Intimate Partner Violence and Firearm Suicide: Using Text Mining to Bridge a Critical Gap in United States Data Systems

Overview

There are no national data that document the link between intimate partner violence (IPV) and firearm suicide. This dissertation will use the National Violent Death Reporting System (NVDRS) and apply supervised machine learning to assess the proportion of firearm and non-firearm suicides that are precipitated by IPV in the United States.

Status

Complete

Purpose

We will create a new IPV variable specifically for suicides in NVDRS that can be used for future firearm policy research. By improving available data about the contribution of IPV to suicide, this dissertation may reveal new program and policy opportunities for prevention.

Approach

This project will examine over n=123,000 suicides in NVDRS (2013-18), representing data from 38 states, Washington DC, and Puerto Rico. Using text from NVDRS death narratives, we will apply Natural Language Processing (NLP) to develop an algorithm that can accurately classify suicides as IPV-related or not. This project will leverage nearly 8,000 cases that have already been hand-reviewed during pilot studies as our ‘gold standard’ for classification. After developing and deploying the algorithm, we will then explore the relative contribution of IPV to firearm and non-firearm suicide.

Significance

This will be the first study to enumerate the contribution of IPV to suicide outcomes using national data. This dissertation may catalyze improvements in death investigation and NVDRS data abstraction processes while also helping inform policy decisions to holistically prevent firearm violence.

Published Research

• Detecting Intimate Partner Violence Circumstance for Suicide: Development and Validation of a Tool Using Natural Language Processing and Supervised Machine Learning in the National Violent Death Reporting System

  View research

• Intimate Partner Violence and Suicide Mortality: A Cross-Sectional Study Using Machine Learning and Natural Language Processing of Suicide Data From 43 States

  View research

• Intimate Partner Violence Circumstances for Fatal Violence in the U.S.

  View research

Investigator Bio

Julie M. Kafka, MPH, is a doctoral candidate in the Department of Health Behavior at the Gillings School of Global Public Health at UNC Chapel Hill. She is also a research fellow at the UNC Injury and Violence Prevention Research Center. Kafka conducts applied research, often using administrative data, to inform practical solutions for averting violence. She is a recipient of the Jamie Kimble Scholarship for Courage.