This cohort study estimates the risk of suicide among US adults returning to the community after release from incarceration.
Key Points
Question
What proportion of US adults who died by suicide spent at least 1 night in jail shortly before their death?
Findings
In this cohort modeling study involving nearly 7.1 million US adults released from incarceration in 2019, nearly 20% of suicides occurred among those who were released from jail in the past year and 7% were by those in their second year of jail release.
Meaning
Findings of this study suggest that focused suicide prevention efforts could reach a substantial number of adults who were formerly incarcerated within 2 years, when death by suicide is likely to occur.
Abstract
Importance
Although people released from jail have an elevated suicide risk, the potentially large proportion of this population in all adult suicides is unknown.
Objective
To estimate what percentage of adults who died by suicide within 1 year or 2 years after jail release could be reached if the jail release triggered community suicide risk screening and prevention efforts.
Design, Setting, and Participants
This cohort modeling study used estimates from meta-analyses and jail census counts instead of unit record data. The cohort included all adults who were released from US jails in 2019. Data analysis and calculations were performed between June 2021 and February 2024.
Main Outcomes and Measures
The outcomes were percentage of total adult suicides within years 1 and 2 after jail release and associated crude mortality rates (CMRs), standardized mortality ratios (SMRs), and relative risks (RRs) of suicide in incarcerated vs not recently incarcerated adults. Taylor expansion formulas were used to calculate the variances of CMRs, SMRs, and other ratios. Random-effects restricted maximum likelihood meta-analyses were used to estimate suicide SMRs in postrelease years 1 and 2 from 10 jurisdictions. Alternate estimate was computed using the ratio of suicides after release to suicides while incarcerated.
Results
Included in the analysis were 2019 estimates for 7 091 897 adults (2.8% of US adult population; 76.7% males and 23.3% females) who were released from incarceration at least once, typically after brief pretrial stays. The RR of suicide was 8.95 (95% CI, 7.21-10.69) within 1 year after jail release and 6.98 (95% CI, 4.21-9.76) across 2 years after release. A total of 27.2% (95% CI, 18.0%-41.7%) of all adult suicide deaths occurred in formerly incarcerated individuals within 2 years of jail release, and 19.9% (95% CI, 16.2%-24.1%) of all adult suicides occurred within 1 year of release (males: 23.3% [95% CI, 20.8%-25.6%]; females: 24.0% [95% CI, 19.7%-36.8%]). The alternate method yielded slightly larger estimates. Another 0.8% of adult suicide deaths occurred during jail stays.
Conclusions and Relevance
This cohort modeling study found that adults who were released from incarceration at least once make up a large, concentrated population at greatly elevated risk for death by suicide; therefore, suicide prevention efforts focused on return to the community after jail release could reach many adults within 1 to 2 years of jail release, when suicide is likely to occur. Health systems could develop infrastructure to identify these high-risk adults and provide community-based suicide screening and prevention.
Introduction
Unlike most high-income countries, the US has a 2-tiered incarceration system. All people in state and federal prisons already received a trial and conviction, primarily of felonies, and are serving their sentence. In contrast, jails, which are operated by local governments, primarily detain unconvicted people awaiting arraignment or held over for trial. Just 35% of people released from jail serve sentenced time.1 Those sentenced to jail receive short sentences (<1 year), primarily for low-level felonies or violating terms of sentenced supervision in the community. Individuals who are unsentenced often spend only a few days in jail. In 2019, US jails had 10 322 570 admissions, with many people admitted multiple times, and a mean daily census of 734 470.2 The weekly turnover rate was 53%.1 Rapid return from jail to the community is the norm.
People in jail have a high annualized suicide rate (48 per 100 000 people) and a standardized mortality ratio (SMR) of 2.2 compared with demographically similar US adults.3 Growing global literature suggests an even higher suicide rate in the year after release from incarceration; meaning that more suicides per day occur after release than during jail time.4,5 Only 1 study with high-quality death registration, however, is specific to US jails: among Philadelphia Medicaid beneficiaries who died by suicide between 2003 and 2018, 25% had a history of Philadelphia county jail incarceration.6
In the US, no data exist on suicide risk during an actionable period of 1 to 2 years after jail release. This cohort modeling study aimed to estimate the percentage of adults who died by suicide within the year or 2 years after jail release who could have been reached if the jail release triggered community suicide risk screening and prevention efforts. It applied meta-analytic and epidemiologic modeling to published cohort studies of suicide after release from prison or incarceration in the US or abroad. It also examined the size of the adult jail population and modeled the risk of suicide within 1 to 2 years after a jail stay in this population compared with adults with no recent incarceration and the portion of adult suicides during that vulnerable period. This study built models from publicly available aggregate counts; it used no unit record data.
Methods
In accordance with the Common Rule, this cohort study was exempt from ethics review and informed consent requirement because it was not considered human participant research. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Data Sources and Calculations
Jail Counts
Every year, federal reports provide the mean daily population in US jails and prisons2 and annual counts of admissions.1 We computed mean length of jail stay by dividing annual jail admissions1 by 365 times the mean daily jail census.2
Many people discharged from jail are readmitted in the same year. Given that arrest counts have been virtually identical to jail admission counts,7 dividing annual admissions by annual arrests per person arrested approximates the number of people jailed annually. The National Survey on Drug Use and Health variable NoBookYr2 reported annual arrests per person arrested and booked during the past 12 months from 2015 to 2018.8
Suicide Risk and Counts
Federal reports count suicide deaths that occur, separately, in jail and in prison.3 Multiplying the death counts by 100 000 divided by the mean daily population yields a crude mortality rate (CMR) per 100 000 person-years by setting.
Federal reports calculate the suicide CMR for cohorts of US adults matched to the jail and prison populations by age, sex, and race and ethnicity. Dividing the suicide CMR for those in jails or prisons by the demographically matched suicide CMR for US adults with no recent incarceration yields the federally reported SMRs by setting. Meanwhile, dividing the suicide CMR for those in jails or prisons by the unadjusted national rate for adults not incarcerated in the past year yields a relative risk (RR) of suicide death.
