Editor—We enjoyed the article on practical management of hyperinsulinism by Aynsley-Green et al.1 It re-emphasises the importance of accurate measurement of blood glucose and insists on an accurate laboratory method and not a bedside screening test for diagnosing hypoglycaemia. However, in certain situations, the use of a bedside test is unavoidable—for example, if there will be a long delay before a laboratory result can be obtained, in general practice, home visits, or during transport. In many hospitals, bedside tests are used to identify high risk babies with suspiciously low values who need accurate laboratory measurements of blood glucose.

Most rapid bedside blood glucose measuring devices have been validated in the range above 2.6 mmol/l, using adult blood. We conducted a study to test the accuracy of two commonly used bedside methods of glucose estimation in the clinically important range of 0.5–4 mmol/l, using neonatal cord blood with a packed cell volume over 0.5.

Cord blood samples were allowed to stand for various periods of time to allow the glucose levels to fall in the range of 0.5–4 mmol/l. A total of 103 samples were analysed simultaneously in duplicate by (a) the Cobas hexokinase method in the laboratory, (b) Hemacue, and (c) Precision QID.

With the laboratory hexokinase method as the standard, the sensitivity and specificity of Precision QID for detecting hypoglycaemia (blood glucose less than 2.6 mmol/l) were 86% and 89% respectively; for Hemacue they were 83% and 100%. On average, blood glucose measured by Precision QID was 0.21 (0.32) mmol/l (mean (SD)) higher than when measured by the hexokinase method on paired samples in this low range. Blood glucose measured by Hemacue was on average 0.34 (0.23) mmol/l higher than when measured by the hexokinase method.

Both these bedside methods tend to overestimate the blood glucose slightly in relation to the standard laboratory method. Our study suggests that in situations in which bedside glucose is the only available estimate of blood glucose, a value of over 3.0 mmol/l would be needed for hypoglycaemia to be confidently excluded.