Validation of the PediBIRN-7 clinical prediction rule for pediatric abusive head trauma.

BACKGROUND: The PediBIRN-7 clinical prediction rule incorporates the (positive or negative) predictive contributions of completed abuse evaluations to estimate abusive head trauma (AHT) probability after abuse evaluation. Applying definitional criteria as proxies for AHT and non-AHT ground truth, it performed with sensitivity 0.73 (95 % CI: 0.66-0.79), specificity 0.87 (95 % CI: 0.82-0.90), and ROC-AUC 0.88 (95 % CI: 0.85-0.92) in its derivation study. OBJECTIVE: To validate the PediBIRN-7's AHT prediction performance in a novel, equivalent, patient population. PARTICIPANTS AND SETTINGS: Consecutive, acutely head-injured children <3 years hospitalized for intensive care across eight sites between 2017 and 2020 with completed skeletal surveys and retinal exams (N = 342). METHODS: Secondary analysis of an existing, cross-sectional, prospective dataset, including assignment of patient-specific estimates of AHT probability, calculation of AHT prediction performance measures (ROC-AUC, sensitivity, specificity, predictive values), and completion of sensitivity analyses to estimate best- and worst-case prediction performances. RESULTS: Applying the same definitional criteria, the PediBIRN-7 performed with sensitivity 0.74 (95 % CI: 0.66-0.81), specificity 0.77 (95 % CI: 0.70-0.83), and ROC-AUC 0.83 (95 % CI: 0.78-0.88). The reduction in ROC-AUC was statistically insignificant (p = .07). Applying physicians' final consensus diagnoses as proxies for AHT and non-AHT ground truth, the PediBIRN-7 performed with sensitivity 0.73 (95 % CI: 0.66-0.79), specificity 0.87 (95 % CI: 0.82-0.90), and ROC-AUC 0.90 (95 % CI: 0.87-0.94). Sensitivity analyses demonstrated minimal changes in rule performance. CONCLUSION: The PediBIRN-7's overall AHT prediction performance has been validated in a novel, equivalent, patient population. Its patient-specific estimates of AHT probability can inform physicians' AHT-related diagnostic reasoning after abuse evaluation.

Retinal hemorrhage variation in inertial versus contact head injuries.

BACKGROUND: Abusive head trauma (AHT) is frequently accompanied by dense/extensive retinal hemorrhages to the periphery with or without retinoschisis (complex retinal hemorrhages, cRH). cRH are uncommon without AHT or major trauma. OBJECTIVE: The study objectives were to determine whether cRH are associated with inertial vs. contact mechanisms and are primary vs. secondary injuries. PARTICIPANTS AND SETTING: This retrospective study utilized a de-identified PediBIRN database of 701 children <3-years-old presenting to intensive care for head trauma. Children with motor vehicle related trauma and preexisting brain abnormalities were excluded. All had imaging showing head injury and a dedicated ophthalmology examination. METHODS: Contact injuries included craniofacial soft tissue injuries, skull fractures and epidural hematoma. Inertial injuries included acute impairment or loss of consciousness and/or bilateral and/or interhemispheric subdural hemorrhage. Abuse was defined in two ways, by 1) predetermined criteria and 2) caretaking physicians/multidisciplinary team's diagnostic consensus. RESULTS: PediBIRN subjects with cRH frequently experienced inertial injury (99.4 % (308/310, OR = 53.74 (16.91-170.77)) but infrequently isolated contact trauma (0.6 % (2/310), OR = 0.02 (0.0004-0.06)). Inertial injuries predominated over contact trauma among children with cRH sorted AHT by predetermined criteria (99.1 % (237/239), OR = 20.20 (6.09-67.01) vs 0.5 % (2/339), OR = 0.04 (0.01-0.17)). Fifty-nine percent of patients with cRH, <24 h altered consciousness, and inertial injuries lacked imaging evidence of brain hypoxia, ischemia, or swelling. CONCLUSIONS: cRH are significantly associated with inertial angular acceleration forces. They can occur without brain hypoxia, ischemia or swelling suggesting they are not secondary injuries.

Skeletal survey yields in low vs. high risk pediatric patients with skull fractures.

