pharmacoepidemiology and drug safety 2016

Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/pds.4011

ORIGINAL REPORT

Benefit-risk reassessment of medicines: a retrospective analysis of all

safety-related referral procedures in Europe during 2001–2012
Jacoline C. Bouvy1,2, Lotte Huinink2 and Marie L. De Bruin1*
1
Division of Pharmacoepidemiology and Clinical Pharmacology, Faculty of Science, Utrecht University, Utrecht, The Netherlands
2
Institute for Medical Technology Assessment (iMTA), Faculty of Health, Policy and Management, Erasmus University Rotterdam,
Rotterdam, The Netherlands

ABSTRACT
Purpose The aim of this study was to determine the outcomes and timing within the product life cycle of all benefit-risk reassessment
procedures for marketed products that were completed by the committee for medicinal product for human use during 2001–2012.
Methods A cohort of all referral procedures for benefit-risk reassessment (Article 20, Article 31, Article 36, Article 107 procedures) for
which committee for medicinal product for human use issued an opinion between 1 January 2001 and 31 December 2012 was created.
The European Medicines Agency website and the Dutch Medicines Evaluation Board website were used to collect all data.
Results There were a total of 73 benefit-risk reassessments during the study period; 61 reassessments for a single product and 12
reassessments for multiple products or an entire product class. Nineteen reassessments resulted in the recommendation to remove the product
from the market. On average, a benefit-risk reassessment was performed 18.7 years after the product was first marketed. Seventeen products
were marketed 5 years or less when the reassessment procedure was completed; six of these products were subsequently removed from the
market.
Conclusions The majority of all benefit-risk reassessments that were performed during the study period did not result in removing the
product from the market, but rather, in confirming the positive benefit-risk of the product, conditional to changes to the product’s marketing
authorisation. About half of all products that were removed from the market during the 2000s had been marketed for more than 20 years.
Copyright © 2016 John Wiley & Sons, Ltd.

key words—drug regulation; referrals; pharmacovigilance; benefit-risk assessment; regulatory science; pharmacoepidemiology

Received 20 November 2015; Revised 04 March 2016; Accepted 19 March 2016

INTRODUCTION is monitored throughout the product life cycle, and

the occurrence of unexpected or serious adverse events
In Europe, a market license for a new medicinal prod- in the post-marketing setting may compel a reassess-
uct is only granted when the product’s benefit-risk ment of a product’s benefit-risk profile. Until 2012,
profile is deemed positive by expert committees, benefit-risk reassessments for all medicinal products
which means that there is sufficient evidence in sup- that are authorised in the European Union (EU)—
port of the product’s quality, safety and efficacy. Such regardless of whether products are registered in all
data originate from preclinical and clinical studies that EU countries or only in a number of member
are usually performed in limited and selected patient states3,4—were performed by the committee for me-
populations1 and are designed to establish efficacy dicinal product for human use (CHMP), but since
rather than safety. As a result, less frequent adverse 2012, these reassessments are performed by the
drug reactions of the medicine, or those adverse drug pharmacovigilance risk assessment committee
reactions occurring only after long-term treatment, re- (PRAC). During recent years, calls from industry,
main unobserved.2 Therefore, the safety of medicines regulatory authorities and academics for a switch to
a more rational and explicit decision-making proce-
dure with regard to the benefit-risk assessment of
medicines have been made5–10 in order to improve
* Correspondence to: M. L. De Bruin, Division of Pharmacoepidemiology and
Clinical Pharmacology, Faculty of Science, Utrecht University, Utrecht, The the transparency and accountability of regulatory
Netherlands. E-mail: m.l.debruin@uu.nl decision-making.

Copyright © 2016 John Wiley & Sons, Ltd.

