Abstract
Aim
This study analyzed whether retino‐dural hemorrhages in infants are markers of the degree of intracranial pathology, rather than evidence of violent shaking.
Methods
Using data from 420 infants with acute intracranial pathologies, comparison of clinical findings is made between cases diagnosed as abusive head trauma (AHT) and four categories: cases where caregivers report no trauma; cases of witnessed or admitted AHT; cases where caregivers report accidental trauma; and witnessed accidents. The data are then controlled for degree of intracranial pathology by only comparing cases in each category that have evidence of hypoxic‐ischemic swelling.
Results
Although categories differ in clinical findings when all data are considered, they do not differ when the data are controlled for degree of intracranial pathology.
Conclusions
The data suggest that the clinical findings widely considered to be indicative of shaking are instead markers of the degree of intracranial pathology. Previous results showing differences were driven by selection effects, whereby different categories have different fractions of serious cases. Most notably, caregiver and witnessed reports of accidental head trauma led doctors to explore intracranial pathologies across a broader spectrum of severity, including milder cases, as opposed to situations where no head trauma is reported.
Keywords: abusive head trauma, evidence‐based medicine, retinal hemorrhage, shaken baby syndrome
Introduction
A range of clinical findings are often associated with abusive head trauma (AHT) that involves shaking, including encephalopathy, extensive, widespread retinal hemorrhages (RH), bilateral and interhemispheric subdural hemorrhages (SDH), respiratory compromise, and seizures. 1 , 2 When a combination of such clinical findings is present, the process for diagnosing AHT involves eliminating other known causes such as bleeding disorders or metabolic disease. In particular, the combination of encephalopathy, SDH, and RH is often referred to as the “triad,” but also as retino‐dural hemorrhages in infants. 3
However, the evidentiary association between these clinical findings and traumatic shaking has been criticized for its heavy reliance on circular reasoning and/or admissions made during the investigative or judicial process. 4 , 5 , 6 , 7 , 8 , 9 , 10 Neither diagnoses nor admissions made during the investigative or judicial process make reliable scientific reference standards to robustly categorize cases as AHT. These weaknesses in methodology have made some researchers question the validity and accuracy of the manner in which such clinical findings are used to diagnose AHT. 11
A further problem with studies that claim to establish an evidentiary link between particular clinical findings and AHT is the failure to control for the degree of cerebral pathology. 1 , 12 , 13
An alternative theory suggests that clinical findings normally associated with AHT may be markers of the degree of intracranial pathology rather than indicators of shaking. 14 , 15
This study controls for the degree of intracranial pathology when comparing various categories of cases and investigates whether it explains the different rates of clinical findings between the different categories of cases.
Methods
Data
The data set used for this study comprises 420 infants sourced from a publication by the Pediatric Brain Injury Research Network (PediBIRN) (https://www.pedibirn.com/home.php). PediBIRN is a collaborative initiative involving multiple university‐affiliated medical centers in the United States, aimed at establishing a clinical decision rule for evaluating AHT. The cases incorporated in the data set were drawn from eight medical centers participating in the pediBIRN collaboration, spanning the period between February 2011 and March 2021. The median age of infants was 5 months, ranging from 0 to 36 months, with 265 (63%) male and 155 (37%) female. The following summarizes the methodology used by the pediBIRN collaboration to create this publicly available data set.
All included infants had undergone intensive care hospitalization and exhibited symptomatic, acute, closed traumatic cranial or intracranial injuries, as confirmed through computerized tomography (416 infants) and/or magnetic resonance imaging (234 infants). Individuals with preexisting brain abnormalities or those involved in motor vehicle collisions were systematically excluded. Furthermore, those with medically diagnosed causes were omitted, meaning that all cases in the data set were assumed to be traumatic.
To standardize the classification of the data, investigators from the participating medical centers addressed a series of inquiries:
-
1.
