Transcript Pharmacovigilance and the WHO Collaborating Centre for

Pharmacovigilance and the WHO
Collaborating Centre for International
Drug Monitoring in Uppsala
Technical Briefing Seminar in Essential
Medicines Policies, Geneva, October 2007
Ronald Meyboom, MD, PhD
The Uppsala Monitoring Centre
www.who-umc.org
16% of hospital admissions are drugrelated (medical ward). Nelson KM, Talbert
RL. Pharmacotherapy 1996;16:701-7.
Adverse drug reactions are the 5th
leading cause of death in a hospital.
Lazarou J. Pomeranz BH, Corey PN. JAMA
1998;279:1200-5.
Avoidable in ca. 50 %
Definition of Pharmacovigilance:
(WHO, 2002, ISBN 9241590157)
• The science and activities relating to the
detection, assessment, understanding
and prevention of adverse effects or any
other drug-related problems
• Treatment evaluation science
Why pharmacovigilance?
• Clinical trials focus on demonstrating
efficacy and tolerability (selected ‘healthy’
patients, limited in number and duration)
• Incomplete knowledge, e.g. effectiveness,
rare but serious adverse reactions,
interactions, ‘real-live’ patients, subpopulations
• Do not produce all the information needed
for the balance of benefit and harm
How pharmacovigilance?
•
•
•
•
•
Spontaneous Reporting
Intensive monitoring (hospital)
Prescription Event Monitoring
Case Control Surveillance
Comprehensive population databases,
data-mining
• Patient series
• Observational studies
Formal studies
• Defined aim (identified
problem)
• Hypothesis testing
(problem solving)
• Established methods
(clinical trial, case
control, cohort)
• Comparison
• Limited in time, drugs,
population, place
Vigilance
• Open question:
looking for the
unexpected
• Hypothesis generation
(‘problem raising’)
• Exploratory,
controversial (SR,
PEM, CCS)
• No comparison
• Continuous, all drugs,
total population
Emphasis on
•
•
•
•
Early warning
Generation of knowledge
Dissemination of information
Rational and safe use of medicines
– Benefit and harm together
Three categories of adverse drug reactions
Need different methods for detection
An ABC of drug-related problems. Drug Saf 2000;22:415-23
Type A
(pharmacological)
Type B
(hypersensitivity)
Type C
(more frequent in
exposed than in
unexposed)
Clinical trial
Spontaneous reporting
Pharmacoepidemiology
challenge
Spontaneous Reporting
• A country-wide system for the reporting of
suspected adverse reactions to drugs
• A case report is a notification from a
health care professional, describing the
history of a patient with a disorder that is
suspected to be drug-induced
• Limitations because of privacy protection
and medical secrecy
Spontaneous Reporting
• When different doctors independently
report the same unknown and unexpected
adverse experience with a drug, this may
be a valid early signal
• Quantitative: more frequently reported
than expected from the background
Advantages of Spontaneous Reporting
•
•
•
•
•
•
Effective
‘All’ patients; ‘all’ drugs; many ADRs
Continuous
Rapid
Cheap
Not much health care infrastructure
needed
Limitations of Spontaneous Reporting
• Suspicions, incomplete, uncertain
• Underreporting is vast but unknown
and variable
• Exposure data available?
• No frequency measurement
• Comparison of drugs difficult
• Insensitive to type C adverse effects
• Further study for signal testing and
explanation
Data assessment in pharmacovigilance
1. Individual case report assessment
•
•
•
•
Interest, relevance (new, serious?)
Medical, pharmacological; coding
Follow-up
Causality assessment
2. Aggregated study and interpretation
•
•
•
•
Signal detection
Risk factors, interactions
Serial (clinicopathological) study
Frequency estimation
General design of systems for
causality assessment
Drug Safety 1997;17:374-389
• Basic questions
– Sub-questions
• Scores
• Overall score
• Causality category,
e.g. possible, probable, etc
None of the available systems has been
validated (i.e. shown to consistently and
reproducibly gives a reasonable
approximation of the truth)
• Validation = ‘proving that a procedure
actually leads to the expected results’
• No gold standard
• Causality category definitions
• What causality
• What causality
assessment can do
assessment cannot do
– Decrease disagreement between
assessors
– Classify relationship
likelihood (semiquantitative)
– Mark individual case
reports
– Education / improvement of scientific
assessment
– Give accurate quantitative
measurement of
relationship likelihood
– Distinguish valid from
invalid cases
– Prove the connection
between drug and event
– Quantify the contribution
of a drug to the
development of an
adverse event
– Change uncertainty into
certainty
Underreporting
•
•
•
•
•
•
•
Vast (> 90%)
Unknown
Variable
Biased
Difficult to adjust for
No frequency calculation
Delays signal detection
A signal is a set of data constituting a
hypothesis that is relevant to the
rational and safe use of a medicine
Hypothesis, data, arguments
•
•
•
•
•
Pharmacological
Clinical/pathological
Epidemiological
Quantitative / qualitative
Dynamic; develops over time
Drug Safety 1997;17:355-65.
