Executive Session on
Community Corrections
This is one in a series of papers that are being
published as a result of the Executive Session on
Community Corrections (2013-2017).
The Executive Sessions at Harvard Kennedy
School bring together individuals of independent
standing who take joint responsibility for
rethinking and improving society’s responses to
an issue. Members are selected based on their
experiences, their reputation for thoughtfulness,
and their potential for helping to disseminate the
work of the Session.
Members of the Executive Session on Community
Corrections came together with the aim of
developing a new paradigm for correctional policy
at a historic time for criminal justice reform. The
Executive Session worked to explore the role of
community corrections and communities in the
interest of justice and public safety.
Learn more about the Executive Session on
Community Corrections at:
Harvard’s website: http://www.hks.harvard.edu/
criminaljustice/communitycorrections
Integrated Health Care and Criminal Justice Data — Viewing the
Intersection of Public Safety, Public Health, and Public Policy
Through a New Lens: Lessons from Camden, New Jersey
Anne Milgram, Jerey Brenner, Dawn Wiest, Virginia Bersch, and Aaron Truchil
Introduction
At the intersection of public safety and public
health lies the potential to view crime prevention
through a new lens: the lens provided by
analyzing integrated data from the many
agencies that serve vulnerable populations. is
study involved the integration of health care and
criminal justice data for people who cycle in and
out of hospitals and police precincts in Camden,
New Jersey. Working pursuant to a grant from the
Laura and John Arnold Foundation, researchers
from the Camden Coalition of Healthcare
Providers (the Coalition) integrated existing
data sets to break down traditional information
silos, identifying and analyzing the experiences
of people who showed an extreme number of
contacts with both systems.
By analyzing these cross-sector data, Coalition
researchers found that a small number of
Camden residents have an enormous and
disproportionate impact on the health care
and criminal justice sectors, neither of which
is designed to address the underlying problems
they face: housing instability, inconsistent
or insufficient income, trauma, inadequate
nutrition, lack of supportive social networks,
Papers from the Executive Session on Community Corrections
APRIL 2018
Cite this paper as:
mental illness, and substance abuse disorders.
ese unaddressed social determinants of behavior
appear to drive a cycle of repeated arrests and
hospitalizations.
But the study’s potential impact goes well beyond
the identication of a population that frequently
cycles through the health care and criminal justice
systems. Cross-sector data oer a more holistic view
of the challenges these individuals face, telling a
dierent story than the one we typically hear — a
story with far-reaching public policy implications.
When we overlay data to view the trajectories of
lives through consecutive cross-sector contacts,
we begin to see that crime most often happens
after, and not before, contacts with hospitals and
other government agencies. During these earlier
encounters, we could find potential markers
that would allow us to identify individuals at risk
of future criminal justice involvement. In large
part because agencies are not sharing data in
the collaborative ways needed to gain a holistic
understanding of individuals, opportunities to
intervene earlier in their trajectories are lost.
Most interventions to prevent recidivism currently
occur during the community corrections and
re-entry phases, well after a crime has happened and
the individual’s case has ended. e study suggests
that we should shift from a mindset of reacting
to immediate health and crime crises as distinct
events to focusing on holistic approaches that result
in better individual outcomes, increased public
safety, and reduced system costs. e holistic view
provided by integrated data will allow researchers,
policymakers, and practitioners to design earlier
interventions to prevent crime and the avoidable
use of jails and emergency departments. The
Coalition’s researchers plan to design and test such
interventions in the next phase of this study.
is paper is organized in two parts.
Part I sets out the Camden study’s key ndings
from the analysis of integrated hospital and
police data:
•
A small percentage of arrestees account for a
disproportionate share of total arrests.
•
There is a relationship between high use of
hospital emergency departments (EDs) and
frequent arrests.
•
A small subset of 226 individuals had extreme
numbers of contacts with both hospital EDs and
police.
Part II outlines the potential impact of integrated
data analysis on public safety, public health, and
public policy:
•
Cross-sector data that look beyond the
criminal justice system, including data on
health, housing, employment, and other
socio-economic characteristics, provide a
holistic view of individuals and their contacts
with multiple systems over time.
Milgram, Anne, Jerey Brenner, Dawn Wiest, Virginia
Bersch, and Aaron Truchil. Integrated Health Care and Criminal Justice
Data — Viewing the Intersection of Public Safety, Public Health, and Public
Policy Through a New Lens: Lessons from Camden, New Jersey. Program in
Criminal Justice Policy and Management, Harvard Kennedy School,
April 2018.
Integrated Health Care and Criminal Justice Data | 3
• Integrated data reveal that, even within groups
of individuals who have frequent contacts with
multiple systems, there is signicant variety in
their experiences and behaviors, suggesting that
interventions will need to be tailored to meet
their unique needs.
•
Integrated data analysis is possible only
through cross-sector collaboration, which
breaks down data silos between agencies that
serve vulnerable populations. e information
gleaned by analyzing integrated data will
provide policymakers and practitioners with
tools and ideas to address the root causes of
crime by finding earlier intervention points
and designing strategies that can go beyond the
criminal justice system to include social services,
health care, and other community partners to
stop system cycling and prevent criminal justice
involvement in the rst place.
Study Overview
The Coalition, which comprises more than 25
members that include hospitals, primary care
providers, and community organizations working
to deliver high-quality health care to vulnerable
residents of Camden, began its work in the health
care sector. By many measures, Camden (a city
of approximately 77,000 people) is one of the
most challenged cities in America. According to
the American Community Survey (U.S. Census
Bureau, 2010–2014), the median household income
in Camden was estimated to be $26,201, with
approximately 39 percent of the population living
below the poverty line, including more than half
of the residents younger than age 18. Although
Camden has seen deep drops in violent crime over
the past decade, the city continues to struggle with
violence and other types of crime.
Since the Coalition’s founding by Dr. Jerey Brenner
in 2002, the use of integrated data has been one
of its core strategies for observing failures in the
health care system and driving system change. By
linking hospital claims data from multiple health
systems to create the Camden Health Database, the
Coalition sought to better understand and address
health trends in the city.
In 2011, a team of Coalition researchers, data
scientists, and programmers working with
integrated data sets from the city’s three hospitals
found that 10 percent of patients accounted for 75
percent of the more than $130 million in annual
hospital costs — nearly $100 million — which is
primarily paid for by Medicare and Medicaid. ese
high-cost, high-need patients tended to be older, to
suer from multiple chronic conditions, and to face
social challenges such as addiction, mental illness,
and housing instability, all of which posed barriers
to consistent medical care. In addition, they were
functioning within a highly fragmented and
uncoordinated health care delivery system, where
information was not shared across hospitals, EDs,
and primary care oces. Faced with such barriers,
these individuals cycled through multiple hospitals,
relying on expensive, hospital-based services
to treat medical problems as well as advanced
illnesses that could be better managed with
greater social supports and stronger relationships
to primary care.
