Jail Conditions And Mortality: Death Rates Associated With Turnover, Jail Size, And Population Characteristics

Abstract

In 2019, there were approximately ten million admissions to more than 3,000 US jails—facilities that had become increasingly deadly in the prior decades. Between 2000 and 2019, jail mortality rose by approximately 11 percent. Although incarceration is widely viewed as a health hazard, relationships between jail conditions and jail deaths are understudied. Using data from the Bureau of Justice Statistics and Reuters journalists, we assessed mortality rates and conditions in approximately 450 US jails in the period 2008–19. During those years, certain facility characteristics were related to mortality. For example, high turnover rates and high populations were associated with higher death rates. Greater proportions of non-Hispanic Black people in jail populations were associated with more deaths due to illness, and the presence of larger shares of non–US citizens was associated with lower overall mortality rates. Our findings suggest that heavy reliance on incarceration and the prevalence of broad health disparities escalate jail mortality.

There were approximately 10.3 million admissions to more than 3,000 US jails in 2019.¹ Unlike prisons, which are run by federal and state governments and hold people serving relatively long sentences after conviction, jails are locally managed, and the majority of their populations are unconvicted and awaiting trial.² In 2019, people detained pretrial made up two-thirds of the 734,500 people serving time in US jails¹ and 76 percent of jail deaths.³

Although jails are diverse in terms of size, conditions, oversight, and demographics,⁴ certain populations face a heightened risk of being detained in them. People in jails have been found to be “significantly poorer” than people outside of jails,⁵ and more than 30 percent of those who are detained pretrial remain incarcerated because they cannot afford to pay bail.⁶ As of 2019, Black people were jailed at a rate more than three times higher than White people,¹ and Black and Hispanic people were disproportionately likely to face pretrial detention.⁷

Sometimes called the “front door of the criminal justice system,”⁴ jails are widely viewed as hazards to individual and population health.⁸ People may receive some medical services in jails that could otherwise be inaccessible, but care is often resource deprived, inconsistent, or negligent.^9-12 Rates of chronic and infectious diseases among people in jails are higher than rates in the general population; behind bars, chronic diseases are often mismanaged or untreated, and contagious diseases can easily spread.^13-15 Beyond threats posed by disease, the violence that people encounter in jails can place them in physical danger and worsen symptoms of mental illness.^16,17 Furthermore, the health-related consequences of incarceration extend outward, affecting people’s postrelease lives and their home communities.^8,18-20

A particularly acute health risk—dying while incarcerated—became more commonplace during the period 2000–19, when overall jail mortality increased by approximately 11 percent, drug- or alcohol-related deaths almost quintupled, and the suicide death rate hovered at more than double the adjusted national average.³ Although mortality due to some illnesses fell, there were increases in deaths due to heart disease and cancer. Those conditions disproportionately affect older adults, who account for a significant share of the incarcerated population.^3,21,22

News reports, lawsuits, and other sources offer disturbing insights about deaths among people in custody,²³ and scholars have underscored the imperative of “establishing transparency about death in jail.”^24(p24) A study published in 2019 maintains that the prevalence of avoidable jail-based deaths, such as suicides, reflect “failures inside the jail system.”^24(p17) Other research reveals that jail mortality is higher than prison mortality and that although the reported in-jail mortality rate is lower than that of the adjusted general population, life expectancy is reduced for those exposed to incarceration.^25-28 Studies also focus on particular causes of death in jails; how jail authorities assess and report cause of death; and tactics that may reduce the risk for certain types of deaths, including those due to suicide or drug overdose.^29-32 Other work has examined mortality in specific institutions, focusing, for example, on why cause-specific death rates change over time and how they vary among demographic groups.^33-35

Research on the relationship between jail characteristics and jail mortality across facilities is limited. A 2012 article maintained that environmental factors such as poor health care, unsanitary living conditions, and high levels of violence make jails “increasingly like a death sentence.”³⁶ Seven years later, a scholar noted that although it is widely acknowledged that “bad outcomes like disease or death often result from the interaction of environmental and personal risk factors,”^24(p20) data on jail mortality fail to offer “an apples-to-apples comparison” of circumstances in multiple facilities.^24(p18)

To address this issue, we examined how a variety of jail characteristics, including turnover rate, population size, and demographics, relate to mortality in facilities nationwide.

