Bruising in Children: Practice Patterns of Pediatric Hematologists and Child Abuse Pediatricians

Abstract

The evaluation for children with bruising may be affected by the specialty to which they are referred. We conducted a 3-year retrospective review of subjects referred for bruising to Child Abuse Pediatrics (CAP) or Pediatric Hematology to identify characteristics associated with referral to each specialty and to compare the diagnostic evaluations and diagnoses based on specialty. Of 369 subjects, 275 were referred to CAP and 94 to Hematology. Clinical exam findings were similar in both groups. Hematology referrals were significantly more likely to have laboratory evaluations. Among those referred to CAP, 9.5% had head computed tomography scans and 27.3% had skeletal surveys. No children referred to Hematology had these imaging studies performed. Hematology never diagnosed child physical abuse, and CAP never diagnosed bleeding disorders. Pediatric hematologists and CAPs perform different evaluations and reach different diagnostic conclusions for similar patients with bruising. Further investigation of these practice patterns is warranted.

Introduction

Bruising is the most common presenting feature of child physical abuse and is also a common presenting symptom of bleeding disorders.^1–4 Distinguishing between bruising due to abuse and that due to a bleeding disorder can be challenging, as bruising due to bleeding disorders can present in a manner that is nearly indistinguishable from that due to abuse.⁵ As such, children with bruising as the primary clinical finding often require thorough medical evaluations for both bleeding disorders and abuse.⁶

Social and clinical factors may play a role in referral to medical specialists for patients with bruising and in the evaluations performed and diagnoses made by these specialists. Previous studies have demonstrated disparities with respect to race/ethnicity and socioeconomic status (SES) factors in referral patterns to Children’s Protective Services (CPS) and to medical providers.^7–9 Additionally, multiple studies have shown that SES and/or race/ethnicity may affect the evaluation and/or diagnosis of child physical abuse.^10–12 Children with bruising may be referred to a specialist, such as a pediatric hematologist or a child abuse pediatrician (CAP), and differing expertise and experience based on specialty of training may have an impact on evaluations and end diagnoses provided by these specialists.

Currently, little has been published about referral patterns to pediatric hematologists and CAPs for bruising in children, and how pediatric hematologists and CAPs may differ in their evaluations and diagnoses of children with bruises. The objectives of this study are to (a) identify factors associated with referral to CAP versus Pediatric Hematology and (b) compare CAP and Pediatric Hematology evaluations and diagnoses in children with bruising as the primary clinical finding.

Methods

Subjects and Setting

The study was conducted at a large urban tertiary academic children’s hospital with a catchment area of approximately 5 million people. All specialty clinics, including the CAP and Pediatric Hematology clinics, serve the same population. Additionally, all specialty clinics receive referrals through a centralized hospital intake service. The intake service does not choose the specialty clinic for a given referral; rather, the specialty clinic is chosen by the person (parent, medical provider, or other person) making the referral.

The authors reviewed charts for subjects birth to18 years of age that were referred to CAP or Pediatric Hematology solely for bruising from January 1, 2008, to December 31, 2010. Subjects were identified through the CAP physical abuse clinic log and the Pediatric Hematology clinic log. Exclusion criteria included patients referred for other bleeding manifestations (eg, menorrhagia, oral bleeding after dental surgery, intracranial hemorrhage), patients with a preexisting hematologic diagnosis, and patients with other exam findings concerning for physical abuse (eg, intracranial hemorrhage, fractures, burns, patients with concerns of sexual abuse).

All subjects referred to Pediatric Hematology for bruising were evaluated by 1 of 18 board-eligible/board-certified pediatric hematologist/oncologists. All subjects referred to CAP for bruising were evaluated by 1 of 4 board-eligible/board-certified child abuse pediatricians. This study was approved by the hospital institutional review board.

Procedures

The authors collected subject data from chart review including the following: age (including classifying <9 months of age, as bruising in children <9 months of age is concerning for abuse)¹³; gender; self-reported race/ethnicity; primary language; insurance status; family history of a known, named bleeding disorder (as provided by the family); referral source; referral specialty (either CAP or Pediatric Hematology inpatient or outpatient clinic); disclosure of abuse; prior labs obtained; specialty laboratory evaluation and radiographic evaluation; referral to CPS; anatomic location of bruising and presence of patterned or object bruising on physical exam; and final diagnosis regarding bruising as documented directly in the medical record by CAP or Pediatric Hematology.

