BEYOND THE ABSTRACT

1. Wheels within wheels    This is a rambling review, liable to go off on tangents and to proceed slowly in instalments. This part focuses on the background to a recently published clinical trial report about withdrawal from fluoxetine, paroxetine and sertraline.

The trial was funded by Eli Lilly, the manufacturers of fluoxetine (Prozac), and the results were first presented at a conference, held last year on Amelia Island ("Florida's Timeless Treasure" etc). Under a grant from Eli Lilly, the conference was sponsored by the Society of Biological Psychiatry, which has just published the paper in its journal:

J F Rosenbaum, M Fava, S L Hoog, R C Ascroft, W B Krebs: Selective Serotonin Reuptake Inhibitor Discontinuation Syndrome: A Randomised Clinical Trial, Biol Psychiatry, 1998, 44, 77-87.

The bottom line of the trial was that Prozac won hands down, with Lustral/Zoloft and Seroxat/Paxil lagging far behind. Eli Lilly has been advertising this advantage for some time. "Dear Doctor … (etc etc etc) … "You can see that Prozac can help improve depression management on stopping therapy and when occasional doses are missed".

The principal author of this report, Dr Jerrold F Rosenbaum, is also co-author of three other recently-published reports; his colleague, Dr Maurizio Fava, is co-author of two. These related reports not only press the point but build on it, seeking to show how effective Prozac is in the long-term:

J Zajecka, M D Fawcett, J Amsterdam, F Quitkin, F Reimherr, J Rosenbaum, D Michelson, C Beasley: Safety of Abrupt Discontinuation of Fluoxetine: A Randomised, Placebo-Controlled Study, J. Clin Psychopharmacol., 1998, 18, 3, 193-197.

F W Reimherr, J D Amsterdam, F M Quitkin, J F Rosenbaum, M Fava, J Zajecka, C M Beasley, D Michelson, P Roback, K Sundell: Optimal Length of Continuation Therapy in Depression: A Prospective Assessment During Long-Term Fluoxetine Treatment: Am J Psychiatry, 1998, September, 155, 9, 1247-1253.

J W Stewart, F M Quitkin, P J McGrath, J Amsterdam, M Fava, J Fawcett, F Reimherr, J Rosenbaum, C Beasley, P Roback: Use of Pattern Analysis to Predict Differential Relapse of Remitted Patients with Major Depression During 1 Year of Treatment with Fluoxetine or Placebo: Arch Gen Psychiatry, 1998, September, 55, 334-443.

All four studies list authors who work for Eli Lilly (see table), and financial support from the company is acknowledged in three. No reference to funding appeared in the report by Zajecka and colleagues.

Zajecka and Rosenbaum were also members of the seven-man SSRI Discontinuation Consensus Panel set up by Eli Lilly, at the closed symposium in Arizona, in December 1996. This Panel's report later appeared in a 40-page supplement, also sponsored by Eli Lilly, and published in the Journal of Clinical Psychiatry - where Rosenbaum and Quitkin are members of the editorial board.

More recently, the same journal published in its "Academic Highlights" section, another 8-page review about discontinuing antidepressant drugs. This included a detailed summary by Rosenbaum of the findings reported in Biological Psychiatry - plus a report by Dr Peter Haddad, another member of the Discontinuation Consensus Panel. Both papers had previously been read at a 1997 symposium at the Geneva Congress of the World Psychiatric Association. The symposium was sponsored by Eli Lilly, as was the 8-page "Academic Highlights" section of J. Clin Psych.

The authorship of the Highlights of the Rosenbaum paper is unclear. If the suggested citation is to be believed, Rosenbaum is no shrinking violet; if not, the report was perhaps written by someone from Eli Lilly, on the basis of what Rosenbaum said:

"Dr Rosenbaum and colleagues investigated the effects of this intermittent non-compliance in the first gold-standard, double-blind, placebo-controlled, parallel design study comparing the effects of treatment interruption among SSRIs" [Rosenbaum J, Risk of Adverse Events and Depressive Symptom Breakthrough Following Brief Interruption of SSRI therapy, pp 555-557. In Thompson C, chairperson, Discontinuation of Antidepressant Therapy: Emerging Complications and Their Relevance (ACADEMIC HIGHLIGHTS), J. Clin Psychiatry, 59, 541-548.]

