Subdural haemorrhages in infants

In this issue Jayawant et al report the results of a study of the incidence, causes, and outcome of subdural haemorrhages in infancy in a defined geographical area in England and Wales from 1993 to 1995 (p 1558).¹ This subject is important because, as this study confirms, most subdural haemorrhages are due to abuse. The subdural haemorrhage is just one element of the brain injury in infants who have suffered non-accidental head injury (caused either by shaking alone or by shaking and impact).²

Ascertainment seems to have been thorough and the results are likely to be generalisable to the rest of the United Kingdom. The results suggest that a large district general hospital can expect, on average, to see an infant with a subdural haemorrhage every year. Most of these infants needed intensive care and, as in other studies, the outcome was poor.³,⁴ Assuming that the results are generalisable, this paper contains important messages for paediatricians, general practitioners, social workers, and neurosurgeons.

Seven of the infants had previously been abused, and six had siblings who had been abused. These findings raise the question whether more could have been done to protect the infants from the assaults that caused their subdural haemorrhages. The authors also state that six infants had histories of “repeated admissions to hospital with symptoms of drowsiness and lethargy before a subdural haemorrhage was diagnosed.” For most of these babies a subdural haemorrhage is likely to have been the cause of their earlier symptoms. However, these symptoms are non-specific, and if the baby improves while under observation it is understandable that the diagnosis is sometimes missed. A recognised pitfall is that blood staining of cerebrospinal fluid can be wrongly attributed to trauma from the lumbar puncture needle.⁵

Perhaps the most important message from the paper, however, is that when subdural haemorrhage was identified abuse was often not diagnosed when it should have been. One of the 33 cases of subdural haemorrhage was caused by a road accident; 21 of the other 32 cases had been attributed to abuse. The authors reviewed the evidence in the remaining 11 cases and considered that in six the evidence was “highly suggestive of abuse.” This evidence included coexisting fractures and salt poisoning and leaves little room for doubt.

The authors classify the five remaining cases as showing “no obvious evidence of child abuse.” Yet some or all these cases will arouse suspicion in the minds of many readers. There was no history of trauma in four of the five cases. The fifth infant (who also had a retinal haemorrhage) had allegedly “tipped from a bouncy chair.” In the absence of an underlying cause (such as a coagulopathy) the presence of a subdural haemorrhage without a history of substantial trauma means that abuse is likely and should lead to thorough investigation.²,⁶ The extent to which minor trauma can cause subdural haemorrhages remains controversial.⁷

Some infants were not investigated adequately. Two of the four infants with “no obvious evidence of child abuse” did not undergo a skeletal survey. Fractures visible on skeletal survey have been reported in 32-70% of infants with subdural haemorrhages due to abuse.⁸,⁹ Similarly, two of the four did not undergo ophthalmoscopy. Again, retinal haemorrhages have been reported in 38-89% of infants with subdural haemorrhages due to abuse.³,¹⁰,¹¹

Only 22 of the infants had undergone all the investigations that the authors recommend (multidisciplinary social assessment, ophthalmoscopy, a skeletal survey (which may need to be repeated), a coagulation screen, and computed tomography or magnetic resonance imaging). Furthermore, ophthalmoscopy should be performed by an ophthalmologist; apart from the greater expertise of ophthalmologists, some retinal haemorrhages may be missed if indirect ophthalmoscopy is not carried out.¹² Only 14 of the infants in this study were examined by an ophthalmologist.

In summary, almost all the subdural haemorrhages in this study were either definitely or probably due to abuse. The fact that seven infants had been previously abused shows that measures taken to protect them after their earlier abuse had been inadequate. An earlier opportunity to make the diagnosis had probably been missed in at least some of the six infants who had previously been admitted with drowsiness and lethargy. Three infants were apparently considered by those caring for them not to have been abused despite not having been adequately investigated. This paper thus provides evidence that British paediatricians are sometimes not diagnosing child abuse even when investigation shows that the diagnosis seems inescapable. These failures are important. If we do not recognise child abuse no action will be taken to protect the child and the child’s siblings from further assaults.

Papers p 1558