You are on page 1of 28

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/343139445

Illness spillovers of lethal police violence: the significance of gendered


marginalization

Article  in  Ethnic and Racial Studies · July 2020


DOI: 10.1080/01419870.2020.1781913

CITATIONS READS

14 179

6 authors, including:

Alyasah Ali Sewell Rashawn Ray


Emory University University of Maryland, College Park
28 PUBLICATIONS   423 CITATIONS    48 PUBLICATIONS   882 CITATIONS   

SEE PROFILE SEE PROFILE

Keon Gilbert
Saint Louis University
46 PUBLICATIONS   525 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Deregulated Disparities View project

Surveilled Health View project

All content following this page was uploaded by Alyasah Ali Sewell on 17 May 2021.

The user has requested enhancement of the downloaded file.


Ethnic and Racial Studies

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/rers20

Illness spillovers of lethal police violence: the


significance of gendered marginalization

Alyasah Ali Sewell , Justin M. Feldman , Rashawn Ray , Keon L. Gilbert , Kevin
A. Jefferson & Hedwig Lee

To cite this article: Alyasah Ali Sewell , Justin M. Feldman , Rashawn Ray , Keon L. Gilbert , Kevin
A. Jefferson & Hedwig Lee (2020): Illness spillovers of lethal police violence: the significance of
gendered marginalization, Ethnic and Racial Studies, DOI: 10.1080/01419870.2020.1781913

To link to this article: https://doi.org/10.1080/01419870.2020.1781913

Published online: 22 Jul 2020.

Submit your article to this journal

Article views: 562

View related articles

View Crossmark data

Please cite article as: Sewell, Alyasah Ali, Justin M. Feldman, Rashawn Ray, Keon L. Gilbert, Kevin A.
Jefferson & Hedwig Lee (2021) Illness spillovers of lethal police violence: the significance of gendered
marginalization, Ethnic and Racial Studies, 44:7, 1089-1114, DOI: 10.1080/01419870.2020.1781913

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=rers20
ETHNIC AND RACIAL STUDIES
https://doi.org/10.1080/01419870.2020.1781913

Illness spillovers of lethal police violence: the


significance of gendered marginalization
Alyasah Ali Sewell a, Justin M. Feldman b, Rashawn Ray c
,
Keon L. Gilbert d, Kevin A. Jefferson a and Hedwig Lee e

a
Department of Sociology, Emory University, Atlanta GA, USA; bDepartment of Public Health,
New York University, New York, USA; cDepartment of Sociology, University of Maryland,
College Park, USA; dDepartment of Behavioral Science and Health Education, St. Louis
University, St. Louis, USA; eDepartment of Sociology, Washington University in Saint Louis,
St. Louis, USA

ABSTRACT
Police violence is a pressing public health problem. To gauge the illness
associations of police killings – the most severe form of police brutality, we
compile a unique multilevel dataset that nests individual-level health data
from the 2009–2013 New York City Community Health Survey (nij = 39,267)
within neighbourhood-level data from 2003 to 2012 EpiQuery Vital Statistics
(nj = 34). Using weighted hierarchical generalized linear models, we assess
main and gendered associations between neighbourhood exposures to lethal
policing and five illnesses. Holding all else constant, living in lethally surveilled
areas is linked to a greater risk of high blood pressure and obesity for all
neighbourhood residents and to a greater risk of obesity for women.
Furthermore, illness risks are also gendered: Women face a 30–54 percent
greater risk of diabetes, high blood pressure, and obesity compared to men.
Lethal police brutality is an important neighbourhood risk factor for illness
and, especially, for women’s health.

ARTICLE HISTORY Received 10 April 2019; Accepted 5 June 2020

KEYWORDS Police brutality; health; neighbourhoods; gender; women’s health; police violence

Introduction
Police violence is both a measure and driver of population health. The dispro-
portionate lethal brutality faced by men like Eric Garner, Michael Brown, Alton
Sterling, and George Floyd reveal the need for the sophisticated, mandatory
reporting of police violence (Gilbert and Ray 2016; Krieger et al. 2015; Lewis
2014). Moreover, the deaths of Sandra Bland, Korryn Gaines, Erica Garner,
and Breonna Taylor (Horton 2018; Neyfakh 2015; Wang 2017; Whaley 2020)
reveal multifaceted gendered processes at play. We assess the illness

CONTACT Alyasah Ali Sewell aasewel@emory.edu www.aasewell.com @aasewell


© 2020 Informa UK Limited, trading as Taylor & Francis Group
2 A. A. SEWELL ET AL.

spillovers of living in lethally surveilled neighbourhoods – specifically, places


where more deaths are officially attributed to interactions with police.
To estimate lethal police violence, we deploy vital records categorized by
the International Classification of Diseases, Tenth Revision (ICD-10). Coroners
and medical examiners classify killings as “legal intervention deaths” (LIDs)
when they designate the underlying cause of death to be the result of injuries
inflicted by enforcers of the law during the course of legal action (DeGue,
Fowler, and Calkins 2016; Sikora and Mulvihill 2002). This medicalization
process signals the formal recognition and attribution of state-sanctioned
police killings.
We focus on New York City (NYC) – a city with a long history of police bru-
tality and anti-brutality movements to curb lethal policing and its effects
(Johnson 2004). We pinpoint associations apparent during the stop-and-
frisk era of NYC, which formally ends with the 2013 Floyd, et al. v. City of
New York decision. Using neighbourhood effects research designs, we esti-
mate multilevel and gendered associations of living in lethally surveilled
areas for five illnesses: poor/fair self-rated health, diabetes, high blood
pressure, asthma, and obesity.

Background
High levels of police surveillance damage the fabrics of communities (Brunson
2007; Geller and Fagan 2019). Lethal policing likely impairs the body through
the production of social emotions that compromise feelings of belongingness
and safety and enhance distrust of and resentment toward police. Legal inter-
vention deaths brand bereavement as a product of the state, a product of
“carceral grief”.
Research associates police intervention with emotional and physical health
using both individual-level and multi-level data. For example, discriminatory
policing worsens a person’s emotional and physical health (Forman, Williams,
and Jackson 1997; Kessler, Mickelson, and Williams 1999; Williams et al. 1997).
Individuals who are stopped more by the police report worse psychological
and general well-being than those who have had less contact with the police,
especially if the encounter is forceful, aggressive, or violent (DeVylder et al.
2018; Geller et al. 2014). Even vicarious police contact hastens aging and contrib-
utes to obesity and worse self-assessed health (McFarland et al. 2018; McFarland,
Geller, and McFarland 2019; McFarland, Taylor, and McFarland 2018).
Furthermore, residents of communities with a higher concentration of frisk-
ing and police use of force report more illness conditions (Sewell 2017; Sewell
and Jefferson 2016). State-level exposure to police killings of unarmed Black
men weakens the mental health of Black people (Bor et al. 2018). Police vio-
lence is understudied as a mode of community violence (Boyd, Ellison, and
Horn 2016; Haldipur 2018). Nevertheless, violence’s place-based ecological
ETHNIC AND RACIAL STUDIES 3

form diminishes emotional, psychobiological, and physical health (Assari et al.


2016; Margolin and Gordis 2000; Sternthal et al. 2010). So, there are reasons to
believe illness associations exist with neighborhood police violence.

Spillovers of lethal police violence


Aggressive policing elicits “surveillance stress” (Sewell 2017) – the ongoing
strain borne from routine scrutinization to extract information through institu-
tionalized technologies. The loss of network members to violent policing
emits surveillance stress that is not readily evident in nonlethal policing.
Several research strands suggest disease burden is higher in lethally surveilled
neighbourhoods.
First, police contact has long been considered an unpleasant, stressful life
event that requires adaptation and change (Holmes and Rahe 1967; Lewin-
sohn and Talkington 1979) and emits deleterious physical and emotional con-
sequences (Thoits 2010). For example, those in close proximity to police
violence are aware of it and change their behaviour because of it
(Desmond, Papachristos, and Kirk 2016). Trauma studies, furthermore,
suggest that being a witness to or even hearing about violence is linked to
greater psychopathology and physiological reactions (Eisenberg et al. 1991;
Lutgendorf and Costanzo 2003). We consider the traumas of police violence
occur through broad neighbourhood-based social networks.
Second, police surveillance, through its links to mass incarceration, exerts
social control over ethnoracially- and economically-subjugated people and
communities (Alexander 2020; Browne 2015). Ethnoracially-marginalized
neighbourhoods have a stronger police presence (Stults and Baumer 2007),
undergo more frequent and aggressive policing (Johnson et al. 2019; Lersch
et al. 2008; Smith and Holmes 2014), and register more complaints of police
misconduct (Smith and Holmes 2003). Economically-disadvantaged neigh-
bourhoods, meanwhile, face more forceful police supervision and further
deterioration via the revolving door of mass incarceration (Clear 2009;
Sampson and Loeffler 2010). Furthermore ethnoracially- and economically-
subjugated people are more likely to be or know someone who has been har-
assed by the police, sustained police violence, or been incarcerated (Geller
and Scott 1992; Goff et al. 2016; Lee et al. 2015; Ross 2015; Terrill and Reisig
2003; Weitzer and Tuch 2014).
Third, aggressive and abusive policing fuels perceptions of injustice (Mac-
Donald et al. 2007; Skolnick and Fyfe 1993), and perceptions of injustice and
discrimination are linked to poorer health (Kessler, Mickelson, and Williams
1999; Williams et al. 1997). Greater exposure to police surveillance and
abuse creates a “climate of fear” (Shedd 2012). Where legal surveillance back-
drops social life, brutal and unfair police encounters - personally or vicariously -
engineer resistance to the state (Gau and Brunson 2010).
4 A. A. SEWELL ET AL.

