ARTICLE IN PRESS

0277-9536/$ - se

doi:10.1016/j.so

�CorrespondE-mail addr

Social Science & Medicine 61 (2005) 881–892

www.elsevier.com/locate/socscimed

A comparative analysis of drug safety withdrawals in the UKand the US (1971–1992): Implications for current regulatory

thinking and policy

John Abraham�, Courtney Davis

Centre for Research in Health and Medicine (CRHaM), Department of Sociology, University of Sussex, Arts E Building, Falmer,

Brighton BN1 9SN, England, UK

Available online 2 March 2005

Abstract

By going beyond individual case studies and solely quantitative surveys, this paper systematically examines why there

were over twice as many new prescription drugs withdrawn from the market on grounds of safety in the UK as there

were in the US between 1971 and 1992. Drawing on interviews with regulators, industry scientists and others involved,

and on regulatory data never before accessed outside governments and companies, five key hypotheses which might

explain this difference in drug safety withdrawals are analysed. These are: (1) simply because the UK approved more

new drugs than the US; (2) because of an industrial corporate strategy to seek approval of ‘less safe’ drugs in the UK

earlier; (3) because British regulators were more vigilant at spotting post-marketing safety problems than their US

counterparts; (4) because the slowness of the US in approving new drugs enabled regulators there to learn from, and

avoid, safety problems that had already emerged in the UK or European market; and (5) because more stringent

regulation in the US meant that they approved fewer unsafe drugs on to the market in the first place. It is concluded

that the main explanation for fewer drug safety withdrawals in the US is that the regulatory agency there applied more

stringent pre-market review and/or standards, which took longer than UK regulatory checks, but prevented unsafe

drugs marketed in the UK from entering the US market. Contrary to the claims frequently made by the pharmaceutical

industry and regulatory agencies on both sides of the Atlantic, these results imply that it is likely that acceleration of

regulatory review times in the US and the UK since the early 1990s is compromising drug safety.

r 2005 Elsevier Ltd. All rights reserved.

Keywords: Drug regulation; Safety withdrawals; Pharmaceutical industry; Drug lag; International comparison; Review times; UK;

USA

Introduction

International studies of pharmaceutical regulation

have examined the political contexts of regulatory

institutions in order to characterise regulatory change

e front matter r 2005 Elsevier Ltd. All rights reserve

cscimed.2005.01.004

ing author. Tel.: +440 1273 678883

ess: j.w.abraham@sussex.ac.uk (J. Abraham).

and policy (Abraham & Reed, 2001, 2003; Olson, 1999;

Wiktorowicz, 2003). However, they are not conclusive

about the implications of regulatory context for actual

drug safety and regulatory outcomes. For example, the

implications for drug safety and public health of

regulatory agencies’ acceleration of new drug ap-

proval/review times over the last decade, are not self-

evident—indeed, they are disputed (Lexchin, 1995;

Rawson & Kaitin, 2003).

d.

ARTICLE IN PRESSJ. Abraham, C. Davis / Social Science & Medicine 61 (2005) 881–892882

Research on regulatory outcomes regarding drugs

themselves has been dominated by comparative surveys

of drug regulation across a large number of products,

often concerned with the ‘drug lag’ thesis, the claim that

many drugs were approved faster in the UK than in the

US. Building on the ‘drug lag’ thesis, these surveys have

focused on how many and how quickly new patented

drugs, known as new active substances (NASs), i.e.

new chemical or biological entities, come to the market

in different countries, and whether regulation has

hampered industrial innovation or denied doctors’

significant therapies for their patients (Andersson,

1992; Kaitin, Mattison, Northington, & Lasagna,

1989; Parker, 1989; Schweitzer, Schweitzer, & Sourty-

Le Guellec, 1996; Wardell, 1973). However, some have

provided comparative data on drug safety withdrawals

in the US and the UK (Bakke, Wardell, & Lasagna,

1984; Bakke, Manocchia, de Abajo, Kaitin, & Lasagna,

1995; Jefferys, Leakey, Lewis, Payne, & Rawlins, 1998).

Nevertheless, their emphasis on counting regulatory

outcomes provides little understanding of the processes

of regulatory assessment. Yet such a social scientific

understanding is vital to make rational judge-

ments about the public health protective perfor-

mance of regulation. For example, if one regulatory

agency has more drug safety withdrawals than another,

then this may be because it is better at spotting and

withdrawing unsafe drugs on the market or because it

approves more unsafe drugs on to the market in the

first place.

In this paper, we go beyond all previous research

by conducting systematic comparative analyses of

the regulatory processes and outcomes regarding all

drug safety withdrawals in the UK and US between

1971 and 1992 inclusive. This period has been chosen

because modern regulation of drug safety and efficacy

did not begin in the UK until 1971 with the imple-

mentation of the 1968 Medicines Act, though it had

existed in the US since 1962. The cut-off point of

1992 was chosen because since then the US drug

regulatory agency, the Food and Drug Administra-

tion (FDA) has been subject to the Prescription Drug

Users Fee Act (PDUFA) and subsequent legisla-

tion, which has altered its regulatory context signifi-

cantly, including a partial dependence on industry fees.

Also since 1992, the Europeanisation of UK drug

regulation has become significant, culminating in the

establishment of the European Medicines Evalua-

tion Agency and EU-wide licensing in 1995 (Abraham

& Lewis, 2000). However, as we shall demon-

strate, lessons from this period provide critical insight

into the appropriateness of subsequent policy shifts and

direction.

Specifically, we seek to explain why there have

been approximately twice as many drug safety

withdrawals in the UK as in the US between 1971

and 1992 by reference to the following five hypotheses

(H1–H5):

H1. Simply because the UK approved more drugs than

the US.

H2. Because of an industrial corporate strategy to seek

approval of ‘less safe’ drugs in the UK early, but not to

seek approval, or to seek approval later, for such drugs

in the US.

H3. Because British regulators were more vigilant in

spotting safety problems with drugs after entering the

market than their American counterparts.

H4. Because of the ‘drug lag’, that is, because the slow

and overly bureaucratic US regulatory system lagged

behind the UK in making decisions to approve new

drugs which meant that US regulators could see safety

problems, which could only be detected post-marketing

anyway, and which emerged in the markets of the UK

and/or Europe before the FDA had to decide on

marketing approval in the US.

