In its 2011 summer issues, Behavioral Healthcare ran a two-part interview with me about my book, Anatomy of an Epidemic. This stirred William Glazer, a well-known psychiatrist who has served as a consultant to Eli Lilly since 1992 (and to other pharmaceutical companies during that time as well), to pen a two-part “rebuttal” to Anatomy of an Epidemic. This is the second in-depth attempt by a psychiatrist with close industry ties to “rebut” the book” and so it should be worthwhile to take a closer look at it.
In particular, we will want to make this assessment: Is Dr. Glazer able to point to research findings that show that psychiatric medications improve the long-term outcomes of major mental disorders? And is his rebuttal intellectually honest?
Dr. Glazer’s Industry Ties
In his Behavioral Healthcare rebuttal, Dr. Glazer notes that he has “had a longstanding collaborative relationship with the pharmaceutical and insurance industries,” and that he receives funding from Eli Lilly and Merck. In the past, he has also been a consultant to Johnson & Johnson, Bristol-Myers Squibb, and Astra Zeneca. He also is owner of Glazer Medical Solutions, which promotes continuing medical education programs (which often are funded by pharmaceutical companies.)
Dr. Glazer’s “Argument Number One.”
In Anatomy of an Epidemic, I detail how there has been a dramatic rise in the number of “disabled” mentally ill since the beginning of the “psychopharmacological revolution,” and that this has been particularly true since the arrival of Prozac and the other second-generation drugs. I then note that this data raises a question: Is it possible that our drug-centered paradigm of care is fueling this epidemic? And to answer that question, I then spent several hundred pages reviewing the outcomes literature for major mental disorders to see whether this history of science tells of drugs that improve long-term outcomes, or worsen them.
Although Dr. Glazer’s opening argument is, in my opinion, rather confused, he basically states this: Although it is true that there has been a sharp rise in the number of people on disability due mental illness, there is no “evidence available to suggest that the increase in psychiatric disability claims was caused by the use of psychiatric medications.” Instead, he argues that the rise is due to the inadequacy of funding for community-based services, the growing disparity in wealth in the United States, and changes in disability rules that have made it easier for people to qualify for a disability payment. He also states that epidemiological surveys conducted in 1990 to 2003 did not find any dramatic increase in the prevalence of mental disorders, and thus, there is no “epidemic.”
Now, the interesting part of this argument is that, if we are to accept Dr. Glazer’s statement that there is “no available evidence” to assess whether psychiatric medications are fueling an epidemic of disabling mental illness, then it shows that psychiatry has an utter lack of evidence on how medications shape long-term outcomes. Indeed, Dr. Glazer states that it would be worthwhile to study whether the drugs are causing the surge in the number of disabled mentally ill. “This is a “hypothesis in need of testing, not a fact. A properly designed study to test this hypothesis would have to rule out many possible confounding factors. And no such study has been proposed or conducted.”
I would think that the American people, faced with the mounting cost of the disability epidemic, would indeed welcome such a study. But, in fact, one could argue that the prevalence and treatment figures cited by Dr. Glazer for the years between 1990 and 2003 provide the raw data for such a study.
Here is the data. The prevalence of psychiatric disorders was 29.4% in 1990-1992 and 30.5% in 2001-2003. Thus, the percentage of adults in the population experiencing psychiatric distress remained the same during these two periods. However, what changed during this period was that the percentage of people with a psychiatric disorder who were treated for it rose from 20.3% to 32.7%.
Thus, this is the hypothesis that can now be tested: If treatment increases the likelihood that a person struggling with psychiatric distress will end up on disability, then the total number of people on disability should have climbed sharply during that 13-year period.
Let’s do the math. If you then look up census data for the number of adults in 1990 and 2003, and do the relevant calculations, you find that the number of people treated rose from 11.16 million adults in 1990 to 21.77 million in 2003. And what happened as more people got treated? The number of people receiving SSI or SSDI due to mental illness during this period more than doubled, from 1.47 million people in 1990 to 3.25 million people in 2003.
