cause of retinal hemorrhage which when present is usually mild [54]. The history is usually obvious and characteristic findings, such as orbital fracture, multiple skull fracture, basilar skull fracture, and cranial nerve palsy should be present.

Both optic nerve sheath and intraocular hemorrhages are frequently reported findings in postmortem examinations of AHT victims. Optic nerve sheath hemorrhages frequently involve multiple layers, but often show a preponderance of hemorrhage in the subdural space [54, 55]. Other evidence of acceleration– deceleration induced orbital trauma may include optic nerve intradural hemorrhage, and hemorrhage in the orbital fat, cranial nerve sheaths, or extraocular muscles [56]. Postmortem clinical examination may be useful to help document the pattern of hemorrhages. Protocols for forensic ophthalmic autopsy are available and should include gross photography of the open fixed globe, orbital exenteration (preferably with the globe en bloc), as well as microscopic examination [57]. The role of hemosiderin in dating retinal hemorrhages remains unclear.

Documentation is an integral part of the evaluation of a case of possible AHT. Many examiners may use quickly crafted, hand-made drawings that may not adequately reflect the number, distribution, and types of retinal hemorrhages. Detailed descriptions preferably using both words and careful drawings are more useful. The value of ocular photography has been recognized almost as far back as its availability and it is perhaps the gold standard for documenting retinal hemorrhages, although it requires expensive equipment not readily available in all centers. The inability to perform photography either because of the lack of availability of equipment or assistance should not be considered a flaw in the medical/forensic evaluation of the child as long as the hemorrhages are well detailed by other forms of documentation. Even when the equipment is available, retinal photography in an awake child can be quite difficult. There are reports of contact retinal photography [58] or examination without photography [59] causing scattered retinal hemorrhages in premature children being evaluated as a part of routine surveillance for retinopathy of prematurity, but these situations are clearly recognizable and should not discourage the use of retinal photography in suspected AHT. Several eye findings seem to have prognostic significance. The extent of intraocular hemorrhage, the presence of macular

retinoschisis lesions, and the presence of pupillary abnormalities at presentation have been correlated with fatal outcome and permanent neurological impairment [49, 60–63]. The correlation between the severity of ocular injury and neurological outcome suggests a relationship between the mechanism of brain and ocular injury in AHT.

17.8  Dating of Retinal Hemorrhages

The age of intraocular hemorrhage is very difficult to assess clinically. It has been assumed that the hemorrhages occur immediately at the time of injury although there are reports of critically ill children initially found to have unilateral hemorrhages advancing to bilateral hemorrhages or an increase in severity over the early days of hospitalization [64]. Some evolution, including darkening of the retinal hemorrhages, organization of vitreous hemorrhages, and disappearance of the retinal hemorrhages occurs over 2–4 weeks following the acute injury, although there is widespread variability in the time it takes for retinal hemorrhages of all types to disappear. Even extensive intraretinal hemorrhages may resolve within the first few days following injury, whereas preretinal and macular schisis hemorrhage may sometimes remain for more than a month. Retinal hemorrhages in living children cannot be used to time and date the injury except at the end of the spectrums: fading hemorrhages are unlikely to have occurred within the last few hours and widespread intraretinal hemorrhages are unlikely to last for weeks or months. There is some evidence for a delay of 1–2 days between injury and the presence of vitreous hemorrhages although cases of acute onset vitreous hemorrhage at the time of injury are recognized [65].

17.9  Treatment of Retinal Hemorrhages

The management of acute intraocular hemorrhage is primarily supportive. Gradual resolution is generally seen usually without significant retinal or visual sequelae. With prolonged vitreous hemorrhage, intrafoveal hemorrhage, or hemorrhage in front of the fovea, young children do have a risk of deprivational amblyopia. In

these cases, patching the better eye may be needed to treat amblyopia after the hemorrhage resolves. Surgical intervention to remove vitreous blood or blood within a traumatic schisis cavity is controversial and may be difficult, but it may be helpful in some cases. The ophthalmologist is sometimes faced with a difficult dilemma: choosing between waiting for spontaneous resolution with the risk of amblyopia, myopia, and other vision-threatening issues vs. surgical intervention with risks of surgery such as cataracts and retinal tears. In our experience, surgery is rarely indicated unless significant nonresolving vitreous hemorrhage is present though the threshold for intervention should be earlier in very young children. The visual prognosis for children with untreated macular schisis is remarkably good as the cavity usually clears with few foveal changes. Although the perimacular folds persist (and indicate prior injury), they rarely involve the fovea.

