Medicare Provider Analysis and Review (MEDPAR) file includes information on 100 percent of Part A inpatient discharges (1986 to present). This file contains person-level data with unique identifiers (to allow linkage with other files), demographic data, clinical data, and information on days of care, charges, and provider of service. The diagnoses and procedures in this file are coded with ICD-9-CM codes (International Classification of Diseases, Ninth Revision, Clinical Modification). The MEDPAR file contains approximately 10 million records per year and is updated quarterly.

Second, the Medicare Automated Data Retrieval System (MADRS) contains 100 percent of Part A and B files, linked together. This file also contains person-level data with unique identifiers, demographic data, information on all institutional services, Part B payment records (summary level only), and limited medical information. MADRS is triggered by the inpatient admission, contains 250 million records, and is updated monthly.

Third, the Health Insurance Master Accretion (HIMA) contains 100 percent information on all Medicare enrollees (approximately 32 million). This file contains person-level data with unique identifiers, demographic data, name and address of the beneficiary, and dates of entitlement. HIMA is updated daily.

Finally, the Part B Medicare Annual Data (BMAD) file contains a 5-percent sample of beneficiaries and includes over 21 million records. The data are on a person level with unique identifiers. They include physician and supplier Part B claims for services and expense information (dates, place, and type of services rendered). The procedures in this file are coded with the HCFA Common Procedure Coding System (HCPCS), which is based on the CPT-4 (Current Procedure Terminology) system. BMAD is updated annually.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement