‘The evidence base for retinal haemorrhages in shaken baby syndrome’


* john.elston@orh.nhs.uk

SIR–Dr Squier’s review1 analyses the evidence base for subdural haemorrhages which we agree is an important finding in ‘shaken baby syndrome’ (non-accidental head injury [NAHI]). However, it provides a skewed perspective on the subject by focusing on subdural haemorrhages in isolation. It ignores another important finding – the presence of retinal haemorrhages which, along with the history, clinical examination, radiology, and pathology findings help arrive at the diagnosis.2

The severity of retinal haemorrhages in NAHI closely parallels the degree of brain injury, and the neurological outcome.3,4 There is ample evidence that the same mechanism is responsible for the eye and brain findings. Of the theories that have been proposed for the mechanism of retinal haemorrhages in NAHI, rotational and shearing traction between the vitreous gel and the retina is the likeliest mechanism. The infant eye has strong adhesions between the vitreous gel and the internal limiting membrane of the retina, especially around the optic disc and in the retinal periphery. The differential movement at the interface between the vitreous and retina due to acceleration-deceleration forces4–6 is believed to result in shearing of retinal vessels, which may be dilated as a result of increased intracranial venous pressures secondary to raised intrathoracic pressure.7 Moreover, macular retinoschisis and retinal folds, characteristic but not invariable findings in NAHI,5 result from mechanical separation of the retinal layers. This cannot occur without repeated traction between the retina and the vitreous gel, lending further support to this theory.

Dr Squier has suggested that the subdural haemorrhages in NAHI may occur from vessels damaged by hypoxia and haemodynamic disturbances, rather than from traumatic rupture of bridging vessels. The author refers to Larroche et al 8 who reported the association of findings of venous congestion and hypoxic–ischaemic injury with the presence of subdural haemorrhages in a study of 700 autopsies of perinatal deaths, not cases of head injury. The author has interpreted this association to imply causation. This theory, previously proposed by Geddes9,10 (referred to as Geddes III, an autopsy study of deaths from various causes including but not restricted to head injury) and later retracted in the UK Court of Appeal,11 provides no explanation for the characteristic retinal haemorrhages of NAHI.

The abnormal haemodynamic forces referred to by the author include venous hypertension, systemic arterial hypertension, and episodic surges in blood pressure. The retinal haemorrhages in NAHI are large, extend throughout the retinal layers, and into the periphery. In contrast, the retinal haemorrhages in hypoxia and haemodynamic instability, when present, are flame shaped, superficial, and do not extend to the peripheral retina. This difference may be explained by the direct transmission of the abnormal haemodynamic forces to the posterior retinal vesels, and the greater density of vessels at the posterior pole. For example, the combination of severe hypoxia, and a sustained, severe rise in venous pressure is seen after strangulation or hanging. Reports of fundus examinations in survivors or autopsy findings do not show haemorrhages of the same severity as NAHI.12 Macular retinoschisis and retinal folds are never seen in cases of hanging or strangulation. Similarly, conditions like phaeochromocytoma that can produce a sudden rise in systemic arterial pressure are not associated with the retinal findings seen in NAHI.13 Also, vigorous cardio-pulmonary resuscitation, which is necessarily associated with hypoxia and raised venous pressure, is almost never associated with retinal haemorrhages.14 Haemodynamic changes may thus contribute,7 but cannot be the sole mechanism for the retinal changes seen in NAHI.

There has to be another mechanism to account for the ophthalmological findings seen in these cases. Anterior tracking of sub-arachnoid haemorrhage was proposed as a possible mechanism,15 but this theory has been disproved by clinico-pathological studies.16,17 The most likely mechanism that would explain all these findings is the tractional forces generated by repeated abusive handling.

The author quotes Ommaya’s study of adult primates,18 and the conclusion that manual shaking of a child cannot generate the same magnitude of forces as single high impact trauma as in a motor vehicle accident. There is good evidence that high speed single impact head trauma of sufficient severity to produce serious brain injury or even death, is only rarely associated with retinal haemorrhages. The ALSPAC study confirmed that falls from beds and settees in young infants did not result in skull fractures, and serious injury was the result of complex accidents.19

From consideration of the likely mechanism of retinal haemorrhages in NAHI and the rare description of the trauma by perpetrators,20 it is evident that sufficient shearing forces can be generated by abusive handling.21

The author suggests that in contrast to diffuse axonal injury which produces immediate brain swelling, hypoxic–ischaemic damage produces slower onset of brain swelling, with resultant damage to the intracranial vessels and subdural bleeding. This slower onset of brain swelling has been proposed as an explanation for a lucid interval in such cases. However, the characteristic history given by the carers admitting to NAHI is that the infant collapsed immediately.20,22,23 The lucid interval suggested by the author is not seen in clinical practice, undermining the validity of her hypothesis.

Furthermore, the author suggests that awake apnoea may be caused by aspiration of feed/vomit, or even nappy changing. She proposes that the apnoea may produce hypoxia and increased blood flow to the brain, which in turn may result in dural haemorrhages. However, we know that severe hypoxia from acute life-threatening events in infants is not associated with retinal haemorrhages.14 This theory therefore cannot explain the eye findings in NAHI.24

In conclusion, the hypotheses advanced in this review for the causation of subdural haemorrhages cannot explain the retinal haemorrhages in NAHI. This could have serious implications, both scientific and legal on the mechanism of non-accidental shaking injuries.