Advertisement

The “New Science” of Abusive Head Trauma

  • Daniel M. LindbergEmail author
  • Howard Dubowitz
  • Randell C. Alexander
  • Robert M. Reece
Original Paper
  • 508 Downloads

Abstract

Claims that new science is changing accepted medical opinion about abusive head injury have been made frequently in the media, legal publications, and in legal cases involving abusive head trauma (AHT). This review analyzes recently published scientific articles about AHT to determine whether this new information has led to significant changes in the understanding, evaluation, and management of children with suspected AHT. Several specific topics are examined as follows: serious or fatal injuries from short falls, specificity of subdural hematoma for severe trauma, biomechanical explanations for findings, the specificity of retinal hemorrhages, the possibility of cerebral sinus thrombosis presenting with signs similar to AHT, and whether vaccines can produce such findings. We conclude (a) that the overwhelming weight of recent data does not change the fundamental consensus, (b) that abusive head trauma is a significant source of morbidity and mortality in children, (c) that subdural hematomas and severe retinal hemorrhages are commonly the results of severe trauma, (d) that these injuries should prompt an evaluation for abuse when identified in young children without a history of such severe trauma, and (e) that short falls, cerebral sinus thrombosis, and vaccinations are not plausible explanations for findings that raise concern for abusive head trauma.

Keywords

Abusive head trauma Retinal hemorrhages Subdural hematoma 

New scientific research has cast doubt on the forensic significance of this triad, thereby undermining the foundations of thousands of … convictions. (Tuerkheimer 2009)

New scientific evidence that changes the generally accepted understanding of the medical community may justify an appeal in criminal cases where medical evidence was important in the court’s conclusion. Recent claims that new science has changed the mainstream understanding of the forces and mechanisms of brain injuries in children, such as the quote above, are made in the popular media, legal articles, and criminal proceedings (“Cavazos 2011; Cenziper 2015; “Del Prete 2014; “State of Wisconsin v and Edmunds 2008; Tuerkheimer 2009, 2010). Because the medical diagnosis of abusive head trauma (AHT) also implies that a crime has occurred, scientific progress that substantially changes the way that AHT is understood or diagnosed can have profound effects for people accused or convicted of child physical abuse. However, in other areas relevant to child physical abuse, scientifically unsupported hypotheses such as temporary brittle bone disease or infantile rickets have also been submitted as new science, only to be widely discredited (Keller and Barnes 2008; Mendelson 2005; Paterson and Monk 2011; Slovis et al. 2012; Spivack and Otterman 2010; Sprigg 2011; Strouse 2013).

The objective of this review is to examine the concepts frequently submitted as “new science” in medical-legal cases of alleged abusive head trauma. The authors chose the topics based on their experience in legal settings. These include serious or fatal injuries from short falls, specificity of subdural hematoma for severe trauma, biomechanical explanations for findings, the specificity of retinal hemorrhages, the possibility of cerebral sinus thrombosis presenting with signs similar to AHT, and whether vaccines can produce such findings. With the exception of one author (RMR), who retired in 2010 after practicing child abuse pediatrics for more than 40 years, all authors are practicing child abuse physicians who have testified in US courts in cases of alleged AHT.

Serious or Fatal Injuries from Short falls

Case Presentation

A 4-month-old boy reportedly rolled off his parents’ bed and was found crying on a hardwood floor. He was inconsolable, refused his bottle, and vomited. His parents called 911, and emergency medical personnel recorded that he had a seizure en route to the hospital. On arrival, he was comatose and swelling was noted on his scalp in the right parietal region. A head CT scan revealed a thin, bilateral subdural hematoma (SDH) with loss of gray-white differentiation. The retinal exam showed bilateral extensive retinal hemorrhages (RHs). The skeletal survey was negative. The patient died 3 days later. An autopsy confirmed the radiographic findings, and no other injuries were identified.

Analysis

The core question is whether the history of a short fall represents a plausible explanation for the injuries. Falls are the initial explanation in up to 70% of cases where children are ultimately diagnosed with abusive injuries (Duhaime et al. 1992; Leventhal et al. 1993; Scherl et al. 2000; Strait et al. 1995).

Three publications have suggested that short, accidental falls can be life-threatening or fatal. First, Plunkett and others (Plunkett 2001) examined records from the United States Consumer Product Safety Commission National Injury Information Clearinghouse to identify all head and neck injuries involving playground equipment from January 1988 through June 1999. This study identified 18 cases where fall-related head injuries led to fatalities, with 5 of them occurring at home. Cases included children aged 12 months–13 years and the height of falls (measured from the lowest part of the body to the ground) was 0.6–3 m (2–10 ft). The authors reported that 12 falls were witnessed by someone other than the caretaker, and 12 had a “lucid interval” (i.e., a period where the child appeared mostly well). Among six children with the retinal examination, 4 were reported to have had bilateral RHs (the nature of which was not described).

The authors noted several limitations to their study: six falls were not independently observed, eight falls did not have a reported height, and other included falls were from significant heights—at least 5 falls were from 5 ft or more, and the highest fall was from 10 ft. The study has also been criticized because none of the subjects was less than 12 months old, and only nine were less than 5 years old. None had formal retinal examinations, and only 1 of the 5 children less than 2 years of age had a witnessed fall.

The second commonly cited article (Hall, Reyes, Horvat, Meller, & Stein, 1989) was a retrospective study of pediatric deaths from the Cook County Medical Examiner’s office over a 4-year period (1983–1986). Fatalities in 18 children with falls less than 3 ft and 18 children with falls between 3 ft and five stories were described, including one 8-month-old child who was dead on arrival after a reported fall from a couch to a hardwood floor, and who was found to have a large SDH.

There are several limitations to this study. The brief “Methods” section reveals that not all subjects had a full autopsy, that medical records were not available to the authors, and that radiographs were not used to evaluate for fractures. Further, because only the mean ages of subjects were given, it is not possible to know how many children were infants, and few details were provided about the reported mechanisms of injury.

Finally, a case report (Gardner 2007) involves an 11 month old who was reported to have fallen backwards from a sitting position to a carpeted floor. The child’s 5-year-old brother reportedly witnessed this event. The infant was found to have a SDH requiring surgery and diffuse multi-layer RHs. A skeletal survey was negative. Treating physicians were divided about whether the injury could have been the result of the history provided. The author concluded that “if RHs can [ever] occur without shaking … , they are of no value in determining etiology.” (p. 435) The clear concern with this article is whether the account attributed to the young sibling was accurate. Some skepticism seems warranted. The fall described is particularly minor. If such falls resulted in serious injuries, one would expect this to be a rather common occurrence. It is not.