Gendered spillovers of lethal police violence


Lethal policing fosters shared experiences of illness among all neighbourhood
members. However, gender status patterns marginalization and criminaliza-
tion systems differently for men and women across all neighbourhoods.
Lethal policing thus may pattern illness both between and within neighbour-
hoods by gender status.
On one hand, processes of multiple marginalization indicate men are both
targets of and burdened by the carceral state. Men have more contact with
the police than women (Goff et al. 2016). Men’s routine activities are more
likely to be interfered and criminalized by police (Boyd and Clampet-Lund-
quist 2019; Brunson and Miller 2006). They also are more likely to judge
their contact with police as discriminatory (Kessler, Mickelson, and Williams
1999). Furthermore, men show more empathy for male relatives who are
incarcerated (Brown, Bell, and Patterson 2016). Last, men living in aggres-
sively surveilled neighbourhoods (characterized as increased frisking and
force use by police) report greater psychological distress than women
(Sewell, Jefferson, and Lee 2016). Hence, men may be unhealthier than
women in lethally surveilled areas.
On the other hand, processes of network marginalization indicate women
are connected to people targeted by the carceral state. Women are often
the primary contact for someone arrested or incarcerated (Comfort 2009; Hal-
dipur 2018). Women’s caretaking responsibilities increase in the face of the loss
of a network member (Lee and Wildeman 2013), as they assume greater con-
tributions to household earnings (Davis 2016) and supervision of grieving and
displaced network members (Burton and Tucker 2009; Stack 1997). Girls and
women are more likely than boys and men to express worry about vicarious
reports of police misconduct (Hurst, McDermott, and Thomas 2005) and
police sexual abuse (Brunson and Miller 2006; Cooper et al. 2004). They
report being more frustrated about and frightened by police neglect and mis-
behaviour (Cooper et al. 2004; Jones 2014). Having an incarcerated family
member reduces cardiovascular health among women (Lee et al. 2014). Corre-
spondingly, women may be unhealthier than men in lethally surveilled areas.

Research design
This study examines whether living in lethally surveilled neighbourhoods
impairs health. We create a measure that identifies whether someone lives
in a neighbourhood where at least three legal intervention deaths have
been documented between 2003 and 2012. We then deploy multilevel
models that enable us to compare reports of five (5) illness conditions for
people living in neighbourhoods with a high count of LIDs to the illness con-
ditions of their counterparts living in neighbourhoods where LIDs are rarer.
We consider the relative roles of within-neighbourhood and between-
ETHNIC AND RACIAL STUDIES 5

neighbourhood factors that co-occur with lethal policing and illness as well as
the collective roles of both sets of factors on the illness risks of lethal police
surveillance. We evaluate the illness risks of living in lethally surveilled neigh-
bourhoods for all community residents, for women residents, and for men
residents, separately.
We then consider whether the illness associations of lethal police violence
differ according to gender status. Specifically, we create a cross-level inter-
action term that codifies the difference in the illness association of police vio-
lence present between women and men (Women:Men). This interaction term
fuses the status of being a woman with the experience of living in lethally sur-
veilled neighbourhoods (Women X LID). When the cross-level interaction term
is not statistically significant (i.e. close to 1), men and women similarly experi-
ence illness while living in lethally surveilled neighbourhoods. Statistically, if
the term exceeds 1, women experience more illness than men; if the term
is less than 1, men experience more illness than women.
We are unable to evaluate the mechanisms connecting processes of mul-
tiple or network marginalization to gender differences in the illness risks of
living under the weight of lethal police violence. However, we do consider
individual- and neighbourhood-level covariates that could account for gen-
dered-continencies in the illness associations of lethal police violence. Specifi-
cally, we identify the competing role of these covariates in contributing to the
illness risks of living in lethally surveilled areas for all neighbourhood resi-
dents, for women, for men, and for women relative to men. The contribution
of individual- and neighbourhood-level covariates are estimated separately
and, then, together in a final model. Essentially, we assess whether lethal
police surveillance is a neighbourhood feature that embeds illness risks that
are not reflective of compositional and contextual forces.

Methods
Data
We create a unique multilevel database that merges individual-level data
from a community survey with neighbourhood-level data from multiple
sources describing features of NYC. To connect data sources with overlap-
ping periods of observation, we capitalize on a consistent neighbourhood
indicator available across all data sources. Neighbourhoods used in this
study cover all U.S. postal zip codes of NYC and identify 34 unique neigh-
bourhoods that are proprietary to the United Hospital Fund (NYC Depart-
ment of Health and Mental Hygiene 2006). Appendix A indicates the
names of NYC neighbourhood areas comprising the United Hospital Fund
neighbourhoods (UHFs). The use of publicly available data is exempt from
IRB review by the authors’ institutions.
6 A. A. SEWELL ET AL.

Neighbourhood level
Data from EpiQuery provide neighbourhood-level indicators of lethal surveil-
lance in New York City. EpiQuery compiles data on the health of New Yorkers
from a variety of sources, including surveys, surveillance data, and vital
records. Neighbourhood-level counts of deaths due to selected causes,
including legal intervention, are available by UHFs from 2003 to 2016.
Yearly neighbourhood-specific data for LIDs are aggregated to the neigh-
bourhood-level. Neighbourhood-varying data on LIDs occurring between
2003 and 2012 are pooled to deidentify cases of LID, which creates
overlap with the other datasets and allows lag time for illness to develop
after the earliest exposure to LIDs. Each neighbourhood is assigned a
single numerical value for number of LIDs that does not change in value
over the study period. Other neighbourhood-level indicators are culled
from data provided to the authors by the New York Police Department
(NYPD) for the 2010–2012 period or data compiled by the authors from
the IPUMS National Historical Geographic Information System (NHGIS)
using the 2010 U.S. Census.

Individual level
The analysis is based on a cross-sectional repeated measurement of a
sample of adults (n ij = 39,627) who participate in the New York City Com-
munity Health Survey (NYC-CHS). This study relies on surveys collected
by the NYC Department of Health and Mental Hygiene between 2009
and 2013 (New York City Department of Health and Mental Hygiene
2009–2013). NYC-CHS is a representative sample of New York City resi-
dents utilizing an annual random-digit-dial health survey of non-institu-
tionalized adult (18+) New Yorkers. Each year, the NYC-CHS interviews
approximately 10,000 people. All data are collected by telephone or
cell phone.
NYC-CHS compiles information about NYC residents citywide, by neigh-
bourhood, and across sociodemographic populations. The survey employs
validated instruments to assess a wide array of self-assessed well-being indi-
cators. These instruments are based on the national Behavioral Risk Factor
Surveillance System coordinated by the U.S. Centers for Disease Control and
Prevention.
We pool adults across annual waves of the NYC-CHS using methods
common to cross-sectional repeated research designs (e.g. NORC’s General
Social Surveys, the American National Election Surveys). NYC-CHS includes
UHFs identifiers for respondents surveyed. Respondents live in 1 of the 34
UHFs, which are identified for all respondents. Inferences in this study are
derived from a sample of NYC residents occupying the population of NYC
neighbourhoods.
ETHNIC AND RACIAL STUDIES 7

Measures
Neighbourhood-level lethal police violence
Neighbourhood-level lethal police violence is measured using EpiQuery vital
statistics data on “legal intervention deaths” (LIDs). ICD-10 defines legal inter-
vention as “injuries inflicted by police or other law-enforcing agents … in the
course of arresting or attempting to arrest lawbreakers, suppressing disturb-
ances, maintaining order, … and other legal action”. Based on death certifi-
cate data from the New York City Office of Chief Medical Examiner, death
by legal intervention is an external cause that captures all deaths coded as
“Y35” or “Y89.0” by the New York City Bureau of Vital Statistics. Annually, Epi-
Query identifies the number of LIDs in a UHF based on the neighbourhood in
which the death occurred but does not provide any demographic data con-
cerning the decedent. We create a dichotomous indicator of a UHF having
more than two LIDs (“>2” or “3+”) during the 2003–2012 period.

