Transcript M Andersen - 16th Nordic Congress

Prescription registers in Denmark
Morten Andersen
Senior Researcher, PhD
Clinical Pharmacologist
Nordic Congress of General Practice
Copenhagen, May 2009
Prescription registers in Denmark used for
pharmacoepidemiologic research
• Odense University Pharmacoepidemiologic
Database, OPED, Funen (1990)
• Northern Jutland, PDNJ (1991)
• Aarhus (1996)
• Viborg (1998)
• Research registers in Statistics Denmark:
National register of drug statistics of the Danish
Medicines Agency (1995)
Sources of pharmacy
dispensing data
• Regional health insurance registers
– Data from pharmacies to regional health insurance
– Drugs in the general reimbursement scheme: All
dispensings, regardless of copayment
– Drugs with individual reimbursement: Only reimbursed
dispensings
• National register of drug statistics
– Data from pharmacies to the Danish Medicines Agency
– All prescriptions dispensed at community pharmacies
– Drugs on prescription regardless of reimbursement
Geographical bias
Odense PharmacoEpidemiologic Database
(OPED)
• All computer-registered purchases of reimbursed
prescription drugs in pharmacies of Funen County
(population 470,000) since October 1990
• Complete for the whole county since November 1992
• West Zealand 2000
• Region of Southern Denmark 2007 (1.2 million)
• Data on the individual level (CPR-number)
• Anonymised version available
• Research registry maintained by the university
Data recorded in OPED
Prescription data
CPR-number of patient
Date of purchase
Package number
Package
Volume
Strength
Dispensing form
ATC-code and DDD
Number of packages
Pharmacy
Prescribing practice
Price and reimbursement
Population data
CPR-number
Date of birth
Sex
Municipality of residence
Dates of migration
Date of death
National register of drug statistics
• Data collected by the Danish Medicines Agency
• Available under the research registers in Statistics
Denmark
• Anonymous data, person identifier not accessible
• Record linkage to other registers in Statistics Denmark
– Health registers
– Demographic data (residence, migration, death, family)
– Socioeconomic data (education, occupation, employment
status, income)
National register of drug statistics
• Authorised research institutions offered remote
access
• Externally acquired data with CPR-number can be
linked to the research registers (one-way procedure)
• Programs for data processing and analysis can be emailed and placed on server
• On-line access (secure connection)
• Results e-mailed back to user (screened for misuse:
single records or person identifiable data)
Incomplete coverage of dispensing registers
• Non-reimbursed drugs
(regional registers)
–
–
–
–
Benzodiazepines
Oral contraceptives
Certain antibiotics
ASA (only when prescribed to
aquire reimbursement)
– Paracetamol (only when
prescribed to aquire
reimbursement)
• OTC use
• In-hospital use
• Drugs dispensed through
hospital
pharmacies/outpatient clinics
– HIV treatment
– Anti-tuberculosis drugs
– Biologicals
Record linkage of register data
PRESCRIPTION
REGISTER
HOSPITAL
REGISTER
ID
ID
Date
Date
Drug
Diagnoses
Dose
Procedures
POPULATION REGISTER
ID, date, residence, birth, death, migration
Letigen (ephedrine/caffeine)
marketing suspended 2002 in DK
Confounding by indication
Letigen
Obesity
Myocardial infarction
Case-crossover design
Each person serves as his/her own control,
adjusting for time-independent confounders
Exposure status
Case / Control
No / Yes
Yes / No
Yes / No
Case time: MI
Exposure: Letigen
Control time (1 year before)
Effect period
Ephedrine/caffeine study results
• Among 2,316 case subjects, 282 (12.2%) were
current users of ephedrine/caffeine
• Case-crossover OR 0.84 (95% CI: 0.71, 1.00)
• After adjustment for trends in ephedrine/caffeine
use OR 0.95 (95% CI: 0.79, 1.16).
• Subgroup analyses: no strata with significantly
elevated risk
• Case-control substudy: no increased risk among
naïve users or users with large cumulative doses
Important information on medication and
patient factors missing
• Confounding factors in register-based
epidemiological studies
• Indication for drug (diagnosis)
• Recommended dosage
• Patient’s medical history, co-morbidities
• Lifestyle factors (BMI, physical activity, alcohol,
smoking, diet)
Information in patient records
GENERAL
PRACTICE
ID
Date
Diagnoses
Procedures
Prescriptions with
indications
Other clinical and
lab data
Lifestyle factors
PRESCRIPTION
REGISTER
HOSPITAL
REGISTER
SOCIO-ECONOMIC
DATA
SPECIALISED
CLINICAL
REGISTERS
POPULATION REGISTER
HOSPITAL
RECORDS
ID
Date
Clinical
examination
Lab data
Diagnostic
procedures
Drug use
Discharge
summary
Other current research examples
• Quality indicators for asthma treatment (patient
questionnaires and spirometry)
• Treatment of hypertension in general practice (GP
clinical information, patient questionnaires)
• Generic substitution, patient concerns and
compliance (patient questionnaires and interviews
)
Conclusions
• Prescription databases are important sources of
information on medication use, including the quality
of prescribing, and adverse effects
• General practice is responsible for the majority of
prescribing, treatment initiations and follow-up in the
population
• Important patient characteristics and information on
drug use are captured in the GP patient record
systems
• Pharmacoepidemiological studies should more often
have general practice as the starting point