Two further things to be said about the main paper, (Rosenbaum, Fava et al, 1998), are that it is not clear who did the actual work, and what roles the different authors played. The evidence suggests their study was very much orchestrated by Eli Lilly and clearly its marketing value is high.

Its value was enhanced by publishing in the names of the two independent authors, but it is not certain what credit they are due. The situation is complicated because an earlier version of the same study was published a year earlier under different names. In the preliminary report, the first four author credits went to employees of Eli Lilly, while the two big names brought up the rear. Unaccountably, in the final version, the two names underlined (including the principal author) have now disappeared:

S.L. Blomgren, W. Krebs, M. Wilson, R. Ascroft R, M. Fava M, J. Rosenbaum, SSRI Dose Interruption Study: American Psychiatric Association, 1997 Interim Data, American Psychiatric Association, 150th-Annual-Meeting, San-Diego, 1997

The roles of Rosenbaum & Fava are also uncertain, in the light of a small-print acknowledgement to a commercial research organisation::

Lilly Research Laboratories would like to acknowledge Clinical Studies Ltd, 2 Charles Street, Providence, Rhode Island, for assistance in conducting this study at 12 of its community-based research centers.

Whether and to what extent Drs Rosenbaum and Fava actually treated the patients in this study is therefore also unclear. Perhaps the most positive thing to be said about their prominence as authors is that it introduces an element of independent oversight. As both Rosenbaum and Fava both work for the Massachusetts General Hospital (MGH), the influence of colleagues, ethical committees and institutional standards cannot be overlooked. Enquiries will be made.

Other questions arising from all this relate to the quality of the research and relevance of the study findings, and to the extensive involvement of Eli Lilly in this small but influential world.

 

2. Three buckets, two sizes of hole   One of the requirements in the MGH guidelines on conflict of interest is that an investigator should notify, in writing, the manufacturers of other products being evaluated against the sponsor's own. In this case, it would have meant writing to SmithKline Beecham (paroxetine) and Pfizer (sertraline), to tell them that Eli Lilly was comparing their products with fluoxetine (Prozac). Tact would have been needed, because the design of the trial guaranteed Prozac would come out on top. As contests go, it was like seeing which held water longest - two buckets with small holes in the bottom, or one with a pinhole.

The general idea was that patients who were established users of the three study drugs would have their medications secretly withdrawn for about a week (ie a double-blind procedure), before being reinstated on the active drug. The job of the investigators was to record what happened, and compare the results with those of a matched group of patients whose drug supply had been maintained. Aside from ethical questions, this was a quite legitimate race to run, and the study in question seems to have been well designed and properly controlled (Rosenbaum, Fava, 1998). For brief details of the procedure, see (nee Blomgren et al, 1997).

Perhaps the main problem with the study would be to do with its scope and real clinical significance - given that it carefully measures what happens when people come off paroxetine or sertraline, but doesn't establish what happens when they stop fluoxetine. The investigators took measurements which compared the effects of a nearly complete discontinuation from sertraline or paroxetine, with just the beginnings of a withdrawal from fluoxetine.

The ostensible reason for the study - the authors make much of it - was that people on antidepressants sometimes skip a dose, or leave their pills behind. And the results of the study show that, if this happened, patients would indeed be likely to suffer more with either of the two competing brands. The basic reason for this is simple and predictable: after a break of about a week, the level of fluoxetine in the (average) body might still be around 80% of what it was. After the same length of time, users of paroxetine or sertraline would be more or less drug free.

To this extent, the study doesn't investigate what happens when people come off fluoxetine altogether - though this is a live question, discussed later. The immediate point is that Rosenbaum, Fava and colleagues (1998) have produced probably the most reliable data there is about withdrawal from either paroxetine or sertraline. Sure enough, they confirm that the MCA/CSM have underestimated the risk of withdrawal symptoms by something like 100,000% - the result of gross misinterpretation of data from 'Yellow Card' spontaneous adverse reaction reports. 