Outcomes of interest
The NYC-CHS is used to consider five (5) chronic illness conditions: self-
reported health, diabetes diagnosis, high blood pressure diagnosis, asthma
episode in the past year, and BMI-indicated obesity. These are well-being indi-
cators used in prior research on neighbourhood police surveillance (Sewell
and Jefferson 2016) and thus provide a good comparison base for illness
associated with the lethality of police interactions. Self-reported health
status is determined by responses to the question: “Would you say in
general that your health is: excellent, very good, good, fair, or poor?” The
ordinal distribution is dichotomized: 1 = poor or fair; 0 = excellent, very
good, or good. Diabetes and high blood pressure are determined by
responses to the question: “Have you ever been told by a doctor, nurse or
other health professional that you have [condition]? (1 = Yes; 0 = No)”.
Asthma episodes are determined by responses to the question: “In the last
12 months, have you had an episode of asthma or an asthma attack? (1 =
Yes; 0 = No)”. Obese body weight is based on categorizations of body mass
index (BMI) provided by the New York City Department of Health and
Mental Hygiene through calculations of self-reported height and weight.
This outcome is dichotomous, such that normal or overweight (BMI ≥ 18.5
to < 30 kg/m2) bodyweight is the reference category, with BMIs ≥ 30 classified
as obese. Those who are underweight are removed from the sample.

Individual-level covariates
We consider key sociodemographic (gender, age, ethnoracial status, nativity,
marital status, educational attainment, income, work status, and health insur-
ance status) variables of importance to health. Except for age, all individual-
level variables are categorical. Age is mean-centered at 52.6 years.
8 A. A. SEWELL ET AL.

Neighbourhood-level covariates
We account for three neighbourhood features. The NYPD provides neighbour-
hood-level data on the average count of robbery complaints between 2010
and 2012 in a neighbourhood. Zip code-level data from the 2010 U.S.
Census culled by IPUMS NHGIS is aggregated to UHFs to create indicators
of the proportion of the population that identifies as Black or Latino and
the proportion of families with incomes below the federal poverty line.

Sample
We deal with missing data through dummy variable replacement, since only
one covariate contributed to more than 10 per cent of missing data: income in
relation to the federal poverty line. People whose income level are missing
comprise 16.9 per cent of the sample of individuals who provide data
between 2009 and 2013. The PIs for NYC-CHS provide a dedicated category
for the 4,256 individuals who did not know their income, a group comprising
about 9.6 per cent of the sample. An additional 3,079 people – about 6.9 per
cent – refused to provide income data. Less than 1 per cent of people was
missing income for some other reason (n = 152). To reduce bias due to sys-
tematic missingness on income, we retain a category that indicates people
who did not know their income and create a category for people who
refused to provide income data.
Only 5.2 per cent of the sample was deleted due to missing data on out-
comes of interest (n = 2,295). An additional 6 per cent of the sample was
deleted due to missing data on at least one individual-level covariate.
Deleted participants did differ from the included participants; included partici-
pants were more likely to be male, born in the United States, and insured and
have higher educational degrees and incomes. No systematic variation in
missingness was related to neighbourhood-level covariates. Removing partici-
pants with missing data still left us with sufficient power; thus, listwise del-
etion was appropriate for the remaining covariates. After using listwise
deletion to exclude participants with missing data on dependent covariates
and all independent covariates (except income to federal poverty line), our
analytical sample was comprised of 39,627 individuals (86.9 per cent of
sample) nested in 34 UHFs.

Statistical analysis
The statistical modelling framework employed anticipates that individual
reports of illness are partly a function of the UHF in which an individual
resides (Raudenbush and Bryk 2002). UHF-level variables are appended to
each individual observation in the NYC-CHS database, and post-stratification
survey weights are applied. Individuals are a representative sample of NYC
ETHNIC AND RACIAL STUDIES 9

residents. UHFs, meanwhile, are the population of NYC neighbourhoods, as all


residential areas are present.
Stata 14.2 is used for all analyses. Weighted multilevel models for binary
response outcomes are employed that allow for a random-intercept associ-
ated with UHF neighbourhoods that assume an unstructured covariance
structure. A logit-link is used for poor/fair self-rated health measure; the dia-
betes and high blood pressure diagnosis report; the asthma episode indicator
in the past year; and the obesity indicator for BMI. We expect non-specific
physical illness associations with neighbourhood legal intervention deaths.
For each outcome-LID pair, we conduct three sets of regressions. The first
set of regressions (overall) evaluates the main association between general/
physical health and living in lethally surveilled areas (i.e. >2 LIDs). The next set
of regressions (stratified) stratifies the association between general/physical
health and living in lethally surveilled neighbourhoods for women and men
separately. The stratified models include a gender-specific covariate for
living in a lethally surveilled area, such that each covariate (i.e. Women: >2
LIDs; Men: >2 LIDs) represents the illness risks for women and for men,
respectively, living in a neighbourhood with more than two legal intervention
deaths. The last set of regressions (moderated) estimates whether the illness
associations of living in lethally surveilled neighbourhoods is contingent on
(women’s) gender while assuming that other covariates have gender-invariant
associations. The moderated model includes an interaction term between
women’s gender and LIDs that characterizes the added illness women
endure living in lethally surveilled neighbourhoods. The interaction term in
the moderated model (i.e. Women X > 2 LIDs) represents the difference in
the illness risks of LIDs between women and men – a statistical comparison
of police killings covariates in the gender-stratified models.
Within each main form of regressions, four estimates of the illness risks of
living in lethally surveilled areas are shown. The first set of estimates adjusts
for gender in the overall models, differentiates by gender in the stratified
models, and examines the added risks for women in the moderated
models. The second set of estimates removes variation in the illness risks of
living in lethally surveilled neighbourhoods that is a function of neighbour-
hood variation in individual-level characteristics of neighbourhood residents
(or, “compositional” effects). The third set of estimates removes variation in
the illness risks of living in lethally surveilled neighbourhoods that is a func-
tion of neighbourhood features – specifically, crime exposure, ethnoracial seg-
regation, and poverty concentration (or, “contextual” effects). The fourth set of
estimates removes variation in the illness risks of living in lethally surveilled
neighbourhoods that is a function of both individual- and neighbourhood-
level characteristics. Adjusted odds ratios are reported for all analyses.
Sensitivity analysis uses pooled individual-level data as early as 2006 to
coincide with consistently available data for self-reported health and
10 A. A. SEWELL ET AL.

chronic condition outcomes, as well as pooling data from 2012 and 2013 to
coincide with the decline in policing that began at the onset of the Floyd
case. Estimates from both sets of sensitivity tests reveal consistency in the esti-
mates presented in the following analysis. Sensitivity analyses also examine
alternative LID thresholds (1, 2, 5) and their combinations. The categorization
of communities into 0–2 deaths vs. 3 or more deaths provided the most par-
simonious models, identifying unique illness associations from those of other
functional forms for neighbourhood LIDs (e.g. count index, dissimilarity
measure). Sensitivity analysis conducted using the ordinal (self-rated health)
and linear (BMI) distributions yielded similar associations as those reported.

Results
Are lethally surveilled neighborhoods distinct?
A legal intervention death in NYC occurred every 48 days and 16 hours, total-
ling 75 deaths between 2003 and 2012. There is substantial neighbourhood
variation in where such deaths occur (Table 1). Eleven UHFs have no LIDs,
while nine neighbourhoods have at least three deaths. On average, 26.5 per
cent of UHFs have more than two LIDs over the 10-year study period.
Figure 1 depicts neighbourhood variation in police killings in New York
City. The map reveals clusters of lethal police violence in Brooklyn and the
Bronx. In Brooklyn, the Bedford-Stuyvesant/Crown Heights neighbourhood
has the greatest number of LIDs – 14. Similarly, in the Bronx, lethal police vio-
lence is highest in the South Bronx neighbourhood, where there are 11 LIDs.
This dichotomous categorization also yields a clear areal distribution of police
killings that is distinct enough to detect neighbourhood-level associations
(Figure 1).
There is also sufficient variation in illness in NYC to detect neighbourhood-
level associations. Intraclass correlations indicate that 4.0 per cent of the vari-
ation in self-rated health, 3.6 per cent of the variation in diabetes, 1.3 per cent
of the variation in high blood pressure, 3.8 per cent of the variation in asthma,

Table 1. Frequency Distribution of legal intervention deaths in New York city, 2003–
2012, by united hospital fund neighbourhood.
Total LIDs Nj Frequency Cumulative Frequency
0 11 32.35 32.35
1 6 17.65 50.00
2 8 23.53 73.53
3 3 8.82 82.35
4 1 2.94 85.29
5 3 8.82 94.12
11 1 2.94 97.06
14 1 2.94 100.00
Note: j denotes United Hospital Fund neighbourhoods (total = 34 areas).
Source: EpiQuery, 2003–2012 New York City Vital Statistics (total = 75 deaths).
ETHNIC AND RACIAL STUDIES 11

Figure 1. Map of legal intervention deaths by united hospital fund neighbourhoods.

and 3.7 per cent of the variation in obesity is attributable to residential neigh-
bourhoods. On average residents of neighbourhoods with three or more
police killings between 2003 and 2012 have a higher risk of illness than
those with less than three (Figure 2).
Sociodemographic characteristics of residents who live in lethally-sur-
veilled neighbourhoods (i.e. 3+ LIDS) are also distinct from those of residents
12 A. A. SEWELL ET AL.