The Camden Coalition’s “Health Care Hotspotting”
Identifying the outsized effect of these “super-
utilizers” on health care system costs was the rst
step. us, the Coalition employed a form of “health
care hotspotting,” which is the strategic use of
data to identify individual patients who are heavy
users of the health care system, engage and assess
them in real time guided by an authentic healing
relationship (Grinberg et al., 2016), and deploy
tailored evidence-based strategies to holistically
address their constellation of needs. e term was
inspired by the hotspotting data strategy used in
policing to identify geographic locations where
high levels of crime occur. Although health care
hotspotting is based on a premise similar to police
hotspotting, the Camden Coalition has most often
used the methodology to focus on individuals going
through the system rather than on places.
Once a high-needs, high-cost individual is
identied, the Coalition proactively engages them
with an intervention designed to address their
needs and change their patterns. Using new care
coordination and care management models to
work with this high-cost subset of the population,
the Coalition directs community-based support
teams of nurses, social workers, behavioral health
specialists, and community health workers to
identify and engage individuals in real time when
they are readmitted to the hospital — a catalytic
moment for behavior change — to mitigate
many of the underlying barriers contributing to
hospital readmissions. The work begins in the
hospital, but the team’s efforts extend into the
community, where they visit individuals in their
homes and accompany them to appointments
with primary care providers and social service
agencies, providing assistance in navigating the
fragmented and complicated health and social
service landscape that surrounds them.
Early attempts to quantify the impact of the
Coalition’s Care Management Intervention on
reductions in hospital use and health care spending
have shown promise. In a 2010 nonrandomized
evaluation, the intervention was found to improve
health outcomes, decrease the use of emergency
and inpatient services, and decrease charges for a
cohort of 36 high-cost members from $1.2 million
per month to $534,000 per month, a savings of
56 percent over five years (Green, Singh, and
O’Byrne, 2009). In 2012 and 2013, the Coalition
partnered with one of New Jersey’s managed care
organizations (MCOs) to test the model on a subset
of its most complex patients. Consistent with the
earlier study, the MCO found cost savings of more
than $10,000 per patient per year from reduced
hospitalizations and ED visits.
1
Camden ARISE: Integrating Hospital and Police
Data
Despite the success of the health care hotspotting
work, Coalition researchers realized that
health care data only tell one small piece of an
individual’s story. Taken alone, health care data
are woefully inadequate for understanding the
broader circumstances determining individuals’
heavy use of the hospital system. erefore, the
researchers began to pursue administrative data
from a variety of other sources to gain a deeper
Integrated Health Care and Criminal Justice Data | 5
understanding of their patients and to better
comprehend the drivers of repeated, avoidable
hospitalizations and poor health outcomes.
Camden ARISE (Administrative Records
Integrated for Service Excellence) was launched in
January 2015 to supplement the Coalition’s health
care data with information from other human
service domains.
e Camden County Police Department provided
the first non-health-care data in the form of
individual-level arrest information (Camden
Coalition of Healthcare Providers and Camden
County Police Department, 2014), which was
integrated into the Camden Health Database (the
citywide hospital claims data). Since the launch
of this study, Coalition researchers have entered
into further data-sharing agreements with two
additional health systems, the Camden City
School District, CamConnect (a local nonprofit
that conducted a citywide vacancy survey), and the
Southern New Jersey Perinatal Cooperative. It also
received data from the Camden County jails. is
paper focuses on ndings with respect to integrated
police and hospital claims data.
Part I: Findings
Working with J. Scott omson, Chief of Police for
Camden County, and the team from the Arnold
Foundation, the Coalition’s researchers developed
a hypothesis: ere is a relationship between the
factors that contribute to both negative public
safety outcomes and negative public health
outcomes. The researchers believed that many
of the same households and individuals who
had a disproportionate number of contacts with
Camden’s EDs would also have a higher number
of interactions with the criminal justice system.
If such individuals and households could be
identied, interventions could be designed and
tested to help reduce crime, improve health care,
and reduce system costs. In other words, the team
recognized that health care hotspotting could play
a dual role in both advancing public health and
preventing crime.
e study’s integrated data analysis revealed just
such a relationship. Several signicant patterns
have been identied in arrests, hospital use, and
the experiences of individuals who are involved in
dual-system cycling:
1. A small percentage of arrestees account for a
disproportionate share of arrests.
Health care hotspotting showed that individuals
who were frequently cycling through the health
care system accounted for a disproportionate share
of costs; crime hotspotting showed that frequent
recidivists accounted for a disproportionate
number of arrests. In Camden, 5 percent of adults
arrested by the police accounted for 25 percent of
all arrests over the data period.
2. There is a relationship between frequent use of
the hospitals and arrest.
The study found that many of the factors that
correlate with frequent hospital use also correlate
with a high risk for crime and criminal justice
involvement. Figure 1 shows the overlap between
those who had contacts with both systems. Of the
Figure 1. Individuals overlapping across police and hospital data, 2010-2014
Hospital
claims data
93,344
individuals
Police
arrest data
18,755
individuals
Overlap
12,541
individuals
(67%)
More arrests associated with a higher likelihood of hospital encounters
1 arrest 2 to 6 arrests 7 or more arrests
53%
72%77%
47%
28%
23%
Percent of individuals
with no hospital
encounters
Percent of individuals
with 1 or more
hospital encounters
18,755 individuals who were arrested, 12,541 (67
percent) also overlapped with the hospital’s claims
data set.
e study found that:
• A majority of all Camden arrestees (67 percent)
made a trip to the ED at least once during the
study’s timeframe (2010 to 2014), with more
than one-half (54 percent) of this group making
ve or more visits.
•
In addition to significant overlap across
populations, the study also found a relationship
between high use of EDs and frequent arrests;
only 11 percent of the individuals with one
visit to an ED over the period had an arrest,
compared with 20 percent of individuals with
two to ve ED visits and 32 percent of individuals
with six or more ED visits.
3. A small subset of individuals have high levels
of involvement with both hospitals and the
criminal justice sector.