BACKGROUND: To assess for occult fractures, physicians often opt to obtain skeletal surveys (SS) in young, acutely head-injured patients who present with skull fractures. Data informing optimal decision management are lacking. OBJECTIVE: To determine the positive yields of radiologic SS in young patients with skull fractures presumed to be at low vs. high risk for abuse. PARTICIPANTS AND SETTING: 476 acutely head injured, skull-fractured patients <3 years hospitalized for intensive care across 18 sites between February 2011 and March 2021. METHODS: We conducted a retrospective, secondary analysis of the combined, prospective Pediatric Brain Injury Research Network (PediBIRN) data set. RESULTS: 204 (43 %) of 476 patients had simple, linear, parietal skull fractures. 272 (57 %) had more complex skull fracture(s). Only 315 (66 %) of 476 patients underwent SS, including 102 (32 %) patients presumed to be at low risk for abuse (patients who presented with a consistent history of accidental trauma; intracranial injuries no deeper than the cortical brain; and no respiratory compromise, alteration or loss of consciousness, seizures, or skin injuries suspicious for abuse). Only one of 102 low risk patients revealed findings indicative of abuse. In two other low risk patients, SS helped to confirm metabolic bone disease. CONCLUSIONS: Less than 1 % of low risk patients under three years of age who presented with simple or complex skull fracture(s) revealed other abusive fractures. Our results could inform efforts to reduce unnecessary skeletal surveys.

Child Abuse Pediatrics Research Network: The CAPNET Core Data Project.

OBJECTIVE: Examine the epidemiology of subspecialty physical abuse evaluations within CAPNET, a multicenter child abuse pediatrics research network. METHODS: We conducted a cross-sectional study of children <10 years old who underwent an evaluation (in-person or remote) by a child abuse pediatrician (CAP) due to concerns for physical abuse at ten CAPNET hospital systems from February 2021 through December 2021. RESULTS: Among 3667 patients with 3721 encounters, 69.4% were <3 years old; 44.3% <1 year old, 59.1% male; 27.1% Black; 57.8% White, 17.0% Hispanic; and 71.0 % had public insurance. The highest level of care was outpatient/emergency department in 60.7%, inpatient unit in 28.0% and intensive care in 11.4%. CAPs performed 79.1% in-person consultations and 20.9% remote consultations. Overall, the most frequent injuries were bruises (35.2%), fractures (29.0%), and traumatic brain injuries (TBI) (16.2%). Abdominal (1.2%) and spine injuries (1.6%) were uncommon. TBI was diagnosed in 30.6% of infants but only 8.4% of 1-year old children. In 68.2% of cases a report to child protective services (CPS) was made prior to CAP consultation; in 12.4% a report was made after CAP consultation. CAPs reported no concern for abuse in 43.0% of cases and mild/intermediate concern in 22.3%. Only 14.2% were categorized as definite abuse. CONCLUSION: Most children in CAPNET were <3 years old with bruises, fractures, or intracranial injuries. CPS reports were frequently made prior to CAP consultation. CAPs had a low level of concern for abuse in majority of cases.

The association of subcortical brain injury and abusive head trauma.

BACKGROUND: Abusive head trauma (AHT) remains a major pediatric problem with diagnostic challenges. A small pilot study previously associated subcortical brain injury with AHT. OBJECTIVES: To investigate the association of subcortical injury on neuroimaging with the diagnosis of AHT. PARTICIPANTS AND SETTING: Children <3 years with acute TBI admitted to 18 PICUs between 2011 and 2021. METHODS: Secondary analysis of existing, combined, de-identified, cross-sectional dataset. RESULTS: Deepest location of visible injury was characterized as scalp/skull/epidural (n = 170), subarachnoid/subdural (n = 386), cortical brain (n = 170), or subcortical brain (n = 247) (total n = 973). Subcortical injury was significantly associated with AHT using both physicians' diagnostic impression (OR: 8.41 [95 % CI: 5.82-12.44]) and a priori definitional criteria (OR: 5.99 [95 % CI: 4.31-8.43]). Caregiver reports consistent with the child's gross motor skills and historically consistent with repetition decreased as deepest location of injury increased, p < 0.001. Patients with subcortical injuries were significantly more likely to have traumatic extracranial injuries such as rib fractures (OR 3.36, 95 % CI 2.30-4.92) or retinal hemorrhages (OR 5.97, 95 % CI 4.35-8.24), respiratory compromise (OR 12.12, 95 % CI 8.49-17.62), circulatory compromise (OR 6.71, 95 % CI 4.87-9.29), seizures (OR 3.18, 95 % CI 2.35-4.29), and acute encephalopathy (OR 12.44, 95 % CI 8.16-19.68). CONCLUSIONS: Subcortical injury is associated with a diagnosis of AHT, historical inaccuracies concerning for abuse, traumatic extracranial injuries, and increased severity of illness including respiratory and circulatory compromise, seizures, and prolonged loss of consciousness. Presence of subcortical injury should be considered as one component of the complex AHT diagnostic process.