Although these publications are often offered as “new science,” more recent data contradict these results. Chadwick and colleagues published data from 317 children who presented after a reported fall to a single trauma center in San Diego (Chadwick et al. 1991). Among 65 children who fell from 5 to 9 ft, and 118 children who fell from 10 to 45 ft, there were zero and one fatality, respectively. However, among 100 children with falls reported from less than 4 ft, there were seven deaths. The child who died from the high fall died of sepsis after a prolonged hospital course. Conversely, the seven short fall deaths died of head injury and all had associated injuries or concern about the accuracy of the history provided. If death is a rare outcome from a “high” fall, it seems quite unlikely that short falls would be fatal. The authors concluded that deaths reported as the result of falls less than 4 ft raise the suspicion that the history is inaccurate.

In 2008, Chadwick and colleagues performed a systematic review to determine the risk of death resulting from falls less than 1.5 m (approximately 5 ft) in children younger than 5 years old. This review examined data from five book chapters, seven literature reviews, three public injury databases, and 177 peer-reviewed articles categorized according to sample characteristics and methodology. This included a review of California’s statewide injury database, which reported a maximum of 13 short fall deaths among 2.5 million California children over a 5-year period. In this population, six fatalities were identified that might have been the result of short falls. Five articles described reliably witnessed falls in 560 children with hospital falls; none died. Twenty-five studies of children in large licensed daycares had only two fatalities attributed to falls in this setting. After data from all the above data sources, the authors concluded that the risk of death from a short fall is less than 1 in 2,000,000 per year.

Although science cannot prove that something is “impossible,” these data suggest that death or serious injury in a young child from a short fall is exceedingly rare and that all such deaths should be carefully investigated with a detailed history, physical examination, radiographic and laboratory studies, and a full autopsy.

The Specificity of SDH

Proponents of “new science” use a straw man to argue that new science is changing mainstream opinion. This straw theory—“triad theory”—asserts that clinicians simplistically diagnose abuse in any case that has SDH, any RHs, and any cerebral edema. While this straw theory is ridiculous on its face (abuse is not raised when these findings are seen in children who present after being hit by a car, or with a known, severe crush injury), the underlying argument deserves to be addressed. Most doctors view most significant SDHs to be the result of significant trauma. New science advocates suggest that SDHs are commonly the result of minor trauma or other non-traumatic causes that doctors fail to consider.

Many theories have been postulated as to how SDHs can happen without trauma, and it is beyond the scope of this review to address all of these. While Ehlers-Danlos syndrome, dysphagic choking, and osteogenesis imperfecta (among others) are occasionally put forward as explanations for SDHs, the most common arguments are rooted in three concepts: (a) that children commonly have asymptomatic SDHs resulting from birth, (b) that SDHs can occur spontaneously in children with enlarged extra-axial spaces, and (c) that sudden and catastrophic clinical decline can result from bleeding into an existing asymptomatic SDH.

Although some of these concepts are supported by the literature, ultimately, the conclusion that significant SDHs are commonly atraumatic is unsupported. To understand why, it is important to ask two questions. First, can SDH occur with little or no trauma? Second, when it does, what are the symptoms and prognosis?

SDH from Birth

With the evolution of fast magnetic resonance (MR) imaging techniques that image the infant brain without radiation or sedation, two groups of investigators have identified asymptomatic SDHs in a significant proportion of neonates. Looney et al. (2007) performed an MRI on 96 asymptomatic neonates. They identified SDH in 16 cases including two children who also had sub-arachnoid hemorrhages and five with intra-parenchymal hemorrhages. All affected children were born by vaginal deliveries and all hemorrhages were infra-tentorial or low in the occipital lobe.

A second group had similar results published 1 year later (Rooks et al. 2008). Of 101 asymptomatic neonates, 46 were found to have SDH on MRI. All were supra-tentorial and posterior, and < 3 mm in width. In contrast to Looney et al. (2007), this group did not identify other intracranial injury types. These authors also performed serial imaging to assess the natural history of these SDHs and showed that 94% had resolved by 1 month, and 100% resolved within 3 months. At 2-year follow-up examinations, six children (14%) had a speech delay and one was being evaluated for an autism spectrum disorder, but none had deficits clearly related to trauma. None had a sudden collapse.

Enlarged Extra-Axial Spaces

It has been well-described that the extra-axial space enlarges in older adults as the brain shrinks relative to the skull. That minor trauma can rarely cause SDH in elderly people led some to suggest that this could also be true in children with enlarged subdural spaces. If tearing of the bridging veins is the source of traumatic SDH, the theory goes; the widened extra-axial spaces may increase the risk for bridging veins to stretch and break, causing SDH.

Two articles have addressed this theory in children who had an MRI for macrocrania. The first, by McKeag et al. (2013), reviewed records from 177 children < 2 years old from the Children’s Hospital of Philadelphia, where MRI for macrocrania showed enlarged extra-axial spaces. Of these, four (2.3%) had SDH. Each of these had a thorough evaluation for abuse. This was negative in three of the children. One was thought to have been abused based on the finding of multiple rib fractures in different stages of healing.

The second article, by Greiner et al. 2013a, identified 108 children less than 2 years old with enlarged extra-axial spaces among 168 children imaged for macrocrania at the Cincinnati Children’s Hospital. Six children (3.6%) had SDH, all asymptomatic. Only two children with SDH were evaluated for abuse, and 1 was found to have characteristic RHs and was diagnosed as abuse.

An article from investigators in France (Vinchon et al. 2004), who prospectively collected data for several years in cases of pediatric SDH, found 16 cases in which they concluded that SDH had occurred without trauma. Twelve of these children had macrocrania, and others had illnesses such as severe dehydration. These authors conclude that “spontaneous” SDH does exist but it is rare and cannot be diagnosed without thorough evaluation for AHT. They state that the absence of traumatic features is insufficient to diagnose spontaneous SDH and conditions like macrocrania, severe dehydration, or arachnoidomegaly are necessary for the diagnosis.

