This report was prepared by the U.S. Census Bureau, Center for Economic Studies (CES), using
federal funding provided by the Bureau of Justice Statistics.
Document Title:

Mortality in a Multi-State Cohort of Former State Prisoners, 2010-2015

Authors:

Leticia Fernandez, U.S. Census Bureau
Sharon Ennis, U.S. Department of Veterans Affairs
Sonya R. Porter, U.S. Census Bureau
Elizabeth Carson, Bureau of Justice Statistics

Document No.:

304203

Publication Date:

February 2022

Award No.:

This project was supported by award number DJO-BJS-21-RO-0005.

Abstract:
This report was produced by the U.S. Census Bureau, Center for Economic Studies (CES), under award
number DJO-BJS-21-RO-0005. It explores the role that race/Hispanic origin, other demographic
characteristics, and custodial/criminal history factors have on post-release mortality, including on the
timing of deaths. It also assesses whether conditional release to community supervision or
reimprisonment may explain the higher post-release mortality found among non-Hispanic whites. In the
second part of the analysis, the report estimates standardized mortality ratios by sex, age group, and
race/Hispanic origin using the U.S. general population as a reference. The data come from state prison
releases from the Bureau of Justice Statistics’ National Corrections Reporting Program (NCRP). The NCRP
records were linked to the Census Numident to identify deaths occurring within five years from prison
release. NCRP records were linked to previous decennial censuses and survey responses to obtain selfreported race and Hispanic origin if available.
Disclaimer
The Bureau of Justice Statistics funded this third-party report. It is not a BJS report and does not release
official government statistics. The report is released to help inform interested parties of the research or
analysis contained within and to encourage discussion. BJS has performed a limited review of the report
to ensure the general accuracy of information and adherence to confidentiality and disclosure
standards. Any statistics included in this report are not official BJS statistics unless they have been
previously published in a BJS report. Any analysis, conclusions, or opinions expressed herein are those of
the authors and do not necessarily represent the views, opinions, or policies of the Bureau of Justice
Statistics or the U.S. Department of Justice.

This page intentionally left blank.

Mortality in a Multi-State Cohort of Former State Prisoners, 2010-2015
by
Leticia Fernandez
U.S. Census Bureau
Sharon Ennis
US Department of Veterans Affairs
Sonya R. Porter
US Census Bureau
Elizabeth Carson
Bureau of Justice Statistics

CES 22-06

February 2022

The research program of the Center for Economic Studies (CES) produces a wide range of
economic analyses to improve the statistical programs of the U.S. Census Bureau. Many of these
analyses take the form of CES research papers. The papers have not undergone the review
accorded Census Bureau publications and no endorsement should be inferred. Any opinions and
conclusions expressed herein are those of the author(s) and do not necessarily represent the
views of the U.S. Census Bureau. All results have been reviewed to ensure that no confidential
information is disclosed. Republication in whole or part must be cleared with the authors.
To obtain information about the series, see www.census.gov/ces or contact Christopher Goetz,
Editor, Discussion Papers, U.S. Census Bureau, Center for Economic Studies, 4600 Silver Hill Road,
Washington, DC 20233, CES.Working.Papers@census.gov. To subscribe to the series, please click
here.

Abstract
Previous studies report that individuals who have been imprisoned have higher mortality rates
than their demographic counterparts in the general population, particularly non-Hispanic white
former prisoners. Most of these studies have been based on a single state’s prison system, and
the extent to which their findings can be generalized has not been established. In this study we
explore the role that race/Hispanic origin, other demographic characteristics, and custodial/
criminal history factors have on post-release mortality, including on the timing of deaths. We also
assess whether conditional release to community supervision or reimprisonment may explain the
higher post-release mortality found among non-Hispanic whites. In the second part of the
analysis, we estimate standardized mortality ratios (SMRs) by sex, age group, and race/Hispanic
origin using as reference the U.S. general population. The data come from state prison releases
from the Bureau of Justice Statistics’ (BJS) National Corrections Reporting Program (NCRP). The
NCRP records were linked to the Census Numident to identify deaths occurring within five years
from prison release. We also linked NCRP records to previous decennial censuses and survey
responses to obtain self-reported race and Hispanic origin if available. We found that nonHispanic white former prisoners were more likely to die within five years after prison release and
more likely to die in the initial weeks after release compared to racial minorities and Hispanics.
Reimprisonment, age at release, and a history of multiple prison terms had a similar influence on
the odds of dying across all race/Hispanic origin groups. Other factors, such as the type of release
and the duration of the last term in prison, were associated with higher risks of mortality for
some groups but not for others.

*

The views expressed are those of the authors and not necessarily those of the U.S. Census Bureau, the Bureau
of Justice Statistics, the U.S. Department of Justice, or the Department of Veterans Affairs. The statistical
summaries reported in this paper have been cleared by the Census Bureau's Disclosure Review Board under
release authorization numbers CBDRB-FY21-123, CBDRB-FY21-CES010-011, CBDRB-FY22-CES014-004, and CBDRBFY22-CES010-003. We thank Mariel Alper, Renuka Bhaskar, Brad Foster, Cheryl Ann Grim, and Danielle H. Sandler
for helpful comments and discussions.
1

Mortality in a Multi-State Cohort of Former State Prisoners, 2010-2015
Leticia Fernandez, US Census Bureau
Sharon Ennis, US Department of Veterans Affairs
Sonya R. Porter, US Census Bureau
Elizabeth Carson, Bureau of Justice Statistics

Introduction
A growing number of studies report that the experience of imprisonment may have long-lasting
and significant consequences on a person’s life expectancy, and that race and Hispanic origin
seem to mediate the imprisonment-mortality relationship (Binswanger et al. 2013; Graham et
al. 2015; Patterson 2013; Pridemore 2014; Rosen et al. 2011; Spaulding et al. 2011; Wildeman
et al. 2016a; Wildeman et al. 2016b).
Overall, state prisoners have lower mortality rates than the U.S. resident population2 for all
causes except cancer, suicide, and homicide, when the U.S. resident population is adjusted for
age, sex, and race/ethnicity (Carson 2021a). Non-Hispanic black prisoners, and to a lesser
extent non-Hispanic white prisoners, tend to have lower all-cause mortality rates than
individuals in the same demographic groups in the general population (Wildeman et al. 2016a,
2016b). After release from prison, however, former prisoners have higher mortality rates than
their demographic counterparts in the general population, and the magnitude of these
differences vary by race/Hispanic origin, sex, and age. Former prisoners who are non-black,
female, or young tend to have the highest standardized mortality ratios (SMRs), suggesting that
imprisonment heightens their risk of death after release more so than for former prisoners who
are black, male, or released from prison at older ages (Binswanger et al. 2007; Rosen et al.
2008; Spaulding et al. 2011; Zlodre and Fazel 2012).
Additional insights about the relationship between imprisonment and mortality have been
gained from studies that assess relative mortality within cohorts of former prisoners
(Binswanger et al. 2013; Patterson 2013; Testa et al. 2018). While these studies provide
valuable information about custodial and criminal history factors that influence mortality, most

2

Except where noted, in this report, former prisoner mortality is compared to the general population, which
consists of all U.S. residents, including those currently or formerly in prison. This report uses the terms “general
population” and “resident population” interchangeably.

2

have focused on a single state’s prison system, and the extent to which their findings can be
generalized to other states or nationwide has not been established.
In this study, we examine post-release mortality in a large and racially diverse sample obtained
from the 2010 cohort of released state prisoners as reported by states to the Bureau of Justice
Statistics’s (BJS) National Corrections Reporting Program (NCRP). We explore the role that
demographic characteristics, selected custodial/criminal history factors, and geographic region
of imprisonment have on post-release within-cohort mortality, with particular emphasis on
how timing and rates of mortality vary by race and Hispanic origin. In addition, we estimate
separately SMRs for Hispanics (of any race), non-Hispanic whites, non-Hispanic blacks, and a
group that combines individuals of all other non-Hispanic races using two reference
populations, the U.S. general population and the population from the 35 states that provided
the release data in 2010 included in this study. To our knowledge, this is one of a few multistate studies of mortality among former state prisoners. Findings from this study contribute to
the growing body of research on the interrelationships among race/Hispanic origin,
imprisonment, and mortality outcomes. In addition, this study serves to explore the extent to
which previous findings reported by state-level studies are generalizable to the U.S. population
of former prisoners.
Background
Studies of populations during imprisonment report that their mortality rates are consistently
lower than groups with similar demographic characteristics in the resident population (Noonan
and Ginder 2015). In particular, non-Hispanic black males in prison tend to have much lower allcause mortality rates relative to their non-imprisoned counterparts. Although the mortality
advantage is not as large for other groups, non-Hispanic black female prisoners, Hispanic male
prisoners, and non-Hispanic white male and female prisoners also exhibit favorable mortality
rates relative to their counterparts in the general population. Hispanic females in prison are an
exception in that they may experience similar or even higher mortality rates than their
counterparts in the general population (Wildeman et al. 2016a, 2016b). However, further
studies are needed to confirm these findings.3
The lower mortality rates during imprisonment compared to the general population may result
from the increased supervision in prisons and better access to health care while in prison,
especially for individuals from the most disadvantaged backgrounds (Rosen et al. 2011;
Spaulding et al. 2011). Some researchers have suggested that compassionate release policies in
which moribund prisoners are released prior to death may also explain inmates’ lower mortality
3

In one of these studies, the SMR relative to Hispanic females in the general population was not statistically
significant (SMR=1.18, 0.93-1.43) (Wildeman et al. 2016b); and in another study, the SMRs was statistically
significant only for Hispanic females in New York prisons (SMR=3.32, 1.82-4.88) but not in California prisons
(SMR=1.17, 0.72-1.64) (Wildeman et al. 2016a).