Figure 2. Proportion of New Yorkers in ill health by LID community type. Source: 2003–
2012 EpiQuery NYC Vital Statistics; 2009–2013 NYC Community Health Survey.
Note: LID = Legal Intervention Deaths.

who live in less lethally surveilled areas (i.e. 0–2 LIDS). Means, proportions, and
frequency distributions, respectively, for all outcomes of interest and individ-
ual-level and neighbourhood-level covariates are shown. All comparisons are
statistically significant (p < 0.001) (Table 2).

Is illness associated with legal intervention deaths?


The analysis indicates people living in areas with a high count of LIDs between
2003 and 2012 are at an increased risk of reporting poor health along some of
the investigated dimensions of illness, holding constant gender status (Panel
A, Table 3). In communities with more than two LIDs between 2003 and 2012
compared to less lethally surveilled communities, there is: a 138 per cent
increase in the odds of reporting poor/fair health (OR = 2.375; 95 per cent
CI: [2.207,2.554]; p < 0.001); a 25 per cent increase in the likelihood of
having diabetes (OR = 1.251; 95 per cent CI: [1.197,1.306]; p < 0.001), a 53
per cent increase in the likelihood of reporting high blood pressure (OR =
1.530; 95 per cent CI: [1.504,1.556]; p < 0.001), and a 52 per cent increase in
the likelihood of having an asthma episode in the past year (OR = 1.523; 95
per cent CI: [1.459,1.590]; p < 0.001). No association exists with obesity;
however, supplemental analysis indicates that living in lethally surveilled
ETHNIC AND RACIAL STUDIES 13

Table 2. Means for variables of interest by community type (<3 LID vs. 3+ LID).
<3 LIDs 3+ LIDs Total
ILLNESS OUTCOMES
Poor/Fair Health 0.2159 0.2913 0.2373
Diabetes 0.1132 0.1568 0.1256
High Blood Pressure 0.3335 0.3850 0.3482
Asthma Episode 0.0382 0.0693 0.0470
Obese 0.2132 0.3183 0.2431
BMI 26.5716 28.3074 27.0655
GENDER STATUS
Women 0.5613 0.6366 0.5828
Mena 0.4387 0.3634 0.4172
ETHNORACIAL STATUS
Black Non-Latino 0.1672 0.3820 0.2283
Latino 0.1838 0.4608 0.2626
Asian/Pacific Islander 0.1074 0.0227 0.0833
a
White 0.5416 0.1344 0.4257
Age (in Years) 52.8657 49.2152 51.8268
MARITAL STATUS
Formerly Married 0.2760 0.3361 0.2931
Never Married 0.2471 0.3141 0.2661
Currently Married/Cohabitatinga 0.4769 0.3498 0.4408
EDUCATIONAL STATUS
a
Less than High School 0.1115 0.2404 0.1482
High School 0.2052 0.2670 0.2228
Some College 0.1985 0.2325 0.2082
College Degree 0.4848 0.2601 0.4208
EMPLOYMENT STATUS
Employed 0.5701 0.5225 0.5565
Unemployeda 0.4299 0.4775 0.4435
INCOME LEVELS
a
Income Less than Poverty Line 0.1576 0.3046 0.1994
Income 1–2X Poverty Line 0.1443 0.2058 0.1618
Income 2–4X Poverty Line 0.1451 0.1543 0.1477
Income 4–6X Poverty Line 0.1538 0.1146 0.1426
Income More than 6X Poverty Line 0.2524 0.0871 0.2053
Don’t Know Income 0.0807 0.1011 0.0865
Poverty Data Missing 0.0661 0.0325 0.0566
NATIVITY STATUS
U.S. Born 0.6161 0.5886 0.6083
Not Born in U.S.a 0.3839 0.4114 0.3917
HEALTH CARE INSURANCE
Insured 0.8889 0.8511 0.8781
Uninsureda 0.1111 0.1489 0.1219
NEIGHBORHOOD-LEVEL CONTROLS
Average Robbery Complaint Rate, 2010–2012 23.6765 135.6219 55.5338
Proportion Black or Latino, 2010 0.3706 0.8402 0.5042
Proportion Below Federal Poverty Line, 2010 0.1483 0.2923 0.1893
OBSERVATIONS 28,350 11,277 39,627
Source: 2009–2013 Community Health Survey; 2003–2012 EpiQuery.
Note: All t-statistics and are statistically significant (p < 0.001) using a two-tailed test.
a
Reference category.

areas is associated with a 73 per cent increase in the odds of being over-
weight/obese (OR = 1.734; 95 per cent [1.643,1.830]; p < 0.001), when only
gender status is held constant. Lethal police surveillance substantially
increases the risk of illness.
14 A. A. SEWELL ET AL.

Table 3. Multilevel mixed-effects model of the illness associations of legal intervention


deaths with adjustments for complex survey design.
PR(>2 LIDs)
Panel A: Gender Only
Poor/Fair Health 2.375***
[2.207,2.554]
Diabetes 1.251***
[1.197,1.306]
High Blood Pressure 1.530***
[1.504,1.556]
Asthma Episode 1.523***
[1.459,1.590]
Obese 1.023
[0.969,1.080]
Panel B: Gender + Individual-Level Covariates
Poor/Fair Health 1.292***
[1.175,1.422]
Diabetes 1.159***
[1.062,1.264]
High Blood Pressure 1.125***
[1.070,1.183]
Asthma Episode 1.059
[0.975,1.149]
Obese 1.168***
[1.102,1.239]
Panel C: Gender + Neighborhood-Level Covariates
Poor/Fair Health 0.833***
[0.790,0.877]
Diabetes 0.760***
[0.715,0.809]
High Blood Pressure 0.997
[0.955,1.042]
Asthma Episode 1.087***
[1.050,1.126]
Obese 1.238***
[1.143,1.341]
Panel D: Gender + All Covariates
Poor/Fair Health 0.797***
[0.725,0.876]
Diabetes 0.820***
[0.777,0.865]
High Blood Pressure 1.070*
[1.001,1.143]
Asthma Episode 0.889**
[0.820,0.962]
Obese 1.138***
[1.044,1.222]
Note: Adjusts for variation due to time fixed-effects and ceteris paribus gender differences. Panel A pro-
vides “Baseline Estimates”.
*p < 0.05; ** p < 0.01; *** p < 0.001 (two-tailed test; 95 per cent confidence interval in brackets).

Holding individual-level covariates constant reduces the odds of illness


associated with living in lethally surveilled areas for all outcomes of interest
(Panel B, Table 3). The odds of experiencing an asthma episode in the past
ETHNIC AND RACIAL STUDIES 15

year is reduced to statistical insignificance (OR = 1.059; 95 per cent CI


[0.975,1.149]; p = 0.174). Yet, in neighbourhoods with more than two LIDs
between 2003 and 2012, there remains a 29 per cent increase in the odds
of having poor/fair health (OR = 1.292; 95 per cent CI: [1.175,1.422]; p <
0.001), a 16 per cent increase in the odds of having diabetes (OR = 1.159; 95
per cent CI: [1.062,1.264]; p < 0.001), and a 13 per cent increase in the odds
of reporting high blood pressure (OR = 1.125; 95 per cent CI: [1.070,1.183];
p < 0.001). Holding individual-level covariates constant, the obesity risks of
living in lethally surveilled areas are stronger: the odds of being obese
increase by 17 per cent in these neighbourhoods (OR = 1.168; 95 per cent
CI: [1.102,1.239]; p < 0.001) compared to living in less lethally surveilled
neighbourhoods.
Neighbourhood-level controls, meanwhile, play a larger role in reducing
the odds of some illnesses (Panel C, Table 3). For instance, poor/fair health
and diabetes are statistically significant; yet, the direction of the associ-
ation indicates a reduced odds of reporting poor/fair health (OR = 0.833;
95 per cent CI: [0.790,0.877]; p < 0.001) and being diagnosed with diabetes
(OR = 0.760; 95 per cent CI: [0.715,0.809]; p < 0.001). Holding constant eth-
noracial composition, poverty concentration, and crime levels, there
remains, however, a 9 per cent increase in the odds of experiencing an
asthma episode in the past year (OR = 1.087; 95 per cent CI:
[1.050,1.126]; p < 0.001). LID concentration has a strong impact on the rela-
tive risks of reporting poor/fair health, diabetes, and asthma episodes.
Holding neighbourhood-level covariates constant also reveals the
obesity risks of living in lethally surveilled areas: the odds of being
obese increase 24 per cent in these neighbourhoods compared to living
in less lethally surveilled neighbourhoods (OR = 1.238; 95 per cent CI:
[1.143,1.341]; p < 0.001).
Once both individual- and neighbourhood-level controls are accounted for
(Panel D, Table 3), the likelihood of reporting poor/fair health in lethally sur-
veilled neighbourhoods is 20 per cent lower (OR = 0.797; 95 per cent CI:
[0.725,0.876]; p < 0.001), reporting diabetes is 18 per cent lower (OR = 0.820;
95 per cent CI: [0.777,0.865]; p < 0.001), and reporting asthma episodes is 11
per cent lower (OR = 0.899; 95 per cent CI: [0.820,0.962]; p < 0.01). However,
the illness associations of LIDs in the community remain for some illnesses
– high blood pressure and obesity. Even considering communities with
similar ethnoracial compositions, poverty concentrations, and robbery
reports, residents in lethally surveilled neighbourhoods face 7 per cent
increased odds of reporting high blood pressure (OR = 1.070; 95 per cent CI:
[1.001,1.143]; p < 0.05) and 14 per cent increased odds of being obese
(OR = 1.138; 95 per cent CI: [1.081,1.199]; p < 0.001). The findings suggest
illness-specific associations for individuals living in lethally surveilled
neighbourhoods.
16 A. A. SEWELL ET AL.

