Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Dec 1.
Published in final edited form as: Prev Med. 2020 Oct 3;141:106269. doi: 10.1016/j.ypmed.2020.106269

Exceptional mortality risk among police-identified young black male gang members

David C Pyrooz 1,*, Ryan K Masters 1, Jennifer J Tostlebe 1, Richard G Rogers 1
PMCID: PMC7704767  NIHMSID: NIHMS1634923  PMID: 33022317

Abstract

Gang membership is associated with many risky behaviors but is often overlooked as a source of mortality among young Americans. Gang Member-Linked Mortality Files (GM-LMFs) match St. Louis, Missouri gang members listed in a law enforcement gang database to mortality records in the National Death Index. We created three analytic samples composed of black males aged 15–35 years by merging cases of the GM-LMFs with National Vital Statistics System and Census data in years 1993–2016. Mortality rates standardized to the 15–35-year-old 2010 U.S. male population were estimated for all-cause (1477.4, 99% CI=1451.5–1503.3), homicide (950.1, 99% CI=932.2–967.9), non-homicide injury (314.0, 99% CI=308.8–319.2), and non-injury (213.3, 99% CI=202.3–224.4) deaths in the GM-LMFs. We fitted Poisson rate models to estimate mortality rate ratios (RR) between gang members and demographically-matched comparison groups. Black male gang members in St. Louis were at an elevated mortality risk from all causes of death, and homicides contributed substantially to this risk. Compared to black males in St. Louis, gang members experienced greater relative risk of all-cause (RR=2.9, 99% CI=2.4–3.5), homicide (RR=3.2, 99% CI=2.5–4.1), and non-homicide injury (RR=4.0, 99% CI=2.8–5.8) mortality between 1993 and 2016. Relative risk was greater when compared to black males in St. Louis MSA, Missouri, and the USA. These results identify a key source of excess mortality among young black Americans. Health policies and interventions may be most efficacious when they acknowledge, address, and incorporate information about and target high-risk populations, including gang members, that contribute to relatively high mortality risk in the USA.

Keywords: gang membership, assault, homicide, intentional injury, police, St. Louis, National Death Index

Introduction

U.S. life expectancy only modestly improved over the past decade, recently declined, and lags behind other high-income peer countries (National Research Council and Institute of Medicine, 2013; Rogers et al., 2020). Comparatively low U.S. life expectancy is due to a combination of factors, including high-risk sociodemographic subpopulations, geographic variations, firearm prevalence, and high rates of accidents and violence (Grinshteyn and Hemenway, 2019; National Research Council and Institute of Medicine, 2013). For example, age-specific mortality among Americans aged 15–34 exceeds that of 16 high-income peer countries; over two-thirds of the 92,000 deaths in 2017 at these ages were attributable to preventable causes, such as homicides (12%), suicides (15%), and unintentional injuries (43%) (Kochanek et al., 2019). There is a crucial need to better understand the underlying sources of preventable deaths among young Americans, especially high-risk demographic subpopulations.

Age-adjusted U.S. mortality rates for preventable causes are higher among males than females (Kochanek et al., 2019). In 2017, there was a 3.5 year overall black/white racial gap in life expectancy that was greater among select geographic areas and causes of death. Importantly, homicide was the top cause of death among black males aged 15–34, whose homicide rate was 11 times greater than white males. Murray and colleagues (2005) examined life expectancy among eight U.S. subpopulations, including Middle America (primarily whites with average incomes and levels of education), Black Middle America, and Black High-Risk Urban residents. Compared to Middle America males (75.2 years of life expectancy), males in Black Middle and Black High-Risk Urban America had 2.9 and 8.5 years lower life expectancies, respectively. Further, Rogers and colleagues (2001) found that central city residence increased the risk of homicide mortality among blacks but reduced it among whites. Indeed, compared to whites living in central cities, black central city residents experienced an 8.3-fold increased risk of homicide mortality, controlling for age and sex. Thus, it is vital to more fully understand mortality in high-risk subpopulations such as urban black males.

One potentially overlooked source of elevated mortality risk among young Americans is gang membership. Gangs are durable, street-oriented youth groups whose illegal activities constitute part of their collective identity (Klein and Maxson, 2006). Although gangs exist in other high-income countries, they maintain an enduring and pernicious presence in the United States, particularly in urban areas. They are one of the few social groups that operate within communities and institutions, including schools and prisons, which broadens their reach and impact. As high as eight percent of the U.S. population has a history of gang involvement by their early twenties, found disproportionately among subpopulations at high-risk for preventable causes of death, such as males, Blacks and Latinos, and juveniles and young adults (Pyrooz, 2014).

