Mary D. Fan

Violence perpetration among patients hospitalized for unintentional and assault-related firearm injury: a case-control study and a cohort study

An estimated 41024 persons aged 15 years or older were hospitalized because of nonfatal firearm injuries in 2014 in the United States (1). About 79% of those injuries were intentional and resulted from interpersonal violence (assault), and 11% were unintentional (accidental). The remaining 10% were self-inflicted, were due to legal interventions, or had an undetermined intent. Regardless of intent, many patients with nonfatal firearm injuries have short-term, long-term, or permanent physical and psychological sequelae (2). Illness associated with such trauma translates to a notable loss of healthy life-years and considerable societal costs (3, 4). Effective primary, secondary, and tertiary prevention strategies are critically needed to reduce the heavy burden of firearm injuries. Such strategies should preferably integrate pertinent elements of clinical care, public health, and/or the criminal justice systems. Previous investigations have highlighted overlapping risks between becoming a victim and perpetrator of violence (511). Population-based research to specifically examine violence perpetration before and after firearm injury can inform interventions in both community and health care settings. Because only about half of all violence victimizations (that is, becoming a victim of violence) are reported to police (12), hospital admission is an important sentinel event that could present a valuable opportunity for violence risk reduction. In 2009, the National Network of Hospital-based Violence Intervention Programs was formally established (13). These programs consider the in-hospital recovery period as a valuable opportunity or teachable moment during which patients can be connected with principal community services to help reduce retaliation and recidivism. Programs typically focus on patients whose injury was assault-related because of their presumed involvement in a cycle of interpersonal violence. Whether patients without assault-related injuries would also benefit from such programs is unclear. Of note, empirical evidence is lacking on prior involvement in, and subsequent risk for, violence perpetration among patients with unintentional injuries. We conducted 2 statewide studies to examine violent crime perpetration both before and after hospitalization for a firearm injury among patients aged 15 years or older. Injury and crime were studied together to add to the existing body of knowledge on gun violence by using data from both clinical and criminal justice system encounters. Injury intent was a central theme of both studies and was separated into 2 categories (assault vs. unintentional) to examine the association between intent-specific firearm injury and violence perpetration. We focused on these categories because they constitute most firearm injuries requiring hospitalization. Most patients with self-inflicted firearm injuries die before presenting to the hospital, and the number of patients in other injury intent categories (for example, legal interventions) is also relatively small within the hospitalized population. Methods Design, Setting, and Participants We conducted a casecontrol study and a retrospective cohort study. In the casecontrol study, we compared the odds of violence-related arrest before hospitalization between persons hospitalized for firearm injuries and those hospitalized for other reasons. In the cohort study, we compared rates of violence-related arrest after hospitalization between persons hospitalized for firearm injuries and those hospitalized for other reasons. We used data previously assembled in a larger investigation for both of these studies (14). In that investigation, we first identified all patients hospitalized for an injury by any mechanism from 2006 to 2007 in Washington by using International Classification of Diseases, Ninth Revision, codes. Then we chose a random sample of patients hospitalized for a noninjury reason (that is, the no injury group) and frequency-matched them with those in the injury group on age and year of hospitalization in a 2:1 ratio. For the analyses presented here, we separated the injury group into 2 mutually exclusive subgroups: patients hospitalized for a firearm injury (firearm injury group), and those hospitalized for an injury not caused by a firearm (other injury group), resulting in 3 groupsfirearm injury, other injury, and no injury. Figure 1 depicts the design of the 2 studies. In the casecontrol study, the firearm injury group served as the case population and the other injury and no injury groups served as 2 separate control populations. The exposure of interest was arrest for a violent crime before hospitalization. In the cohort study, the firearm injury group served as the exposed population and the other injury and no injury groups served as 2 separate unexposed populations. The outcome of interest was time to first arrest for a violent crime after hospital discharge. Figure 1. Design of the 2 studies. Information on hospitalizations was obtained from the Washington State Department of Health Comprehensive Hospital Abstract Reporting System (15). This system contains coded discharge information and is used to collect various data, such as age, sex, payer status, and diagnosis and procedure codes. Consistent with the literature, an injury-related hospitalization was defined as a discharge with a primary diagnosis of an acute injury (International Classification of Diseases, Ninth Revision, codes 800 to 959). Records containing injuries from medical and surgical misadventures (E870 to E879), late effects of injury (E929 or E999), and adverse effects of substances in therapeutic use (E930 to E949) were excluded (16). Codes for external causes of