So here is the bottom line: The prevalence remained the same between 1990 and 2003, but the percentage of people treated for the disorder rose, and as that happened, the number of people on disability soared.
There is one other way we can parse this data. In 1990, there were 55 million adult Americans with a “psychiatric disorder” (based on the prevalence data cited by Dr. Glazer.) There were 1.47 million adults who collected a federal disability check that year due to mental illness. Thus, one in every 37 adults (2.7%) with a psychiatric disorder in 1990 was on long-term disability. By 2003, the adult population had increased, and thus although the prevalence of psychiatric disorders had remained the same as in 1990, the number with a psychiatric disorder had risen to 66 million. There were 3.25 million adults who were collecting a federal disability payment that year due to mental illness, or one in every 20 with a psychiatric disorder (5%).
Thus, during this period of increased treatment, the percentage of those with a psychiatric disorder who ended up on long-term disability rose from 2.7% to 5%.
In sum, we already have data that can be used to test the hypothesis that Dr. Glazer thinks is worthy of investigation. Unfortunately, the results are exactly what you would expect if psychiatric drugs are indeed fueling an “epidemic” of disabling mental illness. And I should note that this epidemic of disabling mental illness is showing up in country after country that has adopted this paradigm of care.
Dr. Glazer’s “Argument Number Two”
In the second part of his rebuttal, Dr. Glazer discusses several of the scientific studies I cited in Anatomy of an Epidemic.
1. Supersensitvity psychosis
In Anatomy, I detail how, by the end of the 1970s, various studies had raised the worry that antipsychotics might be making people diagnosed with schizophrenia more biologically vulnerable to psychosis over the long run. This is how the NIMH’s William Carpenter framed this worry:
“There is no question that, once patients are placed on medication, they are less vulnerable to relapse if maintained on neuroleptics. But what if these patients had never been treated to drugs to begin with? We raise the possibility that antipsychotic medication may make some schizophrenic patients more vulnerable to future relapse than would be the case in the natural course of the illness.”
Two Canadian investigators, Guy Chouinard and Barry Jones, then proposed an explanation for why this might be so. The first-generation antipsychotics had been found to block dopamine receptors in the brain, particularly a subtype known as D2 receptors. In an effort to compensate for this response, the brain increased the density of its D2 receptors. This made the brain “supersensitive” to dopamine, Chouinard and Jones argued, and this in turn made patients more likely to relapse when they withdrew from antipsychotics. At the same time, when patients stayed on the drugs long-term, this process could lead to a “tardive psychosis,” and when this happened, the illness “appears worse than ever,” they wrote.
In his review, Dr. Glazer states that there is no need to worry about this any longer. He writes that “an authoritative review of the literature concluded in the early 1990s that ‘research has not established that neuroleptics caused the proposed effect, and considerations of mechanism have not been separated from those of causation.’ . . . This idea did not bear out. ”
So let’s see if that is indeed true. The 1992 study cited by Dr. Glazer was titled “The concept of supersensitivity psychosis.” Dr. Glazer has quoted part of the abstract correctly, but he has left off the key concluding line, in which the authors state that more research is needed on this subject. “The focus of research in this area should be the establishment of a causal relationship between chronic neuroleptic use and psychotic relapse.”
In other words, this paper, which Dr. Glazer has cited as reason to think the concern had been put to rest, actually stated that more research was needed. That was in 1992, and fortunately there has been further research on “supersensitivity psychosis” by Philip Seeman, at the University of Toronto.
First, in 2005, Seeman reported that agents that triggers psychotic-like behavior in animals—amphetamines, angel dust, lesions to the hippocampus, and gene-knockout manipulations—all cause an increase in D2 receptors that have a “high” affinity for dopamine (meaning that they are very sensitive to the neurotransmitter.) These results “imply that there may be many pathways to psychosis, including multiple gene mutations, drug abuse, or brain injury, all of which may converge via D2 HIGH to elicit psychotic symptoms,” Seeman wrote.