When hospital claims data were overlaid with police
data, another trend became clear: A small subset of
individuals includes those who are both criminal
justice recidivists and super-utilizers of the hospital
system. ese 226 individuals had high contacts
with both systems, falling into the top 5 percent for
both the number of arrests and ED visits, with 16 or
more ED visits and seven or more arrests over the
ve-year study period. e researchers chose this
top 5 percent parameter for the subset because
they wanted to capture a sample that would be
large enough to identify meaningful proles of
individuals who would potentially be eligible
for and benet from prearrest diversion into an
intensive case management program, without
overestimating the size of this population or
complicating the interpretation of the subgroups
that emerged through data analysis. e 5 percent
cuto point generated a sample of individuals who
had extremely high contacts with both systems to
serve these exploratory purposes.
Figure 2 shows the distribution of individuals’
contacts with each system; those in the top 5
percent are indicated in green.
Integrated Health Care and Criminal Justice Data | 7
High levels of police involvement. e individuals
in the top 5 percent for both ED visits and arrests
were arrested primarily for nonviolent, low-
level offenses such as disorderly conduct, drug
possession, drinking in public, and loitering. Most
arrests in this group were for disorderly conduct
(42 percent of arrests) or for technical violations
(34 percent of arrests), which include arrests for
unspecied outstanding warrants. Over the ve
years, the 226 individuals with high dual-system
use were arrested a total of 3,686 times. Of these
Figure 2. Distribution of arrests and emergency department visits, 2010-2014
Arrests
Emergency department visits
top 5%
(16 or more emergency department visits)
top 5%
(7 or more
arrests)
Individual in police or hospital data Individual with high dual-system involvement
(7 or more arrests and 16 or more emergency department visits)
0
5
10
15
20
25
30
35
40
40
60
80
100
Note: Individual arrest totals were truncated at 40 and emergency department visit totals were truncated at 100 to suppress extreme outlier values.
arrests, only 176, or 5 percent, were for a violent
oense. Table 1 breaks down arrests in this group
by oense type.
2
Table 2 shows the number and percentage of the
226 individuals with extreme dual-system cycling
who had at least one arrest in each of the specied
oense categories. Over a ve-year period, nearly
all 226 individuals (96 percent) had at least one
arrest for disorderly conduct. A majority (65
percent) had at least one drug-related arrest, and
a minority (35 percent) were arrested at any point
for a violent crime, which includes weapons-related
oenses (35 percent, or 79 individuals) and property
crime (30 percent, or 64 individuals).
High numbers of hospital encounters. Among the
226 individuals, the median number of ED visits
over the ve-year period was 25. is group was
also far more likely to be admitted to the hospital
than the average arrestee: 67 percent of individuals
with high dual-system involvement were admitted
to the hospital during the study period, compared
with 17 percent of all arrestees.
High levels of socio-behavioral complexity. ese
individuals presented higher levels of overall
physical, mental, and social challenges than others
in the data set:
•
They were 50 percent more likely than
those with as many ED visits but fewer
arrests to have mental health or substance
use-related illnesses.
•
Seventy-ve percent received at least one mental
health-related diagnosis at the hospital.
•
Forty-two percent experienced homelessness at
least once during the study period.
3
•
The multifaceted issues facing this population
are shown in figure 3, which compares the
prevalence of socio-behavioral complications
of all individuals who overlapped both hospital
and police data against those 226 individuals
in the top 5 percent of ED visits and arrests. e
data show significantly higher occurrences
of substance use diagnoses, mental health
diagnoses, hospitalization as a result of violence
or assault, and homelessness among those with
extreme multisystem involvement.
Part II: Lessons From Camden
1. Data integration provides a new and more
holistic lens through which to view and improve
individuals’ lives.
Integrated data systems that link individual
information across sectors using common
identifiers provide a more complete look at an
individual’s life and thus a more meaningful and
complete understanding of the challenges he
or she faces. Abe
4
is one such individual who is
caught in this cycle; he has been arrested many
times and made repeated trips to Camden’s EDs.
Abe’s experience is typical of those identified
Table 1. Arrests by offense type for individuals with high dual-system involvement,
2010-2014
Table 2. Number and percent of 226 individuals with high dual-system involvement
arrested (1 or more times) by offense type, 2010-2014
Integrated Health Care and Criminal Justice Data | 9
Figure 3. Prevalence of socio-behavioral complexity among individuals with
dual-system involvement, 2010-2014
Total overlapping police/hospital population individuals with high dual-system involvement (7 or more
arrests and 16 or more emergency department visits).
*Evidence of homelessness is defined as having a homeless or shelter address at least once in either
data set.
by the Camden researchers as outliers who
have a disproportionate number of contacts
with both hospitals and the police. An excerpt
from the case history taken by researchers tells
Abe’s story (Camden Coalition of Healthcare
Providers, 2016):
Abe, a forty-year-old-man, was living on the
streets when he arrived at a Camden, New
Jersey emergency department. Earlier that day,
he had been involved in a minor altercation
with a friend. At the hospital, Abe was treated
for his injuries — a number of lacerations and
contusions — and was also diagnosed with major
depressive disorder. A few weeks later, he showed
up again at the emergency department. Because
of his unstable situation, he never completed the
six-week course of antibiotics that he had been
prescribed and his wounds had become infected.
Two months following these episodes, Abe was
walking across the street when he was hit by a
car. He was transported by ambulance to the
hospital where he remained for three days with a
fractured ankle and concussion. Over the course
of the hospitalization, he presented with suicidal
ideations and received treatment for alcohol
dependency.
Over five years, Abe was seen in every one of
Camden’s three emergency departments for a
total of more than two dozen times. Because of
his complex medical history — hypertension,
diabetes, and other chronic conditions; addictions
that escalated from alcohol dependency to
heroin use with sporadic overdoses; and mental
illness — a number of these hospital visits
resulted in admission. In total, Abe spent more
than 45 days in the hospital, accumulating over
$400,000 in bills. The hospitals were able to
recoup $27,000 through his occasional Medicaid
coverage, but the vast majority of these costs were
passed on to the system as charity care.
But Abe’s hospitalizations tell only one part of
the story. During the same ve years, Abe was
arrested more than fteen times, mostly for low-
level oenses. He was picked up several times
by police for repeatedly shoplifting from a store
outside the hospital, just moments after he had
walked out of the emergency department. In
between these run-ins, Abe was also commonly
booked for being severely intoxicated in front of
the same handful of corner liquor stores. Each
time, he was booked by police, given a summons
to appear in court, and then released. After
failing to show up in court and pay his fees, a
warrant was put out for his arrest, and the next
time Abe was picked up he spent a month in the
Camden County Jail. Over the ve-year period,
Camden police officers spent over 120 hours
either directly with Abe or writing up encounter
notes. He spent about as much time in jail as
he did in the hospital, to say nothing of the as
yet untallied cost of courts, fees, and nes, and
the devastating human cost to Abe and those
around him.