The answer to the question “Is there new evidence that suggests that SDH can occur from birth, or without major trauma?” is “Yes.” But this simple conclusion is not the end of the discussion. It bears emphasis that each child with a SDH in these series was asymptomatic: none presented in a coma, and no child was thought to have non-abusive SDH when other injuries or RHs were present. Each article cited contains statements emphasizing the need to consider AHT in cases with unexplained SDH. Small, asymptomatic SDHs in the absence of traumatic injuries are not the focus of most abuse investigations. In most cases of AHT, the child has sudden, obvious symptoms of traumatic brain injury or other signs of trauma. To change the accepted opinion of AHT, the onus is on “new science” proponents to show how asymptomatic SDHs can rapidly evolve, leading to sudden neurological collapse.

Re-Bleeding into SDH

Evidence of re-bleeding in adults and children has been known for decades and is based on follow-up imaging showing enlarged SDH or new hyperdense blood products in an existing or old SDH. Although recent studies disproved the dogma that CT is able to determine the age of a SDH by density (Bradford et al. 2013; Tung et al. 2006; Vinchon et al. 2004), it is accepted that re-bleeding can occur in a SDH without significant additional trauma. Whether this is because bridging veins are stretched or whether vascularized neomembranes bleed (or both) is currently unknown, but does not affect the understanding of whether re-bleeding is likely to produce the findings of AHT.

The key question is whether re-bleeding causes symptoms likely to be mistaken for AHT. There is less direct evidence to address this question. One article describes two cases where re-bleeding is suggested to have caused symptom progression (Hymel et al. 2002). In one case, a 20-month-old child with a known, 3-month-old SDH presented with 1 week of fussiness after hitting a windowsill. In the next, an 11 months old presented with vomiting and was found to have a SDH. Later that night, increased swelling and a more hyperdense SDH was seen on CT. These cases do not directly address the question at hand. In the first case, it is not clear that the child had increased symptoms, and in the second, it is not clear that the mildly increased symptoms were not the result of the simple progression of the acute SDH. In neither case did the child have sudden severe symptoms after being asymptomatic.

Although there are no new data that directly address the question of whether minor trauma into a known SDH results in re-bleeding and sudden collapse, this concept is implausible, given the natural history of known SDHs. While SDHs commonly resolve over weeks to months, children with known asymptomatic or mildly symptomatic SDH are discharged from the hospital after a few days. They are discharged without protective helmets or other devices to prevent the minor trauma that is ubiquitous in the life of normal children. Yet children have not been reported to return with sudden, devastating symptoms after being discharged. Bradford et al. reported that, of 105 children with SDH, 17 (16%) were found to have re-bleeding, but none had symptoms (Bradford et al. 2013). The Pediatric Emergency Care Applied Research Network (PECARN) (Kuppermann et al. 2009) examined more than 10,000 children less than years old who presented to emergency departments for evaluation of traumatic brain injury. Robust follow-up procedures were employed to detect missed traumatic brain injuries. Results showed that asymptomatic children without loss of consciousness or severe mechanism of injury have < 0.02% chance of clinically important brain injury—including the potential for that brain injury to be identified days or weeks after presentation. In short, sudden symptoms as a result of re-bleeding into a SDH is not a known phenomenon among children with demonstrated SDH—it is only proposed for children where the original SDH is itself a theoretical entity.

Although there is new science about pediatric brain injury, it does not change the accepted understanding of AHT. In sum:
  1. 1.

    Small, asymptomatic SDH can occur from normal birth;

     
  2. 2.

    Small, asymptomatic, non-abusive SDH rarely occur in children with enlarged sub-arachnoid spaces;

     
  3. 3.

    CT imaging does not reliably estimate the age of SDH based on density.

     

Nevertheless, the vast majority of symptomatic SDH are the result of significant trauma; sudden, significant neurological symptoms in the setting of SDH imply a recent, significant traumatic event.

Biomechanics

It is sometimes suggested that biomechanical data offer a more pure and scientific approach than studies of human beings. Some proponents of “new science” have suggested that biomechanical evidence should change our understanding of the forces that are necessary or capable of causing traumatic brain injury.

In one of the first studies looking at the nature of the forces involved when a child is shaken, Duhaime et al. (1987) used two arguments to say that serious or fatal AHT injuries require more force than shaking alone. In the first phase, they carefully examined 13 consecutive dead infants diagnosed with the shaken baby syndrome (SBS) and found some evidence of head impact as well. They concluded that all cases of SBS must have some component of impact. In the second part, they constructed a doll model and measured peak accelerations. They first shook the doll (no impact) and compared peak accelerations to a trial in which they shook the doll and also allowed it to hit a bar (impact). Unsurprisingly, the sudden stop of an impact created a larger peak acceleration (deceleration). When the values were compared against monkey data and extrapolated, the authors concluded that shaking alone could not yield sufficient forces to cause serious brain injury.

Alexander et al. (1990) looked prospectively at 24 cases of diagnosed SBS. In addition, they were carefully examined for signs of trauma, often with MRI. Nine of the children died. There was not even marginal evidence of impact trauma in 12 of the cases, including five of the children who died. Shaking with impact (12 cases) and shaking without impact (12 cases) were both found. If impacts were significant in these AHT cases, as hypothesized by Duhaime et al. (1987), then all of the dead children should have shown impact. This was not found. Following this study, there has been no large study of children with AHT that has found all children to have signs of impact. Typically, about 50–70% of children have impact evidence along with shaking. Thus, the Duhaime hypothesis has been disproved. Shaking alone can cause serious and fatal injuries (Starling et al. 2004).

Duhaime et al.’s doll experiment, like others, have also been criticized for comparing only a single shake to a single impact—concluding that a single shake alone could not cause the injuries seen in AHT. However, perpetrators who confess to shaking have reported multiple shakes within seconds (Adamsbaum et al. 2010; Starling et al. 2004). The cumulative effect of these repetitive shakes was not modeled and would seem to be greater than that of a single shake.

Using a pig model, Raghupathi et al. (2004) shook 3–5-day-old anesthetized piglets once or twice with a 15-min gap. The mechanical shakes were rapid (< 15 m/sec) and did not involve impact. At 6 h post-injury, the brains were examined. More injury was seen with the double shake trial, showing that immature brains are vulnerable to repeated, relatively mild, non-impact loading conditions.

Bandak (2005) compared calculated structural failure limits of the cervical spine and deduced that the spine should be injured with less force than the brain. Clinically, this phenomenon is not seen; hence, the mechanism of shaking should be re-evaluated.