17.10  Late Ophthalmic Findings

In contrast to the dramatic and relatively specific acute findings, late changes associated with AHT are neither consistent nor specific to AHT except for perimacular cirumlinear hypopigmented lines or folds which have a high likelihood of indicating prior AHT. Survivors should be reexamined for amblyopia, refractive errors, and other late complications that require treatment. Permanent visual impairment is frequent and cortical visual impairment secondary to hypoxic ischemic brain injury, infarction, or occipital lobe contusion is the most common cause followed by optic atrophy [66].

17.11  Differential Diagnosis of Retinal

Hemorrhages in Childhood

Perhaps, the most important obstacle to understanding the cause of retinal hemorrhages is the failure to accurately describe them. Assertions that retinal hemorrhages are or are not the result of child abuse may often be clarified with descriptions that will offer more specificity. Retinal hemorrhages should be described in terms of where they lie topographically, the layers at which they lie, their pattern (e.g., perivascular), and their number. Intraretinal and preretinal hemorrhages

may appear to have white centers. However, while this appearance is classically associated with bacterial endocarditis (Roth spot), virtually, any cause of retinal hemorrhage may be associated with white-centered hemorrhages. Depending on the underlying cause of hemorrhagic retinopathy, the white center may represent central clearing, embolus, leukemic infiltrate, or even reflection from the examining illumination [67]. Traumatic retinoschisis has particular diagnostic significance in recognizing AHT.

Papilledema occurs in less than 10% of cases of AHT [68]. Retinal hemorrhages are often one of the distinguishing signs of papilledema though these hemorrhages tend to be small, few in number, flameshaped, and located in a radiating pattern around the obviously swollen optic nerve. These hemorrhages do not necessarily indicate trauma and can be seen in any cause of papilledema. It is perhaps incorrect to include these hemorrhages in the category of retinal hemorrhages when discussing AHT as an etiology. Other retinal hemorrhages from elevated intracranial pressure in the absence of shaking appear to be very rare. When they do occur, there are characteristically a few intraretinal or preretinal hemorrhages in the posterior pole, particularly on or around the optic nerve.

Retinal hemorrhages of newborns, likely related to obstetrical and perinatal hemodynamic changes, and perhaps most importantly, peripartum prostaglandin release, are the most common cause of retinal hemorrhage in children [14]. They have been extensively studied both retrospectively and prospectively in tens of thousands of babies with most authors finding an incidence of 20–30% if evaluated in the first 24 h of life and 10–15% if evaluated before 72 h [69]. From this data, it can be concluded that superficial retinal hemorrhages resolve by 1 week (usually in less than 3 days) postpartum and deeper retinal hemorrhages resolve by 6 weeks (usually in less than 2–3 weeks). Intrafoveal, preretinal, and vitreous hemorrhage following birth may last longer.

Retinal hemorrhages have also been reported in association with severe accidental injury. Multiple clinical and postmortem studies of eyes in patients with severe head injury suggest that the rate of retinal hemorrhage is less than 3% [70, 71]. In most studies, the reported incidence is zero, particularly when one considers short falls. When retinal hemorrhages do occur following accidental injury, the injury history is usually obvious, the clinical condition of life-threatening and the retinal hemorrhages typically are confined to

the posterior pole, few in number, and rarely subretinal. More severe retinal hemorrhages, even with extension to the retinal periphery, can be seen in up to 14% of children involved with severe motor vehicle accidents, usually characterized by multiple acceleration– deceleration events such as vehicle roll-overs [48, 72].

