Traumatic shaking: The role of the triad in medical investigations of suspected traumatic shaking.

The Swedish Agency for Health Technology Assessment and Assesment of Social Services (SBU) is an independent national authority, tasked by the government with assessing methods used in health, medical and dental services and social service interventions from a broad perspective, covering medical, economic, ethical and social aspects. The language in SBU's reports are adjusted to a wide audience. SBU's Board of Directors has approved the conclusions in this report. The systematic review showed the following graded results: There is limited scientific evidence that the triad (Three components of a whole. The triad associated with SBS usually comprises subdural haematoma, retinal haemorrhages and encephalopathy.) and therefore, its components can be associated with traumatic shaking (low-quality evidence). There is insufficient scientific evidence on which to assess the diagnostic accuracy of the triad in identifying traumatic shaking (very low-quality evidence). Limited scientific evidence (low-quality evidence) represents a combined assessment of studies of high or moderate quality which disclose factors that markedly weaken the evidence. It is important to note that limited scientific evidence for the reliability of a method or an effect does not imply complete lack of scientific support. Insufficient scientific evidence (very low-quality evidence) represents either a lack of studies or situations when available studies are of low quality or show contradictory results. Evaluation of the evidence was not based on formal grading of the evidence according to GRADE but on an evaluation of the total scientific basis. ©2018 Foundation Acta Paediatrica. Published by John Wiley & Sons Ltd.

Abusive Head Trauma: damage to the skull caused by maltreatment of the child

Accidental Trauma

Attenuation: absorption of radiation in the body, which varies in accordance with the density of the tissues.

Brain Oedema/Oedema

Interdisciplinary team which investigates cases of suspected child abuse

Computed Tomography

Increased volume of cerebrospinal fluid in the cavities of the brain

Accumulation of fluid, possibly arising after an earlier episode of bleeding

Inflicted Head Injury or Intentional Head Injury: head injury caused by abusive maltreatment of the child

Inflicted Head Trauma: injury to the head resulting from abusive maltreatment of the child

Metaphyseal Fracture: a fracture in the growth zone of a long bone, for example in the shinbone just below the knee

Magnetic Resonance Imaging

Magnetic Resonance Tomography (see MRI)

Retinal Bleeding/haemorrhage, intraocular bleeding

Subarachnoidal Haemorrhage: bleeding in the subarachnoid space, i.e. between the soft meninges of the brain)

Shaken Baby Syndrome; a syndrome comprising three components, the triad

Subdural Haemorrhage, Subdural Haematoma: bleeding under the dura Subarachnoid spaceThe space between the soft meninges

The injurious mechanism when a child is shaken violently (not to be confused with the medical findings, ‘the triad’)

Three components of a whole. The triad associated with SBS usually comprises subdural haematoma, retinal haemorrhages and encephalopathy

Aim

In cases of suspected traumatic shaking, the diagnosis has conventionally been based on three findings, referred to collectively as the triad, namely: subdural haematoma (bleeding between the dura mater and the brain), retinal haemorrhages and various forms of brain symptoms (encephalopathy). The presenting history is often that of lethargy, seizures and apnoea. The purpose of this evaluation was to determine how reliably the triad or its components can be explained by traumatic shaking of children up to one year of age.

Background

Child abuse was described in the medical literature as early as in the 1800s 1, but it was only much later that awareness of the practice became more widespread 2, 3. Child abuse can often be concealed within the family, and there is a risk of underdiagnosis, in part because the child is unable to speak for itself. At the same time, overdiagnosis can have serious consequences, because families can be split apart on false grounds. Adherence to the healthcare principle that the triad is attributable exclusively to traumatic shaking can lead to overdiagnosis, because of failure to consider other possible causes of the child's condition.

What is traumatic shaking?

Traumatic shaking occurs when a child is shaken in such a way that its head is flung backwards and forwards. In 1971, Guthkelch, a neurosurgeon, hypothesized that such shaking can result in a subdural haematoma, in the absence of any detectable external signs of injury to the skull 4. The article describes two cases in which the parents admitted that for various reasons they had shaken the child before it became ill. Moreover, one of the babies had retinal haemorrhages. The association between traumatic shaking, subdural haematoma and retinal haemorrhages was described by Caffey in 1972 and referred to as Whiplash Shaken Infant Syndrome 2. The injuries were believed to occur because shaking the child subjected the head to acceleration – deceleration and rotational forces. In 1987, this theory was queried by Duhaime et al. 5 in a biomechanical study which concluded that isolated shaking, in the absence of direct violence, is probably not of sufficient force to cause the injuries described above.

The name of the condition has since been changed to Shaken Baby Syndrome (SBS). There are a number of studies on the association between various clinical and radiographic findings and injuries caused by violent shaking of a child 6, 7, 8, 9, 10, 11, 12, 13.

In recent years, the term abusive head trauma (AHT) has been introduced (see section ‘Terminology’). The project group decided to apply the term ‘traumatic shaking’ to the trauma mechanism and the ‘triad’ to the actual signs and symptoms 13.

Signs and symptoms

In the scientific literature, various signs and symptoms are described in association with traumatic shaking. The collective name ‘triad’ has been adopted for the most frequently occurring injuries (subdural haematoma, retinal haemorrhage and encephalopathy). The main focus of this report is the triad (see section ‘Terminology’). Other signs are occasionally reported in association with traumatic shaking, including bruising to the chest, fractures of, for example, the ribs and shinbone (metaphyseal fractures), but these injuries are not included in the present review.

Presenting medical history

When medical attention is sought for the affected child, the presenting history includes various clinical signs such as seizures, lethargy or other symptoms of encephalopathy. The initial clinical and radiographic examination can disclose the presence of, for example, subdural haematoma, or various symptoms of brain dysfunction. Subdural haematoma, retinal haemorrhages and various forms of encephalopathy can have serious sequelae, with permanent damage to the brain and/or the eyes. Permanent damage can comprise serious impairment of cognitive and/or motor function, with widespread adverse effects on the child's health, development and future quality of life and can ultimately even be fatal.

Healthcare personnel is encouraged to be alert to the findings which comprise the triad and are required by law (Social Services Act: Chapter 14, Section 1) to notify the Board of Social Welfare if they become aware, or suspect, that a child is being abused, or otherwise may need protection. In a frequently quoted article by the American Academy of Pediatrics, 1993 14, physicians documenting trauma affecting the brain in newborns are encouraged to conduct a thorough examination and to be familiar with the clinical and radiographic findings which can confirm damage caused by traumatic shaking. The regional healthcare plan published by the Stockholm County Council 2011 stated as follows:

If there is no history of a traffic accident or a fall from a considerable height, the combination of subdural hematoma and encephalopathy with edema or hemorrhage strongly suggests that the child has been abused. If there are also retinal hemorrhages then from the medical point of view the diagnosis of abuse is quite clear 15.

