Databases or Registries?Databases or Registries? Points to Consider Points to Consider

Mary Lou Skovron, DrPHMary Lou Skovron, DrPHGroup Director, Global EpidemiologyGroup Director, Global Epidemiology

Bristol-Myers SquibbBristol-Myers SquibbFDA/Industry Statistics WorkshopFDA/Industry Statistics Workshop

September 29, 2006September 29, 2006

OverviewOverview

Claims databasesClaims databases• TypesTypes• AdvantagesAdvantages• LimitationsLimitations

RegistriesRegistries• TypesTypes• AdvantagesAdvantages• LimitationsLimitations

ExamplesExamples ConclusionsConclusions

Claims DatabasesClaims Databases

Claims DatabasesClaims DatabasesTypesTypes

Open-Plan eg UHC, PharMetricsOpen-Plan eg UHC, PharMetrics• Pharmacy-basedPharmacy-based• Practice-basedPractice-based• Hospital-basedHospital-based

HMO, eg Kaiser-PermanenteHMO, eg Kaiser-Permanente Government eg Medicare, MedicaidGovernment eg Medicare, Medicaid

Claims Databases:Claims Databases:AdvantagesAdvantages

Already exist, accruing patient Already exist, accruing patient information…useful when a relatively information…useful when a relatively quick answer is neededquick answer is needed

Potentially large populations from Potentially large populations from which to draw treated patient and which to draw treated patient and comparison samples…statistical comparison samples…statistical power not usually an issuepower not usually an issue

May include subgroups not included May include subgroups not included in clinical trials…expand knowledgein clinical trials…expand knowledge

Claims Databases:Claims Databases:LimitationsLimitations

ICD-9 coding… potential for misclassification ICD-9 coding… potential for misclassification • Limited sensitivity/specificity Limited sensitivity/specificity • Coding affected by reimbursementCoding affected by reimbursement

Short ‘residence’ in the database Short ‘residence’ in the database Clinical, lab, imaging data not usually presentClinical, lab, imaging data not usually present Rare events in rare diseases require huge Rare events in rare diseases require huge

“electronic” populations to identify adequate “electronic” populations to identify adequate numbers treatednumbers treated

May not represent important sub-populationsMay not represent important sub-populations Surveillance bias and channeling biasSurveillance bias and channeling bias Inpatient drug exposures not usually recordedInpatient drug exposures not usually recorded

Claims Databases: Claims Databases: ApplicationApplication

Acute events (short-term events)Acute events (short-term events) Events that come to medical attention, eg Events that come to medical attention, eg

CVACVA Validated algorithm to identify indication Validated algorithm to identify indication

and event of interest OR medical record and event of interest OR medical record review to verify eventsreview to verify events

Statistical approaches for confounders (eg Statistical approaches for confounders (eg propensity scores, instrumental variables, propensity scores, instrumental variables, risk factor scores, multivariate analyses) risk factor scores, multivariate analyses)

RegistriesRegistries

RegistriesRegistries

““A systematic collection of defined events A systematic collection of defined events or product exposures in a defined patient or product exposures in a defined patient population for a defined period of time”population for a defined period of time”11

““A registry A registry per seper se is not a study. It is an is not a study. It is an organized collection of data in humans organized collection of data in humans within a particular disease group or other within a particular disease group or other special group…”special group…”22

1Arlett P, Moseley J, Seligman PJ p 119 in Pharmacoepidemiology Fourth Edition, Strom BL, ed 200522 CIOMS V Section II.h. CIOMS V Section II.h.

