Evaluation of the Reliability of Bedside Glucose Testing

Abstract

A review of six months of patients' records showed that glucose values obtained on the wards by nurses with various chemistry reagent strips differed significantly from those obtained by the laboratory. The differences could be as much as 12mmol/L(216mg/dL). The bias found seemed to apply to all brands of strips used; 58.7% of glucose results obtained by strip with manual reading had a bias exceeding 20% of those results obtained by the laboratory, and 15.7% of glucose results had a difference of greater than 10mmol/L(180mg/dL). When meters were used to read the strip color, the percentage of discrepancies was 47.0%. Similar results were obtained in the concurrent study, although the discrepancies were less when meters were used for reading the strip color. Our studies suggest that user-to-user variability, lack of quality control, lack of stringent staff training, and education criteria may all contribute to the variability and poor performance found. These points must be addressed if bedside glucose monitoring is to be of practical value, thus ensuring reliability in the management of diabetic patients.