H5. Because regulators’ safety standards for drug

approval were more stringent in the US than in the

UK, so they approved fewer unsafe drugs.

In drawing this UK/US comparison, we do not

assume that these regulatory agencies operated in

isolation from one another, but they did make different

decisions, which need to be explained, despite informal

collaboration between the two regulatory agencies. Nor

do we discount the possibility that the pharmaceutical

industry may have influenced the regulatory process to

serve its own interests, but it is reasonable to assume

that the regulatory agencies in both countries face

broadly similar industry influences.

Methods, secrecy and data sources

Our research was conducted from 1998 to 2003. To

compile a definitive list of drug safety withdrawals in the

UK and US between 1971 and 1992 we reviewed all

relevant published surveys and collected further doc-

umentary data on individual drug products from the

following sources: MedLine, PubMed, Web of Science

and BIDS; the pharmaceutical trade publications, Scrip

(UK-based) and The Pink Sheet (US-based); The United

Nations’ Consolidated List of Products Whose Consump-

tion and/or Sale have been Banned, Withdrawn, Severely

Restricted or Not Approved by Governments; Hansard;

product labeling and safety data sheets contained in the

ARTICLE IN PRESSJ. Abraham, C. Davis / Social Science & Medicine 61 (2005) 881–892 883

Association of the British Pharmaceutical Industry

(ABPI) Compendia; publications by the Committee on

Safety of Medicines (CSM), an expert advisory commit-

tee to the UK regulatory authority throughout this

period and publications by the UK regulatory authority

itself, known as the Department of Health’s Medicines

Division (DoHMD) from 1971–1989, and as the

Medicines Control Agency (MCA) after 1989; docu-

ments obtained from the FDA by US Freedom of

Information Act (FOIA) requests; US Congressional

Hearings; documents held by the US public health

advocacy group, Public Citizen Health Research Group;

and information obtained through Open Government

requests to the MCA.

Following a successful appeal against the MCA’s

arbitrary claim that dates of licence applications were

commercially confidential, we obtained dates when

product licence applications, for all the drugs withdrawn

on safety grounds between 1971 and 1992 in the UK,

had been submitted, approved and cancelled in the UK,

Consequently, we were able to calculate review times for

individual drugs in the UK—something not done in

previous surveys of drugs in this period, but important

in determining whether a ‘lag’ in regulatory review,

as distinct from a lag in marketing approval,

existed between the UK and the US in relation to these

drugs.

The FDA has published data relating to drugs

approved and withdrawn for safety reasons in the US.

However, no official information was publicly available

relating to drugs for which marketing approval was

sought but not obtained in the US. Whilst more drugs had

to be withdrawn from the UK market on grounds of

safety than in the US, it is necessary to know whether or

not this could be because marketing approval was never

actually sought in the US for the drugs in question.

Initial attempts to obtain basic information about drugs

not approved for marketing in the US using FOIA

requests failed because the FDA maintained that it was

‘proprietary information’, even though the FOIA did

not prohibit its release. Nobody has ever obtained this

information from the FDA before, so the statutory issue

of its disclosure had never been resolved through

litigation. After we threatened to make a legal challenge,

the FDA agreed to release data relating to the FDA’s

review of drugs approved and subsequently withdrawn

in the UK, but non-approved by the FDA, if the drugs

had reached the clinical trial phase in the US.

We reviewed the documentary sources listed above to

examine the regulatory processes for each of the drug

products withdrawn on safety grounds. To clarify and

supplement our findings from these sources, we also

conducted interviews in the UK and the US with current

and former regulators, pharmaceutical industry scien-

tists/representatives and clinical investigators, concerned

with the drugs in question, and other interested parties,

such as current and former Congressional staff, Parlia-

mentarians, medical professionals/specialists, public

health advocacy groups and patient/victim support

groups. Disaggregated response rates were as follows:

�

UK regulators—74% (14/19),

�

US regulators—71% (24/34),

�

Industry—44% (15/34),

�

Others—100% (20/20).

H1. Simply because the UK approved more drugs than

the US.

Between 1971 and 1992, 26 pharmaceutical products

(drugs, biologicals and vaccines, including NASs, new

dosage formulations, or new routes of administration

for NASs), were approved for marketing after 1

September 1971, and subsequently withdrawn for safety

reasons in the UK or US or both (Table 1). While 10 of

the 26 products were marketed in both countries, 15

drugs, which were approved for marketing and then

withdrawn in the UK, were never approved in the US.

By contrast, there were only two drugs marketed in the

US and later withdrawn on safety grounds that were

never approved for marketing in the UK. Also triazolam

and Factor VIII both remained on the US market

despite their withdrawal from the UK market for safety

reasons. Of these 26 products, 24 were approved for

marketing in the UK and then withdrawn, whereas only

12 of them were approved for marketing in the US, of

which nine were subsequently withdrawn. The absolute

difference in drug safety withdrawals between the UK

and the US is, therefore, 24 to 9 or a ratio of 2.67.

However, Bakke et al. (1995) and Jefferys et al. (1998)

argue that this is simply a function of the fact that more

drugs were approved in the UK during the relevant

period.

According to Bakke et al. (1995, p. 115), the drug

safety withdrawal rate (the number of drugs approved

and withdrawn on safety grounds as a percentage of the

total number of new drugs approved in the same period)

between 1974 and 1993, was roughly the same—around

3% in the US and 4% in the UK. Similarly, some UK

regulators, such as Jefferys et al. (1998, p. 156), claim

that ‘there is no evidence of markedly different risk-

benefit withdrawal rates between the two countries’ and

hence that, if drug safety withdrawals represent a failure

of drug regulation, then the record of the UK regulatory

authorities is no worse than that of the FDA.

Using Bakke et al.’s own data and inclusion criteria

(NASs only), for the period 1974 to 1993 inclusive, there

were 10 drug safety withdrawals in the US and 18 in the

UK, while 377 NASs were introduced into the US

market and 430 into the UK. This yields a drug safety

withdrawal rate of 2.6% in the US and 4.2% in the UK.

