It often seems that psychiatry today, as an institution, is like a miner in the mid-1800s desperately sifting for gold. The “biological psychiatry” revolution has clearly failed, and with increasing evidence that psychiatric drugs do more harm than good over the long term, the field often seems focused on sifting through the mounds of research data it has collected, eager to at last sit up and cry, here’s a shiny speck of gold! Our drugs do work!
We saw that recently with the Cipriani study published in the Lancet, which led to headlines in British newspapers proclaiming the wonders of antidepressants. While even a quick look at the study revealed that it provided nothing new, and that the “evidence base” for assessing the risks and benefits of antidepressants hadn’t changed, the soundbite headlines were the “holler” heard round the world.
Lancet Psychiatry is back this month with another such holler, which is now making its way around psychiatric circles. The study tells of how discontinuing antipsychotic medication following recovery from an initial episode of psychosis increases the risk of “poor long-term clinical outcome.” That is a conclusion certain to provide comfort to prescribers today, as it will be seen as evidence that there is no need to change their current practices, which emphasize continual drug maintenance.
However, there is a newly published study in the Journal of the Norwegian Medical Association that tells of an opposite result. The Norwegian researchers report that a treatment they call “basal exposure therapy” has helped chronic patients taper from their medications (either down to a lower dose or off psychiatric drugs altogether), and that the best long-term outcomes are seen in those who got completely off the drugs. Unfortunately, I am pretty sure this one won’t soon be circulating in psychiatric circles.
On the surface, it might seem that these two studies have focused on the same question: how does an effort to discontinue from antipsychotics and other psychiatric drugs affect long-term outcomes? But, in fact, the studies had two very different purposes.
The first study was designed to test whether maintenance drug therapy was good for patients, with that proof, in large part, dependent on patients withdrawn from their medications faring poorly. The second study was designed to assess whether discontinuation could be done successfully and provide a long-term benefit.
It’s a tale of two studies, and a quick review of both helps frame an important ethical question for psychiatry: which of the two studies should guide their thinking? The one that is dependent on drug-withdrawn patients doing poorly, or the one designed to help drug-withdrawn patients succeed?
The Lancet Psychiatry Study
It takes a bit of detective work to fully understand the Lancet Psychiatry study, as you have to dig up an earlier published article to make sense of it. Only then can you understand how the initial discontinuation trial was conducted, and how that “discontinuation” exposure might have translated into a risk for a worse outcome at 10 years, which is the conclusion drawn in the Lancet Psychiatry article.
The earlier discontinuation study was conducted in a very select group of patients. The researchers identified 178 patients—out of a pool of 1,606 first-episode patients—who had been particularly good responders to antipsychotics for up to two years after their initial episode of psychosis. To be eligible for this study, the patients had to be symptom free, have suffered no relapses since their initial episode, and have been taking medication continually for at least one year. Only 11% of the 1606 patients met these criteria.
The 178 good-responders were randomized to treatment with quetiapine or to placebo. Astra Zeneca, the manufacturer of quetiapine, partially funded the study. A cross-taper protocol was used to transition patients from the antipsychotic they had been on to their randomized treatment (quetiapine or placebo), with this cross-taper lasting four to six weeks. As such, it seems that the patients in the placebo group were gradually withdrawn from their antipsychotic over this period.
Relapse was defined as the “reappearance” of psychotic symptoms, even if such symptoms were not severe enough to require hospitalization. This “lower threshold for defining relapse” would then lead to quick resumption of antipsychotic use in the placebo group. It also led to the classification of patients in the placebo group who experienced any noticeable return of psychotic symptoms as treatment failures.
It is easy to see the result that could be expected from this design. By enrolling only “best responders” to medication, all of whom had been symptom free while on an antipsychotic medication for at least a year, the investigators could expect that there would be a low rate of relapse in those randomized to quetiapine. Meanwhile, the drug-discontinuation group could be expected to do poorly. It is well known that there is a high risk of relapse for patients following withdrawal of an antipsychotic, and that is particularly true when the withdrawal is done through a one-size-fits-all protocol, as opposed to a patient-centered tapering process. This is a study design setting up the “discontinued” patients to fail, and by setting a low-threshold for defining relapse, the investigators could expect to give that failure rate an additional statistical boost.
The announced results fit the expectation. Nearly half of the placebo group relapsed during the first four months following randomization, and by the end of one year, 63% of the placebo patients had relapsed versus 30% of the quetiapine patients. The one surprise was that, overall, the quetiapine group didn’t fare so well either. In addition to the 30% that relapsed, another 31% discontinued due to adverse events (which was greater than the discontinuation rate in the placebo group), such that only a minority in the quetiapine group—39%—had stayed well during the year.