Other regional care programmes as well as the statement by the Swedish National Board of Health and Welfare on children who are being abused or are at risk of abuse include guidelines on the care of infants in cases of suspected abuse 16, 17.

In recent years, however, the certainty with which it can be determined that the findings of the triad are in fact attributable to traumatic shaking has been questioned 18, 19, 20, 21, 22, 23, 24, 25, 26. Many articles which have debated the subject of traumatic shaking and the symptoms and signs of the triad have been published in international and national journals and in the media. In this context, it is important to ascertain whether the conclusion that traumatic shaking is the cause of these signs and symptoms is based on evidence of the highest possible scientific quality. However, grading of scientific supporting evidence is based on the assessment of groups – not of individuals. In order for the justice system or social services to make a statement on the association between exposure and disease or injury, assessment of the individual case is required, with other observations and conditions also taken into account.

Terminology

The English term for the triad is Shaken Baby Syndrome (SBS), which refers to the signs and symptoms which allegedly can arise after an episode of isolated traumatic shaking, that is shaking without the head impacting on any object.

In 2009, the American Academy of Pediatrics recommended a broader term, Abusive Head Trauma (AHT), which includes also direct trauma to the head 27. Also, to be found in the literature are several other terms, which partly or completely overlap the terms Shaken Baby Syndrome and Abusive Head Trauma (see Chapter 9). The terms are used in a variety of ways in the scientific literature, and this contributes to the lack of methodological clarity in studies of the effects of traumatic shaking. The project group has therefore decided to limit the scope of the project to isolated traumatic shaking, thereby including only studies of cases in which there is no evidence of direct trauma (external injury) to the head. Furthermore, the authors have avoided the terms SBS and AHT because they imply both the signs and symptoms and the alleged mechanism behind the findings, even the intent. Instead, the authors chose to make a clear distinction between the injurious mechanism (‘traumatic shaking’) and the medical findings (‘the triad’).

Investigation of injuries which may be attributable to traumatic shaking

Diagnosis of suspected brain injury is based on computerised tomography (CT) and/or magnetic resonance imaging (MRI). The presence of retinal haemorrhages is determined by examination of the fundus by ophthalmoscopy or fundoscopy.

Other possible causes (differential diagnoses) of the triad and its components

In cases presenting with the triad, it is important to determine whether these can be attributed to causes other than traumatic shaking.

Subdural haematoma, retinal haemorrhages and encephalopathy have been described after delivery and in association with such conditions as various convulsive states, certain haemorrhagic diseases, infectious diseases, metabolic disorders, immunological diseases, skeletal diseases and vascular malformations (see Appendix 1 for details).

The triad – Signs and symptoms

Subdural haematoma

It is well known that trauma to the head can give rise to subdural haematoma. In an adult, the underlying mechanism is rupture of one or more of the bridging veins, with bleeding into the subdural space. In many such cases, there are often also external signs of trauma to the head in the form of soft tissue bleeding, but in other cases, a CT scan may disclose internal injury in the absence of any evidence of external trauma.

Shaking an infant causes the movement of the brain to be out of synchrony with the movement of the skull. However, there is lack of consensus about the mechanism underlying the bleeding. It has been proposed that it may result from capillary injury 28, 29. As isolated traumatic shaking does not involve direct trauma to the head, there will be no external signs of head trauma such as swelling of soft tissues, contusions, lacerations or skull fractures. Hence, an incident is not classified as isolated traumatic shaking when soft tissue injuries or skull fractures are detected. Soft tissue injury and skull fracture(s) are therefore findings which exclude isolated traumatic shaking.

Imaging techniques have shown that subdural haematoma can occur in association with vaginal delivery but is usually resorbed within a few weeks 22.

In the space created by the haematoma, effusion (leakage of fluid) may result in the development of a so‐called subdural hygroma, which contains cerebrospinal fluid. It has been proposed that further bleeding in this space could occur spontaneously, or as a result of minor trauma 30, 31, 32, 33. It has also been proposed that an enlarged subarachnoid space could increase the risk of subdural haemorrhage 19, 34, 35, 36.

Retinal haemorrhages

Retinal haemorrhages associated with traumatic shaking have been attributed to transfer of shearing forces in the vitreous body of the eye to the retina, due to increased pressure in the venous blood vessels in the retina, resulting in rupture of the vessels 37, 38, 39. It has also been proposed that during shaking, repetitive acceleration and deceleration create shearing forces between the vitreous body and the retina, as well as direct injury to the eyeball. However, bleeding in the fundus of the eye has also been demonstrated in association with subdural haematoma considered to be caused by disease and it is therefore possible that retinal haemorrhages can arise as a sequel to subdural haematoma. One possible explanation is that increased intracranial pressure caused by oedema of the brain leads to increased pressure in the central optical vein, with congestion in the retina 40, 41. The relationship between subdural haematoma and retinal haemorrhages is supported by studies showing that isolated incidents of retinal haemorrhages are very rare 38, 41. Retinal haemorrhages have also been observed after normal vaginal deliveries 42.

Encephalopathy

Encephalopathy can present with such signs as lethargy, seizures and dyspnoea, among others. These signs may be attributable to frictional damage in the brain or the cervical medulla, and/or brain oedema. Brain oedema and brain hypoxia can cause irreversible brain damage. Increased intracranial pressure, for example, due to brain oedema or subdural haematoma, can also result in seizures, apnoea and lethargy 43, 44. Brain oedema can be revealed by both computed tomography (CT) and magnetic resonance imaging (MRI) and appears as effacement of the sulci and compression of the cerebral ventricles. These may be temporary conditions, which resolve without any permanent brain damage.

The most serious condition can be revealed by CT and MRI as reduced differentiation between the white and grey matter of the brain, representing a global irreversible ischaemic injury.

Diagnostic methods

Intracranial examination

While CT is based on the differences in absorption of x‐ray radiation of substances and tissues of varying density, MRI exploits a number of different properties of substances and tissues and thereby offers a richer and often more specific characterisation of the tissue being examined.

Both techniques allow observation of thin ‘sections’ through the entire brain, with reconstructions in several planes, and also assessment of the intracranial vessels (using contrast medium in the vessels). Both techniques provide similar information on changes in the brain ventricles and basal cisterns, such as compression of the ventricles in brain oedema, widening in hydrocephalus, displacement due to haemorrhage and the risk of brain herniation.

