Critique of evidence base and practice
In a 'Perspectives' article in the Journal of the Norwegian Medical Association, Knut Wester expresses his concern about the uncertainty involved in diagnosing 'shaken baby' cases (3). He claims that a number of people may have been convicted of having shaken a child based on insufficient medical evidence. He is currently working on a research project in which he will review previous court convictions to see whether this could be the case. Wester bases his assertions partly on a Swedish literature review from 2016 that appears to conclude that there is insufficient scientific evidence for shaken baby syndrome (4).
Wester's concerns must be taken seriously. It is our opinion, however, that the medical evidence for the diagnosis of physical abuse of children in general, and abusive head injury in particular, is far stronger than he proposes. In our view, both Wester and the authors of the Swedish literature review have used definitions that deviate from the real issue in the diagnostic workup.
Wester defines the diagnosis of the shaken baby as a triad of medical findings: subdural haematoma, extensive retinal haemorrhages and encephalopathy. This is a drastic oversimplification of the clinical presentation, and it is completely misleading when used as the sole grounds for the diagnosis. A finding of the so-called triad is not in itself sufficient proof of child abuse. It is true that in a forensic setting the triad has been used as an indicator that violent shaking may have been the mechanism of injury (5, 6). This assumes, however, that other findings point in the same direction and that an extensive workup has been conducted that critically assesses other potential mechanisms of injury and differential diagnoses. The three triad variables have distinctive features that may indicate abuse, but they cannot be used as binary yes-no variables.
Firstly, a subdural haematoma has a distribution, form, size and a radiological/pathological pattern that varies from case to case depending on the injury mechanism, pathophysiology, coexisting pathology and age. In a typical case of traumatic shaking, the haematomas are multifocal/bilateral and occur over the hemispheres, posteriorly and/or along the falx cerebri (7). There are often signs of injury to bridging veins at the midline (tadpole signs). At autopsy, injury to bridging veins can also be noted by careful dissection (8, 9).
Secondly, in typical cases of traumatic shaking, retinal haemorrhages have a characteristic appearance with numerous haemorrhages in multiple layers of the retina, located both centrally and peripherally in all four quadrants, but with a normal optic nerve papilla (10). This is substantially different from the retinal haemorrhages described in relation to an acute increase in intracranial pressure in which the haemorrhages are typically found only centrally and papilloedema is clearly present (11). In a prospective, population-based study with 45 cases of known/admitted traumatic shaking, there was a specificity of 97 % for traumatic shaking in cases of major preretinal haemorrhages together with other extensive retinal haemorrhages with or without retinoschisis (formation of folds around the macula) (12).
Thirdly, the term encephalopathy encompasses any conceivable type of brain injury. Very few shaken children have structural signs of brain injury (2). When disclosed, such signs often entail rapid development of cytotoxic oedema and diffuse brain tissue damage that may indicate a hypoxic-ischemic mechanism related to trauma. Such brain injuries may be asymmetrical, but are most often bilateral. In some cases an MRI may reveal a resemblance to hypoxic-ischemic injuries subsequent to circulatory failure resulting from a specific incident or illness (7). The child's medical history must be carefully reviewed to rule out such incidents. Less frequently, traumatic shaking can lead to more widespread/extensive brain injury of the type known as 'diffuse axonal injury' associated with high-energy trauma (6, 7).
When focus is placed solely on the triad, this implies a choice to disregard the other accompanying findings such as skeletal fractures and bruises as well as signs of caregiver neglect. (2). The Swedish literature review from 2016 concluded that there was insufficient scientific evidence to claim that children with the triad of findings had been violently shaken (4). However, the committee that conducted the study ignored all other clinical aspects and findings that are given weight when making a diagnosis (13). Since its publication, the Swedish report has been criticised for its methodological weaknesses and flawed reasoning (14, 15), and the Royal College of Paediatrics and Child Health in the UK has urged the authors to withdraw the publication or allow it to be subjected to international scrutiny (14).