The Challenges of

Pharmacovigilance
I. Ralph Edwards

‘’Courage is the human virtue that counts

most—courage to act on limited knowledge and
insufficient evidence. That's all any of us have.”
~ Robert Frost
20th century American poet and three time
Pulitzer prize winner (1924, 1931, 1937)
 The problem

(all red text indicates need for

improvement)
Size and severity of the ADR problem
Patient safety signals.
• 39 prospective studies from US hospitals
• Overall incidence of serious ADRs = 6.7%
• Overall incidence of fatal ADRs = 0.32%
(106 000 individuals)
• 4th - 6th leading cause of death
Lazarou et al JAMA 1998;279: 1200 - 1205
6.7% of hospital patients have serious adverse
drug reactions (medication error excluded)

Lazarou J. Pomeranz BH, Corey PN. JAMA 1998;279:1200-5

16.2% of hospital admissions are drug-related

Therapeutic failure 54.8%
Adverse reactions 32.9%
Overdose 12.3%
Avoidable 49.3%
Nelson KM, Talbert RL. Pharmacotherapy 1996;16:701-7
Pirmohamed M JS, Meakin S, Green C, Scott AK,
Walley TJ, Farrar K, Park BK, Breckenridge AM.
Adverse drug reactions as cause of admission to
hospital: prospective analysis of 18 820 patients.
BMJ 2004;329(7456):15-19.

• ….identifies the main drug culprits in a large

hospital based study. They are old drug
groups (low dose aspirin, diuretics, warfarin,
and non-steroidal anti-inflammatory drugs
other than aspirin: the most common reaction
being gastrointestinal bleeding) on which we
have much information……
 Managing a signal ?
Public
Public
Information
Information
22
Public
Public
Information
Information
11

Increasing information and knowledge

A (also misinformation)
first
signal

SCARE

Media coverage
 Five broad activities essential
to pharmacovigilance:
• Suspected ADR signal detection and formation of hypotheses
Decisions
• Analysis of all issues around the signal, particularly
confirmation (or refutation) of hypothesis, estimation of the
size of the risk and whether susceptible patients exist

Decisions
• Consideration of possible effectiveness-to-risk issues in
therapy (comparative)
– How to do it?
– Economics Decisions

• Communication of information to health professionals and

patients in a useful way. And possible regulatory action.
Decisions
• Consequence evaluation.
 Problems of withdrawal &
regulation (examples)

Consequences - Outcomes research needed

 Examples
• Slow action • Cisapride, Digesic
• Precipitate action • Many old examples
– Lack of evidence – also Sertindole
• Influence of media • Many eg. Vioxx
• Complexity • COX2 & NSAIDS
• Effectiveness & risk • Phenylpropanolamine
– Confusion – Dose and indication
• Pseudo-safety • Nefazodone
withdrawals – monitoring
– Company finances
 Examples
• Off-label usage • Vioxx (dose)
– Dosage – Cerivastatin (int-act)
– Failures of guidance – Troglitazone (monit)
• Long term effects • HRT (breast ca.)
• Pre-marketing • SSRI
information – Company problem, not
suppressed drug?
• Scientific confusion • HRT (heart dis. or ?)
• Herbals safety • Aristolochia
– Species?
Example of a decision problem to
be solved [1] ...
Cisapride - heart rhythm disorders ?
• 1986: double blind study “cisapride ted
rke
produced tachycardia” a
m
e in Br Med J
• 1992 WHO Signal published b
on serious arrhythmia t i l l
s
Je“no epidemiological support”
– letters to Br Medid
p r
sa
• 1995 case ireport published, Lancet “QT
d c
prolongation and tachycardia”
l
u Doctor letter in USA by manufacturer
o
– Dear
h
• S
 Example of a decision problem to
be solved [2] ...
l a?
e u
z
• Piroxicam is the most GI toxicenof e the
V
NSAIDS in several comparative i b i n studies
x
– GI toxicity is a majorfecause co of morbidity
r o
c e
p l a
re
a m
x ic
ir o
d p
u l
ho
S
 ry ?
s a
Discussion es s?
e c e lp
points y n i t h
a ll a n
a l
l r e tih
g n n
w a l e s t io
ra o opm ma
h d p ef r o r
i t r e
m e n f
g w oti , i
r u m
e
h us e d
s d rm t
e ly
n i h
u a
c
te a
edl e r
of raewr rop
o w h dw d p
H i
a t n an
s wc l,
e
o w fu
o
DH use
to
Visions and goals of
patient/drug safety
Prevention in drug safety

Effectiveness and risk

Need for comparison between
medicines
Risk assessment
 The true balanced concepts
• Efficacy (hard data) • Hazard
NO!
• Effectiveness • Risk
Yes!
• Benefit (what the patient
• Harm (soft data)
feels)

t ter
ma
e s
is do
Th
 Goals and tools
• Problems with
medicines
– Find them
• Causation
• Spontaneous reports
– Data mining
– Analyse them
• Put them in context
– Quantify them – Epidemiology
• Effectiveness and risk
– Comparisons
– Careful assessment of
all relevant data
• How? Who?
 Goals and tools
• Problems with medicines • Risk management
– Prevent or limit them • Communicate information to
– Manage them HPs & patients
– Maximise effectiveness
with minimum risk to • Check for response
individual patients
• Communicate to the health
• Offer best treatment authorities, public & others
options • Check media: surveys
• Most effective
• Safest
• Cheapest

• Check result
 Do we really use our tools?
• Case reports can tell us a huge amount about
what concerns individual
doctors/h.p’s/consumers
– We could get much more information on what
happens to people, medication error, interactions,
patients at risk, how to diagnose ADRs, etc.

