Social Science & Medicine 55 (2002) 1671–1690

Transnational industrial power, the medical profession and theregulatory state: adverse drug reactions and the crisis over the

safety of Halcion in the Netherlands and the UK

John Abraham*

School of Social Sciences, Centre for Research in Health and Medicine (CRHAM), Arts E Building, University of Sussex,

Falmer, Brighton BN1 9QN, UK

Abstract

Taking the controversy over the safety of the hypnotic, Halcion, in the Netherlands and the UK, as a case study, this

article examines the problems for public health associated with responses to warnings about drug hazards by regulatory

agencies, governmental expert advisers, the pharmaceutical industry and the medical profession. It is argued that

regulators and the medical profession rely too heavily on manufacturers to investigate warnings from doctors’

spontaneous reporting of adverse effects of drug products on the market. It is demonstrated that a pharmaceutical firm’s

commitment to search effectively for evidence against the safety of its own product in order to confirm doctors’ warnings

can have severe limitations. Deficiencies in the socio-institutional responses to post-market ‘early warning systems’ about

drug hazards imply that the regulatory policies of ‘early licensing’ and minimal pre-market checks for new drugs are

misconceived and threaten public health. To improve public protection from drug injury, the regulators should abandon

their conviction that compelling evidence of drug hazards are required to confirm doctors’ warning signals prior to

regulatory intervention. Instead, they should adopt a policy of measured regulatory intervention as an immediate

response to warning signals, while investigators, independent of the manufacturers, assess the significance of the signal.

r 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Halcion (triazolam); Adverse drug reactions; Medicines regulation; Post-marketing surveillance; Pharmaceutical industry;

Netherlands; UK

Adverse drug reactions (ADRs) are a major public

health problem. Lazarou, Pomeranz, and Corey (1998)

estimate that 106,000 hospital patients died from an

ADR in the US in 1994, making ADRs the fourth

leading cause of death among hospitalised patients in

the country after heart disease, cancer and stroke. Since

1969, the British and Dutch governments have estab-

lished national systems for the spontaneous reporting of

suspected ADRs (also known as adverse medical events)

to help them monitor the safety of drugs after they have

reached the market (Id.anp.a.an-Heikkila, 1985, p. 128).

These systems, such as the famous ‘yellow card system’

in the UK, are voluntary for doctors, who are asked to

report all suspected adverse events to the national

medicines regulatory agencies, while the drug manufac-

turers are required to do so by law. They are called

‘spontaneous’ to distinguish them from ‘organised’ post-

marketing surveillance in the form of epidemiological

studies with sampling.

National systems for monitoring spontaneous ADR

reporting signalled the realisation that pre-market drug

testing in animals and humans do not detect all the

important risks associated with new drugs. There is a

great deal of uncertainty about how to extrapolate from

animal studies to humans situations, and about whether

pre-market, controlled trials with patients or healthy

volunteers in a few hundred, or even a few thousand,

people can detect the risks resulting from the exposure

of hundreds of thousands of patients in ‘non-controlled’

clinical practice (Abraham, 1997; Burley & Glynne,

1985). Hence, national spontaneous reporting systems*Tel.: +44-1273-606-755; fax: +44-1273-623-572.

E-mail address: j.w.abraham@sussex.ac.uk (J. Abraham).

0277-9536/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.

PII: S 0 2 7 7 - 9 5 3 6 ( 0 2 ) 0 0 1 6 0 - 0

have become a central part of post-marketing surveil-

lance for ADRs. In particular, they were developed in

order to provide an ‘early warning system’ to regulators

and the medical profession about the risks of new drugs

(Mann, 1984, p. 656).

Regulators in the UK and the Netherlands acknowl-

edge that the extent of pre-market checks on manufac-

turers’ drug testing may be weighed up against the

strength of post-marketing surveillance (Dukes, 1979).

For example, Professor Sir William Asscher, formerly

the Chairman of the UK’s expert advisory Committee

on Safety of Medicines (CSM), stated:

One of the real problems with animal experimenta-

tion is that if a drug is toxic in animals it will never be

given to man so you never really know whether there

is actual correspondence between animal toxicology

and human toxicology, so by the time a drug is

licensed we really know very little in the case of a new

chemical entity about its possible risks. Nevertheless

it is the policy, and I hope it will continue to be the

policy, of the CSM to go for early licensing, but of

course this must mean good post-marketing surveil-

lance (Asscher, 1987, p. 8).

He further developed this view in relation to pre-

market clinical trials:

yyou really cannot judge what a drug is going to be

like at the time of licensing. You haven’t really got a

clue as to what the drug is like.yWe [the British

regulatory authorities] license early in this country

because we are well aware that you can’t really get to

know a drug until it is used the way it will be used by

doctors in the rough and tumble of clinical practice.

Clinical trials are very artificial; patients are carefully

selected, the indications are very carefully observed.

This is not so in ordinary clinical practice.ybut

don’t you start saying that the British public are

being used as guinea pigs because they benefit from

the early licensing of medicines (Personal Commu-

nication, 1994a).

On this view, the ‘early warning system’ of sponta-

neous reports is seen as a reassurance that new drugs can

be released with relatively fewer, rather than more, pre-

market checks, which would delay licensing of medi-

cines, whose properties might deliver benefits to some

patients. However, such early licensing is justified on the

assumption that the early warning system is effective.1

This approach to medicines regulation in the UK and

the Netherlands invests considerable trust in pharma-

ceutical companies, who, unlike the regulatory agencies,

organise and conduct the pre-market testing of new

drugs. As Professor Graham Dukes, formerly Head of

the Department of Pharmacotherapy in the Dutch

regulatory agency’s Central Inspectorate for Drugs,

put it:

Always present is the obligation on industry, created

by law, custom and the industry’s own declaration of

good faith and ethical behaviour, constantly to

examine and re-examine the evidence of drug safety

and to present it as objectively as possible; since so

much material is unpublished and becomes available

earlier to the manufacturer, society is heavily

dependent upon his good faith and meticulous

accuracy in these matters (Dukes, 1994, p. 34).

Thus, there are three key groups, whose responses to

post-market ADRs, determine the capacity of national

spontaneous reporting systems to protect public health

from unnecessary drug injury, namely, the regulators,

the medical profession and the pharmaceutical compa-

nies.

Numerous books in medicine and pharmacy studies

discuss the general features of post-marketing surveil-

lance and its role in overall drug safety evaluation, but

do not involve any empirical social scientific analysis of

how these key actors relate to the ‘early warnings’ in

practice (Inman, 1986; Mann, 1987). Social scientific

research has focused on the regulation of animal

toxicology in assessing the cancer risks of drugs, the

implications of post-market clinical practice for innova-

tive drug development, the relationships between phy-

sicians and pharmaceutical companies in drug

evaluation or inadequate standards in drug advertising

(Abraham, 1998; Hemminki & Pesonen, 1977; Vos,

1991; Wieringa et al., 1992). Given the dearth of social

scientific literature in the field, this article takes a case

study approach in order to define the depth and scope of

the problems associated with responses to early warn-

ings about drug hazards by regulators, companies and

the medical profession.

These actors’ handling of the ADRs associated with

the hypnotic (sleeping pill), known as Halcion (triazo-

lam), has been chosen for several reasons (see Appendix

A for chronology of key events). First, it has been a very

important drug both commercially for the manufacturer

and for public health because so many patients have

taken it. If one wishes to understand the social science of

post-marketing surveillance, then that understanding

must take account of such important cases. Second, for

various reasons, it is possible, with perseverance, to

obtain more detailed documentary data on this case

than any other in the UK and the Netherlands because

1Spontaneous reporting systems are not without their own

problems. It is estimated that only about 5–10 per cent of

ADRs are actually reported and, when reported, they can be

difficult to interpret because, unlike adverse events recorded

during controlled trials, spontaneous ADRs cannot be com-

pared with a control population, such as placebo (Id.anp.a.an-

Heikkila, 1985; Walker & Lumley, 1987).

J. Abraham / Social Science & Medicine 55 (2002) 1671–16901672

normally manufacturers and regulatory agencies man-

age to keep secret the thinking behind their responses to

drug hazards (Abraham & Sheppard, 1997). Conse-

quently, it is an extremely instructive and revealing case.

And third, it permits some informative comparisons

between various regulatory bodies in the Netherlands

and the UK.

Background

Halcion was first synthesised by the American

company, Upjohn, as a triazolobenzodiazepine in

1969, code-named U-33,030. It is a benzodiazepine, of

the same therapeutic category as alprazolam, diazepam,

flurazepam, nitrazepam and temazepam. In the early

1970s it was shown to have hypnotic, anticonvulsant

and muscle relaxant properties in animals (Kales, Bixler,

Scharf, & Russek, 1976). By 1975 a number of clinical

trials, in which Halcion was given to patients for up to a

week, had demonstrated that a daily dose of 0.5–1.0mg

was initially effective in inducing and maintaining sleep

(Rickels et al., 1975; Vogel et al., 1975). When the drug

was clinically tested in the 1970s, it became clear, not

only that it had therapeutic potential as a hypnotic, but

also a very short (metabolic) half-life—a measure of the

time taken for the drug (or its metabolites) to be cleared

(eliminated) from the body. For example, Halcion has

an elimination half-life of about 3 h compared with

nitrazepam’s half-life of about 24 h (Wheatley, 1992).

This can be significant because sleep maintenance and

next-day carry over, sometimes referred to as ‘the

hangover effect’, are related to half-life as well as

dosage. It was thought that a benzodiazepine with a

short half-life was likely to produce less hangover

effects, such as grogginess, than one with a longer

half-life (Dement, 1983).

In this respect, Halcion was thought to confer several

advantages over some other hypnotics on the market in

the late 1970s, such as nitrazepam and flurazepam.

Medical experts had become concerned that the long

half-lives of nitrazepam and flurazepam were contribut-

ing to adverse effects in the elderly, who, tending to take

longer to eliminate drugs, were particularly prone to the

risks of toxic drug accumulation. These adverse effects

might include daytime confusion, disorientation and/or

lack of coordination, potentially leading to dangerous

falls or occupational hazards. With its short half-life,

Halcion was cleared rapidly from the body and

promised to reduce the risk of toxic accumulation,

leaving little or no hangover effect the following

morning.

Halcion was first launched as a prescription-only

hypnotic for the treatment of insomnia in Belgium and

the Netherlands in 1977 and subsequently in the UK in

1978 (Royal Courts of Justice, 1994, pp. 17–19). In

November 1977, it was licensed in the Netherlands at

daily doses from 0.25 to 1.0mg, according to the

following instructions for doctors on the label (data

sheet):

The dose varies from 0.25 to 1.0mg and should be

adjusted to ensure optimum effect. The following

may be taken as a guideline.

* 0.25–0.5 mg for elderly patients and patients who

have not previously used hypnotics or tranquillisers* 0.5–1.0 mg for hospitalised patients, psychiatric

patients, chronic alcoholics and patients who have

already used other hypnotics or tranquillisers (Royal

Courts of Justice, 1994, p. 183).

Upjohn applied for an UK product licence to market

the drug in October 1976 with a recommended daily

dose of 0.5 mg. The application was considered by the

British regulatory authorities, then known as the

Medicines Division of the Department of Health, and

the expert scientific advisory body, the Committee on

Safety of Medicines (CSM). On 13 May 1977, the CSM

advised Upjohn by letter that the Committee ‘had a

mind to refuse’ the licence application because it was

particularly concerned about the therapeutic margin of

safety of Halcion and the short duration of the clinical

trials used to test it (Royal Courts of Justice, 1994,

p. 142).

