TIM SOON CHEOK 1, HENRIK FALHAMMAR 2, 3, 4, SANDAWANA WILLIAM MAJONI1, 4
1Northern Territory Medical Program, Flinders University, Darwin, Northern Territory; 2Department of Endocrinology, Metabolism and Diabetes, Karolinska University Hospital, Stockholm, Sweden; 3Department of Molecular Medicine and Surgery, Karolinska Institutet, Stockholm, Sweden; 4Menzies School of Health Research and Royal Darwin Hospital, Tiwi, Northern Territory
Aim: to determine the accuracy of HbA1C and Glycated Albumin versus continuous glucose monitoring in diabetic patients on dialysis
Background: HbA1c is poorly correlated with glycaemic control in patients with diabetes undergoing dialysis. Glycated albumin (GA) is better estimate of glycaemic control.
Methods: We performed a systematic search of MEDLINE, CIHNAL, The Cochrane Library, Web of Science and Google Scholar, and extracted studies providing measures of two or more of HbA1c, GA and continuous glucose monitoring (gold standard) in diabetic patients on dialysis. All the studies recorded mean of the glucose monitoring, HbA1C or glycated albumin. No study recorded tests of diagnostic accuracy. Bland-Altman plot analysis was therefore performed to assess the agreement of these measures of glucose levels.
Results: 37 studies with 44 data points from 17 150 participants out of 314 studies were analysed. The results showed both tests as acceptable. However, there were significant proportional errors for either method against mean blood glucose. There was no significant errors for HbA1C vs glycated albumin; mean blood glucose vs HbA1C[ Mean difference: 119.9 (95% CI 111.8 to 128), Pitman’s Test of difference in variance: r = 0.71, n = 23, p<0.001], mean blood glucose vs glycated albumin [(Mean difference: 144.8 (CI 129.54 to 160.0) Pitman’s Test of difference in variance: r = 0.89, n = 13, p < 0.001)] HbA1C vs glycated albumin [Mean difference: 25.8 [(CI 22.2 to 29.4) Pitman’s Test of difference in variance: r = 0.21 , n = 28, p = 0.274]
Conclusion: Both HbA1C and glycated albumin seem to measure mean blood glucose with acceptable bias and precision. However, studies with tests of diagnostic accuracy are lacking.