To evaluate the correlation between the plasma glucose-to-glycated haemoglobin ratio (GAR) and clinical outcome during acute illness.

This retrospective observational cohort study enrolled 661 patients who visited the emergency department of our hospital between 1 July 2008 and 30 September 2010 with plasma glucose concentrations>500mg/dL. Systolic blood pressure, heart rate, white blood cells, neutrophils, haematocrit, blood urea nitrogen, serum creatinine, liver function and plasma glucose concentration were recorded at the initial presentation to the emergency department. Data on glycated haemoglobin over the preceding 6 months were reviewed from our hospital database. The glucose-to-HbA_1c ratio (GAR) was calculated as the plasma glucose concentration divided by glycated haemoglobin.

The GAR of those who died was significantly higher than that of the survivors (81.0±25.9 vs 67.6±25.0; P<0.001). There was a trend towards a higher 90-day mortality rate in patients with higher GARs (log-rank test P<0.0001 for trend). On multivariate Cox regression analysis, the GAR was significantly related to 90-day mortality (hazard ratio [HR] for 1 standard deviation [SD] change: 1.41, 95% confidence interval [CI]: 1.22–1.63; P<0.001), but not to plasma glucose (HR: 0.89, 95% CI: 0.70–1.13; P=0.328). Rates of intensive care unit (ICU) admission and mechanical ventilator use were also higher in those with higher GARs.

GAR independently predicted 90-day mortality, ICU admission and use of mechanical ventilation. It was also a better predictor of patient outcomes than plasma glucose alone in patients with extremely high glucose levels.