As a proxy for suicide RRs after jail release, we used RRs after prison release, with emphasis on studies from countries that house unconvicted people in prison. To identify linked data on deaths of people discharged from incarceration, we started with 5 systematic reviews identifying studies published through October 2021.4,5,9,10,11 We updated from there by scanning all studies that cited the systematic reviews from 2021 onward; the 191 summaries from a Google Scholar search from 2021 through January 2024 using the terms suicide, linkage, (incarcerat* or jail or prison), post-release, and (death or mortality); and first-screen hits when we used Google Scholar or PubMed to retrieve each relevant study.
The systematic reviews focused on lifetime rather than acute suicide risk. Seeking acute risk data and recent studies, 1 of us (T.R.M.) read all studies in the systematic reviews and the 23 potentially relevant studies that we identified. One of us (T.R.M.) also verified the systematic review coding and the quality of study CMR or SMR computations.
Included studies had to provide a demographically matched SMR or the data to compute an SMR using the Centers for Disease Control and Prevention’s Web-Based Injury Statistics Query and Reporting System (WISQARS) and at least 2 of the following: a CMR, a suicide count, and person-years of exposure. We included only 1 study per jurisdiction, giving preference to studies that provided CMRs at year 1 or year 2 after release, with next priority going to studies that covered the shortest duration beyond 2 years. We excluded studies involving juveniles, subpopulations, or fewer than 40 suicide deaths over 2 years after release. For supplemental analyses, we also sought CMRs per person-year during incarceration.
Statistical Analysis
Using Excel version 2401 (Microsoft Corp), we applied published formulas for the variance of CMRs and SMRs and Taylor expansion formulas for the variance of other ratios,12,13 and then we ran random-effects restricted maximum likelihood (RE-ML) meta-analyses using RStudio 4.3.2 (RStudio Team). We built 10 jurisdictional estimates from 8 adult incarceration studies14,15,16,17,18,19,20,21 identified in the 2023 meta-analysis by Janca et al5 plus 3 studies we identified.22,23,24 All studies were dominated by decades-old data. All except 3 studies14,20,24 gave partial information on suicide timing in the 2 years after release. We used ratios of suicide CMRs in postrelease year 1 and year 2 and narrow assumptions or calculations (eTable 1 in Supplement 1) to adjust study data from diverse time periods to estimates in postrelease year 1 and year 2. We calculated the CMR of demographically similar people without recent incarceration for 1 study24 (eTable 2 in Supplement 1).
We developed estimated counts of suicide deaths after incarceration from the meta-analysis and 4 sensitivity analysis sets of studies with different strengths and limitations. For each set, we multiplied the postrelease SMR by the 2019 US CMR for nonincarcerated adults to estimate a postrelease CMR for the US. The number of postrelease suicides equaled the released person count times the CMR. The study sets comprised (1) our primary estimate; (2) a study from Ontario, Canada17; (3) the SMR for releases in England and Wales23; (4) a consolidated estimate from separate computations by sex; and (5) the meta-analysis by Janca et al.5
First, we estimated suicide SMRs in postrelease years 1 and 2 using meta-analytic RE-ML pooling of SMRs from 10 jurisdictions. Variance for CMRs and SMRs were computed using standard formulas.12 Six of these high-quality studies16,17,18,19,20,21,22 included unconvicted individuals, 4 were from North America,14,17,23,24 and all were adjusted to cover equal periods.
Second, a study by Kouyoumdjian et al17 linked deaths to incarceration releases in 2000 throughout the Canadian province of Ontario and provided a detailed time track for suicide deaths in the year after release and a mean rate for the following 2 years. Its North American data matched US jail data on the percentage of females and on 63% vs 65% of nondeported individuals who were released unconvicted.
Third, the SMR from 1 year of tracking data for all people who were released from incarceration in England and Wales during 2000 to 2002, of which 35% were individuals who were unconvicted or released to the community on sentencing.18 We corrected the published SMR for both sexes combined, which erroneously had been computed by averaging community CMRs for nonincarcerated persons by sex rather than by weighting them to match the sex distribution of the incarcerated population. The unconvicted persons were reasonably well represented, and tracking was for exactly 1 year.
Fourth, we developed a consolidated estimate from separate computations for males and females using SMR estimates for postrelease year 1 using RE-ML meta-analytic pooling of data from 6 adult incarceration studies. We matched this estimate to the US release distribution by sex. Five studies16,18,20,21,22 included people who were released unconvicted, all studies were adjusted to cover 1 year, but only 1 study23 was from North America.
Fifth, the 2023 meta-analysis of 11 studies by Janca et al5 provided an independently developed, peer-reviewed SMR estimate. The meta-analysis gave preference to SMRs over more years rather than 1 to 2 years, diluting the higher risk soon after release; thus we considered its estimates weak. The between-study heterogeneity was high for duration and population, with juvenile detention and studies of males alone included. Two studies had SMR calculation errors.18,25
To calculate the suicide CMR among adults who were not incarcerated, we subtracted the number of suicide deaths during and after a jail or prison stay from the WISQARS adult suicide count and the number of adults incarcerated from its population count.26 The CMR equaled 100 000 times suicide count divided by population count.
We developed another estimate using a distinctly different ratio method (hereafter referred to as ratio method) and meta-analysis of the RR of suicide death in the year after release vs while incarcerated. For example, in a UK study by Phillips and Roberts,27 the RR was 2.55 (CMR of 212 in the year after release divided by CMR of 83 per year incarcerated). Except with the Phillips and Roberts27 study, we used online counts of inmates and suicide deaths during incarceration3 to compute the ratios. We generated 8-study RE-ML meta-analytic RR estimates using this approach for postrelease year 1 and year 2. We multiplied the mean RRs by the 2019 US CMRs while in jail and in prison3 to estimate the postrelease CMRs. The final estimates were independent of the suicide rate in the general population. Half of the estimates were from North America. The CMR estimates while incarcerated, however, were not always definitive or perfectly matched. Data analysis and calculations were performed between June 2021 and February 2024.