Gang membership is associated with criminogenic attitudes and health risk behaviors that may elevate the likelihood of preventable causes of early death. Homicide risk is likely elevated due to gang members’ to normative orientations toward violence, the sale of illicit drugs, and the victim-offender overlap (Decker and Van Winkle, 1996; Thornberry et al., 2003). Carrying a weapon for protection, use of illicit substances, and reckless driving may elevate risk of unintentional fatal injuries (Decker and Van Winkle, 1996; Sanders et al., 2013; Thornberry et al., 2003). The anxiety of a gang context characterized by retaliatory violence and fear of victimization can lead to post-traumatic stress symptoms and poor mental health, which, in turn, is linked to psychiatric morbidity and suicide (Coid et al., 2013; Watkins and Melde, 2016). Moreover, there is evidence that young people with elevated criminogenic attitudes and health risk behaviors are more likely to join gangs (Gilman et al., 2014; Thornberry et al., 2003). Consistent with Thornberry and colleagues’ (2003) theoretical model, the kind of individuals who join gangs (selection) and the kind of environments in which gangs operate (facilitation) combine to increase mortality risk among gang members.

Initial reports suggest an association between gang membership and premature mortality, based on findings from small-cohort ethnographic studies in Chicago, Milwaukee, and St. Louis (Decker and Van Winkle, 1996; Hagedorn, 1998; Levitt and Venkatesh, 2001) and longitudinal panel studies in Philadelphia, Phoenix, and Pittsburgh (Chassin et al., 2013; Loeber and Farrington, 2011). Aggregate-level data from local law enforcement agencies collected in 2012 as part of the National Youth Gang Survey (Decker and Pyrooz, 2010; Egley Jr. et al., 2014), although discontinued in 2013, indicated that under one percent of the U.S. population was composed of gang members, yet 16 percent of U.S. homicides were gang-related.

Prior research on this topic is limited in several respects. First, aggregate-level studies are subject to inter-agency differences in defining gang activity and “[f]ew surveillance systems collect data with the level of detail necessary to gang homicide prevention strategies” (Centers for Disease Control and Prevention, 2012, p. 1018). Second, individual-level studies were not designed to generate reliable estimates of mortality risk among gang members. Ethnographic follow-up and longitudinal panel studies typically contain small sample sizes of gang members (maximum<250) and/or small number of deaths (maximum<90); several lack comparison groups altogether. Finally, all studies lack comprehensive information about cause(s) of death; many rely on informal death reporting, and most focus exclusively on homicide. Gang members are a hard-to-reach, risky population, and very few datasets permit generating reliable estimates of differences in mortality risk between gang and non-gang populations.

We examine all-cause and cause-specific mortality risk among gang members. We overcome the limitations of prior research by linking a large sample of police-identified gang members in the St. Louis area with death records from the National Death Index and provide reliable estimates of mortality risk associated with gang membership. St. Louis is an ideal site for assessing gang member mortality risk because of its long history of gang activity and high rates of violence, especially among young black males (Decker and Van Winkle, 1996; Huebner et al., 2016). We hypothesize that all-cause mortality risk is higher among gang members than demographically-matched populations in the U.S., Missouri, and St. Louis. We further hypothesize that mortality risk from homicide is quite high among gang members. National health objectives and social policies that ignore gangs may miss an important high-risk population (Simon et al., 2013), as well as a potentially crucial yet overlooked source for the United States’ relatively low life expectancy (National Research Council and Institute of Medicine, 2013).

Methods

Data

Gang member data is based on 3,225 individuals living in the St. Louis area and identified as gang members between 1993 and 2003 by the St. Louis police department, law enforcement agencies in the St. Louis County area, or local/state correctional institutions. To generate the Gang Member-Linked Mortality Files (GM-LMFs), males in the gang database were linked to the National Death Index (NDI) (Tostlebe et al., 2020). Females were not submitted for matching because they composed less than two percent (n=61) of the database (Fillenbaum et al., 2009), consistent with longstanding biases in law enforcement records. Mortality status between 1993 and 2016 was determined for 3,120 cases (98.6%). The GM-LMFs contain 285 confirmed deaths and include cause of death codes from the International Classification of Diseases, 9th and 10th revisions (ICD-9 and ICD-10).

We restrict the focus to black males, who composed 88% of the database, and are linked to 206 total deaths at ages 15–35 years. Surveys and ethnographies of gang populations in St. Louis have indicated that black males are overrepresented in gangs (Decker and Van Winkle, 1996; Curry et al., 2002). One study found that black males were 87% of the victims of gang homicide in St. Louis in the mid-1990s (Decker and Curry, 2002). Too few non-black males (n=359) prevent reliable estimates of mortality risk for additional subgroups.

We generated three analytic samples of black males aged 15–35 years in the GM-LMFs (n1=1,314; n2=1,032; n3=921) using exclusion criteria based on missing characteristics used by the NDI to match individuals to death records (see appendix for details). We used the most inclusive criteria to generate the first sample (n1; “Inclusive Sample”) and the most exclusive criteria to generate the third sample (n3; “Restrictive Sample”). The most reasonable estimates of mortality risk among black male gang members in St. Louis are likely found in the second sample (n2; “Preferred Sample”). We report results derived from all three to bound our estimates.