Was the event of the child's head injury witnessed and thoroughly described by an impartial, independent observer? (a) Yes, and the observer characterized it as an "accidental" or "nonabusive" head injury event, (b) Yes, and the observer characterized it as an "inflicted" or "abusive" head injury event, or (c) "No or unknown."
In this study, cases corresponding to response (a) in question 1 are designated as witnessed accidents.
Cases corresponding to response (b) are designated as witnessed AHT.
For cases with response (c), investigators provided additional details by addressing further questions:
-
2.
Was the individual responsible for the child at the time of the head injury—or when the child's illness became evident—asked to provide an explanation? Of the 289 cases that were not witnessed [i.e., responded (c) to question 1], 276 cases answered affirmatively to question 2.
The cases that answered negatively to question 2 are excluded from this study. Among the cases where investigators answered affirmatively to question 2, investigators were subsequently queried:
-
3.
Which of the following statements most accurately summarizes the caregiver's account of the child's head injuries and immediate clinical presentation? (a) The caregiver described an "accidental" or "nonabusive" head injury event, (b) The caregiver explicitly admitted to "inflicted" or "abusive" head trauma, (c) The caregiver specifically denied any head trauma before the child's symptoms emerged, or (d) The caregiver declined to provide an explanation.
Cases corresponding to response (a) in question 3 are categorized as reported accidents. Cases corresponding to response (b) are categorized as admitted AHT. Cases corresponding to response (c) are categorized as reported no trauma.
Investigators were additionally asked to document the final consensus diagnosis provided by treating and consulting medical professionals. Response choices were definitive AHT, probable AHT, undetermined, probable non‐AHT, and definitive non‐AHT.
Case categories
In this study, four categories of cases are explored:
Reported no trauma : The caregiver specifically denied that the child experienced any head trauma before he or she became symptomatic.
Witnessed or admitted AHT : AHT was witnessed by an independent observer, or an admission to AHT was made.
Reported accidents : The caregiver described an accidental or nonabusive head injury event.
Witnessed accidents : The head injury was accidental, as witnessed and described thoroughly by an unbiased, independent observer.
Cases of reported accidents and reported nontrauma were not witnessed by an unbiased, independent observer. There were only three cases of witnessed AHT, so the category of witnessed or admitted AHT is statistically dominated by admitted cases.
These four categories of cases are then compared with cases that are diagnosed AHT , that is, those cases with a final consensus diagnosis of definitive or probable AHT.
The caregiver was defined as “the person responsible for the child when he/she was acutely head‐injured or first became clearly and persistently ill with clinical signs linked to acute traumatic head injuries.” 16
Clinical findings
The clinical findings explored in this study (and the number of cases with each finding) are:
Respiratory compromise: acute respiratory compromise (180).
Circulatory compromise: acute circulatory compromise (147).
Seizure(s): seizure(s) at the scene of injury, during transport, in the emergency department, or prior to hospital admission (139).
Hypoxia‐ischemia: brain hypoxia, ischemia, and/or swelling (142).
Acute encephalopathy: alteration or loss of consciousness at the scene of injury, during transport, in the emergency department, or prior to hospital admission (223).
Acute encephalopathy > 24 hours: as above where loss of consciousness lasts longer than 24 hours (110).
Bilateral SDH: SDH overlying both cerebral hemispheres (167).
Interhemispheric SDH: SDH involving or extending from the interhemispheric space (133).
Severe RH: RH described by an ophthalmologist as dense, extensive, covering a large surface area, and/or extending to the ora serrata (133).
Triad: bilateral and/or interhemispheric SDH plus acute encephalopathy plus severe RH (94).
Such clinical findings are often indicative of AHT and in particular of violent shaking. 2
Statistical tests
The p values were calculated using the chi‐squared test through the Python routine chi2_contingency from the scipy.stats module. Statistical significance was determined with a threshold of p < 0.05, while high significance was denoted by p < 0.01. Tables 1 and 2 present the occurrence rates of various clinical findings within case categories, alongside statistical comparisons between diagnosed cases and each of the other case categories, displaying p values and odds ratios accompanied by their respective 95% confidence intervals.