View of the CSM/MCA
" … overall symptoms due to stopping an SSRI are rare. The absolute risk of a withdrawal reaction with any of the SSRIs may be so low that differences are undetectable except through spontaneous reporting where drug exposure is high. The withdrawal symptoms observed do not appear to be severe" (Price et al., 1996)

This CSM/MCA assessment is largely based on Yellow Card reports of withdrawal symptoms with paroxetine or sertraline representing something of the order of 0.03 to 0.003 per hundred prescriptions. However, this new study shows that about 60% of patients on paroxetine and sertraline (n = 122) experienced something of an SSRI "discontinuation syndrome" and that about 30% had "substantial increases" (8 points or more) on the 28-item Hamilton Depression Rating Scale. The authors described this as a "breakthrough of depressive symptoms" or "a relapse in depression", but on the strength of evidence which suggested anything but. Their evidence clearly points to withdrawal symptoms producing depression - rather than anything to do with the unmasking of some depression that had been lurking there all the time:

"The withdrawal of antidepressants can produce changes in mood, appetite and sleep that are apt to be incorrectly misinterpreted as indicating a depressive relapse ... The probability of depressive relapse is low in the days and weeks after the discontinuation of antidepressants, and the cumulative probability of relapse increases as a function of time when the patient is medication free ... In contrast the frequency of antidepressant withdrawal symptoms is high in the first 2 to 14 days following the last dose." (Dilsaver, 1989)

"It is unlikely that reports of withdrawal reactions represent recurrence of depressive illness rather than ADR because of the short time interval between stopping the drug and the onset of the reaction, and the nature of the symptoms which were unlike those associated with depressive illness" (Price et al, 1996)

True, the authors concede that their study "cannot address the question whether within each patient the constellation of symptoms experienced during the treatment interruption reproduces the patient's original depressive symptoms" - not that this stopped them from assuming that the "re-emergence of depressive symptoms" was what it was really all about. They could, of course, have shown that this was iatrogenic depression with a different study design. If the study had involved patients kept off their drugs for a whole month, the investigators would no doubt have found that those who came off fluoxetine got progressively worse (as the drug increasingly left the body), while those on the comparator drugs tended to get better (as the withdrawal symptoms diminished in time).

There was another clue to what was really happening, when patients went back on active drug ("restabilization"). The withdrawal symptoms experienced by patients on sertraline and paroxetine cleared so fast that, "at the end of restabilization, the mean number of DESS (discontinuation-emergent signs and symptoms) was significantly higher in the fluoxetine-treated than in the paroxetine treated patients".

The question of what happens to people who come off Prozac therefore remains open.

 

3. Slightly withdrawing from fluoxetine   Let us now put aside the central point in this paper (Rosenbaum, Fava et al., 1998). The point the authors want to make is strongly supported by the evidence and well taken: someone on an antidepressant who misses a few doses seems much less likely to suffer withdrawal effects with fluoxetine (Prozac) than other SSRIs. But what does this study otherwise say about susceptibility to withdrawal effects when stopping fluoxetine?

The short answer is not much, yet just enough. Although, after a week, the patients taken off fluoxetine would still have had high levels of drug in the body, the study does give some evidence of withdrawal effects, if read between the graphics, tables and lines. The study shows that relatively few people experienced withdrawal effects after fluoxetine was withdrawn - but it still gives good evidence both of the tip of some iceberg, and also the great variability of individual response.

No one has any really clear idea why some people get bad drug withdrawal symptoms, others none - though, over the years, some pretty odious comparisons have been made between them. The usual inference seems to have been that, if the majority don't suffer, then the rest are abnormal. It used to be called lack of moral fibre; now it is usually described in terms of "dependence-prone personality" or resourcelessness. These have to pretty sweeping assumptions in the light of the failure to analyse how individual metabolic differences may affect experience with drug withdrawal. It is well recognised that there are huge differences, from one person to another, in concentrations of drug in the body, and in rates of drug elimination. What isn't really known is how this influences what happens when drugs are withdrawn.

In this particular study (Rosenbaum, Fava et al., 1998), nine out of 63 (14%) patients starting withdrawal from fluoxetine experienced a "discontinuation syndrome" (defined as an increase of four or more DESS events during the off-drug period). Also four patients (6%) became markedly depressed (defined as an increase of 8 points of more on a 28-point (HDRS) scale. The numbers are small but, given that this was only the beginning of a withdrawal, they clearly show something of a problem. This evidence suggests that:

The aggregation of data in the reported results means one cannot look in detail at the experiences of the individual patients who suffered, though there is some indication of the problems they experienced in the tabulation of reported DESS (discontinuation effects). Seven symptoms were reported by at least 10 per cent of patients after beginning withdrawal from fluoxetine; most indicated psychic distress:

Symptoms (DESS) and percentage of patients reporting them after stopping fluoxetine (n = 63)
Worsened mood 22
Irritability 17
Agitation 16
Fatigue 16
Confusion 14
Headache 14
Emotional lability 13

