Presenter Disclosure Information• Moira Kapral
• Challenges and Opportunities in Linking Administrative Claims Data with Stroke Registry Data
FINANCIAL DISCLOSURE: None
UNLABELED/UNAPPROVED USES DISCLOSURE: None
Challenges and opportunities in linking administrative claims data with registry data: the Registry of the Canadian Stroke Network
Moira K. Kapral MD, MSc, FRCPCMoira K. Kapral MD, MSc, FRCPCMay 2010May 2010
Overview
• Description of the Registry of the Canadian Stroke Network (RCSN)
• Administrative databases available for linkage
• Linkage process
• Advantages and disadvantages of linking registry to administrative data
Ontario, Canada
• Canada's largest province
• Population 13 million
• Urban and rural
• Ethnically diverse
• Universal health coverage with single payer
Ontario Stroke System
• Regionalized stroke care
• Designated stroke centres
• Transfer and bypass protocols
• Funded by Ministry of Health
• Part of a nationwide stroke strategy
• Requirement for measurement of monitoring of the quality of stroke care delivery
Evaluation of the provincial stroke strategy
• Mandate for reporting regional performance on key stroke quality indicators
• 23 indicators including
– Thrombolysis
– Neuroimaging
– Stroke unit care
– Carotid imaging
– Antithrombotic therapy
– Risk factor modification
• Need for high-quality clinical database
Registry of the Canadian Stroke Network
• Clinical database founded in 2001
• Patients with acute stroke or transient ischemic attack admitted to hospital or seen in the ED of acute care institutions
• Four phases with varying methodology
• Funded by the Canadian Stroke Network
and the Ontario Ministry of Health
and Long-Term Care
• www.rcsn.org
Registry of the CSN
Emergency Emergency Department Department
DataData
Discharge Discharge
DataData
Administrative DataAdministrative Data-Hospitalizations-Hospitalizations-Mortality-Mortality-Physician Services-Physician Services-Provincial Drug Formulary-Provincial Drug Formulary
Core Database
Entry Criteria:• ED diagnosis of stroke/TIA
• onset 2 weeks of hospital visit
EMSEMS
DataData
Hospital Hospital AdmissionAdmission
DataData
Data collected
• Demographics
• Pre-hospital/EMS and emergency department care
• Stroke data – type, subtype, severity, scales
• In-hospital interventions, consultations, complications
• Medications – prehospital, during admission, at discharge
• Investigations
• Disposition
Methodology
• laptop computer with custom software for data entry
• electronic transfer of data to coordinating centre
• web-based module also in use (SPIRIT)
Intelligent Data Entry Improves Data Quality• only appropriate fields appear (if … then “pop-ups”)
• few text fields (check boxes or choice fields)
• range checks
• logic checks – e.g. only correct sequence allowed
• data completeness checks
• double entry of critical fields
• display of time intervals, age for reality checks
Characteristics of a high quality clinical database
completeness of recruitment
completeness of data
use of explicit definitions of variables
data validation
Black N, Barker M and Payne M. BMJ 328:1478, 2004
Data Transfer
Institute for Clinical Evaluative Sciences (ICES)
• Established by provincial government to perform research related to equity, access and quality of health care
• Administrative data housed there by special agreement
• Strict data security measures
RCSN phases 1 and 2: 2001-2003
• 21 stroke centres across Canada
• Consent-based with 6-month follow-up interviews for functional status and quality of life
• Problems with consent led to biased sample
RCSN “Prescribed” in PHIPA 2004
• The RCSN is one of only four registries in Ontario that have been granted 'prescription' in the regulations of the Ministry of Health and Long-Term Care under s.39(1)(c )of the Personal Health Information Privacy Act 2004.
• RCSN collects data without consent, “for the purposes of facilitating or improving the provision of health care”
• RCSN is the primary means of monitoring and evaluating acute stroke care and outcomes in Ontario
RCSN Phase 3: 2003 onwards
• Data collected without consent, “for the purposes of facilitating or improving the provision of health care”
• All patients at 11 Ontario stroke centres (core RCSN)
– N > 40,000
• Population-based audit (RCSN Ontario Stroke Audit)– 15-20% of all Ontario stroke cases at all 150
hospitals (n~5,000/year)
• Patients at secondary prevention clinics
Clinics
Stroke centers
Province-wide
audit
Data Security
• laptop computers use finger print readers for password protection
• data encrypted using BestCrypt ® software
• personal patient information stripped before data sent to ICES (encrypted health card number sent separately)
• encrypted data uploaded to ICES by direct unpublished telephone line
• data kept on a secure server without connections to Internet or Intranet
• ICES has physical security barriers
• data security and privacy policies
www.ices.on.ca Investigative Reports
Marked variations in tPA by hospital type, 2002/03 and 2004/05
Regional variation in stroke unit admissionsOverall rate 18.4%
Publications
Why link registry to administrative data?
Rationale for linkage to administrative data
RCSNstroke cohort
Follow up for readmissions, medications, deaths
Characterize based on
geographic and area factors
Pre-stroke conditions, care, drugs
Advantages of linked registry and administrative data
• Registry creates well-characterized cohort of stroke patients, with detailed baseline clinical data
• Long-term follow up through administrative data
– Less expensive than clinical follow up
– Minimal loss to follow up
• For evaluation of stroke systems and regionalized care, permits evaluation of association between interventions and outcomes (mortality, readmissions, patterns of care)
Database Variables
Registered Persons/Vital Statistics
Mortality
CIHI Discharge Abstract DB Hospital separations
CIHI National Ambulatory Care
Emergency department and ambulatory visits
Drug Benefits Database Prescription claims for those aged > 65
Canada Census Area-level income, education
Physician Claims Outpatient visits, procedures
Population-based, comprehensive, validated
Administrative data sources in Ontario
Process for linkages
• Need unique patient identifier: health card number
• Considered most sensitive piece of personal health information – need stringent data security procedures
• Health card number collected in RCSN database
• Not transferred with other data; sent on separate disc to specific data custodian at ICES
• Scrambled to create a new unique ID number; kept on a separate server with no connection to Internet in an area with restricted access
Challenges in using linked data
• Collection of unique patient identifier necessitates stringent, time-consuming and expensive data security measures
– Development of protocols and procedures
– Personnel to implement
– Programming and software
– Security of data facility
• Cannot export or share linked dataset
Challenges in using administrative data
• Population-based data sources not always available
• Not all variables of interest available in existing databases
– Functional status, quality of life, laboratory data, biomarkers, genetic tests
• Coding/miscoding
• Claims may not reflect reality
• Experience required for proper use and interpretation
Conclusions
• Linked registry and administrative data ideal for studying both processes and outcomes of stroke care
• Should be considered for jurisdictions that are establishing regionalized systems of stroke care, to allow evaluation of return on investment
• Main challenges are
– Availability of appropriate databases
– Data security
– Expertise in linkages and analyses
• Worthwhile investment for policy-makers and government
Advertisement for RCSN database
• AVAILABLE TO YOU for research projects
• Analyses done on-site at ICES and funded by RCSN grant
• Need to collaborate with RCSN investigator
• Project request forms available at www.rcsn.org
Thanks and questions