Background Patients with diabetes are recommended to self-monitor their blood glucose levels also at home. Accuracy of a hand-held glucometer and a Continuous Glucose Monitoring (CGM) device were comparatively evaluated. Methods Venous blood samples (for reference laboratory determinations; n = 428) were collected from 18 type 1 patients (35–65 years old), immediately followed by capillary measurement (Bayer ContourLink meter) and CGM readings (Medtronic Paradigm). Results Laboratory values did not differ statistically from ContourLink and CGM readings, mean difference (± SD) being − 0.05 ± 1.06 mmol/L and 0.10 ± 1.84 mmol/L glucose, respectively. A bias ((value − reference)/reference × 100) ≥ 15% was observed in 27.7% and 54.9% of cases, respectively. Notably, below 3.9 mmol/L glucose (hypoglycemic threshold), an absolute error > 0.8 mmol/L was found in 78.9% and 94.1% of cases. The absolute errors of the CGM device were inversely related to the rate of glucose change (r = 0.598, p Conclusions A very large error was observed at the extreme glycemic values, which may lead to erroneous therapy. Consequently, performance of future portable glucometers should be focused in particular under hypo- and hyper-glycemia. Moreover, integrated CGM devices should not disregard the effect of the rate of blood glucose change on the sensor readings.