Multiple cause of death compressed mortality files from the National Vital Statistics System (NVSS) provided counts of death from all- and specific-causes for demographically-matched groups aged 15–35 for years 1993–2016 (National Center for Health Statistics, 2017). Five comparison groups, selected to estimate relative mortality risk among gang members, are increasingly demographically- and geographically-comparable to our sample: (1) U.S. males, (2) U.S. black males, (3) Missouri black males, (4) black males residing in the St. Louis Metropolitan Statistical Area (MSA), and (5) black males residing in St. Louis City. We removed the deaths among black males in the GM-LMFs samples from each comparison group’s mortality counts in the NVSS data to ensure mutually exclusive samples. We used Census estimates (https://seer.cancer.gov/popdata/) of population counts to estimate mortality rates for the comparison populations, 1993–2016.

Measures

All-cause mortality rate is estimated from counts of all deaths in three-year age groups 15–17, …, 33–35 and duration of time lived in the three-year groups. Homicide mortality rate is estimated from counts of deaths from assault/homicides (ICD-9 codes: E960-E978; ICD-10 codes: X85-Y09). Non-homicide injury mortality rate is estimated from counts of deaths from suicides, accidents, and other injuries (ICD-9 codes: E800-E959, E980-E999; ICD-10 codes: V01-X84, Y85-Y87). Non-injury mortality rate is estimated from counts of death not due to homicides or injuries.

Gang membership.

All individuals listed in the gang database were determined by law enforcement to meet criteria to be classified as affiliated with a gang. Gang classification rationale were included for 71.2% of the individuals listed in the GM-LMFs. Criteria with a clear gang nexus included: self-admission (29.9%), gang-related tattoos (3.2%) or clothing/possessions (6.6%), participation in gang activity (2.7%), associations with known gang members/areas (8.8%), and peer/independent sources (0.8%). Other criteria in which we could not always discern a clear gang nexus based on the entry rationale included arrests (53.9%), corrections classification reports (18.5%), police sources requesting entry (28.8%), and other/indiscernible explanations (6.6%). These criteria were not mutually exclusive (Mean=1.6; Median=2.0). Line-level police and correctional officers, as well as those in specialized gang units, typically gathered this information, which was subject to second-level evidentiary review by a superior before entry into the database.

Law enforcement gang data are routinely used by social scientists (Brantingham et al., 2019; Huebner et al., 2016; Lewis and Papachristos, 2019; Ridgeway et al., 2018), although not without criticism. Katz (2003) reported that gang unit officers in a Midwestern agency documented people as gang members if they met minimum thresholds and satisfied officers’ substantiation of gang membership, which was supported by superiors without scrutiny. Others outline the problematic application of gang policing to racial/ethnic minorities and threats to civil liberties in recording practices (Densley and Pyrooz, 2019). However, law enforcement likely undercounts rather than overcounts gang members, as Curry (2000) demonstrated, using a sample of youth in Chicago, that 78% of the police-identified gang members self-reported as gang members, while only 27% of self-reported gang members were recorded as such by the police. Since gang members are a hard-to-reach population, and prior research relies on small sample sizes, these data are a productive next step to study mortality risk with the qualifications associated with law enforcement data.

Analytic Strategy

We aggregated individual survival histories in the GM-LMFs into counts of death and exposure time for ages 15–35 and appended the aggregated GM-LMFs with the NVSS-Census data. We estimated age-specific mortality rates for three-year age groups, 15–17, …, 33–35, among black male gang members in the GM-LMFs as well as among our five comparison groups. All- and cause-specific death rates in the GM-LMFs and comparison populations are age standardized using the 15–35-year old U.S. male population age distribution in 2010.

We fitted Poisson rate models to estimate mortality rate ratios (RR) between black male gang members and comparison populations. The models were structured as:

log(Deathsi)=log(Ei)+β∗agei′+γ∗gangmember

The mortality rate is expressed as a function of a set of i three-year age groups, agei′(age15–17,age18–20,…,agei) and associated parameters, β = (β15–17, β18–20, …, βi), log(Ei) is declared an offset term in the model where Ei is the estimated person-years of exposure for age group i, and γ is the estimated log rate ratio associated with gang membership (1=yes, 0=no). The Poisson rate models estimate the conditional expectation (i.e., mean rate) of the death rate for a given age group (Powers and Xie, 2008).

We fitted fifteen models to estimate the mortality rate ratios between ages 15–35 for black male gang members in our restrictive, inclusive, and preferred samples in relation to all comparison groups. The mortality rate ratios between black male gang members and comparison groups were separately estimated for all causes of death and for deaths from homicide, non-homicide injuries, and non-injury (analytic scripts in appendix).

Results

Table 1 presents cases, duration of exposure, and counts of total deaths, homicides, and non-homicide injuries, and non-injury deaths, by three-year age groups for our three analytic samples of the GM-LMFs. Characteristics of our preferred GM-LMFs sample (bold emphasis added) are bounded by the characteristics of our most inclusive and most restrictive samples. Because age of entry into gangs varies, person-years of exposure time increase with age. Homicide accounts for nearly 60% of all deaths observed over the exposure period, and non-homicide injuries account for nearly 30% of all deaths. The proportion of deaths due to homicide is greatest at the youngest ages, 15–17 (6/7=.86) and 18–20 (15/19=.79), before declining at older ages.

