Intimate partner violence (IPV) is a significant public health and criminal justice problem that negatively impacts millions of victims yearly in the United States, primarily women (85%). Most IPV-related healthcare visits (83%) occurred in an emergency department (ED), and these clinical encounters are unique opportunities to identify IPV victims and potentially provide assistance. Although numerous health professional organizations have endorsed universal screening and counseling for IPV since 1992, actual screening rates, detection of IPV victims, and referrals to IPV services remain low in the ED. As a result, many IPV victims pass through the ED unidentified and untreated. Computerized screening tools have been developed and implemented in clinical settings in order to assist providers in screening and detecting IPV. However, these tools have a great limitation in that they rely on information collected from the patient and do not utilize the longitudinal data in electronic health records (EHR). Recently, researchers demonstrated that a history of IPV diagnoses and associated clinical symptoms highly predict current and future IPV (OR=7.8), and these important IPV data could serve as red flags that trigger providers to assess patients further for IPV. In order to enhance IPV screening in the ED, we propose to develop and assess an automatic clinical data summarizer that extracts, abstracts and synthesizes patient historical IPV data (structured and unstructured), and delivers patient historical IPV data to ED providers through an intuitive interface.
The specific aims are: 1) develop and evaluate natural language processing (NLP) strategies to identify and extract patient historical IPV incidents and timelines from clinic notes; 2) develop and evaluate a web service-based summary tool (IPV-Summary-Service) that synthesizes patient- specific IPV information from both NLP-processed data elements and structured data; and 3) develop and pilot test an enhanced IPV screening strategy that delivers clinical evidence generated by the IPV-Summary- Service through a specific EHR (Epic) to providers during the patient universal IPV screening in the ED. This automatic clinical date summary for IPV will be piloted in one ED at MUSC. There are two major outcomes to be measured for the IPV-Summary enhanced screening for 6 months before and after the index date of pilot testing: 1) rate of successful referral to the IPV 24-hour dedicated IPV nurse; and 2) rate of initiation of referral and identification of persons at high risk of IPV. We will use mixed effect generalized linear regression models to estimate the effects of IPV-Summary on the referral rate and IPV case identification rate. Through survey studies, we will assess secondary outcomes including factors of system feasibility, usability, and providers' satisfaction. These analyses can identify potentially important correlates of the major outcomes and may help us improve the design of the intervention. The results from this study will form the foundation for a broader implementation in a regional health information exchange for EDs.

Public Health Relevance

Computer-based approaches for intimate partner violence (IPV) universal screening have led to significantly higher screening rate and detection rate, as well as receipt of IPV services in the emergency department (ED). However, these approaches rely on information collected from the patient and do not utilize the longitudinal IPV data existing in electronic health records (EHR), which have high predictive power of IPV risk. In order to enhance the effectiveness of IPV screening, we propose to develop and assess an automatic clinical data summary tool that extracts, abstracts, and synthesizes patient historical IPV information from EHR and then delivers that critical information to ED providers at the point of care.

Agency
National Institute of Health (NIH)
Institute
National Library of Medicine (NLM)
Type
Exploratory/Developmental Grants (R21)
Project #
5R21LM012945-02
Application #
9739347
Study Section
Biomedical Computing and Health Informatics Study Section (BCHI)
Program Officer
Sim, Hua-Chuan
Project Start
2018-07-05
Project End
2020-06-30
Budget Start
2019-07-01
Budget End
2020-06-30
Support Year
2
Fiscal Year
2019
Total Cost
Indirect Cost
Name
Medical University of South Carolina
Department
Public Health & Prev Medicine
Type
Schools of Medicine
DUNS #
183710748
City
Charleston
State
SC
Country
United States
Zip Code
29407