Introduction and evaluation of a computerised insulin protocol

Objective: To lower glucose levels in all patients in the intensive care unit (ICU) to the target range of 4.5-7.5 mmol/l using a nurse-driven computerised insulin protocol in combination with bedside glucose measurement. Design: Cohort study. Setting: Mixed adult ICU. Patients and participants: All 182 patients admitted to the ICU during a 3-month period were studied, except for 3 patients admitted for diabetic keto-acidosis. Interventions:Five steps were taken to improve glucose regulation: (1) Nurses were authorised to adjust insulin dosage using a protocol. (2) Glucose was measured more often. (3) Glucose was measured at the bedside. (4) Consecutive protocols aimed for successively lower glucose levels; the final protocol had a target range of 4.5-7.5 mmol/l. (5) The protocol was computerised. Measurements and results: Mean glucose decreased from 9.23 mmol/l without protocol to 7.68 mmol/l with the final protocol. This final protocol with the target of 4.5-7.5 mmol/l was evaluated more extensively. Glucose levels were measured a total of 1854 times in 179 ICU admissions during 552 ICU treatment days. The median glucose level was 7.0 mmol/l, and 53.1% of glucose measurements were within the target range of 4.5-7.5 mmol/l. One episode of hypoglycaemia (glucose en 2.2 mmol/l) occurred, representing 0.5% of patients or 0.05% of glucose measurements. Conclusions: The combined strategy of successively more ambitious nurse-driven (computerised) insulin protocols and bedside glucose measurement resulted in acceptably low glucose levels with very few episodes of hypoglycaemia.