Table 1.

Descriptive statistics of the analytic samples of Gang Member-Linked Mortality Files, by three-year age group, 1993–2016.

Inclusive Sample Preferred Sample Restrictive Sample
Total
 Cases 1314 1032 921
 Age Range 12.79 – 34.29 12.79 – 34.29 12.79 – 34.29
 Exposure Years 19027.4 14252.5 12343.6
 Total Deaths (homicide, injurya, non-injury deaths) 206 (121, 58, 27) 206 (121, 58, 27) 206 (121, 58, 27)
Ages 15–17
 Person-Years 608.1 384.1 312.3
 Total Deaths (homicide, injurya, non-injury deaths) 7 (6, 0, 1) 7 (6, 0, 1) 7 (6, 0, 1)
Ages 18–20
 Person-Years 2085.8 1486.3 1263.7
 Total Deaths (homicide, injurya, non-injury deaths) 19 (15, 0, 4) 19 (15, 0, 4) 19 (15, 0, 4)
Ages 21–23
 Person-Years 3134.1 2334.5 2036.2
 Total Deaths (homicide, injurya, non-injury deaths) 31 (22, 8, 1) 31 (22, 8, 1) 31 (22, 8, 1)
Ages 24–26
 Person-Years 3465.6 2627.1 2302.7
 Total Deaths (homicide, injurya, non-injury deaths) 38 (18, 15, 5) 38 (18, 15, 5) 38 (18, 15, 5)
Ages 27–29
 Person-Years 3477.7 2644.0 2314.1
 Total Deaths (homicide, injurya, non-injury deaths) 38 (23, 10, 5) 38 (23, 10, 5) 38 (23, 10, 5)
Ages 30–32
 Person-Years 3333.6 2551.8 2219.7
 Total Deaths (homicide, injurya, non-injury deaths) 42 (24, 13, 5) 42 (24, 13, 5) 42 (24, 13, 5)
Ages 33–35
 Person-Years 2922.5 2224.7 1894.9
 Total Deaths (homicide, injurya, non-injury deaths) 31 (13, 12, 6) 31 (13, 12, 6) 31 (13, 12, 6)
Open in a new tab
a

Non-homicide injury deaths

Source: Each sample composed of black males aged 15–35 in the GM-LMFs, 1993–2016.

The estimated age-standardized all-cause mortality rate among gang members aged 15–35 in the Preferred Sample is 1,477.4 deaths per 100,000 population (99% CI=1063.2–1091.6). The estimated age-standardized cause-specific rates among gang members in the Preferred Sample is 950.1 per 100,000 population for homicide (99% CI=932.2–967.9), 314.0 for non-homicide injury (99% CI=308.8–319.2), and 213.3 for non-injury (99% CI=202.3–224.4). Across all three analytic samples of the GM-LMFs, the rates of all-cause and homicide mortality exceed those observed among each of the comparison groups (age standardized mortality rates for all analytic samples are available in appendix).

Table 2 presents estimates of mortality rate ratios (RR) between black male gang members in the GM-LMFs and comparison populations. Estimates from the Preferred Sample (bold emphasis added) show that black male gang members experienced an almost three-fold increased risk of all-cause mortality (RR=2.9, 99% CI=2.4–3.5) compared to black males in St. Louis City. The risk ratios increase with comparisons to black males in St. Louis MSA (RR=3.6, 99% CI=3.0–4.3), to black males in the state of Missouri (RR=4.2, 99% CI=3.5–5.1) and in the United States (RR=5.6, 99% CI=4.7–6.7). Furthermore, because U.S. blacks have higher mortality than most other U.S. race/ethnic populations (Kochanek et al., 2019), there is an exceptionally large mortality gap between black male gang members in St. Louis and all males in the United States (RR=9.6, 99% CI=8.0–11.5). The Inclusive and Restrictive Samples provide lower and upper bounds of the RRs, respectively.

Table 2.

Mortality rate ratios between black male gang members and comparison populations aged 15–35, from all causes of death and from homicides, 1993–2016.