At Upjohn’s request, a formal hearing before the

Committee took place on 18 May 1978 at which the

company proposed reductions in the recommended daily

dose, possibly to a maximum of 0.25 mg, rather than

0.5 mg, in response to the CSM’s concern about its

narrow margin of safety. On the duration of (clinical)

trials and especially the inadequacy of long-term data,

Upjohn asserted that:

Triazolam [Halcion] was administered chronically to

60 insomniacs for 91 days. Three hundred patients

received 1.0mg of triazolam for periods up to 42

days. In all cases the drug was well-tolerated and no

serious side effects occurred. Side effects observed

were those usually associated with a CNS [central

nervous system] depressant (Royal Courts of Justice,

1994, p. 143).

This claim was derived from several trials, known as

protocols P321, P6049, P6023 and P6047. The 60

insomniacs who took Halcion for 91 days were 37

subjects on P6049 and 23 patients on P6023. The 300

people who received 1.0mg/day of Halcion for up to 42

days were 17 healthy subjects on P321, while the others

were (supposed to be) subjects on either P6047 or P6023.

Of these 300 people, only the 17 healthy subjects on P321

took the drug for more than 15 days, while 258 of them

took Halcion for one day only—though whether the

CSM perceived the data in this way in 1978 is unknown.

J. Abraham / Social Science & Medicine 55 (2002) 1671–1690 1673

In summarising these longer-term data at the hearing

before the CSM, Upjohn elaborated:

Based upon our experience in 300 patients, it can

be concluded that a dose of 1.0mg of triazolam

[Halcion], twice the highest recommended dose,

is a safe and effective dose for inpatients. Side

effects with 1.0 mg dose in this patient population are

minimal [emphasis added].yFourteen subjects re-

ceived 1.0mg for a period of 42 days. Triazolam was

an effective hypnotic in these subjects and there was

no indication of deleterious effects on laboratory

tests or vital signs (Royal Courts of Justice, 1994, p.

144).

The CSM and the Medicines Division accepted

these reassurances at face value, although they insisted

that the maximum daily dose should be restricted to

0.25mg. Consequently, the British Licensing Authority

granted Upjohn licences for Halcion at the 0.25 and

0.125mg tablet strengths in September 1978. The

starting dose for geriatric patients was to be 0.125mg

‘to decrease the possibility of development of over-

sedation, dizziness, or impaired coordination’, though

it could be increased to 0.25 mg if necessary (Upjohn

Ltd, 1979). Thus, elderly patients in the Netherlands

were receiving up to four times the recommended

starting dose in the UK, where doctors were cautioned

against such high dosage because of the risk of over-

sedation.

Halcion was a phenomenal commercial success and

soon started to challenge nitrazepam and temazepam,

the two major competitors in the hypnotics market. In

November 1982 it gained marketing approval in the US

at a recommended daily dose of 0.5 mg (Upjohn Ltd,

1982). By 1986 it was the world’s leading hypnotic and,

in 1991, world sales of the drug amounted to $237

million—around a quarter of the two top-selling

medicines, the anti-ulcer drug, ranitidine, and the

calcium antagonist, nifedipine (Anon 1988, 1991b;

Royal Courts of Justice, 1994, p. 36). Nevertheless,

reports that serious events, such as aggression, amnesia,

anxiety, homicide, paranoia and suicide were associated

with Halcion made it one of the most controversial

drugs ever marketed in the UK and the Netherlands. On

different occasions, both the British and Dutch regula-

tory authorities have banned the drug as unsafe and, on

each occasion, Upjohn have challenged those decisions

at the highest judicial levels, embroiling the wider

medical profession and even the broadcast media in

the dispute. By examining the controversy over Halcion,

this article examines the roles of transnational industrial

power, the medical profession and the regulatory state in

responding to doctors’ post-marketing warning signals

about the safety of new drugs in the Netherlands and the

UK.

Methods and data sources

The research was conducted between May 1994 and

February 1999. It involved a systematic analysis of the

relevant literature in medical science and the pharma-

ceutical trade, including reports from Dutch doctors and

accounts of the actions of the Dutch regulatory

authorities, accessed via the MEDLINE and DIALOG

computer databases. Most of the research involved

primary documentary data, supplemented by interviews

where necessary and feasible. The main documentary

sources were: official documents produced by the British

regulatory authority and the CSM, including relevant

documents in the public domain pertaining to the their

correspondence with Upjohn, and a CSM report to the

European Union’s Committee on Proprietary Medicinal

Products (CPMP) explaining the British suspension of

Halcion; transcripts and related materials (including

details about the initial Dutch approval) from the court

regarding libel proceedings brought by Upjohn against

Ian Oswald, a Professor of Psychiatry at Edinburgh

University and the British Broadcasting Corporation

(BBC); internal Upjohn memos and communications,

including those pertaining to the Dutch drug regulatory

authorities, which entered the public domain during

litigation in the UK and the US.

The Judgement regarding the libel case brought by

Upjohn against Oswald and the BBC was a particularly

important source. On 24 January 1992 Upjohn took

legal actions for libel in the UK against the BBC and

Oswald for allegations that Upjohn had covered up

ADRs to Halcion made in a BBC Panorama television

programme a few months before, and also made in the

New York Times by Oswald. In May 1994, Justice May

found in favour of Upjohn regarding the above

allegations. In other words, the judge found that Upjohn

had not been involved in a deliberate and dishonest

conspiracy to hide the truth about Halcion’s adverse

effects, that the above allegations were unjustified libels

and awarded damages to the company.

Twenty-one individuals were interviewed (four of

them twice) with the assistance of a pre-prepared

interview schedule containing questions about the safety

and regulation of Halcion. All interviews were tape-

recorded, except on the few occasions that permission

was denied, when notes were written up several hours

later. Sixteen scientific experts were interviewed in the

UK, all of whom were current or former members of

either the CSM or the Medicines Commission, which

acts as an appellate body under the Medicines Act. This

included Professors Sir William Asscher and Sir Michael

Rawlins, the two former chairmen of the CSM during

and since the suspension of Halcion’s product licences in

the UK, and a former chair of the Medicines Commis-

sion. Also interviewed was an expert, who had been

appointed to the special ‘Panel of Persons Appointed’ by

J. Abraham / Social Science & Medicine 55 (2002) 1671–16901674

the Secretary of State to hear Upjohn’s final appeal,

under the Medicines Act, against the British regulatory

agency’s recommendation to revoke Halcion’s licences.

Without exception all were involved in giving advice on

the drug at some stage in its history.

As well as the individuals with direct connections to

the expert advisory bodies, a number of experts, who

had either exercised influence or possessed special

knowledge about Halcion were interviewed, including:

Dr. Patrick Waller, the principal internal scientific

assessor of Halcion at the British regulatory agency in

1991; Professor Ian Oswald, Halcion’s principal critic

and former Professor of Psychiatry at Edinburgh

University; and Professor Ian Hindmarch, a former

consultant to Upjohn and a Halcion supporter at

University of Surrey. Requests to interview Upjohn in

the UK or in the US, the home of the company’s

headquarters, were unsuccessful (Personal Communica-

tion, 1994b, 1995a, b). This is unfortunate and made it

impossible to question Upjohn representatives specifi-

cally about the research in this paper.

The interviewees were predominantly from the UK

because the regulatory ‘crisis’ was much more extensive

there than in the Netherlands and because the main

Dutch actors have made their positions about Halcion

clear in a number of different public fora, whereas many

of the UK experts involved adopted a very low public

profile. While recollection bias and the possibility that

respondents saw things differently with hindsight are

potential limitations of this study, the documentary data

(including court testimonies) and interview data were

confirmatory of each other in general. On the extremely

rare occasions when data sets were inconsistent, those

data were not used unless other documentary sources or

a further interview resolved the inconsistency beyond

doubt.

The entire database was reviewed, systematically

identifying the claims/stances taken by the central actors

towards the regulatory assessment of medicines in

general, and of Halcion in particular, with particular

reference to studies with healthy volunteers, controlled

clinical trials and spontaneous adverse drug reaction

reporting. Actors’ claims/stances were meticulously

related to the chronology of events known to have

occurred with Halcion as verified by both authoritative

documentary sources and between-source consensus. A

synthesis of these analyses made possible an investiga-

tion of the key social scientific factors that may influence

regulatory assessment of drug safety. The analysis was

verified by another researcher.

Early, and not so early, warning signals from doctors

Fifteen months after marketing in the Netherlands,

van der Kroef (1979), a Dutch psychiatrist, wrote to the

Lancet, stating that he had noted some forty side-effects

in twenty-five of his patients prescribed Halcion over a 2

month period. The adverse reactions he reported were

serious, including amnesia, paranoia and aggression, as

follows:

During the past nine months I have been confronted

in psychiatric practice with a syndrome which is

almost certainly induced by the benzodiazepine

triazolam (‘Halcion’). I have made a close study of

25 patients. Triazolam can produce the following

symptoms: severe malaise; depersonalisation and

derealisation; paranoid reactions; acute and chronic

anxiety; continuous fear of going insane; depression

and deterioration of existing depressions y; night-

mares; restlessness; inability to concentrate; verbal

and physical aggression; severe suicidal tendencies;

hallucinations; impulse actions; amnesia; disphagia

[difficulty in swallowing] accompanied by nasty taste,

painful tongue and mucous membranes, dry mouth,

loathing of food, rigid feeling in the throat and

emaciation up to two and a half stone; cervical pains;

headaches that are often extremely sensitive to

sound; pressure on the ears; numb and cold feeling

in fingers and toes, extending to distal parts of the

extremities; tingling feeling, muscular cramps and

paralyses; catatonically impaired motor functioning;

reading complaints and blurred vision; dysfunctional

speaking and writing; and sweating (van der Kroef,

1979).

In early July 1979, van der Kroef’s findings were

published in the Dutch Medical Journal, by which time

the Dutch regulatory authorities, the Ministry of Health

(DMOH), had received 30 reports of adverse reactions

associated with Halcion. He also appeared on Dutch

television, AVRO-TV, airing his concerns about the

drug. Later that month the DMOH sent a letter to

doctors in the Netherlands requesting them to report

any data they might have that was relevant to the safety

of Halcion. During July, the DMOH saw an enormous

increase in the number of doctor’s reports of adverse

reactions associated with the drug. Consequently, on 6

August 1979 the DMOH suspended Halcion from the

Dutch market on grounds of safety after Upjohn refused

to revise the labelling to include mention of adverse

reactions, such as extreme anxiety and fear, suicidal

tendencies, aggression and paranoia (Anon, 1985). The

Dutch regulatory authorities had received over 1000

adverse reaction case reports from about 600 prescribing

doctors concerning Halcion, ‘exceeding the number of

reports on all other drugs [filed in the Netherlands] in

that year’ (Meyboom, 1992).

As Upjohn and some members of the medical

profession suspected, it was possible that the huge

increase in adverse reports about Halcion resulted from

J. Abraham / Social Science & Medicine 55 (2002) 1671–1690 1675

the publicity received by the drug (Lasagna, 1980),

although one prominent medical expert later was of the

opinion that it had been ‘shown beyond all reasonable

doubt’ that the warning signal generated in the Nether-

lands ‘was valid’ (Inman, 1986, p. 8). Yet, even if media

coverage motivated doctors to report adverse reactions

to Halcion more than other drugs, the large number of

reports elicited pointed to the very substantial amount

of under-reporting of problems with the drug in the

absence media attention. Moreover, according to

Dukes, then a leading regulator at the DMOH, some

of the adverse reactions to Halcion were very unusual.

He found that in 63 cases the ‘entire functioning of the

patient had been deranged’, and that several hundred of

cases were ‘rock-solid’. In some cases, he found the

evidence implicating Halcion to be ‘compelling’ because

of ‘positive rechallenge’, that is, when patients stopped

taking the drug their adverse symptoms receded, but

returned on restarting the medication (Dukes, 1994, pp.

151–53).

