NHLBI Workshop

Data Needs for Cardiovascular Events, Management, and Outcomes

Academic Medical Centers - Dr. Veronique Roger

Rochester Epidemiology Project (REP)

• The REP is not a database, it is a records linkage system
• REP studies are labor intensive
• Information in multiple sources, paper, and electronic format. Formats vary across sources and over time. Increasing need for IT…
• Access to over half a century of ~ complete IN and OUT patient data on a geographically-defined population
  

Olmsted County, MN, Laboratory for Epidemiological Studies

• Geographically isolated, home of Mayo Clinic. Almost all care delivered by a few providers.
• Since 1907, Mayo patients assigned a unique identifier, information in a unit medical record (hospitals, offices, ED)
• Since 1945, diagnoses and surgical procedures indexed
• REP expanded indexing and medical records linkage to non-Mayo care providers.
• Lengthy follow-up for fatal/nonfatal outcomes. Follow-up for vital status ~ complete
• Population-based denominators from decennial

Electronic Medical Record (EMR) and Research

• For clinical / QA,QC / financial needs
• Research second step
• Mayo/IBM Life Sciences system (live July 2005) umbrella for:
  
  • Clinical notes (dictated)
  • Lab/imaging tests
  • Resources utilization/Billing data
    

Use of the EMR for surveillance

• Examples
  • Active recruitment of Acute Coronary Syndrome (ACS) using lab data
  • Active recruitment of Heart Failure (HF) Using Natural Language Processing from clinical notes
    
• Need to:
  • Identify population
  • Validate cases using definitions, manually
  • Conduct research-driven measurements
    
• Future directions
  • Tracking of non-fatal outcomes
  • Tracking of heath care delivery patterns
    

Strengths and limitations

• Strengths
  • Opportunity for efficiencies, timeliness
  • High quality detailed clinical documentation
  • Access to outpatient data
  • Active surveillance capabilities
  • The population
    
• Limitations
  • Systems designed for clinical purposes
  • Research applications considered at best as an after thought, often not at all
  • Human interface still needed
  • Still labor intensive
  • The population
    

Lessons learned

• EMR systems in academic medical centers differ from one another, are fragmented and at various stages of maturation
• Sizable challenges for collaborative work
• As study needs/designs differ, customization likely unavoidable
  

Opportunity to partner to address CV data needs

• Clinical and research interfaces between: a) Clinical research - Drug/device trials and b) Epi/HSR research, pharmaco-epi, QA...
  • a) Clinical research - Drug/device trials
    • Study operations
    • Study data
    • Document management
  • b) Epi/HSR research, pharmaco-epi, QA...
    • Patient information environment
• For surveillance, need denominators/defined populations

Value of the model for new studies to address CV data needs

• Linkage system
• Population denominators
• Replication conceptually easier in electronic world
• IT intensive

References

• Bristol N, Lancet, 2005; 365: 1610-1611
• Pakhomov SV, Journal of Biomedical Informatics; 2005 38:145-153
• Academic Health Centers' Clinical Research Forum (http://ahcforum.org)
• Association of American Medical Colleges (Recommendations from conference in IT enabling clinical research, 2002)
  

Back to Workshop Agenda