Inclusive Sample Preferred Sample Restrictive Sample
RR 99% CI RR 99% CI RR 99% CI
All Deaths
Comparison Population
 Black Males, St. Louis City 2.2 (1.8–2.6) 2.9 (2.4–3.5) 3.4 (2.8–4.1)
 Black Males, St. Louis MSA 2.7 (2.3–3.3) 3.6 (3.0–4.3) 4.2 (3.5–5.0)
 Black Males, Missouri 3.2 (2.6–3.8) 4.2 (3.5–5.1) 4.9 (4.1–5.8)
 Black Males, USA 4.2 (3.5–5.1) 5.6 (4.7–6.7) 6.5 (5.4–7.8)
 All Males, USA 7.2 (6.0–8.6) 9.6 (8.0–11.5) 11.1 (9.2–13.2)
Homicides
Comparison Population
 Black Males, St. Louis City 2.4 (1.9–3.1) 3.2 (2.5–4.1) 3.7 (2.9–4.8)
 Black Males, St. Louis MSA 3.3 (2.6–4.2) 4.5 (3.5–5.7) 5.1 (4.0–6.5)
 Black Males, Missouri 4.2 (3.3–5.3) 5.6 (4.4–7.1) 6.4 (5.1–8.2)
 Black Males, USA 7.1 (5.6–9.0) 9.1 (7.2–11.5) 10.5 (8.3–13.3)
 All Males, USA 29.8 (23.6–37.7) 40.0 (31.6–50.6) 46.1 (36.5–58.3)
Non-Homicide Injuries
Comparison Population
 Black Males, St. Louis City 3.0 (2.1–4.3) 4.0 (2.8–5.8) 4.6 (3.2–6.6)
 Black Males, St. Louis MSA 3.2 (2.2–4.5) 4.2 (3.0–6.0) 4.9 (3.4–6.9)
 Black Males, Missouri 3.4 (2.4–4.8) 4.5 (3.2–6.4) 5.2 (3.7–7.4)
 Black Males, USA 4.2 (3.0–5.9) 5.6 (4.0–7.9) 6.5 (4.6–9.1)
 All Males, USA 3.7 (2.6–5.2) 4.9 (3.5–6.9) 5.7 (4.0–8.0)
Non-Injury Deaths
Comparison Population
 Black Males, St. Louis City 1.1 (0.7–1.9) 1.5 (0.9–2.4) 1.7 (1.0–2.8)
 Black Males, St. Louis MSA 1.3 (0.8–2.1) 1.7 (1.1–2.8) 2.0 (1.2–3.3)
 Black Males, Missouri 1.5 (0.9–2.4) 2.0 (1.2–3.2) 2.3 (1.4–3.7)
 Black Males, USA 1.6 (1.0–2.5) 2.1 (1.3–3.4) 2.4 (1.5–3.9)
 All Males, USA 3.1 (1.9–5.0) 4.1 (2.5–6.6) 4.7 (2.9–7.6)
Open in a new tab

Note: RR is the mortality rate ratio between police-identified black male gang members in each GM-LMFs sample and comparison population derived from the NVSS. CI is the 99% confidence interval for the risk ratio. All models include three-year age group fixed effect for ages 15–17, …, 33–35.

For homicides, we observe similar but even larger RRs between gang members and comparison populations. Estimates from the Preferred Sample show black male gang members experienced over a three-fold increased risk of homicide mortality (RR=3.2, 99% CI=2.5–4.1) compared to black males residing in St. Louis City. The RRs increase with comparisons to black males residing in St. Louis MSA (RR=4.5, 99% CI=3.5–5.7), the state of Missouri (RR=5.6, 99% CI=4.4–7.1) and the United States (RR=9.1, 99% CI=7.2–11.5). Remarkably, black male gang members experienced about a 40-fold elevated risk of homicide compared to all males in the United States (RR=40.0, 99% CI=31.6–50.6).

Black male gang members are also estimated to have experienced higher mortality risk from non-homicide injuries than comparison populations, but nonsignificantly different risk from non-injury causes of death. For instance, compared to black males in St. Louis City, black males in the GM-LMF had about a four-fold increased risk of death from non-homicide injuries (RR=4.0, 99% CI=2.8–5.8), but only 1.5 times the risk of death from non-injury causes (nonsignificant, 99% CI=0.9–2.4). Homicide mortality, in contrast to non-homicide and non-injury mortality, exhibited a spatial gradient, where the relative risk of mortality among gang members increases with spatially expansive comparison groups.

Discussion

Medical researchers are interested in the mortality risk associated with various high-risk and hard-to-reach populations, such as homeless youth (e.g., Roy et al., 1998), juvenile offenders (e.g., Chassin et al., 2013), and former prisoners (e.g., Binswanger et al., 2017). Young people who affiliate with street gangs are considered a high-risk population, but prior research has not yet generated reliable estimates of elevated mortality risk. Our results reveal that gang membership imparts a high risk of death, which was consistent with our hypothesis that mortality risk is raised as a result of the “kind of people” who end up in gangs and the “kind of contexts” in which gangs are immersed (Thornberry et al., 2003). The fact that homicides and other injury-related deaths accounted for nearly all of this risk was expected owing to the normative orientations of gangs, extra-individual liabilities assumed with gang membership, ingrained networks of inter-gang conflict, and the health risk behaviors observed among gang members. In contrast to non-homicide injury and non-injury mortality, the spatial gradient was especially strong for homicide among police-identified gang members, illustrating exceptional risk of mortality for young black males in St. Louis generally.

This work takes on added significance in light of the comparatively low life expectancy found among U.S. men generally, and young people specifically. The National Research Council and Institute of Medicine (2013) documented that two-thirds of the U.S. male disadvantage in life expectancy occurs before age 50. Preventing gang membership and reducing gang violence could greatly reduce overall and homicide mortality among young black Americans. Redoubling efforts to eliminate homicide as a leading cause of death for all sociodemographic subpopulations would be beneficial, especially for young black males (Abt, 2019; Grinshteyn and Hemenway, 2019). Targeting risk factors that elevate the likelihood of entry into gangs should be the focus of primary and secondary prevention efforts (Hennigan et al., 2015). Intervention efforts involving conflict resolution and crisis mediation, such as street outreach, could assist in disrupting the cycle of gang violence and altering the social norms that promote the violent resolution of conflict (Butts et al., 2015).