3

rates. However, several reports find that very few prisoners are released each year due to
extreme illness or impending death (HRW & FAMM 2012; Spaulding et al. 2011; USGAO 2012).
The causes of death that are lower among prisoners compared to the resident population vary
by race and Hispanic origin and include accidents, drug/alcohol intoxication, respiratory
disease, and some chronic conditions (Carson 2021a; Noonan and Ginder 2015; Rosen et al.
2011).
In contrast, formerly imprisoned persons tend to have higher mortality rates than their
demographic counterparts in the general population (see Zlodre and Fazel 2012 for a review).
Studies consistently find that female, non-black, and younger former prisoners have higher
SMRs relative to male, black, and those released from prison at older ages when compared to
their counterparts in the general population (Binswanger et al. 2007; Rosen et al. 2008;
Spaulding et al. 2011).
In addition to demographic factors, length of imprisonment, the number of times incarcerated,
and the type of release (conditional or unconditional) have been linked to higher odds of postrelease mortality. For example, several studies report that individuals with a history of multiple
short terms in prison face higher risk of death after they are released from prison than those
with longer but fewer reimprisonments (Binswanger et al. 2013; Graham et al. 2015; Jones et
al. 2017; Testa et al. 2018). In contrast, the influence of community supervision on post-release
mortality has yielded somewhat inconsistent findings. In a cohort of former prisoners from
Washington state prisons, individuals released without community supervision had higher
mortality than those released to community supervision (Binswanger et al. 2013); however,
another study using Pennsylvania state prison data found that being released without
community supervision was associated with lower mortality among non-Hispanic blacks and did
not seem to influence mortality risks for non-Hispanic whites or Hispanics (Testa et al. 2018).
In terms of causes of death, previous studies report that among former prisoners the top three
causes include drug or alcohol overdose, cardiovascular diseases, and cancer. These are
followed in frequency by homicide, suicide, and motor vehicle accidents. Less frequent but also
important causes are non-infectious diseases (liver disease and respiratory illnesses) and
infectious diseases (viral hepatitis, HIV, and tuberculosis) (Binswanger et al. 2007, 2013; Jones
et al. 2017; Kuzyk et al. 2018; Noonan and Ginder 2015; Rosen et al. 2011; Testa et al. 2018).
All studies agree, however, that the risk of death is highest for former prisoners in the first few
weeks after release. In a cohort of prisoners released from Washington state prisons, the risk of
death in the first two weeks was 12.7 times higher than for individuals of the same race, sex,
and age in the same state’s non-imprisoned population (Binswanger et al. 2007). Similarly, a
study of New York parolees reported that the odds of dying declined by approximately two
percent for each month an individual survived parole after release from prison; surviving the
first year on parole was associated with a 24 percent decline in the odds of dying (Patterson
4

2013). In another study, 20.5 percent of all deaths occurring in the first year after release in a
cohort of Scottish former prisoners took place in the first two weeks (Graham et al. 2015).
Researchers suggest that higher mortality in the initial weeks after prison release is associated
with the multiple challenges that individuals must negotiate upon re-entering society, including
finding housing, a stable job, and reestablishing or developing personal relationships (Rosen et
al. 2011). Higher levels of stress during the transition from prison to reentry in a community
may jeopardize the physical and emotional health of recently released former prisoners,
especially if they also have a history of substance abuse, mental health issues, or both (Chang et
al. 2015; Jones et al. 2017). Studies of post-release mortality from Connecticut state prisons
(Kuzyk et al. 2018), Washington state prisons (Binswanger et al. 2007, 2013), New Mexico state
prisons (Krinsky et al. 2009), and North Carolina state prisons (Rosen et al. 2008) consistently
report that among those who die in the first couple of weeks after prison release, the leading
causes of death are opioid overdose, alcohol-related causes, suicide, and homicide. Moreover,
a meta-analysis based on six drug-related mortality studies among former prisoners in three
countries (United States, United Kingdom, and Australia) concluded that individuals were about
4 to 8 times more likely to die of drug-related causes in the first two weeks than between
weeks three to twelve after their release (Merrall et al. 2010).
Age-sex-race-specific comparisons of mortality rates between former prisoners and the general
population consistently find that the former have higher mortality rates. One drawback of these
SMRs is that comparisons across racial/ethnic groups is difficult given their differences in
socioeconomic status and in the prevalence of former prisoners in the general population.
At the end of 2020, for example, there were 1,234 non-Hispanic black prisoners sentenced to
more than one year in state and federal prisons per 100,000 U.S. adult non-Hispanic black
residents, compared to 639 among Hispanics, and 223 among non-Hispanic whites (Carson
2021b). That is, compared to non-Hispanic whites, non-Hispanic blacks were about 5.5 times
more likely to be imprisoned for more than one year and Hispanics were about thrice as likely.
This means that a higher percent of the currently non-imprisoned non-Hispanic black and
Hispanic populations that are used as a reference in the computation of SMRs have been
imprisoned in the past, obscuring the effects of imprisonment on mortality (Spaulding et al.
2011).
Moreover, former prisoners are likely to differ from their demographic counterparts in the
general population because they come disproportionately from economically distressed
neighborhoods that may have higher mortality rates than those in the general population
(Looney and Turner 2017; Rosen et al. 2008). Therefore, SMRs that use as reference the general
population may overestimate the influence of imprisonment on mortality. Impoverished
neighborhoods can influence mortality through several pathways, including exposure to
communicable diseases, low quality housing, limited health services, and high levels of social
fragmentation associated with stress and anxiety (Braveman et al. 2011). Two recent studies on
5

members of the general population have confirmed that living in distressed neighborhoods is
associated with an increased risk of death. One of these studies found higher mortality rates
among adults with unhealthy lifestyles if they were living in the most disadvantaged
neighborhoods (Warren et al. 2018). The other study found higher all-cause mortality among
veterans living in poor neighborhoods compared to more well-off neighborhoods, even after
adjusting for individuals’ demographic and socioeconomic characteristics (Nelson et al. 2016).
Therefore, comparing the mortality of former prisoners to a population that is more similar in
terms of socioeconomic status could potentially attenuate the imprisonment-mortality
relationship (Dirkzwager et al. 2012).
There are also geographic variations in mortality rates that make cross-state differences in
SMRs difficult to explain (Wildeman et al. 2016a). For example, former prisoners in Washington
state have mortality rates that are 3.61 (95% CI: 3.48-3.73) times greater than those in the state
non-imprisoned population (Binswanger et al. 2013). In comparison, the mortality of former
prisoners from North Carolina prisons is 2.10 (95% CI: 1.95-2.24) times greater than for their
state general population (Jones et al. 2017). In Georgia, former state prisoners have a mortality
rate that is 1.54 (95% CI: 1.48-1.61) times greater than in their state general population
(Spaulding et al. 2011). These differences do not necessarily mean that the mortality rate of
former prisoners in Georgia is lower than that of former prisoners in Washington state. It could
also be that mortality rates in the general Georgia population are higher than in Washington
state, or that there are differences in the states’ and/or the prisoners’ socioeconomic or
racial/Hispanic origin distributions that are associated with these disparities in SMRs (Kochanek
et al. 2017).
An additional perspective comes from studies of relative mortality within cohorts of former
offenders, but the findings have been somewhat inconsistent. For example, Hispanics and nonHispanic blacks were more likely to die in a New York cohort of persons on parole compared to
non-Hispanic whites (Patterson 2013). In contrast, and consistent with the studies mentioned
above that use as reference the mortality rates in the general population, two studies of
cohorts of former prisoners have found that non-Hispanic whites are at higher risk of death
after release from prison than other racial/Hispanic origin groups in the same cohort. A study
that followed released prisoners from Washington state found that Hispanics, non-Hispanic
blacks, and non-Hispanic Asians had lower risks of all-cause, overdose, and opioid-related
deaths compared with non-Hispanic whites (Binswanger et al. 2013). Similarly, another study of
a cohort of former prisoners released from a Pennsylvania state prison found that non-Hispanic
whites had a higher risk of death than did Hispanics and non-Hispanic blacks (Testa et al. 2018).
Testa and colleagues (2018) speculated that the higher mortality of non-Hispanic white former
prisoners may be due to their longer exposure to risks outside of prison after release, because
non-Hispanic blacks and Hispanics are more likely to be reimprisoned. This is supported by
findings from a recent study of prisoner recidivism in which 40.2 percent of non-Hispanic
whites, 46.0 percent of non-Hispanic blacks, and 47.3 percent of Hispanics released in 2005
6

were reimprisoned in the first year after release (Alper et al. 2018). In addition, given that the
rate of imprisonment relative to population size is generally lower for non-Hispanic whites, it
may be that non-Hispanic white former prisoners are a more selective group in terms of poor
physical and/or mental health conditions compared to Hispanic and non-Hispanic black former
prisoners (Schnittker et al. 2011; Testa et al. 2018; Zane et al. 2018).
Findings from studies that have followed cohorts of those involved and not involved in criminal
justice activities for decades and into their old age provide support to the argument that the
higher mortality of chronic offenders, compared to non-offenders and to adolescence-only
offenders, seems to be associated with detrimental factors that influence both criminal
behavior and early mortality; namely, low self-control, substance abuse, poor self-care, and
unhealthy lifestyles (Laub and Vaillant 2000; van de Weijer et al. 2016; Zane et al. 2018).
Finally, the state prison population has aged over the past 40 years, through both increased
admissions of older offenders and longer lengths of stay in prison (Carson and Sabol 2016). The
median age in state prison rose from 30 years in 1993 to 36 years in 2013, and releases of
persons age 55 or older increased by 384 percent. Non-Hispanic whites comprised 50 percent
of state prisoners age 55 or more at yearend 2013, an increase of 327 percent from 1993. In
contrast, non-Hispanic blacks made up 35 percent of state prisoners in this age group in 2013
and Hispanics 15 percent, increases of 431 percent and 456 percent, respectively.
In the present study we use a large multi-state sample of released prisoners to explore the
imprisonment-mortality relationship using the two approaches discussed above, SMRs and
intra-cohort relative mortality comparing four groups: Hispanics (any race), non-Hispanic
whites, non-Hispanic blacks, and non-Hispanic other races. To our knowledge, this is one of the
first studies that uses multi-state level data to compute SMRs for Hispanic former prisoners
separate from their non-Hispanics counterparts. In addition, we compare intra-cohort mortality
of former prisoners separately by the type of release (conditional or unconditional) and by
whether individuals were reimprisoned or not after their 2010 release to assess whether
community supervision or the higher reimprisonment rates of racial minorities and Hispanics
seem to explain the higher post-release mortality found among non-Hispanic whites.
In the next section, we describe the data and methods, the sample characteristics, and the
results from record linkage of state prisoner data to death records from the Census Numident
file. In subsequent sections, we report results from logistic regressions examining the role of
demographic and imprisonment factors on the mortality and the timing of deaths occurring up
to five years after release. We also discuss SMRs using two population references, the U.S.
resident population and the population in the 35 states that provided the prison release data in
2010 used in this study.

7

Data and Methods
The data come from state prison releases from the National Corrections Reporting Program
(NCRP) which is administered by the Bureau of Justice Statistics (BJS). State departments of
corrections (DOCs) submit one record per prison admission and release during the calendar
year, as well as a record for each prisoner in their physical custody on December 31. The NCRP
data contain prisoners’ demographic and criminal justice information (dates and types of
admission and release, sentencing and offense details) as recorded in the DOCs’ offender
management information systems.
Using personal identifiers (name and date of birth) in the NCRP records, each individual was
assigned a unique Protected Identification Key (PIK) internal to the Census Bureau, when
available. The PIK was assigned to individuals through probabilistic matching to Census Bureau
reference files (Wagner and Layne 2014). Individuals with little or no personal information on
their records and those who were not found in Census Bureau reference files, including nonU.S. citizens, could not be assigned a PIK, and therefore were excluded from the analysis.
While BJS has collected NCRP data since 1983, not all states submit all types of data each year.
For example, in 2010, 45 states submitted prison release records to BJS. Sixteen states were
removed from our sample because either they did not provide NCRP release data for 2010,
expressed privacy concerns, or because of low PIK rates due to incomplete personally
identifiable information. In all, the sample of former prisoners released in 2010 that we use in
this study combines information from 35 states. In this study, the PIK rates for the states
included in the analysis ranged from about 59 percent to 99 percent.
In the case of persons with two or more prison release records for 2010, we chose as their
release date for this study the record with the latest date in the year. We then linked NCRP data
to the 2019 Census Numident file to identify deaths taking place between 2010 and 2015,
within 5 years of a person’s release. The Census Numident file is created from the Social
Security Administration’s (SSA) Numerical Identification (Numident) file. It contains information
for each person ever issued a Social Security Number and includes name, date of birth, and
date of death if available. The Census Numident file is updated with death information from the
SSA’s Death Master File.
We also linked NCRP records to previous decennial censuses and survey responses to obtain
information about race and Hispanic origin if available (Census 2000, 2010 Census, and 20012014 single-year American Community Survey (ACS) data).
Our analysis consists of modeling the odds that an individual will die within five years after
release from prison. Persons who were reimprisoned after their 2010 release and died in prison
were removed from the sample. The dichotomous dependent variable is coded as dead or alive
8

