External hydrocephalus as a cause of infant subdural haematoma; epidemiological and radiological investigations of infants suspected of being abused

Background Acute and chronic subdural haematomas (ASDH/CSDH) in infants have been regarded as highly specific for abuse. Other causes of CSDH have not been investigated in a large population. Purpose To investigate to what extent external hydrocephalus is present in infants with ASDH and CSDH undergoing evaluation for abuse. Material and methods Eighty-five infants suspected of being abused, with ASDH (n=16) or CSDH (n=69), were reviewed regarding age, risk factor profiles, cranio-cortical width (CCW), sino-cortical width (SCW), frontal interhemispheric width (IHW), subarachnoid space width (SSW) and head circumference (HC). In infants with unilateral SDH, correlations between contralateral SSW and ipsilateral CCW and SDH width were investigated. Results Infants with CSDH had significantly lower mortality, were more often premature and male and had significantly higher CCW, SCW, IHW and SSW than infants with ASDH (p < 0.05). Ipsilateral CCW (R = 0.92, p < 0.001) and SDH width (R = 0.81, p < 0.01) correlated with contralateral SSW. Increased HC was more prevalent in infants with CSDH (71%) than in infants with ASDH (14%) (p < 0.01). Forty-two infants, all with CSDH, had at least one of CCW, SCW or IHW ≥ 95th percentile. Twenty infants, all with CSDH, had CCW, SCW and IHW > 5 mm, in addition to increased HC. Conclusion A substantial proportion of infants with CSDH who had been suspected of being abused had findings suggesting external hydrocephalus.


Introduction
Subdural haematoma (SDH) is encountered more frequently in infants than in older children [1]. A Swedish national registry study found an incidence of 16.5 infants with SDH per 100,000 infants, with a mean age of 3.3 months, median age of 2.5 months, and a male preponderance for all SDH subgroups [2]. One limitation of that study was that the International Statistical Classification for Diseases -Tenth Revision (ICD-10) does not differentiate between acute subdural haematoma (ASDH) and chronic or mixed SDH/hygroma (CSDH). No study on a large population has investigated the possibility that infants referred with suspected abusive head trauma (AHT) could be suffering from external hydrocephalus (benign external hydrocephalus (BEH), benign enlargement of subarachnoid space (BESS), macrocephaly etc.) complicated by a CSDH [3][4][5][6][7]. External hydrocephalus has a marked male preponderance [8] and prematurity is relatively frequent [9]; the condition has also been suggested as a possible pitfall in the diagnosis of AHT [4, [10][11][12].

Purpose
It is reasonable to expect that a traumatic event will cause an ASDH, however there are also rare non-traumatic causes such as bleeding disorders, central venous thrombosis, vessel malformations or genetic diseases. The non-traumatic differential diagnoses are to a large degree possible to confirm by thorough medical examination, although small vessel malformations can be missed in neuroimaging, and central venous thromboses have a range of aetiologies [13,14]. CSDH, on the other hand, has multiple possible aetiologies that may be hard to identify at the time of detection, e.g., birth [15,16] or BEH [5,6,10,12,17].
Based on earlier studies [2,4,5,12,17,18], we hypothesized that BEH would be present more often in infants with CSDH than in ASDH. In Sweden, such cases would have been referred to the National Board of Forensic Medicine due to suspected AHT. The aim of the present study was thus to investigate the proportion of BEH in infants with ASDH and CSDH undergoing evaluation for abuse.

Material and methods
The present study is a descriptive review of a national series of infants aged younger than one year, with an SDH or subdural hygroma, who had been subjected to forensic J o u r n a l P r e -p r o o f investigation due the presence of SDH without an obvious cause. The presence or nonpresence of other medical findings or history of trauma were not considered for inclusion/exclusion. SDHs due to obstetric or neonatal malpractice, traffic accidents, multistory falls or similar high energy trauma were not included. The cases were retrieved from the computerised register of the Swedish National Board of Forensic Medicine and had been registered during the period January 1, 1994 to December 12, 2018.
A total of 1,380 infants were identified from that period, of whom 497 had been subjected to clinical medico-legal investigation and 883 had died and had therefore undergone an autopsy.
In this total, 1,249 infants did not have any SDH or had an SDH as a result of an obvious perinatal or obstetrical complication, a traffic accident or a multi-story fall. Among the remaining 130 infants, neuroimaging at the time of diagnosis was available in 96 cases and 85 of those had subdural fluid that could be classified as ASDH or CSDH. These 85 cases comprised the study population. See the flowchart in Figure 1 for an overview of the database.
Our data consists of information available from the aforementioned register and included forensic reports and, to a varying extent, hospital records and birth records. Some infants had other findings such as skull fracture, extracranial fractures, retinal haemorrhages, subarachnoid haemorrhage, cerebral venous thrombosis, hypoxic ischemic injury or other parenchymal injury. Our aim in the present study was however only to study findings related to BEH on a group level and not to finally conclude whether an infant had suffered abuse or a spontaneous SDH.