Table 1.
Comparison of rates of findings between cases diagnosed as abusive head trauma (AHT) and cases of reported no trauma, cases of witnessed or admitted AHT, cases of reported accidents, and cases of witnessed accidents. The p values and odds ratio (with 95% confidence interval) are shown for the comparison between diagnosed AHT and each of the other categories.
|
Reported no trauma Diagnosed AHT |
Admitted or witnessed AHT Diagnosed AHT |
Reported accident Diagnosed AHT |
Witnessed accident Diagnosed AHT |
|
|---|---|---|---|---|
| Clinical finding |
Rates p value OR (95%) |
Rates p value OR (95%) |
Rates p value OR (95%) |
Rates p value OR (95%) |
| Respiratory compromise |
59/95 (62%) 66/109 (60%) 0.82 1.07 (0.6–1.9) |
9/19 (47%) 121/192 (63%) 0.18 0.53 (0.2–1.4) |
80/232 (34%) 67/106 (63%) <0.001 0.31 (0.2–0.5) |
18/48 (37%) 121/192 (63%) <0.001 0.35 (0.2–0.7) |
| Circulatory compromise |
48/95 (50%) 58/109 (53%) 0.70 0.9 (0.5–1.6) |
9/19 (47%) 103/192 (53%) 0.60 0.78 (0.3–2.0) |
66/232 (28%) 56/106 (52%) <0.001 0.35 (0.2–0.6) |
13/48 (27%) 103/192 (53%) <0.001 0.32 (0.2–0.6) |
| Seizure(s) |
45/95 (47%) 51/109 (46%) 0.93 1.02 (0.6–1.8) |
9/19 (47%) 91/192 (47%) 0.99 1.0 (0.4–2.6) |
64/232 (27%) 50/106 (47%) <0.001 0.43 (0.3–0.7) |
10/48 (20%) 91/192 (47%) <0.001 0.29 (0.1–0.6) |
| SDH bilateral/interhemispheric |
77/95 (81%) 89/109 (81%) 0.91 0.96 (0.5–1.9) |
16/19 (84%) 192/192 (83%) 0.97 1.0 (0.3–3.7) |
94/232 (40%) 90/106 (84%) <0.001 0.12 (0.1–0.2) |
17/48 (35%) 161/192 (83%) <0.001 0.11 (0.05–0.2) |
| Acute encephalopathy |
63/95 (66%) 76/109 (69%) 0.60 0.85 (0.5–1.5) |
12/19 (63%) 192/192 (69%) 0.55 0.74 (0.3–2.0) |
109/232 (46%) 71/106 (66%) <0.001 0.44 (0.3–0.7) |
24/48 (50%) 192/192 (69%) 0.01 0.43 (0.2–0.8) |
| Acute encephalopathy >24 h |
43/95 (45%) 40/109 (36%) 0.21 1.43 (0.8–2.5) |
7/19 (36%) 80/192 (41%) 0.68 0.82 (0.3–2.2) |
47/232 (20%) 45/106 (42%) <0.001 0.34 (0.2–0.6) |
8/48 (16%) 80/192 (41%) <0.001 0.28 (0.1–0.6) |
| Severe RH |
58/92 (63%) 58/107 (54%) 0.21 1.44 (0.8–2.5) |
10/18 (56%) 115/188 (61%) 0.6 0.79 (0.3–2.1) |
51/159 (32%) 67/104 (64%) <0.001 0.26 (0.2‐0.4) |
4/20 (20%) 115/188 (61%) <0.001 0.16 (0.05‐0.5) |
| Triad |
36/92 (39%) 46/107 (42%) 0.58 0.85 (0.5–1.5) |
7/18 (39%) 82/188 (43%) 0.70 0.82 (0.3–2.2) |
40/159 (25%) 43/104 (41%) 0.01 0.48 (0.3–0.8) |
4/20 (20%) 82/188 (43%) 0.04 0.32 (0.1–1.0) |
Abbreviation: OR, odds ratio.