However, Bakke et al. (1995) did not include five other

ARTICLE IN PRESS

Table 1

Drug safety withdrawals in the UK and the US 1971–1992

Drug Year withdrawn in UK Year withdrawn in US

Alphaxalone+Alphadolone 1984 Not marketed

Azaribine Not marketed 1976

Benoxaprofen 1982 1982

Brotizolam 1989 Not marketed

Clomacran 1982 Not marketed

Crescormon 1985 1985

Dilevalol 1990 Approved for marketing but never

marketed

NDA not approved

Domperidone (injectable) 1986 Not marketed

Encainide 1991 Approved for marketing but never

marketed

1991

Factor VIII 1986 Not withdrawn

Fenclofenac 1984 NDA not approved

Feprazone 1984 Not marketed

Indoprofen 1983 NDA not approved

Indomethacin- (osmotic release form) 1983 NDA not approved

Metipranolol 1991 NDAs for 0.1% and 0.6% not approved

0.3% not withdrawn

Nomifensine 1986 1986

Perhexiline 1985 NDA not approved

Pluserix/ Immravax 1992 Not marketed

Polidexide 1975 Not marketed

Suprofen 1987 1987

Temafloxacin 1992 1992

Terodiline 1991 NDA not approved

Ticrynafen Not approved 1980

Triazolam 1991 Not withdrawn

Zimeldine 1983 NDA not approved

Zomepirac 1983 1983

J. Abraham, C. Davis / Social Science & Medicine 61 (2005) 881–892884

UK drug safety withdrawals (clomacran, brotizolam,

metipranolol, Factor VIII and Pluserix/Immravax)

which, as NASs, should have been counted according

to their own inclusion criteria. On the other hand, the

NAS, L-tryptophan, should be excluded from analysis

because, while a UK drug safety withdrawal at the time

of Bakke et al. (1995), it was subsequently reinstated on

the market. This increases the number of UK drug

safety withdrawals (1974–1993) to 22 and yields a UK

drug safety withdrawal rate of 5.1% almost double that

in the US (2.6%).

Bakke et al. (1995) explicitly exclude two NASs

(encainide and dilevalol) because, while approved for

marketing in the UK, they were withdrawn before

actually being marketed there due to safety concerns

overseas. However, the relationship between drug safety

withdrawals and regulatory performance should incor-

porate analysis of the regulatory decision to approve for

marketing as well as actual marketing. Jefferys et al.

(1998) agree because they include these two NASs in

their list of UK drug safety withdrawals. This inclusion

criterion for both countries increases the number of UK

drug safety withdrawals (1974–1993) to 24 and the drug

safety withdrawal rate in the UK to more than double the

comparable rate in the US.

Evidently, whether one quantifies absolute safety

withdrawals or withdrawal rates, the result is that from

the early 1970s to the early 1990s the extent of drug

safety withdrawals in the UK has been about double that

in the US. Hence, the greater number of drug safety

withdrawals in the UK than the US in this period cannot

be explained merely by the fact that the UK approved

more drugs than the US.

Typically these quantitative surveys reduce the defini-

tion of new drugs to NASs. This commercial perspec-

tive, derived from industry concern to define discrete

molecular entities protected from competition by

patents, is not necessarily very relevant to drug safety

(or efficacy). Thus, in our detailed analysis of with-

drawals and regulatory case histories, we include

indomethacin-OROS (slow release delivery of indo-

methacin) and injectable domperidone (not themselves

NASs) because they were new drug products involving

radically new and therapeutically significant ways of

ARTICLE IN PRESS

Table 2

Withdrawal times (months)

Drug UK US

Alphaxalone+Alphadolone 141 —

Azaribine — 14

Benoxaprofen 27 4

Brotizolam 60 —

Clomacran 60 —

Dilevalolol 4 (drug approved but never marketed in the UK) —

Domeperidone (injectable) 40 —

Encainide 12 (drug approved but never marketed in the UK) 60

Fenclofenac 74 —

Feprazone 103 —

Indomethacin-in-OROS 12 —

Indoprofen 19 —

Metipranolol eyedrops 56 Not withdrawn

Nomifensine 111 12

Perhexiline 132 —

Polidexide 12 —

Suprofen 84a 18

Temafloxacin 10 5

Terodiline 64 —

Ticrynafen — 9

Triazolam 156 Not withdrawn

Zimeldine 21 —

Zomepirac 27 29

aAccording to the MCA this drug was approved on 23 January 1980. However, it was not marketed until 1983 (CSM, 1986).

J. Abraham, C. Davis / Social Science & Medicine 61 (2005) 881–892 885

delivering the NASs. However, for reasons of brevity,

our detailed analysis of withdrawals and regulatory case

histories excludes the three biologicals, namely the blood

product Factor VIII, the growth hormone Crescormon

and the vaccine Pluserix/Immravax—leaving 23 new

drug products (see Table 2).

H2. Because of an industrial corporate strategy to seek

approval of ‘less safe’ drugs in the UK early, but not to

seek approval, or to seek approval later, for such drugs

in the US.

The British and US regulatory agencies might

contribute little to the explanation for the greater

number of drug safety withdrawals in the UK than in

the US if pharmaceutical companies chose not to seek

approval in the US for ‘less safe’ products or chose to

seek approval for such products later in the US. For

example, if they chose to seek approval for such

products in the US later than in the UK (an ‘application

lag’), withdrawal on safety grounds may have led to

their abandonment in the US regulatory process before

having the opportunity to enter the US market.

Of the 26 pharmaceutical products withdrawn in

either country between 1971 and 1992 (Table 1),

approval was sought in both countries for at least 19

of these products. In particular, of the 15 drugs

approved for marketing and then withdrawn in the

UK, but never approved in the US, marketing approval

was definitely sought for at least nine of these drugs in

the US (apa, dilevalol, fenclofenac, indomethacin-

OROS, indoprofen, metipranolol, perhexiline, terodiline

and zimeldine)—and probably more than this because

interviews confirmed that pharmaceutical companies

generally sought to penetrate the lucrative US market

whether or not they might come up against stricter

regulatory controls there. Hence, there was a very

considerable degree of overlap of these drugs submitted

for marketing approval in the two countries and the

greater number of drug safety withdrawals in the UK

than in the US cannot be explained by an industrial

strategy not to seek marketing approval for these drugs

in the US.

Tables 3 and 4 provide data on 14 of the drugs for

which marketing approval was sought in both countries,

and for which there was an identifiable ‘application lag’.