Study Data And Methods

JAIL DATA

Our study used jail mortality data made available by Reuters journalists.^37,38 Hereafter, we refer to this data set as “Reuters data.” The journalists filed public records requests for information on jail deaths during the period 2008–19 in all jails or jail systems with an average of 750 or more incarcerated people, plus the ten largest jails in each state, regardless of size. To our knowledge, no other jail-level mortality data covering these years are publicly accessible. From the Reuters data, which include 523 jails or jail systems, we used jail-level total deaths, deaths by cause, jail average daily population, and type of medical provider.

We supplemented the Reuters data with additional jail-level characteristics from the Bureau of Justice Statistics, which periodically conducts a Census of Jails among all local jails in the US except those in combined jail and prison systems in six states (Alaska, Connecticut, Delaware, Hawaii, Rhode Island, and Vermont).³⁹ In the years between census collections, the Bureau of Justice Statistics conducts an Annual Survey of Jails, which samples a proportion of jails.⁴⁰ We extracted jail annual admissions and releases, rated capacity (the maximum number of beds or incarcerated people assigned by a rating official to the facility), number of confined and non-confined people, and demographics of incarcerated people from 2013 and 2019 Census of Jails data and 2008–18 Annual Survey of Jails data.

The process of matching the Reuters data with Bureau of Justice Statistics data was complicated by two issues: Observation units differ across the three data sets, and they lack a common numeric identifier that could be used as a matching key. After undertaking a fuzzy matching process relying on county and jail names, we matched 470 of 523 jails from Reuters data with Census of Jails or Annual Survey of Jails data. (See the online appendix for details about the matching process.)⁴¹

ANALYSIS

Combining the Reuters and Bureau of Justice Statistics data from the period 2008–19, we conducted multivariable regression analyses to explain the variation in jail mortality over time and across facilities. Our outcome of interest was the annual mortality rate. Because jail populations constantly move into and out of facilities, finding an appropriate denominator to calculate the mortality rate was difficult.³³ We followed the Bureau of Justice Statistics definition,³ calculating it as total deaths per jail average daily population. We modeled jail deaths using a negative binomial model, which is a preferred model for overdispersed count variables. We included logarithm of average daily population as an offset variable in the model to represent mortality rates.

In our models of jail mortality rates, we considered a range of characteristics related to jails and incarcerated people based on the two data sources. We generated a weekly turnover rate of incarcerated population to control for the flow of people in and out of jails. Following the Bureau of Justice Statistics definition,¹ we calculated the weekly turnover rate using the sum of weekly admissions and releases divided by average daily population. We also generated the percentage of capacity occupied as a measure of jail fullness. The latter was calculated by taking the number of incarcerated people (measured by confined population) and dividing it by the rated capacity, as used in Bureau of Justice Statistics reports.¹

To distinguish the effect of jail size on mortality, we placed jail-year observations into three groups according to tertiles of average daily populations. Jails (in certain years) with the lowest, medium, and largest average daily populations are referred to as small, medium, and large jails (with 0–364, 365–825, and 826 or more people, respectively). We also distinguished whether a jail’s medical provider was public or otherwise (private or a combination of private and public). Characteristics of incarcerated people include percentages of adult females, juveniles (children under age eighteen), non-Hispanic Black people, Hispanic people, people of other races, those detained pretrial, and non–US citizens. The regressions also included state fixed effects for any unobserved time-invariant state-specific effects and year fixed effects for any common time trends that may affect all facilities.

In addition to total mortality rates, we examined mortality by cause of death. Regressions were conducted separately for five types of death: illness-related or natural death, suicide, acute drug- or alcohol-related death, accident, and homicide.

Last, we examined whether the associations between jail characteristics and mortality differed by jail population size. We reexamined the associations among jails with small, medium, and large populations. For all regression analyses, the standard errors were clustered at county level.

The analyses were conducted using Stata, version 17.0.