Table 2.
Comparison of rates of clinical findings between cases diagnosed as abusive head trauma (AHT) and cases of reported no trauma, cases of witnessed or admitted AHT, cases of reported accidents, and cases of witnessed accidents, but only for cases with hypoxia‐ischemia.
|
Reported no trauma + Hyp‐isch Diagnosed AHT + Hyp‐isch |
Admitted or witnessed AHT + Hyp‐isch Diagnosed AHT + Hyp‐isch |
Reported accident + Hyp‐isch Diagnosed AHT + Hyp‐isch |
Witnessed accident + Hyp‐isch Diagnosed AHT + Hyp‐isch |
|
|---|---|---|---|---|
| Clinical finding |
Rates p value OR (95%) |
Rates p value OR (95%) |
Rates p value OR (95%) |
Rates p value OR (95%) |
| Respiratory compromise |
43/54 (79%) 42/51 (82%) 0.72 0.84 (0.3–2.2) |
4/8 (50%) 84/102 (82%) 0.03 0.21 (0.0–0.9) |
49/60 (81%) 46/57 (80%) 0.89 1.1 (0.4–2.7) |
11/14 (78%) 84/102 (82%) 0.73 0.79 (0.2–3.1) |
| Circulatory compromise |
35/54 (64%) 38/51 (74%) 0.28 0.63 (0.3–1.5) |
4/8 (50%) 72/102 (70%) 0.22 0.42 (0.1–1.8) |
42/60 (70%) 38/57 (66%) 0.70 1.2 (0.5–2.5) |
8/14 (57%) 72/102 (70%) 0.31 0.56 (0.2–1.7) |
| Seizure(s) |
29/54 (53%) 30/51 (58%) 0.60 0.81 (0.4‐1.8) |
5/8 (62%) 57/102 (55%) 0.72 1.32 (0.3‐5.8) |
31/60 (51%) 29/57 (50%) 0.93 1.0 (0.5‐2.1) |
6/14 (42%) 57/102 (55%) 0.36 0.59 (0.2–1.8) |
| SDH bilateral/interhemispheric |
44/54 (81%) 43/51 (84%) 0.70 0.82 (0.3–2.3) |
8/8 (100%) 86/102 (84%) 0.23 – |
44/60 (73%) 46/57 (80%) 0.34 0.66 (0.3–1.6) |
12/14 (85%) 86/102 (84%) 0.89 1.1 (0.2–5.5) |
| Acute encephalopathy |
46/54 (85%) 47/51 (92%) 0.26 0.49 (0.1–1.7) |
6/8 (75%) 91/102 (89%) 0.23 0.36 (0.1–2.0) |
56/60 (93%) 48/57 (84%) 0.12 2.6 (0.8–9.1) |
12/14 (85%) 91/102 (89%) 0.70 0.73 (0.1–3.7) |
| Acute encephalopathy >24 h |
36/54 (66%) 32/51 (62%) 0.67 1.2 (0.5–2.6) |
4/8 (50%) 67/102 (65%) 0.37 0.52 (0.1–2.2) |
37/60 (61%) 38/57 (66%) 0.57 0.80 (0.4–1.7) |
7/14 (50%) 67/102 (65%) 0.25 0.52 (0.2–1.6) |
| Severe RH |
41/52 (78%) 37/51 (72%) 0.46 1.4 (0.6–3.5) |
7/8 (87%) 77/100 (77%) 0.49 2.1 (0.2–17.9) |
35/54 (64%) 44/55 (80%) 0.08 0.46 (0.2–1.1) |
3/6 (50%) 77/100 (77%) 0.14 0.3 (0.1–1.6) |
| Triad |
29/52 (55%) 30/51 (58%) 0.75 0.88 (0.4–1.9) |
5/8 (62%) 59/100 (59%) 0.85 1.2 (0.3–5.1) |
29/54 (53%) 31/55 (56%) 0.78 0.90 (0.4–1.9) |
3/6 (50%) 59/100 (59%) 0.66 0.69 (0.1–3.6) |
Abbreviation: OR, odds ratio.
