Transcript Potential for Extracting Data from Sample Databases

Life Sciences Research Office
Evaluating Adverse Event Systems for
Dietary Supplements
Potential for Extracting Data from Sample Databases
January 31, 2003
Y. Renee Lewis, Chief Operating Officer
Drug Safety Management
Safety is good business.
Agenda
About QED Solutions, Inc.
Data Considerations
Analyzing the Data
Research Capabilities
Summary
Company Proprietary
2
About QED Solutions
QED Overview
Company Focus: Web-based Drug Safety Management Solutions –
product with supporting services.
Sophisticated analytical tools that support
pharmacovigilance and safety surveillance
investigations
Research capabilities to aid medical professionals in
finding and understanding patterns of drug behavior
provided in details of adverse event data
Services:
Implementation, application hosting, data validation,
data extraction/aggregation, data extensions, limited
research (primarily with partners)
Client-base:
22 major pharmaceutical and bio-technology firms.
Primarily Global Safety Office, Medical Affairs or
Epidemiology.
Company Proprietary
4
Product Overview - QscanTM
Products:
One research solution, access multiple data sets
QscanTM FDA – subscription service to AE data released from
the FDA through FOIA
QscanTM World – subscription service to AE’s received by the
WHO at Uppsala Monitoring Center representing 67 countries
(CIOMs and MedWatch forms)
QscanTM PRO – internal application to review data collected at a
pharmaceutical company
Statistics:
Proportional analysis, frequency profiling, correlations,
comparisons
Others:
Automatic detection alerts to simplify monitoring of
hypothesis
Research capabilities to aggregate independent cases into
series for analysis
Company Proprietary
5
Product Maturity
12/1998 Founded by Victor Gogolak to develop solutions for pharmacovigilance
10/1999 Begin development of first product (data and application
06/2000 QscanTM Alpha released at DIA conference, Glaxo first Alpha customer
10/2000 QscanTM FDA 1.0 released to public
12/2000 Roche purchased – first production customer
06/2001 QscanTM FDA 1.2 released
09/2001* QscanTM FDA 1.5 released (General Release Product)
02/2002 QscanTM FDA 1.6 released
04/2002 QscanTM FDA 1.6.1 released
06/2002 QscanTM World alpha released
08/2002* QscanTM World 1.7 released
09/2002 QscanTM FDA 1.7 released
Aggressive release schedule
includes:
• Data manipulation features
• Analytic techniques
• Data sets or updates
03/2003 QscanTM FDA 2.0 and World 2.0
Company Proprietary
6
Using Tools
What you can do with tools:








Monitor and detect requested patterns in the data
Research general patterns and trends
Probe and analyze hypothesis
Show how the reported data compares consistent with proposed
populations
Identify strength of association between elements in a set of cases
Compare two different sets of data
Store saved results and data for future review, comparison
Export data for additional analysis or reporting
What QscanTM does not do:
 Provide “answers” – medical judgment is required
 Show causality – must use other methods
 Directly support regulatory reporting process
Company Proprietary
7
Data Considerations
Data Issues - General
Data Availability – no “good source” for herbals, vitamins or OTC’s
 No regulation to drive data collection – voluntary
 No regulation around data review or analysis
 No source specific to this domain
– Safety is a public policy issue
– Does data makes a company possibly vulnerable
OR does it provide a competitive advantage?
Data Quality
 Limited training on data collection for these items –
data collected by accident!
 Data collection poor, not controlled
 Many consumer reports with no medical follow up
 Lack of integrity of data relationships (e.g., time to onset)
 FDA and World only report “serious and unexpected” – bad things
Company Proprietary
9
Data Considerations with QscanTM Sources
QscanTM FDA (through FOIA or other 3rd party)
 FDA data released under FOI in “raw” form (verbatim terms)
 Released quarterly - approximately 6 month latency
 Nov 1997 to present - Adverse Event Reporting (AERs); MedDRA reaction
dictionary (very old release)
 1969 – October 1997 - Spontaneous Reporting System (SRS); COSTART mapped
to MedDRA
 Regulations to control volume of data – severe and unexpected
QscanTM World