However, in this same report, Seeman found that haloperidol and olanzapine both doubled the density of D2 HIGH receptors. Thus, the drugs cause the very biological abnormality that in animal models had been identified as a final pathway to psychosis.
This finding set up an obvious experiment. Seeman could induce psychosis in rats, and then see how the drugs affected their behavior. Since the drugs blocked D2 receptors, at first the drugs should be effective. But then, over time, since they also caused a doubling of D2 HIGH receptors, that should lead to supersensitivity psychosis and “relapse.”
Seeman did that study, and here is what he reported in 2007: “We show that during ongoing treatment with clinically relevant doses, haloperidol and olanzapine progressively lose their efficacy . . . the loss of efficacy is linked to an increase in D2 receptor number and sensitivity. These results are the first to demonstrate that ‘breakthrough’ supersensitivity during ongoing antipsychotic treatment undermines treatment efficacy.”
I am sure that Dr. Glazer knows of Philip Seeman’s work, as I wrote about it in Anatomy of an Epidemic and on this blog. So we see in this quick review of “supersensitivity psychosis” that the worry arose in the 1970s, and that in recent years, Philip Seeman has shown, in animal studies, that the drugs do cause this effect.
So now we can ask: Given Philip Seeman’s work, was it intellectually honest for Dr. Glazer to tell readers of Behavioral Healthcare that worry about supersensitivity psychosis had been put to rest two decades ago?
2. Martin Harrow’s 15-year outcomes study
If readers of this blog want to know what I actually wrote about this study, instead of what Dr. Glazer claims I wrote, I urge them to read pages 115 to 118 of Anatomy of an Epidemic. But let’s visit the actual findings from that prospective study (yet again.)
From 1975 to 1983, Harrow enrolled sixty-four young people diagnosed with schizophrenia into his study, recruiting them from two Chicago hospitals. One was private and the other public, as this ensured that the group would be economically diverse. He also enrolled 81 young people with milder psychotic disorders. All of these patients were treated conventionally with medications in the hospital, and then Harrow tracked their long-term outcomes, periodically assessing how they were doing. Were they symptomatic? In recovery? Employed? Were they taking antipsychotic medications? And here is what his data shows:
• The 25 schizophrenia patients who got off antipsychotic medications had much better long-term outcomes than those who stayed on the drugs. The off-antipsychotic patients had an eight-fold higher recovery rate; their aggregate global outcomes were much better; they were much less likely to have a “uniformly poor outcome;” and they were much less likely to still have psychotic symptoms at the 10-year and 15-year follow-up assessments.
• Among the 81 patients with milder psychotic disorders, those who stopped taking antipsychotics and other psychiatric medications (nearly half of the cohort) had markedly better global outcomes than those who stayed on the medications.
• The collective outcomes for all of the psychotic patients stacked up like this, from best to worse: milder psychotic disorders off meds, schizophrenia off meds, milder psychotic disorders on meds, and schizophrenia on meds. In other words, those with a milder disorder at the beginning who stayed on antipsychotics did worse long-term than the schizophrenia patients who got off the medications.
• Finally, Harrow grouped his schizophrenia patients into good-prognosis and bad-prognosis patients. While he didn’t provide the specific outcomes data for each group, he did observe that in each of these two sub-groups, it was those who got off antipsychotics who had better long-term outcomes.
Now this is the only study of its type in the outcomes literature, in which a researcher has followed a large cohort of patients over a long period of time, and assessed both their medication usage and outcomes. As Martin Harrow concluded, the results belie the common wisdom that all people diagnosed with schizophrenia and other psychotic disorders need to be on antipsychotics all their lives. At the very least, it shows that there is a subset of patients that can do well off the medications long-term.
