The pathogenesis of retinal and subdural haemorrhage in non-accidental head injury in infancy: Assessment using Bradford Hill criteria

Fatal non-accidental head injury (NAHI) in infancy is characterised by severe bilateral retinal haemorrhage (sbRH), thin film diffuse bilateral subdural haemorrhage (dbSDH) and severe fatal hypoxic ischaemic encephalopathy (sfHIE). The Geddes hypothesis is that sfHIE is the direct cause of sbRH and dbSDH. The conventional hypothesis (post-Geddes) is that trauma precipitates hypoxia to cause sfHIE and independently tears veins to cause sbRH and dbSDH. These hypotheses are assessed using Bradford Hill criteria of causation. The triad (sbRH, dbSDH and sfHIE) is a strong, specific and consistent association which suggests a causative link. Furthermore sbRH and dbSDH are positively correlated with severity of HIE but negatively correlated with objective evidence of trauma. The conventional hypothesis (post-Geddes) fails the test of biological plausibility because trauma causes asymmetrical and focal bleeding not diffuse bilateral capillary and venular oozing. The conventional hypothesis lacks coherence since it was shown that the encephalopathy of the triad is not traumatic. Arguments by analogy also support the Geddes hypothesis. There is a paucity of experimental evidence for either hypothesis but this is likely to be rectified as molecular methods (genomics and proteomics) are used in the infant autopsy.