at the end of the five-year follow up period. Table 1 shows the explanatory variables in the
analysis and how they were constructed. These variables were selected because previous
research has shown them to be associated with post-prison mortality.
The standardized mortality ratios (SMRs) compare mortality of released prisoners to both
national mortality rates and mortality rates in the subset of states that provided data for 2010
releases. We obtained deaths and population counts to compute mortality rates at the national
level and for combined states by calendar year, age group, sex, race, and Hispanic origin from
the Wide-ranging Online Data for Epidemiologic Research (WONDER) system of the Centers for
Disease Control and Prevention (CDC). Mortality rates in the sample were computed by dividing
the number of deaths by the person-years at risk. Person-years at risk were computed from the
date of release to either the date of death or the end of the five-year follow up period,
whichever came first, minus any time spent in reimprisonment. The SMRs were computed by
dividing the observed number of deaths in the sample by the expected number of deaths given
the mortality rate for the same calendar year, age group, sex, race, and Hispanic origin in the
resident population.
In order to meet the Census Bureau’s Disclosure Avoidance requirements, all population counts
are reported as rounded numbers. In Appendix Table A we compare the characteristics of
individuals in the full 2010 released cohort (Column A) with those of individuals who received a
PIK (Column B), and with those in the unduplicated PIK analytic sample (Column D). There were
records for about 605,000 individuals in the national-level cohort released from prison in 2010.
PIKs were assigned to about 508,000 (84 percent). Among individuals assigned a PIK, about 6
percent turned out to be duplicate records for the same individual, some involved different
release dates and some had the same release date. In cases of individuals with more than one
release record in 2010, only the record with the latest release date in the year was retained.
The unduplicated sample in this study consists of about 478,000 individuals, which represents
79.0 percent of the full cohort of state prisoners released nationally in 2010.
For the most part, the released cohort and the two PIKed samples have similar characteristics,
but some groups were less likely to receive a PIK. In particular, Hispanics released from prison
were 18.5 percent of the 2010 cohort (Column A in Appendix Table A), but they were less likely
to be assigned a PIK and therefore they contributed a smaller percent (15.9 percent) to the
unduplicated sample. This is not surprising, given that the state prison population holds a
number of non-U.S. citizens who would not have a record in the Numident file, and the majority
of these prisoners are identified by the state departments of corrections as Hispanic. In
contrast, 40.5 percent of the individuals in the released cohort were non-Hispanic white, and
their contribution as a percent in the unduplicated sample increased to 42.7 percent. In the
Appendix we provide additional discussion of similarities and differences between the 2010
cohort of released prisoners and the analytic sample used in this study.
9

A limitation of this and similar studies that link individual-level data from various sources is the
issue of selectivity. Individuals with complete identifying information that are assigned a unique
identifier for linkage across datasets tend to differ in various relevant respects from those with
missing or incomplete information. Specifically, low income, racial minorities, and Hispanics
tend to have lower PIK rates (Bond et al. 2014). In addition, it may not be possible to assign a
unique identifier to individuals with multiple names or aliases (Kinner and Binswanger 2014). A
second issue, specific to studies of mortality differentials, is that some deaths may not be
reported in the case of individuals who die outside the country, or they may not be matched
due to multiple names or aliases.

Findings
Racial/Ethnic Differences in Mortality
As shown in Table 2, of the 478,000 individuals in the unduplicated sample of released
prisoners, about 18,000 (3.8 percent) died within five years of their release from prison.
Column A in Table 2 shows the number of former prisoners in the unduplicated sample by
demographic and imprisonment characteristics. Column B shows the percent of deaths within
five years from release for each category.
Non-Hispanic white former prisoners were more likely to die than other racial/ethnic groups.
Those released at older ages or who served longer terms were more likely to die in the 5-year
period than their younger counterparts and those who served short prison terms. In addition, a
history of multiple state prison terms was associated with higher mortality among former
prisoners.
The rest of the columns in Table 2 show the percent distribution of the deaths that occurred in
the five years after release, by month for the first 3 months, by the monthly average for the last
three quarters, and by year. Note that among those who died within five years after release,
the highest monthly percentage of deaths occurred in the first month (3.1 percent) compared
to later months. Relative to their counterparts, non-Hispanic whites, those age 65 or older at
release, those who served long terms prior to their 2010 release, and individuals with unknown
release type had higher mortality within the first month as well as the first year after release.
Table 3 shows the results of modeling the likelihood of dying within five years after release
from prison. As a reminder, only deaths outside of prison were considered in the analysis. We
included interactions between race and age at release (Model B), race and reimprisonment
(Model C), and race and type of prison release (Model D) to assess whether these factors
influenced differently the odds of dying for non-Hispanic whites and racial/ethnic minorities
after release. Consistent with earlier studies, we find that net of other factors, particularly age
10

at release, non-Hispanic whites were more likely to die in the five-year post release period than
non-Hispanic blacks, non-Hispanic other races, and Hispanics. As mentioned earlier, racial
minorities and Hispanics in the criminal justice system may spend more time in prison and
subsequent reimprisonment than non-Hispanic whites (Testa et al. 2018; Wildeman et al.
2016a, 2016b). Since healthcare is constitutionally guaranteed to prisoners, prisoners may
receive treatment for chronic or infectious conditions that might go untreated outside prison.
The interaction terms between race/Hispanic origin and age at release in model B show that
the mortality advantage among non-Hispanic blacks and Hispanics relative to non-Hispanic
whites holds for all or most age groups, except for Hispanics ages 45 to 54. In contrast, the odds
of dying did not differ significantly between non-Hispanics of other races and non-Hispanic
whites in the same age group at release.
The coefficients in Model C, Table 3, show that former prisoners who were reimprisoned at any
time within five years of release had lower odds of dying than those who were not
reimprisoned. Part of this relationship may be mechanical since individuals must be alive in
order to be reimprisoned. However, some studies suggest that access to healthcare while
incarcerated may improve health status and reduce post-release mortality (Rosen et al. 2008).
In addition, prison provides some protection from deaths due to drug/alcohol overdose and
homicide. Figure 2 shows that for each race/Hispanic origin group, former prisoners who were
not reimprisoned after release had a higher probability of dying within five years than those
who were reimprisoned at any point during the study window.
The coefficients in Model D, Table 3, show that in general former prisoners released
unconditionally in 2010 were more likely to die than those released to community supervision
(conditionally). However, the increased risks of death associated with unconditional release
were lower for non-Hispanic blacks than for other racial/Hispanic origin groups.
Net of other factors, Table 3 also shows that individuals with less than a high school education,
those whose last state prison term was under a year, and those with a history of multiple
imprisonments were more likely to die within five years from release than were individuals with
more education, those who served longer terms, and those for whom their last release was
their first term in prison.
Probabilities of death within five years after release in 2010 were computed for each
race/Hispanic origin by age group and reimprisonment status using models B, C, and D in Table
3. Figure 1 shows how the probability of dying increases with each age group (age at release),
and that by age 25 and older, non-Hispanic blacks and Hispanics had significantly lower
probabilities of death than non-Hispanic whites. Figure 2 shows that even when comparing only
non-reimprisoned former prisoners, non-Hispanic whites had a significantly higher probability
of dying within five years of release than non-Hispanic blacks, non-Hispanic others, and
11

Hispanics. In contrast, the probability of dying was similar across most racial/Hispanic origin
groups that were reimprisoned after their 2010 release (within five years), and only nonHispanic blacks had a lower probability of dying than non-Hispanic whites. That is, the relative
survival advantages of racial and ethnic minorities outside of prison (or the disadvantage of
non-Hispanic whites) seemed to even out for individuals who were reimprisoned, maybe
because of less time spent outside of prison or because the more controlled environment in
prison reduced risks across all groups. Figure 3 shows that unconditional release is associated
with higher odds of dying for non-Hispanic whites and non-Hispanic other races, but type of
release does not seem to be a factor in the likelihood of death for non-Hispanic blacks and
Hispanics. Further research is needed to understand the mechanisms that may explain nonHispanic blacks’ significantly lower probabilities of dying than their counterparts in the same
age groups, by reimprisonment status and by type of release.
To further explore whether reimprisonment or type of release are implicated in the higher
mortality of non-Hispanic whites relative to other racial and ethnic groups in the study, we
conducted the same analysis as shown in Table 3 separately for individuals who were not
reimprisoned at any time in the five-year follow up period, and by their type of release in 2010
(not shown but available upon request). Considering only individuals who were not
reimprisoned, non-Hispanic whites remained significantly more likely to die than other
racial/Hispanic origin groups in the study. Similarly, regardless of type of release, non-Hispanic
whites were significantly more likely to die in the 5 years of the study than the other groups.
The regression coefficients in the additional analysis were similar in both magnitude and
significance to those obtained using the full sample.
In Table 4 we explored separately by race and Hispanic origin whether the risk factors in our
models influenced differently the odds of dying within five years from release. We found that
reimprisonment, age at release, and a history of multiple prison terms had a similar influence
on the odds of dying across all the race/Hispanic origin groups. Specifically, reimprisonment
after their 2010 release was associated with lower odds of death; and the odds of dying were
significantly higher for individuals released at older ages and for individuals who had served
multiple terms prior to the imprisonment that resulted in their 2010 release, compared to their
younger counterparts and those who had served only the prison term that ended in their 2010
release.
Other factors were significant for some groups and not for others. Non-Hispanic whites and
Hispanics who served a short term (under a year) prior to their 2010 release had higher odds of
dying within five years from release than those who served longer terms. Time served was not a
significant factor in the odds of dying within five years for non-Hispanic blacks or non-Hispanic
others. Confirming findings shown in Table 3 using the full sample, the analysis by
race/ethnicity shows that the odds of dying within five years from release were higher for non-

12

Hispanic whites and non-Hispanic others if they were released unconditionally, but type of
release did not influence the risk of death for non-Hispanic blacks or Hispanics.
To explore differences in the timing of death by race and Hispanic origin among the former
prisoners who died within five years from their 2010 release, we ran two models, shown in
Table 5. In Model 1, the three mutually exclusive categories in the dependent variable are: died
within two months from release (omitted), died between three and twelve months from
release, and died between thirteen months and five years from release. In Model 2, the
mutually exclusive categories in the dependent variable are: died within the first year from
release (omitted), died in the second year from release, and died between three and five years
from release.
The coefficients in Model 1, Table 5, show that race/Hispanic origin, reimprisonment, and
region of release were associated with the odds of dying within two months from release rather
than surviving to the first three to twelve months after release. Non-Hispanic whites, those who
were not reimprisoned, and individuals who were released from a prison in the West were
more likely to die in the first two months compared to non-Hispanic blacks and Hispanics, those
who were reimprisoned, and those released from prisons in the South.
Among the former prisoners who died in the five years after release, non-Hispanic blacks and
Hispanics were more likely than other groups to survive through the first year. Although a
history of multiple prison terms prior to their 2010 release did not increase the odds of dying
within two months after release, it was associated with being less likely to survive to the second
year after release compared to individuals who had served only one term, ending in 2010. In
addition, individuals released at older ages were more likely to survive to the second year after
release than their younger counterparts.
The coefficients in Model 2, Table 5, show that compared to individuals who died within the
first year after release, those who died later were older at release, more likely to be female,
less likely to have a history of multiple terms in prison prior to their 2010 release, and more
likely to have been reimprisoned at some point after their 2010 release.

