In 2009, the NICE-SUGAR study became a reference supporting the use of higher glycaemic target bands for glycaemic control. The important increased risk of hypoglycaemia and mortality associated with lower target band in this study contradicted previous studies showing lower target bands improved outcomes. In this analysis, virtual trials of the NICE-SUGAR protocol and the patient-specific model-based STAR protocol are compared to reported clinical results to evaluate the safety and efficacy of the NICE-SUGAR protocol design.

Simulation results show STAR has higher safety and performance than NICE-SUGAR, with higher time in band, lower glycaemic variability, and lower incidence of both hyper- and hypo- glycaemia, which are all associated with improved outcomes. Compared to clinical results, the important difference in workload (9.4 vs 25.0 measurements per day) and insulin administration (50.2±38.1 vs. 154.0±209.2 U/d) shown in the simulations suggest poor clinical compliance to protocol in the NICE-SUGAR study. Thus, the increased clinical incidence of hypoglycaemia in the NICE-SUGAR study may have resulted from low compliance to protocol, and the interpretation of the results could have been biased by a non-compliant glycaemic control protocol design.

In conclusion, NICE-SUGAR protocol design was not clinically feasible, shown in the low compliance, likely resulting in low safety, efficacy, and highly variable glycaemic outcomes. Hence, the use of intensive insulin therapy for glycaemic control targeting lower glycaemic bands has been wrongly blamed for increased hypoglycaemia and mortality. Glycaemic control must be safe and effective for all patient, before any further study can assess potential beneficial clinical outcomes.