Data Collection Definitions

Further definitions for data collection categories are as follows. “Referral source” was classified as “medical” or “nonmedical.” “Medical” referral sources included primary care physicians, emergency departments, and other medical personnel. “Nonmedical” referral sources included CPS, police, schools, self-referrals/referrals due to parental concern, and all other nonmedical sources. A “disclosure of abuse” was considered to be present when (a) the child verbally reported physical abuse directly to medical providers, CPS personnel, or other referral source, or (b) someone witnessed the child being abused. “Prior laboratories obtained” was defined as having a laboratory evaluation by the referral source specifically due to concerns of this episode of bruising. Prior laboratory evaluation could consist of any blood work obtained to evaluate bruising and must have been documented by CAP or Pediatric Hematology in the medical record. “Specialty laboratory evaluation” performed by CAP or Pediatric Hematology was classified as follows:

• Coagulation testing (coags)—prothrombin time and/or activated partial thromboplastin time

• Platelets—platelet count

• Liver function tests (LFTs)—including total protein, albumin, total/direct/indirect bilirubin, aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase

• Other labs obtained—other hematologic testing, such as coagulation factor levels, von Willebrand antigen, platelet function assay, and other platelet testing (but not including coags, platelets, or LFTs)

“Referral to CPS” was considered to have occurred if a subject was referred to CPS during any point of the evaluation (by medical personnel or otherwise) for this specific instance of bruising.

Bruising

All children evaluated for bruising by either CAP or Pediatric Hematology received a complete physical exam including a detailed skin exam identifying and characterizing bruising, with bruising location and characteristics recorded in the medical record. The authors constructed a classification tool to identify bruising concerning for abuse (BCFA) using data from previously published studies.^12–14 This tool has been used in previous similar work.⁵ Table 1 summarizes characteristics of BCFA. The authors used location and characteristics of bruising as documented in the medical record, and the classification tool to classify bruising as either “BCFA” or “bruising not concerning for abuse.” “Final diagnosis regarding bruising” was classified as child physical abuse, bleeding disorder, or other (encompassing all diagnoses other than bleeding disorders and child physical abuse). Diagnoses regarding child physical abuse and bleeding disorders in this study were based on the diagnosis documented from CAPs and pediatric hematologists.

Table 1.

Classification of Bruising Regarding Concern for Abuse.^12–14

Concerning for Abuse	Not Concerning for Abuse
Face (not including forehead)	Head, not including forehead, face, or ear
Ear	Forehead
Neck	Oral
Proximal arms (above the elbows)	Distal arms (including elbow)
Wrist/hand	Spinous/paraspinous
Anterior or lateral chest	Hips
Abdomen	Anterior and posterior legs (shins, knees, and calves)
Back (excluding spinous/paraspinous)	Ankle
Buttock/genitalia
Anterior/posterior thigh
Foot
Patterned/symmetric
Any bruising in a child <9 months of age

Statistical Analysis

Analysis of Associations of Subject Characteristics and Referral to CAP Versus Pediatric Hematology.

Descriptive statistics of the subjects and their evaluations and final diagnoses were calculated. Then, χ² and Fisher’s exact tests were used to evaluate associations between subject characteristics and referral to CAP versus Pediatric Hematology. Results of the χ² and Fisher’s exact tests were used to identify covariates with a higher likelihood of having independent associations with the outcome measures (P < .25).¹⁵ Exact logistic regression was used to determine the associations of the previously identified higher-likelihood covariates with referral to CAP versus Pediatric Hematology. Exact logistic regression was used because it allows for the inclusion of fields where there is a variable with few or no observations. To assess for multicollinearity (a high correlation between 2 or more predictor variables in a multiple regression that may affect reliability of the regression model), tolerance and variance inflation factor (VIF) were calculated. A tolerance of <0.4 and/or a VIF of >10 indicate high multicollinearity.^16,17 The logistic regression model to discriminate subjects referred to CAP versus subjects referred to Pediatric Hematology was assessed using Hosmer and Lemeshow goodness-of-fit C-statistic (area under the receiver operative characteristic curve).¹⁸ A C-statistic greater than 0.80 indicates good discriminatory ability.¹⁸ Subjects with missing data were excluded from analysis for only the specific missing data point when computing χ² and Fisher’s exact tests. However, subjects with missing data were excluded from the entire exact logistic regression analysis.¹⁹

Analysis of Associations of Medical Evaluation and Diagnoses Based on Referral Specialty.

Chi-square and Fisher’s exact tests were used to compare the laboratory and radiological evaluations in children based on referral specialty (CAP vs Pediatric Hematology). Again, subjects with missing data were excluded from analysis for only the specific missing data point when computing χ² and Fisher’s exact tests.

Statistics were calculated using SPSS, version 18, and SAS, version 9.2.