 j. c. bouvy et al.
The natures of those safety (or efficacy) issues benefit-risk reassessments. Medicinal products in
that trigger a benefit-risk reassessment procedure Europe can be authorised through two different path-
(a so-called referral procedure; Table 1) as well as ways and are either centrally authorised products
the outcomes of these reassessments have not been (authorised in entire EU by the EMA) or non-centrally
systematically studied. Previous studies have assessed authorised products (authorised in one or more mem-
market withdrawals in the USA,11 in the UK,12,13 ber states by a national regulatory authority). As a
or more recently, safety withdrawals in the EU,14 result, there are different so-called referral procedures
but to the best of our knowledge, no recent studies (Table 1), but the benefit-risk reassessment for all
have assessed the outcomes of all benefit-risk these procedures is essentially identical, and all
reassessments—including those reassessments that do reassessments were performed by the CHMP during
not result in a market withdrawal. Furthermore, most the study period.
studies in the past have focused on new active sub-
stances or relatively newly marketed products, but it
was found that the 20-year survival rate for medicines Cohort
marketed since 1972 in the UK was 84%.12 Therefore, We created a cohort of all benefit-risk reassessments
it is important to study the benefit-risk reassessments (i.e. referrals) of medicinal products that were
of all marketed medicines, including those medicines triggered by safety and/or effectiveness concerns and
that have been marketed for many years, as well that were finalised (i.e. a CHMP opinion was issued)
as to consider all possible outcomes of such between 1 January 2001 and 31 December 2012
reassessments, and not merely focus on market (Figure 1). We scanned the minutes of all monthly
withdrawals. The aim of this study, therefore, was to CHMP meetings during the study period—that are
determine the outcomes and timing of all benefit-risk publicly available on the EMA website—to identify
reassessment procedures that were performed by all products that were subjected to an Article 20,
the European Medicines Agency (EMA) during Article 31, Article 36 or Article 107 referral procedure
2001–2012. (Table 1 and Figure 1). Subsequently, the referral data-
base—available on the EMA website as well—was
METHODS used to check the list compiled based on the CHMP
minutes in order to identify any inconsistencies
Benefit-risk reassessment procedures and/or missing information. Furthermore, the EMA
When a benefit-risk reassessment procedure is started, kindly provided a list of referral procedures for CAPs
a medicine—or a medicine class—is referred to the upon request in a personal communication. We
EMA such that (until July 2012) the committee for organised all procedures by international nonpropri-
medicinal products for human use (CHMP) could etary name (INN) and differentiated between proce-
make a harmonised recommendation regarding the dures in which one INN was evaluated (single INN
product’s benefit-risk in light of new evidence. Since reassessments), and procedures in which multiple
July 2012, the PRAC performs all safety-related INNs or a whole class of products were evaluated

Table 1. Types of referral procedures in Europe

Referral procedure Description of procedure

Article 20 Reg. (EC) 726/2004 This type of referral is triggered for medicines authorised via the centralised procedure in case of manufacturing
or safety issues. The procedure is organised by product name level; this means that if one INN is registered under
different brand names and formulations, each form of the product will have a separate procedure.
Article 31 Dir. 2001/83/EC This type of referral is triggered when the interest of the community is involved, following concerns relating to
the quality, safety or efficacy of a medicine not authorised through the centralised procedure. The procedure is
organised by INN level or by product class level.
Article 36 Dir. 2001/83/EC This referral applied for products that were authorised through the non-centralised route. It was triggered when a
Member State considered that action (variation, suspension or withdrawal) was needed on the grounds of the need
to protect public health. This procedure has been replaced by Art. 107i and Art. 31. The procedure is organised by
INN level or by product class level.
Article 107 Dir. 2001/83/EC This referral was triggered when a Member State varied, suspended or revoked the MA for a medicine in its
territory because of a safety issue. This procedure has been replaced by Art. 107i in July 2012. The procedure
is organised by INN level or by product class level.
Article 107i Dir. 2010/84/EU (Amd.) This referral is triggered when a Member State or the European Commission consider that urgent action is
necessary because of a safety issue. This procedure applicable for both centrally and non-centrally authorised
products and is implemented per July 2012. The procedure is organised by INN level or by product class level.

Copyright © 2016 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2016
DOI: 10.1002/pds
 benefit-risk reassessment of medicines

Figure 1. Data selection

(class reassessments). We excluded new procedures Medicines Evaluation Board (www.cbg-meb.nl). In a

that were performed for a product that was already number of cases, no exact year of first authorisation
suspended (i.e. not marketed) at the start of the was given, but only the decade was stated (e.g. has
procedure. been available in member states since the 1960s).
A referral procedure for a marketed product can In such cases, we used the decade’s middle year (e.g.
have four possible types of outcomes: the CHMP can 1965) as an approximation for the year of first market-
recommend to (i) maintain the marketing authorisation ing authorisation in an EU country.
of the product without any changes; recommend to (ii)
change the conditions of the product’s marketing RESULTS
authorisation; recommend the (iii) suspension of the
marketing authorisation of the product until new evi- Outcomes of benefit-risk reassessments
dence for the product’s positive benefit-risk is gener- We identified a total of 120 referral procedures with
ated; or recommend the (iv) revocation of a product’s a CHMP opinion that was issued between 1 January
marketing authorisation. We collected all CHMP deci- 2001 and 31 December 2012 and excluded seven
sions, as well as the reasons for the reassessment, from procedures because of misclassification, resulting
the publicly available referral assessment reports in 113 procedures in total (Figure 1). As multiple
(www.ema.europa.eu). Other data that we collected in- procedures can be started for the same active
cluded the date of the first market authorisation for all ingredient simultaneously and such procedures are
products that underwent a reassessment, as well as the combined in one CHMP review, there were 73 con-
year in which the CHMP Opinion was issued. In 22% solidated procedures, composed of 61 single INN
of cases, the date of first marketing authorisation was benefit-risk reassessments and 12 reassessments
not stated in the assessment reports. In these instances, where either a number of different INNs or an entire
we extracted the date of the first marketing authorisa- medicine class were reviewed for the same safety
tion in the Netherlands from the website of the Dutch issue (Figure 1).

