A Validated Hemoglobin A1C Calculator That Predicts Long Term A1C after Short Term Repeat A1C Measurement

Presentation Number: MON 628
Date of Presentation: April 3rd, 2017

Christopher T Martin*1, Cary N Mariash2, Ameer Khowaja3, Sidney A Jones4 and Rupendra T Shrestha1
1University of Minnesota Medical Center, Minneapolis, MN, 2Methodist Research Institute and Indiana University, Indianapolis, IN, 3Hennepin County Medical Center, Minneapolis, MN, 4Ridgeview Medical Center, Chaska, MN

Abstract

Introduction

Hemoglobin (HgB) A1C is a commonly utilized clinical marker of glycemic control in diabetic patients. Nevertheless, A1C measurements do not accurately reflect mean glucose levels when measured prior to complete turnover of red blood cells (RBC). Consequently, many insurers will only cover the cost of A1C measurements obtained at intervals greater than 90 days. This potentially leaves medical providers without an accurate estimate of glycemic control due to changes in therapy and patient lifestyle. Therefore, it would be desirable to have a prediction of the steady state A1C at any time it was measured. To that end, we developed and validated a calculator available online and on hand-held devices able to provide such information.

Methods and Results

The equation to predict the steady state A1C is derived from the assumptions of steady state kinetics and first order rate reactions. The calculation is based on knowing the days between two A1C measurements and estimated RBC half-life. The half-life estimate defaults to 35 days based on prior studies, but is user adjustable. We validated the calculator with a retrospective chart review of 2,420 patient records from the Fairview Medical System. To assure relevancy of the review, inclusion criteria included age greater than 18 years, three hemoglobin A1C values and a single diabetes medication change during that time. Calling the three HgB A1C results in order of time A1C-1, A1C-2 and A1C-3, we required A1C-2 minus A1C-1 to be greater than 20 and less than 71 days and A1C-3 minus A1C-1 to be greater than 85 days and less than 225 days. We furthermore required a sole diabetes medication change to occur within 14 days of A1C-1 without subsequent diabetes medication changes between A1C-1 and A1C-3. Exclusion criteria included any occurrence in the patient medical record of end stage renal disease as defined by international classification of diseases version 10 diagnosis code N18.6, creatinine greater than 5.0 milligrams / deciliter and HgB less than 8.0 grams / deciliter. The calculator was validated by comparing the predicted steady state A1C based on historical A1C-1 and A1C-2 with historical A1C-3. The correlation between A1C-3 and Predicted A1C was strong and highly significant (P < 10^-7, R = 0.62).

Conclusion

The calculator has the advantage of providing both the clinician and the patient with a predicted steady state A1C and mean glucose concentration. For example, one can repeat an A1C one month after instituting changes in diabetes management and predict what the A1C will be many months later, thereby demonstrating improved glycemic control in a shorter duration.

 

Nothing to Disclose: CTM, CNM, AK, SAJ, RTS