Suicide mortality among individuals in federal prisons compared with the general population: a retrospective cohort study in the USA from 2009 to 2020

Introduction

In 2022, there were nearly 2 million individuals incarcerated in the US prisons and jails.1 Prior to incarceration, these individuals have a high burden of severe stressors including substance use, unemployment and homelessness, all contributing to a disproportionate burden of poor mental health.2 Once incarcerated, carceral facilities often lack comprehensive mental health treatment: only 54% of those meeting criteria for serious psychological distress in prison received treatment from 2011 to 2012.3 Incarceration separates people from their families and communities, and carceral facilities are stressful and unpredictable living environments that remove individuals’ control over their surroundings.4 Time spent incarcerated may therefore create and exacerbate pre-existing mental health issues.5 Furthermore, post release, individuals are often not connected to mental health treatment and face many stressors (eg, housing instability, unemployment or underemployment) that further exacerbate mental health issues and prevent them from achieving stability, thus placing them at a higher risk of reincarceration.6 7 This myriad of problems results in individuals cycling back into prisons and increases the risk of self-harm and suicides behind prison walls—which has long been one of the leading causes of death and the leading cause of unnatural death in US prisons.8

However, with the high degree of surveillance faced by individuals in prisons, suicides should not be possible. For example, prison officials monitor individuals every 30 min if they have expressed suicidality and monitor the general prison population five times per day.9 10 Prisons also have mandatory suicide-prevention programmes and policies that are meant to mitigate self-injurious behaviour.11

While suicides are among the leading causes of deaths in prisons, the US Department of Justice Bureau of Justice Statistics (BJS) reporting is incomplete, lags by approximately 2 years and aggregates data from state and federal prisons.8 The BJS reports also lack demographic information (eg, age) of those dying by suicide, which has prevented a more rigorous analysis comparing the suicide rates of the general and incarcerated populations. During the COVID-19 pandemic, especially in 2020, individuals were subjected to solitary confinement for disease containment, were unable to have visitors and were largely unable to practice social distancing or access personal protective equipment.12 13 However, the BJS released its final report in 2019 and no information is available on suicide deaths during the COVID-19 pandemic.8

We began to address this research gap by comparing the age-standardised suicide rate among those incarcerated to the general population from 2009 to 2020. While prior research has compared crude suicide rates across both populations, it has not accounted for the different age distributions of these populations, which can lead to incorrect conclusions.14 Moreover, prior research has not included data from the COVID-19 pandemic, expected to impact suicide mortality incidence both in the prison and general populations.8 Our objective was to compare the reported suicide rate among those incarcerated in federal prisons to the general population, accounting for age distributions, before and during the COVID-19 pandemic (2009–2020).

Methods

We conducted a retrospective cohort study among those incarcerated in the Federal Bureau of Prisons (BOP) from 2009 to 2020. We estimated age-standardised mortality ratios (SMR) to compare the observed suicide mortality among incarcerated individuals with expected suicide mortality had they experienced the same age-specific suicide rates as the general population. We chose age-standardisation because suicide rates vary by age and the prison population has a substantially younger age distribution than the general population. We also tested for linear trends over calendar year in both the incarcerated and general populations.

Results

BOP death records indicate that for the BOP population, there were 5215 deaths from 2009 to 2020; 245 (5%) of deaths were reported as suicide (data not shown). The age distribution of those incarcerated is younger than in the general population. Specifically, in 2020, those aged 25–54 years made up 78% of the incarcerated population but only 50% of the general population, and those 65 years and older made up 4% of the incarcerated population and 22% of the general population.

Over the study period, among those incarcerated, suicide rates were highest for those in the oldest age cohorts—65 years and older followed by those ages 60–64 years (table 1). This is in contrast with the general population, for whom suicide rates were highest for those slightly younger, ages 50–54 years. Suicide rates were lowest in both cohorts for individuals under the age of 35 years though those in the general population under the age of 35 years had higher suicide rates than the incarcerated population under age 35 years.

The crude suicide death rate in the incarcerated population was relatively unstable due to small counts early in the time period (table 2 and figure 1). The crude suicide death rate in the incarcerated population varied from 10.57 per 100 000 person years in 2009 to 13.13 per 100 000 person years in 2019 before increasing sharply to 19.01 per 100 000 person years in 2020. This amounted to an average increase of 0.79 suicide deaths per 100 000 per year from 2009 to 2020 (table 3). The crude suicide rate in the general population also increased during this time, from 15.41 to 17.26 per 100 000 (table 2), representing an average increase of 0.23 suicide deaths per 100 000 per year (table 3). This corresponded to a difference in trend of 0.57 suicide deaths per 100 000 per year for the incarcerated population relative to the general population (95% CI: 0.16 to 0.98, table 3). Notably, from 2019 to 2020, there was a 45% increase in the crude suicide rate in the incarcerated population compared with a 4% decrease in the crude suicide rate in the general US population.

Request permission

After standardising the age distribution in the general population to reflect the age distribution in the prison population, the expected suicide rate increased from 14.86 in 2009 to 17.71 in 2020 (table 2). This represented an average increase in the expected suicide rate of 0.30 suicide deaths per 100 000 per year from 2009 to 2020 (table 3). The difference in trend between the observed suicide rate vs expected suicide rate was 0.49 suicide deaths per 100 000 per year (95% CI: 0.08 to 0.90, table 3).

