Editor—A 20 day old baby boy was referred to our hospital with a history of right sided facial palsy, nystagmus, and seizures. He had been born by normal vaginal delivery, at 36 weeks of gestation, with good Apgar score (8 and 9 at five minutes, respectively). He had jaundice, for which he required phototherapy for four days at the referring hospital, but was otherwise normal.

Ten days after discharge he presented again with a history of poor feeding and lethargy for one day. On examination, he looked pale with a full and tense anterior fontanelle. He was febrile and tachypnoeic. A presumptive diagnosis of sepsis/meningitis was made and he was given cefotaxime and ampicillin. Spinal tap was deferred as the fontanelle was tense. He developed intractable seizure, requiring diazepam, phenobarbitone, and phenytoin.

In view of his deteriorating condition, he was transferred to the university hospital. On arrival, he was noted to be febrile and lethargic. After stabilisation an emergency computed tomography scan of the brain was performed, which revealed cerebral oedema and a large, left sided, intracerebral bleed, with extension into the subarachnoid spaces and ventricle. Investigations on admission revealed severely deranged electrolytes. His serum glucose was 51.2 mmol/l, sodium 170 mmol/l, potassium 7.5 mmol/l, urea 39 mmo/l and creatinine 246 μmol/l. The white cell count was 17 × 10⁹/l, with a haemoglobin of 156 g/l. The platelet count was obtained from the referring hospital because our specimen had a small clot (125× 10⁹/l). The coagulation profile was normal (PT 15 seconds, APTT 45 seconds, D-Dimer > 0.5 < 1 μg/ml, fibrinogen 3.1g/l). The infant was treated symptomatically with supportive care, but died on the second day of admission.

The common causes for bleeding were ruled out. There was no history of birth asphyxia or trauma. The platelet counts and coagulation profile were normal. Although the chance of arterio-venous malformation as a cause of spontaneous bleed cannot be ruled out completely, the computed tomography scan gave no indication of this. The diagnosis of neonatal diabetes was supported by concomitant hyperglycaemia (51.2 mmol/l), glycosuria, metabolic acidosis (pH 7.29, baseline excess –16.3), dehydration and weight loss (weight at admission was 2050 g compared with birthweight of 2580 g).

Neonatal diabetes is rare. The incidence rates reported from the UK and Germany are 1 in 400 000 and 1 in 500 000, respectively.1As far as we are aware, no reports have been published on the association between neonatal diabetes and intracerebral haemorrhage. However, previous reports have associated hyperglycaemia with haemorrhagic transformation of the cerebral infarct.2 3In a recent study, Scott et al showed an increased incidence of intracerebral haemorrhage with admission hyperglycaemia in adults.4 In an animal model study, de Courten-Myers et al reported 25-fold more extensive haemorrhage in cats with hyperglycaemia compared with normoglycaemic cats.2 Similarly, Brodericket al associated hyperglycaemia with cerebral bleed in two human adults.3 The exact mechanism by which hyperglycaemia induces cerebral bleeding is not clear. It enhances lactic acidosis in brain tissue and the combination of hyperglycaemia and acidosis enhances endothelial damage with subsequent extravasation of red blood cells through the leaky vessels.3 There have been studies on the effect of glucose and acid base changes on brain ischaemia.5 6

A causal link between hyperglycaemia and intracerebral haemorrhage in neonates may be difficult to prove, but we look forward to other prospective studies on the risks and mechanisms of brain haemorrhage in neonates with severe hyperglycaemia.