Abe’s living situation remained unstable over
this entire period. He moved back and forth
between living with family members or friends
to homeless shelters and the streets. Each time
he left the care of the hospital or was released
from criminal justice custody, Abe returned to
the unstable environment that contributed to
his hospitalization or arrest, no better prepared
to deal with the challenges he would face. Each
successive interaction with the healthcare or
criminal justice system likely exacerbated his
volatile state, making the next hospitalization
or arrest even more likely.
Figure 4 illustrates Abe’s cycles through the
criminal justice and hospital systems. It is a map
of system contacts and missed opportunities to
intervene in ways that might have changed his
trajectory. Contacts with the health care system
are reected above the gray bar in gure 4, with
blue lines denoting the times that Abe’s ED visits
resulted in a hospital admission. Housing issues are
noted in the center of the gray bar. Interactions with
the criminal justice system are shown below the
gray bar, with each thin line marking contact with
law enforcement and each thicker line representing
a period of incarceration.
As Abe’s story and figure 4 demonstrate, the
criminal justice and health care systems tend to
function on an event-by-event basis: An incident
occurs, the system responds, and the person is
released into the same setting that prompted the
criminal act or health crisis in the rst place. Both
systems treat the immediate trigger event — an
illness or injury that requires medical treatment
or an arrest for a violation of the law — as the
focus of their eorts. To hospitals, the illness is the
problem; to the criminal justice system, the crime
is the problem. Both systems also typically operate
in information silos, recording and maintaining
data separate not only from each other, but also
in isolation from other agencies that monitor the
larger economic and social circumstances that
exacerbate both crime and poor health. is lack of
cross-sector collaboration shown by the data is not
unique to Camden, but represents the typical siloed
nature of agency information about vulnerable
populations in the United States. Camden is
among the rst cities in the nation to do this type
of integrated data work,
5
and the possibilities for
cross-sector collaboration on solutions are just
beginning to become apparent.
2. The individuals who have an extremely high
number of contacts with both hospitals and the
criminal justice system exhibit significant
variation in their experiences and behaviors.
Even within the small subset of 226 people in the
top 5 percent of dual-system users, researchers
identified four distinct profiles based on the
nature of their medical, behavioral, and social
needs. Although they share some characteristics in
common, the following subgroups reect dierent
experiences and behaviors that suggest they will
require dierent strategies for engagement, whether
they are encountered in an ED, by a police ocer on
the street, or through some type of newly designed
intervention:
Integrated Health Care and Criminal Justice Data | 11
2011 2012
January
2010
December
2014
inpatient
stays
emergency
visits
theft-related
public nuisance
county jail
20132013
assaulted
assaulted
major depression & suicidality
diagnosed as homeless by hospitaldiagnosed as homeless by hospital diagnosed as
homeless
by hospital
struck
by a
car
arrest warrant
violence
accidents, fights,
& social correlates
injuries
other
addiction
mental health
wounds
&
infection
heroin
overdose
Diagnosis Categories
Police Encounter Types
primary diagnosis
Housing Type
residential “homeless”shelter
Figure 4. Case study of a cross-sector complex-care patient
•
Nonviolent, medically complex drug
offenders. These individuals are most likely
to be males between the ages of 18 and 29. In
addition to a history of arrest for drug possession,
they have been arrested many times in the
past for disorderly conduct, but never for a
violent offense. They have a high degree of
medical complexity, including a markedly
higher prevalence of HIV compared to the
other subgroups, and are often admitted to the
hospital. ere is a 50/50 chance that they have
been seen at the hospital with a drug overdose.
ey also have a history of behavioral health
struggles and most likely have been diagnosed
with a serious mental illness or a severe drug
abuse-related condition.
•
Nonviolent individuals with behavioral health
complexity who are arrested predominantly
for petty crimes. These individuals are
typically male, age 40 or older. Although they
frequently visit the city’s EDs, they are less
likely to be admitted to the hospital than those
in other subgroups. They are often arrested,
predominantly for petty crimes (rarely for
drug possession or trafficking). They are the
most likely of individuals in any subgroup to
experience housing instability, with a 50-percent
likelihood of having been homeless at least once
during the study period.
•
Assault victims with mental health challenges
and addictions who commit crimes against
other persons. Individuals in this subgroup
are typically women age 40 or younger. ey are
the most likely of individuals in any subgroup to
be arrested for a violent crime, predominantly
simple assault, and are rarely arrested for drug
oenses. ey are frequently admitted to the
hospital, often for assault-related injuries,
suggesting that while they may perpetrate
violence in some instances, they may also be
heavily victimized.
6
•
Male drug oenders, some with violent crime
arrests, who have few hospitalizations and
a comparatively low prevalence of serious
mental illness. ese individuals are most often
young males. ey have frequent contact with
the police and have been arrested for a wide
array of offenses, including drug trafficking,
property crime, and violent crime. However,
they are less likely than those in the other
subgroups to suffer from mental illness or
alcohol or substance abuse disorder. ey are
also less likely to visit the ED or be admitted to
the hospital.
ese subgroups illustrate that the forces underlying
cycling behavior can differ from individual
to individual, even as some characteristics
and experiences are similar. No “one size fits
all” approach will be sufficient in serving this
population. Instead, an understanding of the
dierent constellations of factors that place these
individuals at risk for prolonged entanglement in
crisis systems, coupled with an enhanced ability
to assess individuals more robustly, could lead to
more eective policies and strategies.
3. Only by breaking down data silos among
agencies that serve vulnerable populations can
we begin to address the root causes of behavior
and prevent individuals from cycling through
multiple systems.
The potential that integrated data holds to shift
both policy and practice is possible only through
collaboration and data sharing among the agencies
that serve individuals and families that overlap
multiple systems. e study’s ndings raise the
question of whether we treat the symptoms of
conditions such as substance abuse, mental illness,
and homelessness — frequent emergency medical
problems and repeat commission of low-level,
nonviolent crimes — as short-term problems to
be solved by hospitals and jails and not as needs
to be met through deeper treatment of underlying
diseases and problems. In other words, have we
chosen a seemingly quick x, where we repeatedly
funnel people who need treatment into our jails
and hospitals, over solutions that foster the long-
term safety and well-being of communities? e
result is seen in stories like Abe’s — a seemingly
unbreakable cycle of hospital stays and arrests and
incarceration, punctuated by periods of housing
instability and homelessness, all of which appear
to be driven largely by untreated substance abuse
and lack of social supports.
e United States makes enormous expenditures
after a crime has been committed, but makes a
much lower comparative investment in working
to prevent crime and recidivism by addressing
the underlying needs of at-risk populations. Yet,
recidivism rates in the United States remain high.
