE world?
What is RAISE?
There are two RAISE studies. Their stated goal was to evaluate what is called coordinated specialty care (CSC), i.e., clinicians with varying expertise who work together as a team. The population of interest was young people who were experiencing their first episodes of psychosis. RAISE Implementation and Evaluation Study (IES) was intended to study the implementation process in several clinical settings. I will not discuss it here but it is important since it influenced the development of On Track New York, a large initiative in that state. Under the direction of Lisa Dixon, M.D., this program is also having significant influence on program development around the country. The one in which I participated is the subject of recent news reports — RAISE ETP. The Principle Investigator is John Kane, M.D. This study was designed to compare the efficacy of CSC as compared to treatment as usual (TAU).
RAISE funding was part of the American Recovery and Reinvestment Act. The late US Senator Arlen Spector allegedly directed money to NIMH for this purpose but I have not been able to confirm that. As you may recall, that was a time when our economy was floundering and the federal government infused it with money to help keep it afloat. Regardless of whether one thinks that funding psychiatric research is a way to save the economy, my sense is that this was a relative windfall to a neglected area of research – clinical trials that include non-medical interventions. An enormous amount of work went into getting this large operation underway quickly. Leading researchers in the fields of family education, individual psychotherapy of psychosis, and psychiatric rehabilitation were brought together to design a protocol. The study was modeled on similar programs that had been developed in other countries and the intent was to integrate treatment approaches that had the best evidence bases. Some of the interventions were designed specifically for RAISE ETP. For instance, the manual for the individual psychotherapy component, Individual Resiliency Training (IRT), while based on existing cognitively based approaches, was written exclusively for this study and clinicians at the participating sites all needed to be trained before the study began. The manual is available on line. A paper has been published that describes the rationale for the particular components of this study .
The experimental treatment was NAVIGATE. It included a number of features: A weekly team meeting including a program coordinator, research assistant, psychiatrist/nurse practitioner, and other clinicians who offered the required treatments. The specific treatments offered were IRT, family psychoeducation, and supported employment. The psychiatrist worked with a program called COMPASS. The study subjects completed an online questionnaire prior to each visit. The answers were available to the psychiatrist when they met. It included questions about voices, mood, unusual thoughts, and a host of possible side effects. Weight and blood pressure were monitored regularly. It also asked about the person’s goals for the meeting. This was intended to foster shared decision-making. Within COMPASS, the psychiatrist had access to suggestions about recommended treatment but there was no fixed treatment algorhythm s/he was forced to follow. For instance, if the person was gaining weight, suggestions would appear for evidenced-based recommendations for next steps. Each individual was followed for two years. An extended study is underway to follow these individual for up to an additional three years but the current reports are on the initial two-year study.
The research team identified 34 sites around the country where investigators were willing to participate. The range of sites provided both geographic and socioeconomic diversity. They were not programs that were affiliated with academic research centers. This was intended to mimic “real life” but it posed an added challenge for the research team, since participants at these sites were not used to following research protocols. Since this was a study intended to compare NAVIGATE to TAU, the researchers needed to assign subjects to one of each treatment. Rather than do this for each person who consented to participate, they randomized the 34 sites, i.e., each site offered only NAVIGATE or TAU to each person who agreed to participate. This had the advantage of only needing to train clinicians at 17 of the sites. I was fortunate to be a participant at a NAVIGATE site.
What were the Assumptions and Hypotheses?
During that same time, I was also studying Open Dialogue (OD). When I finished the RAISE study, I led my clinic in the direction of Open Dialogue. There were several reasons why.
First of all, I no longer am sure I accept the fundamental assumptions inherent in RAISE. Although the model includes a number of important non-medical approaches, the conceptualization of schizophrenia as a medical condition is at the core of the program. In a previous blog, I contrast a medical model approach with a need-adapted approach. In the former, the expert conducts an evaluation and then offers treatments based on the outcome of the evaluation. In need-adapted approaches, the understanding of the problem is derived from the process of meeting together and listening to all perspectives. RAISE adapted a medical model and that was evident in all aspects of the study. For instance, the psychotherapy, Individual Resiliency Training, included modules that address illness management and the emergence of symptoms. This is a decidedly medical formulation: the clinician says to a young person, “You have an illness and I can help you to manage it and cope with its symptoms.” Similarly, the family component is family psychoeducation about the illness.
Secondly, engagement seemed to be our biggest obstacle in working with this group of individuals. Getting people to appointments, even after they went through the rigors of signing an informed consent form, was not easy. There was a lot that we did – drawing mostly on our experience and common sense – that was not captured by RAISE. Open Dialogue, in contrast, is all about engagement. OD has given me a framework that is helpful in many settings but especially when a family is concerned and invites us in but the person at the center of concern does not agree with the way the problem has been defined. These individuals would not come to IRT because they do not think of themselves as having an illness to be managed. This is not to say I have entirely abandoned medical conceptualizations of psychosis. I just am not sure that conceptualization is helpful for everyone.
In RAISE, antipsychotic drugs are considered essential and the recommendation is that they are started early and continued indefinitely, albeit at the lower doses than are typically used. I have written elsewhere why I think this is problematic.