The British authorities took no regulatory action

against Halcion at that time because they supposed that

the high level of serious adverse reports in the Nether-

lands was due to the higher doses licensed there, and due

to the adverse publicity following van der Kroef’s

television appearance. Indeed, no changes were made

to the British data sheet to inform doctors that the van

der Kroef reports had occurred. While it was true that

adverse psychiatric effects were reported more fre-

quently by patients taking the 1.0 mg dose in the

Netherlands, according to Sigelman (1989), one survey

showed that 73.7 per cent of the Dutch patients

reporting psychiatric disturbances in 1979 had taken

0.5mg or less of Halcion. During their respective first 8

years of marketing, Halcion was associated with over

five times as many such adverse reactions as temazepam.

Of course, these post-1979 adverse drug reaction data

were not known to Upjohn or regulators at the time of

van der Kroef’s reports and it is likely that they were

also unaware of the survey cited by Sigelman. Never-

theless, as Table 1 shows, even in 1979, the CSM’s

database indicated that doctors in the UK were

reporting approximately between 25 and 1000 times

more adverse psychiatric reactions associated with

Halcion than with other benzodiazepines. Moreover,

as the CSM was later to acknowledge, ‘reporting rates

for psychiatric reactions have been consistently and

substantially greater for Halcion than any other

benzodiazepine’ in the UK from 1979 to 1991, and ‘34

per cent and 53 per cent of these occurred at the low

doses of 0.125 and 0.25mg, respectively’ (Royal Courts

of Justice, 1994, p. 262), which were licensed in the UK

throughout that period.

As Halcion continued its commercial success on the

UK market, two British doctors, Morgan and Oswald

(1982), argued that the drug’s short half-life, far from

being a benefit, might be a distinct disadvantage. In a

double-blind study with 21 poor sleepers, they found

that 0.5mg/day of Halcion increased the patients’

anxiety and was associated with significantly more

‘rebound anxiety’ than a comparator benzodiazepine

with a long half-life:

Drugs used as hypnotics are the same as those used

to diminish anxiety—for example, alcohol, barbitu-

rates and benzodiazepines—and their presence leads

to adaptive changes in the central nervous system, as

if to counteract the drug. When the drug is stopped

the induced changes persist, with resultant insomnia

and anxiety. These rebound phenomena are features

of the first few weeks after stopping benzodiazepines.

The more rapidly the drug is eliminated the earlier the

rebound [emphasis added]. A measurable rebound in

sleep may occur within a single night. We presume

that the large dose [0.5 mg] of triazolam [Halcion]

each evening was rapidly metabolised and so led to

daytime rebound anxiety, in contrast to the more

familiar reduction of anxiety by the longer-persisting

loprazolam (Morgan & Oswald, 1982).

In another double-blind study Adam and Oswald

(1989) compared the 0.5mg dose of Halcion with

placebo and another benzodiazepine hypnotic, known

as lormetazepam, by studying their effects, over a period

of 45 nights, on 82 women and 38 men. They reported

that:

Halcion-takers became more anxious on self-ratings,

were judged more often to have had a bad response

by an observer, more often wrote down complaints of

distress, and suffered weight loss. After about 10 days

of regular Halcion they tended to develop panics and

depression, felt unreal, and sometimes paranoid

(Adam & Oswald, 1989, p. 115).

In August 1989, Oswald took his criticisms much

further, calling for the withdrawal of Halcion from the

market in the Lancet. According to Oswald (1989),

Halcion had been marketed following controlled trials in

which only 11 patients over the age of 40 were known to

have taken 0.25mg/day for more than 2 weeks even

though hypnotics are mostly taken by the middle-aged

or older, sometimes for many months. While there had

been three post-market double-blind controlled trials

with adequate numbers of subjects, over 40, taking the

drug for longer than 2 weeks, he claimed that these

implied that Halcion regularly caused daytime anxiety,

and concluded:

It is now 10 years since The Lancet published a letter

from a Dutch psychiatrist, Dr van der Kroef, who

described anxiety, derealisation, paranoid ideas, and

other mental changes associated with triazolam

(Halcion). yvan der Kroef was right and the

J. Abraham / Social Science & Medicine 55 (2002) 1671–16901676

Netherlands have been right to ban triazolam

[Halcion]. People who complain of poor sleep are

generally anxious people. If after 3 weeks they are

even more anxious, doctor and patient alike easily

attribute any change to the patient rather than to the

drug. It is a matter for concern that Halcion was

marketed on the basis of deficient research. It should

no longer be sold (Oswald, 1989).

Following Oswald’s call for the withdrawal of

Halcion, the CSM reconsidered the drug, but in January

1990, concluded that, although there was a signal, it was

not proven that severe adverse psychiatric effects were

more common with Halcion than with other benzodia-

zepines. The CSM requested that Upjohn conduct

further studies to allay fears about amnesia and

psychiatric side-effects (Royal Courts of Justice, 1994,

p. 229).

Industrial commitments and medical debate: Upjohn’s

response to doctors’ warnings

In May 1979, Upjohn became aware that van der

Kroef’s findings were soon to be published in the Dutch

Medical Journal. On 23 May 1979, the company’s Dutch

subsidiary made enquiries to the DMOH about the

number of reports of adverse reactions associated with

Halcion, which the Dutch regulators had received.

Within a few weeks, the DMOH informed Upjohn that

they had received 14 reports of ‘psychological symp-

toms’ associated with Halcion (Royal Courts of Justice,

1994, p. 148). In addition, Upjohn’s European Medical

Affairs Director visited van der Kroef and asked to see

details of his records, but he refused (Royal Courts of

Justice, 1994, p. 148).

Following van der Kroef’s reports, Upjohn reviewed

some of their data on Halcion in 1979 and 1980. Indeed,

in June 1982, the company wrote to doctors in the UK

claiming that:

In 1979 and 1980, Upjohn undertook a comprehen-

sive review of all available evidence to determine the

truth regarding allegations against Halcion in Hol-

land. These efforts included: a complete reanalysis of

all available safety data from clinical trials (Royal

Courts of Justice, 1994, p. 165).

However, this ‘review of all available evidence’ did not

include a reanalysis of the case report forms (CRFs) of

individual patients in Halcion trials (Royal Courts of

Justice, 1994, p. 166). Moreover, although Upjohn

reviewed a lot of their reports on clinical trials with

patients, they may not have reviewed even the reports on

trials with healthy volunteers, such as P321, which

certainly included safety data (Royal Courts of Justice,

1994, pp. 165, 166). Evidently, the company’s review,

though extensive and honest, was not as comprehensive

as it could have been.

Some medical staff at Upjohn perceived the decision

by the Dutch regulatory authorities to suspend Halcion

as very threatening. One senior scientist, who seemed to

advocate the suppression of van der Kroef’s warnings

about the safety of the drug, was particularly concerned

that the situation in the Netherlands might jeopardise

the marketing of Halcion in the UK:

The UK CSM were giving credence to the van der

Kroef data.yThey do not, for the moment, see any

problem in the UK data from the monitored release

[post-marketing ‘trials’], but they do believe van der

Kroef [emphases in original]. Thus, the Dutch disease

spreads.yI can see only one plan of action which

will stop the otherwise inevitable. The Dutch adverse

reaction ‘data’ must be shown to be false. We must

stop further publication by van der Kroef in major

journals.y(I think it need not be said that I have no

doubts about Halcion and that I feel a strong

revulsion about the recent happenings). We must

learn everything possible about van der Kroef, and

be prepared to use the evidence. It should be clear

that someone is going to get hurt and this is going to

be a long and tough battle (Kratochvil, 1979, p. 2).

This indicates that a genuine opposition to van der

Kroef’s claims and a strong commitment to the Halcion

product existed within Upjohn. However, as Justice

Table 1

UK spontaneous reports of psychiatric reactions by year for specified benzodiazepines per million prescription

Year 1979 1980 1981 1982 1983 1984 1985 1986

Triazolam [Halcion] 114 44 14 43 20 5.5 1.8 2.5

Flurazepam 0.9 0.9 0.4 1.3 0.5 0.0 2.2 0.0

Nitrazepam 0.1 0.9 0.5 0.0 0.3 0.0 0.8 0.4

Temazepam 4.1 4.5 3.7 1.9 2.2 1.7 1.3 0.4

Lorazepam 2.4 3.3 1.4 2.0 4.3 1.9 6.7 1.9

Diazepam 0.4 0.7 0.1 0.1 0.2 0.4 0.4 1.0

Source: CSM (1991) ‘confidential’ report to the EU’s committee for proprietary medicinal products (CPMP), Table 5.

J. Abraham / Social Science & Medicine 55 (2002) 1671–1690 1677

May later found, it also implies that ‘Upjohn were

considering cut-throat commercial tactics’ as a conse-

quence of that strong, albeit honest, commitment (Royal

Courts of Justice, 1994, p. 168).

On 10 September 1979, a meeting was held at Upjohn

to discuss the establishment of a ‘Halcion thought leader

conference’. The special conference of experts was to be

financed by Upjohn and held in Boston later that year.

According to an internal Upjohn memo, the objectives

of the conference were: to ‘provide the scientific basis on

which the DMOH could gracefully reverse its suspen-

sion of the Halcion Product Licence’; to ‘neutralise the

efforts of the DMOH and van der Kroef to spread the

‘‘Dutch hysteria’’ about Halcion to other countries’; and

to ‘provide ‘‘scientific reassurance’’ to regulatory agen-

cies and lay press’ (Kagan, 1979, p. 1).

Upjohn asked Dr. Frank Ayd, Clinical Professor of

Psychiatry at West Virginia to set up, and chair, the

Boston conference. There were 12 experts, including

Ayd. The experts were asked to evaluate the significance

of the van der Kroef findings based on data supplied to

them by Upjohn (BBC, 1991). Upjohn regarded van der

Kroef’s findings as aberrant and believed that his data

were not supported by their own research on the drug.

According to Ayd, the Upjohn representative at the

conference told the experts that there was no material

from research trials carried out by Upjohn indicating

that van der Kroef might be right (Ayd, 1992, p. 12).

The experts trusted the veracity of this statement and

were consequently sceptical of van der Kroef’s results.

The principal outcome of the Boston conference was a

letter to the Lancet in November 1979, based on

information from Upjohn, and signed by 10 of the 12

experts who attended. It cast doubt on the adverse

effects found by van der Kroef and questioned the

validity of his conclusions, stating that ‘studies with

doses up to 1.0mg have not shown the symptoms noted

by van der Kroef’ (Ayd et al., 1979). The letter cited

clinical trials comparing Halcion with flurazepam and

placebo, as evidence that ‘the frequency and severity of

side effects attributable to triazolam [Halcion] are equal

to or less than those attributable to flurazepam’, and

also noted that the symptoms described by van der

Kroef could have been ‘due to the size of the triazolam

[Halcion] doses authorised in the Netherlands’ (Ayd

et al., 1979). The force of this letter may partly explain

why the British regulatory authorities did not require

Upjohn to alter the 1980 or 1981 UK data sheets for

Halcion to warn doctors about the possibility of adverse

psychiatric reactions to the drug.

Reporting back on the conference to Upjohn, Dr.

Robert Purpura, one of the company’s Medical Direc-

tors and Head of Upjohn’s Psychopharmacology Unit,

believed that ‘several important suggestions came out of

the meeting and either have or will result in actions

favourable to Halcion and the Upjohn company’

(Purpura, 1979, p. 2). His interpretation was that the

conference’s doubts about the appropriateness of the

1.0 mg dose could allow the Dutch regulatory autho-

rities ‘a simple, face-saving way of allowing Halcion

back on the market in Holland’ because ‘it allows the

DMOH an explanation for the alleged adverse effects

(possibly of too high a dose) and allows the Upjohn

company a bargaining point (elimination of the 1.0mg

dose)’ (Purpura, 1979, p. 2). In November 1979,

Upjohn’s Dutch subsidiary offered to withdraw the

1.0 mg tablet. However, the DMOH wished to cancel the

approval of both the 0.5 and 1.0mg tablets, and to

approve the 0.25mg tablet back on the market only if

there was a revised label referring to the adverse events

at higher doses. Upjohn refused to accept this proposal

so Halcion remained banned in the Netherlands for

some years.

