3091 — Improvements in VA Patient Glycemic Control Over Time with Differences by Race and Co-Morbid Diseases
Miller DR (Bedford VAMC), Christiansen C
(Bedford VAMC), Chen X
(Bedford VAMC), Fincke BG
(Bedford VAMC), Tseng C
(East Orange VAMC), Pogach L
(East Orange VAMC)
Hemoglobin A1c as a measure of glycemic control is used by health care systems to monitor performance. We studied repeated measures of A1c in patients from the Department of Veterans Affairs health care system (VA) to determine trends, identify disparities, and better understand how these measures may be used to assess performance.
We used national computerized data from the Diabetes Epidemiology Cohorts (DEpiC) for all VA patients with diabetes and A1c testing in 2000-2004. To evaluate trends, we used longitudinal regression models with random effects for health care facilities and individual patients and with adjustment for age, sex, and seasonal variation.
Our analysis included 2,816,496 A1c measures from 248,768 diabetes patients (average of 2.3 per patient per year). Average A1c was 7.27% in 2000 and there was an overall adjusted decline of 0.06% per year. Levels were lowest in patients who had incident diabetes, were new to VA care, or were older, near death, or had serious co-morbid conditions. Trends in the relative proportions of these patients contributed to the A1c trends. Seasonal variation in A1c (high in winter, low in summer) was striking. Among surviving, prevalent diabetes patients at baseline, A1c levels were higher in African Americans (7.91%) and Hispanics (7.81%) as compared to whites (7.60%). The subsequent decline was greatest in African Americans (0.40% / year) so that 2004 A1c levels were close to those of whites; this was not observed for Hispanics, however, who had only minimal decline (0.13%).
This study shows how A1c levels are associated with many factors extraneous to the affects of health care on patient status and, thus, they may limit inferences regarding “quality of care”. Nevertheless, even after considering these factors, recent improvements in glycemic control are evident across the VA patient population with differences in trends by race, ethnicity, and co-morbidity status.
This research demonstrates that longitudinal analyses can be used to assess changes in glycemic control in patient populations provided that a number of potential confounding factors are considered in the analysis.