Figure 1: Clarke Error-Grid Analysis of data from Guardian RT during the whole study period. Zone A (accurate), within 20% of reference values, zone B (benign erroneous), outside of 20%, but not leading to inappropriate treatment, zones A and B, clinically acceptable accuracy. Zone C (unnecessary correction), leading to overcorrection of acceptable glucose levels, zone D, potentially dangerous failure to detect hypo- or hyperglycemia, zone E (erroneous treatment), erroneous treatment of hypo- or hyperglycemia (for hypoglycemia in case of hyperglycemia and vice versa), zones C+D+E—clinically unacceptable.