OBJECTIVES

: To study the influencing factors of blood glucose index in patients with type 2 diabetes mellitus in real world and establish a predictive model of blood glucose index.

METHODS

: A cross-sectional survey was conducted, screening for patients with type 2 diabetes mellitus randomly. The patient's basic information, self-monitoring blood glucose (SMBG) status and treatment status were causal variables; glycosylated hemoglobin (HbA1c), fasting blood glucose (FBG), and Random blood glucose (RBG) were used as outcome variables. We studied the key factors that may influence the outcome variables, and used the multiple linear regression techniques to establish a patient HbA1c predictive model.

RESULTS

: Factors that may affect HbA1c in patients include diabetes duration (P=0.152, R=0.1546), previous measurement HbA1c values (P<0.0001) and previous HbA1c measurement interval (P=0.0154, R=0.1770), self-measurement frequency of FBG (P = 0.0073) and self-measurement value of FBG (P = 0.0238). After using the stepwise method for variable screening, the HbA1c predictive model for type 2 diabetes patients was: HbA1c=5.8-0.4*sex+1.0* previous measurement HbA1c -0.4* self-measurement frequency of FBG +0.4* self-measurement value of FBG, R=0.5793. The predicted residual is distributed within 2 standard deviations.

CONCLUSIONS

: Measurement of SMBG contributes to improve glycemic control in patients with type 2 diabetes mellitus in real world; the established HbA1c prediction model has certain predictive performance, which can provide reference for glycemic control in diabetic patients.