World Data – 67 countries – CIOMs except US
Includes FDA data, but only “non-consumer” reports
WHO-ART reaction dictionary; WHO-DD
Some noticeable latency in data collection (years in some countries)
QscanTM PRO
 Internal data – Post Market and Clinical Trial
 World Health Organization – Uppsala Monitoring Centre
 Clinical Trials
Company Proprietary
10
Sample Data
Dictionary mappings
and aggregations
Suspect vs
non-suspect
Company Proprietary
11
Verbatims (Show Source)
Company Proprietary
12
Data Examples – Case Listing
Follow-on processing
Duplicates? Twins?
Reactions – MedDRA
terminology
Company Proprietary
13
Sample – Case Detail
All the data is
made available to
review every
known detail.
Additional
elements can be
easily added.
Company Proprietary
14
Other Possible Data Sources
Commonly used for analysis
 Consumer reports to the company
 GPRD (General Practice Research Database)
 Claims data
 Registries
Uncommonly used, commonly used as data extenders
 Prescription data - denominator
 Medical records (closed systems like Kaiser Permanente)
 Medical records with claims data (hard to find)
 Genomic databases, Toxicology data
 Internet sites
Requirements for use
 Structured
 Coded (dictionary, reactions, outcomes, demographics)
Company Proprietary
15
Analyzing the Data
If it’s so bad, why use this data?
Availability ….
Many AE’s include herbals, vitamins and OTCs as a byproduct of the process
 Severely under reported, but can assume that the under
reporting is uniform
 Many herbals, vitamins and OTCs have been around for a
long time – our data goes back to 1969
 Patterns may emerge using more sophisticated techniques:
– Proportional Analysis
– Correlation
 Reactions are coded (MedDRA and WHO-ART)
 Drug names are mapped and can be remapped easily
 Available today, immediately
Company Proprietary
17
Analytics
Common output with these data
 Rates and counts
 Proportional reporting rates – “out of norm”
 Odds Ratio (where appropriate)
 Correlations
 Comparisons – standard backgrounds, other data sets
 Trends
Ability to export data from the system
 Continued analysis and imaging
 Documentation
 Information sharing
Requires structured information
 Dictionaries and terminologies
 Setup for both analysis and research (drill-down)
Company Proprietary
18
Research Social Circumstances – PRR 2.98
Details
• Case 3570383
Drug abuser
• Case 3618733
Refusal of treatment
Primary
suspect on
both!
Positive
Dechallenge
Company Proprietary
19
Research Capabilities
Case Series and Search Criteria
Ability to group cases based on criteria, for example
 Reactions
 Concomitants
 Demographics
 Report dates
 Outcomes
 Medically significant
Methods to find “difficult” groupings, for example
 Cases where these two drugs occur together
 Cases with this drug or that drug
 Cases where this drug occurs, but not that drug
 Cases with only these reactions
Ability to review out put and refine criteria based on results
Facility to share results with interesting information or comments
Company Proprietary
21
What’s it take to make data “analyzable?”
Structured data for elements coded to terminologies, where possible
 Drug/compounds (primary, concomitants, suspect drugs and why)
 Reaction Terms
 Outcomes
 Demographics
Additional information is a plus to increases capabilities and understanding
 Condition data
 Time to onset
 Report dates
Good intake procedures improve the data quality





Handling of consumer reports
Medical review of reports
Data collection tools and automation procedures
Application and data access for review, follow-up and analysis
Methodologies – passive and active
Company Proprietary
22
Summary
Summary
Not much available today, so make the most of what’s there
You can use what is available to your advantage with the
“right” approaches
Tools and software frameworks make the task less arduous,
data readily available
Better data sources for herbals and vitamins are sorely
needed
Qscan-like tools can be used against any spontaneous data
source with some minimal effort
Dictionaries, standard terminologies and proper mapping
techniques make the data systematically available
Company Proprietary
24