Frequency of Extreme Differences and Clinical Performance of Glucose Concentration Measurements Judged from 21 000 Duplicate Measurements

To the Editor: Duplicate measurements of patient samples are not common in routine clinical chemistry. The performance of measurements is continuously monitored by the internal quality control (IQC),1 which is not designed to find occasional dropouts. Therefore, the frequency of such errors is not usually known. In the present study, we attempted to estimate the error frequency at commonly suggested imprecision targets by evaluating the difference between 2 adjacent repeated measurements. The frequency is expected to be very low, and therefore the study of this problem required a large number of samples. We measured glucose concentration over a 6-month period in samples obtained from an unselected population of hospitalized patients and outpatients. We used 3 Dimension Vista 1500 analyzers connected to StreamLAB (Siemens Healthcare Diagnostics), in accordance with the manufacturer's instructions. To include measured values in our study, we required that glucose assay performance comply with the RiLiBAEK (Richtlinien der Bundesarztekammer) IQC limits (1) over the measuring interval of 1 mg/dL (0.06 mmol/L) to 500 mg/dL (27.75 mmol/L). We measured 21 653 samples twice, with each sample randomly allocated to one of the instruments by the laboratory automation system. The repeat measurement, which …