The ADAG Study Group defined the relationship between average blood glucose (AG) and A1C. They derived an equation valid across diabetes types but excluded comorbidities known to impact A1C. The goal of this study is to examine the relationship between AG and A1C in patients omitted from prior analyses. A retrospective chart review was performed of downloaded glucose data from patients prescribed insulin at the UW Diabetes Center between 1/2011-10/2017. Patients were identified who had either 1) documentation and/or laboratory confirmation of anemia; 2) CKD; or 3) NAFLD. Eligible charts required: 1) 2 weeks of fingerstick glucose (SMBG) data with ≥14 checks or 1 month of CGM with ≥14 complete days and, 2) an A1C measured within 1 month of the encounter. The SMBG and CGM means were obtained from downloads. An AG was extrapolated from the A1c using the ADAG equation. The downloaded data and A1C-derived AGs were compared. There was a linear correlation between AG and A1C (R2=0.48) in anemic patients regardless of etiology. Non-anemic patients (N=220) had better ADAG agreement (70% of AGs inside 15% of predicted) than anemic patients (55% of AGs, N=123, p=0.006). Normal GFR yielded 71% of AGs inside 15% of ADAG estimation (N=440). This decreased to 59% if limited to anemic patients (N=83, p<0.001). ADAG agreement decreased as a function of renal impairment. The impact of anemia was only significant in CKD stage 3A (N=49, p=0.02). In CKD stage 4/ESRD there was no correlation between AG and A1C (N=14, R2=0.0004). Abnormal LFTs did not affect ADAG agreement. In NAFLD there was no correlation between AG and A1C (N=14, R2=0.03) but 64% of AGs were inside 15% of ADAG projection. For many, SMBG and CGM data provide an important confirmatory assessment of A1C. However, in certain conditions there is significant discordance between AG and A1C. There is a complete lack of correlation between AG and A1C in advanced renal dysfunction and NAFLD as measured by our lab. These findings suggest glucose downloads are a better assessment of diabetes control in several common comorbidities.


J.E. Perlman: None. J.L. Rosenbaum: None. B.K. McNulty: None. I.B. Hirsch: Research Support; Self; Medtronic MiniMed, Inc.. Consultant; Self; Abbott, Bigfoot Biomedical, Roche Diabetes Care Health and Digital Solutions, ADOCIA.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at