The CAPNET multi-center data set for child physical abuse: Rationale, methods and scope.

BACKGROUND: The pediatric subspecialty of Child Abuse Pediatrics (CAP) was certified by the American Board of Medical Subspecialties in 2006. Relative to its impact on pediatric health, CAP-focused research has been relatively under-funded. Multi-center networks related to CAP-focused research have made important advances, but have been limited in scope and duration. CAPNET is multi-center network whose mission is to support CAP-focused research. OBJECTIVE: To describe the rationale, development, and scope of the CAPNET research network infrastructure, the CAPNET data registry and associated data resources. METHODS: Based on existing priorities for CAP-focused research, we used consensus building and iterative testing to establish inclusion criteria, common data elements, data quality assurance, and data sharing processes for children with concerns of physical abuse. RESULTS: We describe the rationale, methods and intended scope for the development of the CAPNET research network and data registry. CAPNET is currently abstracting data for children <10 years (120 months) old who undergo sub-specialty evaluation for physical abuse at 10 US pediatric centers (approximately 4000 evaluations/year total) using an online data capture form. Data domains include: demographics; visit timing and providers, medical/social history, presentation, examination findings, laboratory and radiographic testing, diagnoses, outcomes, and data for contact children. We describe the methods and criteria for collecting and validating data which are broadly available to CAP investigators. CONCLUSIONS: CAPNET represents a new data resource for the CAP research community and will increase the quantity and quality of CAP-focused research.

An analysis of physicians' diagnostic reasoning regarding pediatric abusive head trauma.

BACKGROUND: Physician diagnoses of abusive head trauma (AHT) have been criticized for circular reasoning and over-reliance on a "triad" of findings. Absent a gold standard, analyses that apply restrictive reference standards for AHT and non-AHT could serve to confirm or refute these criticisms. OBJECTIVES: To compare clinical presentations and injuries in patients with witnessed/admitted AHT vs. witnessed non-AHT, and with witnessed/admitted AHT vs. physician diagnosed AHT not witnessed/admitted. To measure the triad's AHT test performance in patients with witnessed/admitted AHT vs. witnessed non-AHT. PARTICIPANTS AND SETTING: Acutely head injured patients <3 years hospitalized for intensive care across 18 sites between 2010 and 2021. METHODS: Secondary analyses of existing, combined, cross-sectional datasets. Probability values and odds ratios were used to identify and characterize differences. Test performance measures included sensitivity, specificity, and predictive values. RESULTS: Compared to patients with witnessed non-AHT (n = 100), patients with witnessed/admitted AHT (n = 58) presented more frequently with respiratory compromise (OR 2.94, 95% CI: 1.50-5.75); prolonged encephalopathy (OR 5.23, 95% CI: 2.51-10.89); torso, ear, or neck bruising (OR 11.87, 95% CI: 4.48-31.48); bilateral subdural hemorrhages (OR 8.21, 95% CI: 3.94-17.13); diffuse brain hypoxia, ischemia, or swelling (OR 6.51, 95% CI: 3.06-13.02); and dense, extensive retinal hemorrhages (OR 7.59, 95% CI: 2.85-20.25). All differences were statistically significant (p ≤ .001). No significant differences were observed in patients with witnessed/admitted AHT (n = 58) vs. patients diagnosed with AHT not witnessed/admitted (n = 438). The triad demonstrated AHT specificity and positive predictive value ≥0.96. CONCLUSIONS: The observed differences in patients with witnessed/admitted AHT vs. witnessed non-AHT substantiate prior reports. The complete absence of differences in patients with witnessed/admitted AHT vs. physician diagnosed AHT not witnessed/admitted supports an impression that physicians apply diagnostic reasoning informed by knowledge of previously reported injury patterns. Concern for abuse is justified in patients who present with "the triad."

External Validation of the PediBIRN Screening Tool for Abusive Head Trauma in Pediatric Emergency Department Settings.