Neuroimaging procedure
Neuroimaging was collected from the respective hospitals and interpreted by a specialist in neuroradiology (JW). Both magnetic resonance imaging (MRI) and computed tomography (CT) were available in 65 cases, only CT in 27 cases, and only MRI in four cases. Among the cases which had a confirmed SDH on the neuroimaging (n=85), both MRI and CT were available in 61 cases, only CT in 20 cases and only MRI in four cases.

J o u r n a l P r e -p r o o f
Cases were divided into two groups for later analysis of risk factors and neuroimaging comparison. Where CT showed subdural fluid predominantly hyperattenuating compared with adjacent brain parenchyma, the haematoma was defined as an ASDH, while a haematoma with predominantly iso-/hypodensity was considered to be a CSDH. In cases without CT, T1-and T2-weighted sequences on MRI were used to assess the type of SDH, where signal intensity similar to cerebrospinal fluid was interpreted as a sign of a CSDH.
Cranio-cortical width (CCW), sino-cortical width (SCW) and frontal interhemispheric width (IHW) are possible to measure on either CT or MRI. While subarachnoid space width (SSW) could be measured on MRI in most cases, this was possible on CT in only a few cases, as the subarachnoid and subdural spaces are more difficult to distinguish. If the MRI was performed more than four weeks after the CT (n=4), we did not measure SSW on MRI, as the MRI could not be considered representative for the measurements at the time of diagnosis.
Risk profiles were analysed with regard to age at diagnosis, sex, prematurity, multiple birth and death; differences were estimated with Fisher's exact tests and proportion tests.
Infants were reviewed with regard to available HC charts, HC at the time of diagnosis as measured on CT, and radiological characteristics of BEH, such as left and right measures of CCW, SCW, SSW, in addition to IHW. All measurements were made in millimetres at the level of the foramen of Monro. Wilcoxon's signed-rank test was used to calculate differences of means. Spearman's rank-order correlation was used to estimate a correlation between contralateral SSW and ipsilateral CCW and SDH width (CCW-SSW) in infants with unilateral SDH located at the level of the foramen of Monro (n=10), to investigate the possibility that the SDH was simply replacing a pre-existing widened subarachnoid space.

Increased head circumference
In Sweden, HC is routinely measured during infancy at child care centres and HC charts are plotted with standard deviations (SDs). For infants with HC data from growth charts or with repeated measurements of HC, data were plotted manually and the infants were subsequently assigned to one of four categories: 1) normal HC; 2) rapidly increasing HC exceeding two SDs from normal range; 3) rapidly increasing HC exceeding two SDs from J o u r n a l P r e -p r o o f normal range and an HC ≥ 3 SDs from normal range; 4) no rapid increase in HC, but a large head from birth -defined as an HC ≥ 3 SDs from normal range.
Infants with increased HC and ASDH or CSDH, categories 2-4, were compared to infants with normal HC and ASDH or CSDH, category 1, using Fisher's exact test.
HC was also measured on the initial axial CT scan at the level of the foramen of Monro.
SD for males and females were imported into R from the World Health Organization's website (https://www.who.int/childgrowth/standards/hc_for_age/en/) and the HC of each infant was inserted. Data for premature infants were age-adjusted.