100
//
Knowledge of adverse effect (%)
90
signal assessment
80
70
60
50
40
30
20
10
0
//
signal
generation
signal
strengthening
signal Time
follow-up
The balance of evidence in a signal
• Quantitative strength of the association
– number of case reports
– statistical disproportionality
– drug exposure
• Consistency of the data (pattern)
• Exposure-response relationship
– site, timing, dose, reversibility
• Biological plausibility of hypothesis
– pharmacological, pathological
• Experimental findings
– e.g. dechallenge, rechallenge, blood levels,
metabolites, drugdependent antibodies
• Analogies
• Nature and quality of the data
– objectivity, documentation, causality assessment
• Signal detection is searching for the unknown.
The same data can lead to different
conclusions. Since the truth is unknown it is
uncertain who is right, but nobody is wrong!
• Dilemma: a signal should be early and credible
at the same time
• Signals may consist of only a few cases and
may not be statistically prominent
• A signal is a snapshot and changes over time
• Signal testing and explanation require further
study
• Many signals remain unconfirmed
– scientific limitations
– no funding
WHO Collaborating Centre for
International Drug Monitoring
The Uppsala Monitoring Centre
Stora Torget 3, 75320 Uppsala, Sweden
www.who-umc.org
The Uppsala Monitoring Centre
• 1968 - WHO Collaborating Centre for
International Drug Monitoring, Geneva
• 1978 - Moved to Uppsala after agreement
between Sweden and WHO
• Non-profit foundation with international
administrative board
• WHO Headquarters responsible for policy
• Self-financing
• Global pharmacovigilance
The Uppsala Monitoring Centre
•
•
•
•
•
•
•
•
•
Director: Prof Ralph Edwards
Deputy director: Dr Marie Lindquist
International affairs: Sten Olsson
Pharmacists
(Bio)medics
IT specialists
Financing (‘Products and Services’)
Administrative
Together 45
Members of WHO Drug Monitoring
Programme
No of countries
100
80
60
40
20
0
1969
1979
1989
1999
WHO International Pharmacovigilance
Programme, March 2006
Aims and activities
• Collaboration with National Centres
• World-wide collection, analysis and
distribution of data
– Signal detection and analysis
– Pooling of data, comparing experiences
•
•
•
•
Communication, exchange of information
Technical support
Development of methods and tools
Improvement of pharmacovigilance around
the world
Cumulative number of reports
in ’Vigibase’
TOP 10 COUNTRIES
OTHERS 11%
NLD 2%
THA 2%
SWE 3%
ESP 3%
USA 46%
FRA 4%
AUS 5%
CAN 5%
DEU 6%
GBR 13%
World-wide accumulation and assessment
of data
• 80 participating National Pharmacovigilance Centres around the world
• 3.5 million case reports
• Early warning - acceleration of signal
detection
• Early signal strengthening by
comparing countries
Automated quantitative signal detection
• Extremely large numbers of drug adverse reaction combinations
• Selects automatically high-interest
combinations, using quantitative
disproportionality
• Manageable subsets of data
• No human time needed
• No investigators bias
• Objective, transparent, reproducible
• Flexible / adjustable
• Explorative
Signal detection at the
Uppsala Monitoring Centre
Eur J Clin Pharmacol 1998;54:315-321
A combination of
1. Automated quantitative data mining,
using Bayesian statistics and a
neural network architecture
(Information Component – ‘IC value’)
2. ‘Triage’
3. Human assessment
– National Centres
– Review Panel
– UMC staff
Triage filter, combining quantitative and
qualitative criteria; automatic selection of
associations that
• IC025 > 0; two or more countries