Results
This cohort modeling study included 2019 estimates for 7 091 897 adults (2.8% of US adult population; 76.7% males and 23.3% females) who were released from incarceration at least once, typically after brief pretrial stays. Table 1 and Table 2 describe the 10 studies that we analyzed and the meta-analytic consolidation of their data, including the CMRs and SMRs in year 1 (Table 1) and in the first 2 years combined (Table 2) after release from incarceration as well as the risk of suicide after release vs during incarceration (eTables 3 and 4 in Supplement 1 provide corresponding SEs and 95% CIs). For example, the 10 studies linked mortality data from more than 1 million people for 1 year after incarceration, finding 1409 suicide deaths. Among the 10 studies, the CMR range in the year after incarceration was 42.51 to 361.37 per 100 000 person-years (Table 1); across the 2 years after release, the CMR range was narrower: 42.75 to 242.93 per 100 000 person-years (Table 2).
Table 1. Meta-Analytic Data on Suicide in Year 1 After Incarceration Release .
| Source | Place of incarceration | No. of people released | Release dates | Periods reported | CMR per 100 000 person-years, postrelease year 1 | SMR, postrelease year 1 | No. of suicides, postrelease year 1 | Person-years at risk, postrelease year 1 |
|---|---|---|---|---|---|---|---|---|
| Binswanger et al,14 2007 | Washington, US | 30 327 | July 1999-December 2003 | 14 d, 2 y | 94.05 | 4.57 | 27 | 28 525 |
| Cunningham et al,22 2022a | New Zealand | 87 894 | 1998-2016 | 28 d, 1 y, 2 y, 9.4 y | 361.37 | 16.04 | 315 | 87 169 |
| Fitch et al,23 2024b | North Carolina, US | 474 892 | 2000-February 2020 | 28 d, 1 y, 2 y, 5.7 y | 42.51 | 1.94 | 181 | 425 818 |
| Haglund et al,15 2014 | Sweden | 26 985 | 2005-2009 | 28 d, 2 y | 235.65 | 21.02 | 78 | 33 100 |
| Kariminia et al,16 2007 | New South Wales, Australia | 85 196 | 1988-2002 | 28 d, 6 m, 7.7 y | 203.63 | 6.48 | 229 | 112 335 |
| Kouyoumdjian et al,17 2016 | Ontario, Canada | 46 442 | 2000 | 28 d, 1 y, 2 y | 99.40 | 8.43 | 46 | 46 277 |
| McNeeley et al,24 2023 | Minnesota, US | 36 716 | 2010-2019 | 4.9 y | 112.26 | 4.88 | 41 | 36 466 |
| Pratt et al,18 2006; corrected | England and Wales, UK | 245 029 | 2000-2002 | 1 y | 155.93 | 8.92 | 382 | 245 029 |
| Stewart et al,20 2004 | Western Australia, Australia | 9381 | 1994-1999 | 3.4 y | 291.78 | 7.17 | 27 | 9245 |
| Van Dooren et al,21 2013 | Queensland, Australia | 42 015 | 1994-2007 | 1 y, 7.5 y | 216.67 | 12.00 | 84 | 38 769 |
| 10 Studies combinedc | NA | 1 054 550 | NA | NA | 177.64 | 8.95 | 1409 | 1 062 733 |
Abbreviations: CMR, crude mortality rate; NA, not applicable; SMR, standardized mortality ratio.
Randomly rounded counts for years 1 and 2 were provided by Ruth Cunningham (email communication to Dr Miller, January 25, 2024).
Counts for years 1 and 2 were provided by Kate Vinita Fitch (email communication to Dr Miller, February 8, 2024).
Studies were combined using restricted maximum likelihood random-effects meta-analysis. The Figure and eTable 3 in Supplement 1 show SEs and 95% CIs for CMRs, SMRs, and RRs.
Table 2. Meta-Analytic Data on Total Suicides in Years 1 and 2 After Incarceration Release .