BEH
BEH is usually defined as a rapid increase of HC in combination with increased SSW; however, it is difficult to find a distinct, generally accepted radiological definition [8]. Lam et al. [19] have suggested to use CCW, SCW and IHW as indicative measures of BEH, while also taking age into account. They created a function which can be used to estimate normal values. In their definition, infants with CCW, SCW or IHW ≥ 95 th percentile should be viewed as possibly having BEH.
A plot was created based on the functions for CCW, SCW and IHW (Table 1)

J o u r n a l P r e -p r o o f
Hussain et al. did not find a correlation between SSW and age and described an overlap between infants with and without BEH [20]. They recommended that infants who did not have clinical features of BEH should be considered to belong to the normal population.
We therefore used a definition in which all of the measurements of CCW, SCW and IHW had to be above 5 mm and HC had to be increased for BEH to be indicated. The differences were compared using Fisher's exact test.
Assuming that CSDH could be a complication of BEH and that ASDH to a greater extent would be of traumatic origin, one would logically expect that risk factor profiles and CCW, SCW, IHW and SSW would differ substantially between the CSDH and the ASDH groups.
The correlation between the ipsilateral CCW and SDH width and contralateral SSW in infants with unilateral SDH was used to investigate the hypothesis that the SDH only filled a preexisting space in infants with increased SSW.

Software
The R studio packages tidyverse, ggpubr and rstatix were used for descriptive statistics and statistical computations.

Ethical approval
This study was approved by the Regional Ethical Review Board in Uppsala 2015/039 and 2015/040.

Results
It was possible to determine the type of subdural fluid in 85 of 96 cases. Sixteen infants had ASDH (19%) and 69 had CSDH (81%). Nine cases had been diagnosed with an SDH, but no subdural fluid could be identified at the re-assessment. In two cases, it was not possible to assess whether the fluid was subdural or subarachnoid (only CT imaging was available in these cases).

Risk factor profiles
J o u r n a l P r e -p r o o f There were marked differences between ASDH and CSDH regarding prematurity (p < 0.05more common in CSDH) and mortality (p < 0.001 -higher in ASDH) and a numerical difference for being male (p = 0.39). When comparing sex distribution to an expected frequency of 50 %, infants with CSDH, but not ASDH, had a higher proportion of male infants (p > 0.01) see Table 2.
There was a significantly higher proportion of male infants with CSDH (68%, p > 0.01) compared with an expected frequency of 50%; this was not seen in infants with ASDH (56%, Table 1 for test statistics. Most infants with ASDH were identified during the first month of life (0-30 days), with mean and median ages of 2.6 and 2.0 months, respectively, whereas infants with CSDH were most often identified during the third month of life, with mean and median ages of 3.0 and 2.5 months, respectively. See Figure 2 for age distribution details for the two groups.
Fifty-six percent of the infants with ASDH were males, one (6%) was born preterm (week 31), 13% had a twin, and the mortality was high, at 44%.
See Figure 2 for age distribution among infants with ASDH and CSDH and Table 2 for risk profile comparison.

Increased head circumference
HC charts or repeated HC measurements were available for 53 infants. It was significantly more likely for infants with CSDH (71%) than for infants with ASDH (14%) to have an increased HC (p < 0.01). HCs measured on initial CTs can be seen in Figure 3.

Neuroimaging measurements and correlation
Measurements for CCW, SCW and IHW were obtained in all 85 cases. SSW was possible to measure in 68 cases (CSDH = 57, ASDH = 11). Infants with CSDH had significantly higher CCW, SCW, IHW and SSW than infants with ASDH, see Figure 4. In infants with unilateral SDH there were significant correlations between contralateral SSW and ipsilateral CCW (p < 0.001) and SDH width (p < 0.01), respectively, which can be seen in Figure 5.

CCW, SCW and IHW ≥ 95 th percentile
No infants with ASDH had a CCW, SCW or IHW ≥ 95 th percentile, as suggested by Lam et al. [19], but 42 infants with CSDH (49% of all infants, 60% of the infants with CSDH) had at least one measurement which was ≥ 95 th percentile. See Table 3 and Figure 6.

