Glucose Meter Performance in a Tertiary Care Institution

Abstract

Aim This article aimed to study the agreement between blood glucose levels measured by glucose meters and in the clinical biochemistry laboratory. Materials and Methods One hundred ten patients admitted to the medicine department at SRTR Government Medical College, Ambajogai were selected randomly. Samples from these patients were analyzed for glucose concentration by the glucose meters and by a glucose oxidase–peroxidase method in the biochemistry laboratory simultaneously. The glucose meter was an international brand but its identity could not be revealed for contractual reasons. The results were analyzed by appropriate statistical methods to evaluate their agreement and by using Clarke error grid analysis and Parkes error grid analysis methods to evaluate the utility of the results in clinical settings. ISO 15197:2003 and 2013 guidelines were considered. Results The mean value of the glucose meter results was significantly different from the laboratory method (P < 0.05). Bland–Altman analysis showed disagreement between the 2 methods. Glucose meter results were always higher in comparison to the values from the laboratory method. Error grid analysis confirmed this finding and the test results maximally scattered in C, D, and E areas of the error grid which could result in unnecessary treatment of patients. Conclusions The use of some glucose meters in hospital wards and intensive care for tight glycemic control is not recommended due to differences from laboratory results. The glucose meters were not in agreement with the standard laboratory method for glucose measurements.