Standardized Mortality Ratios (SMRs) Relative to Mortality Rates in the Resident Population
In addition to examining the factors associated with differential mortality within our sample of
former state prisoners, we computed standardized mortality ratios (SMRs) over the five-year
period of the study to compare the mortality rates of former prisoners to those of the resident
population during the same years, as well as to the mortality rates in the combined population
of the 35 states that provided the NCRP with prison release data in 2010. As mentioned earlier,
SMRs across studies differ substantially due to differences in mortality in the reference
13

populations, in the mortality of former prisoners, or both. In this study we held constant the
observed mortality among former prisoners and varied the reference populations. We find
essentially the same SMRs, whether we use as reference the mortality in the subsample of
states or the mortality in the total U.S. population. This suggests that factors not usually
considered, such as the racial/ethnic differences in mortality in each state, may be reduced by
using mortality rates from aggregated populations.
Table 6 shows that the mortality rate measured over the first five years after release from
prison for male former state prisoners was nearly thrice as high (SMR=2.8; 2.7-2.8) as for males
in the general U.S. population. For female former prisoners, the rate was nearly five times
higher than for females in the resident population (SMR=4.8; 4.6-5.0). Moreover, female former
prisoners had higher SMRs than male former prisoners in every age group in the study except
ages 65 and older.
Consistent with other studies, comparing mortality rates between former prisoners and the
resident population for the same years, sex, age group, and race/ethnicity, we found large
differences in SMRs, partly reflecting the mortality disadvantage of non-Hispanic blacks in the
resident population. Non-Hispanic black males (SMR=1.7; 1.7-1.8) and non-Hispanic black
females (SMR=2.8; 2.5-3.0) had the lowest SMRs in the sample, followed by non-Hispanic white
males (SMR=3.6; 3.5-3.7) and Hispanic males (SMR=4.1; 3.9-4.3). The highest SMRs in the
sample were among Hispanic females (SMR=7.9; 6.8-9.0) and non-Hispanic other race females
(SMR=9.6; 7.8-11.5) relative to their counterparts in the general US population.
Across all age and race/Hispanic origin groups in the study, the SMRs were highest within the
first few weeks after release from prison, and although they declined over time, after one year
they had not reached parity with the resident population. In the first month after release, nonHispanic whites had an SMR=13.8 (12.3-15.3); followed by SMR=13.0 (6.2-19.8) for nonHispanic other races; and by SMR=11.7 (8.9-14.5) for Hispanics. By sex, within the first month
after release, females died at much higher rates than males relative to the resident population.
For the first month after release, female former prisoners had an SMR of 11.7 (8.5, 14.9), while
males had an SMR of 8.9 (8.1, 9.7). The SMRs in the first month after release were high across
all age groups. Former prisoners aged 18-34 had an SMR of 8.6 (7.2, 10.1); the SMR was 9.3
(8.2, 10.4) for individuals ages 35-54; and 9.2 (7.7, 10.7) for individuals ages 55 and older. In the
second and later months after release, the SMRs declined faster for individuals released at ages
35 and older compared to those under age 35 at release from prison.

14

Discussion/Conclusions
This is the first study to leverage multi-state prison release data from the National Corrections
Reporting Program (NCRP). Individual records from the NCRP data were linked to the Census
Bureau Numident, decennial censuses, and survey files to explore the relationship between
imprisonment and mortality. The mortality status of former prisoners released in 2010 from 35
states was followed for five years after their release.
Our analysis was three-fold. First, we modeled the odds that an individual would die within five
years after release from prison. Second, among individuals who died in the five-year window,
we modeled the odds that the death occurred within the first two months post release, rather
than at a later time. Third, we computed SMRs to compare mortality of former prisoners to that
of their counterparts in the U.S. resident population.
This study confirms at the multi-state level findings previously reported from state-level
studies. Non-Hispanic white former prisoners were more likely to die within five years after
release, and also more likely to die in the initial weeks after release, compared to other racial
minorities and to Hispanics. In addition, as reported by state-level studies, mortality was higher
among individuals with a history of multiple imprisonments prior to their 2010 release than for
those with only one term in prison.
Because our large sample size allowed separate analysis by race and Hispanic origin of the role
that demographic, custodial/criminal, and geographic factors play on the risk of death of former
prisoners, our findings provide additional detail to what previous state-level studies have
reported. In particular, we found that some factors, such as duration of the prison term served
prior to release in 2010 and the type of release, influenced the odds of death only among nonHispanic whites, but were not always significant factors in the mortality risks of other
racial/Hispanic origin groups.
A contribution of this paper includes findings that the region from which individuals were
released influenced both their risk and timing of mortality. The region of release may not be the
same as the region where the death took place, but region is likely to serve as a marker or
proxy for differences in the type of prison or the provision of services immediately after release,
which may be contributing factors in the successful reintegration of former prisoner into their
communities.
Importantly, age at release, sex, education, time served, having a history of multiple
imprisonments, release type, and most serious offense did not influence the odds of dying
within two months from release rather than surviving through the first post release year.

15

Throughout our analysis, and consistent with previous state-level findings, reimprisonment
after the 2010 release was associated with lower odds of dying across all the racial/Hispanic
origin groups in the sample; and, among those who died, with a later time of death within our
five-year window. Reimprisonment has two seemingly contradictory effects on mortality. On
the one hand, reimprisonment exerts a life-protective effect that may come from reduced
exposure to multiple risks while in prison. On the other hand, as mentioned earlier, a history of
multiple imprisonment terms prior to release was associated with higher mortality, and it is
likely to signal a pattern of risk behaviors that increase the odds of death among former
prisoners.
The large multi-state sample in this study allowed for the computation of SMRs for Hispanics
(any race) separately from non-Hispanic groups. Our findings confirm reports from state-level
studies about the large mortality disadvantage between former prisoners relative to the U.S.
resident population across all race/Hispanic origin groups. The largest differences took place in
the first month after release from prison, when non-Hispanic other race, Hispanic, and nonHispanic white former state prisoners had mortality rates that were twelve to fourteen times
those found among individuals with similar characteristics in the resident population. In
addition, in the first month after release from prison, the mortality rates of female former
prisoners were about twelve times higher than for females in the resident population.
The large mortality differentials between former prisoners and the resident population suggest
the need for interventions in the immediate weeks after release from prison. Type of release
and post-release interventions could be tailored to address the needs of former prisoners
found to be most at risk: females, individuals released after a short term in prison, those with a
history of multiple prison terms, and those released unconditionally. Finally, in future research,
cause of death information could clarify the relationship between reimprisonment, type of
release, and mortality. Prisoners released conditionally can be subject to random drug tests,
supervision check-ins, employment, and stable housing requirements, and even electronic
monitoring, which may limit their ability to participate in risky behaviors such as drug and
alcohol intoxication.

16

References
Alper, Mariel, Matthew R. Durose, and Joshua Markman. 2018. 2018 Update on Prisoner
Recidivism: A 9-year Follow-up Period (2005-2014). Bureau of Justice Statistics Special Report.
Washington, DC: U.S. Department of Justice, Office of Justice Programs. Accessed 8/31/2018.
Available: https://bjs.ojp.gov/content/pub/pdf/18upr9yfup0514.pdf.
Binswanger, Ingrid A., Marc F. Stern, Richard A. Deyo, Patrick J. Heagerty, Allen Cheadle, Joann
G. Elmore, and Thomas D. Koepsell. 2007. Release from Prison — A High Risk of Death for
Former Inmates. The New England Journal of Medicine, 356(2): 157-165.
Binswanger, Ingrid A., Patrick J. Blatchford, Shane R. Mueller, and Marc F. Stern. 2013. Mortality
after Prison Release: Opioid Overdose and Other Causes of Death, Risk Factors, and Time
Trends from 1999 to 2009. Ann Intern Med, 159(9):592-600.
Bond, Brittany, J. David Brown, Adela Luque, and Amy O’Hara. 2014. The Nature of the Bias
When Studying Only Linkable Person Records: Evidence from the American Community Survey.
Working Paper CARRA-WP-2014-08. Washington, DC: U.S. Census Bureau.
Braveman, Paula, Susan Egerter, and David R. Williams. 2011. The Social Determinants of
Health: Coming of Age. Annual Review of Public Health, 32:381–398.
Carson, E. Ann. 2021a. Mortality in State and Federal Prisons, 2001-2019 – Statistical Tables.
Bureau of Justice Statistics Statistical Tables. Washington, DC: U.S. Department of Justice, Office
of Justice Programs. Accessed 12/1/2021. Available:
https://bjs.ojp.gov/content/pub/pdf/msfp0119st.pdf
Carson, E. Ann. 2021b. Prisoners in 2020 -- Statistical Tables. December 2021. Bureau of Justice
Statistics Bulletin NCJ 302776. Washington, DC: U.S. Department of Justice, Office of Justice
Programs. Accessed 12/22/2021. Available: https://bjs.ojp.gov/content/pub/pdf/p20st.pdf
Carson, E. Ann. 2018. Prisoners in 2016. Bureau of Justice Statistics Bulletin. Washington, DC:
U.S. Department of Justice, Office of Justice Programs. Accessed 8/24/2018. Available:
https://bjs.ojp.gov/content/pub/pdf/p16.pdf.
Carson, E. Ann, and William Sabol. 2016. Aging of the State Prison Population, 1993-2013.
Bureau of Justice Statistics Special Report. Washington, DC: U.S. Department of Justice, Office
of Justice Programs. Accessed 11/23/2021. Available: Aging of the State Prison Population,
1993–2013 (ojp.gov).