METHODS: We conducted a retrospective, secondary analysis of an existing, deidentified, prospective data set captured to derive a bruising CDR. Subjects were patients under 3 years with bruising and confirmed acute head trauma. An expert medical panel had previously identified patients with AHT. Measures of the CDR's AHT screening performance (sensitivity, specificity, likelihood ratios) were calculated with 95% confidence intervals (CIs). RESULTS: Expert medical panel members had classified 78 of 117 eligible patients (67%) as AHT, 38 (33%) as non-AHT, and 1 as indeterminate. Excluding the indeterminate case, the PediBIRN-4 demonstrated a sensitivity of 0.96 (95% CI, 0.88-0.99), specificity of 0.29 (95% CI, 0.16-0.46), positive likelihood ratio of 1.35 (95% CI, 1.10-1.67), and negative likelihood ratio of 0.13 (95% CI, 0.04-0.46). Close inspection of the data revealed that 1 of the CDR's predictor variables had lowered specificity without impacting sensitivity. Eliminating this variable would have increased specificity to 0.84 (95% CI, 0.68-0.93). CONCLUSIONS: The PediBIRN 4-variable CDR demonstrated AHT screening sensitivity in the pediatric ED equivalent to pediatric intensive care unit and other inpatient settings, but lower specificity. Further study of a simplified 3-variable PediBIRN AHT screening tool for the ED setting is warranted.

Screening for pediatric abusive head trauma: Are three variables enough?

BACKGROUND: The PediBIRN 4-variable clinical decision rule (CDR) detects abusive head trauma (AHT) with 96% sensitivity in pediatric intensive care (PICU) settings. Preliminary analysis of its performance in Pediatric Emergency Department settings found that elimination of its fourth predictor variable enhanced screening accuracy. OBJECTIVE: To compare the AHT screening performances of the "PediBIRN-4" CDR vs. the simplified 3-variable CDR in PICU settings. PARTICIPANTS AND SETTINGS: 973 acutely head-injured children <3 years hospitalized for intensive care across 18 sites between February 2011 and March 2021. METHODS: Retrospective, secondary analysis of the combined, prospective PediBIRN data sets. AHT definitional criteria and physicians' diagnoses were applied iteratively to sort patients into abusive vs. other head trauma cohorts. Outcome measures of CDR performance included sensitivity, specificity, predictive values, likelihood ratios, ROC AUC, and the correlation between each CDR's patient-specific estimates of AHT probability and the overall positive yield of patients' completed abuse evaluations. RESULTS: Applied accurately and consistently, both CDR's would have performed with sensitivity ≥93% and negative predictive value ≥91%. Eliminating the PediBIRN-4's fourth predictor variable resulted in significantly higher specificity (↑'d ≥19%), positive predictive value (↑'d ≥8%), and ROC AUC (↑'d ≥5%), but a 3% reduction in sensitivity. Both CDRs provided patient-specific estimates of abuse probability very strongly correlated with the positive yield of patients' completed abuse evaluations (Pearson's r = 0.95 and 0.91, p = .13). CONCLUSION: The PediBIRN 3-variable CDR performed with greater AHT screening accuracy than the 4-variable CDR. Both are good predictors of the results of patients' subsequent completed abuse evaluations.

Traumatic Head Injury and the Diagnosis of Abuse: A Cluster Analysis.

OBJECTIVES: Data guiding abusive head trauma (AHT) diagnosis rest on case-control studies that have been criticized for circularity. We wished to sort children with neurologic injury using mathematical algorithms, without reference to physicians' diagnoses or predetermined diagnostic criteria, and to compare the results to existing AHT data, physicians' diagnoses, and a proposed triad of findings. METHODS: Unsupervised cluster analysis of an existing data set regarding 500 young patients with acute head injury hospitalized for intensive care. Three cluster algorithms were used to sort (partition) patients into subpopulations (clusters) on the basis of 32 reliable (κ > 0.6) clinical and radiologic variables. P values and odds ratios (ORs) identified variables most predictive of partitioning. RESULTS: The full cohort partitioned into 2 clusters. Variables substantially (P < .001 and OR > 10 in all 3 cluster algorithms) more prevalent in cluster 1 were imaging indications of brain hypoxemia, ischemia, and/or swelling; acute encephalopathy, particularly when lasting >24 hours; respiratory compromise; subdural hemorrhage or fluid collection; and ophthalmologist-confirmed retinoschisis. Variables substantially (P < .001 and OR < 0.10 in any cluster algorithm) more prevalent in cluster 2 were linear parietal skull fracture and epidural hematoma. Postpartitioning analysis revealed that cluster 1 had a high prevalence of physician-diagnosed abuse. CONCLUSIONS: Three cluster algorithms partitioned the population into 2 clusters without reference to predetermined diagnostic criteria or clinical opinion about the nature of AHT. Clinical difference between clusters replicated differences previously described in comparisons of AHT with non-AHT. Algorithmic partition was predictive of physician diagnosis and of the triad of findings heavily discussed in AHT literature.

