172 - Neighborhood deprivation index and risk of fatal violent injury recidivism in adolescent assault survivors

Publication Number: 172.5964

Ashkon Shaahinfar, University of California, San Francisco, School of Medicine, Oakland, CA, United States; Sun-Ming J. Pan, VisionForward Enterprises, New York, NY, United States; Jasmine Solola, UCSF Benioff Children's Hospital Oakland, Oakland, CA, United States; Irene Yen, University of California, Merced, Merced, CA, United States; Harrison Alter, Andrew Levitt Center for Social Emergency Medicine, Berkeley, CA, United States; Ann Lazar, University of California, San Francisco, School of Medicine, San Francisco, CA, United States; Jahan Fahimi, UCSF, San Francisco, CA, United States; Charles Branas, Columbia University, New York, NY, United States

Background: Pediatric survivors of firearm injury and assault are at high risk for violent trauma recidivism and mortality. The literature has underscored that homicide and violent reinjury are associated with male sex, Black race, low socioeconomic status, and neighborhood-level deprivation. However, few studies have examined the role of neighborhood level factors in the risk of subsequent homicide among adolescents surviving assault.

Objective: To determine whether adolescents surviving assault (including by firearm) are at increased risk of homicide relative to those surviving unintentional trauma and to explore the impact of neighborhood deprivation in relation to this outcome.

Design/Methods: This was a secondary analysis of adolescents patients age 12 to 16 years, within a broader multicenter, retrospective cohort study, who presented to the three trauma centers in San Francisco and Alameda counties, California, between 1/1/2000 and 12/31/2009 after assault and unintentional trauma (UT). The Social Security Death Master File and the California Department of Public Health Vital Statistics were queried through December 31, 2014, to identify those who died after surviving their initial hospitalization and to delineate death by homicide. A principal component analysis was used to determine neighborhood deprivation index (NDI) by census tract. Multivariable Cox proportional hazards models were specified, with and without NDI, to determine associations between exposure to assault and subsequent homicide.

Results: Among 752 assault and 2183 UT patients who survived the index hospital visit, 37 (1.3%) died by subsequent homicide. In the multivariable Cox model including only individual-level variables, assault exposure, male sex, Black race and non-private health insurance were independent risk factors for subsequent homicide. However, after inclusion of NDI in the model, Black race was no longer independently associated with subsequent homicide risk. Assault exposure conveyed twice the hazard of homicide compared to UT in both models (AHR 2.2, 95% CI = 1.11-4.2 and AHR 2.1, 95% CI = 1.04-4.3, respectively). NDI demonstrated a trend towards association with subsequent homicide risk, which did not reach statistical significance (AHR 7.4, 95% CI = 0.83-66).

Conclusion(s): Adolescents who survive assault, including by firearm, are more likely to die by homicide compared to those who survive unintentional, non-violent trauma. Neighborhood deprivation may play an important mediating role and, in the context of structural racism, may in part confound the association observed between Black race and subsequent homicide risk.

Table 1: Characteristics of Cohorts
UT, unintentional trauma; SD, standard deviation. 1 based on Mann-Whitney. 2 based on Chi-square.

Table 2: Cox proportional hazards models (with and without NDI) with unadjusted and adjusted hazard ratios for death by homicide after surviving index visit†
†Excluding subjects with inadequate identifying information for long-term follow-up.
VCI not included in multivariable models as models would be overspecified for number of outcomes.
*P-value < 0.05
**P-value < 0.01