In order to statistically compare categories of cases with diagnosed cases, the two must be independent variables. Presumably, independently witnessed accidents, independently witnessed abuse, and admissions may have an impact on diagnostic decisions, meaning that the diagnosis may be dependent on such information as well as on the clinical findings. Therefore, such cases are excluded from the diagnosed AHT cases when comparisons are made with the case categories. In practice, this exclusion did not affect the results or conclusions of this article: including them made minimal quantitative difference and no qualitative difference.
When comparing case categories with diagnosed cases, some cases will fall into the particular case category and will also be diagnosed as AHT. These cases are also excluded from the statistical tests. So, for example, cases in the category reported no accident that were diagnosed with AHT are excluded from the diagnosed AHT cases when these two populations are compared. This ensures that the two groups being compared are mutually exclusive. This means that the number of diagnosed AHT cases in Tables 1 and 2 will differ in each comparison. As with witnessed and admitted cases, including these cases does not affect the conclusions. Such exclusions are quantitively minimal and qualitatively make no difference to the results or conclusions: no p value in Table 1 or 2 changes from significant to insignificant or from insignificant to significant when these cases are included.
Results
Table 1 compares rates of clinical findings in cases diagnosed AHT to cases in each category, namely cases in the categories reported no trauma, witnessed or admitted AHT, reported accidents, and witnessed accidents.
There are no significant differences between cases diagnosed as AHT and those in the category of reported no trauma, nor with cases in the category admitted/witnessed AHT. By contrast, there are significant (or highly significant) differences in all clinical findings between cases diagnosed as AHT and cases in the categories reported accidents and witnessed accidents.
Table 2 summarizes the comparison of rates of clinical findings between cases diagnosed AHT and the same four categories of cases summarized in Table 1, but now including only cases with high degrees of intracranial pathology, as evidenced by the existence of hypoxia‐ischemia. The rates of the clinical findings increase in each of the categories in Table 2 as compared with those in Table 1, reflecting the correlation between hypoxia‐ischemic swelling and these clinical findings that are often associated with AHT.
No statistical significance is found when comparing clinical findings in cases diagnosed as AHT with those in cases in any other category, with the single exception of a slightly significant (p = 0.03) lower rate of respiratory compromise in the category admitted/witnessed AHT than in diagnosed AHT.
Discussion
Table 1 summarizes the statistical similarities in clinical findings between diagnosed and the category of admitted AHT and documents differences between diagnosed cases and the category witnessed accidents (non‐AHT). These results have been documented in several studies and have been interpreted as demonstrating the specificity of these clinical findings for AHT, and validating the methods used to diagnose AHT.
It is also well established in the literature that cases where caregivers report no trauma (they “deny” that trauma has occurred) demonstrate statistically similar clinical findings as cases of diagnosed AHT and admitted AHT, as is also presented in Table 1. The “denial” of trauma is therefore considered a red flag suggestive of AHT. 17 , 18
The statistical similarities presented in Table 1 between the categories of witnessed accidents and reported accidents have also been documented and have recently been interpreted as corroborating evidence that caregivers are generally telling the truth when they report a traumatic accident. 19
However, when comparing clinical findings across different categories of cases, it is crucial to carefully consider the methods employed for case selection within each category. The scrutiny of selection methods ensures that any comparisons drawn are equitable and unbiased. What might initially appear to be results stemming from disparities in clinical findings between categories can instead reflect how cases were selected to be categorized. These phenomena are commonly referred to as “selection effects.”