What Are the Results?
There have been several papers published thus far, and undoubtedly there are more to come. The early ones describe the study and offer a description of those who participated. One notable finding is that most study subjects were already on antipsychotic drugs at entry and many were already overweight with elevated lipids (cholesterol and triglycerides). Many were taking antipsychotic drugs – and in some instances more than one – at doses that were above the guidelines established by the RAISE team (based on their review of the available data). This demonstrates that common prescribing practices do not follow expert recommendations.
The paper that caught so much attention was the first to publish outcome results. While the study was touted as showing greater improvement among those in the study group, those effects were decidedly modest. They reported that NAVIGATE participants were more likely to engage in and remain connected to services. As for clinical outcomes, their major measure was the Quality of Life Scale. It is worth noting that in both settings there was overall improvement. The NAVIGATE group fared better on the overall score as well as on the subscales of interpersonal relations and intrapsychic foundations. The NAVIGATE group also showed overall greater improvement on the Positive and Negative Symptom Scale, a standard symptoms rating scale. A major finding was that the NAVIGATE treatment had its impact primarily in the group of subjects who had been symptomatic for less than 74 weeks. The nominal differences between the groups was small. This study was large. That allowed the researchers to detect differences that while statistically significant may not be as clinically meaningful.
There was a high dropout rate and its impact on the analysis remains unclear to me. The NAVIGATE group started out with 223 individuals, of whom 104 (47%) had full or partial 24-month assessments with no gaps. Forty-three (19%) had partial 24-month assessments with gaps, and 76 (34%) left the study early. In the TAU group, 181 entered the study, 59 (33%) completed the 24-month assessment with no gaps, 25 (14%) completed it with gaps, and 95 (52%) left the study early. This is a reflection of the challenges inherent to this kind of study but it has important implications for how we understand the results. It also bolsters my impression that engagement was a challenge in both settings.
There has been some discussion about the use of medications since in some of the initial press reports John Kane was quoted as saying that the subjects in NAVIGATE were on 20-50% lower doses than in the TAU group but these results were not reported in the initial papers. I suspect these data are available and will be included in future reports.
Science, Spin, and Politics
The RAISE study has been celebrated since its inception. I was surprised to find NIMH touting the presumed efficacy of NAVIGATE even before the study was fully analyzed. For an organization that prides itself on scientific rigor, that struck me as strange. Spin dominates so much public discourse that we seem not even to notice it anymore. Those who fail to spin their results or programs risk losing out on funding and I suspect this was the main driver of the hype in this instance. It is no secret that NIMH has decidedly deemphasized research that is not aimed primarily at elucidating brain function. The researchers on this study have had to fight their way to get funding, especially those among them who study psychosocial treatment approaches. This was a huge windfall and opportunity for them. It was a lot of money and perhaps it was too big to fail.
While I welcome the attention this study has received, there is risk in the hype. Rather than understand what might have been done better, we focus on why this study was so good. I have come to the conclusion that Open Dialogue offers something that is missing in NAVIGATE. But leaders of my field, and this likely includes the leaders of RAISE, would suggest there are inadequate data to make this conclusion. I agree. But I would argue that the results of RAISE are inadequate to conclude that NAVIGATE is the best model. What I am calling for here is to slow any rush to judgment.
I have another concern that relates to conflicts of interest. COMPASS, the decision making tool that guided prescribing decisions, leaned toward the newer antipsychotic drugs. Studies cited to support the use of the newer antipsychotics over the older ones were industry-sponsored studies. The history in this area is not good and suggests that our current evidence base provides a distorted picture of drug efficacy.
The lead author of RAISE is currently promoting the use of the new patented antipsychotic drugs that are given by injection on a monthly basis, including among those who have experienced single episodes of psychosis. He likely does this with the sincere belief that this is the best approach. However, his list of industry associations is long. He is a man with a significant stature in psychiatry and this study, and the attention it has received, only enhances that. What he says has impact whether he is talking about a government-funded study or a pharmaceutical study. If the Wunderink study on intermittent drug use is correct, then we may be making a tragic error with many young lives when we encourage patients to take long-acting injectable antipsychotic medicines. When I asked the RAISE leaders when was a good time to try to stop the drugs, the emphatic answer was “Never.”
A Post-RAISE World
In the U.S., there is currently much attention given to reform of the mental health care system. One area of focus has to do with early intervention with young people who are experiencing psychosis. RAISE is held up as a model. The outcomes from this study are not good enough to support the notion that we have all of the answers.
While we do not yet know if the outcomes from Finland can be replicated in other settings, there are enough positive data to suggest that we continue to evaluate that program. There are projects underway in the US and elsewhere and reports are forthcoming. But RAISE and Open Dialogue are not the only models. Oregon has implemented a wonderful program called Early Assessment and Support Alliance (EASA). It is a person-centered, flexible approach to working with young people that warrants broad attention. Questions regarding the roles of antipsychotic drugs and optimizing engagement are far from settled. There may be reason to celebrate the RAISE study, but while the results favored the experimental protocol, they did so only modestly. We can do better. I hope we remain open to exploring other promising models of care.