Barriers to screen for domestic violence among women in emergency department

The National Institute of Nursing Research (NINR) awarded a $865,084 project grant (CFDA 93.361) to the Medstar Health Research Institute Inc. to investigate barriers to screening for domestic violence among women in emergency department settings. The project aims to leverage advanced data science methods to identify injury patterns and use natural language processing to find verbal cues in electronic health records related to domestic violence. It also involves conducting interviews with emergency department providers to understand the barriers to screening for domestic violence during patient encounters. This research has the potential to provide insights to improve domestic violence screening processes and the care provided to patients in emergency departments. A sub-award of an unspecified amount was provided to George Washington University to contribute subject matter expertise on domestic violence and emergency department operations.

Abstract

Abstract domestic violence (DV) includes physical and sexual violence, threats, economic, and emotional/psychological abuse, or other abusive behavior as part of a systematic pattern of control and power perpetrated by one intimate partner against another. IT causes a significant burden for the healthcare systems by increasing morbidity and mortality among victims. Women are disproportionately affected, although men may experience DV as well. The recent covid-19 pandemic led to movement restrictions and stay at home orders. While these decisions were essential to prevent spread of the virus, such extended domestic stays may exacerbate the number the total as well as reported incidents of DV. As a result, in recent years, DV has transformed into a shadow pandemic, which further complicated this public health issue and increased the need to perform accurate and timely interventions. DV often forms a pattern, and many of the victims experience repeated acts of physical or mental abuse. Victims of DV may seek care in hospital settings which makes timely interventions critical and even lifesaving. While there is a serious need for government to reinforce commitments made to eliminate all forms of DV against women, the health sector plays an essential role in breaking the cycle of abuse. Health providers can prevent reoccurrence of such violent incidents by identifying women who are experiencing DV, and then provide comprehensive services and train health providers in responding to the needs of survivors in addition to caring for physical injuries. Abused women rarely disclose the reason for emergency department (ED) visit due to various reasons including shame, fear of the perpetrator or financial dependencies. While these factors form patient-specific barriers to screen for DV, the barriers to screening, detecting and helping DV victims can be recognized at different levels during an ED visit. Since these barriers are not clear, more exploration is needed to understand important features by analyzing EHR data to gain further understanding of the clinical experience and environment. In aim 1 of this proposal, we will use the DV-related ICD-9/ICD-10 diagnosis codes to find positive cases of DV among the visits to ED. Then adapt market-basket analysis, which is a data mining method originated in the field of marketing, to our objective and identify patterns of injury and health problems which are observed together frequently. Then, we will utilize state-of-the-art deep learning-based natural language processing (NLP) models to learn the patterns in electronic health records clinical notes related to DV. In aim 2, we will conduct semi- structured interviews with ED health providers to investigate the barriers to screening for DV during patient- provider encounter. The outcomes of this study have the potential to add significant insights to improve the screening process and the care we provide our patients in the ED. 1