| Source | Place of incarceration | % Of inmates unconvicted | CMR per 100 000 person-years, during incarceration | Release dates | RR (CMR after release divided by CMR in custody) (95% CI) | CMR per 100 000 person-years, combined years 1 and 2 | SMR, combined years 1 and 2 | No. of suicides, combined years 1 and 2 | Person-years at risk, combined years 1 and 2 | |
|---|---|---|---|---|---|---|---|---|---|---|
| Year 1 | Year 2 | |||||||||
| Phillips and Roberts,27 2019 | UK | NA | 83.00 | Fiscal year 2018-2019 | 2.55 (2.51 to 2.60) | 1.98 (1.94 to 2.02) | 164.42a | NA | 237 | 144 340 |
| Binswanger et al,14 2007 | Washington, US | 0 | 12.81 | July 1999-December 2003 | 7.34 (–23.00 to 37.68) | 5.46 (–10.63 to 21.55) | 70.00 | 3.40 | 40 | 57 049 |
| Cunningham et al,22 2022b | New Zealand | 47.2 | 83.73 | 1998-2016 | 4.32 (3.70 to 4.93) | 2.90 (2.64 to 3.17) | 242.93 | 10.42 | 423 | 174 124 |
| Fitch et al,23 2024c | North Carolina, US | 0 | 9.80 | 2000-February 2020 | 4.34 (2.69 to 5.99) | 4.36 (2.79 to 5.94) | 42.75 | 1.93 | 327 | 764 900 |
| Haglund et al,15 2014 | Sweden | 0 | NA | 2005-2009 | NA | NA | 204.00 | 18.20 | 127 | 48 157 |
| Kariminia et al,16 2007 | New South Wales, Australia | 19.9 | 125.22 | 1988-2002 | 1.63 (1.57 to 1.69) | 1.32 (1.28 to 1.35) | 164.88 | 5.25 | 324 | 196 651 |
| Kouyoumdjian et al,17 2016 | Ontario, Canada | 62.7 | 48.65 | 2000 | 2.04 (1.50 to 2.58) | 1.38 (1.16 to 1.61) | 67.23 | 5.70 | 62 | 92 223 |
| McNeeley et al,24 2023 | Minnesota, US | NA | 12.76 | 2010-2019 | 8.80 (–7.54 to 25.14) | 6.59 (–0.67 to 13.85) | 84.13 | 3.66 | 61 | 72 558 |
| Pratt et al,18 2006; corrected | England and Wales, UK | 35.0 | 82.78 | 2000-2002 | 1.88 (1.83 to 1.94) | 1.51 (1.48 to 1.53) | 124.66 | 7.13 | 611 | 489 976 |
| Stewart et al,20 2004 | Western Australia, Australia | 15.8 | NA | 1994-1999 | NA | NA | 236.11 | 5.80 | 44 | 18 762 |
| Spittal et al,19 2014 and Van Dooren et al,21 2013 | Queensland, Australia | 19.5 | NA | 1994-2007 | NA | NA | 173.01 | 9.60 | 128 | 73 985 |
| 10 Studies combinedd | NA | NA | NA | NA | NA | NA | 138.52 | 6.85 | 2147 | 1 994 641 |
| 8 Studies combinedd | NA | NA | NA | NA | 2.70 (1.78 to 3.31) | 2.12 (1.38 to 2.87) | NA | NA | NA | NA |
Abbreviations: CMR, crude mortality rate; NA, not applicable; RR, relative risk; SMR, standardized mortality ratio.
The CMR in year 1 was 212, based on 153 suicides and 72 170 person-years at risk.
Randomly rounded counts for years 1 and 2 were provided by Ruth Cunningham (email communication to Dr Miller, January 25, 2024).
Counts for years 1 and 2 were provided by Kate Vinita Fitch (email communication to Dr Miller, February 8, 2024).
Studies were combined using restricted maximum likelihood random-effects meta-analysis. The Figure and eTables 3 and 4 in Supplement 1 show SEs and 95% CIs for CMRs, SMRs, and RRs.
One study27 was included only for the RR analysis. The Figure shows the associated meta-analytic forest plots. For example, the SMR during the first year after release was 8.95 (95% CI, 5.54-12.36), with the study with the lowest SMR23 receiving the largest weight (10.43%) and the study with the highest SMR15 receiving the lowest weight (8.73%).
Figure. Meta-Analytic Estimates of Standardized Mortality Ratios (SMRs) for Suicide After Incarceration Release and Crude Mortality Rate (CMR) Ratios After vs During Incarceration.
Error bars represent 95% CIs; vertical dashed line at 0 represents the line of no effect; square size represents the study weight; and diamond represents the overall estimate from the meta-analysis, with the width representing the 95% CI of the point estimate.
Jail Stay and Release
From 2015 to 2018, data from the National Survey on Drug Use and Health indicated that people who were arrested had a mean number of 1.456 arrests annually, with 73.9% arrested only once; 10% of these people were rearrested the next year.23 In 2019, 7 091 897 US adults (2.8% of adult population) were released from jail, with another 734 470 in jail on an average day.2 The mean jail stay (including both convicted and unconvicted individuals) was 24.7 days.
The best meta-analytic estimate (Table 3) showed that 9121 (95% CI, 7440-11 034) people died by suicide during the year after jail release (sensitivity analysis range, 8706-10 470 people; ratio method estimate, 9255 [95% CI, 6114-12 396] people). In 2019, 355 people died by suicide in US jails (Table 4).2
Table 3. Rate, Relative Risk (RR), and Percentage of Suicides in Adults Released from Incarceration or Not Recently Incarcerated, 2019.
| Data Source and calculation method | No. of adult suicides | Suicide rate per 100 000 adults | RR of adult suicidesa | % Of adult suicides (95% CI)b | % Of adult population |
|---|---|---|---|---|---|
| Postrelease year 1, estimate from 10-study meta-analysis; best estimate | |||||
| Year 1 after jail release | 9121 | 128.61 | 8.95 | 19.9 (16.2-24.1) | 2.8 |
| Year 1 after prison release | 777 | 128.61 | 8.95 | 1.7 (1.4-2.0) | 0.2 |
| Age ≥18 y, no recent jail or prison | 35 273 | 14.38 | 1.00 | 76.9 (73.0-80.9) | 96.1 |