One response by Margulies and colleagues (Margulies et al. 2006) noted that Bandak made multiple numerical errors in his analysis, often off by a factor of 10. “There is no single, simple explanation responsible for the errors that appear in every value in Table 3.” (Margulies et al. 2006, p. 278). When accurately calculated, they argued that the brain can indeed be injured without neck injury—even using Bandak’s (2005) suppositions. They concluded that neck injury could occur during severe shaking without impact but that it would not necessarily occur if the shaking caused brain injury. Rangarajan and Shams (2006) also pointed out these numerical miscalculations and unclear assumptions. They noted that Bandak referenced three presentations at a conference in which the chair of the workshops said the material should not be used as references.

Using a mannequin, Wolfson et al. (2005) showed that concussion levels of energy could be achieved and raised concern about using models of single trauma to explain a phenomenon (shaking) that is often repetitive. “It is unlikely that further gross biomechanical investigation of the syndrome will be able to significantly contribute to the understanding of SBS.” (p. 70) “Current injury criteria are based on high-energy, single-impact studies. Since this is not the type of loading in SBS it is suggested that their application here is inappropriate and that future studies should focus on injury mechanism in low-energy cyclic loading.” (p. 70)

Biomechanical arguments asserted as part of the “new science” may also rest on extreme assumptions that do not seem to fit what we know of the natural world:
  1. 1.

    Heads are often treated as a single mathematical point. With 80–100 million neurons, layers of the brain of different densities, and a non-spherical shape, the human brain is over simplified when reduced to a simple point or shape.

     
  2. 2.

    Forces are complex. Repetitive shaking can result in translational movement in 3 dimensions, as well as spinning and shear forces, simultaneously. It is not clear that a measurement such as “peak acceleration” is the key factor to consider versus any number of forces acting over time.

     
  3. 3.

    Reductionist thinking often begins by quoting Newton’s second law: force equals mass times acceleration. This basic concept considers only simple, solid shapes (instead of complex biological tissue), and non-complex forces (straight back and forth) and does not account for repetitive forces. A common history in cases of AHT is that the child fell off the sofa. Compare that to an adult falling off a sofa. Acceleration = gravity which is a constant on earth. The equation then reduces to force is proportional to mass. For a tumble off a sofa, this means the bigger the force the harder the fall. One would expect an adult with a far greater mass to hit the floor far harder. Yet no one claims sofa falls create adult fatalities, and suicidal adults would never choose a couch in favor of bridges or high heights.

     

Biomechanical modeling attempts to simplify and understand forces and consequences in nature. Such modeling can be useful but is limited as to whether it faithfully represents the real world. The “new science” proponents selectively cite some biomechanical hypotheses as representing what happens rather than adjusting those models to what is actually observed. At most, biomechanics is a long way from explaining the injuries seen with shaking and shaking with impact.

Retinal Hemorrhages

Just as SDH have been suggested to be non-specific for severe trauma, so too have RHs. Proponents of “new science” assert that RHs can be caused by a wide array of medical and traumatic injuries. Although this fact has been known for decades, it ignores the types and patterns of RH and their significance in determining their origin.

The contemporary understanding of RH is summarized in an article comparing the prevalence and types of RHs in victims of abuse and those with non-abusive traumatic brain injury. Bechtel et al. (2004) examined 15 children with AHT, and 67 with non-abusive brain injuries. Although RHs were present in nine (60%) abused children, they were only found in seven (10%) non-abused subjects. Furthermore, none of the non-abused cohort had RHs that extended to the periphery of the retina. These data suggest that while RHs can occur from non-abusive trauma, these RHs are mild. Mainstream opinion has therefore held that RHs that are multi-layered, numerous, and extend to the periphery of the retina are specific for abuse (Levin 2010).

The “new science” addressing RHs began with a case report of a child who was found to have severe RHs and perimacular folds after sustaining severe brain injury because of a crush injury from a heavy television (Lantz et al. 2004). Another case report from the same group reported a fatal SDH and severe RHs after an unwitnessed stairway fall (Lantz and Couture 2011). As discussed above, a fatality from such a stairway fall is at least very unusual, though this child was also noted to have a partial thromboplastin time (PTT) > 200, implying a severe coagulopathy existed either before or after the severe brain injury.

The prevalence of RHs has also been addressed in children who are critically ill without trauma. Agrawal et al. (2012) examined 159 consecutive patients less than weeks old in a London pediatric intensive care unit. Subjects with AHT were excluded. The number of RHs was rated as mild (1–4), moderate (5–20), or severe (> 20). Of the159 subjects, 24 (15%) were found to have any RHs. Mild RHs were seen in 16 subjects, most of whom had sepsis. Two subjects had moderate RHs—one with sepsis and one with a television crush head injury. Six children were found to have severe RHs—three had leukemia and severe sepsis; one had hemorrhagic disease and a short fall and two had fatal traumatic brain injury (non-abusive).

Similarly, Longmuir et al. (2014) reported results of retinal examinations in 85 intubated PICU patients. Six patients (7%) had any RHs, including four children with AHT; all of whom had severe RHs. The other two children with RHs included one with moderate RHs from a TV crush injury, and one with mild RHs after a cardiac arrest attributed to SIDS.

Without evidence, the hypothesis has been raised that vaccines may cause retinal hemorrhages in the absence of trauma (Clemetson 2004; Gardner 2005; Squier 2011). Binenbaum et al. (2015) reasoned that if vaccines were a cause of RHs, then RHs would be seen frequently and would be temporally associated with immunization. They conducted a retrospective cohort study from June 1, 2009 through August 30, 2012, at the Pediatric Ophthalmology Clinics of the Children’s Hospital of Philadelphia. They examined 5177 children 1–23 months old who were undergoing dilated funduscopic exam for any reason.

The outcomes and measures they examined were the prevalence and cause of RHs and the temporal association between vaccine injections within 7, 14, and 21 days preceding these examinations and RHs. Reasons for dilated examinations in these children included strabismus, amblyopia, red eye, trauma, tear duct obstruction, poor visual behavior, or systemic diseases with ocular findings. The inclusion criterion was the availability of vaccine records. Excluded from the study were a history of direct eye trauma, intraocular surgery, and retinopathy of prematurity stage 3 or worse.