The literature is replete with case reports of medical diseases that have been misdiagnosed as child abuse. There are a wide number of systemic and ocular ­conditions, which may be associated with retinal hemorrhages, although the absence of supportive findings on ocular examination, physical examination, history, or laboratory evaluation makes their consideration often equivocal. The incidence of retinal hemorrhages in children with the following conditions is known to be very low if at all possible and characterized by hemorrhages that are few in number and confined to the posterior pole or with other easily recognizable unique ocular and/or systemic features. Most of these entities are readily excluded from the differential diagnosis on the basis of history and or physical examination. It must also be remembered that these diagnoses do not render a child “immune” to any form of child abuse. The presence of retinal hemorrhage more characteristic of AHT, even in a child with a coexisting condition that may be associated with retinal hemorrhage, must lead the ophthalmologist to consider the possibility that the child is also a victim of AHT. Coagulopathies and other bleeding disorders, including thrombocytopenia, severe anemia, leukemia, factor deficiencies, and vitamin K deficiency, should be considered in the differential diagnosis of intraocular hemorrhage in infants. In general, retinal hemorrhages related to hematologic abnormalities are less numerous and less extensive and usually do not extend peripherally in the retina with the exception of leukemia which would easily be recognized by a routine blood count. Although brain injury itself may result in a mild coagulopathy, even with disseminated intravascular coagulation and other more severe coagulopathies, retinal hemorrhage is very uncommon even in the setting of trauma. Ocular syndromes associated with retinal hemorrhage in childhood, including Norrie disease, Coats disease, persistent hyperplastic primary vitreous, hypotonous retinopathy, CMV retinitis, toxoplasmosis, and retinopathy of prematurity, are easily distinguished from nonaccidental head injury by the distinctive clinical appearance as well as the clinical setting. Vitreous hemorrhage may also be the result of juvenile X-linked retinoschisis or protein C/S deficiency. But the typical macular retinoschisis

lesion of AHT has not been reported in these conditions and their diagnosis is clear by examination, family history, and/or laboratory evaluation.

Basic hematological evaluation should be performed in all suspected cases of AHT although one could argue that if fractures or a clear history of abusive injury are present, systemic workup is not necessary. In the ­setting of suspected abuse, guidelines for appropriate laboratory screening do not exist and screening practices vary considerably. Universal recommendations included complete blood count, platelet count, PT, and PTT (or INR). Additional DIC evaluation including a fibrinogen, D-primer, von Willebrand factor, and platelet function tests are used in some centers. Additional testing should reflect the results of an initial screening test in search for diseases that are known to cause intracranial and retinal findings. In reality, these include only a few hematological disorders. However, physicians often embark on unnecessary medical evaluations to eliminate the remote possibility of alternate explanations in cases of child abuse.

Glutaric aciduria is an autosomal recessive metabolic disorder, which is sometimes associated with subdural hemorrhage after minor head trauma. Preexisting macrocephaly is a hallmark of the condition [73]. Subdural hematoma is thought to be a result of macrocephaly placing bridging veins on stretch thus making them more prone to shearing forces induced by mild head injury. Although this disorder eventually results in serious neurological compromise, affected children may have normal development in early childhood. Early neurological signs are often subtle. Characteristic basal ganglia disease can develop even in the absence of significant changes in electrolytes or glutaric acid concentrations and the neuroimaging findings help to distinguish this disorder from AHT [73]. Retinal hemorrhages may occur, but are usually no more than a few preor intraretinal hemorrhages confined to the posterior pole. Many infants with severe abusive head injury have cardiopulmonary resuscitation including chest compressions and artificial ventilation. Retinal hemorrhages have rarely been reported after prolonged cardiopulmonary resuscitation, but never as numerously or extensively as seen in AHT [74–77]. From case reports and prospective studies, it can be concluded that retinal hemorrhages only rarely, if at all, occur from cardiopulmonary resuscitation and when they do, they are few in number and confined to the posterior pole [78].