Comparing the observed suicide mortality rate with the age-standardised expected suicide mortality, the SMR remained below 1.00 until 2020, when it jumped to 1.07 (95% CI: 0.74 to 1.57) in 2020 (table 2). This means that in 2020, we observed 1.07 times the number of suicide deaths in the prison population than we would have expected if the prison population had the same age-stratified rates of suicide deaths as the general population. This corresponds to 27 suicide deaths in 2020, the most in any single year in the prior decade. SMR and CIs were unchanged in sensitivity analyses that subtracted the population in federal prisons from the general population (data not shown).

Discussion

The overall trend of reported suicide mortality in federal prisons has been increasing since 2009 by an average of 0.79 suicide deaths per 100 000 person years per year, although trends fluctuated year to year. This increase in suicide mortality in federal prisons was more pronounced than in the general population. The suicide mortality rate in federal prisons surpassed that of the general population in 2020. 2020 was thus the first year that we saw an excess number of suicide deaths among the incarcerated population compared with what would be expected if they had experienced the same age-specific suicide death rates as the general population. The increasing suicide mortality in prisons is a public health crisis. Part of this increase may be explained by the exacerbation of mental health stressors during the COVID-19 pandemic. For example, quarantine18 and medical isolation19 were used in ways functionally similar to solitary confinement—internationally recognised as torture,20 individuals’ feared contracting COVID-19 in this congregate setting, prison visits were discontinued, and mental health services were reduced.21

Prior to the COVID-19 pandemic, we found that the observed suicide mortality rate was lower than the expected age-standardised rate had the individuals living in Federal BOP experienced the same age-specific suicide rates as the general population. While this finding seems contradictory to the fact that incarcerated persons have a higher burden of mental health issues than the general population and live in a traumatising environment, this finding may be explained by several factors.3 First, incarcerated persons, and particularly those who are identified as at increased risk of suicide, are under high surveillance9 10 and lack access to firearms (the most common mechanism for suicide completion in the US general population).22 Second, there is known under-reporting of suicide deaths by carceral agencies; this under-reporting is likely greater than the under-reporting of suicide deaths in the general population.23 In 2020, investigative reporting found that 30 individuals incarcerated in Georgia prisons died by suicide while the Georgia Department of Corrections reported none.24 Regardless of these limitations, the observed suicide rates in the incarcerated population are high and increasing and warrant attention and intervention.25

Carceral agencies have recognised this crisis and responded by advocating for more robust mental health treatment in prison.26 However, the traumatising prison environments are inappropriate for such treatment. By placing individuals in prisons, society removes them from communities that could have otherwise supported them. Incarcerated individuals often face different forms of violence and stressors, including overcrowding, unpredictable living conditions, potential time in solitary confinement and physical violence from both carceral staff and fellow prisoners.4 27 Thus, the harms of the prison environment outweigh the potential benefits of in-prison mental health programming.28 Further, these harms persist despite long-term programming to provide mental health services and mitigate suicidal and self-injurious behaviour in the BOP and there is longstanding evidence that incarceration harms mental health.2 11 The response to this mental health crisis should therefore consider decarceration, including necessary treatment (eg, substance use treatment; mental health treatment) and support (eg, housing) for individuals within their communities.28 29 It is important to note that the suicide rate postrelease from incarceration is quite high, particularly soon after release.30 31 While suicides occurring postrelease are not captured in the rate for the currently incarcerated population, time spent while incarcerated is likely a strong risk factor in these cases. It is thus critical that decarceration is coupled with individual and community-level support.

We must also address the issue of who owns these data and who is responsible for these analyses. Not only is BJS reporting incomplete and delayed, their mortality in custody reporting ended in 2021, meaning that its 2021 report on 2019 data will be its last.32 Furthermore, while state and federal carceral agencies are required to report deaths in custody by the Death in Custody Reporting Act, most fail to do so, including the federal system.33 There is a need for transparent legislation that forces these systems to accurately report, in real-time, deaths and causes of death occurring in their custody. There is also a need for an independent national archive that collates these data and makes them available to the public. Such legislation and an independent data archive would improve reporting quality, public data access and transparency. There is evidence that reports from carceral agencies are systematically undercounting suicides—deaths that carceral agencies play a key role in mitigating. News sources including the Atlanta Journal-Constitution have documented far more suicides in prisons than carceral agencies report through searching death certificates and additional records.24 A limitation of our analysis is, therefore, that it is likely an undercount of suicides in federal custody. Additionally, our analyses used age distributions in federal and state prisons due to data availability, and this age distribution may look different in federal prisons alone.

Our study was descriptive in nature and there is a need for future work with more comprehensive data that would allow us to produce causal estimates. While age is an important confounder, age distributions were reported for state and federal facilities combined rather than federal facilities alone, which may differ. Further, using age alone is insufficient for knowing what the suicide rate would be if individuals currently incarcerated were released to the general population. It is also well-known that the sex and race distributions in prison populations differ substantially from the distributions in the general population (eg, Black men make up 13% of the US population but 35% of the incarcerated population34 ; sexual and gender minority populations have higher suicide mortality rates than others35 and are overrepresented in the incarcerated population36). Additionally, we lack data after 2020 from the prison population, preventing us from more fully understanding the full impact of the COVID-19 pandemic on prison suicides and from assessing if the sharp increase in suicides in 2020 was sustained.