However, MRI can provide different information from CT, for example with respect to the presence of fresh blood, deposition of hemosiderin (a decomposition product of haemoglobin) and early ischaemic and axonal injuries 45, 46.

While an acute subdural haematoma in a small child comprises fresh blood, a subacute subdural haematoma is usually composed of a mixture of an upper layer of fluid and a sediment of coagulated blood 47. The development of the haemorrhage over time results in different patterns on CT and MRI. The time frames for the development and duration of these patterns can overlap; hence, determination of the age of the injury is uncertain 48. In rare cases, calcifications can be mistaken for fresh blood, particularly in the brain tissue. On a CT scan, haemorrhage has a more robust pattern than that seen on various MRI sequences, which have a varying and partly overlapping appearance, depending on the composition of the bleeding and the time elapsed since the injury. CT assessment of the age of a subdural haemorrhage is therefore considered to be more reliable than assessment by MRI 49, 50. The ability to determine the age of a subdural haematoma can be important for correlation with the alleged time of injury.

Both CT and MRI can be used to determine brain oedema, which appears as effacement of the sulci on the surface of the brain and compression of the ventricles and basal cisterns. CT is more reliable than MRI for assessing fractures.

Retinal examination

Two methods can be used for examination of the ocular fundus. The most common is fundoscopy with or without dilatation of the pupil. More recently, a photographic method has been developed (RetCam). This method allows subsequent assessment of the findings by other observers who are not aware of the case history or the purpose of the examination 51, 52.

At autopsy, the entire eye can be examined, and other findings can then be described 53, 54.

Haemorrhage in the ocular fundus cannot usually be assessed by CT or MRI. However, in a recently published MRI study, a particular imaging sequence was compared with ophthalmoscopy and it was shown that in 83% of cases, retinal haemorrhages could be detected by MRI 55, 56.

In this context, it is important to be aware that interpretations of CT, MRI and ocular fundoscopy findings are somewhat subjective and the experience of the individual observer can influence the final assessment.

Systematic evaluation: method

Question to be addressed

The aim of the present investigation was to address the following question: With what certainty can it be claimed that the triad, subdural haematoma, retinal haemorrhages and encephalopathy is attributable to isolated traumatic shaking (i.e. when no external signs of trauma are present)?

PIRO

P (Population): Children ≤12 months of age.

I (Index test): The triad in cases of suspected traumatic shaking.

R (Reference test/gold standard): Admitted or witnessed traumatic shaking or other trauma.

O (Outcome measure): Diagnostic accuracy.

The project has been conducted in accordance with the method described in SBU's manual 57.

Criteria for inclusion and exclusion

Inclusion criteria

Study design

Case–control, cohort and registry studies and studies applying qualitative methods of analysis.

Observations

To reduce the risk of random errors of selection, only studies comprising 10 or more cases were included. With respect to possible alternative explanations (differential diagnoses), the project group was of the opinion that one published case was sufficient to question the hypothesis that the triad is always caused by traumatic shaking. Articles on differential diagnoses were not quality assessed and are therefore not included in the basis for the results. If a subgroup of children who had been subjected to traumatic shaking and/or a subgroup aged ≤12 months (median and/or mean age) was included in AHT studies, then these were included by the project group. The peak age of children subjected to traumatic shaking is stated to be 2 months 58 and the project group therefore decided to limit the review to studies of children with a mean or median age of ≤12 months.

Language

Articles written in English, German, French, Swedish, Danish and Norwegian were included.

Other criteria

The project group decided to include only cases of traumatic shaking which were witnessed (e.g. video recorded) or in which someone had confessed to shaking the child.

Exclusion criteria

The project group excluded studies of fewer than 10 cases and AHT studies which included external injury to the head and/or fractures and other injuries.

Studies identified in the literature search as biomechanical studies and studies which deal with other possible causes of the triad have been considered separately and are presented in Appendices 1 and 2.

Methodology for selection of studies

Based on the question to be addressed by the project, the literature databases were searched systematically, in close collaboration between the information specialist and the experts in the project group. The literature search encompassed the databases PubMed, Embase and Cochrane Library through October 15th, 2015. Further studies were searched for manually, through the reference lists of individual studies and systematic reviews. For a detailed description of search terms and limitations, see Appendix 4, www.sbu.se/255e.

Assessment of relevance

The lists of abstracts generated by the literature search were scrutinised independently by two experts. Studies deemed by at least one of the experts to be relevant to the questions to be addressed by the project were retrieved in full text and scrutinised independently by two experts with reference to the project's inclusion criteria. Articles which were scrutinised in full text and did not meet the inclusion criteria were excluded as follows: the main reason for exclusion was recorded (see Appendix 5, www.sbu.se/255e). Disagreements were addressed initially by discussion between the two experts who had read the article. In certain cases, the entire project group was involved in the discussion and the decision about inclusion or exclusion was resolved by consensus.

Assessment of the quality of individual studies

Because of the specific field of research, the project group modified SBU's template to assess the quality of the included studies and to determine the risk of bias (circular reasoning: see the section ‘Circular reasoning in clinical and research settings’ in Chapter 5). The template includes i.a. the type of study (prospective, diagnostic, biomechanical, etc.), the main focus of the study and whether the study addressed subdural haematoma, retinal haemorrhages and/or encephalopathy. In accordance with SBU's guidelines, only studies of moderate or high quality were considered in the results and discussion 11.

Systematic reviews of the field were quality assessed using the AMSTAR instrument 57. The results in the present report were based on original studies and not on other systematic reviews (see Chapter 5).

Method for synthesis of the results

Meta−analysis is a statistical method for quantitatively appraising the results of several studies to obtain data from a larger sample and to achieve a more reliable assessment of the statistical uncertainty. To pool the results, the studies must have been conducted using similar methods and it must be possible to adjust the analyses for similar background factors. As only one of the included studies used a reference group, it was not possible to undertake a meta−analysis.

Assessment of the quality of the evidence

The quality of the evidence indicates the level of reliability of the results and is based on the assessment of study quality (risk of bias), inconsistency, imprecision, risk of publication bias and indirectness.

As no meta−analysis was possible, the results were based on a narrative synthesis of the included studies. Evaluation of the evidence was not based on formal grading of the evidence according to GRADE but on an evaluation of the total scientific basis. The quality of the evidence was deemed to be limited (low) when combined assessment of studies of high or moderate quality disclosed factors which markedly weaken the evidence. The quality of the evidence was deemed to be insufficient (very low) when there was a lack of studies, when the available studies were of low quality or when studies of similar quality showed contradictory results. It is important to note that limited evidence for the reliability of a method or an effect does not imply complete lack of scientific support.