Purtscher retinopathy may occur following severe­ acute compression injuries to the thorax with characteristic manifestations including white retinal patches, retinal hemorrhages, and retinal edema most commonly surrounding the optic disk. A similar retinopathy can be seen in the setting of pancreatitis [67]. Although Purtscher retinopathy is rarely seen in AHT, presumably due to the thorax compression which can be severe enough to fracture ribs, it is not a characteristic finding [79]. The lack of evidence to support cardiopulmonary resuscitation with chest compression as a cause of retinal hemorrhage is also supported by the absence of a single reported case of Purtscher retinopathy following resuscitation. Terson syndrome – retinal and/or vitreous hemorrhages associated with intracranial hemorrhage – is well recognized in adults. The lack of correlation­ between the side of involvement of the subarachnoid hemorrhage and ocular hemorrhage suggests that this is not a sufficient explanation for the retinal hemorrhages seen in AHT as the pathophysiology of Terson syndrome is presumed by some to be related to the direct tracking of blood from the intracranial space into the optic nerve sheath [20]. One report did find only unilateral retinal hemorrhages ipsilateral to the unilateral subdural hemorrhages in 8 patients reported, although four of the eight patients sustained direct trauma to or around the involved eye leaving only four with clear unilateral retinal hemorrhages ipsilateral to the unilateral subdural hemorrhages [80]. Retinal hemorrhage has been found to be uncommon in children with intracranial hemorrhage due to causes other than AHT [20]. Even when Terson syndrome appears to occur in children, the retinal hemorrhages are not in the pattern or quantity usually seen in AHT.

Sudden infant death syndrome (SIDS) is the most common nonabusive cause of death between 1 and 12 months of age. Children who die from SIDS may share several features in common with those who are victims of AHT including the absence of explanatory history and the possibility of a prodromal syndrome such as gastrointestinal complaints. However, SIDS victims, by definition, have no evidence of trauma by history, clinical examination, or autopsy. The American Academy of Pediatrics states that it is uncommon for death due to child abuse to be confused with SIDS and their position statement lists retinal hemorrhages as an exclusion factor for SIDS [81].

Ruptured aneurysm is a well-recognized cause of retinal hemorrhages in adults, although this is a very

rare occurrence in children. Fahmy found no retinal hemorrhages in individuals less than 20 years of age with ruptured aneurysms although the number of studied children was less than 5 [82]. In another study of both adults and children, only 19% had retinal hemorrhages although many of the retinal hemorrhages were probably due to papilledema and did not have a pattern of retinal hemorrhages consistent with those seen in AHT [83]. Most importantly, the diagnosis of aneurysm is usually readily made by neuroimaging or autopsy.

There is no evidence to support a link between childhood immunizations and retinal hemorrhages in children [84, 85]. Normal play activities such as parents jostling their infants, jogging with them in backpacks, or throwing them in the air are not associated with intracranial or ocular hemorrhage despite the initial assumptions by Caffey that this might occur [2]. Coughing and vomiting do not cause a hemorrhagic retinopathy in children [86]. Although Geddes et al. suggested that hypoxia alone could result in intracranial and retinal hemorrhage, there is no clinical evidence to support this conclusion as the authors did not directly study ocular findings, and she retracted much of their own work under oath in United Kingdom court [87]. Sickle cell disease, hypertension, and diabetes mellitus, perhaps the most common causes of retinal hemorrhage in adults, do not cause hemorrhagic retinopathy in the AHT age range. This is a further evidence of the unique status of the pediatric retinal vasculature that should lead to caution when applying adult literature to children.

There are additional conditions, which may be associated with retinal hemorrhages including but not limited to arachnoid cysts [88], arteriovenous malformations [89], endocarditis, hemolytic uremic syndrome, meningitis, and vasculitis. Once again, the incidence of retinal hemorrhages in children, especially those in the AHT age range, with these conditions is known to be rare if at all possible and characterized by retinal hemorrhages that are few in number and confined to the posterior pole or with other recognizable unique ocular and systemic features.