Our study demonstrates the power of linking datasets to the NDI. We examined a non-traditional data source, law enforcement gang intelligence, which fits in the tradition of relying on criminal justice data sources to study mortality risk (Binswanger et al., 2017) and gang violence (Papachristos et al., 2013). In fact, the GM-LMFs is the largest extant contemporary dataset of individual-level records of gang members that includes a long follow-up of survival status. The use of law enforcement data allowed us to overcome many of the drawbacks of prior research, including a large sample of gang members and number of deaths, as well as a control group and information on the causes of death. Future linkages between the NDI and data of these kind might reveal mortality risks among other high-risk and justice-involved populations, such as political extremists or organized crime groups. Our age-standardized mortality rates afford substantive comparisons with, and a meaningful benchmark for, such additional populations.

With these strengths in mind, future research might improve upon five areas of the present study, including causation, heterogeneity, censoring, generalizability, and matching. First, while our theoretical framework anticipates a gang-mortality association due to selection and facilitation, we cannot test the causal effect of gang membership on mortality risk. Unobserved characteristics may predispose individuals to both gang membership and mortality risk. Second, accounting for heterogeneity in gang membership would improve the understanding of the gang-mortality association. Some gang members may be more deeply embedded in gangs compared to others—such as individuals at lower rank, shorter tenure, or who left the gang—thus heightening susceptibility to violence.

Third, the GM-LMFs data are left-truncated and right-censored. Some gang members die before being identified and entered in the GM-LMFs; others may experience higher risk of death after age 35. Alternative surveillance systems, such as large-scale representative panel surveys, could address issues with truncation and censoring, although few data sources will generate reliable estimates. Nevertheless, national surveys could be enriched with questions that ascertain, for examples, whether respondents had ever been arrested for gang-related crimes, closely associated with gang members, participated in gang activities, or were members of gangs. Fourth, our results may not generalize to other U.S. cities. Our RRs may be low given the high mortality rates among young black males in St. Louis. Conversely, our RRs may be high if the St. Louis gang environment uniquely elevates homicide and all-cause mortality risk.

Finally, our estimates were derived from analytic samples of the GM-LMFs that were restricted to cases with high-quality data necessary to link to the NDI. If the gang registry data are not missing at random or if mortality risk is associated with the exclusion criteria used to create our GM-LMFs analytic samples, the estimates may be biased. Although we address this concern by providing estimates from three separate analytic samples that vary in the severity of exclusion criteria, additional research should be conducted to validate or contest our estimates.1

Conclusion

Young black male gang members in St. Louis are at an elevated mortality risk from all causes of death, and homicides contribute substantially to this risk. Our results extend prior research (Murray et al., 2005; Rogers et al., 2001) by providing specific mechanisms that contribute to increased mortality risks in high-risk urban environments. Gang membership prevention and intervention could lower mortality among young black males, which could in turn reduce the racial and gender gaps in U.S. life expectancy.

It is of paramount importance that epidemiologists have a complete understanding of high-risk subpopulations, including gang members, to better estimate mortality and successfully implement health promotion and disease prevention strategies (Centers for Disease Control and Prevention, 2012). Surprisingly, most public health research that identifies social factors associated with increased U.S. mortality risk ignores gangs. For instance, the National Research Council and Institute of Medicine’s (2013) report never mentions gangs despite emphasizing that social conditions are responsible for premature deaths in the United States.

Health policies and interventions may be most efficacious when they acknowledge, address, and incorporate information about and target selected high-risk populations, including gang members, who contribute in unique and important ways to the U.S. mortality disadvantage. The increased risk of all-cause and homicide mortality associated with gang members illuminates a social and public health problem that merits additional attention from U.S. public health policymakers and researchers. Indeed, our results identify a key source of excess mortality associated with urban black males, and thereby provides needed insights and detail into ways to greatly reduce homicide mortality and increase life expectancy.

Supplementary Material

1

Highlights.

  • Young black males aged 15–35 years identified by law enforcement as gang members experienced exceptional risk of mortality;

  • Relative risk of all-cause mortality was about 3 times greater for gang members than young black males in St. Louis;

  • Intentional injury, or homicide, accounted for the majority of deaths of police-identified gang members;

  • Gang member age-standardized rate of homicide was 950 per 100,000, 3 times greater than young black males in St. Louis;

  • Gang membership is a source of excess mortality among young black Americans

Acknowledgement:

We thank the NICHD-funded University of Colorado Population Center (Award Number P2CHD066613) for administrative and computing support and Dr. Beth Huebner for data guidance.

