BACKGROUND

Child maltreatment is a leading cause of morbidity and mortality in the United States. Many children who are injured or die from abuse have been previously evaluated by a physician. Numerous studies demonstrate that physicians fail to consistently screen for abuse and that persistent and pervasive disparities in screening exist related to patient and hospital characteristics. The electronic medical record (EMR) is useful in improving screening rates in a wide variety of diseases, thereby allowing for early intervention and improved outcomes.

OBJECTIVES

1. To develop and validate an EMR-based Child Abuse Clinical Decision Support System (CA-CDSS) in a pediatric hospital emergency department (ED), determine whether physician compliance with evidence-based guidelines for evaluation of child physical abuse improves with CA-CDSS, and compare compliance rates by patient and provider characteristics when physicians do and do not receive CA-CDSS. 2. To develop a CA-CDSS in the EMR at 2 general EDs and determine whether physician compliance with evidence-based guidelines for evaluation of child physical abuse improves with CA-CDSS. 3. To assess whether embedding a validated 5-item child abuse screen (CAS) in the EMR in 13 general EDs increases identification and reporting of suspected abuse.

METHODS

OBJECTIVE 1: To identify children <2 years old at risk for physical abuse, we coded 30 triggers into the EMR at a pediatric hospital ED. We measured characteristics of the trigger system (eg, accuracy) using the decision of a multidisciplinary child protection team as the reference standard for diagnosis followed by a 7-month randomized controlled trial (RCT) to assess physician compliance with American Academy of Pediatrics (AAP) guidelines for physical abuse evaluation. For patients randomized to be cases, the treating physician received a pop-up alert containing a link to the physical abuse–specific order set. For patients randomized to be controls, no alert or link appeared, but physicians could search for the order set in the order catalog. Compliance with AAP guidelines was compared between groups, and associations with patient and physician demographics were evaluated. OBJECTIVE 2: We developed a trigger system at 2 general EDs using a similar approach to the pediatric ED. A 9-month baseline period was followed by a 10-month postintervention period. During the baseline period, physicians were unaware of the CA-CDSS. During the postintervention period, providers received a pop-up alert for patients who triggered; the alert suggested use of the physical abuse–specific order set. OBJECTIVE 3: A prospective observational study compared rates of reports to Child Protective Services (CPS) among children who did and did not have a completed 5-item CAS. This was done in all 13 general EDs, including the 2 described above.

RESULTS

OBJECTIVE 1: Sensitivity and specificity of the trigger system for identifying children <2 years old with physical abuse were 96.8% (95% CI, 92.4%-100.0%) and 98.5% (95% CI, 98.3%-98.7%), respectively. Positive and negative predictive values were 26.5% (95% CI, 21.2%-32.8%) and 99.9% (95% CI, 99.9%-100.0%), respectively. Compliance with AAP guidelines was high in both cases and controls without a difference between groups. Physicians were more likely to be compliant when patients had public insurance and if physicians were pediatric ED fellowship trained, had more than 10 years' experience, or were male. OBJECTIVE 2: A total of 242 children <2 years old triggered the CA-CDSS, with 86 during baseline and 156 during the postintervention. Of the triggers, 81% (195 of 242) were considered appropriate (eg, not overtriggers). Compliance with the AAP guidelines was low and did not change in the postintervention group. OBJECTIVE 3: A 5-item CAS was completed in 11 612 children, with a completion rate of 68%; 1.9% were positive. The rate of reports to CPS was higher when children were screened than when they were not (1.3% vs 0.5%; < .0001). No difference in the screening rates occurred based on patient or hospital characteristics.

CONCLUSIONS

OBJECTIVE 1: An EMR-based trigger system can identify young children who need to be evaluated for physical abuse. The lack of a difference in compliance with AAP guidelines in the RCT was likely because of the high baseline compliance and patient-level randomization, which allowed for contamination of the groups. The rapid uptake of the child abuse–specific order sets demonstrated the high acceptance of this clinical decision support. OBJECTIVE 2: The much lower baseline compliance with AAP guidelines at the general EDs was expected, but the low compliance after implementation of the CA-CDSS and the lack of use of the physical abuse–specific order set was disappointing. Different approaches may need to be used in general EDs vs pediatric EDs to improve compliance with AAP guidelines. OBJECTIVE 3: An EMR-based child abuse screening tool can be successfully implemented across multiple locations. The increasing screening rate over the course of the study suggests clinical acceptability. The lack of a difference in the odds of screening or reports to CPS according to race or income is encouraging.

LIMITATIONS AND SUBPOPULATION CONSIDERATIONS

OBJECTIVE 1: The quality of the reference standard, which was the decision of a multidisciplinary child protection team, may have provided an overestimate of sensitivity because not every child evaluated in the ED undergoes evaluation by the child protection team. The design allowed for group contamination and limited the ability to detect group differences. OBJECTIVE 2: The before-and-after design was a limitation, although this would have been more significant had we seen group differences. OBJECTIVE 3: The observational study design was a limitation; there could have been differences between the children who were and were not screened, which may have affected the likelihood of reporting to CPS.