An evaluation of three glucose meter systems and their performance in relation to criteria of acceptability for neonatal specimens

Background: Prior to making a selection for our hospital, a pediatric institution, we deemed it necessary to evaluate, concurrently, three recently available glucose meter systems, claimed to be suitable for use with neonatal samples. Methods: Comparisons were with laboratory plasma analyses. Linearity and precision were also determined. Results: All meters gave linear responses. Precision determined using quality control material was acceptable. For an in-laboratory side-by-side evaluation, meter 1 showed a small bias but significant result scatter while meter 3 showed a negative bias; mean differences from reference (S.D.) were: -0.30 mmol/l (0.56), 0.06 mmol/l (0.39) and -0.49 mmol/l (0.35) for meters 1, 2 and 3, respectively. Clinical unit testing results gave mean differences from reference (S.D.) of: -0.19 mmol/l (0.56), 0.06 mmol/l (0.48) and -0.12 mmol/l (0.48) for meters 1, 2 and 3, respectively. Using 15% of reference as acceptability thresholds, 61%, 79% and 72% of results for meters 1, 2 and 3 respectively, were within limits. At 20%, the corresponding figures were 81%, 90% and 91%, respectively. All meters showed a sample-hematocrit effect with either negative (meters 1 and 3) or positive (meter 2) bias. Conclusions: Regardless of the performance criteria chosen, meter 1 had the worst performance while meter 2 was slightly better in overall than meter 3. Based on performance, general characteristics and user feedback, meter 2 was selected by us. In light of our results, we nonetheless suggest that performance of the meters tested is less than ideal, especially in the context of clinical utility in neonates. (C) 2002 Elsevier Science B.V. All rights reserved.