Funding: Research reported in this publication was supported by the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health (Award Number R03HD099360), as well as seed funding from the following organizations at the University of Colorado Boulder: the Institute of Behavioral Science, the Center to Advance Research and Training in the Social Sciences, and the Department of Sociology. The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or other organizations that supported this research.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

Sensitivity analyses estimated mortality risk among young black males in the GM-LMFs separately by various criteria used to identify the men as gang members. Results suggest that relative mortality risk is not associated with self-admission, other gang-related indicators (i.e., tattoos, associations, gang activity, peer reports), and criteria that lacked a clear gang nexus. Also, the estimated RRs among GM-LMF cases with only one criterion of gang association did not differ from estimated RRs among GM-LMF cases with two or more criteria. See appendix for detailed results from these sensitivity analyses.

Human Participant Protection statement: This research was reviewed and approved by the Institutional Review Board of the University of Colorado Boulder (protocol # 17–0152).

Conflicts of Interest: The authors declare no conflicts of interest.

References

  1. Abt T, 2019. Bleeding out: The devastating consequences of urban violence—and a bold new plan for peace in the street. Basic Books, New York. [Google Scholar]
  2. Binswanger IA, Morenoff JD, Chilcote CA, Harding DJ, 2017. Ascertainment of vital status among people with criminal justice involvement using Department of Corrections Records, the US National Death Index, and Social Security Master Death Files. Am. J. Epidemiol 185, 982–985. 10.1093/aje/kww221 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Brantingham PJ, Valasik M, Tita GE, 2019. Competitive dominance, gang size and the directionality of gang violence. Crime Sci. 8, 7 10.1186/s40163-019-0102-3 [DOI] [Google Scholar]
  4. Butts JA, Roman CG, Bostwick L, Porter JR, 2015. Cure violence: a public health model to reduce gun violence. Annu. Rev. Public Health 36, 39–53. [DOI] [PubMed] [Google Scholar]
  5. Centers for Disease Control and Prevention, 2012. Gang homicides — Five U.S. cities, 2003–2008. Morb. Mortal. Wkly. Rep 61, 46–51. [PubMed] [Google Scholar]
  6. Chassin L, Piquero AR, Losoya SH, Mansion AD, Schubert CA, 2013. Joint Consideration of Distal and Proximal Predictors of Premature Mortality Among Serious Juvenile Offenders. J. Adolesc. Health 52, 689–696. 10.1016/j.jadohealth.2012.11.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Coid JW, Ullrich S, Keers R, Bebbington P, DeStavola BL, Kallis C, Yang M, Reiss D, Jenkins R, Donnelly P, 2013. Gang membership, violence, and psychiatric morbidity. Am. J. Psychiatry 170, 985–993. [DOI] [PubMed] [Google Scholar]
  8. Curry GD, 2000. Self-reported gang involvement and officially recorded delinquency. Criminology 38, 1253–1274. 10.1111/j.1745-9125.2000.tb01422.x [DOI] [Google Scholar]
  9. Curry GD, Decker SH, Jr AE, 2002. Gang involvement and delinquency in a middle school population. Justice Q. 19, 275–292. 10.1080/07418820200095241 [DOI] [Google Scholar]
  10. Decker SH, Curry GD, 2002. Gangs, gang homicides, and gang loyalty: Organized crimes or disorganized criminals. J. Crim. Justice 30, 343–352. [Google Scholar]
  11. Decker SH, Pyrooz DC, 2010. On the validity and reliability of gang homicide: A comparison of disparate sources. Homicide Stud. 14, 359–376. 10.1177/1088767910385400 [DOI] [Google Scholar]
  12. Decker SH, Van Winkle B, 1996. Life in the gang: Family, friends, and violence. Cambridge University Press, Cambridge, UK. [Google Scholar]
  13. Densley JA, Pyrooz DC, 2019. The matrix in context: Taking stock of police gang databases in London and beyond. Youth Justice 1473225419883706. 10.1177/1473225419883706 [DOI] [Google Scholar]
  14. Egley A Jr., Howell JC, Harris M, 2014. Highlights of the 2012 National Youth Gang Survey. US Department of Justice, Office of Juvenile Justice and Delinquency Prevention, Washington, DC. [Google Scholar]
  15. Fillenbaum GG, Burchett BM, Blazer DG, 2009. Identifying a National Death Index match. Am. J. Epidemiol 170, 515–518. 10.1093/aje/kwp155 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gilman AB, Hill KG, Hawkins JD, 2014. Long-term consequences of adolescent gang membership for adult functioning. Am. J. Public Health 104, 938–945. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Grinshteyn E, Hemenway D, 2019. Violent death rates in the US compared to those of the other high-income countries, 2015. Prev. Med 123, 20–26. 10.1016/j.ypmed.2019.02.026 [DOI] [PubMed] [Google Scholar]
  18. Hagedorn JM, 1998. People and folks: Gangs, crime and the underclass in a rustbelt city., 2nd ed. Lake View Press. [Google Scholar]
