Diabetes treatment is intended to maintain near-normal glycemic levels. Self-monitoring of blood glucose (SMBG) allows patients to track their BG levels compared with glycemic targets and is associated with improved health outcomes. Because of the importance of SMBG, it is essential that results are accurate to prevent errors in nutritional intake and drug dosing. This study presents a new methodology to evaluate the accuracy of BG monitoring systems (BGMSs).Methods:
Sensitivity analyses were performed using real and simulated BGMS data to compute probabilities that, for any BG value, the BGMS result would be within prescribed error bounds and confidence limits compared with laboratory reference values. Multiple BG value ranges were used.Results:
Probability curves were created using data from 3 simulated BGMSs and anonymized data from 3 real-world BGMSs. Accuracy probability curves from capillary fingertip blood samples (actual clinical data) showed that all 3 real-world BGMSs met EN ISO 15197:2015 accuracy criteria, since 99.63%, 99.63%, and 99.81% of results from the 3 BGMSs were within ±15 mg/dL or ±15% of reference for BG <100 mg/dL and ≥100 mg/dL, respectively. However, there was identifiable variability between BGMSs if BG was <70 mg/dL; one BGMS showed further reductions in accuracy if BG was <50 mg/dL.Conclusions:
Probability curves highlight the importance of BGMS accuracy to help achieve optimal glycemic control while avoiding hypoglycemia or hyperglycemia. This may be especially significant in very low BG ranges where small errors in BGMS measurements can have substantial impacts on patient-related outcomes, including hypoglycemia risk.