Results

The literature search yielded 3 773 abstracts, of which 1 065 were retrieved in full text. Of these, 1 035 were excluded because they did not meet the inclusion criteria. Of the 30 remaining studies, two were assessed to have moderate quality and none of high quality. The main reason that so few studies met the quality requirements was that the published papers failed to provide information as to whether traumatic shaking was confessed to by the perpetrator or had been witnessed. Thus, the results are based on only two studies of confessed traumatic shaking and a meta−analysis was therefore not possible. However, agreement between the results of the included studies was discussed in the project group (Fig. 1).

Figure 1

Flow chart of literature search.

Quality of the evidence

The systematic review showed the following graded results:

There is insufficient scientific evidence on which to assess the diagnostic accuracy of the triad in identifying traumatic shaking (very low‐quality evidence).

There is limited scientific evidence that the triad and therefore, its components can be associated with traumatic shaking (low‐quality evidence).

The two included studies of moderate quality, both conducted in France, were based on cases in which the perpetrator confessed to subjecting the child to traumatic shaking. The study by Vinchon et al. was a prospective study. It was based on a register of traumatic head injury in children aged under two years, who were admitted to hospital between May 2001 and February 2009, in a catchment area with a population of about 4 million 59. The material comprised 412 cases, of which 124 were classified as Inflicted Head Injury (IHI) and 288 as Accidental Trauma (AT).

In the group with inflicted injury (IHI group), there were 45 confessed cases: 30 by traumatic shaking and 15 in which the perpetrator admitted to other external trauma. However, the article does not include detailed descriptions as to how the perpetrator inflicted the injuries, nor the circumstances under which the confession was obtained. This group of children was compared with 39 cases in which accidental trauma was witnessed in a public place (AT group).

In the group with inflicted trauma, 37 of 45 (82%) had a subdural haematoma, compared with 17 of 39 (44%) in the accidental trauma group; 37 of 44 (84%) had retinal haemorrhages, compared with six of 35 (17%) in the accidental trauma group and 12 of 45 (27%) had cerebral ischemia, compared with one of 39 (3 per cent) in the group with accidental head trauma.

The study by Adamsbaum et al. was a retrospective observational study, comprising 29 confessed cases of traumatic shaking (in which direct trauma to the head was described in five cases) and a comparative group of 83 unconfessed cases 60. The criteria for inclusion in the study were subdural haematoma disclosed by a CT scan and confession by the suspected perpetrator. As subdural haematoma was one of the criteria for inclusion in the traumatic shaking group, only the results for retinal haemorrhages could be used in this investigation.

In the group in which traumatic shaking was confessed to (Group A), 24 children (83%) had retinal haemorrhages. In all cases where the perpetrator had confessed, the shaking was described as violent (100%) and in some cases (55%), the perpetrator admitted to repeated episodes of shaking. No correlation was established between the density of the subdural haematoma and the number of repeated episodes of shaking. In 14 of 29 cases in Group A, there was a detailed description of how the suspect had committed the act. In the other group (Group B), there were children who had been shaken in an attempt at revival or had suffered accidental injury and some children for whom no explanation of the condition was presented as follows: thus, this group cannot be considered an acceptable reference group.

The studies by Vinchon et al. and Adamsbaum et al. both demonstrate that traumatic shaking can cause subdural haematoma and retinal haemorrhages. In the study by Vinchon et al., the group in which traumatic shaking was confessed to comprise a larger proportion of children with subdural haematoma, retinal haemorrhages and cerebral ischemia than the group of children who had been injured in a witnessed accident. Adamsbaum et al. compared a group of children in which the perpetrators had confessed to traumatic shaking, with a group of children in which the suspects had not confessed: this can result in inclusion bias in one or both groups. As only one of these two studies had a relevant reference group, it has not been possible to conduct a meta−analysis.

There are also other published cases which have been excluded (wrong population, wrong study design), but contain detailed descriptions of confessions which are in accordance with the two studies of moderate quality 61, 62. Because of the low number of studies of moderate or high quality, it was not possible to determine the diagnostic accuracy of the triad in identifying traumatic shaking (Table 1).

Table 1

Characteristics of included studies

Author Year Reference Country	Aims/focus Outcome	Study design Setting Inclusion and exclusion criteria	Number of cases Medical examinations	Results	Methodological considerations/ comments	Study quality
Vinchon 2010 59 France	Aims/focus: To compare the clinical, ophthalmological and radiological features of inflicted head injury (IHI) and accidental trauma (AT) and to test the diagnostic value of these findings Outcome: Registered medical findings including SDH (density), RH and brain ischemia	Study design: Prospective cohort study collecting medical information on all traumatic head injuries in infants from May 2001 to February 2009 Setting: Diagnosis in paediatric intensive care unit or neurosurgical unit Inclusion criteria: Children under the age of 24 months referred alive to the emergency room due to confessed inflicted head trauma (IHT), or witnessed accidental trauma (AT) Exclusion criteria: Obstetric trauma	Number of cases: IHI group: 45 cases of confessed inflicted head injury, 29 boys and 16 girls, mean age 3.8 month (range 0.8–18.3). IHI was caused by shaking in 30 cases and beating in 15 cases AT group: 39 cases of witnessed accidental injury, 23 boys and 16 girls, mean age 8.1 months (range 0–23.9) 19 cases where the baby was a car passenger Three cases where the babywas in a carriage hit by a vehicle Five cases of defenestration 11 cases injured by a short fall One case without description Medical examinations: Clinical examination including neurological symptoms CT scanning Fundoscopy X‐ray and/or isotopic bone scanning skeletal survey	IHI group: SDH: 37/45 (82%) RH: 37/44 (84%) Brain ischemia: 12/45 (27%) AT group: SDH: 17/39 (44%) RH: 6/35 (17%) Brain ischemia: 1/39 (3%) Diagnostic value SDH: Sensitivity: 0.822 Specificity: 0.552 PPV: 0.685 NPV: 0.724 Severe RH: Sensitivity: 0.556 Specificity: 0.974 PPV: 0.961 NPV: 0.655 Brain ischemia: Sensitivity: 0.267 Specificity: 0.971 PPV: 0.921 NPV: 0.505	The report does not contain a detailed description of the method nor of the IHI cases. It is not possible to extract the results for the shaken baby cases in the IHI group There are limitations in the definitions of the diagnostic criteria The triad used by the authors consists of subdural haematoma, retinal haemorrhage, and absence of sign of impact No details about what the suspect had confessed and under which circumstances the confession was obtained	Moderate
Adamsbaum 2010 60 France	Aims/focus: To correlate the history of confessed abusive head trauma (AHT) cases with medical findings and to compare medical findings in confessed cases of AHT with nonconfessed cases Outcome: Density and location of SDH, RH, fractures and skin ecchymosis	Study design: Retrospective observational study with examined forensic evidence, from January 2002 to May 2009 Setting: Not specified Inclusion criteria: Children with AHT (the presence of SDH on CT scan with or without traumatic skin lesions) and a perpetrator conviction, with or without confession Exclusion criteria: Accidental trauma and metabolic or infectious pathology	Number of cases: Group A: Twenty‐nine cases where the perpetrator confessed violent shaking of the child. In five cases, a final impact of the infant's head on a bed was described by the perpetrator. 22 boys and 7 girls, mean age 4.7 ± 2.9 months. 27 cases were younger than 1 year Group B: Eighty‐three cases without confession of a causal relationship between violence and the child's symptoms. 63 boys and 20 girls, mean age 6 ± 5.3 months 19 cases where the children were shaken to revive it from a life‐threatening event 28 cases where a minor accident was described 36 cases where no mechanistic description was given Medical examinations: Clinical examination CT scanning Fundoscopy X‐ray skeletal survey MRI (only in confessed cases)	Group A: RH: 24/29 (83%) SDH hyperdensity: 11/29 (38%) SDH hypodensity: 2/29 (7%) SDH mixed density: 16/29 (55%) Nine infants in Group A died. In four of the cases, the perpetrator admitted head impact Group B: RH: 75/83 (90%) SDH hyperdensity: 28/83 (34%) SDH hypodensity: 0/83 (0%) SDH mixed density: 55/83 (66%) 16 infants in group B died No statistically significant difference between the two regarding incidence of RH, density of SDH, or location of SDH	A retrospective study design without an appropriate control group The inclusion criterion SDH was part of the triad, why the prevalence of SDH after shaking cannot be evaluated The report contains a detailed description of 14 cases in group A. No details under which circumstances the confession was obtained	Moderate

