NHLBI Workshop

Data Needs for Cardiovascular Events, Management, and Outcomes

Data Management Issues - Dr. Wayne Rosamond


Some data management issues to consider

  • Infrastructure
    • What minimal structure needs to be in place to monitor case presentation, treatment patterns, and outcomes for 4 different event types (sudden cardiac arrest, acute coronary syndrome, stroke, heart failure)?
    • How do we balance needs for high quality data with practical (and budget) constraints on infrastructure?
    • How can current infrastructure be best incorporated with new initiatives?
  • Coordination
    • How best to oversee, organize, communicate, and otherwise coordinate data collection, validation, quality control, analysis, and data distribution?
    • How best to balance centralized versus local control and access to data?
  • Analysis
    • How best to manage the data to address various analytic issues we can expect to face?
  • Promotion
    • How can we promote the use of the data on a broad scale?
    • Howe can widespread use of data be balanced with need to maintain high analytic standards?
  • Ethics
    • What policies should be established for communicating clinical data (diagnostic) to subjects?

How does data management fit into workshop goals?

  • National, population-based data needs
  • Acute coronary syndrome, cardiac arrest, stroke, heart failure
  • Surveillance systems that are feasible and sustainable
  • Cardiovascular treatments and outcomes
  • Ascertainment of incidence and prevalence
  • Registries, quality improvement systems, research studies

Data management and flow -- can be a complex web of procedures and system, starting from data collection to ascertainment of outcomes and determination of incidence rates

Infrastructure/coordination issues

  • Data collection
    • Retrospective vs. prospective
    • Concurrent with care vs. chart-based
    • Real time vs. batch mode
    • Validation

Example of Patient ID and Data Collection System: NC Coverdell Acute Stroke Registry

NC Coverdell Acute Stroke Registry

Prehospital delay time for stroke patients as recorded from interview and medical records (Evenson E, Rosamond W, Vallee J, Morris D. Concordance of stroke symptom onset time. Ann Epidemiol 2001;11:202-207.):

  Interview Medical Record
Mean (hr) 9.8 8.9
Median (hr) 3.3 3.1
Interquartile range 1.3-9.1 1.2-9.0
Percent with onset time recorded 95% 60%

EMS Trip Sheet Can Capture (Rosamond W. et al. Calling emergency medical services for acute stroke. A study of 9-1-1 tapes. Prehospital Emergency Care 2005;9:1-5.):

  • Time arrived at scene
  • Time departed from scene
  • Time arrive at hospital
  • Response code
  • Transport code
  • Treatments provided

Example -- Validation of sudden cardiac death

  • Comparison (n) of Reynolds sudden cardiac death review and ARIC sudden death classification using 1 hours definition
 
Reynolds Sudden Cardiac Death Classification
ARIC Sudden Death Classification (1 hour definition)
Definite Sudden Death
Possible Sudden Death
Not Sudden Death
Total
Yes
129
21
32
182
No
125
38
143
306
Unknown
5
3
21
29
Total
259
62
196
517

Analysis issues

  • Sampling
    • Can increase efficiency
    • Straightforward methods exists
  • Local vs. central control
    • Quality improvement requires local access to analysis

Promoting the use of data

  • Public use datasets
  • Meta-analyses
  • Advertising
  • Symposium on how to use data
  • Local sites, real-time access to analysis
  • Fellowships and student involvement

Ethical issues: A framework (Principles and Practice of Public Health Surveillance. 2nd Edition. Teutsch and Churchil Eds. Oxford University Press, 2000 )

  • Respect for autonomy
  • Beneficence
  • Nonmaleficence
  • Justice

Ethical Framework applied to surveillance

  • Reasons for undertaking activity?
  • Benefits vs. harms vs. costs?
  • Resolution of similar ethical problems in the past?
    • Informing subject of test results?
      • Learn from past/current cohort studies
    • Under real time methods should you inform subjects at discharge that their care didn't meet guidelines?
      • Less experience in retrospective community surveillance
  • Community involvement?
  • Violation of rights?
  • Demonstratable virtues?

Summary

  • Attention to data management critical to insure quality data collection and analysis.
  • Reality of real-time data collection and analysis create new data management challenges.
  • Data management systems must balance practical, ethical, and data quality issues.
  • Data management systems must allow for broad access to and dissemination of analyses.

Back to Workshop Agenda

Skip footer links and go to content
Twitter iconTwitterExternal link Disclaimer         Facebook iconFacebookimage of external link icon         YouTube iconYouTubeimage of external link icon         Google+ iconGoogle+image of external link icon