17.12  Pathophysiology of Retinal

Hemorrhages

Many theories as to the cause of retinal hemorrhages in the AHT have been suggested. It is unknown how

much force it takes to create the ocular injuries seen in AHT. Different types of retinal hemorrhages may have different mechanisms and more than one theory may operate in any given instance. One group has suggested, based on incidence statistics, that relatively less force is needed to create intraretinal, subhyaloid, or optic nerve sheath hematoma than that needed for retinal detachment, choroidal hemorrhage, or vitreous hemorrhage [90]. However, this was a study of children who died from AHT, implying that the forces were still in a severe range.

One theory postulates that increased intracranial pressure due to cerebral edema and compressive subdural hemorrhage causes increased venous pressure and resultant obstruction in the retinal vasculature with rupture as the source of retinal hemorrhages. Sudden increases in chest or head pressure are hypothesized to be contributing factors as well. However, branch or central retinal vein occlusion are very uncommon manifestations of AHT, retinal hemorrhage is rarely seen in other causes of increased intracranial pressure (and even then, the hemorrhages are few in number and confined to the posterior pole), and the veins of the orbit are not valved and allow for extensive distribution of pressure. One group found no correlation between signs of elevated intracranial pressure and retinal hemorrhage in AHT [14]. A puzzling scenario is created by two case reports of severe hemorrhagic retinopathy with paramacular retinal folds in children who died after a well-documented head crush injury [52, 53]. Although one case [52] had features which separate it from other crush injuries reported in the literature, a larger study of crush injuries in children did find that some had milder retinal hemorrhages without folds or schisis [54]. Perhaps, the very severe and acute sudden elevation in intracranial pressure from this unique form of injury has a pathophysiologic role. Alternatively, particularly in view of the orbital fractures often seen in crush injury, shear forces to the globe may be important as with AHT-induced retinal hemorrhage [91]. Crush injury has a characteristic clinical profile which differs greatly from AHT, so there is not likely to be a diagnostic confusion. The victims in the case reports may or may not have had underlying biologic factors that predisposed to the unusual retinal findings.

The body of literature suggests that it is the shaking itself, with resultant repetitive acceleration–deceleration forces that induces shearing forces at the vitreo-retinal

interface, which is the primary factor in the generation of retinal hemorrhages seen in SBS [54, 91]. Subdural hemorrhage in AHT is thought to be caused by the shearing of small vessels from repeated acceleration– deceleration movements. When a child is shaken, the vitreous is also shaken which could cause shearing forces to be applied to the retina at points of firm attachment, which include the macula, retinal vessels, and peripheral retina. The high frequency of hemorrhages at the vitreous base (i.e., peripheral retina) and the unique macular retinoschisis of AHT support this theory.

Orbital shaking injury, including disruption of autonomic supply to the retinal vessels, may play a role. The optic nerve is longer than the distance between the apex of the orbit and the back of the globe, which allows the eyeball, and orbital contents to move when the child is shaken. As the optic nerve and other intraorbital structures are firmly attached to the eyeball and the apex of the orbit, injury may occur at these tethering locations as a result of translational and rotational movements of the globe. This may be one explanation for the findings of optic nerve sheath hemorrhage predominately anteriorly suggesting that blood did not arise from communication within the intracranial space [92]. The optic atrophy often seen in survivors may also be best explained by direct optic nerve injury within the orbit. Postmortem orbital findings show an increased rate of orbital hemorrhage in AHT victims when compared to victims of accidental head injury [48]. Blood in cranial nerve sheaths suggest a role for autonomic dysregulation, but direct vessel damage may also have a role in causing retinal hemorrhages. Disrupted autoregulation has been noted in a cat model of retinal shear injury [92].

The roles of the compounding effects of anoxia/ hypoxia, anemia, thrombocytopenia, mild coagulopathy, obstruction of retinal venous flow, or possible ­age-related anatomic variations in the retinal vasculature are not well understood. The adjunctive role of increased intracranial pressure needs further exploration. The role of additional potential contributing factors including underlying biologic factors which may modulate the hemorrhagic response, such as thrombophilic factors, needs investigation. However, each of these factors does not appear to play a primary role in the generation of the unique hemorrhagic retinopathy of AHT, as they do not do so singly or in combination in children who are not AHT victims.