Such results seem at least to demand further investigation but, in the event, they were pretty much dimissed. Long before the main conclusions of this study had been reported in another Lilly-sponsored paper (Zajecka et al., 1998), albeit attributed to Blomgren et al., 1997. The conclusion was that fluoxetine (Prozac) was in the clear:

" … abrupt discontinuation of paroxetine and sertraline, but not fluoxetine, was associated with increased numbers of both spontaneously reported and solicited adverse events as well as increased (depression scores)"

 

4. Now you see it; now you don't

The study reported in June 1998 by Zajecka et al (including Rosenbaum) was based on the hypothesis "that the long half-life of fluoxetine would be protective" after drug discontinuation. The argument is attractive: if the active drug left the body very slowly, it would mimic the effects of tapering - ie progressively reducing the dose over several weeks before finally stopping. On the other hand, there is no shortage of evidence that some patients get bad withdrawal symptoms even when drug dosage is slowly reduced.

This second trial involved a group of about 400 patients, 12 weeks on fluoxetine (20mg/day) and responding well. Three-quarters continued taking fluoxetine; the others (n = 96) were abruptly switched to placebo. Patients were seen at intervals of 1, 2, 4 and 6 weeks later. What happened?

"Mild, self-limited lightheadedness or dizziness occurred in a small percentage of patients who discontinued fluoxetine treatment but was of little clinical significance. No cluster of symptoms suggestive of a discontinuation syndrome was observed. Abrupt discontinuation of fluoxetine treatment was well tolerated and did not seem to be associated with significant clinical risk"

So, one minute there is clear evidence of the early onset of a fluoxetine discontinuation syndrome (Rosenbaum, Fava et al., 1998, nee Blomgren et al., 1997); next minute, it has all but disappeared (Zajecka … Rosenbaum et al., 1998). This extraordinary feat was accomplished by using two quite different trial designs.

The whole point of the first trial was to detect and scrutinise withdrawal symptoms - but only for a week. The one-week cut-off slanted the trial in favour of Prozac, 'positioning' it ahead of its two main rivals, Paxil/Seroxat and Zoloft/Lustral. At the same time, in order to rise above the competition, the design built in some sharp instruments for measuring withdrawal events during that week. The authors even developed a specially designed instrument, the "DESS checklist", for recording withdrawal events.

However, with the second trial: every time a withdrawal reaction was recorded, it would help to disprove the hypothesis - and provide unwanted results. Indeed, the trial methodology is sloppy compared with the first. Much less effort was made to assess the nature and extent of problems on withdrawal, and correspondingly gratifying results were obtained. The data in the table below indicates how sloppy it was. 

The trial by Zajecka et al (1998) used a soft selection of patients: all had been on the lowest dose of fluoxetine, for only a short time (12 weeks), also the investigators knew that everyone was on fluoxetine (or not). But the main difference between the two trials was to do with how withdrawal effects were investigated. In short, in this trial, it was left to patients to complain they had they had been feeling bad, in response to some unspecific question, like: "How have things been since I last saw you? How has the treatment been working? - and perhaps just a hint of "Are you happy with the pills?".

There is a profound difference between this method of eliciting information about withdrawal effects - and the searching enquiries made in the first, comparative trial. The difference is between what works and what doesn't. When patients are closely and systematically questioned, using checklists to record their response, they are much more likely to respond, and to reveal symptoms of psychic distress.

Fortunately, Rosenbaum, Fava, et al., (1998) give dramatic evidence of this. Determined to reveal acute withdrawal reactions, they adopted a belt-and-braces approach - using both research methods to elicit information about withdrawal effects. They found patients volunteered only a fraction of what was really going on:

ADVERSE EFFECTS REPORTED BY 10% OR MORE OF PATIENTS AFTER ONE WEEK INTERRUPTION OF FLUOXETINE TREATMENT

[a] reported by patients spontaneously :

headache (16%)

 

[b] reported by patients in response to clinicians with checklists:

worsened mood (22%);
irritability (17%)
agitation (16%)
fatigue (16%)
headache (16%)
confusion (14%)
emotional lability (13%)

The full extent of the difference can be seen by comparing what happened when the two different research methods were used to record withdrawal effects with the two comparator drugs.

