Claims story: How do you analyze most granular Patient claims data available commercially in US Healthcare? - 06/24

Patients claims data is the most valuable sources of data for healthcare analytics. Payor claims contain detailed diagnosis and procedure information for any billable patient visit. It enables to plot patient journey, find cohorts of patients for treatment or new drug discovery, find KOL doctors, target the right doctors, etc. 

But while analyzing this data, do we ever think how is it prepared and what are its components?

• Drug table: NDC code, manufacturer, brand name, drug generic name, strength, form, package size
  

• Physician table: Physician name, address, specialty, DEA (Drug Enforcement Administration) number, ME (Medical Education) number, NPI (National Provider Identifier) number, Opt out and No contact information
  

• Claims Prescription table: Claim ID, Drug ID, Patient ID, Practitioner ID, Plan ID, Plan pay, Patient pay, Quality, Days supply, Rx fill date
  

• Patient demographics table: Patient id, Sex, Date of birth year, Zip, state, Start & End year of treatment, Start & End year – month of Rx
  

• Claims Procedure table: Claim ID, Drug ID, Procedure code, Procedure date, Units administered, Charge amount, Service date, Plan type code 
  

• Plan table: Plan ID, Plan name, Plan type & description, Plan subtype, National insurer name, Organization name, Admin name, Admin type
  

• Claims diagnosis table: Patient id, claim id, Claim type, service date, service to date, Diagnosis code, Practitioner id, Diagnosis type code, Year month, load date

All these tables are joined to analyze the data and develop actionable insights. There are 100+ different metrics and 10+ tables which claims data is built, did you think about this before analyzing it?

Follow me on LinkedIn