In some cases, a very significant amount of time elapsed

between applications for marketing approval in the UK

and the US for the same product. However, these

‘application lags’ operated in both directions. There was

an ‘application lag’ to US submission for eight of these

14 drugs, and an ‘application lag’ to UK submission for

the other six. For those six drugs whose applications

were submitted to the FDA first, there was an average

ARTICLE IN PRESS

Table 3

Submission and review times: drug applications first submitted to FDA

Country in which

first marketed

‘Lag’ to UK

submission

(months)

UK review times

(months)

US review times

(months)

Difference between

UK/ US review

times (months)

Dilevalol UK only 31 9 — —

Encainide US 62 21 35 +14

Indoprofen UK only 9 5 — —

Perhexiline UK only 31 9 — —

Temafloxacin UK 5 16 26 +10

Zomepirac US 6 18 22 +4

Table 4

Submission and review times: drug applications first submitted to UK Regulatory Authority

Country in which

first marketed

‘Lag’ to US

submission

(months)

UK review times

(months)

US review times

(months)

Difference between

UK/US review

times (months)

Benoxaprofen UK 2 6 27 +21

Fenclofenac UK only 48 14 — —

Metipranolol UK 63 36 14 �22

Nomifensine UK 35 10 72 +62

Indomethacin-

OROS

UK only 3 11 — —

Suprofen UK 2 17 52 +35

Terodiline UK only 39 18 — —

Zimeldine UK only 36 18 — —

J. Abraham, C. Davis / Social Science & Medicine 61 (2005) 881–892886

‘lag’ of 24 months before application to the UK

regulatory authority, whereas for those eight drugs,

whose applications were first submitted to the UK

authority, there was a marginally greater average ‘lag’ of

28.5 months before application to the FDA. Thus, the

significantly greater number of drug safety withdrawals

in the UK cannot be explained (or can be explained only

marginally) by an industrial strategy to seek approval for

‘less safe’ products in the US later than in the UK.

H3. Because of more vigorous post-marketing surveil-

lance by British regulators than their American counter-

parts.

The purpose of post-marketing surveillance is to

detect patients’ adverse reactions to drugs after they

have entered the market (Abraham, 2002). In both the

UK and the US, pharmaceutical companies are required

to report ADRs to their products. In addition, the UK

regulatory authority pioneered a system in the early

1970s for collecting reports of ADRs made voluntarily

by prescribing doctors, known as the yellow card

system. Several years later, the FDA also impressed

upon prescribing doctors in the US the importance of

reporting ADRs to the agency.

Interviews revealed that UK regulators, who were

influential in the period from 1971–1992, believed that

there were more drug safety withdrawals in the UK than

the US because the UK regulatory authorities were

willing to approve drugs earlier than the FDA on the

grounds that the UK post-marketing system was better

than its US counterpart at detecting safety problems,

and that UK regulators were sharper than the FDA at

withdrawing drugs which turned out to be problematic

on the market. The US regulators, whom we inter-

viewed, disputed these claims. To test the validity of this

argument we examined the commitment to post-market-

ing surveillance in the UK and scrutinised the actual

marketing histories of the drug safety withdrawals in the

UK and US.

We found that in the very early years of UK drug

regulation, post-marketing surveillance of ADRs was

neither given much priority nor well resourced. For

example, requests by some UK regulators for a

computer and extra staff between 1965 and 1975 were

refused. Throughout the 1960s and into the 1970s data

on ADRs had to be transported by road to be entered

into the Department of Health’s computer in Reading.

Generally weeks would elapse before processed print-

outs of these data were returned to the regulators

ARTICLE IN PRESSJ. Abraham, C. Davis / Social Science & Medicine 61 (2005) 881–892 887

(Inman, 1999). By 1982, when a major public con-

troversy broke over the safety of benoxaprofen, in the

UK (Abraham, 1994), the CSM still did not have its own

computer and the ADR data obtained through yellow

card reports were not being analysed in a timely manner

(interview, former CSM member, 19 June 2001).

After the embarrassment caused by the benoxaprofen

controversy, a computer and software were made

available to the ADRs section of the DoHMD (inter-

view, former CSM member, 19 June 2001). Several other

changes were introduced. Regulatory staff would look at

the ADR reports every month for a newly marketed

drug, and then quarterly or 6 monthly for the first 5

years of marketing (interview former DoHMD member,

30 May 2001). In addition, prescription data were used

to estimate prescribing volume so that the relative risks

of drugs could be estimated by comparing the yellow

cards for various drugs with their associated prescribing

volume. (CSM, 1986). However, the CSM did not press

for any centrally funded system of post-marketing

surveillance to supplement the yellow card scheme, nor

any system whereby new drugs could be subject to

restricted release before full marketing. Instead, it

recommended that pharmaceutical companies should

conduct post-marketing studies under voluntary ar-

rangements that would be agreed with the CSM (Waller,

Wood, Langman, Breckenridge, & Rawlins, 1992).

Consequently, in 1988 a set of guidelines for such

studies were published by the CSM, the ABPI, the

British Medical Association and the Royal College of

General Practitioners. A regulatory review in 1992 of 31

company-sponsored post-marketing surveillance studies,

which had been conducted since 1988, revealed that they

made only a limited contribution to the assessment of

drug safety because they were generally poorly designed

and companies failed to recruit sufficient patients or to

provide results to the regulatory authorities quickly

enough (Waller et al., 1992). Hence, our research does

not support the claim that there was a particularly high

commitment to post-marketing surveillance of ADRs

within the UK regulatory system.

Moreover, the times that elapsed between approval

and drug safety withdrawals in the UK and the US

between 1971 and 1992 were, on average, much longer in

the UK than the US (Table 2). The average time for the

21 drugs withdrawn in the UK was 58 months, but only

19 months for the eight withdrawals in the US. Only

23% of UK drug safety withdrawals were made within

12 months of approval, whereas this was so for 50% of

drug safety withdrawals in the US. Thus, drugs were

actually withdrawn on grounds of safety much more

quickly in the US than the UK. It might be argued that,

because the FDA was slower to approve new drugs than

the UK regulatory authorities, drugs were withdrawn in

the US on the basis of post-marketing ADR data from

the UK, that is, the FDA learned from the more lengthy

post-marketing experience in the UK. If this were the

case, one would expect the three drug safety withdrawals

in the US, which were never marketed in the UK

(azaribine, encainide and ticrynafen) to have higher

approval to withdrawal times than UK drug safety

withdrawals. However, the average approval to with-

drawal time for these three drugs is 28 months, more

than the overall average for US drug safety withdrawals,

but still less than half the average for UK drug safety

withdrawals.

