Title: Evaluation of Data Sources to Support Pharmacosurveillance
Investigator Name: Dr. Anne Holbrook
Project Completion Date: March 2004
Research Category: Research
Institution: St. Joseph's Healthcare, Hamilton
Project Number: 6795-15-2001/4410013
Help on accessing alternative formats, such as PDF, MP3 and WAV files, can be obtained in the alternate format help section.
(110K)
Summary
Background and Objectives: Regular monitoring of medications
in real clinical practice for benefits, harms and costs (pharmacosurveillance)
has been widely advocated as necessary and feasible to optimize
care at both the individual patient level (patient safety) and
the public policy (post-marketing surveillance and cost-effectiveness)
level. Our study determined the information needs, information
availability and methodologic and information privacy challenges
for pharmacosurveillance in Canada.
Target Audience: Primarily drug regulators, drug benefit
plans, formulary managers and advisors, pharmacoepidemiology researchers.
Methods: Using nominal group consensus technique with experts
in
11 domains from across the country, we determined the information required
to provide optimum routine pharmacosurveillance (gold standard dataset).
Three main data sources - large, linked administrative databases
(LADs),
electronic health records (EHRs)
and patient registries (PRs), were then analyzed to determine how
much of the gold standard dataset requested by the expert panel,
was actually available. This was measured in terms of a) data field
present and b) completeness (data in the field).
One of the data resources, LAD,
was used to explore methods to adjust for channelling bias (patient
allocation bias) by comparing results before and after application
of 3 main adjustment methods with those of landmark randomized
trials (RCTs)
- using hormone replacement therapy and cardiovascular disease
as the clinical scenario. As well, techniques for data mining large
health databases for early signals of unexpected benefit and harm,
were reviewed and applied.
Finally emerging regulations and guidelines regarding the use
of health information, both identified and de-identified, and their
impact on pharmacosurveillance research and policy development,
were examined.
Results: The "gold standard" dataset contained 138 information items, ranging from 0 items of family history to 64 items related to current and past medication use. None of the data resources contained all of this information, nor the information subset that the consensus panel universally supported. LADs tend to have very high rates of completeness (data in the field) but have less than half of the required fields. HRs hold many more of the data fields (potentially all but costing information) but are more variable in their field content and data completeness. PRs tend to be disease or drug-focused and are not catalogued anywhere, making them difficult to access or to use other than for selected questions.
None of the 3 channeling bias adjustment methods examined were
able to fully reproduce the RCT results.
Data mining techniques within large electronic health record systems
or administrative databases, are worth exploring for early signals
of drug effectiveness or harm. Data mining methods used in commercially
available software require further validation.
Regarding health information privacy, both patient and physician
remain concerned about who has access to the patient's information,
even when it is anonymized. Both groups are the least concerned
about access by university or hospital-based research groups. The
lack of consistency and oversight of privacy regulations and interpretation
of regulations between the provinces, has led to unacceptable delays
in accessing health data for research or policy development purposes.
Implications:
- A data resource suitable for routine, rapid investigation of
pharmacosurveillance issues would likely require linkage across LADs
and HRs.
- A registry of health data resources in Canada would be helpful
to understand what data resides where and to ensure that the
data are properly managed, supported and protected.
- The technical feasibility of pharmacosurveillance is a major
problem, given the lack of data standards or data integration
standards in Canada.
- Coordination and support for Canada's many data resources,
particularly research-quality electronic health records and linked
provincial databases, will be required to develop their full
value for research and policy purposes.
- Further work is required to adjust for the effect of biases
inherent in observational data. Without this, interpretations
of cause and effect are flawed and may be fatal, and the databases
will only be useful for retrospective evaluation of practices
or for generating hypotheses for further rigorous, randomized
controlled trial ( RCT)
testing.
- For the foreseeable future, RCTs will remain the "gold standard" for understanding benefit and harm.
- Information privacy guidelines stipulate very strong data security,
data confidentiality and research competence be in place.
- " Health information privacy legislation and guidelines are evolving swiftly. Analyses performed in academic settings with high quality privacy rules and practices, are the least problematic.
The views expressed herein do not
necessarily represent the views of Health Canada
In addition to the above summary, the full report can be accessed
in the following ways:
- The print version of the full report can be obtained in the
language of submission from the Health Canada Library through
inter-library loan.
- An electronic version of the report in the language of submission is available upon request from Health Canada by contacting the Research Management and Dissemination Division.
This research has been conducted with a financial contribution
from Health Canada's Health Policy Research Program. For permission
to reproduce all or part of the research report, please contact
the Principal Investigator directly at the following address: holbrook@mcmaster.ca.
The Health Policy Research Program (HPRP)
funds research that provides an evidence base for Health Canada's
policy decisions. The HPRP is
a strategic and targeted program with a broad socio-economic orientation
and connections to national and international endeavours. The research
can be primary, secondary or synthesis research, a one-time contribution
to a developing research endeavour, or a workshop, seminar or conference.
|