17

Chang, Zheng, Paul Lichtenstein, Henrik Larsson, and Seena Fazel. 2015. Substance Use
Disorders, Psychiatric Disorders, and Mortality after Release from Prison: A Nationwide
Longitudinal Cohort Study. Lancet Psychiatry, 2:422-30.
Dirkzwager, Anja, Paul Nieuwbeerta, and Arjan Blokland. 2012. Effects of First-time
Imprisonment on Postprison Mortality: A 25-year Follow-up Study with a Matched Control
Group. Journal of Research in Crime and Delinquency, 49(3): 383-419.
Dusch, G. and F. Meier. 2012. 2010 Census Content Reinterview Survey Evaluation Report. 2010
Census Program for Evaluations and Experiments. Washington, DC: U.S. Census Bureau.
Graham, Lesley, Colin M. Fischbacher, Diane Stockton, Andrew Fraser, Michael Fleming, and
Kevin Greig. 2015. Understanding Extreme Mortality among Prisoners: A National Cohort Study
in Scotland Using Data Linkage. European Journal of Public Health, 25(5): 879–885.
Human Rights Watch & Families Against Mandatory Minimums (HRW & FAMM). 2012. “The
Answer Is No: Too Little Compassionate Release in US Federal Prisons.” Accessed 11/23/2021.
Available: https://famm.org/wp-content/uploads/The-Answer-is-No-compassionaterelease.pdf.
Jones, Mark, Gregory D. Kearney, Xiaohui Xu, Tammy Norwood, and Scott K. Proescholdbell.
2017. Mortality Rates and Cause of Death among Former Prison Inmates in North Carolina. N C
Med J, 78(4):223-229.
Kinner, Stuart A. and Ingrid A. Binswanger. 2014. Mortality after Release from Prison. In:
Bruinsma, Gerben, and David Weisburd (Eds.) Encyclopedia of Criminology and Criminal Justice.
New York, NY: Springer.
Kochanek, Kenneth D., Sherry L. Murphy, Jiaquan Xu, and Elizabeth Arias. 2017. Mortality in the
United States, 2016. NCHS Data Brief No. 293. Hyattsville, MD: National Center for Health
Statistics.
Krinsky, Clarissa S., Sarah L. Lathrop, Pamela Brown, and Kurt B. Nolte. 2009. Drugs, Detention,
and Death: A Study of the Mortality of Recently Released Prisoners. The American Journal of
Forensic Medicine and Pathology, 30(1):6-9.
Kuzyk, I., K. Baudoin, and K. Bobula. 2018. Mortality among ex-prisoners. CT OPM, Criminal
Justice Policy & Planning Division. The Connecticut Statistical Analysis Center.
Laub, John H. and George E. Vaillant. 2000. Delinquency and Mortality: A 50-Year Follow-Up
Study of 1,000 Delinquent and Nondelinquent Boys. Am J Psychiatry, 157:96–102.
18

Looney, Adam and Nicholas Turner. 2017. Work and Opportunity Before and After
Incarceration. Washington, DC: The Brookings Institution.
Luallen, Jeremy, Kevin Neary, Brendan Rabideau, William Rhodes, Gerald Gaes, and Tom Rich.
2014. National Corrections Reporting Program (NCRP) White Paper Series. A Description of
Computing Code Used to Identify Correctional Terms and Histories. Prepared by Abt Associates,
Inc., for the Bureau of Justice Statistics, Office of Justice Programs, U.S. Department of Justice.
Available: https://www.icpsr.umich.edu/files/NACJD/ncrp/white-paper-computing-code.pdf
Merrall, Elizabeth L.C., Azar Kariminia, Ingrid A. Binswanger, Michael S. Hobbs, Michael Farrell,
John Marsden, Sharon J. Hutchinson, and Sheila M. Bird. 2010. Meta-analysis of Drug-related
Deaths Soon after Release from Prison. Addiction, 105: 1545-1554.
Murphy, P. 2014. American Community Survey 2012 Content Reinterview Survey. 2014
American Community Survey Research and Evaluation Report Memorandum Series #ACS14RER-01. Washington, DC: U.S. Census Bureau.
Nelson, Karin, Greg Schwartz, Susan Hernandez, Joseph Simonetti, Idamay Curtis, and Stephan
D. Fihn. 2016. The Association between Neighborhood Environment and Mortality: Results from
a National Study of Veterans. J Gen Intern Med, 32(4):416–422.
Noonan, Margaret E. and Scott Ginder. 2015. Understanding Mortality in State Prison: Do Male
Prisoners Have an Elevated Risk Of Death? Justice Research and Policy, 16(1):65-80.
Patterson, Evelyn J. 2013. The Dose-Response of Time Served in Prison on Mortality: New York
State, 1989-2003. American Journal of Public Health, 103(3):523-528.
Pridemore, William Alex. 2014. The Mortality Penalty of Incarceration: Evidence from a
Population-Based Case-Control Study of Working-Age Males. Journal of Health and Social
Behavior, 55(2): 215-233.
Rosen, David L., Victor J. Schoenbach, and David A. Wohl. 2008. All-Cause and Cause-Specific
Mortality among Men Released from State Prison, 1980-2005. American Journal of Public
Health, 98(12): 2278-2284.
Rosen, David L., David A. Wohl, and Victor J. Schoenbach. 2011. All-Cause and Cause-Specific
Mortality among Black and White North Carolina State Prisoners, 1995-2005. Ann
Epidemiology, 21(10): 719-726.

19

Schnittker, Jason, Michael Massoglia, and Christopher Uggen. 2011. Incarceration and the
Health of the African American Community. Du Bois Review: Social Science Research on Race, 8:
133-141.
Singer, Phyllis and Sharon R. Ennis. 2003. Census 2000 Content Reinterview Survey: Accuracy of
Data for Selected Population and Housing Characteristics as Measured by Reinterview. Census
2000 Evaluation B.5. Washington, DC: U.S. Census Bureau.
Spaulding, Anne C., Ryan M. Seals, Victoria A. McCallum, Sebastian D. Perez, Amanda K.
Brzozowski, and N. Kyle Steenland. 2011. Prisoner Survival Inside and Outside of the Institution:
Implications for Health-Care Planning. American Journal of Epidemiology, 173(5):479-487.
Testa, Alexander, Lauren C. Porter, and Kiminori Nakamura. 2018. Examining All-Cause and
Cause-Specific Mortality among Former Prisoners in Pennsylvania. Justice Quarterly, 35(5):782815.
United States Government Accountability Office (USGAO). 2012. “Bureau of Prisons: Eligibility
and Capacity Impact Use of Flexibilities to Reduce Inmates’ Time in Prison.” Report to
Congressional Requestors (“GAO February BOP Report”), GAO 12-320.
van de Weijer, Steve, Catrien Bijleveld, and Doreen Huschek. 2016. Offending and Mortality.
Advances in Life Course Research, 28: 91-99.
Wagner, Deborah and Mary Layne. 2014. The Person Identification Validation System (PVS):
Applying the Center for Administrative Records Research and Applications’ (CARRA) Record
Linkage Software. Center for Administrative Records Research and Applications Internal Paper,
Washington, DC: U.S. Census Bureau.
Warren Andersen, Shaneda, William J. Blot, Xiao-Ou Shu, Jennifer S. Sonderman, Mark
Steinwandel, Margaret K. Hargreaves, and Wei Zheng. 2018. Associations Between
Neighborhood Environment, Health Behaviors, and Mortality. Am J Prev Med, 54(1): 87-95.
Wildeman, Christopher, Margaret E. Noonan, Daniela Golinelli, E. Ann Carson, and Natalia
Emanuel. 2016a. State-Level Variation in the Imprisonment-Mortality Relationship, 2001-2010.
Demographic Research, 34(12):359-371.
Wildeman, Christopher, E. Ann Carson, Daniela Golinelli, Margaret E. Noonan, and Natalia
Emanuel. 2016b. Mortality among White, Black, and Hispanic Male and Female State Prisoners,
2001-2009. SSM - Population Health, 2:10-13.

20

Zane, Steven N., Brandon C. Welsh, and Gregory M. Zimmerman. 2018. Criminal Offending and
Mortality over the Full Life‑Course: A 70‑Year Follow‑up of the Cambridge–Somerville Youth
Study. Journal of Quantitative Criminology, 34: 1-23.
Zlodre, Jakov and Seena Fazel. 2012. All-Cause and External Mortality in Released Prisoners:
Systematic Review and Meta-Analysis. Am J Public Health, 102:e67–e75.

21

Table 1. Construction of Independent Variables from the National Corrections Reporting
Program (NCRP) Linked to Census Bureau Data
Demographic Factors
Age at release

Sex
Hispanic origin
& Race

Education
Custodial Factors
Time served

Additional
terms served
prior to 2010

= age of the person in 2010 when released from prison, coded in seven
categories (under 25, 25-34, 35-44, 45-54, 55-64, and 65 and older, and
missing)
= sex of prisoner, male, female, or missing
= coded as Hispanic (any race), non-Hispanic white, non-Hispanic black,
non-Hispanic other (combines Asian, American Indian or Alaska Native,
Native Hawaiian and Other Pacific Islander, and multiracial individuals), and
a missing category. Race and Hispanic origin information in NCRP records
rely on different methods of collection across the states, and BJS has found
that Hispanic and multiracial individuals are underrepresented in these data
(Carson 2018). In contrast, race and Hispanic origin responses in decennial
censuses and the ACS have been found to be of high quality (Singer and
Ennis 2003; Dusch and Meier 2012; Murphy 2014). Comparing NCRP race
and Hispanic origin responses with those in Census data, we found that 75.3
percent of the records had the same race and Hispanic origin; 4.3 percent
were assigned the race and Hispanic origin from census records because
they differed or were missing in NCRP; 20.2 percent retained race and
Hispanic origin data from the NCRP due to missing census data, and 0.2
percent were missing these data in both files. Individuals in the nonHispanic other category were more likely to have missing information in
Census and NCRP records (disclosure of racial/ethnic differences between
NCRP and Census was approved by CBDRB-FY2022-CES010-003).
= as reported in NCRP, coded into five categories (less than high school,
high school or GED, some college, college or higher, and missing).
= time served in last imprisonment prior to 2010 release, coded as mutually
exclusive categories (less than 1 year, 1 year to less than 2 years, 2 years to
less than 3 years, 3 years to less than 5 years, 5 years to less than 7 years, 7
years to less than 10 years, and 10 or more years).
= the number of imprisonment terms served between 2005 and 2010 in
addition to the one that ended with their release in 2010. Coded as 0
additional terms, one additional term, two or more additional terms. We
only go back to 2005 because that is when the records contain more
complete information across all states. A caveat is that for 2005-2010 terms
of imprisonment, the NCRP data are better at capturing within state prior
prison terms than out-of-state prior prison terms. Therefore, our estimates
of additional terms served prior to the one ending in 2010 release are likely
to be incomplete. Linking individual terms of imprisonment over time
within a state in the NCRP is dependent on the quality of identifiers. In
some cases, states have submitted data with unique and consistent
identifiers back to the 1970s, while other states have only recently adopted
consistent numbers to capture individuals who come back to prison in that

22

(Continued) Table 1. Construction of independent variables from the National Corrections
Reporting Program (NCRP) linked to Census Bureau Data
(Continued)
Additional
terms served
prior to 2010

same state. Identifying individuals who are imprisoned in two different
states is achieved by the use of the FBI number, a biometric identification
number that is unique to the entire country. BJS began collecting this
number as part of NCRP in 2012, although multiple states have resubmitted
prior years’ data with this identifier. For more details on the construction of
NCRP term records, please see the NCRP white paper by Luallen et al.
(2014) available at:
(https://www.icpsr.umich.edu/icpsrweb/content/NACJD/guides/ncrp.html).
Release type
= criminal offenders may be conditionally released from prison to serve the
remaining portion of their sentence in the community. Individuals under
conditional release may or may not be required to fulfill certain conditions
and adhere to specific rules of conduct while in the community, including
regularly reporting to a community supervisor in person, by mail, or by
telephone. Failure to comply with any of the conditions can result in
reimprisonment. Unconditional releases do not have any requirements
after release and are the result of a prisoner serving the duration of his/her
sentence, being pardoned, or having the sentence commuted. Releases due
to transfers, escapes, and death were excluded from the analyses. Coded as
conditional, unconditional or missing.
Reimprisonment = whether the individual was reimprisoned after their 2010 release and
within our five-year follow up window ending 2015 (yes, no).
Criminal History
Most serious
= most serious offense for which an individual was serving time during the
offense
term ending in release from prison in 2010. The categories are coded as:
• violent (murder, manslaughter, sexual assault, robbery, assault, other
violent crimes),
• property (burglary, larceny, motor vehicle theft, fraud, possession of
stolen property, and other property crimes),
• drugs (sale, manufacturing, possession),
• public order (disorderly conduct, obstruction of justice, driving while
intoxicated, weapons, commercialized vice, immigration offenses),
• other/unspecified (juvenile offenses, unspecified felonies or
misdemeanors), and
• missing information.
Geographic location
Region
= region from where individuals were released (South, Midwest, Northeast
or West).