Copyright © 2016 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2016
DOI: 10.1002/pds
 j. c. bouvy et al.
The most common decision (N = 39; 64% of all sin- had been marketed 10 years or less. The products that
gle INN reassessments) made by the CHMP following were withdrawn from the market following a benefit-
a benefit-risk reassessment was to allow the product to risk reassessment had been marketed on average for
remain on the market, but under the condition that 27.4 years (N = 9; median 35 years), and the suspended
changes to the product’s marketing authorisation products had been marketed for 20.3 years (N = 10;
would be made (Table 2). In only two cases (3%) median 12 years) (Table 2).
did, the CHMP decide to maintain the marketing There were 17 products in total for which a benefit-
authorisation of a product without any changes. Fur- risk reassessment procedure was finalised within the
thermore, in 16% (N = 10) of the single INN first 5 years of the marketing authorisation (Table 2).
reassessments, the CHMP decided to suspend the Three products were withdrawn, and three products
product’s marketing authorisation. Finally, in 16% were suspended, but 11 of these newly marketed prod-
(N = 10) of the single INN reassessments, the CHMP ucts were allowed to remain on the market; for one
decided either to revoke the product’s marketing product, the CHMP decided that no change to the mar-
authorisation, or the market authorisation holder keting authorisation was necessary, and 10 products
decided to withdraw the product before the reassess- required a change to the conditions of the marketing
ment was finalised. One out of the 12 (8.3%) class authorisation. One of the products with a first
benefit-risk reassessments resulted in the suspension benefit-risk reassessment 4 years after the first market-
of one of the products in the class, and although none ing authorisation (sibutramine; marketed in 1999, first
of the class reassessments resulted in the revocation of referral in 2002) was eventually withdrawn from the
a marketing authorisation by the CHMP, in one case, market in 2010, 11 years after the first marketing
two of the products belonging to the class that was authorisation.
reassessed were voluntarily withdrawn by the manu- In the majority of the single INN assessments, the
facturers before the reassessment procedure was positive benefit-risk profile of the product was con-
finalised. firmed (Table 3). In two cases, the MA was main-
tained without any changes, and in 39 reassessments,
the recommendation that the MA needed to be
Single international nonproprietary name benefit-risk changed was made. Table 3 lists all these benefit-risk
reassessments reassessments, including the year of first MA, the
The average number of years that passed between first trigger for the reassessment procedure and the classifi-
marketing authorisation and the reassessment for sin- cation of the adverse event that triggered the referral
gle products was 18.8 year (median 13.0 years; (MedDRA system organ class (SOC)). The triggers
N = 59). On average, the timing of a benefit-risk reas- were general disorders and administration site
sessment that resulted in the decision to maintain the conditions (nine reassessments), cardiac disorders
marketing authorisation without any changes was (eight reassessments);,hepatobiliary disorders (six
30 years after market entry (N = 2; Table 2). The aver- reassessments); nervous system disorders (five
age time between first authorisation and the end of the reassessments); neoplasms benign, malignant and
reassessment was 15.8 years (N = 38; median unspecified (five reassessments); gastrointestinal dis-
13.0 years) for the products that required a change to orders (three reassessments); skin and subcutaneous
the conditions of the marketing authorisation. Further- disorders (three reassessments); immune system
more, 10 of these products had been marketed 5 years disorders (two reassessments); congenital, familial
or less at the time of the reassessment, and 16 products and genetic disorders (two reassessments); vascular

Table 2. Outcomes of all benefit-risk reassessments during 2001–2012

Maintain MA Change MA Withdraw MA Suspend MA Total

Number of single INN reassessments: 2 39 10 10 61

Reassessments within 6 years of first MA 1 10 3 3 17
Years between first MA—referral, mean 30.0 15.8* 27.4† 20.3 18.4‡
Years between first MA—referral, median 30.0 13.0* 35.0† 12.0 13.0‡
Number of class reassessments 0 10 1 1 12

MA, marketing authorization; INN, international nonproprietary name.

*N = 38; one date could not be retrieved.
†
N = 9; one date could not be retrieved.
‡
N = 59; two dates missing.