Our analysis of marketing histories of the drug safety

withdrawals confirms that these UK/US differences in

drug safety withdrawals cannot be explained by the FDA

learning from an ostensibly more vigilant post-market-

ing surveillance in the UK. Taking first drug safety

withdrawals in the US. Both azaribine and ticrynafen

were withdrawn due to spontaneous ADR reports from

US physicians, not on the basis of post-marketing data

from the UK or Europe (US District Court, 1984;

interview, FDA epidemiology consultant, 21 May 2001;

interviews, current and former FDA regulators, 2 and 24

May 2001). Encainide, approved for the treatment of

arrhythmias (abnormal heartbeats), was withdrawn on a

truly voluntary basis by the manufacturer following the

post-market mortality Cardiac Arrhythmia Suppression

Trial (CAST) conduced by the US National Heart, Lung

and Blood Institute, which showed that patients taking

the drug, who had previously survived a heart attack,

were twice as likely to die from any cause as patients on

placebo and three times more likely to suffer sudden

death or cardiac arrest (CAST Investigators, 1989).

Regarding the other five US drug safety withdrawals,

zomepirac, suprofen and temafloxacin were withdrawn

on the basis of ADRs first reported in the US, leaving

only two, nomifensine and benoxaprofen, which were

withdrawn as a consequence of post-marketing reports

and regulatory action in the UK or Europe (Abraham,

1995; Anon, 1986, 1988, 1992; CSM, 1986; FDA, 1992;

US Congress, 1986; interview, former CSM member, 19

June 2001; interview, FDA epidemiologist, 10 May

2001).

Except for encainide, all of the eight drugs withdrawn

from the US market were withdrawn within two and a

half years of approval. By contrast, only nine of the 21

drugs approved and subsequently withdrawn in the UK

were withdrawn within two and a half years of approval.

Two of these, temafloxacin and zomepirac, were with-

drawn on the basis of ADRs that were first detected in

the US; dilevalol was withdrawn worldwide by the

manufacturer following pre-approval recognition of the

drug’s potential hepatotoxicity by the FDA (interview,

senior FDA regulator, 2 May 2001) and encainide due to

the CAST findings. Hence, only five of the 21 UK drug

safety withdrawals were within two and a half years of

marketing and withdrawn on the basis of data generated

by the UK’s system of post-marketing surveillance.

ARTICLE IN PRESSJ. Abraham, C. Davis / Social Science & Medicine 61 (2005) 881–892888

There were long delays before withdrawal of the

remaining 12 drugs from the UK market. Suprofen was

withdrawn because of renal toxicity not detected by the

yellow card system. Similarly, physicians in the UK did

not recognise the cardiotoxicity of terodiline until

potentially fatal cases of heart rhythm disturbances

(torsades de pointes) associated with the drug were

reported in the medical literature in 1991 (Anon, 1991;

Wild, 1992; interviews, industry scientists, 13 and 17

February 2001). Brotizolam was withdrawn on the basis

of a positive rodent carcinogenicity study that ought,

arguably, to have been completed before the drug was

approved (Jefferys et al., 1998). In the case of

nomifensine there was under-reporting of cases of

drug-induced immune haemolytic anaemia for nearly 7

of the 9 years that the drug was marketed in the UK

(Anon, 1988; Mann, 1988). Thus, post-marketing

surveillance in the UK did not detect serious ADRs to

suprofen and temafloxacin, while there were consider-

able delays before the toxicity of nomifensine and

terodiline were recognised in the UK.

Even when adverse reactions associated with specific

drugs were detected, significant delays could ensue

before withdrawal. The cases of apa, fenclofenac,

feprazone and perhexiline suggest a UK regulatory

tolerance of a high and serious degree of toxicity in

drugs for use in non-life-threatening conditions (House

of Commons, 1984; LaPlane & Bousser, 1981; Pieterse,

Rowland, & Dunn, 1983; interviews, former CSM

member 19 June 2001; former DoHMD member, 30

May 2001; industry scientists, 4 October 2000 and 19

January 2001). The cases of clomacran, injectable

domperidone and metipranolol also fall into this

category, but we could not discover definitively why

there was such a delay before these drugs were

withdrawn in the UK. Hence, neither regulatory case

histories of drug safety withdrawals nor average

approval to withdrawal times for these drugs support

the hypothesis that the larger number of drug safety

withdrawals in the UK than the US is a reflection of a

more vigilant system of post-marketing surveillance in

the UK.

H4 and H5. Because of ‘lag’ in approvals between the

US and the UK caused by lengthy, inefficient and overly

bureaucratic regulation at the FDA or alternatively

because the FDA imposed higher regulatory standards

which have kept more unsafe drugs off the US market in

the first place.

Critics of the FDA in the US and some UK regulators

have argued that the FDA’s lengthy reviews of new drug

applications (Tables 3 and 4) were unnecessary and/or

inefficient because the ADRs that cause drugs to be

withdrawn from the market typically cannot be identi-

fied from pre-market scrutiny. These assertions can only

be tested by analysing the regulatory case histories of the

drugs involved. Table 5 presents such an analysis for 19

of our 23 case study drugs, beginning with those

approved for marketing in the US but not the UK

(azaribine and ticrynafen), then the drugs approved for

marketing in both countries (benoxaprofen, encainide,

nomifensine, suprofen, temafloxacin, triazolam, and

zomepiric) and finally those approved for marketing in

the UK but denied approval in the US when it was

sought there. For the remaining four case-study drugs

(brotizolam, clomacran, injectable domperidone and

polidexide—approved in the UK, but not the US), we

have no detailed comparative data because either

approval was not sought in the US or the application

was abandoned so early in the process that no

information is publicly available.

The regulatory case histories (Table 5) show that of

the 10 UK drug safety withdrawals for which new drug

applications were made to, but refused by, the FDA, for

at least four of them (apa, dilevalol, fenclofenac

indoprofen), US regulators denied approval on to the

US market partly or wholly for the safety reasons that

ultimately led to their withdrawal from the UK market.

