Is Psychotherapy Effective for Severe Mental Illness?
Controversial meta-analyses spark hope.
Posted April 2, 2014
The article begins with a story about a man named Terry who started hearing voices that told him to harm himself and his family members at 13 years old. He later became addicted to heroin and his marriage ended in divorce. Terry credits psychotherapist Jessica Arenella, Ph.D., whom he met when he was 46, with saving his life. For more than a decade, Dr. Arenella has helped Terry follow the gentle voices and ignore the nasty ones.
The article explains how some clinical psychologists are successfully treating psychosis with “talk” psychotherapy. Bucking a decades-old trend in which antipsychotic drugs have been the first line of defense against illnesses such as schizophrenia, they are engaging with patients’ symptoms, such as hearing voices or experiencing hallucinations, and taking them seriously.
As a research scientist, the first question I asked myself was: “Is there evidence that psychotherapy works for people with severe mental illness?”
I thought I’d find the answer easily, since two meta-analyses examining this issue were published this year. The first, by David Turner and his colleagues, 2 examined the relative efficacy of six different interventions for patients with psychosis. Studying 48 outcome trials that involved 3,295 patients, they concluded that cognitive-behavioral therapy (CBT) was significantly more efficacious than other interventions in reducing positive symptoms of psychosis. This finding was robust in all sensitivity analyses for risk of bias, but lost significance in sensitivity analyses for investigator blinding, which the investigators suggest may have been a function of reduced power.
The second meta-analysis, by Jauhar and colleages, 3 examined findings from 52 studies. It focused on people with schizophrenia who participated in either CBT or treatment as usual. That analysis concluded that CBT has a therapeutic effect on schizophrenic symptoms in the “small” range, effects that are reduced when sources of bias, particularly masking (treatment group is not hidden from those conducting the assessments), are controlled for. These investigators note that since 2008, only two published studies found a significant advantage for CBT when overall symptoms was the outcome and only one study found an advantage for CBT when positive symptoms was the outcome. Their conclusion: “The UK government’s continued vigorous advocacy of this form of treatment might be considered puzzling.”
These meta-analyses and an NIH-funded study based on 74 people, many of whom dropped out of the study before it ended, published last month in The Lancet 4 sparked controversy in the scientific world as well as in social media. There has been a lot of finger pointing and jumping up and down. Scientists questioning the value of CBT point to the flaws in existing studies, including: small samples; high drop-out rates; inclusion of people at all phases of illness; failure to blind treatment; outcomes that are not independent of one another; failure to measure treatment dose; and follow-up periods that are too brief. Meanwhile, scientists convinced of the value of CBT counter that the meta-analyses undervalue the efficacy of CBT by unfairly removing studies from consideration, including studies that use different forms of CBT, and failing to control for inclusion and exclusion criteria that can bias study findings. 5
Those favoring CBT and those believing it is ineffective for persons with severe mental illness agree about the magnitude of the effect size (a measure of the strength of a finding) in these meta-analyses. The two recent meta-analyses found effect sizes for CBT to be approximately .3 – similar in magnitude to that of studies examining the effects of clozapine for controlling psychotic symptoms. But drugs have serious side effects, and at least 50% of patients either refuse or fail to take them. This is not true of CBT.
As a scientist, I have to conclude that we have a long way to go before we understand whether CBT is an effective treatment for people with severe mental illness. We need adequately powered studies that assess proximal as well as distal outcomes. We need more sophisticated study designs in which assessors are blind to treatment status. We need studies with more explicit inclusion and exclusion criteria. And we need studies with designs that replicate one another so that findings are comparable. Maybe researchers have been asking the wrong questions. Maybe the question is not whether CBT is effective, but for whom and when is it effective? The bottom line is we need more high-quality research studies before we know whether or not CBT is effective.
As the mother of an adult child with severe mental illness, I’m delighted to see such controversy roiling the scientists. It means people are passionate about treating mental illness. It means we’re looking beyond medications for ways to treat some of our sickest citizens. For now, I have to agree with Jessica Arenella, Ph.D., the psychologist treating Terry who says, “In the end it doesn’t matter whether talk therapies work because of the theory behind them or just because someone is taking the patient and their symptoms seriously. It may be a placebo effect, but I will go for all the placebo effect I can get.”
It may be a placebo effect, but Dr. Arenella has seen it work. Maybe someday we’ll figure out how to do the science right. And maybe someday someone will figure out how to help my daughter.
- D.T. Turner, M. Van der Gaag, E. Karyotaki, and P. Cuijpers (2014). Psychological interventions for psychosis: A meta-analysis of comparative outcome studies. American Journal of Psychiatry, AiA, 1-16.
- S. Jauhar, P.J.McKenna, J. Radua, E. Fung, R. Salvador, and K.R. Laws (2014). Cognitive-behavioural therapy for the symptoms of schizophrenia: Systematic review and meta-analysis with examination of potential bias. British Journal of Psychiatry, doi: 10.1192/bjp.bp.112.116285
- A.P. Morrison, et al. (2014). Cognitive therapy for people with schizophrenia spectrum disorders not taking antipsychotic drugs: A single-blind randomized controlled trial. The Lancet, doi.org/10.1016/S0140-6736(13)62246-1