Abstract

AIMS: To compare two cotside methods of blood glucose measurement (HemoCue and Reflolux II) against a standard laboratory method for the detection of neonatal hypoglycaemia in a developing country maternity hospital where hypoglycaemia is common. METHODS: 94 newborn infants and 75 of their mothers had blood glucose assessed on the same venous sample using three different methods in the Special Care Baby Unit and postnatal wards, Prasuti Griha Maternity Hospital, Kathmandu, Nepal: HemoCue and Reflolux II at the cotside; Roche Ultimate glucose oxidase method (GOM) in the laboratory. RESULTS: The mean (SD) values for blood glucose in newborn infants were GOM 2.5 (1.1) mmol/l; Reflolux II 2.1 (0.9); and HemoCue 4.2 (1.2). For mothers the values were GOM 5.3 (1.2) mmol/l; Reflolux II 3.6 (1.2); and HemoCue 5.6 (1.0). Bland-Altman plots showed that Reflolux II consistently underreads GOM blood glucose in neonates by 0.5 mmol/l (SD 0.7) and that HemoCue overreads glucose by 1.7 mmol/l (SD 0.8). For the detection of hypoglycaemia (< 2.0 mmol/l), Reflolux II achieved a sensitivity of 83%, a specificity of 62%, and a likelihood ratio of 2.2. HemoCue produced a sensitivity of 0% and a specificity of 100% using measured values. If 2.0 mmol were subtracted from all Hemocue values this rose to 81% and 68% and a likelihood ratio of 2.5. CONCLUSION: Although more accurate than Reflolux II for the measurement of blood glucose in mothers, HemoCue overreads glucose concentrations in neonates and is therefore potentially dangerous as a screening method for neonatal hypoglycaemia. Reflolux II is useful as a screening method for high risk infants (low birthweight, post-term) and could achieve a post-test probability of detecting hypoglycaemia in a high risk setting like Nepal of 50-60%.