LIMITATIONS

This study had several limitations. First, as the Reuters data only cover jails with 750 or more incarcerated people plus the ten largest jails in each state, findings might not be applicable to smaller jails. Second, inconsistent naming of facilities in the Reuters and Bureau of Justice Statistics data challenged the matching process. We took several steps to exclude ambiguous matches and managed to match 470 of 523 (or 90 percent) of jails in the Reuters data with Bureau of Justice Statistics information (see the appendix).⁴¹ Still, because we lacked an exact matching key for the jails in the data sets, matching facilities by name could have led to some imprecision. Third, some information available in the Reuters data (for example, demographics of the deceased person) and the Bureau of Justice Statistics data (for example, staff counts) was not used in the study because of considerable shares of missing values. Fourth, health and mortality data for people who are incarcerated or in police custody have been shown to be “incomplete…incorrect… [and] anachronistic,”⁴² and jail data may underestimate deaths or contain inaccuracies related to causes of death.⁴³ Finally, the associations found in the study do not suggest causality.

Study Results

Exhibit 1 presents summary statistics of jail mortality rates and jail characteristics during the period 2008–19 based on observations used in the jail mortality rate regression. On average, about 1.40 deaths occurred annually per 1,000 incarcerated people. About 0.62 deaths per 1,000 incarcerated people were illness-related deaths, 0.53 were suicides, and 0.15 were acute drug- or alcohol-related deaths. Accidents or homicides were less common (0.03 deaths each). The average daily population was close to 1,000 people per facility. The average weekly turnover rate was about 67 percent, and the capacity occupied was more than 90 percent. The majority of adults in jails were men; 14 percent were women. Almost 1 percent of incarcerated people were juveniles. Non-Hispanic White people accounted for 46 percent of the jail population, followed by people who were non-Hispanic Black (30 percent), those who were Hispanic (13 percent), and those of other races (11 percent).

EXHIBIT 1.

Mortality rates, conditions, and inmate demographics in approximately 450 US jails, 2008–19

	No. of jail-year observationsa	Variable mean	SD
Mortality rate (per 1,000 inmates)
All-cause	3,267	1.40	2.06
Illness	3,167	0.62	1.11
Suicide	3,167	0.53	1.47
Acute drug- or alcohol-related	3,146	0.15	0.75
Accident	3,137	0.03	0.37
Homicide	3,140	0.03	0.20
Other	3,134	0.03	0.33
Average daily population	3,267	994	1,399
Weekly jail turnover rate, %	3,267	66.71	70.85
Percent of capacity occupied, %	3,267	92.47	63.21
Inmate demographics, %
Adult female	3,267	13.79	5.81
Juvenile	3,267	0.71	2.12
Non-Hispanic White	3,267	45.87	24.27
Non-Hispanic Black	3,267	29.86	23.91
Hispanic	3,267	12.88	17.16
Other races	3,267	11.38	25.35
Detained pretrial	3,267	60.83	27.20
Non–US citizen	3,267	5.41	10.48
Public medical provider, %	3,267	38.63

SOURCE Authors’analysis of Reuters data and Bureau of Justice Statistics data. NOTES The summary statistics are based on observations used in the total death regression. Mortality rates, average daily population, and public medical providers are from Reuters data; all other variables are from Bureau of Justice Statistics Census of Jails and Annual Survey of Jails data.

^aNumber of available jail-year observations for each variable.

More than 60 percent of people in observed jails in this period were detained pretrial. Non–US citizens accounted for about 5 percent. Among all jail-year observations, about 39 percent of facilities had a public medical provider. The others had either a private medical provider or a mix of private and public providers.

Exhibit 2 shows the associations between jail characteristics and mortality rates. Jail turnover rates were positively associated with mortality rates. A ten-percentage-point increase in the turnover rate, for example, was associated with about 0.01 more deaths per 1,000 inmates. The percentage of capacity occupied, however, was negatively related to the mortality rate, although the association was significant at only the 10 percent level. Compared with small jails, medium or large jails were associated with higher mortality rates. The presence of higher percentages of adult females was also related to higher death rates. Higher shares of non–US citizens were associated with lower mortality rates. No significant association was found between other factors and mortality rates.

EXHIBIT 2.