Copyright © 2016 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2016
DOI: 10.1002/pds
 benefit-risk reassessment of medicines
Table 3. All single INN assessments that resulted in the recommendation to maintain or change the MA (N = 41)
Year Year first
Product referral Outcome MA Suspected ADR SOC

Rotavirus vaccine 2010 Maintain 2006 Contamination PCV-1 virus General disorders and administration
(Rotarix) site conditions
Pholcodine 2011 Maintain 1955 IgE-sensitisation to neuromuscular blocking agents Immune system disorders
resulting in increased risk anaphylactic reactions
Cisapride 2001 Change 1988 QT-prolongation: risk severe arrhythmia Cardiac disorders
Calcitonin 2002 Change 1973 Doubt efficacy General disorders and administration
site conditions
Bupropion 2002 Change 1999 Seizure and fatalities Nervous system disorders
Sibutramine 2002 Change 1999 Two-fatal cases due to cardiovascular events Cardiac disorders
Loratadine 2003 Change 1989 Hypospadias in newborn when used during Congenital, familial and genetic
(+ pseudoephedrine) pregnancy disorders
Nimesulide 2003 Change 1985 Serious liver problems Hepatobiliary disorders
Paroxetine 2004 Change 1991 Suicidal behaviour adolescents Psychiatric disorders
Gadobutrol 2006 Change 2000 Doubt efficacy General disorders and administration
site conditions
Hepatitis B vaccine 2006 Change — Reassess benefits General disorders and administration
site conditions
Bicalutamide 2007 Change 1995 Cardiovascular complications Cardiac disorders
Nimesulide 2007 Change 1985 Serious liver problems Hepatobiliary disorders
Piroxicam 2007 Change 1991 Gastrointestinal side effects Gastrointestinal disorders
Etoricoxib 2008 Change 2002 Cardiovascular risks Cardiac disorders
Moxifloxacin 2008 Change 2002 Hepatotoxicity Hepatobiliary disorders
Norfloxacin 2008 Change 1985 Reassess benefits Administration site conditions
Methylphenidate 2009 Change 1982 Cardiovascular and cerebrovascular events Cardiac and Nervous system disorders
in children
Valproate 2009 Change 1965 Doubt efficacy in manic episodes General disorders and administration
site conditions
Becaplermin (gel) 2010 Change 1999 Increased risk cancer Neoplasms benign, malignant and
unspecified (incl cysts and polyps)
Bevacizumab 2010 Change 2005 Doubt efficacy in combination therapy General disorders and administration
(systemic) site conditions
Ketoprofen topical 2010 Change 1978 Skin photosensitivity Skin and subcutaneous tissue disorders
Modafinil 2010 Change 1992 Neuropsychiatric disorders + skin hypersensitivity Psychiatric and Skin and subcutaneous
tissue disorders
Natalizumab 2010 Change 2006 PML after >2 years of use (1 : 1000) Infections and infestations
Saquinavir 2010 Change 1996 QT-prolongation: risk severe arrhythmia Cardiac disorders
Somatropin 2011 Change 1985 Increased mortality General disorders and administration
site conditions
Dexrazoxane 2011 Change 2006 Risk cancer (AML, MSD) in paediatric patients Neoplasms benign, malignant and
unspecified (incl cysts and polyps)
Dronedarone 2011 Change 2009 Liver-, lung- and cardiovascular events Cardiac, Hepatobiliary, and Respiratory,
thoracic and mediastinal disorders
Lenalidomide 2011 Change 2007 Increased risk cancer (secondary) Neoplasms benign, malignant and
unspecified (incl cysts and polyps)
Nimesulide 2011 Change 1985 Gastrointestinal + hepatic safety concerns Gastrointestinal and Hepatobiliary
disorders
Pioglitazone 2011 Change 2000 Increased risk cancer (bladder) Neoplasms benign, malignant and
(and combinations) unspecified (incl cysts and polyps)
Orlistat 2012 Change 1998 Severe hepatotoxicity (rare) Hepatobiliary disorders
Calcitonin 2012 Change 1973 Increased risk cancer Neoplasms benign, malignant and
unspecified (incl cysts and polyps)
Doripenem 2012 Change 2008 Not effective at currently approved dose General disorders and administration
site conditions
Fibrin sealants 2012 Change 1997 Risk air embolism Vascular disorders
Fingolimod 2012 Change 2011 Cardiovascular risks Cardiac disorders
Influenza vaccine 2012 Change 2008 Narcolepsy children and adolescents Nervous system disorders
MMRV vaccine 2012 Change 2004 Risk congenital rubella syndrome Congenital, familial and genetic
= malformations newborn disorders
Strontium ranelate 2012 Change 2004 Cardiovascular + skin reactions Vascular and Skin and subcutaneous
tissue disorders
Tolperisone 2012 Change 1960 Gastrointestinal + neurological Gastrointestinal, Immune system, and
events; hypersensitivity Nervous system disorders
Trimetazidine 2012 Change 1970 Worsening Parkinson Nervous system disorders

MA, marketing authorization; ADR, adverse drug reaction; SOC, system organ class.
The nature of the changes to the conditions of the marketing authorization for each referral procedure can be found in publicly available documents on the
EMA website.