Given that this was conducted in a cohort of very good responders to antipsychotics, this trial could be said to have harmed both groups: a majority of patients in both cohorts either relapsed or dropped out due to adverse effects. Still, the relapse rate was higher in the placebo group, and psychiatry now had one more study supporting its current prescribing practices. “These findings are decisive in supporting maintenance treatment, even after one year of maintenance and in patients with no positive symptoms of psychosis,” the investigators wrote.
Seven years later, the Hong Kong investigators followed up on these 178 patients, and this led to their article in the March issue of Lancet Psychiatry. They reported that 21% of the patients randomized to maintenance therapy in the earlier study had a poor outcome at 10 years (which was defined as the presence of psychotic symptoms), compared to 39% of those who had been randomized to placebo. And with that data in hand, they concluded that “medication continuation for at least the first three years after starting treatment decreases the risk of relapse and poor long-term clinical outcome.”
That is the “holler” now circulating in psychiatry circles. Antipsychotic maintenance therapy works! But the devil of course is in the details.
The first detail is this: there was little difference in medication use in the two groups during the seven years following the randomization trial. If anything, drug use was higher in the “discontinuation” cohort.
Nearly two-thirds of the discontinuation group in that earlier study relapsed and had been put back on an antipsychotic, which meant that most patients from both cohorts had exited that earlier study on antipsychotics. The next seven years represented the “naturalistic” phase of the study and at the end of that time, 83% from the drug-discontinuation cohort were on antipsychotics, versus 76% of the drug-maintained cohort.
In other words, this isn’t a 10-year study related to the long-term use of antipsychotic medications. It is a study of whether the randomization assignment in that earlier study produced a difference in clinical outcomes seven years later.
The second detail is this: the researchers, for some unknown reason, switched their method for measuring outcomes. The original protocol for the 10-year study called for researchers to assess “recovery” rates for the two cohorts, which would have included assessment of their functional outcomes. But the investigators abandoned that measurement, and instead switched the primary outcome to the same “positive symptoms” measurement used in the initial relapse study. Patients at the ten-year mark experiencing even a lower level of psychotic symptoms would now be categorized as having a “poor outcome,” even though they might have been functioning okay.
The third detail involves the investigators choosing to include outcomes for patients who didn’t actually participate in the follow-up. The investigators were able to interview 142 of the original 178 in the study. Another eight had died, and the final 28 either refused to be interviewed or couldn’t be located. Yet, the investigators, rather than report the outcomes only for the 142 patients they were able to interview, added in the “predicted” outcomes for the 28 patients who couldn’t be found or refused to be interviewed, based on how they fared in the initial relapse study, and then categorized all 178 patients as either having a poor or good outcome. It’s unclear how the addition of the 28 “predicted” outcomes affected the comparison between the two cohorts, but it certainly makes for an odd element when reporting long-term clinical outcomes.
So what conclusion can be fairly drawn from this long-term follow-up of the patients that had been the subjects in an earlier discontinuation trial?
Here is my take: Putting first-episode psychotic patients who have had a good response to medication, lasting at least one year, into a study that switches them, according to a study-imposed timeline to another antipsychotic or to “placebo,” is a bad idea. The study exposes both groups to an increased risk of relapse (both the switching and drug withdrawal can be problematic), and this risk is particularly pronounced for the antipsychotic-withdrawn group. Moreover, it is possible that this relapse led to an increased risk of a “poor outcome” seven years later, because it does mean first-episode patients have now experienced a “recurrence” of symptoms (and perhaps this exposes them to subsequent greater medication use).
As such, this study tells of possible long-term harm done to patients by their participation in a study designed to produce a favorable result for the drugs, and nothing about the possibility that first episode patients can be successfully tapered from an antipsychotic. The study can’t provide any insight into that latter question, because there was no effort to support people in that process, or to manage the withdrawal symptoms that could be expected to appear.
One more point: if the researchers had wanted to provide some insight into the effects of long-term antipsychotic usage, they could have compared the outcomes for the 28 patients off medication at the end of 10 years to the 114 reported to be on medication. I am willing to bet that the outcomes for the unmedicated patients were better. That is the result seen in all other naturalistic long-term studies that have compared outcomes in this way. But that would have led to a different message, and not one that would lead to the study being circulated in psychiatric circles.
The Norwegian Study
The study published in the Norwegian journal was a report on the outcomes for patients treated with basal exposure therapy, which a Norwegian health provider—Vestre Viken Hospital Trust—has been developing and refining for the past 20 years as a treatment for chronic hospitalized patients. The patients admitted into the program typically have low functioning scores, numerous hospital admissions, and long-term use of psychiatric drugs (including polypharmacy). Most are described as “treatment resistant” when they enter the program.
