Leveraging payer data to enhance
longitudinal capture of outcomes
Lesley Curtis
Center for Pragmatic Health Systems Research
Consider a patient with 8 health care encounters in
a given health system…
What might we have missed?
Consider a patient with 8 health care encounters in
a given health system with linked payer data…
Claims vs payer data
Consider a patient with 8 health care encounters in
a given health system with linked payer data…
Claims vs payer data
Inpatient & Outpatient EHR Payer Claims
Overlap
Gregory E. Simon, MDGroup Health Research Institute
Outpatient EHR C Outpatient
EHR A Inpatient EHR B Payer #1 Payer #2 Inpatient EHR B
Overlap
Gregory E. Simon, MDGroup Health Research Institute
•
Complete capture less important for health states that arestable (e.g. weight) or care events that leave an enduring signal (e.g. reversal of gastric bypass)
•
More important for health states that are labile (e.g.hemoglobin) or care events that leave only a transient signal (e.g. suicide attempt)
When does completeness matter?
Gregory E. Simon, MD
• High expectations when exposure outcome occur in the same care setting and close in time
– comparing in-hospital mortality for alternative bariatric procedures
• Lower expectations when exposures and outcomes occur in different care settings and/or different time periods
– comparing incidence of anemia over 2 years for alternative bariatric procedures
When should we expect completeness?
Gregory E. Simon, MD
Completeness through the lens of the ACO
• Medicare FFS beneficiaries are not required to select a primary care physician
• Assignment of patients to providers based on utilization data (e.g., plurality of E&M claims)
Loyalty to assigned hospitals
Bynum JPW HSR 42:1, Part I (Feb 2007)
• 30% of care received by Medicare beneficiaries occurs outside of the hospital to which they are assigned
– Lower for surgical admissions at small hospitals, likely due to narrower spectrum of services
Stability to assigned physicians
Pham N Engl J Med 2007; 356:1130-9
• Assignment changed for 46% of Medicare beneficiaries
Stability of assignment and leakage of outpatient
care
McWilliams JAMA Intern Med 2014;174(6):938-945
• 8.7% of office visits were provided by PCPs outside of the assigned ACO
• 67% of office visits with specialists were provided outside of the assigned ACO.
Challenges of linkage with payer data
• Governance and privacy trump the technical
– Release of PHI outside of institutional firewalls
– Rules of engagement around the use of linked data set
• Proprietary (business privacy) concerns are not far behind • Many, many payers represented in each EHR
– 26+ for Duke health system patients – Data use agreements required for each
• Ignores the uninsured, some of whom may be highly complex • Death remains a major problem
