The CAFE Study: Dr. Lieberman’s High Moral Ground


The CAFE Study, conducted by Jeffrey Lieberman, MD, et al between 2002 and 2005, has been the subject of much comment.  Carl Elliott, in particular, has written extensively on the matter, including his article The Deadly Corruption of Clinical Trials in Mother Jones.

In order to address the issues involved in the CAFE study, we must first take a brief look at the CATIE study.  This was also conducted by Dr. Lieberman et al (not the same et al as CAFE, but with some overlap).  CATIE was conducted between 2001 and 2004.

CATIE (Clinical Antipsychotic Trials of Intervention Effectiveness) has been written up in the New England Journal of Medicine.

CAFE (Comparison of Atypicals in First Episode of Psychosis) has been written up in two parts in the American Journal of Psychiatry, here and here.

CATIE was funded by the NIMH.  CAFE was funded by Astra Zeneca, the manufacturer of quetiapine (Seroquel).


CATIE was a double-blind randomized trial of the tolerability and effectiveness of five neuroleptic drugs.  Four of these drugs were second-generation neuroleptics.  The fifth was an older, first-generation neuroleptic.  The specific drugs compared were:

olanzapine (Zyprexa)
quetiapine (Seroquel)
risperidone (Risperdal)
ziprasidone (Geodon)

perphenazine (Trilafon)

1493 participants were followed for 18 months, and the study was conducted in three phases.

The National Institute of Health (of which NIMH is a section) issued a summary of CATIE’s phase 1 results in September 2005.  Here are some quotes:

“Overall, the medications were comparably effective but were associated with high rates of discontinuation due to intolerable side effects or failure to adequately control symptoms. One new medication, olanzapine, was slightly better than the other drugs but also was associated with significant weight-gain as a side-effect. Surprisingly, the older, less expensive medication (perphenazine) used in the study generally performed as well as the four newer medications.”

“Contrary to expectations, movement side effects (rigidity, stiff movements, tremor, and muscle restlessness) primarily associated with the older medications were not seen more frequently with perphenazine than with the newer drugs. The advantages of olanzapine — in symptom reduction and duration of treatment — over perphenazine were modest and must be weighed against the increased side effects of olanzapine.”

“Thus, taken as a whole, the newer medications have no substantial advantage over the older medication used in this study.”

This result is particularly important in the light of two contextual factors.  Firstly, the newer neuroleptics are a good deal more expensive than the older drugs.  Secondly, for years psychiatrists and their pharmaceutical allies had been promoting the falsehood that the side-effects with the newer drugs were not as severe.

CATIE phase 2 results:

Results: The time to treatment discontinuation was longer for patients treated with risperidone (median: 7.0 months) and olanzapine (6.3 months) than with quetiapine (4.0 months) and ziprasidone (2.8 months). Among patients who discontinued their previous  antipsychotic because of inefficacy (N=184), olanzapine was more effective than  quetiapine and ziprasidone, and risperidone was more effective than quetiapine. There were no significant differences between antipsychotics among those who discontinued their previous treatment because of intolerability (N=168).”

Conclusions: Among this group of patients with chronic schizophrenia who had just discontinued treatment with an atypical antipsychotic, risperidone and olanzapine were more effective than quetiapine and ziprasidone as reflected by longer time until discontinuation for any reason.”

CATIE phase 3 results:

Results: Time until treatment discontinuation for any reason was significantly longer for clozapine … than for quetiapine … or risperidone … but not for olanzapine …. Time to  discontinuation because of inadequate therapeutic effect was significantly longer for clozapine than for olanzapine, quetiapine, or risperidone. At 3-month  assessments, Positive and Negative Syndrome Scale total scores had decreased more in patients treated with clozapine than in patients treated with quetiapine or risperidone but not olanzapine. One patient treated with clozapine developed agranulocytosis, and another developed eosinophilia; both required treatment discontinuation.”

Conclusions: For these patients with schizophrenia who prospectively failed to improve with an atypical antipsychotic, clozapine was more effective than switching to another newer atypical antipsychotic. Safety monitoring is necessary to detect and manage clozapine’s serious side effects.”

I realize that this makes somewhat tortuous reading, and readers might want to refer back to the journal articles for clarification.  But the bottom line, for our present purposes, is that the newer neuroleptics showed no advantage over the older, and that within the newer group, clozapine (the first of the newer drugs) outperformed the more recent products.  And quetiapine (Seroquel) did particularly poorly.


The stated purpose of CAFE was to compare the effectiveness and tolerability of three second-generation neuroleptics in individuals with early psychosis.  The three drugs compared were:

olanzapine (Zyprexa)
quetiapine (Seroquel)
risperidone (Risperdal)

These are three of the four newer neuroleptics that were compared in the CATIE study.

CAFE was a randomized, double-blind trial.  400 participants were followed for one year, and the study was written up in two parts.