In fact, during P321, one of the longest placebo-

controlled Halcion trials conducted by Upjohn, sig-

nificantly more subjects on 1.0mg of the drug, than on

placebo, complained of anxiety, depression, hallucina-

tions, loss of mental ability and paranoia (Royal Courts

of Justice, 1994, pp. 57–61; Veldkamp, Rudzik, &

Metzler, 1973). This double-blind trial was carried out

with healthy prisoners in Michigan in 1972–1973, 30 of

whom took 1.0mg, while 16 took placebo. Unaware of

significant errors in the reporting of this trial, Upjohn

did not make it available to the experts at Boston in

1979. After reviewing P321 in 1991, Ayd concluded:

Upjohn had information, which would have sup-

ported some of van der Kroef’s criticisms, informa-

tion, which should have been disclosed to the experts

in Boston.yIf I had known then what I know now, I

would not have written and signed the [Lancet] letter

(Ayd, 1992, pp. 15, 16).

Furthermore, this should be seen in the context of

Justice May’s assessment that the credibility and stature

of Ayd’s evidence was second to none in the libel case

involving Upjohn, Oswald and the BBC. Although Ayd

was concerned that he and the other experts at the

Boston conference had been misled by Upjohn, he did

not believe that Purpura or Upjohn had been deliber-

ately misleading. While Upjohn possessed information

from P321, which Ayd and other medical experts

regarded as significant, it does not follow from this that

Upjohn appreciated or knew of its significance for those

experts at the time of the Boston conference. Addition-

ally, two of the other signatories to this Lancet letter,

Drs. Greenblatt and Gardner, continued to stand by the

letter in the evidence they gave in the libel case after the

revelations about P321. For example, Greenblatt argued

that because P321 involved a healthy prison population,

which was very different from those who took Halcion

in a therapeutic context, it was not relevant to his

J. Abraham / Social Science & Medicine 55 (2002) 1671–16901678

assessment of van der Kroef’s claims about the drug

(Royal Courts of Justice, 1994, p. 160). Nonetheless, for

Ayd, the Lancet article undermining van der Kroef’s

findings in 1979 became invalid as soon as he had access

to P321, yet it bore his prestigious and leading signature.

The warnings by Oswald and Adam were equally

unwelcome at Upjohn. These doctors first presented

their findings at a conference in Philadelphia in

September 1985, but it was not until a year later at the

Eighth European Congress on Sleep Research in

Hungary that an abstract of the paper was published,

stating:

We attribute the daytime anxiety caused by Halcion

mainly, but not wholly, to withdrawal symptoms

during each day as a consequence of its rapid

metabolism. The clinical observations of van der

Kroef and our own earlier report are confirmed

(Royal Courts of Justice, 1994, p. 270).

This research threatened to stir up the old worries

generated by the van der Kroef reports of the late 1970s

in the Netherlands. Medical scientists at Upjohn

responded by objecting to some of the methods used

by Adam and Oswald in this study, particularly the

scales used to measure the subjects’ experiences and

anxiety. However, these scales had been used in some of

Upjohn’s own clinical studies of Halcion, and had been

cited supportively by company researchers in those

studies (Royal Courts of Justice, 1994, pp. 265–270).

There is also evidence that the company strove

‘behind the scenes to discourage the publication of

Adam and Oswald’ with ‘little, if any, justification’

(Royal Courts of Justice, 1994, p. 272), though it is not

being suggested that Upjohn actually managed to

influence the editorial decisions of journals. For

example, an Upjohn document entitled, ‘Summary of

Worldwide Halcion Committee Activities, March

1986–March 1987’ noted:

Oswald paper indicated high percentage of Halcion

users have problems. So far we have been successful

in having it stopped. It is scientifically a poor paper

(Royal Courts of Justice, 1994, p. 271).

Evidently, some officials at Upjohn sought to

suppress adverse reports about Halcion. Justice May

described such conduct as ‘unacceptable’ and commen-

ted that ‘the public would not expect a pharmaceutical

company to try to suppress adverse report about their

product; the public would expect such reports to be

properly and openly addressed by whatever means was

appropriate to the subject matter’ (Royal Courts of

Justice, 1994, p. 272).

Meanwhile, Upjohn had appealed to the Dutch

Council of State to get the DMOH’s decision to ban

the drug in the Netherlands reversed. In 1985, Upjohn

were informed that the Crown Court, acting on the

recommendations of the Dutch Council of State,

decided to cancel the decision taken by the Dutch drug

regulatory authorities to withdraw the product licence.

However, the Crown Court also held that the approval

of Halcion tablets should not be reinstated under

the labelling in force in 1977 and suggested that the

company should apply to reregister the product in the

0.25 and 0.5 mg tablet forms (Anon, 1985).

Protocol 321: the courts as a ‘late’ warning signal?

The significance of P321 was discovered by lawyers

acting for Ilo Grundberg, who, in 1988, embarked on a

$21 million civil suit seeking damages against Upjohn

for allegedly failing to warn the public adequately about

Halcion’s adverse reactions. After taking Halcion,

Grundberg had shot her 83-year-old mother in Utah

eight times in the head, but charges against her were

dropped because the criminal court determined that she

was ‘involuntarily intoxicated with Halcion’ when she

killed her mother. Upjohn settled Grundberg’s civil suit

out of court, but not before the lawyers on both sides

discovered that the report on P321 submitted to

regulatory agencies in the UK contained wholesale

omissions of medical events—some 330 medical events

regarding the subjects taking the 1.0 mg dose (CSM,

1991, p. 3). Only 67 per cent of Halcion-related and 79

per cent of placebo-related medical events were con-

tained in the report (Royal Courts of Justice, 1994, p.

63). It is not being suggested in any way whatsoever that

Upjohn covered up these events. Rather, as Justice May

concluded, the report was ‘the product of a time when

standards and techniques were nothing like as rigorous

as they are today’ and ‘obviously the product of shoddy

work even by its own contemporary standards’ (Royal

Courts of Justice, 1994, p. 62). Indeed, according to

Upjohn, these omissions also went unnoticed to anyone

in the company until their lawyers discovered them

during the Grundberg litigation (Royal Courts of

Justice, 1994, p. 56). Writing to doctors in a letter to

the Lancet, Keith Krzywicki, the Managing Director of

Upjohn UK Ltd, referred to these omissions as ‘a

clerical error’ (Krzywicki, 1991).

Oswald was asked to analyse P321 by Grundberg’s

lawyers. After completing his analysis, he argued that

seven subjects taking Halcion developed symptoms of

paranoia, whereas the company, mistakenly, reported

only two such cases (Royal Courts of Justice, 1994, pp.

57, 58). According to Oswald, this gave the erroneous

impression that Halcion was no more likely than a

placebo to cause paranoia (Royal Courts of Justice,

1994, pp. 57, 58). Only one, or possibly two, subjects

taking placebo reported feelings of paranoia (Royal

Courts of Justice, 1994, pp. 57, 58). For nine subjects

J. Abraham / Social Science & Medicine 55 (2002) 1671–1690 1679

taking the drug, the subject responses recorded on the

case report forms and questionnaires were: (1) ‘nervous-

ness/restlessness’; (2) ‘paranoia, loss of memory’; (3)

‘paranoia, loss of mental ability, nervousness/restless-

ness’; (4) ‘paranoia’; (5) ‘nervousness/restlessness, weird

feeling in the head, amnesia’; (6) ‘paranoia, severe

depression, bizarre thoughts’; (7) ‘depression, paranoia,

severe nervousness/restlessness’; (8) ‘hallucinations,

blurred vision, ringing ears’; and (9) ‘paranoia, loss of

memory’ (Royal Courts of Justice, 1994, pp. 57–61).

Consequently, Oswald believed that a major finding of

P321 was that Halcion caused mental illness (BBC, 1991;

Royal Courts of Justice, 1994, pp. 23–26).

Even after the revelations about the errors and

omissions in P321, Dr. Oster, the principal clinical

investigator contracted by Upjohn to conduct the trial,

did not regard these subjects’ reports as serious (Royal

Courts of Justice, 1994, pp. 53, 54, 71). He was not in the

full employment of Upjohn, but he had worked for the

company as a consultant physician on clinical research

projects since 1957, mostly at the Jackson Prison clinic

(Royal Courts of Justice, 1994, p. 54). According to

Oster, he did not disregard subjects’ reports of paranoia

and depression during the trial, but he did not think that

they were indications of psychiatric disorder. He had not

had reports of paranoia in other drug trials, but he felt

that psychologically there was nothing abnormal about

these reports, given the prison environment (Royal

Courts of Justice, 1994, p. 71). Mr. Veldkamp, Upjohn’s

main author of the company’s 1973 report on P321, was

also aware that there had been reports of paranoia, but

he did not think they were unexpected, despite their

absence in previous drug studies with which he had been

concerned (Royal Courts of Justice, 1994, p. 59). Indeed,

this perspective is reflected in the one and only

paragraph within the report, which mentions paranoia:

Approximately half the subjects who had been on

U-33,030 [Halcion] considered the drug as non-

acceptable and would not be willing to take it again.

One reason that may account for the high percentage

of non-acceptance among these prison subjects is that

most of these men were not insomniacs and were

receiving a relatively large dose of U-33,030. This

caused them to experience some depression or

drowsiness and a feeling of some loss of mental

ability.yThis gave them a feeling of being less able

to defend themselves if some trouble developed and

was the reason for some subjects reporting feelings of

paranoia (Veldkamp et al., 1973, p. 8).

Similarly, even with the benefit of hindsight in the

1990s, Hindmarsh disputed that P321 showed genuine,

clinical incidences of paranoia. He argued that, because

the subjects went back to the general prison at

night, having spent their days in the special clinic,

they returned to their cells with ‘a powerful sedative

on board’, which made them feel less alert and less able

to cope with the hazards of their immediate

environment:

Just because someone says they are feeling paranoid

does not mean they’re suffering from paranoia—in

fact, quite the reverse. Within Jackson Prison there

were something like four homosexual rapes every

month, there were murders, three guards had been

killed, the prisoners were always robbing each

otherySaying, ‘‘man, I feel paranoid’’ was just part

of the cultural argot of the place (Personal Commu-

nication 1994c).

Thus, Oster, Veldkamp and Hindmarsh took the view

that the adverse psychiatric experiences of the subjects

were caused by them being in prison, rather than by

Halcion. The implication of this is that Upjohn’s design

of P321 was flawed as, on this view, significant

differences between the test drug and placebo could

not be attributed to the drug because of ‘noise’—the

prison environment. Yet, as noted above, at a hearing

before the CSM in 1978, Upjohn explicitly cited safety

data from P321 as evidence in support of the claim that

Halcion was ‘well-tolerated’ at the 1.0mg dose.

The Grundberg revelations and the re-regulation of

Halcion

Ironically, as Halcion was about to face its biggest

regulatory challenge in the UK, in 1990, the Dutch

regulatory authorities reapproved the drug for treatment

of insomnia at the doses licensed in the UK, 0.125 and

0.25 mg, recommending the lowest possible effective

dose for the shortest possible duration. The ‘reapproval’

resulted, in part, from the fact that the State Council had

ruled that some of the views of the Dutch regulatory

authorities in suspending the product’s licences were

inappropriate (Anon, 1990). However, it should be

understood that just prior to banning Halcion in the

Netherlands, the Dutch regulatory authorities offered to

allow the 0.25mg tablet to remain on the market with a

revised label recommending the reduced dose of 0.25mg

and warning about additional ADRs, but Upjohn

refused to accept that offer (Royal Courts of Justice,

1994, pp. 18, 150, 151). Thus, the ‘reapproval’ was

substantially what the company had been offered 10

years before. While the assessment by the Dutch

regulatory authorities in 1979 was made before US data

on Halcion became available, it may be noted that by

1987 the US drug regulatory authorities required that

the American label for Halcion should recommend the

reduced daily dose of 0.25mg, rather than 0.5mg, for

the typical adult (Upjohn Ltd, 1987).