AHT = Abusive head trauma; AT = Accidental trauma; CT = Computed tomography; IHI = Inflicted head injury; IHT = Inflicted head trauma; MRI = Magnetic resonance imaging; NPV = Negative predictive value; PPV = Positive predictive value; RH = Retinal haemorrhage; SDH = Subdural haematoma.

Discussion

Although relatively many studies met the criteria for inclusion, the literature search identified only two studies of moderate quality. This is disconcerting, because traumatic shaking is very serious and has dramatic consequences for both the child and its family. The research field is complex, but this does not excuse, for example, circular reasoning and inadequate presentation of data collection. It is important that reviews of the field include consideration of the methodological flaws which characterise this field of research.

The studies by Adamsbaum et al. and Vinchon et al. were deemed to be of moderate quality. Although both studies have methodological limitations, they support the hypothesis that isolated traumatic shaking can give rise to the triad.

The prospective study by Vinchon et al. was based on more than 400 cases: 124 were classified as inflicted and 288 as accidental injuries to the skull. Forty‐five were cases of confessed inflicted skull injury, of which 30 were cases of confessed isolated traumatic shaking (IHI group).

Thirty‐nine cases were witnessed accidents (AT group). The advantage of this study is that all trauma cases presenting at the hospital were registered prospectively for many years. The study also has a clearly defined reference group, of children who had accidentally sustained injuries in the presence of a witness. However, this group of children is significantly older. One of the limitations of the study is the lack of detailed description of how and when the shaking incident occurred. Vinchon et al. analysed the components of the triad separately, but the authors also introduced a different combination for the triad, namely subdural haematoma, retinal haemorrhages and the ‘absence of scalp swelling’.

However, data on effects on the brain are registered in the study, in the form of seizures, lethargy and coma, among others. By definition, the group of children with isolated traumatic shaking will comprise only those cases without signs of external trauma, while the group with accidental injury will include cases with signs of external trauma. At an early stage of the investigation, the project group contacted Dr. Vinchon to clarify certain ambiguities, but not all queries were answered. The questions included i.a. how retinal haemorrhages were defined, how the authors calculated sensitivity, specificity and the predictive value of the triad, why they chose a different triad component (‘absence of scalp swelling’ instead of encephalopathy) and under what circumstances the suspected perpetrator had confessed and what had been confessed.

The study by Adamsbaum et al. was a retrospective observational study which included 29 cases in which a suspect confessed to traumatic shaking. While detailed confessions were presented for 14 cases, it cannot be discounted that among the cases for which no detailed confession was forthcoming, there could be some in which shaking occurred after the child exhibited symptoms of brain damage. The study group was compared with a reference group comprising 83 unconfessed cases. However, this is not a ‘true’ reference group, as there may be cases of traumatic shaking among the unconfessed cases. As subdural haematoma is a criterion for inclusion of all cases in the study, only the results for retinal haemorrhages can be considered.

During the literature review, the project group identified other conditions or events which can also give rise to the three components of the triad. Some of these conditions or events do not result in permanent disability or are very rare, but it should be noted that the triad or its components can be attributable to causes other than shaking. It is therefore important to consider these possible differential diagnoses in investigations of suspected traumatic shaking. Decisions made by social services or the court system are based not only on medical findings, but also on other evidence.

An analysis of biomechanical studies (Appendix 2) disclosed contradictory results and no conclusions can be drawn as to the minimal forces capable of generating these injuries in children.

Methodological issues

This review of the scientific evidence for diagnosis of traumatic shaking in children under the age of 12 months (mean or median age ≤12 months) disclosed a number of methodological issues in the published studies.

Definition of traumatic shaking

The project was limited to studies in which traumatic shaking was considered to be the primary cause of the child's injuries, but several studies have adopted a wider definition, for example inflicted head injury. Thus, it has not always been possible to distinguish between an injury attributable to traumatic shaking and an injury attributable to direct trauma to the head.

Classification of subjects into groups

Another methodological problem was that traumatic shaking was not always witnessed or confessed to by the suspected perpetrator; hence, correct classification of cases into a traumatic shaking group or a reference group was uncertain. Thus, there is a risk of incorrect evaluation of the association between the triad and traumatic shaking. Although there is a risk of false confessions, apart from film documentation, this is the only means of gaining an insight into what has actually happened to the infant. Because of the risk of false confessions, all confessions in these studies must be considered with caution.