However, Dr. Glazer presents none of this data in his discussion of the study. He doesn’t mention that every subset of patients who got off antipsychotics, including those with a bad prognosis, had better outcomes. Instead, he simply claims that this is just a matter of those with a better prognosis getting off the meds doing better. “Standard clinical logic would conclude that poor outcomes in the antipsychotic treated patients were a function of their illness, not of their treatment.”
As readers of this blog know, I presented at a Grand Rounds at Massachusetts General Hospital last January, and in response Dr. Andrew Nierenberg presented a “rebuttal” of my book. In this rebuttal, Dr. Glazer follows Dr. Nierenberg’s lead time and again, and he ends his dismissal of the Harrow study with this line, which he borrowed from Dr. Nierenberg: “Mr. Whitaker needs a basic course on principles of epidemiologic research, specifically on the concept of susceptibility bias.”
So readers can now ask this question: Why didn’t Dr. Nierenberg and Dr. Glazer discuss the actual data, which showed that every subset of patients off medications did better? Why didn’t they discuss the dramatic differences between every subset of patients? And what does it say about the integrity of their argument that, having failed to discuss the actual data, they conclude that I, a journalist who reported the actual data, should be considered a dunce, in need of remedial education?
3. Nancy Andreasen’s MRI study on brain volumes
Next, we turn to the issue of whether antipsychotics shrink the brain, and in particular, studies on this subject by Nancy Andreasen, who was the editor in chief of the American Journal of Psychiatry from 1993 to 2005.
Here’s a quick review of the relevant literature. First, animal studies have found that antipsychotics do indeed shrink the brain. For instance, in a study in macaque monkeys, researchers reported in 2005 that treatment with either haloperidol or olanzapine for 17 to 27 months led to an “8 – 11% reduction in mean fresh brain weights” compared to controls. The differences in brain weights and brain volumes “were observed across all major brain regions, but appeared most robust in the frontal and parietal regions.”
Nancy Andreasen began her large study of schizophrenia patients in 1989. She periodically used MRI technology to measure their brain volumes, and in 2003 she reported that their brains shrunk over time. At that point in her research, she blamed the shrinkage on the disease. She reported that schizophrenia was a “progressive neurodevelopmental disorder” characterized by “progressive reduction in frontal white matter volume.” Next, in 2003 and 2005, she reported that this brain shrinkage was associated with a worsening of negative symptoms, increased functional impairment, and, after five years, cognitive decline.
But there was an obvious question for Andreasen to address, given the animal studies: Was this shrinkage due, at least in part, to the drugs? In February 2011, she reported that this shrinkage was indeed drug-related. Use of the old neuroleptics, the atypical antipsychotics, and clozapine were all “associated with smaller brain tissue volumes,” with decreases in both white and grey matter. The severity of illness and substance abuse had “minimal or no effect” on brain volumes.
Here is what she told the New York Times: “What exactly do these drugs do? They block basal ganglia activity. The prefrontal cortex doesn’t get the input it needs and is being shut down by drugs. That reduces psychotic symptoms. It also causes the prefrontal cortex to slowly atrophy.”
That’s the story told by Andreasen. However, in his rebuttal, Dr. Glazer doesn’t report her conclusions, or mention the animal research. Instead, he writes that “it must be noted that these brains were compared to those of normal control subjects, not to those of untreated schizophrenia patients. Thus, it is impossible to examine whether it is the medications or the illness that lead to the smaller brain size.”
That is not what Andreasen concluded. She stated that the drugs cause “the prefrontal cortex to slowly atrophy.” And so, once again, we have the same question to ask: why didn’t Dr. Glazer honestly present her research findings?
4. Evidence that antipsychotics improve long-term outcomes
When Dr. Nierenberg and now Dr. Glazer sought to rebut Anatomy of an Epidemic, the obvious thing for them to do would be to cite studies that show that psychiatric medications improve long-term outcomes. Both claimed that they had identified studies that did just that.