Does gender moderate the illness associations of legal intervention


deaths?
Table 4 presents the illness associations of LIDs from gender-stratified models
(i.e. separate models for women (Column 1) and men (Column 2)). The first

Table 4. Multilevel mixed-effects model of the gendered illness associations of legal


intervention deaths with adjustments for complex survey design.
Interaction Term (PR
Women: PR(>2 LIDs) Men: PR(>2 LIDs) [Women X > 2 LIDs])
Panel A: Baseline Estimates
Poor/Fair Health 1.391 *** 1.268 *** 1.102
[1.269,1.523] [1.131,1.421] [0.927,1.309]
Diabetes 1.426 *** 0.993 1.424 ***
[1.312,1.551] [0.880,1.121] [1.188,1.707]
High Blood Pressure 1.441 *** 1.172 *** 1.228 **
[1.351,1.537] [1.091,1.260] [1.077,1.400]
Asthma Episode 1.268 *** 1.310 0.965
[1.107,1.452] [0.945,1.816] [0.627,1.485]
Obese 1.566 *** 1.080 1.453 ***
[1.402,1.750] [0.967,1.207] [1.179,1.791]
Panel B: Individual-Level Covariates Only
Poor/Fair Health 0.913 0.825 ** 1.106
[0.818,1.018] [0.713,0.955] [0.900,1.359]
Diabetes 1.294 *** 0.861 1.502 ***
[1.138,1.470] [0.738,1.006] [1.216,1.855]
High Blood Pressure 1.181 *** 0.909 1.303 **
[1.075,1.296] [0.805,1.025] [1.100,1.545]
Asthma Episode 1.044 1.095 0.954
[0.909,1.201] [0.797,1.504] [0.627,1.451]
Obese 1.474 *** 0.955 1.542 ***
[1.331,1.633] [0.819,1.114] [1.230,1.934]
Panel C: Neighborhood-Level Covariates Only
Poor/Fair Health 0.875 * 0.794 *** 1.096
[0.787,0.974] [0.695,0.907] [0.923,1.301]
Diabetes 1.201 ** 0.835 ** 1.436 ***
[1.073,1.345] [0.748,0.933] [1.200,1.718]
High Blood Pressure 1.119 *** 0.912 * 1.227 **
[1.051,1.191] [0.833,0.997] [1.077,1.399]
Asthma Episode 1.232 *** 1.275 0.968
[1.092,1.389] [0.926,1.755] [0.629,1.487]
Obese 1.332 *** 0.914 1.457 ***
[1.190,1.490] [0.808,1.034] [1.182,1.797]
Panel D: All Covariates
Poor/Fair Health 0.916 0.832 * 1.103
[0.820,1.024] [0.714,0.970] [0.896,1.356]
Diabetes 1.026 0.681 *** 1.502 ***
[0.932,1.130] [0.596,0.779] [1.218,1.853]
High Blood Pressure 0.973 0.746 *** 1.304 **
[0.892,1.061] [0.670,0.830] [1.100,1.547]
Asthma Episode 1.017 1.063 0.959
[0.888,1.165] [0.793,1.426] [0.631,1.457]
Obese 1.262 *** 0.818 *** 1.539 ***
[1.112,1.433] [0.727,0.921] [1.227,1.930]
Note: Adjusts for variation due to time fixed-effects and ceteris paribus gender differences. Adjusted odds
ratios shown.
*p < 0.05; ** p < 0.01; *** p < 0.001 (two-tailed test; 95 per cent confidence interval in brackets).
ETHNIC AND RACIAL STUDIES 17

column shows that women living in lethally surveilled areas face a higher
prevalence of illness than women living in less lethally surveilled areas. By
contrast, in the second column, men display no such association and even
show evidence of reduced rates of some illnesses in lethally surveilled
areas. Women’s increased risk of poor/fair health (OR = 1.391; 95 per cent
CI: [1.269,1.523]; p < 0.001) is a function of both individual- and neighbour-
hood-level covariates, while women’s increased risk of asthma episodes
(OR = 1.268; 95 per cent CI: [1.107,.452]; p < 0.001) is a function of individ-
ual-level covariates. Both individual-level covariates (Panel B) and neighbour-
hood-level covariates (Panel C) reduce women’s increased risk of diabetes
and high blood pressure; yet, the risk of obesity remains (Panel D). Similarly,
men’s increased risk of poor/fair health (OR = 1.268; 95 per cent CI:
[1.131,1.421]; p < 0.001) and high blood pressure (OR = 1.172; 95 per cent CI:
[1.091,1.260]; p < 0.001) due to living in lethally surveilled neighbourhoods
(Panel A) is a function of both the sociodemographic characteristics of neigh-
bourhood residents (Panel B) and neighbourhood features (Panel C).
To ascertain gendered vulnerability in the health associations of living in leth-
ally surveilled neighbourhoods, we focus on the panels of final column – these
present estimates of the interaction term from the moderated model. This inter-
action term compares women and men living in the same neighbourhood and
evaluates the extent to which the illness associations of living in lethally sur-
veilled areas are the same for women and men. Living in areas with a high
count of LIDs does not have the same illness association for women and men:
This pattern holds for diabetes, high blood pressure, and obesity. Holding con-
stant gender status only (Panel A, Interaction Term), living in neighbourhoods
with more than two LIDs increases the odds that women are diagnosed with dia-
betes by 42 per cent compared to their male counterparts (OR = 1.424; 95 per
cent CI: [1.188,1.707]; p < 0.001). Similarly, in relation to their male counterparts
in lethally surveilled neighbourhoods, the odds of women being diagnosed with
high blood pressure are 23 per cent greater (OR = 1.228; 95 per cent CI:
[1.077,1.400], p < 0.01) and the odds of women being obese are 45 per cent
greater (OR = 1.453; 95 per cent CI: [1.179,1.791], p < 0.01). Apart from these
three conditions, LIDs do not differentiate the health of women and men.
Holding constant either individual-level covariates (Panels B, Table 3) or
neighbourhood-level covariates (Panel C, Table 3) does not reduce the mag-
nitude or statistical significance of the gendered moderation of diabetes, high
blood pressure, and obesity. Women’s increased likelihood of illness due to
living in lethally surveilled communities is not a function of the sociodemo-
graphic characteristics of the people who live in surveilled neighbourhoods
or a function of neighbourhood effects mechanisms. Once all controls are
considered (Panel D, Table 3), women’s risk of diabetes, high blood pressure,
and obesity associated with living in lethally surveilled areas remains higher
than that of comparable men’s risk.
18 A. A. SEWELL ET AL.