Two factors contribute to the selection effects that underlie the results observed in Table 1, when comparing categories of cases involving intracranial pathologies in infants and not controlling for degree of intracranial pathology.
First, when accidental head trauma is witnessed—be it by a caregiver in the reported accidents category or by an independent observer in the witnessed accident category—caregivers are inclined to seek medical attention even if symptoms are mild or absent. Conversely, in scenarios without traumatic incidents, caregivers are less likely to take an infant with minor or absent symptoms to the hospital. This inherent selection effect leads to cases involving witnessed accidents or reported accidents exhibiting a broader spectrum of intracranial pathologies, with such cases having a larger tail toward lower degrees of intracranial pathology compared with cases where there is no head trauma, such as the reported no trauma category.
Second, the symptoms associated with intracranial pathologies often lack specificity. Signs such as fussiness, colicky crying, vomiting, and apnea can stem from various conditions, resulting in multiple potential diagnoses. Consequently, in the absence of head trauma reports, the exploration of intracranial pathologies tends to be less common unless the pathology's severity is remarkable. Conversely, when instances of nonspecific symptoms are coupled with reported and/or witnessed head trauma, medical professionals are more likely to investigate intracranial pathologies. This, in turn, leads to identifying cases at the lower end of the intracranial pathology spectrum within the reported accidents or witnessed accidents categories than in the reported no trauma category.
The cases diagnosed AHT are selected to have high rates of clinical findings associated with AHT because diagnosis is predicated on the prevalence of these clinical findings. If such clinical findings are instead markers of degree of intracranial pathology, then cases with relatively low degrees of intracranial pathology will not be selected.
Similarly, cases with clinical findings associated with AHT are investigated by child protection units and subsequently by police. Indeed, in many jurisdictions it is mandatory for health professionals to report cases for investigation when clinical findings associated with AHT are present. These investigations can lead to suspicion falling on caregivers, their questioning, and in some instances leading to admissions. Even if all the admissions are true (there are a range of reasons why many of the admissions are likely false 10 ), these cases have been flagged for investigation due to clinical findings thought to be associated with AHT, which then “selects” cases with such clinical findings.
These factors and selection effects on degrees of intracranial pathologies associated with the various categories explain why the rates of clinical findings associated with AHT vary between the categories in Table 1. When the selection effects are controlled for by only selecting cases with high degrees of intracranial pathology, the rates of clinical findings associated with AHT are essentially the same in all categories, as seen in Table 2.
The results suggest that these clinical findings are not indicators of violent shaking but are instead markers of the degree of intracranial pathology, and that various selection effects are responsible for different categories of cases comprising broader or narrower ranges of such degrees of intracranial pathology.
Strengths of this study include the use of prospectively acquired and uniform pediBIRN data, and a methodology that is free from circular reasoning. Limitations include the inability to independently verify clinical findings that are accepted at face value, and a lack of detailed information as to the degree of independence of witnesses.
In summary, the theory supporting the notion that clinical findings linked with AHT, such as the triad of acute encephalopathy, diffuse SDH, and severe RH, serve as markers of the degree of intracranial pathology is supported by the data.
Author Contributions
Chris Brook: Conceptualization; formal analysis; methodology; writing—original draft.
Conflict of Interest
The author declares no conflicts of interest.
Data Availability Statement
Used publicly available data.
