The Role of International Classification of Diseases 10 Codes in Intimate Partner Violence Surveillance: Examining the Sensitivity and Specificity of Current Practice

May Shum,Dorothea Robertson,Caitlin Ryus,Gunjan Tiyyagura
DOI: https://doi.org/10.1016/j.annemergmed.2024.03.019
IF: 6.762
2024-04-26
Annals of Emergency Medicine
Abstract:Intimate partner violence is a significant public health issue that affects at least 1 in 3 women, 1 in 4 men, and a disproportionate number of transgender individuals in the United States. 1 The emergency department (ED) is often the first place where patients experiencing intimate partner violence seek health care. In fact, victims of intimate partner violence are more likely than nonvictims to access EDs, often after exposure to violence. 2 Administrative medical data, including the International Classification of Diseases, Tenth Revision (ICD-10), have been posited as effective surveillance tools for monitoring and responding to intimate partner violence. 3 The Uniform Data System (UDS) is a standardized data set released by the Health Resources and Services Administration that collects data from and provides reporting requirements to federally funded health centers. The 2021 UDS report outlined 5 ICD-10 codes required to identify intimate partner violence (Table 1). 4 However, there is substantial variation in clinician use of these codes. 5 Variability in ICD-10 code use in intimate partner violence literature has highlighted the tradeoff between increasing sensitivity and minimizing false positives. 3 There is no consensus about which codes are most appropriate for identifying this patient population in the ED. As such, our team has previously relied on manual review of encounters identified by a large list of codes to minimize underestimation of intimate partner violence, which poses significant resource burden. 6 Therefore, we aimed to examine which ICD-10 codes were the highest yield in retrospectively identifying cases of intimate partner violence.
emergency medicine
What problem does this paper attempt to address?