In regards to antipsychotics, Dr. Nierenberg turned to an obscure study of older patients diagnosed with schizophrenia in China who weren’t treated with antipsychotics. I have written about how he misrepresented that study and its findings, and so won’t repeat that explication here.
For his part, Dr. Glazer didn’t try to point to that Chinese study in his rebuttal (likely seeing it as ridiculous in kind), and so he sought a different one. He cited a study of 104 patients treated at Hillside Hospital in Queens during the late 1980s for a first episode of schizophrenia or schizoaffective disorder, who were then followed for five years. He reported that those treated with “antipsychotics were five times less likely to relapse than were patients who stopped such medication,” and thus this study “is one of the best demonstrations of the beneficial impact of long-term antipsychotic therapy.”
I looked up that study, and I have to confess that I gasped. I had written about this research in 1998, when I co-wrote a series for the Boston Globe on abuse of patients in psychiatric research settings.
Here’s the background to the study. Starting in the 1970s, researchers investigating the “dopamine hypothesis” of schizophrenia reasoned that if too much dopamine could cause psychosis, then a drug that caused brain neurons to release dopamine—amphetamine, methylphenidate, L-dopa—should make psychotic patients much worse. They began running experiments to see if that was so, and in the 1970s, David Janowsky, a physician at the University of California at San Diego School of Medicine, reported that dopamine-releasing drugs did indeed make psychotic patients much worse. Methylphenidate, which caused a doubling in the severity of symptoms, was the most potent of the dopamine-releasing agents when it came to making patients more psychotic.
In the late 1980s, psychiatrists at Hillside Hospital in Queens decided to repeat this experiment in first-episode patients who came stumbling into the hospital’s emergency room for help. Rather than treat them with neuroleptics, they gave them methylphenidate, expecting this drug would make them worse. In studies published in 1993, the researchers reported that methylphenidate caused 59 percent of them to temporarily become “much worse” or “very much worse.” The patients were then placed on neuroleptics, but they took longer than usual to stabilize. “We were surprised by the length of time required for patients to recover,” the investigators reported.
The researchers then followed that patient group for another five years. Eighty-two percent relapsed at least once during the five years (after recovering from their initial episode), and a significant percentage suffered several relapses during that period. Those patients who discontinued their medications and remained in the study—the study doesn’t state how many patients were in this category—relapsed at a five-fold higher rate than those who continued on the drugs. However, there were also 13 stable patients off medication who dropped out of the study, and researchers don’t know whether this group subsequently relapsed, or stayed well off medication. (In his study, Martin Harrow discovered that the off-med group that stayed well dropped out of treatment, and thus their success remained hidden to clinicians.)
Researchers have been studying neuroleptics for 50 years, and this is the study that Dr. Glazer cites as one of the “best demonstrations of the beneficial impact of long-term antipsychotic therapy.” And quite apart from its egregious ethical nature, the fact that patients who eventually stabilized on the drugs and then went off them relapsed at high rates is not surprising either; there is a long line of evidence that once patients are exposed to antipsychotics they are at great risk of relapse when withdrawing from them. That, in fact, is the whole point of the dopamine supersensitivity problem: the drugs induce changes in the brain that increase the risk of relapse upon drug withdrawal.
Let us now weigh the respective evidence reviewed by Dr. Glazer, and his assessment of it. The 15-year study by Martin Harrow that found that patients off antipsychotics had much better long-term outcomes is to be dismissed. Worries about drug-induced supersensitivity psychosis are to be dismissed, in spite of the recent findings by Philip Seeman. The fact that Nancy Andreasen reported that antipsychotics shrink the brain is to be dismissed. Instead, if our society wants to know whether antipsychotics improve long-term outcomes, it should look to a study of 104 first-episode patients that, once you investigate its design, involved deliberately making them worse by giving them methylphenidate when they arrived at the emergency room. That study then found after the patients had gotten worse on that drug, they took a longer time than normal to stabilize on neuroleptics. The group as a whole had high relapse rates over five years, with the highest relapse rate seen in those patients who went off the drugs and stayed in the study. We don’t know what happened to 13 off-med patients who were stable and then dropped out of the study. The study also doesn’t tell us anything about how well the on-med patients functioned during that five-year period.