A Cluster Randomized Trial to Reduce Missed Abusive Head Trauma in Pediatric Intensive Care Settings.

OBJECTIVE: To estimate the impact of the PediBIRN (Pediatric Brain Injury Research Network) 4-variable clinical decision rule (CDR) on abuse evaluations and missed abusive head trauma in pediatric intensive care settings. STUDY DESIGN: This was a cluster randomized trial. Participants included 8 pediatric intensive care units (PICUs) in US academic medical centers; PICU and child abuse physicians; and consecutive patients with acute head injures <3 years (n = 183 and n = 237, intervention vs control). PICUs were stratified by patient volumes, pair-matched, and randomized equally to intervention or control conditions. Randomization was concealed from the biostatistician. Physician-directed, cluster-level interventions included initial and booster training, access to an abusive head trauma probability calculator, and information sessions. Outcomes included "higher risk" patients evaluated thoroughly for abuse (with skeletal survey and retinal examination), potential cases of missed abusive head trauma (patients lacking either evaluation), and estimates of missed abusive head trauma (among potential cases). Group comparisons were performed using generalized linear mixed-effects models. RESULTS: Intervention physicians evaluated a greater proportion of higher risk patients thoroughly (81% vs 73%, P = .11) and had fewer potential cases of missed abusive head trauma (21% vs 32%, P = .05), although estimated cases of missed abusive head trauma did not differ (7% vs 13%, P = .22). From baseline (in previous studies) to trial, the change in higher risk patients evaluated thoroughly (67%→81% vs 78%→73%, P = .01), and potential cases of missed abusive head trauma (40%→21% vs 29%→32%, P = .003), diverged significantly. We did not identify a significant divergence in the number of estimated cases of missed abusive head trauma (15%→7% vs 11%→13%, P = .22). CONCLUSIONS: PediBIRN-4 CDR application facilitated changes in abuse evaluations that reduced potential cases of missed abusive head trauma in PICU settings. TRIAL REGISTRATION: ClinicalTrials.gov: NCT03162354.

A changing history: When is it a red flag for child abuse?

BACKGROUND AND OBJECTIVE: While a changing history is frequently cited as a red flag for child abuse, no data support which changes are significant, nor the degree to which concern should be increased. We sought to measure the impact of changing caregiver histories on expert assessments of abuse likelihood. METHODS: We used a vignette survey to measure the impact of a changing history on child abuse expert assessments of abuse likelihood and willingness to undertake testing and protective interventions. By randomly varying the presence and magnitude of history changes, we determined their impact on perceived abuse likelihood. RESULTS: Of 494 invited participants, 267 (54 %) completed the survey. The presence of historical changes significantly affected experts' level of concern for abuse and willingness to test or report abuse, though to variable degrees. For example, while a minor change in the timing of an injury did not significantly increase willingness to perform a skeletal survey (OR: 1.5, 95 % CI: 0.8-2.9), a major change in the timing of an injury did (OR: 2.0, 95 % CI: 1.1-3.6). In addition, a change from having no initial history of trauma to then giving a history of accidental trauma significantly lowered the mean estimate of abuse likelihood and triggered significantly less reports to child protective services (OR: 0.02, 95 % CI: 0.003-0.2). CONCLUSION: For abuse experts, some history changes are more concerning than others, with major changes in history, and an initial denial of trauma having the largest impact. Future research regarding changing histories should consider details of the change, rather than treating all changes equally.

Validation of a Clinical Decision Rule to Predict Abuse in Young Children Based on Bruising Characteristics.

