Outcome of neonatal hypoglycaemia

British Medical Journal, Jan 16, 1999 by Marvin Cornblath, Robert Schwartz, Alan Lucas, Ruth Morley

Complete data are needed

EDITOR--Ten years ago Lucas et al reported a retrospective multicentre study determining whether asymptomatic neonatal hypoglycaemia results in neurological damage.[1] They found that moderate hypoglycaemia ([is less than] 2.6 mmol/l) may have serious neurodevelopmental consequences if present on five or more different days during the first two months of life.

Subsequently, in October 1989, a Ciba Foundation discussion meeting was held in order to clarify the data and conclusions presented by Lucas et al.[1-2] Sinclair and Steere, using specific criteria, recognised that all the published follow up studies were too flawed and inadequate to provide a definite conclusion. Lucas et al had assembled a large inception cohort in their study, with objective outcome criteria and with assessments done by people who were blind to category of exposure. Follow up was almost complete. However, they had not specified a criterion for hypoglycaemia, and sampling bias was not avoided.[1 3] Lucas et al's study therefore did not provide strong evidence that neonatal hypoglycaemia impairs later development.

Despite this caveat the study by Lucas et al has received widespread attention and acceptance in both North America and Europe. Implications for the feeding and care of all neonates have been important; attempts have been made to maintain these comparatively high blood glucose concentrations as acceptable.[4] In the United States these criteria have resulted in a fourfold increase in cases of medical litigation, and of awards (MC). Similar increases have occurred in England (A F Williams, personal communication).

Lucas et al have indicated to us that neurodevelopmental follow up of the 1988 cohort would clarify the importance of hypoglycaemia as stated in their preliminary report.[1] The long term outcomes of all of their clinical trials of early nutrition on later health and development have been published.[5] The original BMJ article[1] is not mentioned, nor are any of the later follow up studies that compare glucose concentrations in the blood early in life and neurodevelopmental outcome. These data are available. We deserve to have the benefit of these observations, flawed as they may be.

Marvin Cornblath Lecturer in paediatrics Neonatology Division, Johns Hopkins Hospital, Baltimore, MD 21987, USA

Robert Schwartz Professor of pediatrics Brown University School of Medicine, Division of Pediatric Endocrinology and Metabolism, Providence, RI 02903, USA

[1] Lucas A, Morley R, Cole TJ. Adverse neurodevelopmental outcome of moderate neonatal hypoglycaemia. BMJ 1988;297:1304-8.

[2] Cornblath M, Schwartz R, Aynsley-Green A, Lloyd JK, Hypoglycaemia in infancy: the need for a rational definition. Pediatrics 1990;85:834-7.

[3] Department of Clinical Epidemiology and Biostatistics, McMaster University Health Sciences Centre. How to read clinical journals. III: To learn the clinical course and prognosis of disease; IV: To determine etiology and causation. CMA Journal 1981; 124:869-72;985-90.

[4] National Childbirth Trust. Hypoglycaemia of the breastfed newborn. Mod Midwife 1997 Oct 7:31-3.

[5] Lucas A. Long term outcome trials of early nutrition on later health and development. In: Perman JA, Rey J, eds. Clinical trials in infant nutrition. Philadelphia, PA: Lippincott-Raven, 1998:181-201.

Authors' reply

EDITOR--Our study design was not "flawed"--we had intended to explore the relation between glucose concentrations in blood in neonates and neurodevelopment at 18 months, based on over 6800 blood samplings in 661 preterm infants. We did not use a definition of hypoglycaemia as our purpose was to explore what concentrations might relate to adverse outcomes. The cut off we derived (2.6 mmol/l) was the same as the glucose concentration below which acute electrophysiological changes in the brain were observed elsewhere.[1] It is at least biologically plausible that at such a sensitive stage in brain development, hypoglycaemia occurring on three to five days (a substantial time) could influence outcome.

We do, however, entirely accept the study's principal limitation--the difficulty of proving causation when an observational approach is used, as emphasised clearly in our paper. The BMJ publishes many articles that show worrying or provoking associations--for example, observations on small size in early life and ischaemic heart disease in later life.[2] Such observations generate hypotheses or legitimate clinical concerns that should stimulate future studies.

The article Cornblath and Schwartz refer to, in Clinical Trials in Infant Nutrition, does not mention our hypoglycaemia study since it was not a trial. We have, however, now related neonatal hypoglycaemia, as defined and modelled previously, to overall intelligence quotient,[3] reading and arithmetic,[4] mid motor performance[5] at 7.5-8 years. We have found evidence of persisting associations between neonatal hypoglycaemia and lower test scores, clinically significant in two of the four outcomes (arithmetic and motor tests), with approximately 0.5 SD reduction in scores (adjusted for respiratory support, birth weight and gestation) after the neonatal concentration of blood glucose was [is less than] 2.6 mmol/l on [is greater than] 3 days (P [is less than] 0.005). Although these data share the earlier same limitation, it seems that a plausible concern remains.