| Kouyoumdjian et al,17 2016 (Ontario, Canada) | |||||
| Year 1 after jail release | 8706 | 122.76 | 8.43 | 19.0 (13.5-24.5) | 2.8 |
| Year 1 after prison release | 741 | 122.76 | 8.43 | 1.6 (1.1-2.1) | 0.2 |
| Age ≥18 y, no recent jail or prison | 35 723 | 14.56 | 1.00 | 77.9 (71.9-83.9) | 96.1 |
| Pratt et al,18 2006, corrected (England and Wales, UK) | |||||
| Year 1 after jail releasec | 9098 | 128.29 | 8.92 | 19.8 (18.1-22.1) | 2.8 |
| Year 1 after prison releasec | 775 | 128.29 | 8.92 | 1.7 (1.5-1.9) | 0.2 |
| Age ≥18 y, no recent jail or prison | 35 297 | 14.39 | 1.00 | 77.0 (75.2-78.9) | 96.1 |
| 6-study meta-analysis, calculated by sex | |||||
| Year 1 after jail release | 10 740 | 151.44 | 11.09 | 23.4 (15.1-31.7) | 2.8 |
| Year 1 after prison release | 925 | 153.21 | 11.22 | 2.0 (1.3-2.7) | 0.2 |
| Age ≥18 y, no recent jail or prison | 33 505 | 13.66 | 1.00 | 73.1 (64.7-82.0) | 96.1 |
| 11-study meta-analysis by Janca et al,5 2023; weak estimate not based on risk at year 1 | |||||
| Year 1 after jail release | 7842 | 110.57 | 7.40 | 17.1 (12.2-21.2) | 2.8 |
| Year 1 after prison release | 668 | 110.57 | 7.40 | 1.5 (1.0-1.8) | 0.2 |
| Age ≥18 y, no recent jail or prison | 36 661 | 14.94 | 1.00 | 79.9 (76.2-85.3) | 96.1 |
| Ratio method, year 1 estimate from 8-study meta-analysis | |||||
| Year 1 after jail release | 9255 | 130.50 | 9.09 | 20.2 (13.3-27.0) | 2.8 |
| Year 1 after prison release | 708 | 117.31 | 8.18 | 1.5 (1.0-2.1) | 0.2 |
| Age ≥18 y, no recent jail or prison | 35 207 | 14.35 | 1.00 | 76.8 (70.0-84.1) | 96.1 |
| Postrelease year 2, estimate from 10-study meta-analysis; best estimate | |||||
| Year 2 after jail release | 12 483 | 92.64 | 6.98 | 27.2 (18.0-41.7) | 5.3 |
| Year 2 after prison release | 1063 | 92.64 | 6.98 | 2.3 (1.5-3.6) | 0.4 |
| Age ≥18 y, no recent jail or prison | 31 624 | 13.26 | 1.00 | 69.0 (53.8-78.9) | 93.4 |
| Ratio method, year 2 estimate from 8-study meta-analysis | |||||
| Year 2 after jail release | 13 821 | 102.57 | 8.07 | 30.1 (19.6-40.7) | 5.3 |
| Year 2 after prison release | 1058 | 92.20 | 7.26 | 2.3 (1.5-0.0) | 0.4 |
| Age ≥18 y, no recent jail or prison | 30 292 | 12.70 | 1.00 | 66.0 (58.5-77.4) | 93.4 |
The RR was computed using the variable age 18 years or older, no recent jail or prison as the denominator.
Percentage of adult suicides was computed from No. of adult suicides.
Included any readmission stays.
Table 4. Population, Suicides, Crude Mortality Rate (CMR), and Relative Risk (RR) of Suicide in US Adults by Sex, 2019a.
| Group | No. of adults | No. of adult suicides | CMR, % | RR of adult suicidesb | % Of adult suicides (95% CI)c | % Of adult populationc |
|---|---|---|---|---|---|---|
| Total population | ||||||
| Year in jail, based on daily census | 734 470 | 355 | 48.33 | 3.32 | 0.8 (0.8-0.8) | 0.3 |
| Year after jail release | 7 091 897 | 10 740 | 151.44 | 11.09 | 23.4 (15.1-31.7) | 2.8 |
| Year in state prison custody | 1 311 100 | 311 | 23.72 | 1.63 | 0.7 (0.7-0.7) | 0.5 |
| Year in federal prison custody | 147 000 | 29 | 19.73 | 1.35 | 0.1 (0.1-0.1) | 0.1 |
| Year after prison release | 603 844 | 925 | 153.21 | 11.22 | 2.0 (1.3-2.7) | 0.2 |
| Age ≥18 y, no recent jail or prison | 245 352 967 | 33 168 | 13.52 | 1.00 | 72.3 (70.5-73.9) | 96.1 |
| Age ≥18 y | 255 241 278 | 45 865 | 17.97 | 1.23 | 100 (100-100) | 100 |
| Females only | ||||||
| Year in jail, based on daily census | 110 735 | 42 | 38.11 | 6.91 | 0.4 (0.4-0.4) | 0.1 |
| Year after jail release | 1 652 253 | 2342 | 141.73 | 25.13 | 24.0 (19.7-36.8) | 1.3 |
| Year in prison custody | 135 735 | 17 | 55.00 | 9.78 | 0.2 (0.2-0.2) | 0.1 |
| Year after prison release | 56 213 | 80 | 141.73 | 25.13 | 0.8 (0.7-1.3) | 0.04 |
| Age ≥18 y, no recent jail or prison | 128 922 632 | 7251 | 5.62 | 1.00 | 74.4 (61.3-79.0) | 98.5 |
| Age ≥18 y | 130 877 568 | 9752 | 7.45 | 1.35 | 100 (100-100) | 100 |
| Males only | ||||||
| Year in jail, based on daily census | 623 735 | 313 | 50.15 | 2.23 | 0.9 (0.9-0.9) | 0.5 |
| Year after jail release | 5 439 644 | 8398 | 154.39 | 6.85 | 23.3 (20.8-25.6) | 4.4 |
| Year in prison custody | 1 322 335 | 323 | 24.45 | 1.09 | 0.9 (0.9-0.9) | 1.1 |
| Year after prison release | 547 631 | 846 | 154.39 | 6.85 | 2.3 (2.1-2.6) | 0.4 |
| Age ≥18 y, no recent jail or prison | 116 430 365 | 26 071 | 22.39 | 1.00 | 72.2 (61.3-79.0) | 93.6 |
| Age ≥18 y | 124 363 710 | 36 113 | 29.04 | 1.29 | 100 (100-100) | 100 |
Abbreviations: CMR, crude mortality rate; RR, relative risk.
Year after jail and prison release based on meta-analytic estimates from 6 studies.16,18,20,21,22,23 Data on people in jail were from Zeng and Minton2; data on people in prison (adjusted to include people in privately operated federal prisons) were from Substance Abuse and Mental Health Services Administration8 and Jones and Maynard9; and data on US population by age group were from Kinner et al.10
The RR was computed using the variable age 18 years or older, no recent jail or prison as the denominator.
Percentages of adult suicides and adults were computed from No. of adult suicides and No. of adults, respectively.