Among the 5177 children included (with 7675 fundus examinations), nine (0.17%) had RHs and all of these were victims of abusive head trauma (AHT) diagnosed with non-ocular findings including intracranial hemorrhage (9), skull fracture (5), bruises (3), hypoxic-ischemic brain injury (2), and spinal fracture, spinal hematoma, or perpetrator confession (1 each).

Vaccination records were available for 2210 (with 3425 fundus examinations). These vaccines included pneumococcus; diphtheria, tetanus, and pertussis (DTP); Haemophilus influenza type B (HIB); polio; hepatitis B; measles-mumps-rubella (MMR); and varicella. Four of the 2210 children (0.18%) with vaccination records had RHs. None of these 4 had vaccinations within 7 days preceding the diagnosis of RHs. One had vaccination within 14 days.

The conclusion of the investigators was that there was no association between receiving a vaccination injection and the presence of RHs in the subsequent 7, 14, or 21 days.

“New science” has not changed the understanding of RHs and its association with AHT. As has been known for decades, RHs can be caused by a host of traumatic and medical causes. However, severe RHs—numerous (> 20), multi-layered, and extending to the ora serrata—are very specific for severe traumatic brain injury. Although AHT is undoubtedly the most common source of severe RHs, serious crush injuries and children with leukemia and coagulopathy can also have severe RHs. Vaccination, however, does not cause RHs.

Cerebral Sinus Thrombosis

Not a single peer-reviewed paper has been identified linking cerebral sinovenous thrombosis (CSV) with other signs of AHT. Nevertheless, this hypothesis is suggested in legal proceedings. CSV is a rare disorder in children (0.67/100,000). The single largest age group with CSV is hospitalized neonates, comprising 43% of all cases up to 19 years of age. Less than 10% of cases have extra-cerebral hemorrhages. CSV is usually associated with systemic illness (84%) such as dehydration, metabolic acidosis, central nervous system infection, cyanotic heart disease, head injuries, craniotomy, renal, and thromboembolic disorders.

Choudhary et al. (2015) reported data from all children at their center younger than 36 months of age who were diagnosed with abusive head trauma (AHT) and who had both magnetic resonance imaging and MRI venography. The purpose of the study was to define the incidence and characteristics of venous and sinus abnormalities in abusive head trauma cases. The study was conducted from 2001 to 2012. Neuroradiologists independently analyzed MRI and MRI venography.

Forty-five children with a median age of 3 months (range 15 days to 31 months) were included. Sixty-two percent were boys. RHs were seen in 71%, extra-cranial fractures in 55%, and in 91%, a CT or MRI showed SDH. AHT was diagnosed by consensus of the treating medical team, perpetrator confession, and/or a judicial ruling. An experienced pediatric radiologist reviewed all of the skeletal surveys.

Thrombosis was defined as the absence of veins or sinuses on 3-D phase contrast MRI venography. MRI venography showed a mass effect on the venous sinuses or cortical draining veins in 69% (31/45). This mass effect was either displacement or partial or complete effacement of the venous structures from an adjacent SDH or brain swelling. The “lollipop” sign occurs when the bridging vein terminates in a sub-arachnoid blood clot and consequently does not drain into the sinus. This represents direct trauma to the cortical bridging veins and was seen in 44% (20/45) of the children.

The authors describe the limitations of the study, including technical limitations of MRI venography, the variations in venous anatomy, and in the appearance of venous thrombosis as well as the imaging peculiarities in children, mainly in terms of size of vessels.

The authors conclude that primary cortical sinus and venous thrombosis is a rare disorder, occurring in two to seven cases per million people. In children, predisposing factors are present in up to 95%. Outside the perinatal period, contributing factors include dehydration, malignancy, chemotherapy, iron-deficiency anemia, infection or sepsis, thrombophilia, gene mutations, and oral contraceptives. The clinical presentation of venous thrombosis is one of progressive, sub-acute decline, often over several days, usually in the context of another identifiable illness.

The Real New Science

For the reasons above, the understanding of AHT has not significantly changed in recent years. One recent survey of hundreds of pediatric specialists at leading children’s hospitals directly assessed which causes were considered most likely to cause the findings associated with AHT. Short falls, vaccines, or choking remained fringe theories as explanations for SDH, RH, and coma or death (Narang et al. 2016). But with the relatively recent recognition of child abuse pediatrics as a new subspecialty within pediatrics, it would be surprising to conclude that the science of AHT has not advanced. Indeed, remarkable progress has been made, especially in the recognition, prognosis, and treatment of AHT.

Centers have identified sentinel injuries that should routinely prompt consideration of AHT, including bruises in young infants, unexplained oral injuries, long-bone fractures, and abdominal injuries (Lindberg et al. 2015; Pierce et al. 2016; Sheets et al. 2013). Multi-center networks and systematic reviews have identified features of brain injuries that are most concerning for abuse (Hymel et al. 2014; Kemp et al. 2011). Recently, a decision rule to identify subtle signs of abusive and non-abusive brain injuries in young infants was validated in four pediatric emergency departments (Berger et al. 2016). Other studies have identified broad variability in screening practices for occult abusive injuries and have recommended best practices to move toward an objective, standardized approach to diagnosing abuse (Greiner et al. 2013b; Harper et al. 2013; Lindberg et al. 2013; Wood et al. 2015a, 2012, 2015b, 2010). Although confidence is waning in CT estimates of the age of SDHs, preliminary work suggests that someday estimation of the timing of RH may improve (Binenbaum et al. 2016). Development of new, fast MRI techniques suggests that screening for AHT will soon be done without the risks of radiation or sedation (Berger 2014; Cohen et al. 2015).

There is much new science with respect to AHT. Without exception, valid, and reproducible methods support the commonly held understanding of AHT that children who present with severe symptoms and who are found to have SDH and RHs are very likely to have been victims of significant trauma (Choudhary et al. 2018).

In a world of finite research resources, the real shame of this review is that so much time and effort is being devoted to research whose only purpose is to counter fringe theories. In any other field, hypotheses and theories based only on case reports, limited biomechanical theory, and aberrant interpretations of radiographs would have self-limited effects. Without reproduction, citations or enduring influence on the field, they would join tens of thousands of forgotten pieces of medical scholarship. It is only the fact that these articles are used in court gives them continued relevance. Ideally, this could be countered by broadly increasing the data literacy of lawyers, judges, jurors, and the public. However, until there is a reliable method by which courts can distinguish the validity of scientific data and thereby differentiate valid methods from fringe theories, it is left to the individual ethics of expert witnesses to communicate correctly the current state of the science.