Importance: Bruising caused by physical abuse is the most common antecedent injury to be overlooked or misdiagnosed as nonabusive before an abuse-related fatality or near-fatality in a young child. Bruising occurs from both nonabuse and abuse, but differences identified by a clinical decision rule may allow improved and earlier recognition of the abused child. Objective: To refine and validate a previously derived bruising clinical decision rule (BCDR), the TEN-4 (bruising to torso, ear, or neck or any bruising on an infant <4.99 months of age), for identifying children at risk of having been physically abused. Design, Setting, and Participants: This prospective cross-sectional study was conducted from December 1, 2011, to March 31, 2016, at emergency departments of 5 urban children's hospitals. Children younger than 4 years with bruising were identified through deliberate examination. Statistical analysis was completed in June 2020. Exposures: Bruising characteristics in 34 discrete body regions, patterned bruising, cumulative bruise counts, and patient's age. The BCDR was refined and validated based on these variables using binary recursive partitioning analysis. Main Outcomes and Measures: Injury from abusive vs nonabusive trauma was determined by the consensus judgment of a multidisciplinary expert panel. Results: A total of 21 123 children were consecutively screened for bruising, and 2161 patients (mean [SD] age, 2.1 [1.1] years; 1296 [60%] male; 1785 [83%] White; 1484 [69%] non-Hispanic/Latino) were enrolled. The expert panel achieved consensus on 2123 patients (98%), classifying 410 (19%) as abuse and 1713 (79%) as nonabuse. A classification tree was fit to refine the rule and validated via bootstrap resampling. The resulting BCDR was 95.6% (95% CI, 93.0%-97.3%) sensitive and 87.1% (95% CI, 85.4%-88.6%) specific for distinguishing abuse from nonabusive trauma based on body region bruised (torso, ear, neck, frenulum, angle of jaw, cheeks [fleshy], eyelids, and subconjunctivae), bruising anywhere on an infant 4.99 months and younger, or patterned bruising (TEN-4-FACESp). Conclusions and Relevance: In this study, an affirmative finding for any of the 3 BCDR TEN-4-FACESp components in children younger than 4 years indicated a potential risk for abuse; these results warrant further evaluation. Clinical application of this tool has the potential to improve recognition of abuse in young children with bruising.

Estimating the Relevance of Historical Red Flags in the Diagnosis of Abusive Head Trauma.

OBJECTIVE: To replicate the previously published finding that the absence of a history of trauma in a child with obvious traumatic head injuries demonstrates high specificity and high positive predictive value (PPV) for abusive head trauma. STUDY DESIGN: This was a secondary analysis of a deidentified, cross-sectional dataset containing prospective data on 346 young children with acute head injury hospitalized for intensive care across 18 sites between 2010 and 2013, to estimate the diagnostic relevance of a caregiver's specific denial of any trauma, changing history of accidental trauma, or history of accidental trauma inconsistent with the child's gross motor skills. Cases were categorized as definite or not definite abusive head trauma based solely on patients' clinical and radiologic findings. For each presumptive historical "red flag," we calculated sensitivity, specificity, predictive values, and likelihood ratio (LR) with 95% CI for definite abusive head trauma in all patients and also in cohorts with normal, abnormal, or persistent abnormal neurologic status. RESULTS: A caregiver's specific denial of any trauma demonstrated a specificity of 0.90 (95% CI, 0.84-0.94), PPV of 0.81 (95% CI, 0.71-0.88), and a positive LR (LR+) of 4.83 (95% CI, 3.07-7.61) for definite abusive head trauma in all patients. Specificity and LR+ were lowest-not highest-in patients with persistent neurologic abnormalities. The 2 other historical red flags showed similar trends. CONCLUSIONS: A caregiver's specific denial of any trauma, changing history of accidental trauma, or history of accidental trauma that is developmentally inconsistent are each highly specific (>0.90) but may provide weaker support than previously reported for a diagnosis of abusive head trauma in patients with persistent neurologic abnormalities.

Estimating the probability of abusive head trauma after abuse evaluation.

