Univariate and multivariate analysis of factors affecting tight glycaemic control

Introduction

Implementation of an intensive insulin protocol (IIP) aiming for tight glycaemic control (TGC) (blood glucose [BG] 4.4–6.1 mmol/l) resulted in a significant mortality reduction from 8% with conventional treatment to 4.6% in the IIP group [1].

Objective

To determine the key factors that influence the degree of control achieved by IIP.

Method

We implemented an IIP in mechanically ventilated surgical and medical patients, aiming for a target BG of 4.4–6.1 mmol/l. Ethics approval was gained to conduct a prospective, observational study in 50 patients who were treated with the IIP. All measured BG results were analysed to calculate the time spent in predefined BG bands. Univariate and multivariate analyses were conducted on key baseline characteristics and the percentage time in the target band, using SPSS version 13 software.

Results

In the univariate analysis, significant results were seen for BMI (R² = 13.1% P = 0.01), previous diabetes (R² = 10.2% P = 0.02) and gender (R² = 7.1% P = 0.06). The APACHE II score, age, drugs affecting glycaemia and percentage of correctly timed assays did not explain the variability in the percentage time in the target TGC range.

The multivariate analysis identified BMI as a significant factor (Table 1). Taken together, the factors in Table 1 explain 22.4% of the variability.

Table 1 abstract P244

Conclusion

BMI was inversely related to control of glycaemia. Diabetics had worse glycaemic control than nondiabetics. Women had better TGC than men. The BMI, previous diabetes, gender and interacting drugs explained 22.4% of the variability in achieving TGC targets.