New Screening Approach Helps Doctors Identify Potential Child Abuse Cases

Twenty years ago, child abuse pediatrician Kent Hymel worked with Carole Jenny on a case involving a child who had suffered severe head trauma. This child had been evaluated nine times before he was correctly diagnosed with head trauma resulting from child abuse. This case motivated Dr. Hymel to create an evidence-based screening approach that could help pediatricians miss or misdiagnose fewer children with abusive head trauma. As they looked into the diagnostic and screening challenges associated with abusive head trauma, Drs. Jenny and Hymel found that 30 percent of their local cases of abusive head trauma had been missed or misdiagnosed. While those numbers have very likely improved since the 1990s, there is still an unacceptably high rate of screening and diagnostic errors. Dr. Hymel, a co-funded faculty member with the Child Maltreatment Solutions Network and Penn State Hershey Children’s Hospital, noted that missing these cases can have severe consequences. “Many of the children whose cases were missed went on to suffer more abuse,” he said. “Five children with missed cases died. Four could have been prevented. A gold standard for the diagnosing of abusive head trauma does not exist. When I considered these numbers, it became my motivation to reduce these tragedies.” To avoid missing cases, Dr. Hymel set out to create a clinical prediction rule—that is, an effective screening test that would help identify abuse among head trauma victims. He and his co-investigators identified four predictor variables that effectively “screen in” the diagnosis. According to Dr. Hymel, a child who is admitted to an intensive care unit and meets any of the following criteria should be thoroughly evaluated for child abuse: (1) Did the child stop breathing? (2) Was there bruising on the torso, ears, or neck? (3) Was there subdural bleeding? (4) Were there complex skull fractures? “Using the prediction rule, we can assign an evidence-based estimate on the probability of abuse,” Dr. Hymel said. “However, this is a screening, not a diagnostic test. It’s designed to cast a wide net so doctors don’t miss cases. If the test is applied accurately, the patients can be carefully assessed through the medical and child protection systems to see if abuse is substantiated.” To test the performance of their new screening tool approach, Dr. Hymel and his co-investigators collected clinical, historical, and radiologic data on 291 additional, acutely head-injured children from 14 pediatric intensive care units (PICU).  Their analyses revealed that their clinical prediction rule (i.e., if any of the four criteria are present in the ICU) accurately predicted 98 percent of the cases where abusive head trauma was ultimately diagnosed. Some children who did not experience abuse were identified as potential victims of abusive head trauma (called a false positive result). However, the goal of Dr. Hymel’s rule is to avoid missing the abuse cases. Even though the screening may result in false positive abuse predictions, it is better to error on the side of caution. “No screening approach is ever completely accurate in its predictions. We use them to help focus clinical attention and to improve the accuracy of the screening process. Our goal was to miss fewer cases” said Dr. Hymel. “In addition to missing fewer cases of abuse, there are several reasons for physicians to use this new screening tool. Perhaps most importantly, the prediction rule provides objective data that will help physicians be more accurate when deciding to proceed with an evaluation for child abuse. It promotes a consistent process that is backed by evidence and addresses a complex problem.” Dr. Hymel added that without this clinical prediction rule, there is no data driving the decision to launch or forgo an evaluation for child abuse. Instead, he says, “there is intuition and the doctor’s experience or, in many cases inexperience, but no data. That is the void we are trying to fill.” Because the screening performance of Dr. Hymel’s clinical prediction rule has now been validated in a new patient population, the next step is to measure its clinical impact when doctors begin to use it in actual practice. Dr. Hymel has begun work on an impact study that will measure the tool’s screening performance in actual clinical practice, and determine the best approaches for getting physicians to adopt the rule as a screening tool. Dr. Hymel leads the Pediatric Brain Injury Research Network, a group of pediatric investigators that assisted in this research. Dr. Hymel can be contacted at khymel@hmc.psu.edu and this study can be found in the December 2014 issue of Pediatrics.