J. Abraham / Social Science & Medicine 55 (2002) 1671–16901680

Upjohn informed the US Food and Drug Adminis-

tration (FDA) of ‘transcription error’ in P321 on 28

June 1991 (Royal Courts of Justice, 1994, p. 242).

However, the British regulatory authorities, now known

as the Medicines Control Agency (MCA), found out

about the omissions in P321 by reading a pharmaceu-

tical industry newsletter, which was reporting on the

Grundberg case on 3 July 1991 (Anon, 1991a). Within 2

weeks, the MCA, asked Upjohn to provide previously

undisclosed data on Halcion, which had emerged in the

Grundberg case, together with a ‘summary of the

implications for [Halcion] product licences’ (Waller,

1994, p. 24). By 12 August 1991, Upjohn had still not

told the MCA about the deficiencies in that drug trial.

Later that month, after further prompting from the

MCA, the company admitted that P321 contained

omissions of adverse medical events due to ‘transcrip-

tion error’ (Royal Courts of Justice, 1994, pp. 236, 237),

but claimed that the omissions did not alter the risk-

benefit assessment of the drug at the doses licensed in the

UK.

The news of errors in P321 provoked the MCA and

the CSM to conduct a review not only of P321, together

with all its case report forms (CRFs) for the first time,

but also the controlled clinical trials submitted in

support of Halcion’s UK product licences and all the

post-marketing spontaneous reports of adverse effects

associated with the drug in the UK. In preparation for

the review, the MCA sought more detailed reports about

clinical studies with Halcion (Royal Courts of Justice,

1994, p. 198). This revealed further problems with the

extent of trust in adequate flows of information about

drug safety from the manufacturer to the UK regulators.

At a meeting with the MCA on 6 September 1991, Dr.

Robert Straw, Upjohn’s Director of Project Manage-

ment for Halcion, provided the agency with copies of the

reports on various Halcion trials, including P6415

without any mention of the fact that the trial was found

to be a fabrication by the FDA (Royal Courts of Justice,

1994, p. 197). On 3 March 1984, the FDA had informed

Upjohn that P6415 was unreliable, and that its clinical

investigator had been disqualified (Royal Courts of

Justice, 1994, p. 196). Nor did Upjohn inform the MCA

about the FDA’s disqualification of one of the clinical

investigators for P6045 and P6041, two major placebo-

controlled trials submitted in support of the initial UK

product licence for Halcion (Royal Courts of Justice,

1994, p. 197). As early as 24 November 1982, the FDA

told Upjohn that the clinical investigator was disquali-

fied for submitting false data, and, consequently, that all

his Halcion data were unreliable (Royal Courts of

Justice, 1994, p. 190).

Notwithstanding the MCA’s limited knowledge about

the unreliability of some important Halcion trials with

patients, Waller, the principal medical assessor for

Halcion at the MCA, presented his analysis to the

CSM in September 1991. It involved three crucial data

sets: P321; controlled clinical trials; and spontaneous

reports of adverse medical events. In reviewing the

Halcion case, the CSM found that the ‘new’ complete

P321 results ‘showed that psychiatric side effects

occurred in 14/30 of subjects who took 1.0mg Halcion

compared to 3/20 who took placebo’ (CSM, 1991, p. 2).

The Committee deduced that ‘there now appears to be a

clear difference between Halcion and placebo in

occurrence of psychiatric symptoms and this study

[P321] estimates that they occur at a very high frequency

of about 50 per cent of patients’ (CSM, 1991, p. 4).

The CSM’s review revealed that initial approval of

Halcion in the UK had been supported by 6 controlled

clinical trials longer than 14 days, two comparing

Halcion with placebo and four comparing it with

another hypnotic, flurazepam. According to the CSM,

almost 20 per cent of the patients taking Halcion in these

studies withdrew—about three times the rate for placebo

or flurazepam. This information was available to the

British regulatory authorities in 1978, but it seems that

they did not pursue the reasons for these withdrawals

thoroughly prior to approval:

In the original application the reasons for withdrawal

were not usually provided on a case-by-case basis and

it was therefore uncertain what side-effects caused

these patients to withdraw and when they first

occurred. Upjohn have now provided these data

and the reasons for an excess of withdrawals due to

side-effects in Halcion-treated patients compared to

those treated with placebo or flurazepam have been

assessed (CSM, 1991, p. 5).

Had the British regulatory authorities demanded these

data prior to approval in 1978, the CSM could have

conducted a similar review then and found that a

substantial number of Halcion withdrawals were due to

potentially serious psychiatric adverse effects. The

CSM’s analysis in 1991 implied that there were 18 such

withdrawals among patients taking Halcion, only one

among patients taking flurazepam and none among the

placebo groups. In short, in controlled clinical studies of

greater than 14 days, ‘the frequency of withdrawal due

to psychiatric side effects was 9.9 per cent on 0.5mg

Halcion, 1.9 per cent on 30 mg flurazepam and 0.5 per

cent on placebo’ (CSM, 1991, p. 2). According to the

CSM, ‘these findings confirm that drug-induced psy-

chiatric side-effects similar to those seen in P321 occur at

a dose of 0.5mg Halcion and in a population of out-

patient insomniacs’ (CSM, 1991, p. 6).

Taking these findings from controlled trials, together

with the spontaneous post-marketing data, which

implied that ‘reporting rates for psychiatric reactions

have been consistently and substantially greater for

J. Abraham / Social Science & Medicine 55 (2002) 1671–1690 1681

Halcion than any other benzodiazepine’ (CSM, 1991,

p. 2), the CSM concluded:

(1) There is a clear causal association between Halcion

and serious psychiatric adverse reactions.

(2) The frequency of serious psychiatric adverse reac-

tions is significantly greater than with comparator

hypnotics.

(3) There is an insufficient margin of safety in relation

to dose.

(4) The risks of treatment with Halcion outweigh the

benefits (CSM, 1991, p. 2).

Despite Upjohn’s claims that the revelations about

P321 did not change the risk-benefit assessment of

Halcion, the CSM advised the MCA to suspend the drug

from the market. On 2 October 1991, the MCA, acting

on behalf of the UK Licensing Authority unilaterally

suspended Halcion’s licences, without consultation with

the EU’s drug regulatory body, the CPMP (Asscher,

1991). By contrast, the Dutch regulatory authorities now

decided to take their lead from the CPMP, who

recommended that Halcion should remain on the

market at low doses and under strict conditions.

Upjohn’s appeal to the CSM

As is their right under the 1968 UK Medicines Act,

Upjohn challenged the MCA’s decision to suspend

Halcion’s licences, at a formal hearing before the CSM

on 3 December 1991. The company contended that the

nature of the P321 data and the environment in which it

was collected made it impossible ‘to arrive at any

conclusion about psychiatric side effects in relation to

triazolam [Halcion] whether potentially serious or

otherwise’ (Royal Courts of Justice, 1994, p. 60).

Furthermore, argued Upjohn, the revelations about

subjects taking 1.0 mg Halcion during P321 did not

affect the safety profile of the drug at the lower doses of

0.25 and 0.125 mg, which had been licensed in the UK

(Wood, 1991, p. 4). Thus, Upjohn now sought to disown

the validity and relevance of the design of their own trial

with healthy volunteers. However, according to Oster, a

medical/ethical review committee approved the design of

P321 when it was first instigated (Royal Courts of

Justice, 1994, p. 54).

Upjohn also sought to disown the design of their own

clinical trials with flurazepam—trials which the com-

pany had previously submitted as valid in order to

obtain marketing approval from regulatory authorities.

The company argued that trials comparing 30mg

flurazepam with 0.5mg Halcion were invalid because,

they claimed, the equiopotent (properly effective) doses

for Halcion and flurazepam were 0.25 and 30mg,

respectively (FDA, 1994, p. 8). The implication of this

argument is that any comparison of 0.5mg Halcion with

30 mg flurazepam was unfair to Halcion because it

overstated the adverse effects of Halcion relative to

flurazepam. Yet 10 of the 16 major clinical trials

submitted to the British regulatory authorities in

support of approval in 1978 involved comparisons of

0.5 mg Halcion with 30 mg flurazepam. Moreover, in

Upjohn’s 1973 report on P321, 0.5mg Halcion was

regarded as equipotent to 30 mg flurazepam, and in

P6049 Upjohn actually stated that 0.5mg Halcion was

equal to 30 mg flurazepam in its effect on sleep (Royal

Courts of Justice, 1994, pp. 187, 188). Notably, prior to

the MCA’s decision to suspend Halcion’s licences,

Upjohn had not told the British regulatory authorities

that they believed that over half of the key clinical trials

in the UK product licence application involved invalid

comparisons because 0.5mg Halcion was not equipotent

to 30 mg flurazepam (Royal Courts of Justice, 1994, p.

187).

As for the spontaneous post-marketing reports of

adverse medical events, the company concentrated on

the general limitations of such data. Specifically, Upjohn

argued that such data were unsuitable for quantitative

between-drug comparisons and were greatly affected by

non-clinical factors such as publicity and reporting rates

of prescribing doctors (Wood, 1991, pp. 8, 9). According

to Upjohn, ‘spontaneous reporting databases can only

generate and cannot confirm signals of possible drug

hazards’ (Wood, 1991, p. 9).

In hearing Upjohn’s appeal, the CSM were aware of

Waller’s analysis of the Halcion trials. He forthrightly

rebutted the argument that the design of P321 was

invalid

This study was well-designed. It had a control group

and random allocation to placebo or triazolam

[Halcion], which is a highly scientific method. It

had what are known as double-blind conditions

whereby patients and the evaluating physicians didn’t

know which treatment they were on. It had appro-

priate ‘exclusion criteria’: subjects were not allowed

to be entered if they had significant mental disorder

and so forth. So I believe it was a valid trial (Royal

Courts of Justice, 1994, p. 60).

Similarly, he argued that it had not been shown that

comparing 0.5mg Halcion with 30mg flurazepam was

inappropriate and so the body of evidence comparing

these doses remained relevant to the clinical risk

assessment of the licensed doses of Halcion (Royal

Courts of Justice, 1994, p. 188):

[the issue is] not at all clear-cutyand I don’t think

we can, therefore, be dismissive of data which show

very striking findings on the basis of the possibility

that they may not be precisely equivalent doses

(Royal Courts of Justice, 1994, p. 74).

J. Abraham / Social Science & Medicine 55 (2002) 1671–16901682

In addition, the CSM did not accept that spontaneous

reporting databases cannot generate quantitative inter-

drug comparisons, especially when the magnitude of

differences between drugs is large (Wood, 1991, p. 8).

They rejected Upjohn’s arguments and advised that all

Halcion licences should be revoked on grounds of safety

(UK Department of Health, 1993a).

Upjohn’s appeal to the Medicines Commission

The company then exercised its right of appeal to the

Medicines Commission under the UK Medicines Act.

The Medicines Commission is an advisory body consist-

ing mainly of expert scientists, but also includes some

non-scientists, such as lawyers. After a formal hearing in

May 1992 the Commission concluded that the licence for

0.25mg Halcion should be revoked, but that the licence

for the 0.125mg product should be allowed to return to

the market. On 17 July 1992, the MCA declined to accept

these conclusions and wrote to Upjohn recommending

the revocation of all Halcion licences. The regulatory

agency recognised that the Commission had reached a

different assessment from its own on ‘substantially the

same evidence’ (Jones, 1992, p. 5), as follows:

1. The Commission took a less serious view of the

number of adverse reactions reported in the UK,

stating that these were similar to those associated

with other benzodiazepines.

2. The Commission took a different view of the clinical

trials arguing that young and middle-aged patients

with no prior mental illness did not appear to be

particularly at risk.

3. Where psychiatric reactions were evident the Com-

mission took the view that they were neither

persistent nor life-threatening.

4. The Commission did not draw any conclusions as to

the frequency or disabling effects of the drug at

different doses when presented with data on patients

withdrawing from clinical trials.

