233 - Developing an ICD Coding Schema to Estimate the Burden of Physical Abuse in Emergency Departments

Poster Number: 233
Publication Number: 233.402

Farah Brink, Nationwide Children's Hospital, Columbus, OH, United States; Charmaine Lo, Nationwide Children's Hospital, Columbus, OH, United States; Junxin Shi, Nationwide Children's Hospital, Columbus, OH, United States; Rachel Stanley, Nationwide Childrens Hospital, columbus, OH, United States; Daniel Lindberg, University of Colorado School of Medicine, Denver, CO, United States

Background: International Classification of Diseases (ICD) billing codes are commonly used to measure disease prevalence but have been shown to be insensitive for physical abuse. This phenomenon is likely more pronounced for emergency department (ED) patients, as the final diagnosis of abuse is often made only after ED ICD codes are assigned, when a patient is admitted to the hospital or discharged. Traumatic injuries highly associated with abuse may be reliably coded in the ED. Our objective was to identify ICD codes that could be used to estimate physical abuse prevalence in an ED population.

Objective: 1. Determine the accuracy of commonly used abuse-specific ICD codes for physical abuse in an ED population. 2. Identify ICD injury codes associated with abuse.

Design/Methods: This retrospective study identified children < 5 years determined by the child protection team (CPT) to be physically abused at a large Midwestern children’s hospital from 2016-2020. The Pediatric Health Information System (PHIS) database was utilized to identify ED and discharge ICD codes for ED visits of children < 5 years. Abuse-specific codes were defined as ICD-10-CM and SNOMED codes included in the Centers for Disease Control and Prevention (CDC) child abuse and neglect syndrome definition. Among children without an abuse-specific ICD code, we identified ICD codes most associated with abuse by surveying the PHIS database during the study timeframe. Codes were reviewed for face validity by a child abuse pediatrician, and the overall proportion related to abuse was determined.

Results: During the study period, there were 635 abused patients < 5 years old, of whom 87% (Nf550) were matched to PHIS records. Of these, 58.4% were male, 65.1% were white and 81.4% had public insurance. Eighty percent (n=442) were < 2 years and 64% (n=352) were < 1 year. Applying the CDC syndrome definition, only 11.6% had an abuse-specific code assigned in the ED. While specific codes varied by age, injury codes most associated with abuse included TEN-4-FACES bruising (S00-T88), rib fractures (S22.3X, S22.4X), and subdural hematomas (S06.5X), with several codes being highly specific. We identified ICD codes with strong face validity that identified 90% of abused children.Conclusion(s): Abuse-specific ICD codes unreliably estimate abuse prevalence in EDs. However, when combined with certain injury codes, abuse may be estimated more readily. The results of this study establish the framework for a larger multicenter study with the overall primary objective of developing an ICD coding schema to estimate prevalence of physical abuse in EDs.
Figure 1.Flowchart of known abused patients with abuse-specific ICD-10 codes per CDC syndromic definition.