Moreover, FDA scientists detected the relevant safety

problem from pre-marketing data for at least five of

these 10 drugs. (apa, dilevalol, fenclofenac, indoprofen,

perhexiline). There is clear evidence that FDA non-

approval was based on post-marketing data from the

UK or Europe for only two of these 10 drugs

(indomethacin-in-OROS and perhexiline). Clearly this

evidence does not support the proposition that there

have been fewer drug safety withdrawals in the US

because the slowness of FDA review has enabled US

regulators to deny approval as a consequence of

discovering the safety problems from marketing experi-

ence in the UK or Europe. On the contrary, it implies

that the FDA’s more meticulous reviewing process

enabled them to prevent some unsafe drugs from

entering the US market.

These cases also demonstrate that of all the drug

safety withdrawals in the UK and US, for at least 10

(apa, azaribine, dilevalol, feprazone, fenclofenac, indo-

profen, perhexiline ticrynafen, triazolam, zomepirac),

evidence of the safety problems that ultimately led to

their withdrawal was definitely contained in the data

submitted before marketing approval, and for at

least a further four (benoxaprofen, nomifensine, tema-

floxacin, zimeldine), evidence of the safety problems

that ultimately led to their withdrawal may have been

contained in the data submitted before marketing

approval. Furthermore, in all of these case histories,

it is clear that lengthy reviewing times at the FDA

were largely because agency scientists were grappling

with new drug applications from companies that fell

below the safety and efficacy standards of US regulators.

For example, FDA scientists had concerns about the

ARTICLE IN PRESS

Table 5

Presence, detection and non-approval of safety problems leading to withdrawal

Drug Reason for Withdrawal PPMT DPMT (by) FDANA

Azaribine Thromboembolism Yes Yes (FDA)a

Ticrynafen Liver and kidney toxicity Yes Yes (FDA)b

Benoxaprofen Liver toxicity Prob Noc

Encainide Mortality No No

Nomifensine Haemotological effects Prob Poss (FDA)d

Suprofen Kidney toxicity No Noc

Temafloxacin Haemolytic-uraemic syndrome Poss Noe

Triazolam Psychiatric effects Yes Nof

Zomepirac Anaphylactic reactions Yes Noc

Apa Anaphylactic reactions Yes Yes (FDA)g Yesh

Dilevalol Liver damage Yes Yes (FDA) Yesi

Fenclofenac GI and skin toxicity and carcinogenicity Yes Yes (FDA and UK) Yesj

Feprazone GI toxicity Yes Yes (UK and poss FDA) Possk

Indoprofen GI toxicity Yes Yes (FDA and UK) Yesl

Indomethacin-in-OROS GI toxicity No No No

Metipranolol Granulomatous anterior uveitis No No No

Perhexiline Peripheral neuropathy and liver damage Yesm Yes (FDA) Nob

Terodiline Torsades de pointes No No Non

Zimeldine Peripheral neuropathy Posso No Nop

PPMT ¼ reason for safety withdrawal present in pre-market testing data submitted (in all but one case this was human trial data);

DPMT ¼ reason for safety withdrawal detected by regulatory agency in pre-market testing data before marketing in that country;

FDANA ¼ FDA non-approved the drug because it detected the safety problem, which subsequently led to withdrawal, from review of

pre-market testing data submitted to it; Poss ¼ possibly; Prob ¼ probably.aUS Congress (1976).bInterview, current senior FDA regulator, 2 May 2001.cAbraham (1995).dUS Committee on Government Operations (1987).eFDA (1992, pp. 144–145); Greene-Reed (1991, p. 300); Interview, FDA epidemiologist, 10 May 2001.fAbraham and Sheppard (1999).gDavis and Pearce (1972).hInterview, senior industry scientist, 22 February 2001.iTemple (1990).jLetter, Acting Director, Division of Oncology and Radiopharmaceutical Drug Products, FDA to RT French Company, 21

February, 1984.kInterviews, industry scientists, 4 October 2000 and 19 January 2001.lBarker (1984).mNon-approvable letter, Henry Simmons, Director, Bureau of Drugs, FDA to Merrell Company, 30 June 1971.nAnon (1991); FDA (2002).oAnon (1983).pFDA (1983).

J. Abraham, C. Davis / Social Science & Medicine 61 (2005) 881–892 889

efficacy of nomifensine and suprofen, and the photo-

sensitivity, carcinogenicity and amnesia associated

with benoxaprofen, suprofen and triazolam, respec-

tively. These concerns were vindicated after marketing

though not the main reason for market withdrawal

(Abraham, 1995; Abraham & Sheppard, 1999; US

Congress, 1986). The FDA did not approve metiprano-

lol (0.1%) or terodiline because of poor efficacy or

metipranolol (0.6%) because of safety concerns other

than those which ultimately led to market withdrawal in

the UK (Botstein, 1991; Carreras, 1991; Chambers,

1989).

Hence, these case histories do not support the

proposition that the FDA’s more lengthy review times

were unnecessary and inefficient because drug safety

problems typically cannot be detected before marketing.

On the contrary, they strongly undermine that proposi-

tion and imply: (i) that a significant proportion of safety

problems were, and could be, detected by regulators

before marketing approval; and (ii) that FDA’s more

lengthy review times may be explained by the fact that

they were applying, for the most part, more stringent

safety and efficacy standards than their UK counter-

parts.

ARTICLE IN PRESSJ. Abraham, C. Davis / Social Science & Medicine 61 (2005) 881–892890

Discussion, conclusion and policy implications

From the 1970s to the early 1990s, the UK and the US

came to represent two contrasting models of drug

regulation. For many commentators, especially those

concerned about the ‘drug lag’, the UK model of

relatively quick marketing approval was favoured over

the slower and more cautious approach at the FDA

(Bakke et al, 1984; Kaitin et al, 1989; Wardell, 1973).

The justifications given for this view by these commen-

tators and UK regulators were that: (1) the drug safety

withdrawal rates in the two countries were about the

same so no safety protection overall was gained from the

FDA’s caution; (2) most ADRs can only be detected

post-marketing among a large population, so there is

little point in undertaking the more exhaustive pre-

market regulatory checks of the FDA; and (3) approval

is only provisional in the sense that new drugs which

turn out to be problematic on the market can be

detected by post-marketing surveillance and promptly

withdrawn.