Results

Associations of Subject Characteristics and Referral to CAP Versus Pediatric Hematology

Figure 1 illustrates the subjects that were included and excluded from the study. There were 927 individual subjects in the CAP physical abuse clinic log and 926 subjects in the Pediatric Hematology clinic log. Of those excluded from CAP (n = 652 or 70.3%), the most common reasons were referral for sexual abuse or presence of other exam findings (eg, burns, fractures). Of those excluded from Pediatric Hematology (n = 832 or 89.8%), the most common reasons were referral for other hematological concerns (eg, epistaxis, menorrhagia).

Figure 1.

Subjects included and excluded.

Table 2 describes characteristics of subjects referred to CAP and Pediatric Hematology. Overall, 53.7% (n = 198) were male, 60% (n = 220) were Caucasian, 92.1% (n = 340) spoke English as their primary language, and 82.4% (n = 304) had BCFA. Subjects referred to either specialty were similar in mean age (CAP 4.7 years vs Hematology 5.1 years), age range (CAP 8 days to 17 years vs Hematology 10 days to 16 years), and proportion who were <9 months of age (CAP 12% vs Hematology 14.9%).

Table 2.

Comparisons of Characteristics of Subjects Based on Referral to Child Abuse Pediatrics Versus Pediatric Hematology Using χ² and Fisher’s Exact Tests.

Characteristic	CAP (n = 275)	Pediatric Hematology (n = 94)	P Valuea
Age
>9 months	242 (88%)	80 (85.1%)	.47
Sex
Male	145 (52.7%)	53 (56.4%)	.54
Race/ethnicityb
White	147 (55.3%)	73 (84.9%)	<.001
Languagec
English	252 (92.3%)	88 (94.6%)	.45
Insuranced
Private	57 (21.3%)	56 (59.6%)	<.001
Family history
Family history of bleeding disorder	1 (0.4%)	19 (20.2%)	<.001
Laboratory
Prior labs obtained	25 (9.1%)	71 (75.5%)	<.001
Referral sourcee
Medical	79 (33.6%)	66 (85.7%)	<.001
Disclosure
Disclosure of abuse	114 (41.5%)	0	<.001
Bruisingf
With BCFA	242 (88%)	62 (68.9%)	<.001

Abbreviations: CAP, Child Abuse Pediatrics; BCFA, bruising concerning for abuse.

^aP values in boldface indicate that the variable was included in the regression analysis.

^bRace/ethnicity category with 9 data points missing from CAP and 8 from Pediatric Hematology.

^cLanguage category with 2 data points missing from CAP and 1 from Pediatric Hematology.

^dInsurance category with 6 data points missing from CAP.

^eReferral source category with 40 data points missing from CAP and 17 from Pediatric Hematology.

^fBruising category with 4 data points missing from Pediatric Hematology.

Initial bivariate analysis identified 7 covariates (P < .25) as “higher-likelihood” for inclusion in logistic regression (Table 2). Logistic regression identified the following 4 variables as independently associated with referral to CAP: no family history of a bleeding disorder (odds ratio [OR] = 132.71; 95% confidence interval [CI] = 3.18->999), no prior labs obtained (OR = 201.9; 95% CI = 46.23->999), referral by a nonmedical source (OR = 29.09; 95% CI = 7.44–174.09), and children with BCFA (OR = 8.91; 95% CI = 2.02–47.44). Disclosure of abuse was not statistically significant (OR = 7.40; 95% CI = 0.95->999). These 4 variables have a low multicollinearity, or are not highly correlated with one another, with a tolerance of each of the variables of >0.89 and a VIF of each of the variables of <1.2. The C-statistic shows that the logistic regression model for children with bruising referred to CAP versus Pediatric Hematology has a good discriminatory ability with a value of 0.96.

Associations of Medical Evaluation and Diagnoses Based on Referral Specialty

Table 3 depicts the differences in the evaluation and diagnoses of patients based on referral specialty. Subjects referred to Pediatric Hematology were significantly more likely to have laboratory evaluations in all of the categories analyzed. The most common specific bleeding disorder diagnosis made by Pediatric Hematology was idiopathic thrombocytopenic purpura (n = 46). The other bleeding disorders diagnosed by Pediatric Hematology were hemophilia A (n = 3), Von Willebrand disease (n = 3), platelet function defect (n = 2), platelet secretion defect (n = 1), and bleeding disorder not otherwise specified (n = 1). Diagnoses, made by either specialty, that were in the “other” category include accidental injury, “medical evaluation for physical abuse” in which no diagnosis was made, bruising secondary to nonsteroidal anti-inflammatory drugs, Depakote use causing thrombocytopenia, bruising not otherwise specified, aplastic anemia, dermal melanosis, hyperpigmentation, irritant dermatitis, subcutaneous fat necrosis of the newborn, and capillaritis.