Copyright © 2016 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2016
DOI: 10.1002/pds
 j. c. bouvy et al.
disorders (two reassessments); psychiatric disorders procedures, we did not include any cases where prod-
(two reassessments); respiratory disorders (onw ucts were withdrawn for other reasons such a market
reassessment); infections and infestations (one withdrawals due to commercial reasons. We also did
reassessment). not include safety reviews that were performed in the
context of a periodic safety update report that will oc-
cur more frequently and might result in label changes.
Suspended and withdrawn products We did not assess the nature of the condition changes
Twenty-two benefit-risk reassessments resulted in the to a product’s MA, but some changes could be more
suspension or revocation of the product’s marketing ‘severe’ than others and range from adding informa-
authorisation (Table 4). As for two of these products, tion on a safety issue to the product’s label to
the suspension was lifted as the result of new evidence, restricting the patient indication. Notwithstanding, we
and it follows that for 19 out of 73 benefit-risk assess- included all benefit-risk reassessments that have been
ments performed the outcome was the removal of performed during a 12-year period and have also in-
the product from the market—decided either by the cluded those procedures where a product was not with-
CHMP or through a voluntary withdrawal of the prod- drawn from the market. Therefore, our study provides
uct(s) by the market authorisation holder. a more complete picture of how CHMP made deci-
The majority of the 19 suspended or withdrawn sions regarding the benefit-risk of medicines during a
products for which we could retrieve the date of first 12-year period.
marketing authorisation had been marketed for An analysis of the market withdrawals of new mo-
10 years or more when the product was removed from lecular entities in the USA during 1980–2009 found
the market, and nine products had been marketed for that the average time until withdrawal was 5.9 years.11
more than 20 years (Figure 2), whereas six products Another study that assessed safety withdrawals in the
were removed from the market within the first 5 years UK during 1971–1992 found that the average time un-
of their market introduction. til withdrawal was 58 months (4.8 years)13 and a third
study found that the average time until withdrawal of
a new active substance from the UK market during
DISCUSSION 1972–1994 was 4.9 years.12 While safety withdrawals
only concerned a subset of the regulatory outcomes
In the majority of all benefit-risk reassessment proce- we studied, we found an average time until withdrawal
dures completed by CHMP between 1 January 2001 of 27.4 years for the 10 products that were withdrawn
and 31 December 2012, it was concluded that the from the market during the 2000s as a result of a refer-
product’s benefit-risk profile remained positive, but ral procedure. Although our sample differs in a num-
that changes to the conditions of the marketing autho- ber of ways from those used in the previous studies,
risation were necessary. The results of this study dem- notwithstanding, these differences indicate that the
onstrate that the majority of products for which types of products that were removed from the market
concerns over product safety or efficacy arose post- from the 1970s until the 1990s were mostly newly
authorisation were allowed to remain on the market marketed products, whereas the majority of products
conditional to changes to the product’s marketing that were removed from the market during the 2000s
authorisation. In addition, about half of the products after a referral procedure were products that had
that were removed from the market during the study already been marketed for many years. A number of
period had been marketed for more than 20 years. other publications have reported timing to market
Our study has several limitations. All data were withdrawals in various regions,16–18 but none of these
manually extracted from documents available through publications used very similar samples, making it
the EMA website. Even though our data cover a long problematic to compare our findings. Comparing tim-
time period, our findings are not necessarily predictive ings of products that were withdrawn in the USA dur-
for future benefit-risk reassessments by regulatory ing the same years as covered by our study could be an
authorities, especially because PRAC now performs area of future research.
all safety referrals. A preliminary analysis of all the Unlike many studies that have focused on safety
referrals that have been finalised in PRAC during its withdrawals of medicines, we assessed all possible
first 18 months of operation, however, shows a similar outcomes of benefit-risk reassessments and found that
proportion of outcomes (approximately 30% with- although newly marketed medicines (marketed 5 years
drawals, and 70% label changes).15 As we intended or less) were still regularly subjected to a reassessment
to study the outcomes of benefit-risk reassessment of the product’s benefit-risk in light of new safety

Copyright © 2016 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2016
DOI: 10.1002/pds
 Table 4. All reassessments that ended in the recommendation to suspend or withdraw the MA (N = 20)

INN ATC code Year referral Outcome Year MA Indication Reasons for start of referral procedure SOC

Cerivastatin C10AA06 2002 Revoked 1998 Hyperlipidaemia CHMP review found cerivastatin was effective in the treatment Musculoskeletal
of hyperlipidaemia but did not present a specific therapeutic and connective
benefit in relation to comparable compounds and there were tissue disorders
concerns related to the risk of rhabdomyolysis.
COX II inhibitors M01AH 2005 Only refecoxib — Rheumatoid Clinical trial data for rofecoxib revealed a risk of thrombotic Cardiac
voluntarily arthiritis; cardiovascular events and resulted in the worldwide withdrawal disorders
withdrawn osteoarthritis of rofecoxib in 2004. Subsequently, other COX II inhibitors
were reviewed by CHMP and were found to have positive
benefit-risk profiles.
Clobutinol R05DB03 2007 Voluntarily 1961 Cough CHMP review started after suspended in Germany in 2007due to Cardiac
withdrawn cardiovascular safety (QT prolongation). MAH decided to disorders
withdraw its product voluntarily from all markets worldwide.
Lumiracoxib M01AH06 2007 Revoked 2005 Rheumatoid CHMP review stated after UK assessment of reports of liver Hepatobiliary