Percentage of patients reporting adverse withdrawal effects, after a one week interruption of treatment with either sertraline (n = 63) or paroxetine (n = 59), when [a] spontaneously reported by 10% or more of patients, or [b] reported to investigators bearing checklists:
SYMPTOM SERTRALINE [a] reported spontaneously SERTRALINE [b] elicited with checklist PAROXETINE [a] reported spontaneously PAROXETINE [b] elicited with checklist
Worsened mood  

28

   45
Irritability   38   35
Agitation   37   31
Dizziness 18 29 29 50
Confusion   23   42
Headache 18 31 17 34
Nervousness 18 31 16 34
Crying   26   40
Fatigue   23 11 1 32
Emotional lability   31   26
Trouble sleeping   22 19 2 39
Dreaming   25   37
Anger   28   29
Nausea 11 14 29 40
Amnesia   17   24
Sweating   17   24
Depersonalisation   17   21
Muscle aches   14   23
Unsteady gait   15   23
Panic   15   21
Sore eyes   14   15
Diarrhea   6 11 24
Shaking   11   21
Muscle tension   14   11
Chills   11   18

1 reported as 'asthenia' 2 reported as 'insomnia' (19%) and 'abnormal dreams' (16%)

The authors (Zajecka et al., 1998) further tweaked the interpretation by counting only 'new' symptoms as withdrawal effects. Thus the 37/96 (39%) who dropped out of the placebo group because of psychic distress were explained away in terms of 'loss of efficacy', and counted as cases of 'relapse'. On this basis, the authors concluded that fluoxetine did indeed have a protective effect.

The Eli Lilly Corporation was obviously impressed with this conclusion. Well before the paper was accepted for publication, the company began to promote this carefully-crafted, sugar-coated version of the punch-line: "fluoxetine appears to lack clinically significant worsened or new symptoms in the 6 weeks after discontinuation". (Michelson D, Abrupt Discontinuation of Fluoxetine: A Randomised, Placebo-Controlled Study, pp 544-545. In Thomson C, Op. Cit., 1998)

It is easy to see why patients who get bad withdrawal symptoms with Prozac might find it very hard to convince their doctor that they were right and Eli Lilly wrong.

 

5. Prozac for ever? From Eli Lilly's perspective, two hurdles had been cleared. Paroxetine and sertraline fell at the first (Rosenbaum, Fava et al., 1998), and Prozac sailed over the second (Zajecka et al., 1998). Having demonstrated, one way and another, that fluoxetine didn't produce withdrawal symptoms, it remained to be shown that patients would benefit from continuing treatment - ie that Prozac was significantly superior to placebo over long periods of time.

This was the last hurdle, and the focus in the third paper from the Lilly-sponsored independent team. The team's findings would be attributed to Reimherr, Amsterdam, Quitkin et al., 1998 - but co-authors included Drs. Rosenbaum, Fava, Zajecka and Michelson, as well.

Under new leadership, the team now reported the results of extended studies on the same pool of 400 patients - in which different groups were exposed to progressively longer treatments with fluoxetine, before having it withdrawn. The team presented its conclusions as if a building block for some future 'treatment guidelines':

"Patients treated with fluoxetine for 12 weeks whose depressive symptoms remit should continue treatment with fluoxetine for at least an additional 26 weeks to minimise the risk of relapse." (Reimherr et al., 1998).

The team had set out to illustrate what they believed to be the "protective effect" of fluoxetine. They sought to consolidate the evidence that (unlike paroxetine or sertraline), fluoxetine was kind to the head, both when taken into the body and when taken out. Notwithstanding the sleight of scientific method, nor the logical improbability of the message, it came down to this: take it or leave it, fluoxetine is best. The idea was to get doctors to understand that, if some of their patients seemed to get depressed again after stopping Prozac, it would be safer to assume the benefits of continued treatment with it.

The irony is that the three trials contain the seeds of their own undoing. Whatever they conclude, they also provide the best evidence there is for believing this to be a potentially dangerous assumption to make:

First, they show that depression is a hallmark symptom of SSRI withdrawal. Symptoms of depression predominated after withdrawal from all three drugs tested, and this was verified on gold standard scales (Rosenbaum et al., 1998).

Secondly, they show that about as many people on fluoxetine got depressed as with the other two, following withdrawal, (Zajecka et al., 1998). The essential difference between the two trials was that paroxetine and sertraline patients got depressed within 5 - 8 days after drug withdrawal whereas, with fluoxetine, it took about as many weeks. Around one in three patients became markedly depressed after withdrawal from all three drugs - but this happened more slowly with fluoxetine and was less plain to see.

