Drug Safety Challenges:
Considerations for Sources of Data
Gerald J. Dal Pan, MD, MHS
Director
Office of Drug Safety
Center for Drug Evaluation and Research
FDA
National Health Policy Conference, Washington, DC
February 6, 2006
1
Sources of Risk From Medical Products
Known Side Effects
Unavoidable
Medication and
Device Errors
Product Defects
Avoidable
Preventable
Adverse Events
Injury or Death
National Health Policy Conference, Washington, DC
February 6, 2006
Remaining
Uncertainties:
•Unexpected side effects
•Unstudied uses
•Unstudied populations
2
Post-marketing Drug Safety Risk Assessment:
Identification of New Adverse Events
• A fundamental goal of post-marketing drug
safety programs
• Must account for many different types of
risk
• Must account for many potentially
confounding factors
• Must account for time course of events
National Health Policy Conference, Washington, DC
February 6, 2006
3
Post-marketing Drug Safety Risk Assessment:
What Pre-marketing Safety Data Tell Us
Pre-clinical
Pharmacology
And Toxicology
Clinical
Pharmacology
Pre-Marketing
Safety Data
Clinical Safety Data
Controlled Studies
National Health Policy Conference, Washington, DC
February 6, 2006
Product
Label
Clinical Safety Data
Open-label Studies
4
Post-marketing Drug Safety Risk Assessment:
Identification of New Adverse Events
Market
Introduction
Pre-marketing
Safety Data
Post-marketing Period
National Health Policy Conference, Washington, DC
February 6, 2006
5
How post-marketing adverse event
reports get to FDA
Patients, consumer, and
healthcare professionals
voluntary
voluntary
Manufacturer
regulatory
requirements
FDA
FDA MedWatch
FDA’s Adverse Event Reporting
System (AERS) database
National Health Policy Conference, Washington, DC
February 6, 2006
6
Post-marketing Drug Safety Risk Assessment:
Identification of New Adverse Events
Market
Introduction
?
?
?
Pre-marketing
Safety Data
Post-marketing Period
National Health Policy Conference, Washington, DC
February 6, 2006
7
Post-marketing Drug Safety Risk Assessment:
Identification of New Adverse Events
Rare but serious adverse event
Market
Introduction
•Aplastic anemia
•Drug-induced liver injury
Pre-marketing
Safety Data
Post-marketing Period
National Health Policy Conference, Washington, DC
February 6, 2006
8
Post-marketing Drug Safety Risk Assessment:
Identification of New Adverse Events
Also common in the population
Market
Introduction
•Myocardial infarction
Pre-marketing
Safety Data
Post-marketing Period
National Health Policy Conference, Washington, DC
February 6, 2006
9
Post-marketing Drug Safety Risk Assessment:
Identification of New Adverse Events
Market
Introduction
Also a manifestation of the underlying
disease
•Myocardial infarction
Pre-marketing
Safety Data
Post-marketing Period
National Health Policy Conference, Washington, DC
February 6, 2006
10
Post-marketing Drug Safety Risk Assessment:
Identification of New Adverse Events
Market
Introduction
How do we separate a potential signal
from the background?
Pre-marketing
Safety Data
Post-marketing Period
National Health Policy Conference, Washington, DC
February 6, 2006
11
Post-marketing Drug Safety Risk Assessment:
Investigation of New Adverse Event Risk
Rare but serious adverse event
Market
Introduction
Intensive case evaluation
Pre-marketing
Safety Data
Post-marketing Period
Look back at pre-marketing
safety database
National Health Policy Conference, Washington, DC
February 6, 2006
12
Post-marketing Drug Safety Risk Assessment:
Investigation of New Adverse Event Risk
Common in the population OR
Market
Introduction
Manifestation of the underlying disease
Intensive case
evaluation
Pre-marketing
Safety Data
Post-marketing Period
Look back at pre-marketing
safety database
National Health Policy Conference, Washington, DC
February 6, 2006
Still hard to establish
and quantify risk
13
Post-marketing Drug Safety Risk Assessment:
Investigation of New Adverse Event Risk
Clinical Trial
Treatment of
Interest
Random
Allocation
Control Treatment
Follow-up Period
National Health Policy Conference, Washington, DC
February 6, 2006
14
Post-marketing Drug Safety Risk Assessment:
Investigation of New Adverse Event Risk
Clinical Trial
Treatment of
Interest
Random
Allocation
Control Treatment
Follow-up Period
National Health Policy Conference, Washington, DC
February 6, 2006
15
Post-marketing Drug Safety Risk Assessment:
Investigation of New Adverse Event Risk
Clinical Trial
Excess Risk
Treatment of
Interest
Random
Allocation
Risk Ratio
Control Treatment
Follow-up Period
National Health Policy Conference, Washington, DC
February 6, 2006
16
Alternative Sources of Information
•
•
•
•
Large health care utilization databases
Electronic medical record systems
Registries
Can be used for active surveillance or to
answer specific drug safety questions
National Health Policy Conference, Washington, DC
February 6, 2006
17
Heath Care Utilization Databases
• Large, population-based, integrated
pharmacy and medical claims databases
– filled prescriptions
– professional services
– hospitalizations
• Can capture real-world practice patterns, in
the context of the system that gives rise to
the data (in US, generally within a given
health insurance plan or set of plans)
National Health Policy Conference, Washington, DC
February 6, 2006
18
Heath Care Utilization Databases
• Strengths
– size
– based on actual care
– data already collected
• Limitations
– specific clinical data not present
– lack of some important health-related
information (eg, smoking status)
– only captures what is billed for
– frequent patient turnover as insurance changes
National Health Policy Conference, Washington, DC
February 6, 2006
19
Post-marketing Drug Safety Risk Assessment:
Investigation of New Adverse Event Risk
Epidemiological Study - Cohort Study
Relative risk
or hazard
ratio
Start
observation
Time
National Health Policy Conference, Washington, DC
February 6, 2006
20
Example of a cohort study:
Statins and hospitalized rhabdomyolysis
Cohort:
Drug-specific inception cohorts of statin and fibrate users, based on data from
11 US health plans using automated claims covering prescription drugs,
outpatient care, hospitalizations, and medical procedures
Exposure:
Algorithm developed to calculate
person-time on drug for each
patient based on prescription
claims. Separate classifications for
monotherapy and statin-fibrate
combination therapy
Outcome:
Medical record review of all
patients based on hospitalization
claims with at least one ICD-9-CM
code suggestive of severe muscle
injury, followed by a blinded
review to determine cases of
rhabdomyolysis.