Copyright © 2016 John Wiley & Sons, Ltd.

arthritis; problems. disorders
osteoarthritis
Veralipride N05AL06 2007 Revoked 1979 Hot flushes CHMP review following reports of serious side effects including Psychiatric
depression and anxiety that resulted in product withdrawn in disorders
Spain in 2005.
Dextropropoxyphene N02AC04 2009 Non-parental 1970 Acute + chronic CHMP review started after concerns for some years over the risk Injury, poisoning
form revoked pain of death from overdose—resulted in safety reviews in several and procedural
and parental member states but these have led to different conclusions. CHMP complications
form suspended concluded that the benefits of non-parenteral forms of product do
not outweigh their risks and recommended their withdrawal.
CHMP noted that safety was less of a concern for the parenteral
form. Therefore, the marketing authorisations for the parenteral
form of product were suspended until further data are available
to show that the benefits outweigh the risks.
Iodocasein/thiamine A11 2009 Revoked 1955 Obesity CHMP review started after Italy suspended product in 2009 Endocrine
because of serious cases of hyperthyroidism and thyrotoxicosis. disorders
Product was only marketed in Italy.
Propacetamol N02BE05 2009 Outcome — Pain + fever CHMP review started after product was withdrawn in France Immune system
(powder for injection) procedure not because of the risk of serious hypersensitivity reactions, cases disorders
clear, likely of thrombosis and administration site reactions. Information is
benefit-risk reassessment of medicines

withdrawn missing from referral database, not clear what outcome of referral
was. Product was only marketed in France.
Benfluorex A10BX06 2010 Revoked 1974 Hyperlipidaemia/ In 2007 indication withdrawn for use in patients with high blood Cardiac and
diabetes levels of triglycerides. Safety review in 2009 by France resulted in respiratory
suspension, after reports of cardiac valvulopathy (thickening of the disorders
heart valves) and pulmonary arterial hypertension. Suspension in
Portugal followed, therefore CHMP started review.
Bufexamac M02AA09 2010 Revoked 1975 Topical NSAID CHMP started review after product was withdrawn in Germany Skin and
because of allergic contact reactions. CHMP noted that the data to subcutaneous
support the effectiveness of bufexamac were very limited. Most of tissue disorders
the studies dated from the original development of bufexamac in
the 1970s and 1980s were of a lower standard than that expected
today. Because of this, no evidence of the effectiveness of
bufexamac could be derived from them. In addition, when looking
at the few more recent, controlled studies, the CHMP noted
that the effectiveness of bufexamac had not been shown.

(Continues)

DOI: 10.1002/pds
Pharmacoepidemiology and Drug Safety, 2016
 Table 4. (Continued)
INN ATC code Year referral Outcome Year MA Indication Reasons for start of referral procedure SOC

Lacosamide (syrup) N03AX18 2011 Syrup revoked, 2008 Epilepsy 2011, Quality defect in several batches of product that could General
tablets remain not be solved. disorders and
available administration
site conditions
Carisoprodol M03BA02 2007 Suspended 1959 Short-term lower CHMP review started after suspended in Norway in 2007 Psychiatric
back pain because of increased risk of abuse and addiction and risk of disorders
causing intoxication and psychomotor impairmen.
Conditions for lifting suspension:
Data demonstrating that the products can be used safely taking
into account information on intoxications from poison centres
in Europe.
Data demonstrating convincing efficacy and safety derived from
appropriately designed clinical trials (including an active

Copyright © 2016 John Wiley & Sons, Ltd.

comparator) and data justifying the proposed dose
Rimonabant A08AX01 2008 Suspended; 2006 Obesity Possible risk of depression was known at MA, but data post- Psychiatric
voluntarily marketing experience and ongoing clinical trials indicated that disorders
withdrawn serious psychiatric disorders may be more common than in the
in 2009 clinical trials used in the initial assessment of the medicine.
Therefore, CHMP review started and product was suspended.
In 2008, MAH voluntarily withdrew product.
Efalizumab L04AA21 2009 Withdrawn 2004 Psoriasis Review at request of the EC following reports of serious side Infections and
voluntarily after effects, including three-confirmed cases of PML in patients who infestations
suspension had taken Raptiva for more than 3 years.
In 2009, MAH decided to voluntarily withdraw the product, as
it did not intend to conduct the clinical trials necessary to fulfil
the requirements for lifting the suspension.
Fentanyl N02AB03 2009 Suspended; 2006 Pain MAH recalled all systems from the EU in 2008 as a precautionary Injury, poisoning
(transdermal) non-renewal of measure due to quality defect resulting in the risk of overdose. and procedural
j. c. bouvy et al.