Thirdly, depression after withdrawal from fluoxetine can be seen to be linked to reducing levels of drug in the body (Reimherr et al., 1998). The raw data in this study showed that, during 14-weeks following withdrawal from fluoxetine, dropouts due to depression (labelled 'lack of efficacy') were about three times higher in the first seven week period than in the second.

It is also worth noting that the Reimherr team convincingly failed to demonstrate that any "protective effect" of fluoxetine was maintained. The difference between so-called "relapse rates" for fluoxetine and placebo showed diminishing (and eventually insignificant) advantage with the passing of time. This may partly explain how the fourth paper came into being.

This paper describes a 'data-dredging' exercise, in which Drs Quitkin, Amsterdam, Fava, Reimherr, and Rosenbaum were all involved (Stewart et al., 1998). The exercise involved notionally dividing the same 400 patients into two sub-groups, then re-analysing the data. On the strength of this new analysis, the team claimed that "true drug responders" did better on fluoxetine, not just for 26 weeks, but for a full year. In this analysis, the distinction was made between the so-called "true drug responders", and patients who showed a marked placebo response.

The distinction was made on two measures. First, how soon after starting drug treatment did the patient's depression seem to clear? And secondly: once depressive symptoms had abated, did they seem to fluctuate? The "true drug" responder was thought to be the patient whose depression did not abate in the first two weeks of treatment, but whose improvement thereafter was consistently maintained. Anything else was considered a response to non-specific factors, rather than to the drug.

Implicitly, the Stewart team was claiming that patients who were genuinely depressed needed to stay on Prozac longest. A less remarkable conclusion would be that patients who respond more strongly to the drug also find it harder to quit.

The wider point - central in The Antidepressant Web - is that the long-term efficacy of antidepressants over placebo seems an illusion. It has been inferred from trials in which the investigators have not doubted that depression following drug discontinuation invariably points not to withdrawal symptoms, but relapse.

If you still believe this, you may have missed something. If not, our mailboxes are open: Is this a true and fair view? Thoughtful comments on this would be welcome, ideas for discussion too.

 

6. Questions and topics for discussion

I am still collecting my thoughts, but this feels to me like something that is going to run and run. Apart from providing (albeit inadvertently) what I can only interpret as solid evidence of the risk of dependence with SSRIs, the papers reviewed raise many questions about the conduct of business, medicine, science and public life.

It is going to take time to unravel all this, and it is easy to imagine a discussion section longer than the review. To contain it, matters arising will be indexed or summarised below, but the issues will be mainly discussed on other parts of the ADWEB site. Stuff posted to the DISCUSSION section, in particular, should be regarded as draft text - musings, subject to amendment and verification.

This list will grow. In the meantime, please CLICK on any title in blue

DEPENDENCE RISK WITH SSRIs

Long-term efficacy of paroxetine - label claim unjustified
MCA & CSM misleading estimates of risk should be withdrawn
SSRI dependence and experience with benzodiazepines (Discussion, 5 Dec 1998)

 

AUTHORSHIP - AND RESPONSIBILITIES OF INDEPENDENT RESEARCHERS
Massachusetts General Hospital ethical guidelines

 

CORPORATE SPONSORSHIP AND CONFLICTS OF INTEREST

 

 

To be continued ...

Author Base No 1 No 2 No 3 No 4
Amsterdam J University of Pennsylvania Y Y Y
Ascroft RC Eli Lilly Corporation Y
Beasley C Eli Lilly Corporation Y Y Y
Fava M Massachusetts General Hospital, Boston Y Y Y
Fawcett J Rush-Presbyterian St Luke's Medical Center Chicago Y Y
Hoog SL Eli Lilly Corporation Y
Krebs WB Eli Lilly Corporation Y
McGrath PJ New York State Psychiatric Institute Y
Michelson D Eli Lilly Corporation Y Y
Quitkin FM New York State Psychiatric Institute Y Y Y
Reimherr F University of Utah, Salt Lake City Y Y Y
Roback P Colorado State University Department of Statistics Y Y
Rosenbaum J Massachusetts General Hospital, Boston Y Y Y Y
Stewart JW New York State Psychiatric Institute Y
Sundell BS Not known, probably Eli Lilly Corporation Y
Zajecka J Rush-Presbyterian St Luke's Medical Center Chicago Y Y

 

Contents page
What's New?