• Quarterly IC increase of 2 or more
• New and serious (WHOART Critical
Terms)
• Target reaction terms (e.g. SJS), two or
more reports, irrespective of IC value
Literature check
Captopril - Coughing
6
5
4
3
2
1
0
-1
-2
IC
79:1 81:1 83:1 85:1 87:1 89:1 91:1 93:1 95:1
Time(year)
"1988:1"
"1989:1"
"1990:1"
"1991:1"
"1992:1"
"1993:1"
"1994:1"
"1995:1"
"1996:1"
"1997:1"
"1998:1"
"1999:1"
"2000:1"
"2001:1"
"2002:1"
"2003:1"
SSRI Neonatal convulsions or neonatal
withdrawal syndrome
All SSRI
6
4
2
0
-2
-4
-6
Example of results in one Quarter (2004)
Total number of combinations:
No. of associations IC025 > 0:
Triage selection:
No. of signals for SIGNAL:
60000
2300
560
28
Signal Review Panel
• 40 Experts around the world
• Evaluate signals, together with UMC
staff and National Centres
• Select associations for follow-up
• Write signals in the SIGNAL document
• Preference for System Organ Class or
drug group (ATC)
The SIGNAL document
• Sent to all National
Centres (national
distribution)
• Individualized
section available to
industry
• All recipients
encouraged to
comment on topics
presented
Presentations at ISOP annual meeting 2006
• HMG-CoA inhibitors and pulmonary fibrosis
• β-2-Adrenoceptor agonists and nocturnal
enuresis
• Systemic effects of intranasal corticosteroids (neuropsychiatric reactions,
spontaneous abortion)
• Taxoids (paclitaxel and docetaxel) and
myocardial infarction
• Hypersensitivity reactions to Umckaloabo
(Pelargonium sidoides and P. reniforme)
• Potentiation of warfarin by glucosamine
Examples of articles
• Clark DW, Strandell J. Myopathy including
polymyositis: a likely class adverse effect
of proton pump inhibitors? Eur J Clin
Pharmacol 2006;62:473-9.
• Sanz E, et al. Selective serotonin reuptake
inhibitors in pregnant women and neonatal
withdrawal syndrome. Lancet 2005;365:
482-7.
• Coulter D, et al. Antipsychotic drugs and
heart muscle disorders in international
pharmacovigilance: data mining study.
BMJ 2001;322:1207-9.
Support to National Centres
•
•
•
•
•
•
•
•
Methodology
Terminologies, guidelines
Software (VIGIFLOW)
Harmonisation, standardisation
VIGIMED email discussion group
Annual meetings
Training
Books and brochures
Terminologies, guidelines
Links with WHO Geneva, CIOMS, ICH
• WHOART
• Drug Dictionary
• Guidelines for setting up and running of
a Pharmacovigilance Centre www.whoumc.org/DynPage.aspx?id=13136&mn=1512#8
• Herbal ATC
• Accepted scientific names of therapeutic
plants. 2005, ISBN 91 974750 3 3.
• WHO guidelines on safety monitoring of
herbal medicines
Harmonisation, standardisation
• Definitions (Biriell C, Edwards IR. Drug
Safety 1994;23:95-9)
• WHO causality categories
• Reporting adverse drug reactions.
Definitions of terms and criteria for
their use (CIOMS Council for International
Organizations of Medical Sciences. WHO,
1999, Geneva. ISBN 92 9036 071 2.)
Herbal and traditional medicines
• UMC Herbals database (Dr Mohamed
Farah)
• Herbal reviewers panel
• Collaboration with
– Uppsala University, Sweden
– WHO Collaborating Centre, Cape Town,
South Africa
– Royal Botanical Garden, Kew, UK
– University of Exeter, UK
– Harvard, US
New development areas
• Integrate Chinese ADR database
• Patient safety focus including medication errors
– World Alliance for Patient Safety
• Improved reporting and analysis of vaccine
reactions (AEFI)
– Flu pandemic planning
• Safety surveillance for other Public Health
Programmes
• Involvement in active surveillance
– Cohort Event Monitoring
• Data mining analysis of longitudinal patient
records
Thank you for your attention
[email protected]
www.who-umc.org