Thus, there are some risks associated with the decision of the project group to include only cases in which someone has confessed. The confession could be false because it was made as part of a plea bargain. It could also be false because the suspect has felt impelled to confess 63, 64, 65.

Circular reasoning in clinical and research settings

Under the Social Services Act, the Board of Social Welfare must be notified not only of all cases of (suspected) child abuse, but also of other cases in which a child may be deemed to be vulnerable to harm and in need of protection. Those required to notify suspected child abuse are personnel within the health and medical services, dental, preschool, school, social and criminal services (Chapter 14, Section 1 of the Social Services Act).

In many cases, it is a child protection team which investigates cases of suspected traumatic shaking. Over the years, these teams have developed criteria based on certain symptoms and signs which can be associated with isolated traumatic shaking, after exclusion of other possible causes of the child's condition 66, 67. Some of these criteria are associated with the carer's credibility. The carer is not considered trustworthy if he/she cannot provide an ‘acceptable’ explanation for the child's condition, for example that the child had fallen from a low height and had not sustained any external injury. A change in statement – for example, the carer first denies shaking the child and later admits to doing so, but only after the child had stopped breathing or lost consciousness – also reduces the carer's credibility. If the child was shaken because it suddenly showed signs of being unwell (such as dyspnoea or apnoea), it is, however, reasonable to assume that the child's condition was already caused for concern before it was shaken and thus, the symptoms were not attributable to the shaking. If, however, such an explanation of events is not deemed ‘acceptable’, the case is still classified as a case of traumatic shaking.

The child protection team's criteria are based primarily on a clinical approach 66, 67. Problems arise later, when and if these criteria are not tested unconditionally by researchers in systematic studies of the association between the triad and traumatic shaking. This means that the interpretation made by the child protection team characterises the scientific investigation and hypothesis testing and this, in turn, means that the conventional approach is reinforced instead of being tested. However, if before the study, it has already been assumed that the question to be addressed by the study has been answered, that is the association between the symptoms and signs of the triad and traumatic shaking has already been described (according to the child protection team's criteria), then circular reasoning occurs. Applied in this context, the reasoning results in a high risk of bias, which in turn results in a situation wherein the researcher does not know what is being compared (the traumatic shaking group may include children who have not been shaken and the reference group may include children who have been shaken). Sensitivity, specificity and predictive values calculated on comparison of such groups will result in incorrect conclusions. It will also result in incorrect calculations of incidence.

To avoid such circular reasoning, study cases and control cases must be identified introcontrovertibly. The project group has chosen to accept as study cases only those in which there was a witness to (or video documentation of) an incident of shaking or where someone has made a detailed confession of shaking the child.

Diagnostic methods

There is uncertainty in determining the time at which a subdural haematoma arose. Moreover, this uncertainty is greater in children under 12 months of age, because the characteristics of subdural haemorrhage at this age differ somewhat from those in adults. A subdural haematoma in a small child or infant usually consists of an upper layer of fluid and a sediment of coagulated blood: if the subdural haematoma is subacute, this layer can exhibit various degrees of attenuation 47. The application of CT and MRI scans has recently reduced this uncertainty somewhat 46, but caution must still be exercised in assessing the age of a haematoma because of the existence of different and partly overlapping patterns 48.

In both controlled experimental and observational studies, systematic errors can occur because various observers do not always make the same observations and/ or interpret the observations differently. Agreement among different investigators in a study can vary according to how well trained the observers are. This applies not only in general to observations and assessments, but of course also to examinations and assessments of the symptoms and signs in cases of suspected traumatic shaking.

In one study, for example, there were major variations among the observers’ interpretation of retinal haemorrhages, that is interobserver agreement was low 51.

Comparison with results of other reviews

The project group identified seven systematic literature reviews addressing the same or partly the same questions as the present report 68, 69, 70, 71, 72, 73, 74. These reviews are not included in the results section of the present paper, but the project group scrutinised and assessed them because they are frequently cited in the scientific literature. All the systematic reviews were assessed by the project group to be of low quality (high risk of bias). Many of them were based on studies in which a team considered that a child had been shaken if it presented with the triad (circular reasoning, see section ‘Circular reasoning in clinical and research settings’). Another weakness in these reviews was that traumatic shaking was not specified and the more general term AHT was used instead, without a detailed description of what this term included.

Issues for future research

It is not possible to conduct randomised experiments in which children of various ages are subjected to various degrees of shaking. Biomechanical studies using dummies or models equipped with various inbuilt measuring instruments have been used to investigate the impact of mechanical forces on a child, but the results are contradictory. Furthermore, for various reasons, it is difficult to extrapolate the results of animal experiments to infants.

The project group was therefore limited to observational studies in which exposure (in this case shaking) was assumed to have occurred. The most reliable are prospective cohort studies and ideally those subjects included in a traumatic shaking cohort should comprise cases in which the perpetrator has confessed in a detailed confession, including documentation of the circumstances under which the confession was obtained.

In many of the scrutinised studies, the children in the reference group were significantly older than those in the traumatic shaking group. The brain, skeleton and neck muscles in a 2−month‐old baby are different from those of an 8 month old. Hence, at the age of 0–2 months, an infant can be assumed to be more vulnerable to injury from shaking than an older baby. Comparison of two groups of children (traumatic shaking and accidental injury groups) which are not age−matched can lead to selection bias and incorrect conclusions. Studies with matched age groups would allow calculation of sensitivity and specificity and predictive values. In this context, an opinion on the probability that the triad was attributable to traumatic shaking could be expressed with greater certainty.

There is a lack of detailed knowledge about the pathophysiology of the development of subdural and retinal haemorrhages associated with vaginal delivery.

Although most bleedings related to delivery are symptomless and disappear (are resorbed) within a few months, occasionally a haemorrhage can degenerate into a hygroma 19, 30, 36. This circumscribed collection of fluid is contained by a membrane in which small vessels form and it is considered that this in turn can lead to renewed bleeding (rebleeding) and a chronic subdural pool of fluid. The possibility cannot be discounted that in certain cases, rebleeding can cause symptoms 19, 36. This could be one reason why a child suddenly exhibits signs of encephalopathy (lethargy, apnoea and/or seizures), causing the carer to seek medical attention. Hypothetically such rebleeding could occur spontaneously or in response to minor trauma. There is therefore an urgent need for research into the pathophysiology and the natural course of subdural and retinal haemorrhages. Our understanding of the sequelae to traumatic shaking could also be improved by the development of better biomechanical models, for example models which take into account the impact of traumatic shaking on both the brain and the cervical vertebrae.

What measures are required to address the scientific uncertainties?

