Improving Adolescent Violent Trauma Outcomes With a Hospital-Based Violence Prevention Initiative

The objective of this study was to determine if evaluation by the SUV program was associated with decreased rates of reinjury and improved outpatient clinic follow-up among young adult patients admitted to a single trauma center for violent trauma.

Retrospective cohort chart review study of patients meeting the inclusion criteria. The study was reviewed and approved to be exempt by the institutional review board of the Albert Einstein College of Medicine.

The study was conducted at NYC Health + Hospitals/Jacobi, a Level I adult and Level II pediatric trauma center in the Bronx, NY. The study included admitted hospital patients for violent trauma from August 1, 2014, to April 31, 2018, and followed until the end of the study period (July 31, 2018).

Patients identified who met the following inclusion criteria: age 15 to 24 years and admitted with hospital International Classification of Diseases, 10th Revision, codes for gunshot wound (GSW) (Y24.9XXA), stab wound (X99.9XXS), or physical assault (T76.1) during the time period of interest. All patient cases were identified through the hospital trauma registry notification system and the corresponding SUV patient log data set.

Review of the electronic medical record was performed for all eligible patients. Injuries were coded as either physical assault, GSW, or stab wound. Charts were reviewed for exposure, outcome, and demographic variables of interest. A deidentified data set was created for the final data analysis.

The exposure variable was documentation of being evaluated by the SUV team during admission. Patients were either seen (yes) or not seen (no) by the team. Patients not seen by the SUV team naturally self-selected themselves from the intervention group and served as the controls. These patients may not have been evaluated by the SUV team because they were admitted and discharged from the hospital on days the SUV team did not have a physical presence in the hospital (ie, holidays/Sundays/after hours) or the SUV team was not notified by the admitting trauma team. Patients who were approached but refused intervention program services were removed from the overall analysis to minimize the effects of selection bias.

The outcome variables included attending >50% of scheduled outpatient follow-up visits 3 months after hospital discharge, and presenting to NYC Health + Hospitals/Jacobi for subsequent violent trauma reinjury (GSW, stab wound, and physical assault) from date of presentation to the end of the study time period (July 31, 2018) for a minimum 3-month follow-up period. The last patient included in the final analysis presented on April 1, 2018, to allow for a minimum 3-month follow-up for outcome variables at the end of the study time period (July 31, 2018). Outpatient follow-up visits included surgery, trauma surgery, and any surgical subspecialty visits (ie, orthopedics; ear, nose, throat; ophthalmology; neurosurgery) 3 months after hospital discharge. The hospital electronic medical record was queried for all subsequent ED, outpatient, and inpatient visits to determine if follow-up and reinjury occurred for a minimum of 3 months after initial hospital discharge. Outcome variables were dichotomized as yes/no for both variables. No follow-up occurred after the end of the study time period.

Demographic data including age, race/ethnicity, and sex were collected. All exposure and outcome variables were dichotomized, and frequencies calculated. χ² test was used to determine association between the main outcome variables (>50% of outpatient department follow-up visits attended, reinjury), the exposure variable (SUV seen), and the other covariates (age, sex, race/ethnicity, injury type). A statistically significant association was defined as 2-sided P value <.05. Any covariate significantly associated with both primary exposure and outcome variables was considered a potential confounder and included in a multivariable logistic regression model to estimate adjusted odds ratios (aORs) with 95% confidence intervals. aORs were compared with unadjusted ORs generated from a univariate regression model without the potential covariates. Variables that altered the unadjusted aORs by a >10% difference were recognized as confounders. These variables were included in the final multivariate logistic regression model and their aOR with 95% confidence interval were reported. All statistical analyses were done using SAS statistical software version 9.4 (Cary, NC).

The SUV team is composed of both a hospital and community outreach team that work in concert to improve patient and community outcomes. The SUV program is staffed with a designated social worker and program director, a community outreach director, a physician medical director, and a hospital responder, along with a network of community outreach workers and supervisors. The community outreach workers, supervisors, and hospital responder are all individuals considered credible messengers, given their presence in the community serviced by the hospital and previous community violence and conflict mediation experience. The SUV team is notified of violent trauma admissions through the ED trauma registry notification system and counsels and supports patients admitted for violent trauma. Patients are evaluated by the SUV hospital team consisting of the physician medical director, social worker, and hospital responder. The medical director assists in medical care coordination, whereas the social worker assists with housing assistance, food insecurity, mental health resources, home care needs, and job resources. The SUV hospital responder utilizes conflict mediation strategies to prevent retaliation and reinjury, and connects patients to medical, social, educational, and vocational services. The community outreach team consists of community outreach workers who identify high-risk individuals in the community and provide mentorship, conflict mediation strategies, and connection to educational/job services.

A total of 528 patients fit inclusion criteria and were analyzed in the final hospitalized patient data analysis. There were 291 patients in the intervention and 237 patients in the control group. Seven patients were excluded from the final analysis that were approached but refused participation. The sample had 493 (92.5%) male patients and 285 (54%) Black patients. The sample had 195 (36%) Latino ethnicity patients. The mean age was 19.1 years. Table 1 lists the demographic characteristics in both the control and intervention groups.

The majority of patients in our cohort during the study period sustained stab wounds (234), followed by GSWs (149) and assaults (145). A total of 278 (51.2%) patients in the cohort attended >50% of scheduled follow-up visits 3 months after discharge and 59 patients presented for reinjury to our trauma center during the study time period with a reinjury rate of 11%. Reinjury follow-up time period ranged from 3 months to 4 years, with each participant having a minimum 3-month follow-up after hospital discharge. Reinjury follow-up was not normally distributed, and median reinjury follow-up was 6 months. No significant differences were found in reinjury follow-up time period between the intervention and control group (Table 2). Table 1 lists the variable frequencies in both the control and intervention groups. Only the variable, injury type fit criteria for confounding with altering the unadjusted exposure–outcome effect by a 10% difference and the ORs reported were derived from an adjusted multivariable regression model.