Fully 77 percent of all individuals released from
Integrated Health Care and Criminal Justice Data | 13
jail or prison are rearrested for a new crime within
ve years of release (Durose, Snyder, and Cooper,
2015). e focus of most interventions to prevent
future crime has been at the prisoner re-entry
and community corrections phases, targeting
individuals released into the community under
parole or probation supervision.
7
Interventions
at this post-disposition phase have been evolving
in recent decades. Research and practice are now
moving away from a sole focus on criminogenic
risk to a growing emphasis on services and
interventions tailored to an individual’s crime-
producing risk factors and responsivity to treatment.
The Camden study makes a compelling case for
moving such interventions earlier. Applying cross-
sector methods to support individuals sooner
may address the determinants that propelled the
criminal behavior in the rst place, such as mental
illness, substance abuse disorder, or homelessness.
8
If we care about public safety, fairness, and cost
eectiveness, we need to better understand the
lived realities of the people in the criminal justice
system by gathering and analyzing data from the
various agencies that serve vulnerable populations.
It is not that the well-being of communities will be
improved by the development of multisector data
systems alone; it is that integrated data oer the
potential for integrated solutions. At present, the
health care and criminal justice systems react
piecemeal to each ED visit or arrest, because neither
system was developed to deal with the underlying
societal problems that drive recidivism and repeat
hospital visits.
9
Yet, each arrest or hospitalization
oers an opportunity to intervene. Each moment
of contact provides an opportunity for change and
a chance to stabilize someone caught in this cycle,
and the failure to do so eectively has enormous
individual and societal costs.
e information provided by integrated data, for
example, could allow for earlier engagement of
at-risk youth, potentially changing their paths
so they never come in contact with the criminal
justice system in the first place (Sampson and
Laub, 1990).
10
Such data also have value as a police
tool that allows agencies to craft a means of law
enforcement-led diversion. This is happening
already on a small scale in Seattle as part of its
Law Enforcement Assisted Diversion (LEAD)
program, a prebooking diversion pilot program
developed with the community to address low-
level drug and prostitution crimes in several
King County neighborhoods.
11
LEAD has been
successful in reducing recidivism in this group by
60 percent (Collins, Lonczak, and Clifase, 2015).
e LEAD program relies on selecting individuals
by geographic location and oense type, without
having access to integrated data. By making law-
enforcement diversion more data driven and
tailored to the underlying needs of individuals,
the Camden project has identied a larger, broader
group of nonviolent oenders for possible diversion,
giving it the potential to prevent more crime.
With the information gleaned from integrated
data, the criminal justice system can begin to
develop targeted approaches to address the issues
underlying crime. There remains much work to
be done to build and test these interventions. Yet,
at a time when crime ranks as the second largest
spending item in most state budgets and as the
largest spending item in many American cities, we
simply cannot aord to run our criminal justice
system as it currently operates. e savings — both
human and nancial — that can be captured by
using integrated data to understand when, where,
and with whom to intervene to prevent crime, can
be tremendous.
Building the frameworks necessary for cross-
sector data sharing presents challenges. Individual
agency data collection procedures rarely result in
uniform identiers that allow researchers to easily
link a person’s contacts with the system across
sectors. Organizational culture may also inhibit
the sharing of data. e public may have further
concerns with privacy and data security. And some
may argue that legal barriers exist to sharing this
kind of information. But the Camden study shows
that these challenges can be overcome and that
the value of the resulting information is worth
any diculty. If we can begin to break down the
silos between the systems that serve vulnerable
populations, we can better focus on improving
outcomes.
Next Steps: The Camden Coalition Model
While the administrative data that the Coalition’s
researchers gathered are useful, they still only tell a
story about individuals once they enter the orbit of
these institutions. We need even more information
from other spheres of life to make the picture as
complete and accurate as possible. To ll in these
missing pieces, the Coalition is beginning to seek
out qualitative data to supplement its quantitative
database and to explore bright spots — people who
are absent from the data because of successes — in
order to identify and learn from individuals who
researchers would expect to have high dual-system
overlap but don’t.
12
Such data hold the potential
to identify system contacts, assess warning signs,
highlight redundancies and inefficiencies, and
reveal more appropriate ways to engage this at-risk
population.
Currently, the Camden study is focused on
gathering and analyzing information. In the next
phase, researchers will turn their attention to
designing and testing interventions to prevent
the cycling the data have shown. The Coalition
is partnering with the Camden County Police
Department, criminal justice agencies throughout
the city, and community organizations to adapt its
existing care management intervention to a more
heavily criminally justice involved population. is
“Camden Model” posits that “if we can identify and
coordinate treatment and services around the risk
factors contributing to repeat hospitalizations
and arrests, we will not only improve outcomes
for Camden’s most at-risk and vulnerable citizens,
we will reduce costs to both the healthcare
and criminal justice systems, unleashing vital
resources for investment in other critical areas”
(Camden Coalition of Healthcare Providers, 2016).
e Camden Model seeks to:
•
Ascertain optimal intervention points by
focusing on how and when at-risk individuals
come in contact with various systems.
•
Design optimal intervention strategies
based on individual typologies that result in
Integrated Health Care and Criminal Justice Data | 15
better outcomes and reduced system costs,
including, where appropriate, alternatives to
arrest and jail.
13
Ultimately, integrated data can lead to better
individual outcomes, reduced crime and system
cycling, and increased efficiency by directing
resources where they will have the most impact.
Further Questions for Research
e integrated data research being done in Camden
is only a starting point. At this time, no causal links
have been found. Many correlations exist, but
there is no proof of causation. Additional research
analyzing cross-sector integrated data in Camden
and beyond is needed to develop interventions to
end cycling between hospitals and jails. Questions
for future research include:
•
What are the burdens of and barriers to data
sharing across systems and sectors?
•
What is causing the overlap between individuals
who have frequent contacts with multiple
systems serving vulnerable populations?
•
What integrated solutions to the problems of
multiple-system cycling can be piloted and
evaluated?
•
To what extent does this research integrate with
research on juveniles, predicting dropout or the
“school-to-prison pipeline”? How early should a
pilot intervention start?
•
What is the right unit to target as a means
of efficiently allocating resources for such
interventions? Is it a residential building? A
family? A neighborhood? Individuals within a
particular subgroup? (See Desmond and An,
2015.)