In their paper, Hammer and colleagues explain the theory behind their novel treatment. The thought is that serious mental disorders “are sustained by avoidance behaviour.” The chronic patients avoid situations or environments that pose a risk of “existential catastrophe,” fearing that this could lead to their disintegration, or their “being engulfed by total emptiness or stuck in eternal pain.”
Thus, serious mental disorders are treated as “phobic conditions, irrespective of formal diagnoses,” and the purpose of basal exposure therapy is to expose them to the very conditions that stir their anxiety (much as other phobic conditions are treated through exposure therapy.) Although this exposure may initially trigger increased anxiety, repeated exposures will reveal to the patients that they can survive such experiences without disintegrating and that the existential threat to their being “is not real.” This becomes the path to recovery.
Psychiatric drugs, Hammer and colleagues write, can hinder this recovery process. The drugs are designed to suppress unwanted internal experiences, and yet it is precisely that internal experience that is needed to help patients get past their fears of existential catastrophes. As such, an important aspect of basal exposure therapy is promoting drug tapering to patients.
The drug withdrawal effort begins with a dialogue focused on the “patient’s own values,” as this promotes their “ownership of the process.” Usually, this leads to the initiation of an effort to gradually reduce the patient’s psychiatric medications, with a tapering plan prepared jointly by doctor and patient. Since many patients are on multiple drugs, the drugs are gradually withdrawn one drug at a time, which is one reason the inpatient treatment may last a fairly long time (six months or more). The patient’s progress through this drug-withdrawal regimen is closely monitored, with an interdisciplinary team—psychiatrists, psychologists, and psychiatric nurses—meeting weekly to review how it is proceeding.
In their article, Hammer and colleagues report on the long-term outcomes of the first 36 patients treated with basal exposure therapy and then discharged from the hospital. They were able to interview or find up-to-date electronic records for 33 of the 36, with these patients having been discharged, on average, five years earlier.
At the time of their enrollment into the basal exposure program, 14 of the 33 had a primary diagnosis in the schizophrenia spectrum; 6 had an affective disorder; 6 had a neurotic disorder; 5 had an emotionally unstable personality disorder; and 2 had other diagnoses. Eighteen of the 33 had multiple diagnoses. All but one were women, and they had quite poor global functioning scores.
The 33 had been discharged from the hospital at least two years earlier, and at followup, 16 of the 33 were off all psychiatric drugs. The 16 who were drug-free had markedly better outcomes than the 17 still taking psychiatric medications: they had much higher “global assessment of functioning” scores, much lower rehospitalization rates since discharge from the basal exposure program, and much higher “full-time” employment rates (56% to 6%). Seven of the 16 in the drug-free group had “fully recovered,” while none of the 17 still using psychiatric drugs achieved this best outcome.
The researchers also found that the patients who had the most exposure to “existential catastrophe anxiety” during their treatment had better social functioning over the long term.
The researchers concluded: “Basal exposure therapy may be a suitable approach for patients who suffer from severe and composite disorders and wish to become drug-free after a long period of psychotropic drug use. The hypothesis that withdrawal of psychotropic drugs combined with exposure may help these patients recover ought to be investigated in controlled prospective studies.”
A Plea To MIA Readers
Although both of these studies might be put into the “withdrawal research” category, they clearly are very different in kind. As I wrote earlier, the first was designed to prove that quetiapine prevents relapse, and exposed the drug-discontinuation group to a one-size-fits-all protocol that had been shown, in previous research, to reliably produce a high relapse rate. The second recorded the results from a therapy designed to help chronic patients get better, with successful drug tapering considered a key aspect of that treatment.
Now which one deserves to promote further thinking about the possible merits of psychiatric drug withdrawal efforts? The first, or the second?
Unfortunately, I know that it is the first study that is going to have the biggest impact. Lancet Psychiatry is the better-known journal, and the conclusion drawn by the researchers is one that most prescribers of psychiatric drugs will be happy to hear. It will confirm their belief that their current prescribing practices lead to better outcomes, even over the long-term. The second study will likely become little known outside Norway, and never enter psychiatry’s “collective consciousness.”
So here’s my plea to MIA readers. If you believe this second study deserves to be known, and that the results should become a topic of discussion in international psychiatric circles, please try to publicize it through social media, internet forums and so forth. A program that enables 50% of chronic patients to get off psychiatric drugs, which in turn leads to marked long-term improvement, needs to become part of the “evidence base” that—at least in theory—is supposed to guide psychiatric care.