BACKGROUND: Evidence-based, patient-specific estimates of abusive head trauma probability can inform physicians' decisions to evaluate, confirm, exclude, and/or report suspected child abuse. OBJECTIVE: To derive a clinical prediction rule for pediatric abusive head trauma that incorporates the (positive or negative) predictive contributions of patients' completed skeletal surveys and retinal exams. PARTICIPANTS AND SETTING: 500 acutely head-injured children under three years of age hospitalized for intensive care at one of 18 sites between 2010 and 2013. METHODS: Secondary analysis of an existing, cross-sectional, prospective dataset, including (1) multivariable logistic regression to impute the results of abuse evaluations never ordered or completed, (2) regularized logistic regression to derive a novel clinical prediction rule that incorporates the results of completed abuse evaluations, and (3) application of the new prediction rule to calculate patient-specific estimates of abusive head trauma probability for observed combinations of its predictor variables. RESULTS: Applying a mean probability threshold of >0.5 to classify patients as abused, the 7-variable clinical prediction rule derived in this study demonstrated sensitivity 0.73 (95% CI: 0.66-0.79) and specificity 0.87 (95% CI: 0.82-0.90). The area under the receiver operating characteristics curve was 0.88 (95% CI: 0.85-0.92). Patient-specific estimates of abusive head trauma probability for 72 observed combinations of its seven predictor variables ranged from 0.04 (95% CI: 0.02-0.08) to 0.98 (95% CI: 0.96-0.99). CONCLUSIONS: Seven variables facilitate patient-specific estimation of abusive head trauma probability after abuse evaluation in intensive care settings.

Racial and Ethnic Disparities and Bias in the Evaluation and Reporting of Abusive Head Trauma.

OBJECTIVE: To characterize racial and ethnic disparities in the evaluation and reporting of suspected abusive head trauma (AHT) across the 18 participating sites of the Pediatric Brain Injury Research Network (PediBIRN). We hypothesized that such disparities would be confirmed at multiple sites and occur more frequently in patients with a lower risk for AHT. STUDY DESIGN: Aggregate and site-specific analysis of the cross-sectional PediBIRN dataset, comparing AHT evaluation and reporting frequencies in subpopulations of white/non-Hispanic and minority race/ethnicity patients with lower vs higher risk for AHT. RESULTS: In the PediBIRN study sample of 500 young, acutely head-injured patients hospitalized for intensive care, minority race/ethnicity patients (n = 229) were more frequently evaluated (P < .001; aOR, 2.2) and reported (P = .001; aOR, 1.9) for suspected AHT than white/non-Hispanic patients (n = 271). These disparities occurred almost exclusively in lower risk patients, including those ultimately categorized as non-AHT (P = .001 [aOR, 2.4] and P = .003 [aOR, 2.1]) or with an estimated AHT probability of ≤25% (P <.001 [aOR, 4.1] and P <.001 [aOR, 2.8]). Similar site-specific analyses revealed that these results reflected more extreme disparities at only 2 of 18 sites, and were not explained by local confounders. CONCLUSION: Significant race/ethnicity-based disparities in AHT evaluation and reporting were observed at only 2 of 18 sites and occurred almost exclusively in lower risk patients. In the absence of local confounders, these disparities likely represent the impact of local physicians' implicit bias.

Classifying Injuries in Young Children as Abusive or Accidental: Reliability and Accuracy of an Expert Panel Approach.

OBJECTIVE: To assess interrater reliability and accuracy of an expert panel in classifying injuries of patients as abusive or accidental based on comprehensive case information. STUDY DESIGN: Data came from a prospective, observational, multicenter study investigating bruising characteristics of children younger than 4 years. We enrolled 2166 patients with broad ranges of illnesses and injuries presenting to one of 5 pediatric emergency departments in whom bruises were identified during examination. We collected comprehensive data regarding current and past injuries and illnesses, and provided deidentified, standardized case information to a 9-member multidisciplinary panel of experts with extensive experience in pediatric injury. Each panelist classified cases using a 5-level ordinal scale ranging from definite abuse to definite accident. Panelists also assessed whether report to child protective services (CPS) was warranted. We calculated reliability coefficients for likelihood of abuse and decision to report to CPS. RESULTS: The interrater reliability of the panelists was high. The Kendall coefficient (95% CI) for the likelihood of abuse was 0.89 (0.87, 0.91) and the kappa coefficient for the decision to report to CPS was 0.91 (0.87, 0.94). Reliability of pairs and subgroups of panelists were similarly high. A panel composite classification was nearly perfectly accurate in a subset of cases having definitive, corroborated injury status. CONCLUSIONS: A panel of experts with different backgrounds but common expertise in pediatric injury is a reliable and accurate criterion standard for classifying pediatric injuries as abusive or accidental in a sample of children presenting to a pediatric emergency department.