Notes

Funding

Dr. Lindberg’s effort was supported by the National Institutes of Health, the Eunice Kennedy Schrivar National Institute of Child Health and Human Development K23HD083559. No other funding support was obtained.

Compliance with Ethical Standards

Conflict of Interest

The authors have each provided expert witness testimony in cases with alleged child physical abuse.

References

  1. Adamsbaum, C., Grabar, S., Mejean, N., Rey-Salmon, C. (2010). Abusive head trauma: judicial admissions highlight violent and repetitive shaking. Pediatrics, 126(3), 546–555.Google Scholar
  2. Agrawal, S., Peters, M. J., Adams, G. G., Pierce, C. M. (2012). Prevalence of retinal hemorrhages in critically ill children. Pediatrics, 129(6), e1388–e1396 (2012)  https://doi.org/10.1542/peds.2011-2772.
  3. Alexander, R., Crabbe, L., Sato, Y., Smith, W., Bennett, T. (1990). Serial abuse in children who are shaken. American Journal of Diseases of Children, 144(1), 58–60.Google Scholar
  4. Bandak, F. A. (2005). Shaken baby syndrome: A biomechanics analysis of injury mechanisms. Forensic Science International, 151(1), 71–79.Google Scholar
  5. Bechtel, K., Stoessel, K., Leventhal, J. M., Ogle, E., Teague, B., Lavietes, S., Duncan, C. (2004). Characteristics that distinguish accidental from abusive injury in hospitalized young children with head trauma. Pediatrics, 114(1), 165–168.Google Scholar
  6. Berger, R. (2014). Development of a Fast MRI Protocol to Screen for Abusive Head Trauma. Annapolis, MD: Paper Presented at the Helfer Society Annual Meeting.Google Scholar
  7. Berger, R. P., Fromkin, J., Herman, B., Pierce, M. C., Saladino, R. A., Flom, L., Kochanek, P. M. (2016). Validation of the Pittsburgh infant brain injury score for abusive head trauma. Pediatrics, 138(1).  https://doi.org/10.1542/peds.2015-3756.
  8. Binenbaum, G., Chen, W., Huang, J., Ying, G. S., Forbes, B. J. (2016). The natural history of retinal hemorrhage in pediatric head trauma. Journal of AAPOS, 20(2), 131–135 (2016).  https://doi.org/10.1016/j.jaapos.2015.12.008.
  9. Binenbaum, G., Christian, C. W., Guttmann, K., Huang, J., Ying, G. S., Forbes, B. J. (2015). Evaluation of temporal association between vaccinations and retinal hemorrhage in children. JAMA Ophthalmol, 133(11), 1261–1265 (2015)  https://doi.org/10.1001/jamaophthalmol.2015.2868.
  10. Bradford, R., Choudhary, A. K., & Dias, M. S. (2013). Serial neuroimaging in infants with abusive head trauma: timing abusive injuries. Journal of Neurosurgery Pediatrics, 12(2), 110–119.  https://doi.org/10.3171/2013.4.PEDS12596.
  11. Cavazos v. Smith, 565 U.S. 1 (2011) (per curiam).Google Scholar
  12. Cenziper, D. (2015). Shaken science: a disputed diagnosis imprisons parents. Washington Post.Google Scholar
  13. Chadwick, D. L., Chin, S., Salerno, C., Landsverk, J., Kitchen, L. (1991). Deaths from falls in children: How far is fatal? The Journal of Trauma, 31(10), 1353–1355.Google Scholar
  14. Choudhary, A. K., Bradford, R., Dias, M. S., Thamburaj, K., & Boal, D. K. (2015). Venous injury in abusive head trauma. Pediatric Radiology, 45(12), 1803–1813 (2015).  https://doi.org/10.1007/s00247-015-3399-4.
  15. Choudhary, A. K., Servaes, S., Slovis, T. L., Palusci, V. J., Hedlund, G. L., Narang, S. K., Offiah, A. C. (2018). Consensus statement on abusive head trauma in infants and young children. Pediatric Radiology, 48(8), 1048–1065 (2018).  https://doi.org/10.1007/s00247-018-4149-1.
  16. Clemetson, C. A. (2004). Elevated blood histamine caused by vaccinations and vitamin C deficiency may mimic the shaken baby syndrome. Medical Hypotheses, 62(4), 533–536 (2004)  https://doi.org/10.1016/j.mehy.2003.12.023.
  17. Cohen, A. R., Caruso, P., Duhaime, A. C., Klig, J. E. (2015). Feasibility of “rapid” magnetic resonance imaging in pediatric acute head injury. American Journal of Emergency Medicine, 33(7), 887–890 (2015)  https://doi.org/10.1016/j.ajem.2015.03.052.
  18. Del Prete v. Thompson, No. 10-5070 (N.D. Ill. Jan. 27, 2014).Google Scholar
  19. Duhaime, A. C., Alario, A. J., Lewander, W. J., Schut, L., Sutton, L. N., Seidl, T. S., et al. (1992). Head injury in very young children: Mechanisms, injury types, and ophthalmologic findings in 100 hospitalized patients younger than 2 years of age. Pediatrics, 90(2.1), 179–185.Google Scholar
  20. Duhaime, A. C., Gennarelli, T. A., Thibault, L. E., Bruce, D. A., Margulies, S. S., Wiser, R. (1987). The shaken baby syndrome. A clinical, pathological, and biomechanical study. J Neurosurg, 66(3), 409–415 (1987)  https://doi.org/10.3171/jns.1987.66.3.0409.
  21. Gardner, H. B. (2005). Immunizations, retinal and subdural hemorrhages: are they related? Medical Hypotheses, 64(3), 663 (2005).  https://doi.org/10.1016/j.mehy.2004.09.009.
  22. Gardner, H. B. (2007). A witnessed short fall mimicking presumed shaken baby syndrome (inflicted childhood neurotrauma). Pediatric Neurosurgery, 43(5), 433–435 (2007).  https://doi.org/10.1159/000106399.
  23. Greiner, M. V., Berger, R. P., Thackeray, J. D., Lindberg, D. M., For the ExSTRA Investigators. (2013a). Dedicated retinal examination in children evaluated for physical abuse without radiographically identified traumatic brain injury. Journal of Pediatrics, 163(2), 527–531 (2013a).  https://doi.org/10.1016/j.jpeds.2013.01.063.
