Improving the quality of near-patient blood glucose measurement

The quality of near-patient blood glucose measurement was audited in our hospitals in 1990, when a diversity of glucose meters were in use, by sending three samples of unknown (to the meter user) concentration to each user and collecting and analysing the results produced. The overall performance was unsatisfactory with a mean coefficient of variation (CV) of 23.5%. A scheme involving training, quality control and external quality assurance was introduced in 1993 based on the Bayer Glucometer II meter. This meter was used exclusively throughout our hospitals. Data from the quality assurance scheme showed that the overall CV fell initially to 14-16% and then settled at about 10-12% for the following 2 years. Unacceptable results (those more than two standard deviations from the mean) were 8-12% of the total. A new meter was introduced in 1995 (the Bayer Glucometer 4) which had the advantages of 'no-wipe' and automatic timing technology and in the subsequent year overall CV fell to 5-6% and has remained at this level. The frequency of unacceptable results fell to 5-7%. The improved precision figures encouraged us to change criteria for acceptability to mean +/- 15%. Using these criteria the level of unacceptable results is now 1-2%. This study shows that introducing training, quality control procedures, a quality assurance scheme and improved meter technology all backed by laboratory expertise can produce significant improvement in the quality of near patient blood glucose measurement.