injury (that is, E codes) were used to determine the mechanism and intent of an injury. The Centers for Disease Control and Prevention recommended this framework of E-code groupings for presenting injury mortality and morbidity data (17). We used E codes to categorize hospitalizations by injury intent: assault, unintentional, self-inflicted, or undetermined. In these studies, we restricted intent-specific analyses to assault-related and unintentional injuries because of the small number of self-inflicted injuries and those with an undetermined intent; however, overall analyses included all injuries regardless of intent. Information on arrests was obtained from Washington State Patrol records. This database provided full arrest history, including juvenile criminal records, and contained information for persons as young as 10 years. We used specific codes in the Revised Code of Washington to identify violent crimes, including homicide, rape, robbery, and assault, according to the Uniform Crime Reporting program of the Federal Bureau of Investigation (18). All patients were aged 15 years or older at the time of hospital discharge. We excluded records for persons younger than 15 years at the time of discharge because they would not have had any criminal records before age 10. Probabilistic algorithms were used to link each patients hospitalization record to his or her arrest record from 2001 through 2011. A subset of identifiers, including the first 2 letters of the first name, first 2 letters of the last name, date of birth, sex, and first 3 digits of the ZIP code, was used for the linkage. Detailed information about data linkage procedures can be found elsewhere (14). The Human Subjects Division of the Washington State Department of Health approved the study protocol and procedures. Statistical Analysis In the casecontrol study, odds of prior violence-related arrest were compared between case and control patients. Odds ratios (ORs) and their corresponding 95% CIs were determined by using multivariable logistic regression models that included covariates for age; sex; payer status; hospital county; and history of diagnosis of a psychiatric disorder, including substance use disorders. In the cohort study, only patients who survived their hospitalization were included. Follow-up began on the day of discharge and ended on the day of the first violence-related arrest, death, or 31 December 2011whichever occurred first. The unadjusted absolute risk for violence-related arrest was estimated by using the cumulative incidence function, with death treated as a competing event. In regression analyses, we used the methods described by Fine and Gray (19) to model violence-related arrest with the subdistribution hazards regression. Subhazard ratios and their corresponding 95% CIs were determined by multivariable models that included the same set of covariates used in the casecontrol study plus history of violence-related arrest. Additional analyses were conducted to compare the firearm injury group with a subset of patients in the other injury group who had sustained injuries through cut or pierce mechanisms (for example, stab wounds) or struck-by or struck-against mechanisms. In terms of the social context in which the injury occurred, these individuals may have been more comparable with the firearm injury group than those who sustained injuries by such mechanisms as motor vehicle crashes or falls. In all analyses, an of 0.05 was used to denote statistical significance. All tests were 2-sided and conducted using SAS, version 10 (SAS Institute), and Stata, version 13 (StataCorp), with the stcrreg and stcurve package for Fine and Gray modeling. Role of the Funding Source This research was funded by the City of Seattle and the University of Washington Royalty Research Fund. The funding sources had no role in the design, conduct, and reporting of this research or the decision to submit the manuscript for publication. Results A total of 245343 hospitalized patients were included in this investigation. Of these, 658, 71855, and 172830 were in the firearm injury, other injury, and no injury groups, respectively. A greater proportion of patients in the firearm injury group than those in the other 2 groups

Risk Magnified: Standing Under the Statist Lens

Why some harms count before the courts and others do not is a matter of acute expressive and practical impact. Judicial refusal to see claimed injuries is an effective denial of legal personhood and a bar from powerful judicial machinery. The issue of “erratic, even bizarre” judicial recognition of supplicants vexed Professor Joseph Vining as early as 1978. Recent scholarship argues that injuries are seen through a subjective lens, reflecting the relative privilege of the judiciary and their concomitant difficulties in perceiving injuries to minorities and the poor. This is a troubling contention. So long as another, objective explanation remains, it should be superimposed, not to conceal and legitimate potentially problematic practices, but to substitute as an alternative rationality and a neutral and transparent principal for future decisions. This comment advances such an alternative explanation: The erratic pattern of judicial sight is partly a refraction of how judges view the risk of probabilistic future injury. Present harm is immediately visible, but the contours of risked injury are less distinct, requiring congressional or constitutional magnification. Aspects of positive law aimed at reducing the risk of prescribed probabilistic future harms are telescopes. Such collectively constructed magnifiers, however, often do not track social risk or vulnerability, since some clout is typically necessary to enshrine interests in positive law. As a result, those whose interests are socially slighted may find themselves similarly slighted before the courthouse doors.