Source: Graham D et al. JAMA 2004;292:25885-2590
National Health Policy Conference, Washington, DC
February 6, 2006
21
Example of a cohort study:
Statins and hospitalized rhabdomyolysis
Analysis:
Relative risk estimates of rhabdomyolysis, adjusted for age, sex, and
diabetes mellitus were calculated using Poisson regression. Incidence rates
per 10,000 person-years of treatment, with 95% CIs, were calculated.
Results:
Rhabdoymyolysis per 10,000 Person-Years of Therapy With Lipid-Lowering Drugs Used as
Monotherapy or as Combination Therapy With Another Drug
Combination Therapy
Monotherapy Incidence
Incidence Rates
Drug
Rates (95% CI)
Combination
(95% CI)
Atorvastatin
0.54 (0.22-1.12)
Atorvastatin + fenofibrate
22.45 (0.57-125)
Cerivastatin
5.34 (1.46-13.68)
Cerivastatin + gemfibrozil
1035 (369-2117)
Pravastatin
0 (0-1.11)
No cases
0 (0-67.71)
Simvastatin
0.49 (0.06-1.76)
Simvastin + gemfibrozil
18.73 (0.47-104)
Fenofibrate
0 (0-14.58)
Fenofibrate + atorvastatin
16.86 (0.43-93.60)
Gemfibrozil
3.70 (0.76-10.82)
Gemfibrozil + cerivastatin
789 (166-2138)
Source: Graham D et al. JAMA 2004;292:25885-2590
National Health Policy Conference, Washington, DC
February 6, 2006
22
Electronic Medical Records
• Contain more information than claims
databases:
– medications prescribed
– detail clinical information (eg, symptoms and
signs)
– physical examination results
– results of diagnostic tests
• Example: General Practitioner Research
Database (GPRD)
National Health Policy Conference, Washington, DC
February 6, 2006
23
Post-marketing Drug Safety Risk Assessment:
Investigation of New Adverse Event Risk
Registry
Persons with disease of interest
Case-control
studies
Study natural
history or
survival
Estimate magnitude
of problem
National Health Policy Conference, Washington, DC
February 6, 2006
24
Post-marketing Drug Safety Risk Assessment:
Investigation of New Adverse Event Risk
Registry
Persons with exposure of interest
Risk factors
for exposure
Outcome of
exposure
Estimate magnitude
of exposure
National Health Policy Conference, Washington, DC
February 6, 2006
25
Use of a Postmarketing Registry:
Antiepileptic Drugs and Teratogenicity
Pregnant women with
epilepsy on valproic acid
Enrollment
7 months Postpartum
Birth
Outcome
ascertainment
149 VPA-exposed, 16 with
major malformations
(10.7%, 95% CI: 6.3-16.9)
Internal comparator rate: 2.9% (95% CI: 2.0-4.1)
External comparator rate: 1.62%
Source: Wyszynski DF et al. Neurology 2005;64:961-965
National Health Policy Conference, Washington, DC
February 6, 2006
26
New Database Acquisitions
• Four organizations with linked pharmacy-medical
claims databases
• Contracts signed September 2005
• Allows for collaborations between FDA epidemiologists
and experts at these organizations
• Four organizations:
–
–
–
–
HMO Research Network/Harvard Pilgrim Health
Kaiser Family Foundation
Vanderbilt University
Ingenix (i3Drug Safety)
National Health Policy Conference, Washington, DC
February 6, 2006
27
New Database Acquisitions
• Four organizations:
– Harvard Pilgrim Health/HMO Research Network
• Eight geographically diverse health plans with 3.2 million members
• Electronic medical records available for 6 of 8 sites
– Kaiser Family Foundation
• 6.1 current members in northern and southern California
• Fully integrated databases, linked to vital statistics and cancer registries
• Unique formulary limited to selected drugs and indications
– Vanderbilt University
• Two state Medicaid populations (Tennessee and Washington)
• 2.2 millions members, some at high medical risk (eg, the poor, nursing home
residents)
– Ingenix
• Geographically diverse insured population of 12 million members
• Some laboratory data also available
National Health Policy Conference, Washington, DC
February 6, 2006
28
Active Surveillance
•
•
•
•
Request for Information issued April 2005
Responses received June 2005
Responses currently under review
Agency will decide on next steps
National Health Policy Conference, Washington, DC
February 6, 2006
29
CMS Interactions
• ODS epidemiogists are working with CMS and AHRQ
staff to understand better the nature of CMS data
• Current efforts focused on using Part B data for a pilot
drug safety study
• Still in learning/exploratory stages
National Health Policy Conference, Washington, DC
February 6, 2006
30