Registries:Registries:TypesTypes

Drug/Device Registry: Includes Drug/Device Registry: Includes subjects receiving the drug or device subjects receiving the drug or device regardless of indicationregardless of indication

Disease Registry: Includes patients Disease Registry: Includes patients with the disease regardless of drug with the disease regardless of drug or device exposureor device exposure

Registries:Registries:StrengthsStrengths

Richer data than in electronic databases: Richer data than in electronic databases: Patient SES, history, treatment, clinical Patient SES, history, treatment, clinical data can be collecteddata can be collected

Define encounter frequency and follow-up Define encounter frequency and follow-up durationduration

Event ascertainment does not depend on Event ascertainment does not depend on ICD-9 codingICD-9 coding

Can address additional questions in the Can address additional questions in the datadata

Registries:Registries:LimitationsLimitations

Accrual can be slow if insufficient Accrual can be slow if insufficient sites engagedsites engaged

Generalizability must be establishedGeneralizability must be established Practical limits on number of patients Practical limits on number of patients

followedfollowed

Registries:Registries:ApplicationApplication

Long-term outcomesLong-term outcomes Rare diseasesRare diseases Potential confounders importantPotential confounders important Multiple objectivesMultiple objectives

Usefulness of RegistriesUsefulness of Registries

Characteristics of patients in the Characteristics of patients in the target population for the new drugtarget population for the new drug

Clinical course of the diseaseClinical course of the disease Treatment patterns, health care Treatment patterns, health care

utilizationutilization Frequency of adverse eventsFrequency of adverse events

Registry LifecycleRegistry Lifecycle Early: Describe patient demographics, Early: Describe patient demographics,

clinical characteristics, practice patterns clinical characteristics, practice patterns (usually cross-sectional analyses); (usually cross-sectional analyses); incidence of AEs with short lag timesincidence of AEs with short lag times

Intermediate: Analyze relationships pf Intermediate: Analyze relationships pf patient characteristics, treatment with patient characteristics, treatment with outcomes; outcomes; incidence of less common incidence of less common AEs, AEs with longer lag times; assess AEs, AEs with longer lag times; assess risk factors for AE incidencerisk factors for AE incidence

Advanced: Evaluate changes in practice Advanced: Evaluate changes in practice patterns; impact on outcomes and patterns; impact on outcomes and AE AE incidence; assess rare AE incidenceincidence; assess rare AE incidence

Registry QualityRegistry Quality

DeCIDE Network currently DeCIDE Network currently developing a reference document on developing a reference document on developing, conducting and developing, conducting and evaluating registriesevaluating registries

Sponsored by AHRQSponsored by AHRQ Document in draft, Outline available Document in draft, Outline available

on the webon the web Report currently in draftReport currently in draft

ExamplesExamples

Example:Example:Cox-2 Inhibitors and Cardiac Cox-2 Inhibitors and Cardiac

EventsEvents Cox-2 use frequent in populationCox-2 use frequent in population Primary care drugPrimary care drug Cardiac events not rare in target Cardiac events not rare in target

populationpopulation Short-term (< 2 years) eventsShort-term (< 2 years) events Clinical history importantClinical history important

One SolutionOne SolutionCox-2 Inhibitors and Cardiac Cox-2 Inhibitors and Cardiac

EventsEvents Claims database analysisClaims database analysis11

• Advantages: Advantages: Data already accrued when study need identifiedData already accrued when study need identified Large populationLarge population

• Limitation offsetLimitation offset Multivariate regression to control confoundingMultivariate regression to control confounding Verified cardiac events by chart review Verified cardiac events by chart review

• Remaining limitationsRemaining limitations Under-represented > 65 yoUnder-represented > 65 yo

1Velentgas P et al 2006

Key Feature of the SolutionKey Feature of the Solution

Numbers readily available: Numbers readily available: • Exposure: ~ 425,000 eligible subjects available Exposure: ~ 425,000 eligible subjects available

for studyfor study at least one dispensing of the 5 study medications at least one dispensing of the 5 study medications

during preceding 18-month periodduring preceding 18-month period first dispensing after minimum of six months without first dispensing after minimum of six months without

any of the medications any of the medications

• Endpoint: ~ 725 confirmed MI/ACSEndpoint: ~ 725 confirmed MI/ACS Verification of endpointsVerification of endpoints

• Medial record review applying accepted criteriaMedial record review applying accepted criteria