The conclusions of the authors, as published in the American Journal of Psychiatry, were as follows:

“Olanzapine, quetiapine, and risperidone all produced significant improvements in neurocognition in early-psychosis patients. Although cognitive improvements were modest, their clinical importance was suggested by relationships with improvements in functional outcome.” (here)

“Olanzapine, quetiapine, and risperidone demonstrated comparable effectiveness in early-psychosis patients, as indicated by similar rates of all-cause treatment discontinuation.” (here)


So quetiapine’s poor performance in the CATIE study is essentially overturned in the CAFE study.  Quetiapine apparently is just as effective as olanzapine and risperidone.

However…if you wade through the actual text of CAFE, you’ll find this interesting disclaimer on p 1058:

“Previous studies suggesting that first-episode patients receive therapeutic benefit from antipsychotic doses lower than those required for chronic patients and that first-episode patients develop unnecessary extrapyramidal side effects at higher doses led us to use lower dose ranges of olanzapine and risperidone in this study than those used in the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE). However, given quetiapine’s low extrapyramidal side effects liability, we expected that the quetiapine doses used in CATIE would be tolerable in our study population. This may have been a factor in the comparable effectiveness demonstrated by quetiapine. In the CATIE phase 1 and phase 2 schizophrenia trials…higher mean modal doses of olanzapine and risperidone but similar mean modal doses of quetiapine were used; in these trials, olanzapine and, less consistently, risperidone proved to be more effective than quetiapine.” [Emphasis added]

In other words, CATIE compared olanzapine, quetiapine, and risperidone, and found that quetiapine did poorly.  CAFE compared olanzapine, quetiapine, and risperidone, and found that they did equally well.  But they reduced the dose of olanzapine and risperidone.

All of these products act as major tranquilizers in the early stages of their use.  They reduce psychotic thinking because they reduce all cognitive activity.  And the more a person takes, the greater the tranquilizing effect.

In the quotation above, CATIE’s authors justify the dosage discrepancy on the grounds that higher doses of olanzapine and risperidone would have placed first episode individuals at higher risk of extrapyramidal side effects (e.g. tardive dyskinesia).  Quetiapine, however, we are told, does not pose such a risk because of “…its low extrapyramidal side effects liability.”  But the CATIE study had found no difference in these products with regards to this risk.  And the CAFE authors don’t quote any references to support this claim.

Given that the study was funded by Astra Zeneca (the manufacturer of quetiapine), it is very difficult to avoid the conclusion that the purpose of CAFE was simply to rescue Astra Zeneca’s product from the negative findings of CATIE.  It looks more like marketing than research.

In this regard, it is perhaps worth noting that in 90% of drug trials that are funded by a pharmaceutical company, the sponsor’s product is found to be superior to the competitor’s product. (Heres et al 2006)

The notion that the CAFE study reflects sponsor contamination has been addressed by other writers.  Bernard Carroll, MD, PhD, past chairman of the FDA psychopharmacologic drugs advisory committee, and past chair of the Department of Psychiatry at Duke University, has written a lengthy comment on the The Carlat Psychiatry Blog on a post concerning the CAFE study..  He offers eleven separate criticisms of the CAFE study, and draws the following conclusions:  “…the entanglement with a commercial entity has influenced the scientific agenda of these academic investigators.”

And, of course, Carl Elliott has been an outspoken critic of the CAFE study for several years.

Two of CAFE’s participants committed suicide during the study period.  Both had been randomly assigned to the quetiapine treatment.  There are very strong indications that one of these individuals – Dan Markingson – was coerced into the study and, because of his psychotic state, was incapable of giving informed consent anyway.  His mother, Mary Weiss, made repeated attempts to have him withdrawn from the study but without success.  (The Department of Psychiatry at the University of Minnesota earned $15,648 for each study participant who completed the treatment course.)  You can find a detailed account of this matter in the Carl Elliott article mentioned earlier: The Deadly Corruption of Clinical Trials.

For a drug company to promote its product is, I suppose, understandable.  It’s what they do, though their tactics are often questionable.

But for psychiatrists – including an eminent psychiatrist such as Dr. Lieberman, currently president of the APA – to lend their names and their reputations to this kind of thing is reprehensible.  And this, incidentally, is the same Dr. Lieberman who on June 18 of this year, told us that psychiatry has the moral high ground, and that he plans to respond to psychiatry’s critics by assuring us that he and his colleagues are planning to rededicate themselves to their Distinguished Fellow pledge!  So that will make everything OK.


To guard against any misapprehension, it is my view that the drugs referred to in these studies as anti-psychotics are more correctly described as neurotoxic tranquilizers with devastating side effects, including tardive dyskinesia, akathisia, and significantly reduced life expectancy.  Comparing the efficacy of these products seems to me a little bit like asking: which is better for your health, a tornado, a hurricane, or a plane crash?