23

Table 2. Time between Release and Death by Selected Characteristics, 2010 Release Cohort

Characteristics

Total
Hispanic origin and race
Non-Hispanic white
Non-Hispanic black
Non-Hispanic all other
Hispanic
Missing
Age at release 1
Less than 25 years
25-34 years
35-44 years
45-54 years
55-64 years
65 years and older
Missing
Sex
Male
Female
Missing

Annual Percent of Total Deaths in the
Percent
Monthly
Percent
of
Total
Deaths
in
Year
1
after
Release
First 5 Years after Release
Number
deceased
released in
within 5
Avg for Avg for Avg for
2010 (with PIK, years of
Months Months Months
no duplicates)
release
Month 1 Month 2 Month 3
4-6
7-9
10-12 Year 1 Year 2 Year 3 Year 4 Year 5
A
B
478,000
3.8
3.1
1.4
1.7
1.7
1.4
1.4 20.1 18.4 19.0 20.7 21.8

204,000
179,000
18,000
76,000
1,100

4.7
3.0
3.6
3.2
3.6

3.7
2.8
(D)
3.0
(D)

1.6
1.5
(D)
1.7
0.0

1.6
1.5
(D)
2.1
(D)

1.6
1.7
(D)
1.7
(D)

1.6
1.7
(D)
1.7
0.0

1.6
1.5
(D)
1.3
(D)

20.0
19.1
23.1
20.8
(D)

19.0
18.1
15.4
18.8
(D)

19.0
19.1
15.4
18.8
(D)

20.0
21.0
23.1
20.8
(D)

22.1
22.9
23.1
20.8
(D)

82,500
173,000
120,000
80,500
19,000
3,000
(D)

1.8
2.3
3.3
6.6
13.7
23.3
0.0

2.6
2.3
3.8
2.9
4.0
5.5

2.0
1.6
1.8
1.2
1.2
(D)

2.0
1.8
1.5
1.3
1.5
(D)

2.0
1.8
1.5
1.5
1.5
(D)

2.0
1.8
1.5
1.3
1.5
2.9

2.0
1.8
1.5
1.3
1.5
(D)

22.6
21.8
20.3
18.3
19.6
26.7

19.4
19.2
17.7
18.3
17.7
13.3

19.4
19.2
19.0
19.2
19.6
20.0

16.1
19.2
20.3
21.2
21.6
20.0

22.6
20.5
22.8
23.1
21.6
20.0

421,000
57,000
<15

3.8
3.5
0.0

3.2
2.5

1.6
1.5

1.6
1.5

1.6
1.0

1.6
1.5

1.6
1.5

20.9
17.1

18.4
17.1

19.0
19.5

20.3
22.0

21.5
24.4

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123 and CBDRB-FY21-CES010-011.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Note: All counts are rounded to be in compliance with the Census Bureau's disclosure avoidance policies.
1
Age at release was calculated using date of birth and release date.
2

Persons who died in prison, were transferred, or escaped from prison were removed from analyses.

(D) indicates that a cell has been suppressed for disclosure avoidance purposes.

24

(Continued) Table 2. Time between Release and Death by Selected Characteristics, 2010 Release Cohort

Characteristics

Education reported in NCRP
Less than high school
High school or GED
Some college
College or higher
Missing
Time served
Less than 1 year
1-1.9 years
2-2.9 years
3-4.9 years
5-6.9 years
7-9.9 years
10 or more years
Additional terms served
between 2005 and 2010
0 terms
1 term
2 or more terms

Number
released in
2010 (with PIK,
no duplicates)
A

Percent
deceased
within 5
years of
release
B

Monthly Percent of Total Deaths in Year 1 after Release

Annual Percent of Total Deaths in the
First 5 Years after Release

Avg for Avg for Avg for
Months Months Months
Month 1 Month 2 Month 3
4-6
7-9
10-12 Year 1 Year 2 Year 3 Year 4 Year 5

142,000
119,000
16,500
2,700
198,000

3.5
3.8
4.6
5.6
3.8

3.0
2.2
2.6
(D)
4.0

1.8
1.4
(D)
(D)
1.3

1.8
1.3
(D)
0.0
2.0

1.8
1.3
(D)
(D)
2.0

1.6
1.6
(D)
(D)
1.3

1.6
1.6
2.7
(D)
1.3

20.2
18.9
20.0
25.0
20.0

18.2
18.9
20.0
18.8
17.3

19.2
20.0
20.0
18.8
18.7

20.2
20.0
20.0
18.8
21.3

22.2
22.2
20.0
18.8
22.7

271,000
96,000
42,500
35,000
13,000
9,600
11,500

3.9
3.7
3.5
3.7
3.5
4.2
5.2

2.9
2.9
2.7
3.1
4.4
(D)
7.4

1.5
1.2
2.0
1.6
(D)
(D)
(D)

1.4
1.4
2.0
(D)
(D)
(D)
(D)

1.4
1.7
1.3
1.5
(D)
(D)
(D)

1.4
1.7
1.3
1.5
(D)
(D)
(D)

1.4
1.4
1.3
1.5
(D)
(D)
(D)

19.4
20.3
20.7
19.2
21.7
22.5
27.3

18.5
17.4
17.2
19.2
19.6
17.5
18.2

19.4
20.3
20.7
19.2
17.4
17.5
18.2

20.4
20.3
20.7
19.2
21.7
20.0
18.2

22.3
21.7
20.7
23.1
19.6
22.5
18.2

340,000
82,000
56,000

3.7
3.7
4.3

2.8
3.4
3.3

1.6
1.7
1.2

1.6
1.7
2.1

1.6
1.7
1.3

1.6
1.3
1.7

1.6
1.3
1.3

20.5
20.3
18.8

18.1
18.6
18.8

18.9
18.6
18.8

20.5
20.3
20.8

22.1
22.0
22.9

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123 and CBDRB-FY21-CES010-011.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Note: All counts are rounded to be in compliance with the Census Bureau's disclosure avoidance policies.
1
Age at release was calculated using date of birth and release date.
2

Persons who died in prison, were transferred, or escaped from prison were removed from analyses.

(D) indicates that a cell has been suppressed for disclosure avoidance purposes.

25

(Continued) Table 2. Time between Release and Death by Selected Characteristics, 2010 Release Cohort

Characteristics

Release type 2
Conditional
Unconditional
Missing
Reimprisoned after 2010
release
No
Yes
Most serious offense
Violent
Property
Drugs
Public order
Other/unspecified
Missing
Region
South
Midwest
Northeast
West

Number
released in
2010 (with PIK,
no duplicates)
A

Percent
deceased
within 5
years of
release
B

Monthly Percent of Total Deaths in Year 1 after Release

Annual Percent of Total Deaths in the
First 5 Years after Release

Avg for Avg for Avg for
Months Months Months
Month 1 Month 2 Month 3
4-6
7-9
10-12 Year 1 Year 2 Year 3 Year 4 Year 5

311,000
145,000
22,000

3.7
3.9
4.1

3.1
2.6
7.9

1.3
1.7
2.3

1.7
1.4
2.2

1.7
1.8
2.2

1.3
1.6
2.2

1.3
1.8
(D)

18.6
21.1
23.5

18.6
19.3
17.7

19.5
19.3
17.7

21.2
19.3
17.7

22.1
21.1
23.5

262,000
216,000

5.2
2.0

4.0
(D)

1.8
(D)

2.2
(D)

2.2
(D)

1.9
0.5

1.9
0.5

25.6
3.5

19.7
13.8

18.3
20.7

18.3
27.6

18.3
34.5

116,000
139,000
128,000
89,000
3,000
3,000

3.7
3.6
3.6
4.4
3.3
3.3

3.5
3.0
3.3
2.6
(D)
(D)

1.9
1.2
1.5
1.5
(D)
(D)

1.6
1.6
1.5
1.8
(D)
(D)

1.6
1.8
1.7
1.5
(D)
(D)

1.6
1.6
1.5
1.5
(D)
(D)

1.4
1.4
1.5
1.5
(D)
(D)

21.2
20.0
19.6
19.2
20.0
18.2

18.8
18.0
18.5
18.0
20.0
18.2

18.8
20.0
19.6
19.2
20.0
18.2

20.0
20.0
20.7
20.5
20.0
27.3

21.2
22.0
21.7
23.1
20.0
18.2

192,000
97,000
53,000
136,000

3.8
3.7
3.4
3.8

3.4
2.7
2.8
3.8

1.4
1.4
1.1
1.5

1.4
1.4
1.7
1.7

2.1
1.7
1.7
1.5

1.4
1.4
1.7
1.4

1.4
1.4
1.7
1.4

20.6
19.2
19.4
19.1

17.8
19.2
19.4
18.1

19.2
19.2
19.4
19.1

20.6
21.9
19.4
21.0

21.9
20.6
22.2
22.9

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123 and CBDRB-FY21-CES010-011.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Note: All counts are rounded to be in compliance with the Census Bureau's disclosure avoidance policies.
1

Age at release was calculated using date of birth and release date.

2

Persons who died in prison, were transferred, or escaped from prison were removed from analyses.

(D) indicates that a cell has been suppressed for disclosure avoidance purposes.

26

Table 3. Logistic Regression with Full Sample Predicting the Mortality of Prisoners
Released in 2010, Odds Ratios
Deceased within 5 years of release
(Alive is omitted category)

Characteristics

Model A
Hispanic origin and race
Non-Hispanic white (omitted)
Non-Hispanic black
Non-Hispanic all others
Hispanic
Age at release
Less than 25 years (omitted)
25-34 years
35-44 years
45-54 years
55 years and older
Sex
Male (omitted)
Female
Education reported in NCRP
Less than high school (omitted)
High school or GED
Some college
College or higher
Missing
Time served
Less than 1 year (omitted)
1-1.9 years
2-4.9 years
5 or more years
Additional terms served between 2005 and 2010
0 terms (omitted)
1 term
2 or more terms
Release type
Conditional (omitted)
Unconditional
Reimprisoned after 2010 release
No (omitted)
Yes

Model B

Model C

Model D

0.65***
0.88***
0.76***

1.10
0.96
0.92

0.66***
0.86**
0.69***

0.69***
0.83***
0.78***

1.14***
1.61***
3.20***
7.66***

1.44***
2.11***
4.07***
9.75***

1.14***
1.62***
3.20***
7.66***

1.14***
1.61***
3.20***
7.65***

0.83***

0.84***

0.83***

0.83***

0.94**
0.92*
0.83*
0.96

0.95**
0.92*
0.83*
0.97

0.94**
0.91*
0.82*
0.96

0.94**
0.92*
0.83*
0.97

0.94**
0.91***
0.81***

0.94**
0.90***
0.81***

0.94**
0.91***
0.81***

0.94**
0.91***
0.80***

1.18***
1.50***

1.18***
1.50***

1.18***
1.49***

1.18***
1.50***

1.06**

1.06**

1.06**

1.11***

0.41***

0.41***

0.39***

0.41***

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123 and CBDRB-FY22CES014-004.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Note: All counts are rounded to be in compliance with the Census Bureau's disclosure avoidance policies.
*p<.05, **p<.01, ***p<.001

27

(Continued) Table 3. Logistic Regression with Full Sample Predicting the Mortality of
Prisoners Released in 2010, Odds Ratios
Deceased within 5 years of release
(Alive is omitted category)