Discussion
Using robust multilevel models, we identify multiple ways that lethal policing
heightens the risks of morbidity for individuals. People living in lethally sur-
veilled areas are sicker than those who live in less lethally surveilled areas.
Moreover, women living in lethally surveilled neighbourhoods face a
greater risk of morbidity than comparable women living in less lethal neigh-
bourhoods. Finally, women face a greater risk of morbidity than their male
counterparts within lethal neighbourhoods.
We extend prior research identifying increased morbidity for those living in
aggressively surveilled areas (Sewell 2017; Sewell and Jefferson 2016) to the
study of lethal policing. Contextual forces associated with the concentration
of ethnoracially marginalized people, impoverished families, and crime
exposure drive much of the neighbourhood effect of lethal policing,
notably so for men. Yet, the gendered nature of morbidity risks related to leth-
ally policed neighbourhoods runs counter to prior research identifying
increased psychological risks for men living in neighbourhoods undergoing
more frisking and use of force (Sewell, Jefferson, and Lee 2016). Network mar-
ginalization appears to play a larger role in shaping illness associated with
lethal policing (Comfort 2009; Lee and Wildeman 2013). Women’s frustration
about police misconduct (Cooper et al. 2004) may also extend to concerns
about the safety of people in their social network. Police brutality “spills
over” onto women’s bodies vis-à-vis the criminalization of not only men
(Jones 2014), but also neighbourhood residents. Lethal policing piles
emotional, economic, and social burdens onto women who survive the carc-
eral state.
The study does have limitations. Primarily, our associations may be con-
founded by unmeasured variables. Compositional and contextual mechan-
isms considered herein may also be confounders. Specifically, they may
stand in for individual-level exposures to violence, police contact, and incar-
ceration, which were not considered due to a lack of data availability. Similarly,
psychological distress was not considered because of inconsistent measure-
ment across the study’s period. Moreover, neighbourhood-level wealth,
work status, and incarceration were not included due to the limited
degrees of freedom to detect areal variation. The presence of confounders
could inflate the possibility of Type I error affecting our study findings,
which is an important consideration since our main effect sizes are weaker
after adjusting for the mechanisms that are measured. Large odd ratios in
null models, however, suggest that these mechanisms play a substantial
role in patterning illness associated with lethally surveilled neighbourhoods.
Furthermore, the analysis relies on legal intervention classifications of
cause of death data, a policing outcome that is underreported (Feldman
et al. 2017). If underreporting occurs at random (i.e. misclassification is not
ETHNIC AND RACIAL STUDIES 19

differentiated by neighbourhood), this suggests a bias towards the null. Sup-


plementary analysis of other datasets identifying “legal intervention deaths”
from vital records indicate that EpiQuery reports mid-ranged estimates of
police killings. Specifically, during the study period, CDC WONDER (also
relying on vital records) reports 59 deaths, while NYPD reports 81 firearm-
related deaths of civilians. This presents a scenario in which 72.8 per cent of
LIDs are properly classified. Given that death investigations and cause-of-
death coding occurs in a centralized office in New York City and that we docu-
ment statistically significant associations for most of the health outcomes ana-
lyzed, misclassification is unlikely to affect the study results.
Moreover, the study employs neighbourhood-level exposure data that are rela-
tively sparse (LIDs between 2003 and 2012), and some outcomes data (2009–2011
health data) were collected before the final exposure to LIDs occurred (2010–2012
deaths). This temporal inconsistency represents a key drawback of pooling cross-
sectional repeated measures. Because the sparseness of the data led us to aggre-
gate data across a long time period, we do not attempt or intend to differentiate
the direction of the effect of living in lethally surveilled areas: Causality methods
are not used, and temporal order cannot be established. Yet, any further disaggre-
gation of the data would reduce our ability to detect variation. It is possible,
however, that people in poor health may deplete their assets in coping with
their illness, thus moving into more disadvantaged neighbourhoods that are
exposed to more lethal surveillance. However, studies of police killings among
communities within a municipality are uncommon given that the sheer rarity of
legal intervention deaths yields identifiable data. Larger scale studies covering
multiple cities for an extended period are necessary to confirm that the context
of legal intervention deaths does increase the likelihood of illness among individ-
uals. Our estimates only reflect correlations.
Future studies should examine whether neighbourhoods where people
spend the majority of their waking hours also matter. Other features of neigh-
bourhood dynamics should be considered. For example, spatial proximity to
areas overexposed to fatal police encounters may carry unique illness risks.
Data identifying legal interventions in death records does not publicly identify
the exact location of legal intervention deaths, so spatial analysis is not poss-
ible with legal intervention data. However, more exhaustive databases record-
ing publicly identifiable police killings does include geocodable data (e.g.
Fatal Encounters) and so would be better suited for spatial analysis. Moreover,
datasets covering a larger swath of areas or smaller neighbourhood units
would enhance the ability to assess the contribution of more neighbourhood
features and to disentangle the contribution of these factors from that of
behavioural factors (e.g. physical activity, smoking).
Future studies should also examine the social mechanisms driving gen-
dered associations between LID and health risk. While men’s health is
shaped by neighbourhood features, gender disparities in the illness risks of
20 A. A. SEWELL ET AL.

lethally surveilled neighbourhoods are independent of community sociode-


mographic factors and neighbourhood features. Morbidity may be higher in
areas with a higher incarceration burden or areas where police surveillance
and police killings are more normative – factors that may contribute to LIDs
being the sixth leading cause of death among men aged 25–29 (Edwards,
Lee, and Esposito 2019). Meanwhile, women’s physical health may be more
sensitive to potential confounders than men’s physical health. Moreover,
accounting for gender inequities in caretaking behaviours and vicarious
experiences of police contact would allow a stronger test of the role of neigh-
bourhood composition in shaping the illness associations of lethal police vio-
lence. Future research should investigate which mechanisms of multiple and
network marginalization underlie gendered associations.
Last, future research should examine the validity of these associations
through use of databases that identify the demographic characteristics of
decedents of police violence. Fatal Encounters (www.fatalencounters.org),
for instance, identifies 160 people killed during interactions with law enfor-
cement in New York City between 2003 and 2012, which suggests that the
state-sanctioned diagnostic category of “legal intervention death” was
only applied to 47 per cent of police killing cases. Investigating the ques-
tions that we pose here with such databases would allow an assessment
of whether the illness associations reported here represent a generic risk
of legal intervention deaths or a specific risk of fatal interactions with
unarmed, minority men. The use of this database would not only provide
a national picture of the associations identified here, but also enable an
estimate of the relative impact of underreporting (Feldman et al. 2017).
Nonetheless, by bridging levels of analysis and linking distinct data sources,
the multilevel design of this study offers inferences yielded from robust esti-
mates of a critical policy-relevant matter. Evidence that, after holding all else con-
stant, high blood pressure and obesity remains associated with living in lethally
surveilled areas suggests a traceable chronic vulnerability to surveillance stress
caused by police. These pathways deserve additional attention from researchers.
Police killings are acts of violence against both the individual and the commu-
nity. Just as mass incarceration has wide-ranging consequences for non-incar-
cerated individuals (Frank et al. 2013; Topel et al. 2018), police killings also
affect the health of the community at large. Lethal police violence is both a
social problem and a public health problem distributed unevenly across the city.
This study paints a chilling picture of women’s health: Living in lethally sur-
veilled areas locks women into a trajectory of morbidity stamped by diabetes,
high blood pressure, and obesity. Erica Garner died of a heart attack three
years after police squeezed her father to death. Yet, this study suggests
there are many more “Ericas”, women who live sick because they are called
upon to pick up the pieces that lethal policing leaves behind. Lethal policing
genders a sociopolitical health phenomenon.
ETHNIC AND RACIAL STUDIES 21

Disclosure statement
No potential conflict of interest was reported by the author(s).

ORCID
Alyasah Ali Sewell http://orcid.org/0000-0002-7146-3657
Justin M. Feldman http://orcid.org/0000-0002-8316-1947
Rashawn Ray http://orcid.org/0000-0003-1735-9194
Keon L. Gilbert http://orcid.org/0000-0001-7830-327X
Kevin A. Jefferson http://orcid.org/0000-0001-9699-1352
Hedwig Lee http://orcid.org/0000-0001-7376-5694

References
Alexander, Michelle. 2020. The New Jim Crow: Mass Incarceration in the Age of
Colorblindness. New York: The New Press.
Assari, Shervin, Maryam Moghani Lankarani, Cleopatra Howard Caldwell, and Marc A.
Zimmerman. 2016. “Fear of Neighborhood Violence During Adolescence Predicts
Development of Obesity a Decade Later: Gender Differences Among African
Americans.” Archives of Trauma Research 5 (2): e31475–e31e75.
Bor, Jacob, Atheendar S. Venkataramani, David R. Williams, and Alexander C. Tsai. 2018.
“Police Killings and their Spillover Effects on the Mental Health of Black Americans: A
Population-Based, Quasi-Experimental Study.” The Lancet 392 (10144):302–310.
Boyd, Melody L., and Susan Clampet-Lundquist. 2019. “‘It’s Hard to Be Around Here’:
Criminalization of Daily Routines for Youth in Baltimore.” Socius 5:
2378023118822888.
Boyd, Rhea W., Angela M. Ellison, and Ivor B. Horn. 2016. “Police, Equity, and Child
Health.” Pediatrics 137 (3): e20152711.
Brown, Tony N., Mary Laske Bell, and Evelyn J. Patterson. 2016. “Imprisoned by
Empathy.” Journal of Health and Social Behavior 57 (2): 240–256.
Browne, Simone. 2015. Dark Matters: On the Surveillance of Blackness. Durham, NC: Duke
University Press.
Brunson, Rod K. 2007. “‘Police Don’t Like Black People’: African-American Young Men’s
Accumulated Police Experiences*.” Criminology & Public Policy 6 (1): 71–101.
Brunson, Rod K., and Jody Miller. 2006. “Gender, Race, and Urban Policing: The
Experience of African American Youths.” Gender & Society 20 (4): 531–552.
Burton, Linda M., and M. Belinda Tucker. 2009. “Romantic Unions in an Era of Uncertainty:
A Post-Moynihan Perspective on African American Women and Marriage.” The
ANNALS of the American Academy of Political and Social Science 621 (1): 132–148.
Clear, Todd R. 2009. Imprisoning Communities: How Mass Incarceration Makes
Disadvantaged Neighborhoods Worse. New York: Oxford University Press.
Comfort, Megan. 2009. Doing Time Together: Love and Family in the Shadow of the
Prison. Chicago, IL: University of Chicago Press.
Cooper, Hannah L. F., Lisa Moore, Sofia Gruskin, and Nancy Krieger. 2004.
“Characterizing Perceived Police Violence: Implications for Public Health.”
American Journal of Public Health 94 (7): 1109–1118.
Davis, Dána-Ain. 2016. “‘The Bone Collectors’ Comments for Sorrow as Artifact: Black
Radical Mothering in Times of Terror.” Transforming Anthropology 24 (1): 8–16.
22 A. A. SEWELL ET AL.