The Commission recommended continued marketing

of the 0.125mg dose, but this advice was based partly

‘on the understanding that 0.125mg is efficacious in

geriatric subjects and in a proportion of non-geriatric

subjects’ (Medicines Commission, 1992, p. 5). However,

the Commission also noted that ‘a dose of 0.125mg was

efficacious in only a minority of non-geriatric subjects

and that the efficacy of 0.25mg was more convincing for

these patients’ (Medicines Commission, 1992, p. 5). The

apparent inconsistency of this was not lost on the MCA,

who commented:

Thus, the restrictions recommended by the Commis-

sion would result in treatment of non-geriatric

subjects being commenced at a dose which they

regard as efficacious in only a minority of such

subjects (Jones, 1992, p. 7).

Upjohn’s appeal to the panel of ‘persons appointed’

Upjohn then decided to exercise its further right of

appeal, under the UK Medicines Act, to a special Panel

of eminent medical experts appointed by the British

Secretary of State for Health, specifically to scrutinise

the case in question. The Panel considered ‘substantially

the same evidence as the CSM and the Medicines

Commission’ (UK Department of Health, 1993a) and

reported to the Department of Health in April 1993

(Anon, 1993a, b, c). It concluded that both 0.25

and 0.125mg dosages of Halcion should be allowed

back on the market with some restrictions in dosage,

indications and duration of use. The Panel’s findings

differed from those of the MCA in five key ways (Anon,

1993a, b, c).

First, additional data presented by the company

included a post-marketing surveillance study carried

out in the UK which allowed the Panel to conclude that

the 0.125 and 0.25mg doses were efficacious and did not

point to any important safety hazard. The MCA, on the

contrary, found the new study to provide ‘insufficient

reassurances when considered in the context of evidence

from several other sources which indicate an important

safety hazard with Halcion’ (UK Department of Health

1993a).

Second, the Panel accepted Upjohn’s assertion that,

when comparing the safety of Halcion with other

benzodiazepines, the doses which had been used, i.e.

0.5 mg of Halcion compared with 30 mg of flurazepam,

could not be considered equivalent. Consequently,

comparative trials were not based on equivalent/

equipotent doses and could not, therefore, be regarded

as useful evidence on which to draw conclusions

about the safety of the drug. In fundamental disagree-

ment, the MCA did not accept that the doses used in

the comparative Halcion/flurazepam trials were non-

equivalent and regarded the findings from these trials as

valid.

Third, the Panel reached different conclusions from

the MCA on the degree of efficacy of Halcion at low

doses. It recommended a starting dose of 0.125mg in

young/middle-aged patients (rising to 0.25 mg if neces-

sary) and 0.0625mg in the elderly (rising to not more

than 0.125mg). Yet, as the MCA noted, there was little

evidence to support the efficacy of the 0.125 mg dose in

the non-elderly and no completed studies at all

concerning the 0.0625 mg dose.

Fourth, while the MCA placed considerable weight on

published spontaneous reporting data from the US

which indicated that adverse psychiatric events had been

J. Abraham / Social Science & Medicine 55 (2002) 1671–1690 1683

reported 22–99 times more frequently with Halcion than

with a comparator drug, the Panel appeared to place no

weight on this evidence.

And fifth, the Panel noted that the information

provided by Upjohn relating to P321 was incomplete,

but accepted Upjohn’s argument that this had only

marginal relevance to the safety of the lower doses in the

UK. Unlike the MCA, the Panel seems to have regarded

Upjohn’s omissions in P321 as irrelevant to the safety of

the 0.125 and 0.25mg doses.

The decision of the MCA on behalf of the UK licensing

authority

Between September 1991 and April 1993, the MCA

had received three different recommendations from

three different expert advisory committees. The CSM

had recommended that all Halcion licences should be

suspended and revoked, the Medicines Commission that

the licence for the 0.25mg dose should be revoked, but

that the 0.125 mg dose should be permitted back on the

market, and the Panel had advised that both the 0.25

and 0.125 mg doses should be permitted back on the

market, subject to low start doses. On 9 June 1993,

disregarding the advice from the Panel and Commission,

but upholding the CSM’s advice, the UK Licensing

Authority finally revoked all Halcion’s licences (UK

Department of Health, 1993a).

In a response going beyond the UK Medicines Act,

Upjohn brought proceedings before the UK High Court

in August 1993 seeking to have the revocation decision

quashed. Upjohn further argued that it was necessary to

involve the European Court of Justice (ECJ) to obtain

guidance on the way the British courts should proceed in

examining the case. When the High Court refused to do

this, Upjohn went to the UK Appeals Court, which

referred the matter to the ECJ in the form of the

question: was it the duty of a national court to decide

whether a licensing authority’s decision was correct, as

opposed to a decision the licensing authority could

reasonably have reached? In January 1999, the ECJ

made the crucial ruling that in making complex

assessments regulatory agencies enjoy a wide measure

of discretion, subject to judicial review, which must be

restricted to an examination of the accuracy of

the findings made by the authority and a verification

that the action taken by that authority was ‘not

vitiated by a manifest error or misuse of powers’ (Anon,

1999).2

Discussion

Making generalisations about medicines regulation

from a single case must be done with caution. On the

other hand, because of the commercial and state secrecy

surrounding medicines testing and regulation in the

Netherlands and the UK, considerable weight should

be given to those cases which enter the public domain

sufficiently to permit social scientific analysis. Further-

more, Halcion was taken by millions of patients within

a relatively short period. If a post-marketing warning

system is deficient in its handling of Halcion, then

that implies that it is likely to be deficient more

generally.

At the very least, the Halcion case raises serious

questions about the supposed strength of post-market-

ing surveillance in the Netherlands and the UK. Since

Halcion’s revocation in the UK, some regulatory

agencies have introduced new computerised systems

for tracking spontaneous ADR reports. In the UK, this

is known as adverse drug reaction on-line information

tracking (ADROIT), which accelerates and improves

data management. Valuable as these developments may

be, data management systems are crucially dependent on

the quality of the data and those producing them, and

on the interpretative and value frameworks of regula-

tors. Thus, the questions that the Halcion case raises

about post-marketing surveillance cannot be reduced to

technical problems of pseudo-epidemiology, which

typically receive all the attention in the regulatory

literature. On the contrary, the foregoing analysis

demonstrates that much more attention needs to be

given to the conduct of the major social, economic and

political institutions involved. Furthermore, the argu-

ment that early licensing in both countries is justified

because there is in place a robust ‘early warning system’

to protect patients once a drug is on the market is not

supported by the Halcion case. This is because of how

the manufacturers, the regulators and the medical

profession interpret spontaneous adverse drug reaction

reports.

The Halcion case suggests that a manufacturer’s

commitment to their product, albeit genuinely held,

may conflict with a commitment to investigate thor-

oughly the dangers to public health implied by doctors’

spontaneous reporting about that product. As Justice

May concluded with respect to Upjohn, a manufacturer

genuinely and strongly committed to their product may

respond to criticism of it with ‘cut-throat commercial

tactics’ designed to discourage the publication of

criticism, which may be seen in a ‘poor light as an

expression of commercial ethic’ and regarded as

‘unacceptable by the public (Royal Courts of Justice,

1994, pp. 168, 272). If Justice May is correct, then under

these circumstances, some staff employed by a manu-

facturer may even give consideration to the suppression

2By the time of writing, the case had returned to the UK

High Court, but Upjohn did not make a case there that the

MCA’s actions were vitiated by a manifest error or misuse of

powers.

J. Abraham / Social Science & Medicine 55 (2002) 1671–16901684

of the doctors’ ‘early warning signal’, as occurred within

Upjohn.

Moreover, the inadequacies in a manufacturer’s

investigation of adverse drug reaction reports may

significantly influence medical debate about the safety

of the product. In particular, large transnational

pharmaceutical companies, who are committed to their

product may produce an alternative view to that of the

doctors who first sound the ‘warning signal’. This may

create additional doubts about an ‘early warning signal’

in the minds of the wider medical profession and the

regulators. It may even undermine it entirely. For this

reason it is crucially important that companies provide

comprehensive information to the medical expert com-

munity. Yet, because of the extent of control which

companies are permitted to exert over drug testing data,

omissions in a manufacturer’s investigations of an ‘early

warning’ may be passed on to, and reproduced by, the

expert medical profession, who rely on the manufacturer

for the provision of the relevant data. In the Halcion

case, this took the form of an article by medical experts

in the Lancet in 1979, which reassured doctors about the

safety of Halcion by undermining van der Kroef’s ‘early

warning signal’. While Upjohn, at the time, believed that

van der Kroef’s data were aberrant, the Lancet article

supporting that view was later regarded as invalid by its

leading author after a full investigation of the drug

testing data existing before 1979. That this can occur

raises questions about how well the relationship between

medical experts and pharmaceutical companies operates

as an investigative mechanism into ‘early warning

signals’—questions which are not entirely obviated by

the fact that two of the co-authors of this article

continue to regard it as valid.

The acceptance or scepticism with which medical

experts treat manufacturers’ responses to ‘warning

signals’ appears to depend on the broad social values

they bring to the situation, rather than merely the

scientific values of the medical profession. For example,

Oswald believed that huge commercial interests in

medicine posed a threat to the independent scientific

studies of new drugs. He wished for more ‘independent’

experts, by which he meant those with no financial links

with pharmaceutical companies and he wanted more

intervention by government regulatory authorities to

protect the public from the risks of medicines (Personal

Communication, 1994d). By contrast, Hindmarsh be-

lieved that consulting and researching for lots of

pharmaceutical companies—the rationality of the mar-

ket—was the best way to ensure ‘independence’. He

claimed that he maintained his independence ‘by having

worked with virtually every drug’ (Personal Commu-

nication, 1994c). As regards medicines regulators and

their expert advisers, he wanted them to be separated

from government so that, as he saw it, politics could be

taken out of medicines risk assessment. From his

perspective the intervention of government is an

undesirable political interference with medical science,

while the commercial and institutional interests of

pharmaceutical companies are not (Personal Commu-

nication, 1994c). Given that such social values, rather

than solely technical calculations, enter into risk-benefit

assessments, it may be appropriate for society at large to

have much greater access to regulatory information than

currently exists in the UK or the Netherlands (Dukes,

1996; Medawar, 1996).

Previous research has demonstrated ‘double-stan-

dards’ in the indications for, and promotion of, drug

products in Western industrialised countries compared

with developing countries (Silverman, 1982; Silverman,

Lydecker, & Lee, 1992). The Halcion case illustrates the

operation of ‘double standards’ in the safe doses

approved in the Netherlands and the UK. Given that

the recommended safe dose for the elderly in the UK

was only a quarter that approved in the Netherlands,

this ought to have sounded an immediate warning signal

to the Dutch regulatory authorities. To be effective in

protecting public health, early warnings need to be

translated into regulatory responses in the absence of

voluntary action by manufacturers. One can define a

‘strong’ warning system as one which leads to immediate

regulatory intervention in response to the risk posed to

public health, while further investigations are con-

ducted. By contrast, a ‘weak’ warning system is one

which relies on extensive confirmation from other

evidence and investigations before any significant

regulatory action is taken.

As regards van der Kroef’s ‘early warning’, the Dutch

regulatory authorities operated a ‘strong’ system by

suspending Halcion from the market in 1979, whereas

the British took a ‘weak’ approach by leaving the drug

on the market, while awaiting possible confirmation of

the ‘warning signal’. This was, and remains, the

regulatory policy approach in the UK. In 1994, the

Chairman of the CSM stated:

You have to have sufficient evidence to make you feel

that the drug should be withdrawn. You don’t have

an option of saying you don’t know—if you don’t

know then you leave the drug on the market by and

large [emphasis added] (Personal Communication,

1994e).