Characteristics

Model A
Most serious offense
Violent (omitted)
Property
Drugs
Public order
Region
South (omitted)
Midwest
Northeast
West
Race * Age at release
Non-Hispanic black * 25-34.9 years
Non-Hispanic black * 35-44.9 years
Non-Hispanic black * 45-54.9 years
Non-Hispanic black * ≥ 55 years
Non-Hispanic all others * 25-34.9 years
Non-Hispanic all others * 35-44.9 years
Non-Hispanic all others * 45-54.9 years
Non-Hispanic all others * ≥ 55 years
Hispanic * 25-34.9 years
Hispanic * 35-44.9 years
Hispanic * 45-54.9 years
Hispanic * ≥ 55 years
Race * Reimprisonment
Non-Hispanic black * reimprisoned
Non-Hispanic all others * reimprisoned
Hispanic * reimprisoned
Race * Release type
Non-Hispanic black * unconditional release
Non-Hispanic all others * unconditional release
Hispanic * unconditional release
Sample size
Deceased within 5 years of release
Pseudo R-Square

Model B

Model C

Model D

1.04
0.98
1.05*

1.05*
0.98
1.05*

1.04
0.98
1.05*

1.04
0.98
1.05*

1.02
0.94
0.94*

1.02
0.95
0.94*

1.02
0.95
0.94*

1.02
0.94*
0.94**

0.61***
0.53***
0.54***
0.59***
0.91
0.94
0.99
0.78
0.77**
0.80**
0.92
0.79*
0.97
1.12
1.45***
0.86***
1.15
0.90
451,000
17,000
0.0766

451,000
17,000
0.0773

451,000
17,000
0.0770

451,000
17,000
0.0767

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123 and CBDRB-FY22CES014-004.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Note: All counts are rounded to be in compliance with the Census Bureau's disclosure avoidance policies.
*p<.05, **p<.01, ***p<.001

28

Figure 1. Probability of Dying Within Five Years from State Prison Release by Race and Age at
Release
0.134 ***

>= 55

0.137**

0.122***

0.175
0.066***
0.074

45-54.9

0.077
0.030***
0.036

35-44.9

19***
0.024
8***
0.027

25-34.9

< 25

0.000

0.017
0.017
0.020
0.018
0.020

0.040

0.060

0.080

0.100

0.120

0.140

0.160

0.180

0.200

Pr(Dying)
■ Hispanic

■

Non-Hispa nic other

■

Non-Hispanic black

■

Non-Hispanic white

Approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Asterisks indicate significant difference in comparison to non-Hispanic white in the same age group *p<.05, **p<.01, ***p<.001

29

Figure 2. Probability of Dying Within Five Years From State Prison Release by Race and
Reimprisonment
0.070

0 .062

0 .054**

0.060

0 .044***

0.050

0.042***
T
.L

tio 0.040
C

·;;.

-0:.
0

0.022

0.030

0.023

0.023
T
l

0.020

0.015***

...
0.010

0.000

Not reimprisoned
■

Non-Hispanic white

■

Reim prisoned
Non-Hispanic black

■

Non-Hispanic other

■

Hispanic

Approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Asterisks indicate significant difference in comparison to non-Hispanic white in the same reimprisonment category
*p<.05, **p<.01, ***p<.001

30

Figure 3. Probability of Dying Within Five Years From State Prison Release by Race and Type
of Release in 2010
0.060

0.047***

0.049***
0.050

T
l

0.042
0.036

0.040

0.030

0.034

0.035

0.030

ti.I)

C

T

-....

.1

·s;.. 0.030
0

a..

0.020

0.010

0.000

Non-Hispanic white

Non-Hispanic black

■ Conditional Release

Non-Hispanic other

Hispanic

■ Unconditional Release

Approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY22-CES014-004.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Asterisks indicate significant differences in comparison between conditional and unconditional release for each race/Hispanic origin
category *p<.05, **p<.01, ***p<.001

31

Table 4. Logistic Regression Predicting the Mortality of Prisoners Released in 2010 by
Race, Odds Ratios
Deceased within 5 years of release
Characteristics

Age at release
Less than 25 years (omitted)
25-34 years
35-44 years
45-54 years
55 years and older
Sex
Male (omitted)
Female
Education reported in NCRP
Less than high school (omitted)
High school or GED
Some college
College or higher
Missing
Time served
Less than 1 year (omitted)
1-1.9 years
2-4.9 years
5 or more years
Additional terms served between 2005 and 2010
0 terms (omitted)
1 term
2 or more terms
Release type
Conditional (omitted)
Unconditional
Reimprisoned after 2010 release
No (omitted)
Yes

(Alive is omitted category)
Model A
NonHispanic
White

NonHispanic
Black

NonHispanic
Other

Hispanic

1.46***
2.14***
4.12***
9.86***

0.86**
1.08
2.15***
5.52***

1.35
2.04***
4.14***
8.02***

1.10
1.70***
3.87***
8.46***

0.81***

0.87**

0.85

0.83*

0.94*
0.95
0.89
0.96

0.91*
0.86*
0.65*
0.98

0.85
1.07
0.35
0.60***

1.06
0.70
0.79
1.00

0.92**
0.88***
0.77***

0.97
0.95
0.89

1.08
0.89
0.70

0.89*
0.90
0.78**

1.20***
1.44***

1.10*
1.39***

1.16
1.42**

1.25***
1.75***

1.08***

0.98

1.22*

1.09

0.39***

0.37***

0.42***

0.54***

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Note: All counts are rounded to be in compliance with the Census Bureau's disclosure avoidance policies.
*p<.05, **p<.01, ***p<.001

32

(Continued) Table 4. Logistic Regression Predicting the Mortality of Prisoners Released in 2010 by
Race, Odds Ratios
Deceased within 5 years of release
(Alive is omitted category)
Model A
Characteristics
NonNonNonHispanic
Hispanic
Hispanic
White
Black
Other
Hispanic
Most serious offense
Violent (omitted)
Property
1.05
0.99
1.13
1.16**
Drugs
0.97
1.00
0.92
0.99
Public order
1.06
1.03
1.13
0.97
Region
South (omitted)
Midwest
1.02
1.05
1.00
0.82
Northeast
0.98
0.96
1.04
0.82*
West
0.85***
1.08
1.32*
0.96
Sample size
Deceased within 5 years of release
Pseudo R-Square

193,000
9,000
0.0812

166,000
4,900
0.0655

17,000
600
0.0716

75,000
2,300
0.0651

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Note: All counts are rounded to be in compliance with the Census Bureau's disclosure avoidance policies.
*p<.05, **p<.01, ***p<.001

33

Table 5. Multinomial Logistic Regression Predicting Time between Release and Death for Prisoners Released
in 2010 who Died within Five Years from Release, Odds Ratios
Time between Release and Death
Model 1

Characteristics

Model 2

3-12 months I 13-60 months
13-24 months I 25-60 months
(< 3 months is omitted category) (< 13 months is omitted category)
Hispanic origin and race
Non-Hispanic white (omitted)
Non-Hispanic black
Non-Hispanic all others
Hispanic
Age at release
Less than 25 years (omitted)
25-34 years
35-44 years
45-54 years
55 years and older
Sex
Male (omitted)
Female
Education reported in NCRP
Less than high school (omitted)
High school or GED
Some college
College or higher
Missing
Time served
Less than 1 year (omitted)
1-1.9 years
2-4.9 years
5 years or more
Additional terms served between 2005 and 2010
0 terms (omitted)
1 term
2 or more terms
Release type
Conditional (omitted)
Unconditional
Reimprisoned after 2010 release
No (omitted)
Yes

1.35**
1.38
1.37*

1.33**
1.08
1.27*

0.97
0.71**
0.96

1.08
0.89
1.01

1.25
0.88
1.10
0.97

1.35*
1.21
1.89***
1.57**

1.06
1.12
1.45***
1.27*

1.16
1.42***
1.89***
1.76***

0.93

1.30*

1.23**

1.44***

1.04
1.27
0.62
0.95

1.15
1.13
0.59
0.93

1.13
0.98
0.75
0.87

1.11
0.92
0.87
1.00

1.11
0.91
0.97

1.00
0.83
0.78

0.89
0.86
0.81*

0.93
0.91
0.81**

0.82
0.79

0.69***
0.54***

0.88
0.77**

0.78***
0.61***

1.09

1.08

1.01

1.01

21.81***

187.6***

5.46***

13.59***

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Note: All counts are rounded to be in compliance with the Census Bureau's disclosure avoidance policies.
*p<.05, **p<.01, ***p<.001

34

(Continued) Table 5. Multinomial Logistic Regression Predicting Time between Release and Death for
Prisoners Released in 2010 who Died within Five Years from Release, Odds Ratios
Time between Release and Death
Model 1
Model 2

Characteristics

3-12 months

I

13-60 months

13-24 months

I

25-60 months

(< 3 months is omitted category) (< 13 months is omitted category)
Most serious offense
Violent (omitted)
Property
Drugs
Public order
Region
South (omitted)
Midwest
Northeast
West
Deceased within 5 years of release
Pseudo R-Square

1.08
1.14
1.23

1.02
1.16
1.24

0.94
1.01
0.98

0.97
1.07
1.09

0.97
1.00
0.74*

1.05
1.10
0.78*

1.13
1.32**
1.12

1.06
1.04
0.94

17,000
0.0723

17,000
0.0544

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Note: All counts are rounded to be in compliance with the Census Bureau's disclosure avoidance policies.
*p<.05, **p<.01, ***p<.001

35

Table 6. Standardized Mortality Ratios (SMRs) among Former Prisoners in the Five Years
after Release, Adjusted for Age Group, Sex, Race/Ethnicity and Year of Death^
Characteristics

SMRs with Total US
as Reference

All deaths in the sample, 2010-2015

2.9

2.9 , 2.9

2.8
4.8

2.7 , 2.8
4.6 , 5.0

3.5
3.2
3.1
2.7
2.5
1.3

3.3 ,
3.1 ,
2.9 ,
2.6 ,
2.4 ,
1.2 ,

3.7
3.3
3.2
2.8
2.6
1.4

5.9
6.7
5.6
4.0
4.2
1.1

4.7 ,
6.1 ,
5.1 ,
3.7 ,
3.6 ,
0.7 ,

7.2
7.3
6.0
4.3
4.8
1.6

3.6
1.7
5.5
4.1

3.5 ,
1.7 ,
5.0 ,
3.9 ,

3.7
1.8
6.0
4.3

5.7
2.8
9.6
7.9

5.4 ,
2.5 ,
7.8 ,
6.8 ,

6.0
3.0
11.5
9.0

95%CI

By sex
Males
Females
By age at release and sex
18-24 years, males
25-34 years, males
35-44 years, males
45-55 years, males
55-64 years, males
65 and older, males
18-24 years, females
25-34 years, females
35-44 years, females
45-55 years, females
55-64 years, females
65 and older, females
By race/Hispanic origin and sex
Non-Hispanic white, males
Non-Hispanic black, males
Non-Hispanic other, males
Hispanic, males (any race)
Non-Hispanic white, females
Non-Hispanic black, females
Non-Hispanic other, females
Hispanic, females (any race)

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123.
^ Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, Previous Census Data and
national and state mortality rates obtained from the CDC Wide-Ranging Online Data for Epidemiologic Research
(CDC-WONDER) for years 2010-2015