  24. Greiner, M. V., Richards, T. J., Care, M. M., & Leach, J. L. (2013b). Prevalence of subdural collections in children with macrocrania. AJNR. American Journal of Neuroradiology, 34(12), 2373–2378.  https://doi.org/10.3174/ajnr.A3588.
  25. Hall, J. R., Reyes, H. M., Horvat, M., Meller, J. L., and Stein, R. (1989). The Mortality of Childhood Falls. The Journal of Trauma, 29(9), 1273-1275.Google Scholar
  26. Harper, N. S., Eddleman, S., Lindberg, D. M., For the ExSTRA Investigators. (2013). The utility of follow-up skeletal surveys in child abuse. Pediatrics, 131(3), e672–e678 (2013).  https://doi.org/10.1542/peds.2012-2608.
  27. Hymel, K. P., Armijo-Garcia, V., Foster, R., Frazier, T. N., Stoiko, M., Christie, L. M., For the Pediatric Brain Injury Research Network I. (2014). Validation of a clinical prediction rule for pediatric abusive head trauma. Pediatrics. 134, e1537–e1544 (2014).  https://doi.org/10.1542/peds.2014-1329
  28. Hymel, K. P., Jenny, C., Block, R. W. (2002). Intracranial hemorrhage and rebleeding in suspected victims of abusive head trauma: addressing the forensic controversies. Child Maltreatment, 7(4), 329–348.Google Scholar
  29. Keller, K. A., & Barnes, P. D. (2008). Rickets vs. abuse: a national and international epidemic. Pediatric Radiology, 38(11), 1210–1216.Google Scholar
  30. Kemp, A. M., Jaspan, T., Griffiths, J., Stoodley, N., Mann, M. K., Tempest, V., Maguire, S. A. (2011). Neuroimaging: What neuroradiological features distinguish abusive from non-abusive head trauma? A systematic review. Archives of Disease in Childhood, 96(12), 1103–1112.Google Scholar
  31. Kuppermann, N., Holmes, J. F., Dayan, P. S., Hoyle, J. D., Jr., Atabaki, S. M., Holubkov, R., et al. (2009). Identification of children at very low risk of clinically-important brain injuries after head trauma: a prospective cohort study. Lancet, 374(9696), 1160–1170.Google Scholar
  32. Lantz, P. E., & Couture, D. E. (2011). Fatal acute intracranial injury, subdural hematoma, and retinal hemorrhages caused by stairway fall. Journal of Forensic Sciences, 56(6), 1648–1653. (2011).  https://doi.org/10.1111/j.1556-4029.2011.01892.x.
  33. Lantz, P. E., Sinal, S. H., Stanton, C. A., Weaver, R. G., Jr. (2004). Perimacular retinal folds from childhood head trauma. BMJ, 328(7442), 754–756.Google Scholar
  34. Leventhal, J. M., Thomas, S. A., Rosenfield, N. S., Markowitz, R. I. (1993). Fractures in young children: Distinguishing child abuse from unintentional injuries. American Journal of Diseases of Children, 147(1), 87–92.Google Scholar
  35. Levin, A. V. (2010). Retinal hemorrhage in abusive head trauma. Pediatrics, 126(5), 961–970.  https://doi.org/10.1542/peds.2010-1220.
  36. Lindberg, D. M., Beaty, B., Juarez-Colunga, E., Wood, J. N., & Runyan, D. K. (2015). Testing for abuse in children with sentinel injuries. Pediatrics, 136(5), 831–838.  https://doi.org/10.1542/peds.2015-1487.
  37. Lindberg, D. M., Shapiro, R. A., Blood, E. A., Steiner, R. D., Berger, R. P., & for the ExSTRA investigators. (2013). Utility of hepatic transaminases in children with concern for abuse. Pediatrics., 131, 268–275.  https://doi.org/10.1542/peds.2012-1952.
  38. Longmuir, S. Q., McConnell, L., Oral, R., Dumitrescu, A., Kamath, S., & Erkonen, G. (2014). Retinal hemorrhages in intubated pediatric intensive care patients. Journal of AAPOS, 18(2), 129–133.  https://doi.org/10.1016/j.jaapos.2013.12.005.
  39. Looney, C. B., Smith, J. K., Merck, L. H., Wolfe, H. M., Chescheir, N. C., Hamer, R. M., Gilmore, J. H. (2007.) Intracranial Hemorrhage in Asymptomatic Neonates: Prevalence on MR Images and Relationship to Obstetric and Neonatal Risk Factors. Radiology, 242(2), 535–541.Google Scholar
  40. Margulies, S., Prange, M., Myers, B. S., Maltese, M. R., Ji, S., Ning, X., ..., Christian, C. (2006). Shaken baby syndrome: a flawed biomechanical analysis. Forensic Science International , 164(2–3), 278–279; author reply 282-273. doi: https://doi.org/10.1016/j.forsciint.2005.12.018.
  41. McKeag, H., Christian, C. W., Rubin, D., Daymont, C., Pollock, A. N., & Wood, J. (2013). Subdural hemorrhage in pediatric patients with enlargement of the subarachnoid spaces. Journal of Neurosurgery. Pediatrics, 11(4), 438–444.  https://doi.org/10.3171/2012.12.PEDS12289.
  42. Mendelson, K. L. (2005). Critical review of ‘temporary brittle bone disease’. Pediatric Radiology, 35(10), 1036–1040.  https://doi.org/10.1007/s00247-005-1573-9.
  43. Narang, S. K., Estrada, C., Greenberg, S., & Lindberg, D. (2016). Acceptance of shaken baby syndrome and abusive head trauma as medical diagnoses. Journal of Pediatrics, 177, 273–278.  https://doi.org/10.1016/j.jpeds.2016.06.036.
  44. Paterson, C. R., & Monk, E. A. (2011). Temporary brittle bone disease: Relationship between clinical findings and judicial outcome. Pediatric Reports, 3(3), e24.Google Scholar
  45. Pierce, M. C., Magana, J. N., Kaczor, K., Lorenz, D. J., Meyers, G., Bennett, B. L., & Kanegaye, J. T. (2016). The prevalence of bruising among infants in pediatric emergency departments. Annals of Emergency Medicine, 67(1), 1–8.  https://doi.org/10.1016/j.annemergmed.2015.06.021.
  46. Plunkett, J. (2001). Fatal pediatric head injuries caused by short-distance falls. American Journal of Forensic Medicine and Pathology, 22(1), 1–12.Google Scholar