Table 3.

Comparisons of Evaluation and Diagnoses of Subjects Referred to Child Abuse Pediatrics Versus Pediatric Hematology Using χ² and Fisher’s Exact Tests.

	CAP (n = 275)	Pediatric Hematology (n = 94)	P Value
Referral to CPS
Yes	258 (93.8%)	0	<.001
Radiology
CT head	26 (9.5%)	0	<.001
Skeletal survey	75 (27.3%)	0	<.001
Laboratorya
Coags and/or platelets	80 (29.1%)	62 (66%)	<.001
LFTs	28 (10.2%)	23 (24.5%)	.001
Other labs obtained	83 (30.2%)	82 (87.2%)	<.001
Diagnosis
Child physical abuse	168 (61.1%)	0
Bleeding disorder	0	56 (59.6%)
Other	107 (38.9%)	38 (40.4%)

Abbreviations: CAP, Child Abuse Pediatrics; CT, computed tomography; LFT, liver function test.

^aCoags—prothrombin time or partial thromboplastin time. Platelets—platelet count. LFTs—including total protein, albumin, total/direct/indirect bilirubin, aspartate aminotransferase, alanine aminotransferase, and alkaline phosphatase.

When specifically looking at children <9 months of age referred for bruising (n = 47), 14 were seen in Pediatric Hematology and 9 (64.3%) were diagnosed with a bleeding disorder. None of this subgroup evaluated by Pediatric Hematology had a head computed tomography (CT) or skeletal survey. The 5 children not diagnosed with a bleeding disorder by Pediatric Hematology were diagnosed with “unknown rash,” “no concerns at this time,” “anemia and thrombocytopenia,” “excessive bruising”, and “no bleeding disorder.” Among the 33 children <9 months of age referred to CAP, 20 (60.6%) had a head CT, 30 (90.9%) had a skeletal survey obtained, and 13 (39.4%) were diagnosed with child physical abuse. Other diagnoses made by CAP in children <9 months of age with bruising include “possible nonaccidental trauma/child abuse,” “medical evaluation for child maltreatment/suspected physical abuse,” “bruise,” and “Mongolian spots.”

Subjects Evaluated by Both CAP and Hematology

Five children were seen by both CAP and Pediatric Hematology. Table 4 illustrates these 5 subjects. Four of the 5 subjects had BCFA. All subjects were first evaluated by CAP and then referred to Pediatric Hematology. Three of the children did not have a change in their final diagnosis whereas 1 had a diagnosis of platelet function defect and 1 child had a diagnosis “concerning for” Henoch–Schonlein purpura made by the Pediatric Hematologist.

Table 4.

Description of Subjects Seen by Both Child Abuse Pediatrics and Pediatric Hematology.

Age	Gender	Race	Insurance	Bruising	Diagnosis
4 years	Female	White	Private	Petechiae to right upper extremity in linear pattern; scattered petechiae under eyes, on back, on abdomen; bruising to iliac crest and lower extremities	Increased bruising and petechiae
8 months	Female	Non-White	Nonprivate	Bruising to bilateral lower extremities that covers multiple planes; right elbow with 2 bruises	Concern for Henoch–Schonlein purpura
9 years	Male	White	Nonprivate	Bruise to face, right upper arm; bruising from axilla down rib cage; bruise to mid-forearm extending to the upper arm and is sleeve-like; left wrist bruise	No concern for bleeding disorder
3 years	Female	Non-White	Nonprivate	Left buttock, back, and posterior/lateral thigh bruising	No concern for bleeding disorder
5 years	Male	White	Private	Bruise over hip	Platelet function defect

Discussion

This study identified characteristics associated with referral to CAP versus Pediatric Hematology in children with bruising as the primary clinical finding and identified significant differences in the subsequent medical evaluations and diagnoses based on referral specialty. The findings suggest that CAPs and Pediatric Hematologists perform different evaluations and reach different diagnostic conclusions in broadly similar patients regarding child physical abuse and bleeding disorders. Comparison of evaluations and diagnoses between the 2 groups suggest that the referral specialty may play a significant role in the end diagnosis in some children referred for bruising.