DeGue, Sarah, Katherine A. Fowler, and Cynthia Calkins. 2016. “Deaths Due to Use of
Lethal Force by Law Enforcement: Findings From the National Violent Death
Reporting System, 17 U.S. States, 2009–2012.” American Journal of Preventive
Medicine 51 (5 Suppl 3): S173–SS87.
Desmond, Matthew, Andrew V. Papachristos, and David S. Kirk. 2016. “Police Violence
and Citizen Crime Reporting in the Black Community.” American Sociological Review
81 (5): 857–876.
DeVylder, Jordan E., Hyun-Jin Jun, Lisa Fedina, Daniel Coleman, Deidre Anglin,
Courtney Cogburn, Bruce Link, and Richard P. Barth. 2018. “Association of
Exposure to Police Violence with Prevalence of Mental Health Symptoms among
Urban Residents in the United States.” JAMA Network Open 1 (7): e184945–e45.
Edwards, Frank, Hedwig Lee, and Michael H. Esposito. 2019. “Risk of Being Killed by
Police use of Force in the United States by Age, Race–Ethnicity, and Sex.”
Proceedings of the National Academy of Sciences 116 (34): 16793–16798.
Eisenberg, Nancy, Richard A. Fabes, Mark Schaller, Paul Miller, Gustavo Carlo, Rick
Poulin, Cindy Shea, and Rita Shell. 1991. “Personality and Socialization Correlates
of Vicarious Emotional Responding.” Journal of Personality and Social Psychology 61
(3): 459–470.
Feldman, Justin M., Sofia Gruskin, Brent A. Coull, and Nancy Krieger. 2017. “Quantifying
Underreporting of Law-Enforcement-Related Deaths in United States Vital Statistics
and News-Media-Based Data Sources: A Capture–Recapture Analysis.” PLOS Medicine
14 (10): e1002399.
Forman, Tyrone A., David R. Williams, and James S. Jackson. 1997. “Race, Place, and
Discrimination.” Perspectives on Social Problems 9: 231–261.
Frank, Joseph W., Clemens S. Hong, S. V. Subramanian, and Emily A. Wang. 2013.
“Neighborhood Incarceration Rate and Asthma Prevalence in New York City: A
Multilevel Approach.” American Journal of Public Health 103 (5): e38–e44.
Gau, Jacinta M., and Rod K. Brunson. 2010. “Procedural Justice and Order Maintenance
Policing: A Study of Inner-City Young Men’s Perceptions of Police Legitimacy.” Justice
Quarterly 27 (2): 255–279.
Geller, Amanda, and Jeffrey Fagan. 2019. “Police Contact and the Legal Socialization
of Urban Teens.” RSF: The Russell Sage Foundation Journal of the Social Sciences 5
(1): 26.
Geller, Amanda, Jeffrey Fagan, Tom Tyler, and Bruce G. Link. 2014. “Aggressive Policing
and the Mental Health of Young Urban Men.” American Journal of Public Health 104
(12): 2321–2327.
Geller, William A, and Michael Scott. 1992. Deadly Force: What We Know: A Practitioner’s Desk
Reference on Police-Involved Shootings. Washington, DC: Police Executive Research Forum.
Gilbert, Keon L., and Rashawn Ray. 2016. “Why Police Kill Black Males with Impunity:
Applying Public Health Critical Race Praxis (PHCRP) to Address the Determinants
of Policing Behaviors and ‘Justifiable’ Homicides in the USA.” Journal of Urban
Health 93 (1): 122–140.
Goff, Phillip A., Tracy Lloyd, Amanda Geller, Steven Raphael, and Jack Glaser. 2016. The
Science of Justice: Race, Arrests, and Use of Force. Los Angeles, CA: Center for Policing
Equity.
Haldipur, Jan. 2018. No Place On the Corner: The Costs of Aggressive Policing. New York:
NYU Press.
Holmes, Thomas H., and Richard H. Rahe. 1967. “The Social Readjustment Rating Scale.”
Journal of Psychosomatic Research 11 (2): 213–218.
ETHNIC AND RACIAL STUDIES 23

Horton, Alex. 2018. “He Watched Police Kill His Mother. A Jury Just Awarded His Family
$37 Million.” The Washington Post, February 17. https://www.washingtonpost.com/
news/local/wp/2018/02/17/he-watched-police-kill-his-mother-a-jury-just-awarded-
his-family-37-million.
Hurst, Yolander G., M. Joan McDermott, and Deborah L. Thomas. 2005. “The Attitudes of
Girls Toward the Police: Differences by Race.” Policing: An International Journal of
Police Strategies & Management 28 (4): 578–593.
Johnson, Marilynn S. 2004. Street Justice: A History of Police Violence in New York City.
Boston, MA: Beacon Press.
Johnson, Odis, Christopher St. Vil, Keon L. Gilbert, Melody Goodman, and Cassandra
Arroyo Johnson. 2019. “How Neighborhoods Matter in Fatal Interactions between
Police and Men of Color.” Social Science & Medicine 220: 226–235.
Jones, Nikki. 2014. “‘The Regular Routine’: Proactive Policing and Adolescent
Development Among Young, Poor Black Men.” New Directions for Child and
Adolescent Development 2014 (143): 33–54.
Kessler, Ronald C., Kristin D. Mickelson, and David R. Williams. 1999. “The Prevalence,
Distribution, and Mental Health Correlates of Perceived Discrimination in the
United States.” Journal of Health and Social Behavior 40 (3): 208–230.
Krieger, Nancy, Jarvis T. Chen, Pamela D. Waterman, Mathew V. Kiang, and Justin
Feldman. 2015. “Police Killings and Police Deaths Are Public Health Data and Can
Be Counted.” PLOS Medicine 12 (12): e1001915.
Lee, Hedwig, Tyler McCormick, Margaret T. Hicken, and Christopher Wildeman. 2015.
“Racial Inequalities in Connectedness to Imprisoned Individuals in the United
States.” Du Bois Review: Social Science Research on Race 12 (2): 269–282.
Lee, Hedwig, and Christopher Wildeman. 2013. “Things Fall Apart: Health
Consequences of Mass Imprisonment for African American Women.” The Review
of Black Political Economy 40 (1): 39–52.
Lee, Hedwig, Christopher Wildeman, Emily A. Wang, Niki Matusko, and James S. Jackson.
2014. “A Heavy Burden: The Cardiovascular Health Consequences of Having a Family
Member Incarcerated.” American Journal of Public Health 104 (3): 421–427.
Lersch, Kim M., Thomas Bazley, Thomas Mieczkowski, and Kristina Childs. 2008. “Police
Use of Force and Neighbourhood Characteristics: An Examination of Structural
Disadvantage, Crime, and Resistance.” Policing and Society 18 (3): 282–300.
Lewinsohn, Peter M., and Joseph Talkington. 1979. “Studies on the Measurement of
Unpleasant Events and Relations with Depression.” Applied Psychological
Measurement 3 (1): 83–101.
Lewis, John. 2014. “Michael Brown, Eric Garner, and the ‘Other America’.” The Atlantic.
http://www.theatlantic.com/politics/archive/2014/12/michael-brown-eric-garner-ot
her-america-john-lewis/383750/.
Lutgendorf, Susan K., and Erin S. Costanzo. 2003. “Psychoneuroimmunology and
Health Psychology: An Integrative Model.” Brain, Behavior, and Immunity 17 (4):
225–232.
MacDonald, John, Robert J. Stokes, Greg Ridgeway, and K. Jack Riley. 2007. “Race,
Neighbourhood Context and Perceptions of Injustice by the Police in Cincinnati.”
Urban Studies 44 (13): 2567–2585.
Margolin, Gayla, and Elana B. Gordis. 2000. “The Effects of Family and Community
Violence on Children.” Annual Review of Psychology 51 (1): 445–479.
McFarland, Michael J., Amanda Geller, and Cheryl McFarland. 2019. “Police Contact and
Health among Urban Adolescents: The Role of Perceived Injustice.” Social Science &
Medicine 238: 112487.
24 A. A. SEWELL ET AL.