Discussion
The aim of the present study was to investigate the hypothesis that BEH would be more prevalent in infants with CSDH than ASDH. The empirical consequences would be different risk profiles, HC and neuroimaging characteristics of infants with CSDH and ASDH who were referred with suspected AHT.
Infants with CSDH were more likely to be premature, had a lower mortality and were more often male than infants with ASDH. These observations indicate that infants with CSDH and ASDH should be viewed as separate groups.
There was a clear difference between infants with CSDH and ASDH regarding HC, CCW, SCW, IHW and SSW, indicating that substantial proportion of infants with CSDH had findings suggestive of BEH.
SDH is known to be a complication of BEH. It is believed that the expansion of the subarachnoid space stretches the bridging veins which extend from the cortex to the dural sinuses. The stretching of the veins may cause spontaneous bleedings [3][4][5][6] or bleedings J o u r n a l P r e -p r o o f following minor trauma [11,21,22]. In the present study, SSW was significantly higher for infants with CSDH than for infants with ASDH.
Measures of CCW, SCW and IHW are affected by ipsilateral SDH thickness, and it might be argued that the differences between ASDH and CSDH infants in this regard could be explained by average differences in SDH thicknesses, with CSDH being thicker than ASDH.
However, the correlation between the contralateral SSW and ipsilateral CCW and SDH width in cases of unilateral CSDH indicates that the SDH replaced a pre-existing expanded subarachnoid space without causing midline shift or cortical or ventricular compression, see Figure 7. A similar presentation can be seen in Figure 8 in an infant who had a birth-related SDH and wide subarachnoid spaces which developed into a unilateral CSDH at the age of 27 days, as described in the study by Rooks et al. [15] and also discussed by Gabaeff [16].
Moreover, the greater observed SSW in CSDH compared with ASDH cannot be explained by thicker CSDH.
It is possible that repetitive rebleeding into a CSDH/hygroma may cause the CSDH, and thus the HC, to grow. Several mechanisms have been described to account for rebleeding into a CSDH, perhaps such a mechanism explains why there were no infants having a purely acute SDH combined with markers of BEH [23,24].

Clinical implications
In addition to the standard protocol of social service investigation, clinical examination and full skeletal survey, for all infants that are diagnosed with an SDH, the investigators should have head circumference charts available and measure CCW, SCW, IHW and SSW in order to not overlook a possible BEH diagnosis.

Strengths and limitations
This study covers all cases referred to the Swedish National Board of Forensic Medicine for suspected AHT during a period of nearly 25 years. This authority manages all medico-legal death investigations in Sweden. Therefore, this study can be regarded as based on the entire population of infants with SDH suspected to be caused by AHT.

J o u r n a l P r e -p r o o f
This study did not aim to conclude finally whether an individual infant has been abused or did suffer from spontaneous SDH or SDH from minor trauma. A control group consisting of infants with known trauma and CSDH and known non-traumatic CSDH was not possible to construct from the present database. Thus, the true positive, true negative rate cannot be ascertained from the study design. A prospective study would be of importance to understand the association between BEH and CSDH and to investigate the possibility of a causation.
Non-deceased infants with physical findings that have led to concern of possible abuse are probably referred to this authority in most instances, but it cannot be completely ruled out that some non-deceased infants were evaluated by clinicians only. However, such a selection is unlikely to result in any kind of systematic bias. We gained access to neuroimaging in 96/131 cases of which 85 had an SDH; this lack of information was mainly due to some cases being from before computerisation of radiology registers; thus, it is unlikely that this caused a skewed selection.
A limitation of the present study was the lack of consensus regarding what criteria should be used to diagnose BEH when using neuroimaging measurements without other clinical signs.
In this study, two different definitions on neuroimaging measurements were used, one with comparison to normal values and one with measurements in combination with information on clinically increased HC. These features were present only for infants with CSDH, regardless of which definition was used. The study by Lam et al [19], which was used to define normal values for these measurements, was performed on Chinese infants, and it is possible that the mainly Caucasian population in the present study had other normal values.

Conclusion
A substantial proportion of infants with CSDH who had been suspected of being abused had findings suggesting external hydrocephalus.

Conflict of interest
Jacob Andersson has no conflicts of interest to declare.      Correlation between contralateral SSW and ipsilateral CCW and SDH width in infants with unilateral SDH located at the foramen of Monro (n = 10). There was a significant correlation between contralateral SSW and ipsilateral CCW (R = 0.92, p < 0.001) and SDH width (R = 0.8, p < 0.01).

Figure 6.
Forty-two infants with CSDH but no infants with ASDH had one or more of CCW, SCW and IHW ≥ 95 th percentile. The thick grey lines indicate means and the dotted lines show 95 th and 5 th percentiles plotted based on the calculations for healthy infants by Lam et al. seen in Table 1.

Figure 7.
A: MRI of infant in the present study with unilateral CSDH with wide subarachnoid space contralaterally to the SDH. Note that there still is a subarachnoid space (marked "SSW"), and that the CCW is the same on both sides. B: MRI of infant in the present study with bilateral CSDH with wide CSDH and visible subarachnoid space.