So, after 50 years of neuroleptics, this study of first-episode patients at Hillside Hospital in Queens, in which researchers initially gave the patients a drug expected to make them worse, is—according to Dr. Glazer—some of the “best” evidence there is that antipsychotics provide a long-term benefit.
5. Evidence that antidepressants improve the long-term course of depression.
Here, Dr. Glazer turns to the same NIMH study that Dr. Nierenberg did in his “refutation” of Anatomy of an Epidemic. He tells readers that in a 20-year study by Leon, “vigorous (antidepressant) treatment resulted in better short-term and long-term outcomes than did suboptimal treatments or no treatment at all.”
Here’s the context for the study. From 1978 to 1981, the NIMH Collaborative Depression Study recruited 955 subjects who sought treatment at a hospital for one of the major affective disorders. Over the next two decades, NIMH-funded researchers conducted many studies of this population. In the Leon study cited by Dr. Nierenberg and Dr. Glazer, researchers were studying 285 patients with major depressive disorder who, at some point during the 20 years, suffered a recurrence of depression. These 285 patients had been treated with antidepressants, and many were taking antidepressants when they suffered a relapse. And what the researchers studied in this particular study was whether those who had suffered a relapse should be treated with a high dose of an antidepressant, a medium dose, or a low dose. They found that the recovery rate from that recurrent episode, over a short period of time, was higher for those given a high dose, and that apparently a medium dose was also better than no somatic treatment (although the study doesn’t provide any information about that no-treatment group, and so it is unclear how that comparison was made.)
Thus, the Leon study doesn’t provide any information on the long-term outcomes of depressed patients who take antidepressants versus those who eschew such treatment. However, and this is the important point, NIMH investigators leading the Collaborative Depression research effort did conduct a well-designed study that made such a comparison. Coryell and his colleagues studied 547 people who suffered a bout of depression, and they found that over the course of six years, those who were treated for the illness were three times more likely than the untreated group to suffer a “cessation” of their “principal social role,” and nearly seven times more likely to become “incapacitated.” Moreover, while many of the treated patients saw their economic status markedly decline during the six years, only 17 percent of the unmedicated group saw their incomes drop, and 59 percent saw their incomes rise. “The untreated individuals described here had milder and shorter-lived illnesses [than those who were treated], and, despite the absence of treatment, did not show significant changes in socioeconomic status in the long term,” the researchers wrote in their 1995 report.
In short, Dr. Nierenberg picked a short-term dosage study from this larger NIMH research effort to assert that antidepressants improve the long-term outcomes of depressed patients. He ignored the actual study from that same research program that assessed six-year outcomes of treated and untreated patients, and had found that the untreated patients fared remarkably better. Now here is what I find remarkable: I had already written about this on my blog, which Dr. Glazer clearly read. And yet still Dr. Glazer chose to follow Dr. Nierenberg’s lead, and present the Leon study in the same disingenuous way that he did.
I should note that if Dr. Glazer is looking for evidence that treatment increases the risk that a depressed patient will become disabled, he might want to read the Coryell study. There, the “treated” group was seven times more likely to become “incapacitated.”
Dr. Glazer’s Argument Number Three
In the last section of his rebuttal, Dr. Glazer argues that the psychiatric establishment has been an honest broker of information, and that, for the most part, those who serve as consultants and advisors to pharmaceutical companies, as he and Dr. Nierenberg do, have not been corrupted by that financial influence. They remain devoted to science, unbiased, and honest in their statements to the public, and thus I am wrong to suggest otherwise.
All I can say is that, having parsed the “rebuttal” made by Dr. Nierenberg back in January, and again here in this two-part rebuttal by Dr. Glazer, I think I can “rest my case” on this particular topic