MA in 2011 Conditions for lifting suspension: complications

Suspended until MAH can robustly demonstrate the quality
of the product.
Rosiglitazone A10BG02 2010 Suspended 2000 Diabetes mellitus Since its first authorisation, rosiglitazone had been recognised to Cardiac
(and combinations) type II be associated with fluid retention and increased risk of heart disorders
failure and its cardiovascular safety was kept under close review.
Recent data from clinical trials and post-marketing data supported
an increased cardiovascular risk of rosiglitazone and resulted in
CHMP review.
Conditions for lifting suspension:
Convincing and robust data to identify a patient population in
which the clinical benefits of rosiglitazone-containing products
clearly outweighs the risks.
Sibutramine A08AA10 2010 Suspended 1999 Obesity Sibutramine previously reviewed in 1999 and 2002, following Cardiac
concerns over its safety, especially cardiovascular side effects disorders
(increased blood pressure and heart rate). As a result of these
reviews, in 2002, a study was started to investigate these risks.
Study was stopped prematurely because of increased
cardiovascular problems in treatment group as compared
with placebo.
Conditions for lifting suspension:

(Continues)

DOI: 10.1002/pds
Pharmacoepidemiology and Drug Safety, 2016
 Table 4. (Continued)

INN ATC code Year referral Outcome Year MA Indication Reasons for start of referral procedure SOC

Copyright © 2016 John Wiley & Sons, Ltd.

Convincing data to identify a patient population in which
sustained and clinically important efficacy of sibutramine-
containing products can be demonstrated, and in which the
benefit clearly outweighs its risks.
Modified-release N02A 2010 One of the 2005 Pain Review was started after concerns that the active substance in the Respiratory,
opioids products was capsules could be released too quickly if the capsules are taken thoracic and
suspended with alcohol, leading to side effects such as respiratory depression mediastinal
(Ethirfin) (and inhibition of breathing). disorders
Market authorisation for Ethirfin was not renewed in 2010 after
product was suspended.
Buflomedil CO4AX20 2011 Suspended 1975 PAOD CHMP review started after product suspended in France in 2011 Nervous system
because of serious and sometimes fatal neurological disorders and cardiac
(convulsions + status epilepticus) and cardiac disorders. disorders
Conditions for lifting suspension:
Convincing data to identify a patient population in which the
benefits of buflomedil clearly outweigh its identified risks.
Meprobamate N05BC01 2012 Suspended 1970 Anxiolytic CHMP review started after suspended 2011 in France due to Psychiatric
safety risks including confusion, loss of consciousness, risk of disorders
addiction and cases of overdose leading to coma or death.
benefit-risk reassessment of medicines

Conditions for lifting suspension:

Convincing data to identify a patient population in which the
benefits of meprobamate clearly outweigh its identified risks.

INN, international nonproprietary name; ATC, anatomical therapeutic chemical classification system; MA, marketing authorization; SOC, system organ class, CHMP, committee for medicinal product
for human use; EU, European Union; PAOD, peripheral arterial occlusive disease; NSAID, nonsteroidal anti-inflammatory drug; MAH, marketing authorisation holder.

DOI: 10.1002/pds
Pharmacoepidemiology and Drug Safety, 2016
 j. c. bouvy et al.

Figure 2. Time (in years) elapsed between first marketing authorisation and suspension or withdrawal of product

information, in the majority of these cases CHMP de- Alternatively, this finding could also mean that regula-
cided that the benefit-risk profile of the product tory tolerability of safety risks has shifted as nowadays
remained positive. The triggers for most of the the tools to better manage the benefit-risk of such
benefit-risk reassessments that did not result in the products are available. In addition, this finding also
recommendation to suspend or revoke the MA of demonstrates that there are two distinct cases of a
the product could still be classified as serious safety safety withdrawal: first, the withdrawal of a new prod-
issues (see Table 3 for an overview of all triggers). uct due to unforeseen safety issues and, second, the
We did not include any potential determinants into withdrawal of a product after having been marketed
our analysis that could explain why some products for many years. This finding indicates that important
were considered to have a positive benefit-risk benefit-risk assessments are made well into the life
whereas others were found to have a negative cycle of medicines and not merely in the first few
benefit-risk (Table 4), but it is likely that the effective- years after marketing authorisation.
ness of the product and the number of alternative treat- Our results indicate that the majority of products
ment options available to patients were taken into that were removed from the EU market during the
account in the benefit-risk reassessment. 2000s after a benefit-risk reassessment procedure
A recent study of safety withdrawals during 2002– had been marketing for more than 5 years. Further-
2011 from EU markets found that the time between more, most of the benefit-risk reassessments that
first authorisation and market withdrawal was were performed during the study period did not re-
23.7 years on average, confirming our findings. This sult in removing the product from the market, but
could mean that regulatory standards for market rather, in confirming the positive benefit-risk of
approval have increased since the 1990s, such that the product, conditional to changes in the product’s
unsafe products no longer receive market approval. marketing authorisation. Finally, compared with