The RRs of suicide among US adults released from incarceration compared with those with no recent incarceration was 8.95 (95% CI, 7.21-10.69) in year 1 after jail release into the community (sensitivity analysis range, 8.43-11.09; ratio method estimate, 9.09 [95% CI, 5.48-13.47]) and 6.98 (95% CI, 4.21-9.76; ratio method estimate, 8.07 [95% CI, 5.25-10.90]) across 2 years after release (Table 3). Table 4 shows mean RRs of 25.13 (95% CI, 11.91-38.34) for females and 6.85 (95% CI, 3.05-10.65) for males. Although elevated, the 3.32 RR of suicide death in jail was lower than after release (Table 4).
People who were released from jail within the year accounted for an estimated 19.9% (95% CI, 16.2%-24.1%) of adult suicide deaths but only 2.8% of the adult population (Table 3) (sensitivity analysis range, 19.0%-23.4%; ratio method estimate, 20.2% [95% CI, 13.3%-27.0%]). The percentages for males and females were roughly equal (23.3% [95% CI, 20.8%-25.6%] and 24.0% [95% CI, 19.7%-36.8%]). The suicide RR was higher for females, but fewer females than males are incarcerated in jails.
Although suicide RR was highest in postrelease year 1, it was also high in postrelease year 2, when another 7.3% of all US adult suicides occurred. The SMR analysis indicated that 27.2% (95% CI, 18.0%-41.7%) of all US adult suicides occurred within 2 years after jail release (ratio method estimate, 30.1% [95% CI, 19.6%-40.7%]). Those in jail accounted for another 0.8% of adult suicides.
Suicide Deaths in Prison and After Prison
People in or released from prison accounted for 2.3% (95% CI, 2.1%-2.6%) of suicide deaths. The suicide RR was 1.63 in state prison and 1.35 in federal prison (Table 4). The postrelease suicide SMR applied to persons released from both prison and jail. Since almost 12 times more people were released from jail (7 091 897) than from prison (603 844) annually, 1.7% (95% CI, 1.4%-2.0%) of adult suicide deaths occurred among people within the year of their prison release (sensitivity analysis range, 1.5%-2.0%); this group represented 0.2% of the US adult population in 2019 (Table 3). Adding the 19.9% estimate after jail release (Table 3), an estimated 21.6% of adult suicides (sensitivity analysis range, 20.6%-25.4%) occurred among the 3% of US adults within postrelease year 1 from jail or prison, with another 0.8% and 0.7% occurring during jail and prison incarceration, respectively (Table 4).
Discussion
Suicide prevention efforts should focus on people who have spent at least 1 night in jail in the past year. Two distinct methods and datasets both indicate that during the year after jail release, an individual’s suicide CMR was 9 times the mean among other US adults, representing 19.9% of adult suicides in 2019, with 27.2% of suicides occurring within 2 years after release. Consistent with the high percentage of jail involvement among US adults who die by suicide, 35% of males and 13% of females who died by suicide in Denmark had gone to court for criminal charges28 and 25% of suicide decedents in Philadelphia had been in jail.6
The results suggest that better integration of suicide risk detection and prevention across health and criminal justice systems (including 911 calls, police contacts, pretrial jail detention, criminal courts, jail sentences, probation, and parole) is critical to advancing population-level suicide prevention efforts. However, high volumes of jail admissions and discharges, short jail stays, and understaffing mean that the 3119 local jails2 in the US generally have limited capacity to coordinate care with outside health agencies.29 Individuals are often arrested when experiencing acute mental health crisis (ie, manic or psychotic), arrests are typically unanticipated, jail stays are usually brief, and releases for individuals in pretrial jail detention are typically unscheduled. Community health systems are often unaware that their subscriber or patient was in jail and may drop them for missing appointments. The suicide rate after the return to the community after jail stay is higher than the suicide rate in jail, but local jails have limited capacity to coordinate postrelease health activities. Thus, a comprehensive approach to reducing the population-level US suicide rate would include health systems screening their subscribers or patients for recent arrest or police involvement and reaching out to those recently released to prevent suicide.30,31,32 In addition, broader use of 1115 Medicaid waivers to keep short incarcerations from disrupting Medicaid coverage could help to ensure that released individuals are insured, reducing barriers to health system reengagement.
It is now possible for health systems to link jail release data (which are typically publicly available) against their patient or plan participant list regularly and to systematize supportive outreach to patients who were recently released. Biomedical informatics advances have made searching text-based data from multiple jails in an area feasible and efficient. For example, CareSource, a large nonprofit managed care organization headquartered in Ohio, developed a data system that surveils booking and release from local jails across the US, identifies subscribers interacting with jails, and leverages that data for improved care coordination with community health care practitioners after release.30 This approach provides a potential solution to identifying a high-risk population for suicide that may not be well connected to standard care; the recently funded National Center for Health and Justice Integration for Suicide Prevention (NCHATS) is evaluating it.
Critical infrastructure problems have historically hindered evidence-based suicide prevention for recently incarcerated individuals. Justice settings often lack resources for needed suicide prevention interventions. Moreover, person-to-person rather than automated handoffs to health systems are currently not feasible at scale. The NCHATS is demonstrating that health systems can automate ethical, Privacy Rule–compliant identification of individuals being released from jails to share data with contracted community practitioners and provide outreach and care connection, solving this long-standing problem.30
Better epidemiologic studies are needed about people who pass through jail. The Bureau of Justice Statistics should identify the demographic distribution of people who are admitted or discharged and should test for demographic biases in length of jail stay. Deaths in a representative cohort also should be tracked for 2 years to validate the estimates in this study.