The reasons for scientific uncertainty in this field vary and should therefore be managed in different ways; from coordination of the entire field of research with respect to the direction future research should take, to conducting studies using correct methodologies and detailed descriptions of how the studies have been conducted.

International coordination

To improve diagnosis within the field, broad coordination at international level is required to ensure a study population of adequate size. Researchers in the field should strive to agree on which research questions are most urgent and collaborate to facilitate larger studies and to use similar study designs, allowing the results to be compared more readily. It should also be possible to establish an international register of confessed and well‐documented cases.

Priority research topics

Of particular importance are studies intended to improve the diagnostic accuracy of diagnostic imaging of the brain, the cervical spine and the eyes 75. There is also a need for better methods of studying the natural course of the observed injuries. Differential diagnosis such as bleeding in neonates associated with delivery also needs to be studied to identify the natural course of events 22, 36, 76, 77. Further research is also required to improve understanding of the pathophysiology underlying the triad. Refined biomechanical models would also contribute to improved understanding of traumatic shaking.

As far as possible, of course, studies should meet all the predetermined quality criteria. It is also important that the researchers are blinded with respect to the suspected mechanism of origin of the injuries and that the results are presented in such a way as to allow diagnostic accuracy to be calculated. This latter requirement thus means that each individual finding must be assessed in both the study group and the reference group.

One of the reasons that it was difficult to find evidence in this field is that in many studies the method and the results were inadequately described. With respect to future studies, the project group presents the following recommendations of requirements to be met, in order that the quality of the studies can be assessed and that meta−analyses can be conducted as follows:

The studies should:

Comprise prospective observational studies of confessed and well‐documented cases with reliable methodology, in which the risk of false confessions was minimised;

Be age−matched (study group and reference group);

Contain detailed presentations of how the study material was collected, including documentation of the examination technique and detailed presentations of any complementary investigations undertaken to exclude differential diagnoses;

Demonstrate that the observers of the symptoms and signs were blinded to (i.e. were unaware of) the suspected or alleged cause of the findings and describe how the blinding was achieved;

Present raw data, sensitivity/specificity and confidence intervals;

Be based on a material of adequate size and apply a uniform method of examination throughout;

Present a detailed account of the confession, what was confessed to and the circumstances under which the confession was made.

Project group, external reviewers, board and councils

Project group

Experts

SBU

The board of SBU has stipulated the conclusions in the report

Correspondence

SOFIA TRANÆUS (Sofia.Tranaeus@sbu.se)

Head of Department, Swedish Agency for Health Technology Assessment and Assessment of Social Services

Scientific reviewers

SBU engages external reviewers in its reports. The reviewers contributed valuable comments, which have improved the report. The final version of the report, however, may not include all the alterations or additions recommended by the external reviewers, partly because they were not always in agreement. Therefore, the external reviewers do not necessarily support all parts of the report.

The following external reviewers were engaged in this report and consented to inclusion of their names here:

Conflict of Interest

In accordance with SBU's requirements, the experts and scientific reviewers participating in this project have submitted statements about conflict of interest. These documents are available at SBU's secretariat. SBU has determined that the conditions described in the submissions are compatible with SBU's requirements for objectivity and impartiality.

Scientific Advisory Committee – Brage

SBU′s scientific advisory committee, Brage, has reviewed the material presented in this report.

Scientific Advisory Committee – Eira

SBU′s scientific advisory committee, Eira, has reviewed the material presented in this report.

SBU's Board of Directors

SBU's board of directors has approved the conclusions presented in this report.

External collaboration

Co‐operation with interested parties

The project management held meetings with members of the Swedish National Association for Families’ Rights at the beginning of the project to inform them of the project and to solicit opinions.

Networking with government authorities

In conjunction with the release of the report, representatives of the following authorities were invited to SBU for information about the results: the Ombudsman for Children in Sweden, the Swedish Health and Social Care Inspectorate, the Swedish Prison and Probation Service, the Swedish National Board of Health and Welfare, the Swedish Association of Local Authorities and Regions, the Swedish Police Authority, the Swedish National Board of Forensic Medicine and the Swedish Prosecution Authority.

The Swedish National Council on Medical Ethics

The analysis of the ethics was conducted by the Swedish National Council on Medical Ethics Appendix 3.

GÖRAN ELINDER Professor of Pediatrics, Karolinska Institutet, Stockholm, and Senior Consultant in Pediatrics	NIELS LYNØE Senior Professor of Medical Ethics, Karolinska Institutet, Stockholm, and specialist in Family Medicine
ANDERS ERIKSSON Professor of Forensic Medicine, Umeå University, and Senior Consultant in Forensic Medicine, the National Board of Forensic Medicine	PIA MALY SUNDGREN Professor of Diagnostic Radiology, Lund University, and Senior Consultant in Neuroradiology, Skåne University Hospital, Lund
BOUBOU HALLBERG PhD, Karolinska Institutet, Stockholm, and Senior Consultant in Pediatrics and Neonatology, Karolinska University Hospital, Stockholm	MÅNS ROSÉN Adjunct Professor of Health Technology Assessment, Karolinska Institutet, Stockholm
	BJÖRN‐ERIK ERLANDSSON External Collaborator, Biomechanical studies (Appendix 2), Professor of Medical Technology, Royal Institute of Technology, Stockholm

FRIDA MOWAFI Project Manager	HANNA OLOFSSON Information Specialist
MARIANNE HEIBERT‐ARNLIND Project Manager until September 2015	ANNA ATTERGREN GRANATH Project Administrator
IRENE EDEBERT Assistant Project Manager	LAURA LINTAMO Analyst
	ANNA BJÖRKLÖF Communication Responsible

STEVEN LUCAS a PhD, MD, Senior Consultant in Pediatrics, Uppsala University Children's Hospital, Sweden	NILS‐ERIC SAHLIN Professor of Medical Ethics, Lund University, Stockholm, Sweden
TIIT MATHIESEN Adjunct Professor of Neurosurgery, Karolinska Institutet, and Senior Consultant in Neurosurgery, Karolinska University Hospital, Stockholm, Sweden	ARNE STRAY‐PEDERSEN a Associate Professor, Institute of Clinical Medicine, University of Oslo, and Senior Consultant in Forensic Pathology and Clinical Forensic Medicine, Oslo University Hospital, Norway
TITTI MATTSSON Professor of Public Law, Lund University, Stockholm, Sweden	INGEMAR THIBLIN a Professor of Forensic Medicine, Uppsala University, and Senior Consultant in Forensic Medicine at the National Board of Forensic Medicine, Sweden

These reviewers have provided open statements – for more information contact SBU.

