2018. Agreement of National Diabetes Database Compared with CHIPS Database
Bessie A Young, MD, MPH
Objectives: to assess the accuracy of the VA Diabetes QUERI (DM-QUERI) data abstracted for VISN 20.
Methods: We conducted a cross sectional comparison of data collected as part of the DM-QUERI national database for VISN 20, and data collected locally as part of the Consumer Health Information and Performance Sets (CHIPS) to determine reliability of data downloaded as part of a VA national quality assurance program. Patients actively treated for diabetes, were identified nationally for fiscal year 1999 (FY99) by the use of diabetes medications (insulin and oral hypoglycemic agents), and diabetes related equipment.
Results: During FY99, 16,422 individual veterans from VISN 20 were identified as having diabetes. Of those, DM-QUERI abstraction identified 15,240 veterans, while CHIPS abstraction yielded 16,396 veterans (agreement=92.6%). From those identified, 205,982 individual prescriptions for diabetes medications and equipment were documented. Of those, 145,552 were found in the DM-QUERI database, while 159,324 were identified from the CHIPS. Diabetes equipment represented 32.6% of the DM-QUERI and 31.1% of CHIPS prescriptions. Insulin represented 24.4% of DM-QUERI and 24.5% of CHIPS prescriptions, while oral hypoglycemic agents accounted for 43.0% of DM-QUERI prescriptions, compared with 44.4% of CHIPS prescriptions. There was substantial agreement between DM-QUERI and CHIPS databases (equipment k=0.59, insulin k=0.64, oral meds k=0.0.68).
Conclusions: National databases are reliable for identification of diabetic patients receiving VA care.