14
Conclusion
e Camden study provides a new high-level view
of a public policy problem that plagues many cities:
How to identify and treat vulnerable individuals
who cycle frequently through multiple systems. It
also raises the question of whether we can provide
long-term solutions to the underlying problems that
drive the frequent use of hospitals and repeat arrests
for crime. In the end, the most important nding
from this study may be that there is enormous value
in fostering collaborative data sharing among
agencies. By highlighting the power of cross-sector
integrated data to unlock key insights into at-risk
populations, this study showcases the potential
of cross-sector collaboration to provide better
outcomes in public safety and public health.
Endnotes
1. Both of these early evaluations are potentially
vulnerable to regression to the mean, which in
this case would be an individual’s hospital use
decreasing on its own due to the natural course
of the disease or another reason unrelated to the
intervention. To combat this methodological
challenge, the Coalition recently partnered with
Massachusetts Institute of Technology’s Abdul
Latif Jameel Poverty Action Lab (J-PAL) to conduct
a randomized controlled trial — the gold standard
for demonstrating causality — to evaluate its
care management intervention. The Coalition
is currently four-fifths of the way through its
anticipated enrollment projects, and the evaluation
is expected to be completed in 2018.
2. Violent crime arrests for this subgroup broke
down as follows: simple assault (38 percent),
weapons charges (30 percent), aggravated assault
(26 percent), and robbery (6 percent).
3. Housing status was determined through a
combination of the patient’s self-reported address
and ICD9 V-Codes, which are known to be
underdocumented and thus an underestimate of
homelessness.
4. To protect privacy, the individual’s name has
been changed and other identifying details have
been modied and/or removed.
5. While the Coalition’s study is unique in
integrating hospital claims data with police data,
projects in several other geographical areas have
integrated health and criminal justice or police data,
including (but not limited to) Antioch, California
Youth Intervention Network; Philadelphia CARES;
Allegheny County Data Warehouse; Los Angeles
Enterprise Linkage Project; Washington State
Integrated Client Database; and Florida Policy and
Services Research Center.
6. When researchers sorted out a cluster of 38
individuals in that group who did show arrests for
violent behavior, they found that before their rst
arrest for a violent oense in the data, 75 percent
had been seen at the hospital for violence directed
at them. Thirty-seven percent (14 individuals)
had been seen two or more times for this reason
before the rst arrest for a violent oense found
in the data. ese individuals also showed socio-
behavioral complexities before their first arrest
for a violent oense; 68 percent had at least one
previous substance use- or mental health-related
hospitalization or ED visit. More than half of these
individuals had two or more such substance use-
or mental health-related hospital visits before their
rst arrest for a violent oense in the data set; ve of
the individuals had 10 or more. All of these contacts
with the system before an arrest for a violent oense
represent opportunities for intervention or chances
to assess warning signs for future violence and
suggest that violence must spring from somewhere;
root causes of the behavior may show themselves
years before violence erupts. In one study of 7,222
seriously mentally ill homeless adults, for example,
the authors found that at least some proportion of
the arrests of this sample were of those who had
been exhibiting antisocial behavior since early
adolescence, and that early antisocial behavior was
a strong predictor of all types of recent arrests in
this population (Desai, Lam, and Rosenheck, 2000).
7. The community corrections population is
roughly double that of America’s jails and prisons.
At the end of 2013, approximately 2.2 million
individuals were incarcerated in U.S. prisons and
jails, and more than 3.9 million people were on
probation (Glaze and Kaeble, 2014).
8. Previous research looking at specific
populations has found linkages among these social
determinants, crime, and health; see, for example,
Tsai and Rosenheck (2012), who found that study
participants with extremely long incarceration
periods had worse substance use outcomes than
Integrated Health Care and Criminal Justice Data | 17
those with no history of incarceration. Greenberg
and Rosenheck (2008) analyzed the 2002 national
survey data on 6,963 inmates, concluding that
homelessness increases the risk of incarceration
and vice versa; mental illness, substance abuse,
and disadvantageous socio-demographic
characteristics amplify this risk. Western (2007)
found a strong relationship between incarceration
and severely dampened economic prospects
among the formerly incarcerated, perpetuating a
damaging cycle of broken families, poverty, and
crime. Metraux and Culhane (2006) showed that
23.1 percent of individuals staying in a Department
of Human Services single adult shelter in New York
City on December 1, 1997, had been incarcerated
within the previous two-year period. McNiel and
Binder (2005) examined archival databases on
the use of psychiatric emergency services in San
Francisco, nding that homeless individuals with
mental disorders accounted for a large proportion
of users and further noting that “[t]he co-occurrence
of homelessness, mental disorder, substance abuse,
and violence represents a complicated issue that
will likely require coordination of multiple service
delivery systems for successful intervention. . . .
Simply diverting individuals with these problems
from the criminal justice system to the community
mental health system may have limited impact
unless a broader array of services can be brought
to bear.” Kushel and colleagues (2005) found that
while there are high levels of health risks among all
homeless and marginally housed people, the levels
among homeless former prisoners were higher.
Martell, Rosner, and Harmon (1995) studied 254
defendants referred for psychiatric examination
by Manhattan courts over a six-month period and
found that homeless, mentally ill persons appear
to be grossly overrepresented among mentally
disordered defendants entering the criminal justice
and forensic mental health systems.
9. e ineciencies created by such a piecemeal
approach are highlighted by the work of Goerge
and colleagues (2010). In an analysis of government
program participation among Illinois families,
Bob Goerge and his team of researchers found
an overlap in five social programs: foster care,
mental health services, substance abuse treatment,
juvenile corrections, and adult corrections. He
found that agencies “tend[ed] to treat all of the
people they serve with their own services and
programs, not with coordinated approaches
across agencies and systems” and that “[s]tate and
local agencies primarily respond to crises dened
by single problems happening at a point in time”
rather than focusing on early intervention to
prevent future problems.
10. Childhood delinquency has been linked to “adult
crime, alcohol abuse, general deviance, economic
dependency, educational failure, unemployment,
and divorce” (see Sampson and Laub, 1990).
11. According to its founders, the LEAD program
“allows law enforcement ocers to redirect low-level
oenders engaged in drug or prostitution activity
to community-based services, instead of jail and
prosecution. By diverting eligible individuals to
services, LEAD is committed to improving public
safety and public order and reducing the criminal
behavior of people who participate in the program.”
LEAD: Law Enforcement Assisted Diversion,
http://leadkingcounty.org.