But then, according to the CSM, one does not know

whether a drug is safe when it is first licensed, especially

given the early licensing policies in the UK and the

Netherlands. Hence, there is always some doubt when a

drug goes on the market. The crucial issue is who

gets the benefit of the doubt: the drug and the

manufacturer’s claims about its benefits or patients’

safety? The result of the regulatory approach in the UK

is that compelling evidence of danger to public health in

J. Abraham / Social Science & Medicine 55 (2002) 1671–1690 1685

support of a ‘warning signal’ is required before a drug is

withdrawn.

Not only do the British regulators require such

compelling evidence, but, in the Halcion case, they left

it up to Upjohn to search for such evidence, rather than

do so themselves. The importance of this is evident from

the fact that the MCA’s final revocation decision stated

that the clinical trials available in 1978 were ‘the most

significant evidence indicating an important safety

hazard with the drug’ (UK Department of Health,

1993a, p. 2). This raises the question: how likely is it that

manufacturers will investigate the risks associated with

their pharmaceutical products so as to produce compel-

ling evidence against their safety? Upon the answer to

this question, rests the verdict on whether the British

drug regulatory system can adequately protect public

health in the face of early, and not so early, ‘warning

signals’ from doctors. Notwithstanding the problems of

generalisation, the Halcion case suggests that manufac-

turers are unlikely to search effectively for compelling

evidence against the safety of their drugs and, therefore,

that a regulatory system, which relies so extensively on

industrial self-examination, is flawed. This does not, and

need not, imply bad faith on the part of the manufac-

turer. As Justice May put it, it may be ‘an attitude where

people who are personally and genuinely committed to a

safe, effective and profitable drug act, speak and write

from that committed position—their actions, words and

writings are honest but are affected by the natural

reaction of the human mind to favour the position to

which it is committed (Royal Courts of Justice, 1994, p.

170).

On the other hand, the experience of Halcion

demonstrates that powerful transnational pharmaceuti-

cal companies have the resources and the inclination to

challenge decisions by regulatory agencies in the courts

in the Netherlands and the UK. Consequently, there are

legal considerations, which militate against a ‘strong

warning system’ because regulators feel the need to be

able to justify in court decisions to suspend or revoke

drug product licences. The ban on Halcion in the

Netherlands was eventually unravelled and to some

extent reversed by the Dutch courts, whereas the ‘weak’

regulatory response in the UK, which waited for

compelling evidence before banning the drug, has

withstood numerous appeals by Upjohn in the British

courts and elsewhere. In short, regulators’ response to

their own ‘early warning systems’ may be influenced by

how they weigh up the importance of forthright patient

protection against the threat to their credibility posed by

potentially successful adversarial legal action by a

powerful transnational company.

Evidently, the British and Dutch regulatory autho-

rities weighed up these considerations differently. This

shows that the power of transnational companies to

resort to legal action against regulators need not

determine whether there is a ‘weak or ‘strong’ regulatory

response to doctors’ warnings about drug safety. On the

other hand, in the UK, the problems, which regulators

may face after suspending a drug product, may be

compounded by the elaborate professional appeals

procedures open to manufacturers before a case may

even go to court. Under these circumstances, regulators

are also confronted with the possibility that the

professional credibility of the medicines licensing system

might be damaged if the various appeals bodies contra-

dict the suspension or revocation decision by the

regulators.

As discussed above, such contradictions featured in

the Halcion case. During the appeals to the CSM, the

Medicines Commission and the Panel, Upjohn’s inter-

pretation of P321 and the Halcion/flurazepam pre-

market clinical trials posed major challenges to post-

marketing surveillance and to the rationality of medicine

more generally. The ‘logic’ of the ‘weak’ post-market

‘warning system’ in the UK is that doctors’ spontaneous

reports are investigated for confirmation from other

sources such as pre-market trials. However, when it was

discovered that those trials did support doctors’ post-

market warnings, Upjohn disowned many of their own

pre-market trials as invalid.

Acceptance of Upjohn’s perspective on these matters

would have three major implications. First, early

licensing may not be advisable because it apparently

permits marketing authorisation on the basis of invalid

trials. Second, regulatory action in response to doctors’

warnings, even in the ‘weak’ sense, becomes almost

impossible because controlled pre-market trials cannot

yield confirmatory (or contrary) evidence. And third,

controlled trials in medicine have no objective status,

but rather their validity depends on socio-historical

context.3 Given that the Medicines Commission and the

special Panel of experts in the UK were persuaded by

some of Upjohn’s interpretations, there is clearly an

extremely wide latitude of interpretative flexibility

among expert government advisers, which is in need of

explanation.

Since the late 1980s, growing signs of patients’

dependence on benzodiazepines led some regulators to

be less sympathetic towards these drugs. In 1991, the

CSM declared that, as a matter of policy, the licensing of

benzodiazepines would be affected by concerns over

their improper use:

The CSM has recommended restrictions on the use of

benzodiazepines in order to reduce the risk of

dependence.yTo discourage the improper use of

benzodiazepines, these products are only licensed for

3 It is not being suggested that this is the case—only that this

is the logical implication of accepting Upjohn’s arguments on

these matters.

J. Abraham / Social Science & Medicine 55 (2002) 1671–16901686

short-term use in patients with severe anxiety and

insomnia (Bottomley, 1991).

It seems likely that this regulatory environment

affected the CSM’s approach to Halcion, whose safety

profile had come under major challenge. As one former

member of the CSM involved in the Halcion decision

put it:

I think there was a bigger, deeper issue there [with

Halcion] which couldn’t be dealt with, and which was

the safety of all the benzodiazepines. The question

was, ‘Should all the short acting benzodiazepines be

phased out? And what is the way of doing it?’

(Personal Communication, 1994f).

This may partly explain why the CSM and the MCA

disagreed with the Panel and the Medicines Commission

about the significance of the adverse reactions during

Halcion trials involving higher doses than on the UK

market. In particular, the MCA considered that the

higher doses recommended by the Medicines Commis-

sion and Panel did not provide an adequate safety margin

for patients because ‘it is well established that benzo-

diazepines are sometimes used at higher doses and for

longer periods than are recommended in data sheets’

(Jones, 1992, p. 6).

Furthermore, unlike the CSM and the MCA, neither

the Medicines Commission nor the special Panel, had

been closely involved in the initial licensing of Halcion.

For the CSM and the MCA, Upjohn’s incomplete

disclosure of P321 data in 1978 was related to public

safety because it misled them into underestimating the

significance of other alerting hazard signals. As Waller

put it:

Had the entire data been presented for this study they

would surely have raised the signal that subsequently

came from van der Kroef in Holland in 1979.yIt

seems inconceivable that the Committee would have

recommended the grant of the licence in 1978 had the

true findings of this study been presented (Wood,

1991, p. 9).

Thus, a breakdown in the trust between the MCA and

the CSM, on the one hand, and Upjohn, on the other,

had occurred—the vital result of which was that these

regulators were no longer willing to give the company

the benefit of the many scientific doubts about Halcion’s

safety. A determination to give the benefit of the doubt

to patients’ safety above those of Upjohn is evident in

the Licensing Authority’s final decision that it ‘remains

not reassured of the safety of Halcion products’ (UK

Department of Health, 1993a, p. 3), that ‘the first

concern must be for patients’, and that ‘it is right to be

cautious’ (UK Department of Health, 1993b, p. 2).

By contrast, the experience of the Halcion case for the

Medicines Commission and the special Panel differed

from that of the CSM and the MCA, even though they

all reviewed substantially the same technical data. Under

the Medicines Act, during the appeal before the

Medicines Commission, the company may address only

the CSM’s reasons for their regulatory judgement and,

importantly, only the appellant’s case is heard. Simi-

larly, the remit of the special Panel was to hear solely the

appeal of the company, but not to hear the MCA’s case

against the company. These factors, combined with a

willingness to give Upjohn the benefit of the doubt,

probably go some way to explaining why the judgements

of the Medicines Commission and Panel conflicted with

those of the CSM and the MCA.

In conclusion, Waller’s coherent analysis of how the

risks of Halcion outweighed its benefits, and the fact

that his, and the CSM’s, interpretations prevailed over

the those of Upjohn, the Medicines Commission and the

Panel in the MCA’s regulatory decision shows that

cynicism about British medicines regulation is out of

place. This comment may be even more true when

considering Dukes’ regulatory intervention in the

Netherlands. Evidently, ‘strong’ regulation to protect

public health is possible within the current legislative

frameworks in both the Netherlands and the UK. On

the other hand, the analysis in this paper suggests that

the regulatory policies of ‘early licensing’ and of ‘weak’

responses to post-market warning signals may be

sociologically naive and medically unsafe.

According to Dukes, ‘it is inconceivable that the

Dutch regulatory agency would have approved Halcion

had it been aware of the ADR data known in 1994 to

have been in the possession of Upjohn at the time’

(Royal Courts of Justice, 1994, p. 151). Similarly, in July

1992, the Chief Executive of the MCA told Upjohn that

‘acceptance of triazolam [Halcion] as safe at the licensed

doses at that time [1978] must be regarded as flawed’

(Jones, 1992, p. 3). In other words, by the safety

standards of the MCA itself, the British regulatory

authorities’ non-detection of Upjohn’s inadequate hand-

ling of Halcion permitted patients to be exposed to an

unsafe drug for nearly 13 years. To avoid such

calamities, regulators in both countries should subject

product licence applications to more rigorous pre-

market scrutiny even if that implies ‘later’ licensing,

unless there is clear evidence that the drug confers

substantial therapeutic benefits over existing medicines.

This case study of Halcion is congruent with the

concerns already expressed by many European regula-

tors that the dramatic shortening of approval times in

the UK and some other EU countries may undermine

extensive pre-market safety checks (Abraham & Lewis,

1999, 2000). In the UK, regarding drugs, such as Halcion,

for which equally effective alternative medications are

available, regulators should adopt the ‘strong’ Dutch

policy approach to post-market ‘warning signals’ from

doctors, involving measured regulatory intervention

J. Abraham / Social Science & Medicine 55 (2002) 1671–1690 1687

while investigators independent of the manufacturer

assess the significance of the reported hazard.

Acknowledgements

I am grateful to the UK Economic and Social

Research Council (ESRC—Grants Refs R000221122

and R000221470), for funding some of the research on

which this paper is based, and to Julie Sheppard for her

assistance in collecting some of the data and in verifying

data analysis. Thanks also to three anonymous referees

for their comments on an earlier version of the paper.

Appendix A. Chronology of Halcion events

1977 Approved for marketing in the Netherlands

at 0.25, 0.5 and 1.0 mg strengths

1978 Upjohn cite P321 data as reassurance of

long-term safety to UK CSM

1978 Approved for marketing in the UK at 0.125

and 0.25mg strengths

1979 Van der Kroef and hundreds of other

Dutch doctors report adverse psychiatric

effects

1979 Banned in the Netherlands

1979 Boston ‘thought leader conference’ chaired

by Ayd

1979 Lancet letter casting doubt on validity of

van der Kroef’s reports

1979/1980 Upjohn undertake ‘comprehensive review

of all available evidence’ pertaining to

safety concerns in the Netherlands

1982 UK CSM states that the adverse experi-

ences seem unique to the Netherlands,

where the dose used was at least double

that recommended in the UK

1982 Approved for marketing in the US at 0.125,

0.25 and 0.5 mg strengths

1982 Oswald clinical trials suggest less safe than

other hypnotics

1982 US FDA inform Upjohn that a principal

clinical investigator on pre-marketing Hal-

cion clinical trials is disqualified for sub-

mitting false data

1984 US FDA inform Upjohn that a clinical

investigator on pre-marketing Halcion clin-

ical trials is disqualified for submitting false

data

1985 Dutch Council of State (Crown Court)

reverses the Dutch regulatory authority’s

banning

1986 Halcion receives far more spontaneous

adverse drug reaction reports than any

other benzodiazepines in the UK since its

marketing

1987 Recommended dosage change in US from

0.5 to 0.25 mg

1989 Oswald calls for withdrawal from market

on safety grounds

1990 Dutch regulatory authorities required to

‘reapprove’ at 0.125 and 0.25 mg strengths

1990/1991 Grundberg litigation in US discovers

wholesale omissions of medical events in

reporting of P321

1991 UK MCA learn that P321 data submitted

initially were unreliable and that the pro-

tocol implies significant association with

adverse psychiatric effects, seemingly con-

firming van der Kroef

1991 UK MCA learn that some key clinical

investigators have been disqualified by the

FDA because of submission of false data

1991 Licence suspended in UK on advice of

CSM

1992 UK Medicines Commission recommends

return to market at 0.125 mg strength only

1993 UK Panel of ‘Appointed Persons’ recom-

mends return to market at strengths 0.0625,

0.125 and 0.25mg

1993 Banned in UK, accepting the view of the

CSM above others

References

Abraham, J. (1997). The science and politics of medicines

regulation. In M. A. Elston (Ed.), Sociology of health &

illness special monograph on sociology of medical science

(pp. 53–82). Oxford: Blackwell.

