Evaluation of the accuracy, precision, and agreement of a glucometer compared to the standard laboratory test in diabetic and non-diabetic patients

Blood glucose monitoring is vital for diabetes management, with nurses playing a key role in glycemic monitoring and patient education. Despite their widespread use, the precision, accuracy and agreement of fingerprick glucometers with laboratory standards remain a significant concern. Discrepancies arise from multiple sources such as differences in sample type (whole blood vs. serum) and enzymatic technologies, potentially leading to clinical errors. Current international standards like ISO 15,197 have been insufficient in ensuring reliable device performance. Therefore, we need more rigorous benchmarking standards to assess their safety and reliability. This study aimed to evaluate the analytical and clinical accuracy, precision, and agreement of the glucosureSTAR point-of-care (POCT) glucometer against the standard laboratory method in diabetic and non-diabetic patients, following the ISO 15197:2013 standards. A cross-sectional study was conducted on 589 participants (136 diabetic and 453 non-diabetic). Paired blood glucose measurements were taken using the glucosureSTAR glucometer (glucose oxidase method) and a Roche Hitachi 917 laboratory analyzer. Analytical accuracy was assessed against ISO 15197:2013 criteria (± 15 mg/dL for values < 100 mg/dL, ± 15% for values ≥ 100 mg/dL). Clinical accuracy was evaluated using Parkes and Clarke Error Grid Analyses. Agreement was statistically analyzed using Bland-Altman plots, Passing-Bablok regression, and the Concordance Correlation Coefficient (CCC). Precision was tested via the coefficient of variation from 20 repeated measurements on three samples. Diagnostic performance was assessed using ROC curve analysis. The analytical accuracy of the glucometer marginally met the ISO 15197:2013 criteria (95.1% of results within ± 15 mg/dL or ± 15%). While clinical accuracy was acceptable (100% of points in Parkes error grid zones A/B), significant systematic error was observed. The device consistently underestimated blood glucose (mean bias: -3.3 mg/dL; 95% LoA: -38.7 to 32.1 mg/dL) with poor absolute agreement (CCC = 0.886). Precision failed ISO requirements at medium and high concentrations. ROC analysis revealed suboptimal diagnostic performance for detecting diabetes (AUC = 0.816; sensitivity = 62.18%, specificity = 93.26% at > 124 mg/dL cut-off). In conclusion, while the glucosureSTAR glucometer demonstrated acceptable clinical risk assessment with 100% of readings in clinically acceptable error grid zones, it showed significant limitations in analytical performance. The device exhibited systematic underestimation, wide variability, and failed to meet precision requirements at medium and high glucose concentrations. With 95.1% of the data points falling within the acceptable range by ISO15197:2013, and suboptimal diagnostic accuracy (sensitivity: 62.18%), this device should not be considered suitable for diagnostic purposes or therapy adjustment. It may be used with caution for trend monitoring in known diabetic patients where rapid results are prioritized over absolute accuracy but cannot replace standard laboratory methods when precise measurement is critical.