On these bases, the FDA model of regulation in the

1971–1992 period was heavily criticised by the pharma-

ceutical industry and some academics and professionals

(often industry-funded) for being irrational and ineffi-

cient because it ostensibly delayed the availability of

beneficial drugs to US patients. Since 1992, this

perspective has become influential in the US Congress

and among some FDA Commissioners (Ault, 2003;

Kessler, Hass, Felden, Lumpkin, & Temple, 1996).

Consequently, regulatory policy has come to be

characterised by a heavy emphasis on the acceleration

of pre-market regulatory review and drug approvals at

the behest of the 1992 PDUFA and the 1997 Moder-

nisation Act in the US, which made the FDA partially

dependent on industry fees in exchange for accelerated

review times. In other words, post-1992 US regula-

tory policy has sought to follow the UK model. Average

approval times for NASs dropped by more than

half between 1988 and 1998 (Kaitin & Healy, 2000,

p. 12).

Yet our research demonstrates that the bases for these

shifts in policy direction since 1992, were unfounded.

Compared with the UK, the FDA’s more thorough pre-

market regulatory checks between 1971 and 1992

provided greater safety of drugs for patients because:

(1) the number and rates of US drug safety withdrawals

were half that in the UK; and (2) those checks prevented

the marketing of some unsafe drugs in the US because

they identified the safety problems due to which they

were withdrawn in the UK. Furthermore, post-market-

ing surveillance is no substitute for rigorous pre-market

regulatory review because of its poor quality and slow

rate of detection of ADRs. Moreover, the fact that the

FDA withdrew unsafe drugs faster than the UK

regulatory authorities suggests that a political culture

of more rigorous and pre-market safety checks may go

hand in hand with a more patient protective approach to

post-marketing surveillance, and that the reputed super-

ior vigilance of the UK post-marketing surveillance

system is probably a myth.

From our findings for 1971–1992, one may predict

that the post-1992 policy shift at the FDA will be

associated with an increase in safety problems with

drugs approved after 1992. There is some evidence to

support this. While there were nine US drug safety

withdrawals in the 21 years between 1971 and 1992,

there have been 10 in the 4 years between September

1997 and August 2001 (Olson, 2002). Whether this is

explained by FDA’s recent approval of more unsafe

drugs or by improvement in post-marketing detection

and withdrawal of unsafe drugs can only be determined

conclusively by an analysis of the regulatory case

histories for the post-1992 period. However, the former

explanation seems more likely in view of the finding by

the US Inspector General’s Office in 2003 that 40% of

FDA scientists believe that the review process has

worsened since 1992 because they do not have enough

time to adequately analyse data (Ault, 2003, p. 380).

Furthermore, Olson (2002) analysed the ADRs reported

to the FDA for the 141 NASs approved by the FDA

between 1990 and 1995 and found that reductions in

FDA review times were associated with increases in

both ADRs requiring hospitalisation and ADRs result-

ing in death.

Regarding the argument that the FDA’s more

cautious approach before 1992 delayed patients’ access

to important therapies, Schweitzer, Schweitzer, and

Sourty-Le Guellec (1996) analysed the approval dates

of 34 pharmaceuticals, marketed in the G-7 countries

plus Switzerland between 1970 and 1988, and designated

especially therapeutically significant (at the time of their

approval) by panels of doctors and pharmacists in the

US and France. The FDA was found to approve more

of these drugs before the UK regulatory authorities and

ranked third out of the eight countries in approving

these drugs on to the market. Hence, the evidence

suggests that current drug regulatory policy develop-

ments in the US and UK may be based on false

premises. If so, then they are irrational and inconsistent

with the legislative purpose of drug regulation in both

countries to protect public health by ensuring that drugs

on the market are safe and effective.

Acknowledgements

Thanks to Peter Davis and two anonymous referees

for comments on a previous version of this paper, to the

UK Economic and Social Research Council (Grant

R000237658) for funding the research, and to all those

who gave their time in interviews.

ARTICLE IN PRESSJ. Abraham, C. Davis / Social Science & Medicine 61 (2005) 881–892 891

References

Abraham, J. (1994). Bias in science and medical knowledge:

The Opren controversy. Sociology, 28, 717–736.

Abraham, J. (1995). Science, politics and the pharmaceutical

industry: Controversy and bias in drug regulation. London/

New York: UCL/St. Martins Press.

Abraham, J. (2002). Transnational industrial power, the

medical profession and the regulatory state: Adverse

drug reactions and the crisis over Halcion in The Nether-

lands and the UK. Social Science & Medicine, 55,

1671–1690.

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

Europe: Competition, expertise and public health. London:

Routledge.

Abraham, J., & Reed, T. (2001). Trading risks for markets: The

international harmonization of pharmaceuticals. Health,

Risk & Society, 3, 113–128.

Abraham, J., & Reed, T. (2003). Reshaping the carcinogenic

risk assessment of medicines: International harmonization

for drug safety, industry/regulator efficiency or both? Social

Science & Medicine, 57, 195–204.

Abraham, J., & Sheppard, J. (1999). The therapeutic nightmare:

The battle over the world’s most controversial sleeping pill.

London: Earthscan.

Andersson, F. (1992). The drug lag issue: The debate seen from

an international perspective. International Journal of Health

Services, 22, 53–72.

Anon. (1983). Zelmid Swedish ADR report. Scrip, 798, 18.

Anon. (1986). Hoechst withdraws nomifensine. Scrip, 1072, 22.

Anon. (1988). Editorial lessons from nomifensine. Lancet, II,

1059–1060.

Anon. (1991). Terodiline withdrawn worldwide. Scrip, 1653, 19.

Anon. (1992). Temafloxacin—anatomy of a withdrawal. Scrip,

1737, 20.

Ault, A. (2003). McClellan’s FDA: Boon to industry, con-

sumers, or both? New FDA chief plans to speed drug

approval and reduce costs. Lancet, 362, 379–380.

Bakke, O. M., Manocchia, M., de Abajo, F., Kaitin, K. I., &

Lasagna, L. (1995). Drug safety discontinuations in

the UK, US, and Spain from 1974 through 1993: A

regulatory perspective. Clinical Pharmacology & Therape-

tics, 58, 108–117.