References
- 1. Hymel KP, Boos SC, Armijo‐Garcia V, et al. An analysis of physicians' diagnostic reasoning regarding pediatric abusive head trauma. Child Abuse Negl. 2022;129:105666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Choudhary AK, Servaes S, Slovis TL, et al. Consensus statement on abusive head trauma in infants and young children. Pediatr Radiol. 2018;48(8):1048‐1065. 10.1007/s00247-018-4149-1 [DOI] [PubMed] [Google Scholar]
- 3. Guthkelch AN. Problems of infant retino‐dural hemorrhage with minimal external injury. Houst J Health Law Policy. 2012;12:201‐208. [Google Scholar]
- 4. Fung ELW, Sung RYT, Nelson EAS, Poon WS. Unexplained subdural hematoma in young children: is it always child abuse? Pediatr Int. 2002;44:37‐42. [DOI] [PubMed] [Google Scholar]
- 5. Donohoe M. Evidence‐based medicine and shaken baby syndrome: part I: literature review, 1966‐1998. Am J Forensic Med Pathol. 2003;24(3):239‐242. [DOI] [PubMed] [Google Scholar]
- 6. Vinchon M. Reply to Pr Charles Hyman: the scientific controversy over abusive head trauma in infants. Childs Nerv Syst. 2011;27:203‐204. [Google Scholar]
- 7. Findley K, Barnes PD, Moran DA, Squier W. Shaken baby syndrome, abusive head trauma, and actual innocence: getting it right. Hous J Health Law Policy. 2012;12(2):258. [Google Scholar]
- 8. Elinder G, Eriksson A, Hallberg B, et al. Traumatic shaking: the role of the triad in medical investigations of suspected traumatic shaking. Acta Paediatr (Stockholm). 2018;107:3‐23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Brook C. Evidence for significant misdiagnosis of abusive head trauma in pediBIRN data. Forensic Sci Int Synergy. 2023;6:100314. 10.1016/j.fsisyn.2023.100314 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Rossant C, Brook C. Why admitted cases of AHT make a low quality reference standard: a survey of people accused of AHT in France. Forensic Sci Int Synergy. 2023;6:100312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Findley KA, Rossant C, Sasakura K, Schneps L, Squier W, Wester K, eds. Shaken Baby Syndrome: Investigating the Abusive Head Trauma Controversy. Cambridge University Press; 2023. [Google Scholar]
- 12. Binenbaum G, Mirza‐George N, Christian CW, Forbes BJ. Odds of abuse associated with retinal hemorrhages in children suspected of child abuse. J Am Assoc Pediatric Ophthalmol Strabismus. 2009;13(3):268‐272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Bhardwaj G, Jacobs MB, Martin FJ, et al. Photographic assessment of retinal hemorrhages in infant head injury: the Childhood Hemorrhagic Retinopathy Study. J Am Assoc Pediatric Ophthalmol Strabismus. 2017;21(1):28‐33. [DOI] [PubMed] [Google Scholar]
- 14. Thiblin I, Andersson J, Wester K, Högberg G, Högberg U. Retinal haemorrhage in infants investigated for suspected maltreatment is strongly correlated with intracranial pathology. Acta Paediatr (Stockholm). 2022;111(4):800‐808. 10.1111/apa.19239 [DOI] [PubMed] [Google Scholar]
- 15. Squier W. Retinodural haemorrhage of infancy, abusive head trauma, shaken baby syndrome: the continuing quest for evidence. Dev Med Child Neurol. 2023;66(3):290‐297. [DOI] [PubMed] [Google Scholar]
- 16. Hymel KP, Willson DF, Boos SC, et al. Derivation of a clinical prediction rule for pediatric abusive head trauma. Pediatric Crit Care Med. 2013;14:210‐220. [DOI] [PubMed] [Google Scholar]
- 17. Hettler J, Greenes DS. Can the initial history predict whether a child with a head injury has been abused? Pediatrics. 2003;111(3):602‐607. 10.1542/peds.111.3.602 [DOI] [PubMed] [Google Scholar]
- 18. Hymel KP, Lee G, Boos S, et al. Estimating the relevance of historical red flags in the diagnosis of abusive head trauma. J Pediatr. 2020;218:178‐183. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Brook C. Data‐driven evidence shows truthful caregiver histories and significant overdiagnosis of abusive head trauma. Ann Child Neurol Soc. 2023;1(4):299‐304. 10.1002/cns3.20035 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Used publicly available data.