Aims/hypothesis: This study investigated the accuracy of blood glucose meters for self-monitoring and its influence on glycated hemoglobin (HbA1c) levels and the frequency of hypoglycemic coma.
Materials and methods: Self-measured and simultaneously obtained laboratory blood glucose values from 9,163 patients with type 1 diabetes <18 years of age in the German/Austrian Diabetes Prospective Documentation Initiative registry were analyzed by investigating their compliance with the International Organization for Standardization (ISO) criteria (versions 2003 and 2013) and by error grid analyses. Regression models elucidated effects on glucose control and hypoglycemia rates.
Results: Depending on the respective subgroup (defined by sex, age, duration of diabetes, mode of insulin therapy), 78.7-94.7% of the self-monitoring of blood glucose (SMBG) values met the old and 79.7-88.6% met the new ISO criteria. In Clarke and Parkes error grid analyses, the percentages of SMBG values in Zone A ranged between 92.8% and 94.6% (Clarke) and between 92.2% and 95.0% (Parkes). The patient group with SMBG devices measuring "far too low" (compared with the laboratory-obtained glucose levels) presented with a higher HbA1c level than those measuring "far too high," "too high," "identical/almost identical," or "too low" (based on quintiles of deviation). Performing "far too high" was associated with the highest rate of hypoglycemic coma in comparison with the other deviation quintiles.
Conclusions: This study showed that current SMBG devices fulfilled neither the previous nor the new ISO criteria. Large deviations of the SMBG values from the "true" glucose levels resulted in higher HbA1c levels and markedly increased rates of hypoglycemic events.