12. While integrated data systems allow us to
combat fragmentation of services and provide a
more holistic understanding of an individual’s well-
being, it is also important to note the limitations
of administrative data. Administrative data
are fundamentally reactive; they only capture
information about individuals who enter the
orbit of an institution, not those who are unable
or unwilling to do so or who are not in need at a
given time. As integrated administrative data
continue to gain traction, it is critical that we strive
to ll in what is potentially missing, both through
qualitative methods to supplement the quantitative
data and through “bright spotting” — identifying
and learning from individuals who we would expect
to have poor outcomes but don’t.
13. Some researchers and localities have been
testing innovative solutions to problems of frequent
recidivism and multisystem cycling. Housing First
programs, for example, provide nonabstinence-
based housing for the chronically homeless, even
those with alcohol or substance abuse issues who
are frequently ineligible for public shelter systems,
as a way to enhance public safety and personal
health and reduce costs. Clifasefi, Malone, and
Collins (2013) found that participants’ criminal
histories reected “symptoms” of homelessness
rather than threats to public safety, and that
exposure to Housing First was associated with
decreased jail time for up to two years (Mackelprang,
Collins, and Clifase, 2014). Perhaps the most well-
known program is 1811 Eastlake in Seattle, a “wet”
housing model that does not require sobriety from
its residents. Researchers have found that 1811
Eastlake has saved taxpayers more than $4 million
in costs for publicly funded services, including jail,
detox center use, hospital-based medical services,
alcohol and drug programs, and emergency medical
services (Larimer et al., 2009). Other intriguing
interventions include assertive community
treatment programs, which are designed to help
individuals with severe mental illness who are at
risk of homelessness and hospitalization to become
integrated into their communities through the
use of round-the-clock mobile services (Lamberti,
Weisman, and Faden, 2004). ProjectLinks is a
program designed to prevent individuals with
severe mental illness from entering the criminal
justice system by integrating criminal justice,
health care, and community support services (see
Weisman, Lamberti, and Price, 2004). e Frequent
Users Service Enhancement or “FUSE” initiative in
New York City, which provided supportive housing
to 200 people with complex involvement in multiple
public systems, resulted in reduced cycling in
and out of jails and homeless shelters (see Aidala
et al., 2013). e Transition Clinic, a health care
program operated by physicians, community
organizations, and representatives of the San
Francisco Department of Public Health’s safety-net
health system, provides transitional and primary
care, as well as case management, for prisoners
returning to the community (see Wang et al., 2010).
14. Desmond and An’s study of Milwaukee
renters is instructive, examining the impact of
neighborhoods versus social networks on various
types of disadvantage.
Integrated Health Care and Criminal Justice Data | 19
References
Aidala, A., McAllister, W., Yomogida, M., and
Shubert, V. (2013). Frequent Users Service
Enhancement “FUSE” Initiative: New York City
FUSE II Evaluation Report. New York, NY: Columbia
University Mailman School of Public Health.
Camden Coalition of Healthcare Providers.
(2016). Camden ARISE: Using integrated data to
unlock key insights into vulnerable populations.
Unpublished report.
Camden Coalition of Healthcare Providers
and Camden County Police Department.
(2014). “Memorandum of Understanding By and
Between the County of Camden (Department
of Police Services) and the Camden Coalition of
Healthcare Providers.” Available at https://www.
camdenhealth.org/wp-content/uploads/2015/10/
Camden-County-Police-Department-and-
Camden-Coalition-Data-Sharing-Agreement.pdf.
Clifasefi, S.L., Malone, D.K., and Collins, S.E.
(2013). Exposure to project-based Housing First is
associated with reduced jail time and bookings.
The International Journal of Drug Policy 24(4):
291-296. Available at http://dx.doi.org/10.1016/j.
drugpo.2012.10.002.
Collins, S.E., Lonczak, H.S., and Clifasefi,
S.L. (2015). LEAD Program Evaluation:
Recidivism Report. Seattle, WA: University
of Washington, Harborview Medical Center,
Harm Reduction Research & Treatment Center.
Available at http://static1.1.sqspcdn.com/
static/f/1185392/26121870/1428513375150/
LEAD_EVALUATION_4-7-15.pdf?token=XmU8bJt
jxSU1PFARuW2Jgw805nM%3D.
Desai, R.A., Lam, J., and Rosenheck, R.A. (2000).
Childhood risk factors for criminal justice
involvement in a sample of homeless people with
serious mental illness. Journal of Nervous and
Mental Disease 188: 324-332.
Desmond, M., and An, W. (2015). Neighborhood
and network disadvantages among urban renters.
Sociological Science 2: 329-350.
Durose, M.R., Snyder, H.N., and Cooper, A.D. (2015).
Multistate Criminal History Patterns of Prisoners
Released in 30 States. Washington, DC: U.S.
Department of Justice, Bureau of Justice Statistics.
Glaze, L.E., and Kaeble, D. (2014). Correctional
Populations in the United States, 2013. Washington,
DC: U.S. Department of Justice, Bureau of Justice
Statistics.
Goerge, R.M., Smithgall, C., Seshadri, R., and
Ballard, P. (2010). Illinois Families and eir Use of
Multiple Service Systems. Chicago, IL: Chapin Hall
at the University of Chicago.
Grinberg, C., Hawthorne, M., LaNoue, M., Brenner,
J., and Mautner, D. (2016). The core of care
management: e role of authentic relationships
in caring for patients with frequent hospitalizations.
Population Health Management 19(4): 248-256.
Available at https://www.ncbi.nlm.nih.gov/
pubmed/26565379.
Green, S.R., Singh, V., and O’Byrne, W. (2009).
Hope for New Jersey’s City Hospitals: e Camden
Initiative. Trenton, NJ: New Jersey Department
of Banking and Insurance, Oce for Electronic
Health Information Technology. Available at http://
www.state.nj.us/dobi/division_insurance/hint/
camdenwhitepaper.pdf.
Greenberg, G.A., and Rosenheck, R.A. (2008). Jail
incarceration, homelessness, and mental health: A
national study. Psychiatric Services 59(2): 170-177.
Kushel, M.B., Hahn, J.A., Evans, J.L., Bangsberg,
D.R., and Moss, A.R. (2005). Revolving doors:
Imprisonment among the homeless and marginally
housed population. American Journal of Public
Health 95(10): 1747-1752.
Lamberti, J.S., Weisman, R., and Faden, D.I.
(2004). Forensic assertive community treatment:
Preventing incarceration of adults with severe
mental illness. Psychiatric Services 55(11): 1285-1293.
Larimer, M., Malone, D., Garner, M., Atkins, D.C.,
Burlingham, B., Lonczak, H.S., Tanzer, K., Ginzler,
J., Clifase, S.L., Hobson, W.G., and Marlatt, G.A.