Factors associated with mortality rates in approximately 450 US jails, 2008–19

	Estimates	Average marginal effects
Weekly jail turnover rate	0.081**	0.113**
Percent of capacity occupied	−0.033*	−0.046*
Medium jail (365–824 inmates)	0.185*	0.234*
Large jail (825 inmates or more)	0.213**	0.274**
Adult female	1.269*	1.779*
Juvenile	0.845	1.185
Non-Hispanic Black	0.267	0.375
Hispanic	−0.032	−0.045
Other races	0.076	0.107
Detained pretrial	−0.025	−0.036
Non–US citizen	−0.898***	−1.259***
Public medical provider	−0.056	−0.079

SOURCE Authors’ analysis of Reuters data and Bureau of Justice Statistics data. NOTES n = 3,267 jail-year observations. The numbers reported are coefficient estimates, average marginal effects from a multivariable negative binomial regression. The average marginal effect for factor levels is the discrete change from the base level. Independent variables also include state and year fixed effects.

^*p < 0.10

^**p < 0.05

^***p < 0.01

Exhibit 3 presents associations by cause of death. More fluid jail populations, measured by higher turnover rates, were associated with higher mortality rates for suicide, drug- or alcohol-related death, and homicide, but not for illness-related death or accident. Percentage of capacity occupied was not significantly associated with most types of deaths, with the exception of suicide. However, the latter association was only marginally significant. Large and medium jail sizes were associated with higher illness-related death rates. Medium jail size was also associated with higher drug- or alcohol-related death rates. Higher shares of juveniles were strongly associated with higher rates of death due to accidents. Among different racial and ethnic groups, higher proportions of incarcerated non-Hispanic Black people showed a strong positive association with illness-related death rates. In addition, higher populations of people who were non-Hispanic Black or other races were associated with higher homicide rates, and higher populations of people who were Hispanic were associated with higher death rates due to accident, although both were only significant at the 10 percent level. Higher percentages of people who were detained pretrial were associated with lower drug- or alcohol-related death rates (at the 10 percent significance level). The presence of higher shares of non–US citizens was associated with lower death rates due to illness, suicide, and drug- or alcohol-related death. Public medical provider oversight was negatively associated with suicide rates.

EXHIBIT 3.

Factors associated with mortality rates in approximately 450 US jails, by cause of death, 2008–19

	All-cause (n = 3,267)	Illness-related (n = 3,167)	Suicide (n = 3,167)	Drug- or alcohol-related (n = 3,146)	Accident (n = 3,137)	Homicide (n = 3,140)
Weekly jail turnover rate	0.081**	0.026	0.124**	0.110***	−0.152	0.184**
Percent of capacity occupied	−0.033*	0.003	−0.066*	−0.068	−0.036	−0.126
Medium jail (365–824 inmates)	0.185*	0.452***	−0.157	0.574**	−0.252	0.257
Large jail (825 inmates or more)	0.213**	0.549***	−0.156	0.291	−0.041	0.566
Adult female	1.269*	1.055	1.670	1.599	1.317	0.925
Juvenile	0.845	0.825	0.874	−6.973	21.359***	2.155
Non-Hispanic Black	0.267	0.650***	−0.404	−0.317	0.778	1.763*
Hispanic	−0.032	0.195	−0.204	−0.327	1.426*	0.739
Other races	0.076	0.178	−0.178	0.118	−0.050	1.192*
Detained pretrial	−0.025	−0.034	0.208	−0.511*	−0.680	−0.238
Non-US citizen	−0.898***	−1.078***	−0.916*	−1.398**	0.097	−1.142
Public medical provider	−0.056	−0.059	−0.171**	0.073	0.321	0.210

SOURCE Authors’ analysis of Reuters data and Bureau of Justice Statistics data. NOTES Sample sizes are jail-year observations. The numbers reported are coefficient estimates from multivariable negative binomial regressions. Independent variables also include state and year fixed effects.