Example:Example:Thrombolytics And BleedingThrombolytics And Bleeding

Important focus: subpopulations eg Important focus: subpopulations eg ethnicity, gender, ageethnicity, gender, age

Short-term eventShort-term event Benefit and riskBenefit and risk Hospital-based treatmentHospital-based treatment

SolutionSolutionThrombolytics and BleedingThrombolytics and Bleeding

Registry approach: National Registry of Registry approach: National Registry of Myocardial InfarctionsMyocardial Infarctions

Salient strengths:Salient strengths:• Clinical and lab data ascertainedClinical and lab data ascertained• Data from medical recordsData from medical records

Limitation OffsetLimitation Offset• Validated completeness against another data sourceValidated completeness against another data source• Large number of hospitals to assure adequate Large number of hospitals to assure adequate

subpopulationssubpopulations Remaining limitationsRemaining limitations

• Short-term data cannot answer long-term questionsShort-term data cannot answer long-term questions

Key Feature of the SolutionKey Feature of the Solution Large number of hospitals participate order to Large number of hospitals participate order to

gather data on~200,000 MIs per yeargather data on~200,000 MIs per year Rich patient, clinical, treatment and outcome Rich patient, clinical, treatment and outcome

informationinformation Answered the question about safety in subgroups Answered the question about safety in subgroups

under-represented in the clinical trialsunder-represented in the clinical trials In its > 10 year lifecycle has contributed to broad In its > 10 year lifecycle has contributed to broad

understanding and improvement of medical understanding and improvement of medical practicepractice

External investigators can propose research; External investigators can propose research; external advisory group reviews and approves all external advisory group reviews and approves all researchresearch

Example: Safety of OrenciaExample: Safety of Orencia®®, a , a New Biological for RANew Biological for RA

Short-term (infections) and longer-Short-term (infections) and longer-term (NHL and other malignancies) term (NHL and other malignancies) eventsevents

Low general population prevalence of Low general population prevalence of RA candidates for biologics RA candidates for biologics treatmenttreatment

RA a specialty-treated disorderRA a specialty-treated disorder Proactive approachProactive approach

Solution: Complementary Solution: Complementary ApproachesApproaches

Claims database study in UHC data Claims database study in UHC data • Event validationEvent validation• Large pool of potential comparatorsLarge pool of potential comparators• Population treatment patternsPopulation treatment patterns• Infections, other possible short-term eventsInfections, other possible short-term events

Registry building on the independently conducted Registry building on the independently conducted National databank for Rheumatic Diseases National databank for Rheumatic Diseases • Large pool of participating rheumatologistsLarge pool of participating rheumatologists• 5,000 Orencia® initiators comparison to >=15,000 5,000 Orencia® initiators comparison to >=15,000

initiators/switchers of other treatmentsinitiators/switchers of other treatments• Patient self-report and event verificationPatient self-report and event verification• Enrollment and retention enhancementsEnrollment and retention enhancements• Short and long-term eventsShort and long-term events

Key Features of the SolutionKey Features of the Solution

External scientific advisory groupExternal scientific advisory group• Individual protocolsIndividual protocols• Statistical analysis plansStatistical analysis plans• Interpretation of results including Interpretation of results including

approaches to integrationapproaches to integration

ConclusionsConclusions

ConclusionsConclusions

One size does not fit all: evaluate One size does not fit all: evaluate optionsoptions

Registries: a useful alternative to Registries: a useful alternative to electronic databaseselectronic databases

Consider complementary approachesConsider complementary approaches

Useful ReferencesUseful References

Strom BL, ed; Pharmacoepidemiology Strom BL, ed; Pharmacoepidemiology 44thth Edition; UK; John Wiley and Sons Edition; UK; John Wiley and Sons Ltd;2005Ltd;2005

AHRQ DeCIDE Network upcoming AHRQ DeCIDE Network upcoming publicationpublication• http://effectivehealthcare.ahrq.gov/decide/http://effectivehealthcare.ahrq.gov/decide/

Thank YouThank You