A mathematical model to estimate mean blood glucose by oral glucose tolerance test

Abstract

Objective To establish a method for estimating mean blood glucose using oral glucose tolerance test (OGTT). Methods A retrospective study was conducted on the patients with type 2 diabetes mellitus (T2DM) in our department between January 2014 and December 2017. All the patients underwent continuous glucose monitoring (CGM) and OGTT. The basic demographic characteristics and results of laboratory tests of the patients were collected. Then relevant mathematical models were established and modified. In the end, external validation was carried out according to the goodness of fit formula. Results Thirty-six diabetic patients who met the enrollment criteria were recruited. The 24-hour mean blood glucose (24 h MBG) was estimated to be 9.66±3.34 mmol/L by OGTT (MBGOGTT), to be 9.40±2.67 mmol/L through CGM calculation. MBGOGTT showed significantly positive correlation with both glycosylated hemoglobin (HbA1c, r=0.582, P < 0.001) and 24 h MBG (r=0.674, P < 0.001). We used linear regression model and established following equation to estimate 24 h MBG by OGTT: 24 h MBGmmol/L=4.166+0.545×MBGOGTT, (r=0.674, P < 0.001). Furthermore, a modified model incorporating HbA1c effect was also obtained: 24 h MBGmmol/L=1.797+0.359×MBGOGTT+0.565×HbA1c (R=0.768, P < 0.001). We selected 18 patients admitted to our department for external validation. In this verification result, we obtained a RNL value of 0.81, which implied a high fit degree. Conclusion A model to evaluate mean blood glucose by OGTT is successfully established, and can be used as a simple method to evaluate blood glucose status in diabetic patients.

Keywords

• diabetes mellitus
• oral glucose tolerance test
• blood glucose monitoring
• mathematical model