^*p < 0.10

^**p < 0.05

^***p < 0.01

Exhibit 4 presents results by tertiles of average daily population.Within each size tertile, higher turnover rates were associated with higher death rates. Percentage capacity occupied did not show any significant association with death rates regardless of jail size. In large jails, higher shares of adult females, non-Hispanic Black people, and people of other races were associated with higher death rates. No such pattern was found in smaller jails. In fact, higher shares of Black people and people of other races were associated with reduced mortality rates in small jails. Higher shares of non–US citizens were associated with lower mortality rates in medium and large jails but with higher mortality rates in small jails.

EXHIBIT 4.

Factors associated with mortality rates in approximately 450 US jails, by jail population size, 2008–19

	Tertiles of average daily population
	Small (n = 772)	Medium (n = 1,198)	Large (n = 1,347)
Weekly jail turnover rate	0.178*	0.055*	0.108**
Percent capacity occupied	−0.268	0.076	−0.038
Adult female	0.341	1.320	1.509*
Juvenile	−9.214	0.014	0.581
Non-Hispanic Black	−1.603*	0.522	0.566***
Hispanic	0.759	−0.467	0.184
Other races	−0.864**	−0.112	0.281**
Detained pretrial	0.131	0.196	−0.073
Non–US citizen	1.110**	−0.916*	−1.037***
Public medical provider	−0.023	0.032	−0.035

SOURCE Authors’ analysis of Reuters data and Bureau of Justice Statistics data. NOTES Small jails were those with 0–364 inmates; medium jails were those with 365–824 inmates, and large jails were those with 825 inmates or more. Sample sizes are jail-year observations. The numbers reported are coefficient estimates from multivariable negative binomial regressions. Independent variables also include state and year fixed effects.

^*p < 0.10

^**p < 0.05

^***p < 0.01

Discussion

Our results show that certain jail conditions relate to mortality rates.

TURNOVER RATE

We found that jails with higher turnover rates were likely to have higher overall mortality than jails with lower turnover. Deaths due to suicide, drugs and alcohol, and homicide showed a significant association with high turnover. The average length of a jail stay in 2019 was twenty-six days.¹ During the period 2000–19, the median time served before a drug- or alcohol-related death was one day, whereas it was nine days before a suicide death and thirty days before a homicide death.³

Connections between turnover and mortality reflect not just an acute risk for death early in confinement but also, more broadly, the dangers of heavy reliance on jails, including the practice of detaining people with mental illness and substance dependence.⁴ Suicide was the leading cause of death for people in jails in 2019,³ and jail suicide rates are more than double both the adjusted national average and the suicide rate in prisons.⁴⁴ A recent study found that people who were homeless, incarcerated for the first time, in psychological distress, or experiencing substance dependence were more likely than others to threaten or attempt suicide or self-harm in jail.⁴⁴ Regardless of preincarceration status, researchers note, individuals are “degraded through every step of the street-to-jail transition,”^44(p3) and jail suicides can result in part from people’s “shock of confinement” and the loss of a “basic sense of normalcy.”⁴⁵ As prior studies suggest and our results show, cycling higher numbers of people into and out of a jail relates to more fatalities.

POPULATION SIZE

Length of incarceration is important to consider alongside our finding that the most populous jails were associated with higher death rates overall, as well as higher rates of mortality due to illness. According to Bureau of Justice Statistics data, length-of-stay increases with jail size. In jails with an average daily population of 50–99 people, the average length-of-stay in 2019 was 18.7 days, as opposed to 29.0 days in jails with 1,000–2,499 people.¹ It follows that illness-related deaths, which the Bureau of Justice Statistics reports often occur after a relatively long period of incarceration (a median of twenty-five days in the case of heart disease, for example, and 138 days in the case of cancer),³ are associated with the largest jails.

