Abstract

We performed a retrospective audit of blood glucose control after introducing a new protocol for insulin infusion. The audit showed a sustained reduction in the median blood glucose, which decreased from 7.0 to 6.8 mmol x l(-1), primarily because of a reduction in the proportion of values above 6 mmol x l(-1). When we examined the proportion of readings with previously published 'acceptable' ranges we found that small, probably clinically insignificant, changes in the accepted range have a major effect on apparent compliance with glycaemic control. This is because a large number of results fall within a narrow range of values. As a result there is a nearly 2.5-fold difference in compliance for a change in the upper limit of the target range of just 2.2 mmol x l(-1). Different target ranges have been recommended for tight glycaemic control. When comparing compliance with tight glycaemic control between intensive care units, target ranges should be identical.