SIR–I would like to thank Drs Parulekar and Elston for their comments and for the opportunity to continue the discussion on this very important topic.

They suggest my perspective is skewed by addressing only subdural haemorrhages; this was intended as subdural haemorrhage is within my daily experience, retinal haemorrhage (RH) is not. Due to the importance of RH in infant head injury, I have asked Dr Pat Lantz who has the appropriate experience, to join me in responding to their specific comments.

We agree that the same mechanisms are likely to lead to intracranial and retinal bleeding; the eye being a direct extension of the brain and subject to the same fluctuations in intracranial pressure. Parulekar and Elston theorize that rotational shearing and shearing traction between the vitreous gel and retina is the likeliest mechanism of retinal haemorrhages in non-accidental head injury (NAHI), although they do not support this contention with any objective scientific evidence.

Their theory is brought into question by observation of similar or identical haemorrhages in situations where no shearing is implicated.1–4 Indeed the anatomy and weight of the eye and its surrounding protective fat suggest that is extremely unlikely that shaking could ever generate sufficient force to tear the vitreous from the retina; this would be in defiance of the basic laws of physics.5

Detailed pathological studies of the eyes of infants considered to have been abused have concluded that ‘retinal haemorrhage in shaking abuse is caused by venous stasis and leakage from retinal vessels which, if prolonged, can lead to circumferential macular and peripheral retinal folds with blood- and protein-filled schisis cavities.’ The authors’ conclusion that macular retinoschisis cannot occur without repeated traction between the retina and the vitreous gel suggests that they are not aware of the recent literature.6–9

With respect to neonatal subdural bleeding I refer to the older classical neuropathological literature where well-documented and numerous cases fail to show that torn bridging veins are the cause. Nowhere did I suggest that hypoxic-ischaemic injury was causative; these studies were cited to indicate an alternative source of subdural bleeding in the infant.

There is substantial evidence that haemodynamic disturbances produce subdural bleeding in infants in the absence of trauma as first noted by Cushing in 1905,10–13 and it would seem reasonable, as noted above, to propose that the same mechanism operates in neonatal RH.

The authors compare the pathophysiology in adult hanging and strangulation to infants with the triad. The clinical circumstances and the literature they cite do not take into consideration the specific anatomical and physiological characteristics of the infant age group in question. Indeed, the statement that ‘macular retinoschisis and retinal folds are never seen in cases of hanging or strangulation’ appears to depend on the autopsies of two patients over the age of 70 years who sustained traumatic brain injuries and not hanging or strangulation.14 It is not only hypoxia or ischaemia or haemodynamic disturbance but the concurrence of these effects in very young infants which is relevant in these cases.

Parulekar and Elston state there has to be another mechanism and conclude that it is something they call ‘abusive handling’. This coy terminology is unhelpful and obfuscating. Its use appears to have more to do with establishing a perpetrator than with understanding the causes of infant intracranial bleeding.

They note that single impact head trauma is rarely associated with retinal haemorrhages, but do not explain the basis for this belief. There is ample evidence that they are incorrect.15–18 Further, the study they do quote is based on a postal questionnaire; there was no documented brain imaging or retinal examination in this study group so it can add nothing to our understanding of the incidence of intracranial or retinal bleeding.

They cite two papers to support their notion that ‘abusive handling’ can generate sufficient shearing forces to cause RH. One is a study of perpetrator confessions of shaking, invalidated by failure to take into account the circumstances of those confessions.19,20 The unreliability of confessions is confirmed by the observation that, even when shaking is admitted, it is usual to find evidence of impact, which generates far greater forces.21 The second, a commentary on the 2005 UK Court of Appeal hearing, is not only irrelevant, but contains no data and is materially incorrect.22,23

Parulekar and Elston suggest that diffuse axonal injury leads to brain swelling. This indicates that they have either misread or misunderstood the article. Diffuse axonal injury may occur in association with brain swelling, but I am unaware of any causative link. That lucid intervals may occur following infant head injury is clearly evident from the literature, where both case series and individual case reports indicate that not all infants collapse immediately after injury.24,25

Detailed and systematic observations of the eyes at autopsy have shown that apnoea is indeed associated with RH (Lantz PE, unpublished data).

The mechanisms of both retinal and subdural haemorrhages in the infant remain incompletely explained. This may have serious implications for the law, more serious still for those affected families. We hope that careful scientific enquiry and open debate will aid in the quest for understanding.


  1. Top of page
  2. References