Statistically significant associations were found between the primary exposure variable (SUV seen) and both outcome variables in our cohort of admitted patients (Table 2). In patients of all injury type, being seen by the SUV team was associated with 2.29 times the odds of attending >50% of scheduled clinic follow-up visits 3 months after hospital discharge as compared with not being seen by the SUV team. In patients of all injury type, being seen by the SUV team was associated with 0.41 times the odds of being admitted for violent reinjury for a minimum of 3 months after discharge within the study period (Table 3).

In our cohort, patients of all injury type evaluated by the SUV team, as compared with those patients not evaluated by the SUV team, were found to have greater odds of attending >50% of their scheduled clinic follow-up visits (2.29 [1.59–3.29]) and decreased odds of presenting to our trauma center for subsequent violent reinjury 3 months from initial presentation within the study period (0.41 [0.22–0.75]) (Table 3).

Our study was limited by all patient data being retrospective and via chart review. It is plausible that some patients presented to other trauma centers with reinjury or for postdischarge follow-up during the study period. The possibility for presentation to another trauma center for reinjury is limited by the fact that our institution is only 1 of 2 adult level I trauma centers, and the only level II pediatric trauma center, in the Bronx, within which 95% of our cohort resides. There is also a possibility that patients may have suffered less-severe reinjuries or fatal injuries not requiring hospital care. Future study should include local trauma center chart review, personal patient phone calls to confirm less-severe reinjuries, and review of city homicide databases to review fatal injuries in previously admitted violent trauma patients.

The study is also limited by selection bias of a convenience sample, with the potential for discrepancy in other hospital resource availability on days the control group presented. For example, hospital social work presence on weekends/holidays is for emergencies only and not for routine coordination of care/services, which could significantly impact postdischarge outpatient follow-up and reinjury rates.

Furthermore, although no significant differences were identified in reinjury follow-up time period between the control and intervention group, the wide range of follow-up in both groups (3–48 months) may result in bias. Further study could encompass a more standardized intervention and follow-up time period to better interpret results.

These findings suggest that incorporating a multidisciplinary approach consisting of physicians, social workers, and community outreach workers may help improve outcomes in adolescent/young adult patients admitted for violent trauma. Improving rates of postdischarge follow-up clinic visits is crucial for optimizing medical outcomes in patients with violent trauma injuries.¹³  The decreased odds of reinjury observed in patients seen by the SUV team not only potentially resulted in decreased medical costs, which can surpass $8 billion, but may also have prevented future long-term psychological-, economic-, and community-level effects that are hard to objectively measure.^2,15–17,19  The findings in our study are consistent with other studies that suggest HVIPs may play a role in preventing reinjury. An analysis of a San Francisco General Hospital HVIP found that expressing the need for education services was associated with reinjury, and that this outcome was reversed with dedicated community-based organization services.¹³ 

Furthermore, improving rates of community shootings prevents hospitalizations, school/work absences, family disruption, and individual/community stress.^19,20  A prevented shooting mitigates a potential retaliatory injury and disrupts the persistent cycle of violence.^21–25  The effect of a prevented violent trauma victimization is difficult to quantify with statistical analysis given its potential positive, long-lasting, unquantifiable effects.^26,27  However, the findings in our study highlight the important role HVIPs can play in creating a safe space at the individual patient and community level. HVIPs can equip individual patients with needed services, along with playing a role in establishing ongoing community relationships through the incorporation of community outreach workers.

The findings in our cohort of improved patient outcomes seem to justify investment in our HVIPs’ funding and can provide direction for future trauma centers’ role in violent trauma prevention. Recent evidence suggests that HVIPs may be a cost-effective way to limit recidivism in admitted violent trauma patients.²³  Monetary savings associated with having an HVIP, such as the SUV program, have a reported range of $82 765 to $4 055 873 in different simulation models and also are shown to have a statistically significant, positive effect on quality-adjusted life-years.^{18,19,23,24,28,29}  Even when accounting for factors that could decrease program cost effectiveness, such as hospitalization and program implementation costs, the presence of an HVIP is still documented to result in an acceptable cost per health outcome gained.^{18,19,23,24,28,29}  Given recent trends finding a notable increase in pediatric violent trauma victimization associated with the coronavirus disease 2019 pandemic, investments in HVIPs may prove to be essential in curtailing the increase in adolescent/young adult violent trauma associated with the coronavirus disease 2019 pandemic.^30,31  Furthermore, every pediatric violent trauma constitutes an early life adversity event, not just for victims, but for their families, friends, and community. Early life adversity is associated with reduced executive functioning and future psychiatric diagnosis in later life.³²  The implementation of an HVIP with the potential of preventing reinjury in pediatric patients admitted after violent trauma may prove to be instrumental in curtailing the effects of current and future early life adversity events. Future studies could include an in-depth cost per health outcome gained for the SUV program and similar HVIPs. An improved HVIP cost analysis will better evaluate cost effectiveness and the direct effect on preventing not just physical reinjury, but also the mental health effects and adversity associated with pediatric violent trauma victimization.

Despite our study design limitations, our findings suggest that the implementation of a multidisciplinary HVIP incorporating community outreach workers may be associated with decreased serious violent reinjury requiring hospital evaluation and improved adherence to postdischarge scheduled appointments in our intervention cohort of patients. The findings in our study suggest that pediatric trauma centers can play a vital role in violence prevention through the implementation of HVIPs to improve both individual patient- and community-level outcomes to curtail the effects of violent trauma in vulnerable adolescent/young adult populations.