  19. Hennigan KM, Kolnick KA, Vindel F, Maxson CL, 2015. Targeting youth at risk for gang involvement: Validation of a gang risk assessment to support individualized secondary prevention. Child. Youth Serv. Rev 56, 86–96. 10.1016/j.childyouth.2015.07.002 [DOI] [Google Scholar]
  20. Huebner BM, Martin K, Moule RK Jr, Pyrooz D, Decker SH, 2016. Dangerous places: Gang members and neighborhood levels of gun assault. Justice Q. 33, 836–862. 10.1080/07418825.2014.984751 [DOI] [Google Scholar]
  21. Katz CM, 2003. Issues in the production and dissemination of gang statistics: An ethnographic study of a large Midwestern police gang unit. Crime Delinquency 49, 485–516. [Google Scholar]
  22. Klein MW, Maxson CL, 2006. Street gang patterns and policies. Oxford University Press, New York, NY. [Google Scholar]
  23. Kochanek KD, Murphy SL, Xu J, Arias E, 2019. Deaths: Final data for 2017 (No. 68, 9), National Vital Statistics Reports. National Center for Health Statistics, Hyattsville, MD. [PubMed] [Google Scholar]
  24. Levitt SD, Venkatesh SA, 2001. Growing up in the projects: The economic lives of a cohort of men who came of age in Chicago public housing. Am. Econ. Rev 91, 79–84. [Google Scholar]
  25. Lewis K, Papachristos AV, 2019. Rules of the game: Exponential random graph models of a gang homicide network. Soc. Forces 10.1093/sf/soz106 [DOI] [Google Scholar]
  26. Loeber R, Farrington DP, 2011. Young male homicide offenders and victims: Risk factors, prediction, and prevention from childhood. Springer. [Google Scholar]
  27. Murray CJL, Kulkarni S, Ezzati M, 2005. Eight Americas: New perspectives on U.S. health disparities. Am. J. Prev. Med 29, 4–10. 10.1016/j.amepre.2005.07.031 [DOI] [PubMed] [Google Scholar]
  28. National Center for Health Statistics, 2017. Compressed Mortality File, 1993–2016 (data file and documentation) as compiled from data provided by the 57 vital statistics jurisdictions through the Vital Statistics Cooperative Program. Hyattsville, MD. [Google Scholar]
  29. National Research Council and Institute of Medicine, 2013. U.S. health in international perspective: shorter lives, poorer health. panel on understanding cross-national health differences among high-income countries. National Academies Press, Washington, DC. [PubMed] [Google Scholar]
  30. Papachristos AV, Hureau DM, Braga AA, 2013. The corner and the crew: The influence of geography and social networks on gang violence. Am. Sociol. Rev 78, 417–447. 10.1177/0003122413486800 [DOI] [Google Scholar]
  31. Powers D, Xie Y, 2008. Statistical Methods for Categorical Data Analysis, 2nd Edition, 2nd ed. Emerald Group Publishing Ltd, Bingley, UK. [Google Scholar]
  32. Pyrooz DC, 2014. ‘From your first cigarette to your last dyin’ day’: The patterning of gang membership in the life-course. J. Quant. Criminol 30, 349–372. 10.1007/s10940-013-9206-1 [DOI] [Google Scholar]
  33. Ridgeway G, Grogger J, Moyer RA, MacDonald JM, 2018. Effect of gang injunctions on crime: A study of Los Angeles from 1988–2014. J. Quant. Criminol 10.1007/s10940-018-9396-7 [DOI] [Google Scholar]
  34. Rogers RG, Hummer RA, Vinneau JM, Lawrence EM, 2020. Greater mortality variability in the United States in comparison with peer countries. Demogr. Res 42, 1039–1056. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Rogers RG, Rosenblatt R, Hummer RA, Krueger PM, 2001. Black-White Differentials in Adult Homicide Mortality in the United States. Soc. Sci. Q 82, 435–452. [Google Scholar]
  36. Roy E, Boivin J-F, Haley N, Lemire N, 1998. Mortality among street youth. The Lancet 352, 32 10.1016/S0140-6736(05)79510-6 [DOI] [PubMed] [Google Scholar]
  37. Sanders B, Valdez A, Hunt GP, Laidler KJ, Moloney M, Cepeda A, 2013. Gang youth, risk behaviors, and negative health outcomes, in: Sanders B, Thomas YF, Deeds BG (Eds.), Crime, HIV and Health: Intersections of Criminal Justice and Public Health Concerns. Springer, New York, pp. 113–127. [Google Scholar]
  38. Simon TR, Ritter NM, Mahendra RR, 2013. Changing course: Preventing gang membership. National Institute of Justice and Centers for Disease Control and Prevention, Washington DC. [Google Scholar]
  39. Thornberry TP, Krohn MD, Lizotte AJ, Smith CA, Tobin K, 2003. Gangs and delinquency in developmental perspective. Cambridge University Press, New York. [Google Scholar]
  40. Tostlebe JJ, Pyrooz DC, Rogers RG, Masters RK, 2020. The National Death Index as a source of homicide data: A methodological exposition of promises and pitfalls for criminologists. Homicide Stud. 10.1177/1088767920924450 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Watkins AM, Melde C, 2016. Bad medicine: The relationship between gang membership, depression, self-esteem, and suicidal behavior. Crim. Justice Behav 10.1177/0093854816631797 [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1634923-supplement-1.docx (132.4KB, docx)

RESOURCES