Bakke, O. M., Wardell, W. M., & Lasagna, L. (1984). Drug

discontinuations in the UK and the US, 1964 to 1983:

Issues of safety. Clinical Pharmacology & Therapeutics, 35,

559–567.

Barker, E. F. (1984). Endyne, internal FDA memorandum

regarding new drug application for indoprofen, 25 May

(END 0154).

Botstein, P. (1991). FDA non-approvable letter by deputy

director of Office of Drug Evaluation, 29 July.

Carreras, J. A. (1991). Medical officer’s review of NDA 20-196

(metipranolol 0.1%), 12 October, FDA (OPT 0015).

CAST Investigators. (1989). Preliminary report: Effect of

encainide and flecainide on mortality in a randomized trial

of arrhythmia suppression after myocardial infarction. New

England Journal of Medicine, 321, 406–412.

Chambers, W. A. (1989). Medical officer’s review of

NDA 19-911 (metipranolol 0.6%), 21 February, FDA

(MET 002).

CSM. (1986). CSM update: Non-steroidal anti-inflammatory

drugs and serious gastrointestinal adverse reactions—2.

British Medical Journal, 292, 1190–1191.

Davis, B., & Pearce, D. R. (1972). An introduction to

Althesin (CT 1341). Postgraduate Medical Journal, June

(Supplement), 13–17.

FDA. (1983). Minutes internal FDA meeting regarding zimeldine

new drug application, 16 September.

FDA. (1992). Forty-second meeting of the Anti-Infective Drugs

Advisory Committee, July 10, Rockville, Maryland.

FDA. (2002). FOIA release, MIC00275, 21 June.

Greene-Reed, T. (1991). FDA Medical Officer’s review of

temafloxacin, 28 August.

House of Commons. (1984). Parliamentary Debates, Hansard,

61, 12 June, w.col. pp. 461–462.

Inman, W. (1999). Don’t tell the patient: Behind the drug safety

net. Bishops Waltham, Hampshire/ Los Angeles: Highland

Park Productions.

Jefferys, D. B., Leakey, D., Lewis, J. A., Payne, S., & Rawlins,

M. D. (1998). New active substances authorised in the

United Kingdom between 1972 and 1994. British Journal of

Clinical Pharmacology, 45, 151–156.

Kaitin, K. I., & Healy, E. M. (2000). The new drug approvals of

1996, 1997 and 1998: Drug development trends in the user

fee era. Drug Information Journal, 34, 1–14.

Kaitin, K. I., Mattison, N., Northington, F. K., & Lasagna, L.

(1989). The drug lag: An update of new drug introductions

in the US and in the UK, 1977 through 1987. Clinical

Pharmacology and Therapeutics, 46, 121–138.

Kessler, D. A., Hass, A. E., Felden, K. L., Lumpkin, M., &

Temple, R. (1996). Approval of new drugs in the US:

Comparison with the UK, Germany and Japan. Journal of

the American Medical Association, 276, 1826–1831.

LaPlane, D., & Bousser, M. G. (1981). Polyneuropathy during

perhexiline maleate Therapy. International Journal of

Neurology, 15, 293–300.

Lexchin, J. (1995). The timeliness of new drug approvals in

Canada: A rejoinder. International Journal of Health

Services, 25, 167–171.

Mann, R. D. (1988). Lessons from nomifensine. Lancet, II,

1490–1491.

Olson, M. K. (1999). Agency rulemaking, political influences,

regulation and industry compliance. Journal of Law,

Economics and Organization, 15, 573–601.

Olson, M. K. (2002). Pharmaceutical policy change and the

safety of new drugs. Journal of Law and Economics, XLV,

615–642.

Parker, J. (1989). Who has a drug lag? Managerial and Decision

Economics, 10, 299–309.

Pieterse, A. S., Rowland, R., & Dunn, D. (1983). Perhexiline

maleate induced cirrhosis. Pathology, 15, 201–203.

Rawson, N. S. B., & Kaitin, K. I. (2003). Canadian and US

drug approval times and safety considerations. The Annals

of Pharmacotherapy, 37, 1403–1408.

Schweitzer, S. O., Schweitzer, M. E., & Sourty-Le Guellec, M-J.

(1996). Is there a US drug lag? The timing of new

pharmaceutical approvals in the G-7 countries and Switzer-

land. Medical Care Research and Review, 53, 162–178.

Temple, R. (1990). Dilevalol. Internal FDA memorandum from

the Director of Drug Evalation I, to C. Graham and R.

Lipicky, April 11.

ARTICLE IN PRESSJ. Abraham, C. Davis / Social Science & Medicine 61 (2005) 881–892892

US Committee on Government Operations. (1987). FDA’s

Regulation of the new drug Merital, Fifteenth Report by the

Committee on Government Operations. Washington: GPO.

US Congress. (1976). FDA’s regulation of the drug ‘‘Triazure’’.

Hearing before a Subcommittee of the Committee on

Government Operations, 27 October. Washington DC:

GPO.

US Congress. (1986). Oversight of the new drug review process

and FDA’s regulation of Merital. Hearing before a Sub-

Committee of the Committee on Government Operations,

22 May, Washington DC: GPO.

US District Court. (1984). United States of America v.

SmithKline et al. Change of Plea Hearing before the

Honorable Edward N Cahn, Eastern District of Pennsylva-

nia, Docket No: CR 84-227, 12 December.

Waller, P. C., Wood, S. M., Langman, M. J. S., Breckenridge,

A. M., & Rawlins, M. D. (1992). Review of company

postmarketing surveillance studies. British Medical Journal,

304, 1470–1472.

Wardell, W. (1973). Introduction of new therapeutic drugs

in the US and Great Britain: An international com-

parison. Clinical Pharmacology and Therapeutics, 14,

773–790.

Wiktorowicz, M. E. (2003). Emergent patterns in the regulation

of pharmaceuticals: Institutions and interests in the US,

Canada, Britain and France. Journal of Health Politics,

Policy and Law, 28, 615–658.

Wild, R. N. (1992). Micturin and torsades de pointes—

experience of a post-marketing alert. Pharmacoepidemiology

and Drug Safety, 1, 147–150.