36

(Continued) Table 6. Standardized Mortality Ratios (SMRs) among Former Prisoners in the
Five Years after Release, Adjusted for Age Group, Sex, Race/Ethnicity and Year of Death^
Characteristics

SMRs with Total US
as Reference

Deaths within the first few months post-release
By months since release and race/Hispanic origin
1 month, non-Hispanic white
13.8
2 months, non-Hispanic white
5.5
3-6 months, non-Hispanic white
5.0
7-12 months, non-Hispanic white
3.4

95%CI

12.3 ,
4.6 ,
4.6 ,
3.2 ,

15.3
6.4
5.4
3.6

1 month, non-Hispanic black
2 months, non-Hispanic black
3-6 months, non-Hispanic black
7-12 months, non-Hispanic black

4.6
2.4
2.3
1.8

3.9 ,
1.8 ,
2.0 ,
1.6 ,

5.4
2.9
2.5
1.9

1 month, non-Hispanic other
2 months, non-Hispanic other
3-6 months, non-Hispanic other
7-12 months, non-Hispanic other

13.0
12.5
8.6
6.7

6.2 ,
5.7 ,
6.2 ,
5.1 ,

19.8
19.3
11.0
8.3

1 month, Hispanic
2 months, Hispanic
3-6 months, Hispanic
7-12 months, Hispanic

11.7
5.7
5.8
4.2

8.9 ,
3.8 ,
4.9 ,
3.7 ,

14.5
7.6
6.7
4.8

11.7
7.0
4.7
4.1

8.5 ,
4.5 ,
3.8 ,
3.4 ,

14.9
9.5
5.7
4.7

8.9
3.9
3.7
2.7

8.1 ,
3.4 ,
3.5 ,
2.5 ,

9.7
4.4
4.0
2.8

By months since released and sex
1 month, females
2 months, females
3-6 months, females
7-12 months, females
1 month, males
2 months, males
3-6 months, males
7-12 months, males

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123.
^ Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, Previous Census Data and
national and state mortality rates obtained from the CDC Wide-Ranging Online Data for Epidemiologic Research
(CDC-WONDER) for years 2010-2015

37

(Continued) Table 6. Standardized Mortality Ratios (SMRs) among Former Prisoners in the
Five Years after Release, Adjusted for Age Group, Sex, Race/Ethnicity and Year of Death^
Characteristics

SMRs with Total US
as Reference

Deaths within the first few months post-release
By months since release and age at release
1 month, ages 18-34 at release
8.6
2 months, ages 18-34 at release
5.7
3-6 months, ages 18-34 at release
5.2
7-12 months, ages 18-34 at release
3.7

95%CI

7.2 ,
4.6 ,
4.7 ,
3.4 ,

10.1
6.9
5.7
4.0

1 month, ages 35-54 at release
2 months, ages 35-54 at release
3-6 months, ages 35-54 at release
7-12 months, ages 35-54 at release

9.3
3.9
3.8
2.6

8.2 ,
3.2 ,
3.5 ,
2.4 ,

10.4
4.6
4.1
2.8

1 month, ages 55+ at release
2 months, ages 55+ at release
3-6 months, ages 55+ at release
7-12 months, ages 55+ at release

9.2
2.7
2.6
2.1

7.7 ,
1.9 ,
2.2 ,
1.9 ,

10.7
3.5
2.9
2.3

Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123.
^ Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, Previous Census Data and
national and state mortality rates obtained from the CDC Wide-Ranging Online Data for Epidemiologic Research
(CDC-WONDER) for years 2010-2015

38

APPENDIX – Comparison of Unduplicated Analytic Sample with Original 2010 Release Cohort
As shown in Appendix Table A, across the 2010 cohort and the two PIK samples (with duplicates
and unduplicated), most individuals (about 60 percent) were between the ages of 25 and 44 at
the time of release, most (about 88 percent) were males, and about half had high school or
lower levels of education at admission to prison. Over half served terms of less than one year in
their last imprisonment, and for most individuals (about 70 percent) that was their first term in
prison. About two thirds of the individuals received a conditional release, and about 45 percent
were reimprisoned at some point between their 2010 release and 2015, the five-year window
of this study.
The percent of duplicate records that were removed varied across the variables as shown in
Appendix Table A (Column E). Compared to their counterparts, higher duplicate rates were
observed among former prisoners who were non-Hispanic of all other races, Hispanics, or had
missing race information. Males, individuals with missing education information, and those
serving terms of less than a year were also more likely to have multiple records than their
counterparts. Individuals with multiple terms in prison, those released conditionally or with
missing release type information, and those who were reimprisoned after their 2010 release
were more likely to have duplicate records compared to individuals with fewer prison terms,
those released unconditionally, and those not reimprisoned between their 2010 release and
the end of our five-year window.

39

Appendix Table A. Selected Characteristics of Persons Released from State Prisons in 2010 and PIK Rates
Released in 2010
Characteristics

Total
Hispanic origin and race
Non-Hispanic white
Non-Hispanic black
Non-Hispanic all others
Hispanic
Missing
Age at release 1
Less than 25 years
25-34 years
35-44 years
45-54 years
55-64 years
65 years and older
Missing
Sex
Male
Female
Missing
Education reported in NCRP
Less than high school
High school or GED
Some college
College or higher
Missing
Time served2
Less than 1 year
1-1.9 years
2-2.9 years
3-4.9 years
5-6.9 years
7-9.9 years
10 or more years

Total Deceased
With PIK (no
%
within 5 Years from
duplicates)
Duplicates
Release
D
Col %
E
F
Col %
478,000
100.0
5.9
18,000
100.0

Total
A
Col %
605,000 100.0

With PIK
PIK rate
B
Col % C = B / A
84.0
508,000 100.0

245,000
224,000
22,000
112,000
2,000

40.5
37.0
3.6
18.5
0.3

215,000
188,000
20,500
82,500
1,200

42.3
37.0
4.0
16.2
0.2

87.8
83.9
93.2
73.7
60.0

204,000
179,000
18,000
76,000
1,100

42.7
37.5
3.8
15.9
0.2

5.1
4.8
12.2
7.9
8.3

9,500
5,300
650
2,400
40

52.8
29.4
3.6
13.3
0.2

103,000
217,000
152,000
103,000
25,000
3,900
250

17.0
35.9
25.1
17.0
4.1
0.6
0.0

87,500
183,000
128,000
86,000
20,000
3,100
(D)

17.2
36.0
25.2
16.9
3.9
0.6
(D)

85.0
84.3
84.2
83.5
80.0
79.5
(D)

82,500
173,000
120,000
80,500
19,000
3,000
(D)

17.3
36.2
25.1
16.8
4.0
0.6
(D)

5.7
5.5
6.3
6.4
5.0
3.2
(D)

1,500
3,900
3,900
5,300
2,600
700
0

8.3
21.7
21.7
29.4
14.4
3.9
0.0

534,000
71,000
200

88.3
11.7
0.0

448,000
59,500
<15

88.2
11.7
(D)

83.9
83.8
(D)

421,000
57,000
<15

88.1
11.9
(D)

6.0
4.2
(D)

16,000
2,000
0

88.9
11.1
0.0

174,000
143,000
19,000
3,200
266,000

28.8
23.6
3.1
0.5
44.0

146,000
123,000
17,000
2,800
219,000

28.7
24.2
3.4
0.6
43.1

83.9
86.0
89.5
87.5
82.3

142,000
119,000
16,500
2,700
198,000

29.7
24.9
3.5
0.6
41.4

2.7
3.3
2.9
3.6
9.6

5,000
4,500
750
150
7,500

27.8
25.0
4.2
0.8
41.7

358,000
113,000
51,000
42,000
15,500
11,500
13,500

59.2
18.7
8.4
6.9
2.6
1.9
2.2

295,000
99,000
43,500
36,000
13,000
9,800
11,500

58.1
19.5
8.6
7.1
2.6
1.9
2.3

82.4
87.6
85.3
85.7
83.9
85.2
85.2

271,000
96,000
42,500
35,000
13,000
9,600
11,500

56.7
20.1
8.9
7.3
2.7
2.0
2.4

8.1
3.0
2.3
2.8
0.0
2.0
0.0

10,500
3,500
1,500
1,300
450
400
600

58.3
19.4
8.3
7.2
2.5
2.2
3.3

0.0
9.9
26.3

12,500
3,000
2,400

69.4
16.7
13.3

Additional terms served between
2005 and 2010
0 terms
N/A
N/A 340,000
N/A 340,000
66.9
71.1
1 term
N/A
N/A
N/A
91,000
17.9
82,000
17.2
2 or more terms
N/A
N/A
N/A
76,000
15.0
56,000
11.7
Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Note: All counts are rounded to be in compliance with the Census Bureau's disclosure avoidance policies.
1

Age at release was calculated using date of birth and release date.

2

Time served was calculated using dates of admission and release.

3

Persons who died in prison, were transferred, or escaped from prison were removed from analyses.

40

(Continued) Appendix Table A. Selected Characteristics of Persons Released from State Prisons in 2010 and PIK Rates
Released in 2010
Characteristics
Total
A

Col %

With PIK
PIK rate
B
Col % C = B / A

Total Deceased
With PIK (no
%
within 5 Years from
duplicates)
Duplicates
Release
D
Col %
E
F
Col %

Release type 3
Conditional
84.2 311,000
399,000
66.0 336,000
66.1
65.1
Unconditional
82.6 145,000
178,000
29.4 147,000
28.9
30.3
Missing
85.7
28,000
4.6
24,000
4.7
22,000
4.6
Reimprisoned after 2010 release
No
N/A
N/A 271,000
N/A 262,000
53.4
54.8
Yes
N/A
N/A 237,000
N/A 216,000
46.7
45.2
Most serious offense
Violent
84.8 116,000
145,000
24.0 123,000
24.2
24.3
Property
84.2 139,000
177,000
29.3 149,000
29.3
29.1
Drugs
82.3 128,000
164,000
27.1 135,000
26.6
26.8
Public order
84.8
112,000
18.5
95,000
18.7
89,000
18.6
Other/unspecified
88.9
3,600
0.6
3,200
0.6
3,000
0.6
Missing
82.1
3,900
0.6
3,200
0.6
3,000
0.6
Region
80.3 192,000
South
243,000
40.2 195,000
38.4
40.2
84.9
Midwest
119,000
19.7 101,000
19.9
97,000
20.3
89.5
Northeast
62,000
10.3
55,500
10.9
53,000
11.1
86.7 136,000
West
180,000
29.8 156,000
30.7
28.5
Table approved for release by the Census Bureau's Disclosure Review Board, CBDRB-FY21-123.
Source: 1983-2016 National Corrections Reporting Program, 2019 Census Numident, and Previous Census Data.
Note: All counts are rounded to be in compliance with the Census Bureau's disclosure avoidance policies.
1

Age at release was calculated using date of birth and release date.

2

Time served was calculated using dates of admission and release.

7.4
1.4
8.3

11,500
5,700
900

63.9
31.7
5.0

3.3
8.9

13,500
4,300

75.0
23.9

5.7
6.7
5.2
6.3
6.3
6.3

4,300
5,000
4,600
3,900
100
100

23.9
27.8
25.6
21.7
0.6
0.6

1.5
4.0
4.5
12.8

7,300
3,600
1,800
5,200

40.6
20.0
10.0
28.9

3

Persons who died in prison, were transferred, or escaped from prison were removed from analyses.
(D) indicates that a cell has been suppressed for disclosure avoidance purposes.

41