(2009). Health care and public service use and costs
before and after provision of housing for chronically
homeless persons with severe alcohol problems.
JAMA 301: 1349-1357.
Mackelprang, J.L., Collins, S.E., and Clifase, S.L.
(2014). Housing first is associated with reduced
use of emergency medical services. Prehospital
Emergency Care 18(4): 476-482. Available at http://
ww w.tandfonline.com/doi/abs/10.3109/10903127.
2014.916020.
Martell, D.A., Rosner, R., and Harmon, R.B. (1995).
Base rate estimates of criminal behavior by
homeless mentally ill persons in New York City.
Psychiatric Services 46(6): 596-601.
McNiel, D.E., and Binder, R.L. (2005). Psychiatric
emergency service use associated with
homelessness, mental disorder, and violence.
Psychiatric Services 56(6): 699-704.
Metraux, S., and Culhane, D.P. (2006). Recent
incarceration history among a sheltered
homeless population. Crime and Delinquency
52(3): 504-517. Available at http://dx.doi.
org/10.1177/0011128705283565.
Sampson, R.J., and Laub, J.H. (1990). Crime and
deviance over the life course: e salience of adult
social bonds. American Sociological Review 55(10):
609-627.
Tsai, J., and Rosenheck, R.A. (2012). Incarceration
among chronically homeless adults: Clinical
correlates and outcomes. Journal of Forensic
Psychology Practice 12(4): 307-324. Available at
http://dx.doi.org/10.1080/15228932.2012.695653.
U.S. Census Bureau. (2010–2014). American
Community Survey 5-Year Estimates: Camden
City, NJ. Available at https://thedataweb.
rm.census.gov/TheDataWeb_HotReport2/
profile/2014/5yr/np01.hrml?SUMLEV=160&stat
e=34&place=10000.
Integrated Health Care and Criminal Justice Data | 21
is paper was prepared with support from the National Institute of Justice, Oce of Justice
Programs, U.S. Department of Justice, under contract number 2012-R2-CX-0048. e opinions,
ndings, and conclusions or recommendations expressed in this publication are those of the
authors and do not necessarily represent those of the Department of Justice.
Wang, E.A., Hong, C.S., Samuels, L., Shavit, S.,
Sanders, R., and Kushel, M. (2010). Transitions
clinic: Creating a community-based model
of health care for recently released California
prisoners. Public Health Reports 125(2): 171-177.
Weisman, R.L., Lamberti, J.S., and Price, N. (2004).
Integrating criminal justice, community healthcare,
and support services for adults with severe mental
disorders. Psychiatric Quarterly 75(1): 71-85.
Western, B. (2007). Punishment and Inequality in
America. New York, NY: Russell Sage Foundation.
Author Note
Anne Milgram, Esq., is a Professor of Practice and
Distinguished Scholar in Residence at New York
University Law School.
Jerey Brenner, M.D., is the founder of the Camden
Coalition of Healthcare Providers and served as
Executive Director until 2017. He now serves as
Senior Vice President of Integrated Health and
Human Services at UnitedHealthcare and leads
the myConnections business unit, which pilots
and scales new models of care that bring traditional
health care services, along with social services
such as housing and transportation, to millions of
Medicaid patients nationwide.
Dawn Wiest, Ph.D., is the Director of Action
Research & Evaluation at the Camden Coalition of
Healthcare Providers.
Virginia Bersch is the Criminal Justice Deputy
Director of National Implementation at the Laura
and John Arnold Foundation.
Aaron Truchil, M.S., is the Director of Strategy and
Analytics at the Camden Coalition of Healthcare
Providers.
NCJ 250934
Learn more about the Executive Session at:
www.hks.harvard.edu, keywords “Executive Session Community Corrections”
Members of the Executive Session on Community Corrections
Marc Levin, Policy Director, Right on Crime;
Director, Center for Effective Justice, Texas
Public Policy Foundation
Glenn E. Martin, Former President and
Founder, JustLeadershipUSA
Anne Milgram, Senior Fellow, New York
University School of Law
Jason Myers, Sheriff, Marion County
Sheriff’s Office
Michael Nail, Commissioner, Georgia
Department of Community Supervision
James Pugel, Chief Deputy Sheriff,
Washington King County Sheriff’s
Department
Steven Raphael, Professor, Goldman
School of Public Policy, University of
California, Berkeley
Nancy Rodriguez, Professor, University of
California, Irvine; Former Director, National
Institute of Justice
Vincent N. Schiraldi, Senior Research
Scientist, Columbia University School
of Social Work; Co-Director, Columbia
University Justice Lab
SandraSusan Smith, Professor,
Department of Sociology, University of
California, Berkeley
Amy Solomon, Vice President of Criminal
Justice Policy, Laura and John Arnold
Foundation
Wendy S. Still, Chief Probation Officer,
Alameda County, California
John Tilley, Secretary, Kentucky Justice
and Public Safety Cabinet
Steven W.Tompkins, Sheriff,
Massachusetts Suffolk County Sheriff’s
Department
Harold Dean Trulear, Director, Healing
Communities; Associate Professor of
Applied Theology, Howard University School
of Divinity
Vesla Weaver, Associate Professor,
Department of Political Science, Johns
Hopkins University
Bruce Western, Daniel and Florence
Guggenheim Professor of Criminal Justice,
Harvard University; Professor of Sociology,
Columbia University; Co-Director, Columbia
University Justice Lab
John Wetzel, Secretary of Corrections,
Pennsylvania Department of Corrections
Ana Yáñez-Correa, Program Officer for
Criminal Justice, Public Welfare Foundation
Molly Baldwin, Founder and CEO, Roca, Inc.
Kendra Bradner (Facilitator), Senior Staff
Associate, Columbia University Justice Lab
Barbara Broderick, Chief Probation
Officer, Maricopa County Adult Probation
Department
Douglas Burris, Chief Probation Officer
(Retired), United States District Court, The
Eastern District of Missouri, Probation
John Chisholm, District Attorney,
Milwaukee County District Attorney’s Office
George Gascón, District Attorney, San
Francisco District Attorney’s Office
Adam Gelb, Director, Public Safety
Performance Project, The Pew Charitable
Trusts
Susan Herman, Deputy Commissioner for
Collaborative Policing, New York City Police
Department
Michael Jacobson, Director, Institute for
State and Local Governance; Professor,
Sociology Department, Graduate Center,
City University of New York
Sharon Keller, Presiding Judge, Texas
Court of Criminal Appeals