Social factors, such as race/ethnicity, insurance status, or primary language, were not found to be driving factors in the referral process. This stands in contrast to other studies suggesting that SES and race/ethnicity influenced CPS and medical evaluations in cases that may be concerning for abuse.^7,10,11 This may be due to the fact that prior studies were not focused solely on bruising, and involved other injury types, such as skull fractures and extremity fractures.^7,10 Characteristics such as “family history of a bleeding disorder” are not applicable to these other conditions. Additionally, children with injuries requiring medical attention, such as fractures, are more likely to be brought to medical attention immediately. Children with bruising as the primary finding may not require medical therapies, and thus may be less likely to be brought to an emergency room and more likely to be referred to CAP and/or Pediatric Hematology as outpatients. Thus, our findings may reflect differing medical-investigative management based on a child’s primary medical complaint.

Four variables were found to be independently associated with referral to CAP (BCFA, no family history of a bleeding disorder, no prior labs obtained, and referral by a nonmedical source). The lack of multicollinearity indicates that these 4 variables were likely independently associated with referral to CAP. The association of BCFA and lack of family history with CAP referral suggests that many referrals to CAP (as opposed to Pediatric Hematology) are likely appropriate. Disclosure of abuse was not found to be independently associated with referral specialty, though referral to CAP did approach significance for these subjects. However, all children who disclosed abuse were referred to CAP. It is possible that disclosure of abuse may actually be associated with referral to CAP, and that this could be statistically supported by increasing the precision of the measurement of the association by increasing sample size. However, our current data do not support such an association.

Our data suggest that CAPs and pediatric hematologists perform different laboratory and radiology evaluations and reach different diagnostic conclusions in the population of children with bruising. Additionally, “cross-referrals” in our population were rare. BCFA may be due to accidental or medical causes; however, in this study, pediatric hematologists never diagnosed child physical abuse, nor referred a child with bruising to CAP. Fourteen (14.9%) of the children with bruising who were <9 months of age were referred to Pediatric Hematology, yet none received any radiographic screening tests for occult injury (skeletal survey, head CT), as recommended for many children in this population.²⁰ It may not be the purview of a pediatric hematologist to conduct child abuse evaluations; however, as none of these patients were referred to CAP, this suggests that child abuse may not be “on the radar” of some pediatric hematologists. Additionally, CAP never diagnosed a child referred for bruising with a bleeding disorder. Coagulation testing was performed much less frequently in CAP referrals, and only 5 children were referred from CAP to Pediatric Hematology for abnormal laboratory evaluations during the 3-year study period. The findings of this study suggest that greater CAP–Hematology collaborative evaluations in children with bruising may be needed to decrease variability in evaluations and diagnoses. The recently published American Academy of Pediatrics Technical and Clinical Reports may assist in increasing collaborative practice.^6,21

Four of the 5 subjects in the study that were seen by both specialties had BCFA and were referred from CAP to Pediatric Hematology due to abnormal laboratory evaluations concerning for possible bleeding disorders. In 2 cases, the pediatric hematologist made a change in diagnosis—1 child was diagnosed with a platelet function defect and 1 child was diagnosed with “concern for Henoch–Schonlein purpura.” In the case of the 5-year-old male diagnosed with a platelet function defect, there was also a positive disclosure of abuse (“thrown to the ground, shoved an egg is his mouth, and kicked by stepdad”), thus illustrating that a diagnosis of a bleeding disorder and child physical abuse are not mutually exclusive. In 3 cases there was no bleeding disorder detected by Pediatric Hematology; however, there was no referral back to CAP. Thus, these 3 children had BCFA but none had appropriate follow-up or a final diagnosis. These 5 cases demonstrate the value of collaborative evaluations and the need for consistent follow-up of cases.

Our study has limitations. First, this is a single-institution study reflecting the practice in an urban, academic children’s hospital with a regional referral base. The findings may not be applicable to other settings with different practitioners. However, this study aligns with previous work showing variability in evaluations and diagnoses in cases that may be concerning for abuse.^7,10,11 Second, this study was not able to collect data on phone and/or “curbside” consultations regarding subjects. As such, it is possible that abuse may have been considered by Pediatric Hematology and bleeding disorders by CAP in more cases than reflected in the data. Third, our data had variables with missing data. Four variables had <15 data points each that were missing and thus would less likely affect our final analysis. However, referral source had 57 missing data points and these subjects were unable to be utilized in our analysis, including our logistic regression. If referral source was known in each subject, our results may have differed. Finally, other factors not considered by this study may contribute to referral specialty. However, the regression analyses in this study achieved high areas under the curve with a C-statistic of 0.96, suggesting that most important predictors were likely included.

Conclusion

Pediatric hematologists and CAPs perform different radiographic and laboratory evaluations and reach different diagnostic conclusions in similar children with bruising. Careful consideration of referrals by medical and nonmedical specialists and, in some cases, collaborative evaluations by CAPs and pediatric hematologists may be warranted to improve care for this population.