Medium-size jails were associated not just with higher illness-related death rates than smaller jails but also with higher rates of drug- or alcohol-related deaths than either smaller or larger jails. It is estimated that just 34 percent of jails provide detoxification services, and even fewer offer maintenance pharmacotherapy.⁴⁶ Future research should examine whether the availability of addiction treatment varies by jail size;^47,48 whether the smallest and largest jails may be particularly prone to undercounting drug- or alcohol-related deaths;³¹ and whether people in jails of certain sizes are better able to access contraband drugs, which could increase the risk for overdose death.⁴⁹

DEMOGRAPHICS

We found that the presence of high proportions of Black people in facilities was associated with higher rates of death due to illness. A similar pattern regarding illness-related deaths could be found in Bureau of Justice Statistics–reported jail mortality rates from the period 2000–19: 69 percent of deaths among Black people in jail were due to illness, versus 42 percent of deaths among White people.³ Although that discrepancy could be due to a variety of factors, including the prevalence of health disparities in broader society,^50,51 an examination of our results alongside Bureau of Justice Statistics data suggests that research is needed on race-based differences in access to care in jails. Our finding that higher proportions of Black people are associated with an increased death rate in large jails could be due to relationships between cause of death and jail size. Black incarcerated people are more likely than others to die because of illness,³ and we found that large jails are strongly associated with higher rates of illness-related deaths.

Our study showed that a higher share of adult females in jail populations was marginally significantly associated with more deaths overall. These results highlight that some facilities lack resources to offer women adequate services¹¹ and that incarcerated women are more likely than incarcerated men to report ever having had a chronic condition.¹³ They also underscore that an increase in the population of women in jails—11 percent between 2000 and 2019—has distinctly health-related implications.¹

According to our results, the presence of a higher proportion of juveniles was associated with more deaths due to accidents. In the past two decades, the number of young people detained in adult jails and prisons has dropped dramatically,⁵² but as of 2019, approximately 2,900 youth were held in jails intended for adults.^30,53 Our findings build on literature showing that incarcerated young people face particular threats related to their health and well-being.⁵⁴

We found the presence of higher proportions of non–US citizens in facilities to be associated with lower death rates. These results could relate to the so-called healthy immigrant effect—the notion that soon after arrival in the United States, many immigrants are healthier than native-born citizens.⁵⁵ But at least two caveats are necessary. First, information about individual deaths of non–US citizens in custody, such as age, length of time in the US, and disease burden, is lacking, and each of those factors can relate to the healthy immigrant effect. Second, incarcerated non–US citizens should not be viewed as a favored subgroup. In relatively small jails, the presence of greater shares of non–US citizens was associated with higher mortality. And according to recent examinations of mortality data from Immigration and Customs Enforcement (ICE) detention centers, deaths occurred in young men with low burdens of preexisting disease, and most fatalities involved violations of “Performance-Based National Detention Standards.”⁵⁶

Also, although our study period predates the onset of COVID-19, it is worth noting that the pandemic led to dramatic increases in deaths in many carceral settings, including ICE facilities.^57,58

HEALTH CARE MANAGEMENT

Our results indicate that health care in jail that is overseen by a public provider, as opposed to a private provider or a hybrid of the two, is related to lower mortality due to suicide. These findings add nuance to the limited previous research examining the relationship between health care management and jail mortality, which has obtained disparate results. An article by the Reuters journalists who gathered the data we used notes that the number of jails contracting with private companies to oversee care increased from approximately 50 percent in 2010 to 62 percent in 2018. During 2016–18, the authors report, jails that hired any of the five “leading jail healthcare contractors” had death rates exceeding those of facilities where care was managed by government agencies.⁵⁹ A subsequent study analyzed jail mortality rates during the period 2008–19 and found “little evidence that switching from publicly to privately provided jail health care increases inmate mortality.”⁶⁰ Further research is needed on how oversight of health services affects health outcomes, including mortality.

Conclusion

We conducted what we believe to be the first national-level study of associations between jail conditions and jail mortality and found that death rates were related to certain facility characteristics: turnover rates, jail size, the proportion of particular demographic groups in the incarcerated population, and health care management.

Well-founded indictments of inhumane conditions and overcrowding in jails and prisons have long been marshaled to justify calls for more carceral capacity, but our findings highlight the need for a different approach. In terms of mortality risk, an underlying cause is a heavy reliance on incarceration: the cycling of people into and out of jails where the impacts of addiction, mental illness, and health inequity can be exacerbated with dire consequences. In light of this, policy makers should resist devoting more funding to incarceration. Instead, they should curtail pretrial detention and provide access to noncarceral community-based social supports, such as mental health and addiction treatment.^61-64