• Studies tell us about populations and not

individuals
– Many importantADRs are rare (< 1/1000)
• Large numbers of exposed patients and controls needed
• We need more, and much better, ways of using health
service databases
 Measuring true effectiveness and
risk with SSRIs?
• Needs knowledge
finding in patient
care databases IC 0
– E.g Data mining in Fluoxetine
IMS Health database
of 2 million patients
• Can show
comparative
effectiveness and IC 0

risk Amitrityline

– Data quality (One

Year)

Suicidality
A new look at spontaneous
reports
Reports of concerns about
therapy

Client dissatisfaction reports

 Reports of concerns about
therapy

• Novel drug signals (traditional

pharmacovigilance)
– Careful clinical evaluation of all the
clinical circumstances a differential
diagnosis, or causal potential, for that p o r ts
s e re
individual. t t h e
r , t h a
w e v e
, h o
io n is
c e p t
is c on y
u s m i o l o g .
r io m
A se or epide
r e p o
a
 Novel drug signals
• E.g. The conversion of cases into database
epidemiological data has cost us time in
recognizing
– the SSRI ‘electric shock’ syndrome (entered into
databases as ‘dysaesthesia’,
– the sumitriptan ‘pain reactivation syndrome’ (
Coulter DM, Passier JL, Clark DW, van Puijenbroek
EP. Activation of pain by sumatriptan.Headache.
2003 Oct;43(9):994-9.),
– statin caused rhabdomyolysis (first recorded in
databases as ‘myopathy’).
 Reports of concerns about
therapy

• Patient safety signals.

– Because pharmacovigilance experts may
know about a particular drug/ADR
relationship does not mean that everyone
knows or uses that information.
– There may be need for other action:
mainly communication/education
 Patient safety signals.
– Continued high levels of reporting of ‘known’
ADRs should lead to an informative newsletter
and also individual helpful responses e.g.
• Some known ADRs may be reported because they
are different in quality than expected (e.g. more
severe).
– A particularly severe skin rash with amoxicillin may be due
to undiagnosed immune disorders. A response suggesting
that might be helpful.
– Drug interactions are often not diagnosed
– a response to a report on a common dose related ADR
might suggest the possibility.
 Patient safety signals.
– ADRs following contraindicated drug use
– Beta-blocking drugs causing severe asthma)
should be followed by a very firm, but not
judgemental, reply: ethical standards demand
that.
– Continued large numbers of reports of
known associations should result in an
educational article in a local journal or
newsletter.
– Benzodiazepine dependence
 Reports of concerns about
therapy

• System signals
– Much criticism is made of spontaneous
reporting defects, but not much intelligent
use is made of information that would lead
to improvements.
–
 System signals
– Knowledge that general practitioner
reports are much greater than hospital
reports should lead to a campaign to
improve the latter.
– Low reporting in some therapeutic areas
should suggest the same need for attention
to those areas.
– Poor quality of reports should lead to
education. This may be individual,
supportive feedback or general via
newsletters
 Other needs
• There are much more active approaches
needed for education of HPs and public in all
countries
– more pro-active use of the media
– much more interest in patient safety issues
• Medication errors
– Root –cause analysis

• There is a great need to know the basis for

regulatory decisions
– particularly those made in the developed world
 Patient involvement
• Patients need more general information
about drugs and their effectiveness and
risk
• VERY IMPORTANT
– We need much more information about
what risks patients are prepared to take
for what benefit to them. Until we know
this we will continue to ‘second-guess’
about what is acceptable or not
 Managing a signal: the future
Public
Public
What do HPs and patients Information
Information
think and want 22

Public
Public
Information
Information
11

A first Increasing information

signal
and knowledge
Pre-marketing risk
management
Media coverage
 Conclusions
• Think less about drug safety: more about
patient safety
– Use and react to concerns

• Think less about regulating (incl.

withdrawal) and automating data input:
more about useful information output

• Think more about impact and consequences

of decisions and non-decisions
 Conclusions
• ‘..Drug safety information must serve the
health of the public. Such information
should be ethically and effectively
communicated in terms of both content and
method. Facts, hypotheses and conclusions
should be distinguished, uncertainty
acknowledged, and information provided in
ways that meet both general and individual
needs...’
• Erice Declaration, 1998.