The Center for Healthcare Improvement and Medical Effectiveness (CHIME) VA Data Core, located at the San Francisco VA, is directed by Katherine Hoggatt PhD and Mary Whooley MD. The CHIME VA Data Core facilitates access to national VA data for clinical research and quality improvement projects conducted by UCSF's VA-affiliated investigators.
The Veterans Health Administration (VA) maintains a central data repository that contains over 20 years of longitudinal electronic health record (EHR) data from 130 health care systems that have provided care to over 21 million United States Veterans. Data fields include (but are not limited to) demographic information, zip code, dates of outpatient and inpatient encounters, International Classification of Disease (ICD) and Current Procedural Terminology (CPT) codes, lab results, vital signs, medications, clinical notes, radiology reports, durable medical equipment, vital status, location of care, costs, and provider specialty. Data files from the Center for Medicare and Medicaid Services (CMS), including Medicare, Medicaid, and the US Renal Data System, are also available.
The CHIME VA Data Core:
- Provides consultation regarding data elements available in the national VA Corporate Data Warehouse and VA/CMS files.
- Assists with identifying, extracting, and merging variables of interest.
- Ensures that use of VHA data meets all regulatory requirements.
- Provides statistical support for projects using national VA and/or VA/CMS data.
- Does not provide statistical support for PI-assembled or hand-curated datasets.
Please note that there is little standardization across the 130 VA EHRs housed in the VA Corporate Data Warehouse. Thus, most of the time invested in national VA data projects is spent cleaning, harmonizing, and validating data elements. CHIME data analysts typically spend 150 to 200 hours assembling an analytic dataset before analyses can begin. Examples of prior projects are available here.
Obtaining a VA appointment and access to VA data may require up to 40 hours of time over a period of 3 to 6 months prior to starting data extraction.
Please feel free to contact Katy Williams with any questions.