Limitations
This analysis is subject to the limitations of the studies that we examined. Only 3 studies included death data beyond 2009. When meta-analyzing SMRs or CMR ratios, however, older data are problematic only if suicide patterns between groups diverge over time. After prison release, among the included 10 studies, suicide CMRs per 100 000 person-years ranged widely from 42.51 to 361.37. All studies ignored deaths that were coded as undetermined intent or unintentional drug overdoses, which were actually suicides or deaths of people who were ambivalent about dying; suicidality is common among drug overdose deaths.33,34 Available data on jail inmates were for the mean daily jail census. The daily census may not be representative of discharges; in large US jails, 5% of discharged individuals with stays longer than 180 days accounted for 41% of the daily census, whereas 62% of individuals discharged in less than a week accounted for 4%.35 Additionally, females made up 23.3% of jail admissions but only 14.7% of the daily census.35 In Ontario, unconvicted people or people released to community supervision on conviction accounted for 62.7% of persons released but 48.8% of the daily census.17 England and Wales18 as well as New Zealand22 experienced similar differentials. Burgeoning equity-driven reductions in cash bail are greatly decreasing the mean jail stay where implemented, potentially affecting suicide rates in jail and after release. We could not calculate CMRs by race and ethnicity because that information on people who were released from jail was unavailable and because reported US suicide CMRs by race and ethnicity are biased.36 Despite these limitations, the study findings were consistent across 2 methods and datasets and across 5 RR estimates with contrasting strengths and weaknesses.
Conclusions
Adults released from jail make up a large, concentrated population at extreme risk for death by suicide. In this study, the nearly 7.1 million US adults released from jail in 2019 accounted for 1 in 5 adult suicides. Thus, identifying adults with recent jail releases can help with community suicide prevention efforts. Jails lack the staffing to manage an effective response to critical behavioral health needs after release. Health systems, including federally qualified health centers, are increasingly positioned to manage care coordination during high-risk transitions from jail to the community and should be recognized as important partners in building the cross-sector infrastructure necessary for identifying high-risk adults and providing comprehensive community-based suicide screening and prevention.
eTable 1. Multipliers and Calculations Used to Infer Suicide Counts in Years 1 and 2 After Release From Incarceration vs Counts for Other Periods
eTable 2. Calculation of Crude Mortality Rate (CMR) for the Population Aged 18-68 in Minnesota Demographically Matched to Its Population of People Released From Prison in 2010-2019
eTable 3. Uncertainty Around Suicide Risk Estimates Over the Year and 2 Years After Release From Incarceration Compared to People Without Recent Incarceration
eTable 4. Uncertainty Around Suicide Risk Estimates During Incarceration and Over the Year and 2 Years After Release From Incarceration Compared to People Who Are Incarcerated
Data Sharing Statement
References
- 1.Zeng Z, Minton T. Jail inmates in 2019. (NCJ 255608). March 2021. Accessed March 31, 2024. https://bjs.ojp.gov/library/publications/jail-inmates-2019
- 2.Zeng Z, Minton T. Census of jails, 2005-2019: statistical tables (NCJ 255406). October 2021. Accessed March 31, 2024. https://www.ojp.gov/library/publications/census-jails-2005-2019-statistical-tables
- 3.Carson EA. Mortality in local jails, 2000-2019 – statistical tables (NCJ 301368). December 2021. Accessed March 31, 2024. https://bjs.ojp.gov/library/publications/mortality-local-jails-2000-2019-statistical-tables
- 4.Zlodre J, Fazel S. All-cause and external mortality in released prisoners: systematic review and meta-analysis. Am J Public Health. 2012;102(12):e67-e75. doi: 10.2105/AJPH.2012.300764 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Janca E, Keen C, Willoughby M, et al. Sex differences in suicide, suicidal ideation, and self-harm after release from incarceration: a systematic review and meta-analysis. Soc Psychiatry Psychiatr Epidemiol. 2023;58(3):355-371. doi: 10.1007/s00127-022-02390-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Candon M, Fox K, Jager-Hyman S, et al. Building an integrated data infrastructure to examine the spectrum of suicide risk factors in Philadelphia Medicaid. Adm Policy Ment Health. 2023;50(6):999-1009. doi: 10.1007/s10488-023-01299-2 [DOI] [PubMed] [Google Scholar]
- 7.Neusteter SR, Subramanian R, Trone J, Khogali M, Reed C. Gatekeepers: the role of police in ending mass incarceration. 2019. Accessed February 27, 2024. https://www.vera.org/downloads/publications/gatekeepers-police-and-mass-incarceration.pdf
- 8.Substance Abuse and Mental Health Services Administration . NSDUH: 4-year RDAS (2015-2018). Accessed June 2021. https://rdas.samhsa.gov/#/survey/NSDUH-2015-2018-RD04YR?column=STNAME&filter=STNAME%3DALABAMA&results_received=true&row=HERPNRYR&run_chisq=false&weight=DASWT_2
- 9.Jones D, Maynard A. Suicide in recently released prisoners: a systematic review. Mental Health Practice. 2013;17(3):20-27. doi: 10.7748/mhp2013.11.17.3.20.e846 [DOI] [Google Scholar]
- 10.Kinner SA, Forsyth S, Williams G. Systematic review of record linkage studies of mortality in ex-prisoners: why (good) methods matter. Addiction. 2013;108(1):38-49. doi: 10.1111/add.12010 [DOI] [PubMed] [Google Scholar]