Abraham, J. (1998). Regulating the cancer-inducing potential

of non-steroidal anti-inflammatory drugs. Social Science &

Medicine, 46, 39–51.

Abraham, J., & Lewis, G. (1999). Harmonising and competing

for medicines regulation: How healthy are the European

union’s systems of drug approval? Social Science &

Medicine, 48, 1655–1667.

Abraham, J., & Lewis, G. (2000). Regulating medicines in

Europe: Competition, expertise and public health. London:

Routledge.

Abraham, J., & Sheppard, J. (1997). Democracy, technocracy

and the secret state of medicine: Expert and non-expert

perspectives. Science, Technology and Human Values, 22,

139–167.

Adam, K., & Oswald, I. (1989). Can a rapidly eliminated

hypnotic cause daytime anxiety? Pharmacopsychiatry, 22,

115–119.

Anon. (1985). Upjohn can resubmit Halcion in NL. Scrip, 1014, 1.

Anon. (1988). Top 50 branded products worldwide. Scrip, 20

(review issue for 1988).

Anon. (1990). Halcion re-approved in the Netherlands. Scrip,

1540, 22.

J. Abraham / Social Science & Medicine 55 (2002) 1671–16901688

Anon. (1991a). Upjohn’s Halcion data to be disclosed. Scrip,

1630, 14.

Anon. (1991b) Zantac leads in US. Scrip, 1669, 25.

Anon. (1993a). UK Halcion hearing set for February 1st. Scrip,

1788, 21.

Anon. (1993b). UK Panel recommends return of Halcion.

Scrip, 1814, 25.

Anon. (1993c). Halcion—why UK authorities rejected panel

advice. Scrip, 1830, 25–26.

Anon. (1999). European court of justice decision in Halcion

case. Scrip, 2407, 2.

Asscher, W. (1987). The perception, measurement and manage-

ment of therapeutic risk. Paper Presented on 11 June at

Metropole Hotel, Brighton.

Asscher, W. (1991). ‘Dear Doctor’ Letter from W. Asscher.

Chairman of Committee on the Safety of Medicines (CSM).

Ayd, F. (1992). Affidavit, Maryland, USA, 7 December.

Ayd, F., Barday, W. R., Callan, J. P., Curran, W. J., Gardner,

E. A., Greenblatt, D. J., Ladimer, I., Papierre, Y.,

Lehmann, H. E., O’Donnell, T. J., van Praag, H. M., &

Shader, R. I. (1979). Behavioural reactions to triazolam.

Lancet, II, 1018.

Bottomley, V. (1991). Secretary of state for health, psycho-

tropic drugs. Hansard (Written Answers), 1568, 266–267.

British Broadcasting Corporation (BBC). (1991). The Halcion

nightmare, Panorama, 14 October.

Burley, D. M., & Glynne, A. (1985). Clinical trials. In D. M.

Burley, & T. B. Binns (Eds.), Pharmaceutical medicine (pp.

126–148). London: Edward Arnold.

Committee on Safety of Medicines (CSM). (1991). Report to

EC Committee on Proprietary Medicinal Products (CPMP),

9 December.

Dement, W. C. (1983). Rational basis for the use of sleeping

pills. Pharmacology, 27(Suppl. 2), 3–38.

Dukes, M. N. G. (1979). The strong, the weak and the wicked.

In J. Z. Bowers, & G. P. Velo (Eds.), Drug assessment

criteria and methods. Amsterdam: Elsevier.

Dukes, M. N. J. (1994). Testimony quoted in Royal Courts of

Justice Judgement between the Upjohn Company and Upjohn

Ltd and Professor Ian Oswald and between Dr Royston

Frederick Drucker and Professor Oswald and between the

Upjohn Company and Upjohn Ltd and the BBC and Tom

Mangold before Mr Justice May, Beverley F. Nunnery,

London, 27 May.

Dukes, M. N. G. (1996). Drug regulation and the tradition of

secrecy. The International Journal of Risk and Safety in

Medicine, 9, 143–150.

Food and Drug Administration (FDA). (1994). Establishment

inspection report on Upjohn. Kalamazoo, MI, 15 April.

Hemminki, E., & Pesonen, T. (1977). An inquiry into

associations between leading physicians and the

drug industry in Finland. Social Science & Medicine, 11,

501–506.

Idanpann-Heikkila, J. E. (1985). Adverse reactions and post-

marketing surveillance. In D. M. Burley, & T. B. Binns

(Eds.), Pharmaceutical medicine (pp. 126–148). London:

Edward Arnold.

Inman, W. H. W. (1986). Monitoring for drug safety. Lancaster:

MTP Press.

Jones, K. (1992). Letter from Director and Chief Executive,

MCA to Lee, Finance Director and Secretary, Upjohn, 17

July.

Kagan (1979). Internal Upjohn memo to Struck, 14 September.

Kales, A., Kales, J., Bixler, E. O., Scharf, M. B., & Russek, E.

(1976). Hypnotic efficacy of triazolam: Sleep laboratory

evaluation of intermediate-term effectiveness. Journal of

Clinical Pharmacology, 16, 399.

Kratochvil. (1979). Internal Upjohn memo from the European

Medical Affairs Director to Struck, 11 September.

Krzywicki, K. (1991). Safety of triazolam. Lancet, 338, 632.

Lasagna, L. (1980). The Halcion story: Trial by media. Lancet,

I, 815–816.

Lazarou, J., Pomeranz, B. H., & Corey, P. N. (1998). Incidence

of adverse drug reactions in hospitalised patients: A meta-

analysis of prospective studies. Journal of the American

Medical Association, 279, 1200–1205.

Mann, R. D. (1984). Modern drug use: An enquiry on historical

principles. Lancaster: MTP Press.

Mann, R. D. (1987). (Ed.). Adverse drug reactions: The scale

and nature of the problem and the way forward (pp. 115–126).

Carnforth, Lancashire: Parthenon Publishing.

Medawar, C. (1996). Secrecy and medicines. The International

Journal of Risk and Safety in Medicine, 9, 133–142.

Medicines Commission. (1992). Report on Halcion.

Meyboom, R. H. B. (1992). The triazolam experience in 1979 in

the Netherlands: A problem of signal generation and

verification. In B. L. Strom, & G. Velo (Eds.), Drug

epidemiology and post-marketing surveillance. New York:

Plenum Press.

Morgan, K., & Oswald, I. (1982). Anxiety caused by a short-life

hypnotic. British Medical Journal, 284, 942.

Oswald, I. (1989). Triazolam syndrome 10 years on’. Lancet, II,

451–452.

Personal Communication. (1994a). Interview with Professor Sir

William Asscher. at St. George’s Hospital Medical School,

University of London, 6 October, at pp. 9–10 of transcript.

Personal Communication. (1994b). Letter to Simon Vane-

Percy, Upjohn, Surrey, UK, 18 August.

Personal Communication. (1994c). Interview with Professor Ian

Hindmarsh, University of Surrey, 10 October.

Personal Communication. (1994d). Interview with Professor

Ian Oswald, Edinburgh, 15 November.

Personal Communication. (1994e). Interview with Professor Sir

Michael Rawlins, University of Newcastle, 14 November.

Personal Communication. (1994f). Interview with former

member of the CSM.

Personal Communication. (1995a). Letter to Dr Assenzo,

Director of Pharmaceutical Regulatory Affairs, Upjohn,

Kalamazoo, MI, 13 September.

Personal Communication. (1995b). Letter to Edward Greissing

Junior, Vice-President, Upjohn, Washington, DC, 3 October.

Purpura, R. (1979). Internal Upjohn memo, 17 September.

Rickels, K., Gingrich, R. L., Morris, R. J., Rosenfeld, H.,

Perloff, M. M., Clark, E. L., & Schilling, A. (1975).

Triazolam in insomniac family practice patients. Clinical

Pharmacology and Therapeutics, 18, 315.

Royal Courts of Justice. (1994). Judgement between the Upjohn

Company and Upjohn Ltd and Professor Ian Oswald and

between Dr Royston Frederick Drucker and Professor Oswald

and between the Upjohn Company and Upjohn Ltd and the

J. Abraham / Social Science & Medicine 55 (2002) 1671–1690 1689

BBC and Tom Mangold before Mr Justice May, Beverley F.

Nunnery, London, 27 May.

Sigelman, D.W. (1989). Halcion: Waking up to the dangers of a

sleeping pill. Trial, November, 40.

Silverman, M. (1982). Prescriptions for death: The drugging of

the third world. Berkeley: University of California Press.

Silverman, M., Lydecker, M., & Lee, P. R. (1992). Bad

medicine: The prescription drug industry in the third world.

Stanford: Stanford University Press.

Upjohn Ltd. (1979). Halcion data sheet.

Upjohn Ltd. (1982). Halcion package insert.

Upjohn Ltd. (1987). Halcion package insert.

UK Department of Health. (1993a). Letter in ‘Current

Problems’, 9 June.

UK Department of Health. (1993b). Withdrawal of Halcion,

Press Release, 9 June.

Van der Kroef, K. (1979). Reactions to triazolam. Lancet, 526.

Veldkamp, W., Rudzik, A. D., Metzler, C. M. (1973). Upjohn

Pharmaceutical Research and Development Technical Report,

U-33,030: Six-Week Tolerance Study, SPSM Protocol No.

321, Phase II Study, 27 August.

Vogel, G., Thurmond, A., Gibbons, P., Edwards, K., Sloan, K.

B., & Sexton, K. (1975). The effect of triazolam on the sleep

of insomniacs. Psychopharmacologia, 41, 65.

Vos, R. (1991). Drugs looking for diseases: Innovative drug

research and the development of the beta blockers and the

calcium antagonists. Dordrecht: Kluwer.

Waller, P. (1994). Testimony of Dr. Patrick Waller. In

Proceedings between the Upjohn Company and Upjohn Ltd.

and Professor Ian Oswald and between Dr. Royston Frederick

Drucker and Professor Oswald and between the Upjohn

Company and Upjohn Ltd. and the BBC and Tom Mangold

before Mr Justice May, Beverley F. Nunnery, London, 25

April, Day 54.

Walker, S. R., & Lumley, C. E. (1987). Reporting and

under-reporting. In R. D. Mann (Ed.), Adverse drug

reactions: The scale and nature of the problem and the way

forward (pp. 115–126). Carnforth, Lancashire: Parthenon

Publishing.

Wheatley, D. (1992). Prescribing short-acting hypnosedatives.

Drug Safety, 7, 106–115.

Wieringa, N. F., Meijer, A. H. R., de Schutjens, M. D. B., &

Vos, R. (1992). The gap between legal rules and practice in

advertising non-registered pharmaceutical products. Social

Science & Medicine, 35, 1497–1504.

Wood, S.M. (1991). MCA Report from Triazolam’s Medical

Assessor to CPMP, 9 December.

J. Abraham / Social Science & Medicine 55 (2002) 1671–16901690