Background Glucometers have become a fundamental tool in measuring and monitoring glucose level, both in healthcare institutions and home care. The accuracy of glucometers affects the quality of management of diabetic patients and is associated with the occurrence of over-treating or under-treating accidents due to inaccurate readings. This study assessed the accuracy of five commercially available glucometers by reference to laboratory venous plasma glucose (PG) measurement.

Methods A cross-sectional study was conducted among patients with diabetes attending King Abdulaziz Medical City laboratory. All participants underwent venipuncture regarding laboratory PG, simultaneously with capillary blood sampling, on which capillary glucose (CG) was measured using the glucometers AccuCheck®, OneTouch®, Freestyle Optium Neo®, Contour Next®, and Contour Next One® in random order. All glucometers were adequately calibrated and verified according to American Diabetes Association guidelines before use. Bias was calculated for each glucometer as the difference between CG and PG (ΔCG-PG). One-sample t-test was used to analyze mean ΔCG-PG by reference to zero for each of the glucometers. Bland–Altman analysis was undertaken by plotting ΔCG-PG against PG. Proportional bias was investigated by analyzing the relationship between ΔCG-PG and PG using linear regression.

Results A total of 203 patients were included, with mean PG 155.22 (SD 64.88) mg/dL. The coefficient of variation of the meters ranged from 37.79% to 41.80%. Mean CGs ranged from 153.01 (SD 57.82) to 163.00 (SD 64.52) depending on the glucometer. Three meters showed negative bias. Mean difference was 2.20 for AccuCheck, -2.26 for One Touch, 0.90 for Freestyle, -2.08 for Contour Next, and -7.78 for Contour Next One. Bias percentage ranged from -5.01 to 1.42. Bland–Altman plots showed proportional bias (an increase in the magnitude of the error as the test result increases). Proportional bias was supported by the significant linear regression analysis for all glucometers.

Conclusion Of all glucometers, Freestyle Optium Neo showed the minimal mean bias, while Contour Next One showed the highest proportional bias. However, all of the glucometers were within 5% difference. High blood glucose readings above 200 mg/dL should be confirmed by venous measurement.