LARS HANSSON Chair, Professor, Care Science, Lund University	LENNART ISELIUS Associate Professor, Healthcare Director, County of Västmanland
CHRISTEL BAHTSEVANI Reg Nurse, Med Dr, Care Science, Malmö University	MUSSIE MSGHINA Associate Professor, Senior Consultant, Psychiatry, Karolinska University Hospital
PER CARLSSON Professor, Health Economics, Linköping University	LARS SANDMAN Professor, Ethics, Borås University
BJÖRN‐ERIK ERLANDSSON Professor, Medical Technology, KTH Royal Institute of Technology, Stockholm	BRITT‐MARIE STÅLNACKE Professor, Senior Consultant, Rehabilitation medicine, Umeå University
ARNE GERDNER Professor, Social work, School of Health and Welfare, Jönköping	SVANTE TWETMAN Professor, Dental Care, Halmstad and University of Copenhagen, Denmark

KJELL ASPLUND Chair, Professor, Stockholm	KRISTINA BENGTSSON BOSTRÖM Associate Professor, Billingens Medical Center, Skövde
HENRIK ANDERSHED Professor of Psychology, Associate Professor of Criminology, Örebro University	CHRISTINA BERGH Professor, Department of Obstetrics and Gynecology, SU/Sahlgrenska, Göteborg
ANNA EHRENBERG Professor of Care Science/Nursing, Dalarna University, Falun	YLVA NILSAGÅRD Associate Professor, Physical therapy, CAMTÖ, Region Örebro County
INGEMAR ENGSTRÖM Professor of Child and Adolescent Psychiatry, Ethics, Örebro University	STEN‐ÅKE STENBERG Professor of Sociology, Stockholm University
NILS FELTELIUS Associate Professor, Medical Products Agency	KATARINA STEEN CARLSSON PhD, Swedish Institute for Health Economics, Lund

NINA REHNQVIST Chair of the Board, professor, Karolinska Institutet	LARS‐TORSTEN LARSSON Associate professor, Director of department, National Board of Health and Welfare
SUSANNA AXELSSON Director General, SBU	STEFAN LINDGREN Professor, Lund University, University Hospital MAS, Chair of the Swedish Society of Medicine
HEIKE ERKERS Chair, Union for Professionals	STIG NYMAN Chair, Swedish Association of Health Professionals
EVA FRANZÉN Director of Research and Development, Swedish National Board of Institutional Care	SINEVA RIBEIRO President, Swedish Association of Health Professionals
ÅSA FURÉN‐THULIN Head of Division, Swedish Association of Local Authorities and Regions	AGNETA VON SCHOTING Chair, NSK−S
JAN‐INGVAR JÖNSSON Secretary General, Swedish Research Council	HEIDI STENSMYREN Chair, Swedish Medical Association
BJÖRN KLINGE Professor, Karolinska Institutet and Malmö University	ANDERS SYLVAN Director, County Council of Västerbotten
HÅKAN SÖRMAN Director General, The Swedish Association of Local Authorities and Regions	KARIN TENGVALD Professor emerita, Linköping University

Table A1

Other possible causes (differential diagnosis) of the triad and its components

Disease/condition	Reported findings from the triad	Reference number (number of cases, or cases/study population size) Reported finding from the triad
Diseases or conditions causing haemorrhagic symptoms
von Willebrand´s disease	SDH, RH	78 (1)
Delta storage pool disease	SDH, BE, RH	79 (1)
Hyperfibrinogenemia	RH (including vitreous haemorrhage)	80 (1)
Haemophilia A	SDH/RH	81 (2)RH 82 (1)SDH
Factor X deficiency	SDH	83 (2)
Idiopathic thrombocytopenic purpura	ICH	84 (1)
Kasabach–Merrit syndrome thrombocytopenia	RH	85 (1)
Hepatitis	RH, BE, SDH	86 (1)
Albers–Schönberg disease	SDH	87 (1)
Vitamin K deficiency	SDH (ICH)/BE/RH	88 (3) SDH 89 (17) SDH 90 (1) SDH 91 (16) SDH 92 (1) SDH, BE, RH 93 (1) SDH, BE
Menkes disease (Copper deficiency)	SDH	94 (1)
Unspecified	RH, SDH	95 (1) 96 (3)
Infections
Infection	RH	97 (4)
Infection with or without hypoxia	SDH (intradural bleeding)	21 (10/30)
Vascular malformations
Aneurysm, Arterio‐venous malformation	SDH (SAH)/BE/RH	98 (1) SDH 99 (1) SDH, BE 100 (1) SDH, BE 101 (1) SDH, RH
Prenatal and birth‐related injuries
Prematurity	RH	102 (11)
Delivery injury	SDH (ICH)/RH	18 (2) SDH 103 (3) ICH 96 (3) SDH 56 (53) RH 104 (10) RH
Normal delivery (or prenatal)	SDH/RH	77 (17/97) SDH 39 (94/252) RH 76 (32/63) SDH
Prenatal trauma	RH	105 (2)
Congenital SDH	SDH	106 (1)
Congenital heart disease	SDH	22 (66/152)
Large head size
Enlarged SA space/	SDH	31 (6/108)
External hydrocephalus/		32 (4/177)
Benign enlargement of the subarachnoid spaces		107 (7)
		33 (3)
		96 (6)
External hydrocephalus	RH, SDH	108 (1)
		109 (6)
Metabolic diseases
Glutamic aciduria	SDH/RH	110 (1) SDH 111 (1) SDH 112 (1) SDH, RH
Immunological diseases
Hemo‐phagocytic lymphohistiocytosis	SDH, SAH (ICH), BE, RH	113 (1)
Transplacental acquisition of anti‐Ro antibodies	SDH	114 (2)
Skeletal diseases
Osteogenesis imperfecta	SDH, RH	115 (3)
Brittle bone disease	SDH/RH	116 (20/20) SDH (11/20) RH
Other
Hypoxia and resuscitation	RH	117 (1/33)
		118 (1)
Hypoxia	SDH/intradural bleeding	21 (20/30)
Choking and resuscitation	SDH (SAH), RH	119 (1)
Resuscitation in patients with retinopathy of prematurity	RH	118 (2)
		117 (1/33)
Hypernatremia and dehydration	ICH, BE	120 (1)
Leukemia	RH	95 (3)
Vaccine‐induced vitamin C deficiency	SDH, BE	121 (2)

BE = Brain oedema; ICH = Intracranial haemorrhage; RH = Retinal haemorrhage; SAH = Subarachnoid haemorrhage; SDH = Subdural haematoma.