  47. Raghupathi, R., Mehr, M. F., Helfaer, M. A., & Margulies, S. S. (2004). Traumatic axonal injury is exacerbated following repetitive closed head injury in the neonatal pig. Journal of Neurotrauma, 21(3), 307–316.  https://doi.org/10.1089/089771504322972095.
  48. Rangarajan, N., & Shams, T. (2006). Re: Shaken baby syndrome: A biomechanics analysis of injury mechanisms. Forensic Science International, 164(2–3), 280–281; Author reply 282-283.  https://doi.org/10.1016/j.forsciint.2005.12.017.
  49. Rooks, V. J., Eaton, J. P., Ruess, L., Petermann, G. W., Keck-Wherley, J., & Pedersen, R. C. (2008). Prevalence and evolution of intracranial hemorrhage in asymptomatic term infants. American Journal of Neuroradiology, 29(6), 1082–1089.  https://doi.org/10.3174/ajnr.A1004.
  50. Scherl, S. A., Miller, L., Lively, N., Russinoff, S., Sullivan, C. M., & Tornetta, P., 3rd. (2000). Accidental and nonaccidental femur fractures in children. Clinical Orthopaedics and Related Research, 376, 96–105.Google Scholar
  51. Sheets, L. K., Leach, M. E., Koszewski, I. J., Lessmeier, A. M., Nugent, M., & Simpson, P. (2013). Sentinel injuries in infants evaluated for child physical abuse. Pediatrics., 131, 701–707.  https://doi.org/10.1542/peds.2012-2780.
  52. Slovis, T. L., Strouse, P. J., Coley, B. D., & Rigsby, C. K. (2012). The creation of non-disease: An assault on the diagnosis of child abuse. Pediatric Radiology, 42(8), 903–905.  https://doi.org/10.1007/s00247-012-2425-z.
  53. Spivack, B. S., & Otterman, G. J. (2010). Does temporary brittle bone disease exist? Not by the evidence offered. Acta Paediatrica, 99(4), 486.  https://doi.org/10.1111/j.1651-2227.2010.01746.x.
  54. Sprigg, A. (2011). Temporary brittle bone disease versus suspected non-accidental skeletal injury. Archives of Disease in Childhood, 96(5), 411–413.  https://doi.org/10.1136/adc.2009.180463.
  55. Squier, W. (2011). The “shaken baby” syndrome: pathology and mechanisms. Acta Neuropathologica, 122(5), 519–542.  https://doi.org/10.1007/s00401-011-0875-2.
  56. Starling, S. P., Patel, S., Burke, B. L., Sirotnak, A. P., Stronks, S., & Rosquist, P. (2004). Analysis of perpetrator admissions to inflicted traumatic brain injury in children. Archives of Pediatrics & Adolescent Medicine, 158(5), 454–458.  https://doi.org/10.1001/archpedi.158.5.454.
  57. State v. Edmunds, No. 2007AP933 (Wis. Ct. App. 2008).Google Scholar
  58. Strait, R. T., Siegel, R. M., & Shapiro, R. A. (1995). Humeral fractures without obvious etiologies in children less than 3 years of age: when is it abuse? Pediatrics, 96(4 Pt 1), 667–671.Google Scholar
  59. Strouse, P. J. (2013). ‘Keller & Barnes’ after 5 years: Still inadmissible as evidence. Pediatric Radiology, 43(11), 1424–1424.  https://doi.org/10.1007/s00247-013-2810-2.
  60. Tuerkheimer, D. (2009). The next innocence project: Shaken baby syndrome and the criminal courts. Washington University Law Review, 87, 1.Google Scholar
  61. Tuerkheimer, D. (2010). Anatomy of a misdiagnosis, Op-Ed. New York Times, p. A31.Google Scholar
  62. Tung, G. A., Kumar, M., Richardson, R. C., Jenny, C., & Brown, W. D. (2006). Comparison of accidental and nonaccidental traumatic head injury in children on noncontrast computed tomography. Pediatrics, 118(2), 626–633.Google Scholar
  63. Vinchon, M., Noule, N., Tchofo, P. J., Soto-Ares, G., Fourier, C., & Dhellemmes, P. (2004). Imaging of head injuries in infants: temporal correlates and forensic implications for the diagnosis of child abuse. Journal of Neurosurgery, 101(1 Suppl), 44–52.Google Scholar
  64. Wolfson, D. R., McNally, D. S., Clifford, M. J., & Vloeberghs, M. (2005). Rigid-body modelling of shaken baby syndrome. Proceedings of the Institution of Mechanical Engineers. Part H, 219(1), 63–70.Google Scholar
  65. Wood, J. N., Fakeye, O., Mondestin, V., Rubin, D. M., Localio, R., & Feudtner, C. (2015a). Development of hospital-based guidelines for skeletal survey in young children with bruises. Pediatrics, 135(2), e312–e320.  https://doi.org/10.1542/peds.2014-2169.
  66. Wood, J. N., Feudtner, C., Medina, S. P., Luan, X., Localio, R., & Rubin, D. M. (2012). Variation in occult injury screening for children with suspected abuse in selected U.S. children’s hospitals. Pediatrics., 130, 853–860.  https://doi.org/10.1542/peds.2012-0244.
  67. Wood, J. N., French, B., Song, L., & Feudtner, C. (2015b). Evaluation for occult fractures in injured children. Pediatrics, 136(2), 232–240.  https://doi.org/10.1542/peds.2014-3977.
  68. Wood, J. N., Hall, M., Schilling, S., Keren, R., Mitra, N., & Rubin, D. M. (2010). Disparities in the evaluation and diagnosis of abuse among infants with traumatic brain injury. Pediatrics, 126(3), 408–414.Google Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  • Daniel M. Lindberg
    • 1
    Email author
  • Howard Dubowitz
    • 2
  • Randell C. Alexander
    • 3
  • Robert M. Reece
    • 4
  1. 1.The Kempe Center for the Prevention & Treatment of Child Abuse & NeglectUniversity of Colorado Anschutz Medical CampusAuroraUSA
  2. 2.Department of PediatricsUniversity of Maryland School of MedicineBaltimoreUSA
  3. 3.Department of PediatricsUniversity of Florida College of MedicineJacksonvilleUSA
  4. 4.Tufts University School of Medicine (Ret.)BostonUSA

Personalised recommendations