Data Governance in Healthcare starts with knowing what data belongs to each patient. But in the United States, where there is no authoritative source of patient identity, that knowledge is elusive.  

Previous methods for identifying a patient use affinity scoring. If two records have enough common information, a scoring machine may find that they represent the same patient. But this approach often undermatches (failing to link records that actually belong to the same patient), overmatches (linking records that actually belong to different patients), or creates non-definitive scores that require a human decision. 

No matching algorithm is perfect. But in healthcare, patient identification must be very intolerant of overmatches, which mix records for two different patients as if they were one patient. 

We present new methods for identifying a patient that leverage demographic information typically available in the healthcare system. Examples of those methods include:

• Splitting the question of identity into two (household v. patient)
• Household linkage cascades
• Contextual uniqueness inference
• Recognizing and remedying weak linkages
• Leveraging continuity of care

Speaker: Howard M Sragow

Director, Enterprise Data Governance Operations, Evernorth, Inc.

Howard is a thought leader in the usage of Data Governance in health care. He is the Director of Enterprise Data Governance Operations at Evernorth, a division of Cigna. He has held leadership roles in technology project management, technical consulting, and clinical research, and has made significant innovations in the field of patient-matching. He is an active member of both the National Council for Prescription Drug Programs and DirectTrust. Howard received his MBA from New York University.