McFarland, Michael J., John Taylor, and Cheryl A. S. McFarland. 2018. “Weighed Down
by Discriminatory Policing: Perceived Unfair Treatment and Black-White Disparities
in Waist Circumference.” SSM - Population Health 5: 210–217.
McFarland, Michael J., John Taylor, Cheryl A. S. McFarland, and Katherine L. Friedman.
2018. “Perceived Unfair Treatment by Police, Race, and Telomere Length: A Nashville
Community-Based Sample of Black and White Men.” Journal of Health and Social
Behavior 59 (4): 585–600.
New York City Department of Health and Mental Hygiene. 2009–2013. “Community Health
Survey 2009, 2010, 2011, 2012, 2013.” Public use Dataset Accessed on June 17, 2015.
Neyfakh, Leon. 2015. “Why Was Sandra Bland Still in Jail?” Slate, July 22. http://www.
slate.com/articles/news_and_politics/crime/2015/07/sandra_bland_is_the_bail_
system_that_kept_her_in_prison_unconstitutional.html.
NYC Department of Health and Mental Hygiene. 2006. “New York City United Hospital
Fund (UHF) Neighborhoods and NYC ZIP Code Areas.” Accessed September 2, 2015.
http://www.nyc.gov/html/doh/downloads/pdf/survey/uhf_map_100604.pdf.
Raudenbush, Stephen W., and Anthony S. Bryk. 2002. Hierarchical Linear Models:
Applications and Data Analysis Methods. Thousand Oaks, CA: Sage.
Ross, Cody T. 2015. “A Multi-Level Bayesian Analysis of Racial Bias in Police Shootings at
the County-Level in the United States, 2011–2014.” PLoS ONE 10 (11): e0141854.
doi:10.1371/journal.pone.0141854.
Sampson, Robert J., and Charles Loeffler. 2010. “Punishment’s Place: The Local
Concentration of Mass Incarceration.” Daedalus 139 (3): 20–31.
Sewell, Abigail A. 2017. “Illness Associations of Police Violence: Differential Associations
by Ethnoracial Composition.” Sociological Forum 32 (S1): 975–997.
Sewell, Abigail A., and Kevin Jefferson. 2016. “Collateral Damage: The Health Effects
of Invasive Police Encounters in New York City.” Journal of Urban Health 93 (1): 42–67.
Sewell, Abigail A., Kevin A. Jefferson, and Hedwig Lee. 2016. “Living Under Surveillance:
Gender, Psychological Distress, and Stop-Question-and-Frisk Policing in New York
City.” Social Science & Medicine 159: 1–13.
Shedd, Carla. 2012. “What About the Other 99%? The Broader Impact of Street Stops on
Minority Communities.” In Key Issues in the Police Use of Pedestrian Stops and
Searches: Discussion Papers from an Urban Institute Roundtable, edited by Nancy La
Vigne, Pamela Lachman, Andrea Matthews, and S. Rebecca Neusteter, 24–29.
New York: The Urban Institute.
Sikora, Andrew G., and Michael Mulvihill. 2002. “Trends in Mortality Due to Legal
Intervention in the United States, 1979 Through 1997.” American Journal of Public
Health 92 (5): 841–843.
Skolnick, Jerome H, and James J. Fyfe. 1993. Above the Law: Police and the Excessive Use
of Force. New York: Free Press.
Smith, Brad W., and Malcolm D. Holmes. 2003. “Community Accountability, Minority
Threat, and Police Brutality: An Examination of Civil Rights Criminal Complaints*.”
Criminology 41 (4): 1035–1064.
Smith, Brad W., and Malcolm D. Holmes. 2014. “Police Use of Excessive Force in Minority
Communities: A Test of the Minority Threat, Place, and Community Accountability
Hypotheses.” Social Problems 61 (1): 83–104.
Stack, Carol B. 1997. All Our Kin: Strategies for Survival in a Black Community. New York:
Basic Books.
Sternthal, M. J., H. J. Jun, F. Earls, and R. J. Wright. 2010. “Community Violence and
Urban Childhood Asthma: A Multilevel Analysis.” European Respiratory Journal 36
(6): 1400–1409.
ETHNIC AND RACIAL STUDIES 25

Stults, Brian J., and Eric P. Baumer. 2007. “Racial Context and Police Force Size:
Evaluating the Empirical Validity of the Minority Threat Perspective.” American
Journal of Sociology 113 (2): 507–546.
Terrill, William, and Michael D. Reisig. 2003. “Neighborhood Context and Police Use of
Force.” Journal of Research in Crime and Delinquency 40 (3): 291–321.
Thoits, Peggy A. 2010. “Stress and Health: Major Findings and Policy Implications.”
Journal of Health and Social Behavior 51 (1_suppl): S41–S53.
Topel, Matthew L., Heval M. Kelli, Tené T. Lewis, Sandra B. Dunbar, Viola Vaccarino,
Herman A. Taylor, and Arshed A. Quyyumi. 2018. “High Neighborhood
Incarceration Rate is Associated with Cardiometabolic Disease in Nonincarcerated
Black Individuals.” Annals of Epidemiology 28 (7): 489–492.
Wang, Vivian. 2017. “Erica Garner, Who Fought Police Brutality, Dies at 27.” A20: The
New York Times.
Weitzer, Ronald, and Steven A. Tuch. 2014. “Race and Perceptions of Police
Misconduct.” Social Problems 51 (3): 305–325.
Whaley, Natalegé. 2020. “#SayHerName: Breonna Taylor, a COVID-19 Frontline Worker,
Killed by Police Violence.” Supermajority News. https://supermajority.com/news/
education-fund/sayhername-breonna-taylor-a-covid-19-frontline-worker-killed-by-
police-violence/.
Williams, David R., Yu Yan, James S. Jackson, and Norman B. Anderson. 1997. “Racial
Differences in Physical and Mental Health: Socio-Economic Status, Stress and
Discrimination.” Journal of Health Psychology 2 (3): 335–351.
26 A. A. SEWELL ET AL.

Appendix A. List of united hospital fund neighbourhoods by


legal intervention deaths.

Borough # of LIDs % of All LIDs


NO (0) LEGAL INTERVENTION DEATHS
Bayside Little Neck-Fresh Meadows Queens 0 0.00
Bensonhurst Brooklyn 0 0.00
Chelsea-Village Manhattan 0 0.00
Flushing Queens 0 0.00
Greenpoint Brooklyn 0 0.00
Jamaica Queens 0 0.00
Long Island City, Astoria Queens 0 0.00
Southwest Queens Queens 0 0.00
Union Square-Lower Manhattan Manhattan 0 0.00
Upper West Side Manhattan 0 0.00
West Queens Queens 0 0.00
ISOLATED (1) LEGAL INTERVENTION DEATH
Borough Park Brooklyn 1 1.33
East Harlem Manhattan 1 1.33
Kingsbridge Bronx 1 1.33
Ridgewood Queens 1 1.33
Southeast Queens Queens 1 1.33
Southern/Willowbrook Staten Island 1 1.33
SPORADIC (2) LEGAL INTERVENTION DEATHS
Canarsie Brooklyn 2 2.67
Coney Island Brooklyn 2 2.67
Downtown-Heights-Slope Brooklyn 2 2.67
East New York Brooklyn 2 2.67
Flatbush Brooklyn 2 2.67
Northern/Stapleton/St. George Staten Island 2 2.67
Sunset Park Brooklyn 2 2.67
Upper East Side-Gramercy Manhattan 2 2.67
EPISODIC (3–9) LEGAL INTERVENTION DEATHS
Fordham-Bronx Park Bronx 3 4.00
Rockaway Queens 3 4.00
Williamsburg-Bushwick Brooklyn 3 4.00
Washington Heights Manhattan 4 5.33
Central Harlem Manhattan 5 6.67
Northeast Bronx Bronx 5 6.67
Pelham-Throgs Neck Bronx 5 6.67
FREQUENT (10+) LEGAL INTERVENTION DEATHS
South Bronx/High Bridge/Morrisania Bronx 11 14.67
Bed-Stuy/Crown Heights Brooklyn 14 18.67
Source: EpiQuery, 2003–2012 (New York City Vital Statistics); Total LIDs = 75.
Note: LIDs = Legal Intervention Deaths (as classified by ICD-10).

View publication stats

You might also like