Copyright © 2016 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2016
DOI: 10.1002/pds
 benefit-risk reassessment of medicines
previous studies, our results indicate that the REFERENCES
number of products that are removed from the mar- 1. Duijnhoven RG, Straus SMJM, Raine JM, de Boer A, Hoes AW, De Bruin ML.
ket because of safety reasons shortly after market Number of patients studied prior to approval of new medicines: a database
analysis. PLoS Med 2013; 10(3): e10001407.
introduction might have decreased compared with 2. Gibbons RD, Amatya AK, et al. Post-approval drug safety surveillance. Annu
the 1970–1990s. Rev Public Health 2010; 31: 4129–437.
3. Rägo L, Santoso B. Drug regulation: history, present and future. Chapter 6 from
‘Drug benefits and Risks: International Textbook of Clinical Pharmacology
WHO’, IOS Press, revised 2nd edition 2008.
CONFLICT OF INTEREST STATEMENT 4. Waller P, Evans S. A model for future conduct of pharmacovigilance.
Pharmacoepidemiol Drug Saf 2003; 12: 17–29.
5. Walker S, McAuslane N, Liberti L. Refining the benefit-risk framework for
No funding was received for the work reported in this assessing medicines: valuing and weighting the parameters. Report on progress
paper. All authors report no conflict of interest. by regulators and industry on the development of a common benefit-risk assess-
ment framework. Clin Pharm Ther 2011; 89(2): 179–182.
The results reported in this paper have not been 6. Coplan PM, Noel RA, et al. Development of a framework for enhancing the
published previously, nor have they been presented transparency, reproducibility and communication of benefit-risk balance of
medicines. Clin Pharmacol Ther 2011; 89(2): 312–315.
at scientific meetings or conferences. 7. Eichler HG, Labadie E, et al. Safe drugs and the cost of good intentions. N Engl J
Med 2009; 360(14): 1378–1380.
8. Hauben M, Bate A. decision support methods for the detection of adverse events
in post-marketing data. Drug Discov Today 2009; 14(7/8): 343–357.
Key Messages: 9. European Medicines Agency. Benefit risk methodology project. 2009. Doc Ref
EMEA/108979/2009
• As the safety profile of medicines is usually 10. Garattini S, Bertele V. Rosiglitazone and the need for a new drug safety agency.
Br Med J 2010; 341: c5506.
incomplete at market entry, the safety of medi- 11. Qureshi ZP, Seoane-Vazquez E, Rodriguez-Monguio R, Stevenson KB,
Szeinbach SL. Market withdrawal of new molecular entities approved in the
cines is monitored throughout the product United States from 1980 to 2009. Pharmacoepidemiol Drug Saf 2011; 20:
lifecycle. If necessary, the benefit-risk profile of 772–777.
12. Jefferys DB, Leakey D, Lewis JA, Payne S, Rawlins MD. New active substances
a marketed medicine will be reassessed by the authorised in the United Kingdom between 1972 and 1994. Br J Clin Pharmacol
committee for medicinal product for human use. 1998; 45: 151–156.
• During 2001–2012, there were a total of 73 13. Abraham J, Davis C. A comparative analysis of drug safety withdrawals in the
UK and the US (1971–1992): Implications for current regulatory thinking and
benefit-risk reassessment procedures for either a policy. Soc Sci Med 2005; 61: 881–892.
14. McNaughton R, Huet G, Shakir S. An investigation into drug products withdrawn
single medicinal product or a product class. The from the EU market between 2002 and 2011 for safety reasons and the evidence
majority of these procedures did not result in used to support the decision-making. BMJ Open 2014; 4: e004221.
15. Arlett P, Portier G, de Lisa R, Blake K, Wathion N, Dogne JM, Spooner A, Raine
the removal of the medicine from the market, J, Rasi G. Proactively managing the risk of marketed drugs: experience with the
but rather, in confirming the positive benefit- EMA pharmacovigilance risk assessment committee. Nat Rev Drug Discov
2014; 13: 395–397.
risk profile, conditional to changes to the market- 16. Mol PGM, Straus SMJM, Piening S, et al. A decade of safety-related regulatory
ing authorization. Furthermore, the majority of action in the Netherlands. Drug Saf 2010; 33(6): 463–474.
17. Issa AM, Philips KA, Van Bebber S, et al. Drug withdrawals in the United
products that were removed from the market States: a systematic review of the evidence and analysis of trends. Curr Drug
were at the end of their lifecycle and not at the Saf 2007; 2(3): 177–185.
18. Lexchin J. How safe are new drugs? Market withdrawals of drugs approved in
beginning. Canada between 1990 and 2009. Open Medicine 2014; 8: e14–e19.

Copyright © 2016 John Wiley & Sons, Ltd. Pharmacoepidemiology and Drug Safety, 2016
DOI: 10.1002/pds