- 11.Borschmann R, Tibble H, Spittal MJ, et al. The Mortality After Release From Incarceration Consortium (MARIC): protocol for a multi-national, individual participant data meta-analysis. Int J Popul Data Sci. 2020;5(1):1145. doi: 10.23889/ijpds.v5i1.1145 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Smyrke N, Dunn N, Murley C, Mason D. Standardized mortality ratios in multiple sclerosis: systematic review with meta-analysis. Acta Neurol Scand. 2022;145(3):360-370. doi: 10.1111/ane.13559 [DOI] [PubMed] [Google Scholar]
- 13.Petitti D. Meta-Analysis, Decision Analysis and Cost-Effectiveness Analysis: Methods for Quantitative Synthesis in Medicine. Oxford University Press; 1994. [Google Scholar]
- 14.Binswanger IA, Stern MF, Deyo RA, et al. Release from prison–a high risk of death for former inmates. N Engl J Med. 2007;356(2):157-165. doi: 10.1056/NEJMsa064115 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Haglund A, Tidemalm D, Jokinen J, et al. Suicide after release from prison: a population-based cohort study from Sweden. J Clin Psychiatry. 2014;75(10):1047-1053. doi: 10.4088/JCP.13m08967 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kariminia A, Law MG, Butler TG, et al. Suicide risk among recently released prisoners in New South Wales, Australia. Med J Aust. 2007;187(7):387-390. doi: 10.5694/j.1326-5377.2007.tb01307.x [DOI] [PubMed] [Google Scholar]
- 17.Kouyoumdjian FG, Kiefer L, Wobeser W, Gonzalez A, Hwang SW. Mortality over 12 years of follow-up in people admitted to provincial custody in Ontario: a retrospective cohort study. CMAJ Open. 2016;4(2):E153-E161. doi: 10.9778/cmajo.20150098 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Pratt D, Piper M, Appleby L, Webb R, Shaw J. Suicide in recently released prisoners: a population-based cohort study. Lancet. 2006;368(9530):119-123. doi: 10.1016/S0140-6736(06)69002-8 [DOI] [PubMed] [Google Scholar]
- 19.Spittal MJ, Forsyth S, Pirkis J, Alati R, Kinner SA. Suicide in adults released from prison in Queensland, Australia: a cohort study. J Epidemiol Community Health. 2014;68(10):993-998. doi: 10.1136/jech-2014-204295 [DOI] [PubMed] [Google Scholar]
- 20.Stewart LM, Henderson CJ, Hobbs MS, Ridout SC, Knuiman MW. Risk of death in prisoners after release from jail. Aust N Z J Public Health. 2004;28(1):32-36. doi: 10.1111/j.1467-842X.2004.tb00629.x [DOI] [PubMed] [Google Scholar]
- 21.Van Dooren K, Kinner SA, Forsyth S. Risk of death for young ex-prisoners in the year following release from adult prison. Aust N Z J Public Health. 2013;37(4):377-382. doi: 10.1111/1753-6405.12087 [DOI] [PubMed] [Google Scholar]
- 22.Cunningham R, King PT, Telfer K, et al. Mortality after release from incarceration in New Zealand by gender: a national record linkage study. SSM Popul Health. 2022;20:101274. doi: 10.1016/j.ssmph.2022.101274 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Fitch KV, Pence BW, Rosen DL, et al. Suicide mortality in formerly incarcerated people compared to the general population in North Carolina, 2000–2020. Am J Epidemiol. 2024;193(3):489-499. doi: 10.1093/aje/kwad214 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.McNeeley S, Clark VA, Duwe G. Risk factors for all-cause, natural, and unnatural deaths among individuals released from state prison. J Crim Justice. 2023;86:102059. doi: 10.1016/j.jcrimjus.2023.102059 [DOI] [Google Scholar]
- 25.Jones M, Kearney GD, Xu X, Norwood T, Proescholdbell SK. Mortality rates and cause of death among former prison inmates in North Carolina. N C Med J. 2017;78(4):223-229. doi: 10.18043/ncm.78.4.223 [DOI] [PubMed] [Google Scholar]
- 26.Centers for Disease Control and Prevention, National Center for Injury Prevention and Control . Fatal injury reports, national, regional and state, 1981–2019. WISQARS—Web-based Injury Statistics Query and Reporting System. 2020. Accessed June 2021. https://webappa.cdc.gov/sasweb/ncipc/mortrate.html
- 27.Phillips J, Roberts R. Deaths of People Following Release From Prison. INQUEST Charitable Trust; 2019. [Google Scholar]
- 28.Webb RT, Qin P, Stevens H, Mortensen PB, Appleby L, Shaw J. National study of suicide in all people with a criminal justice history. Arch Gen Psychiatry. 2011;68(6):591-599. doi: 10.1001/archgenpsychiatry.2011.7 [DOI] [PubMed] [Google Scholar]
- 29.Scheyett A, Vaughn J, Taylor MF. Screening and access to services for individuals with serious mental illnesses in jails. Community Ment Health J. 2009;45(6):439-446. doi: 10.1007/s10597-009-9204-9 [DOI] [PubMed] [Google Scholar]
- 30.Arias SA, Sperber K, Jones R, et al. Managed care updates of subscriber jail release to prompt community suicide prevention: clinical trial protocol. BMC Health Serv Res. 2023;23(1):1265. doi: 10.1186/s12913-023-10249-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Johnson JE, Jones R, Miller T, et al. Study protocol: a randomized controlled trial of suicide risk reduction in the year following jail release (the SPIRIT Trial). Contemp Clin Trials. 2020;94:106003. doi: 10.1016/j.cct.2020.106003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Skopp NA, Smolenski DJ, Bush NE, et al. Caring contacts for suicide prevention: a systematic review and meta-analysis. Psychol Serv. 2023;20(1):74-83. doi: 10.1037/ser0000645 [DOI] [PubMed] [Google Scholar]
- 33.Rockett IR, Smith GS, Caine ED, et al. Confronting death from drug self-intoxication (DDSI): prevention through a better definition. Am J Public Health. 2014;104(12):e49-e55. doi: 10.2105/AJPH.2014.302244 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Rockett IRH, Caine ED, Connery HS, et al. Discerning suicide in drug intoxication deaths: paucity and primacy of suicide notes and psychiatric history. PloS One. 2018;13(1):e0190200. doi: 10.1371/journal.pone.0190200 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Horowitz J, Velázquez T. Small but growing group incarcerated for a month or more has kept jail populations high. Pew Charitable Trust. Accessed April 3, 2024. https://www.pewtrusts.org/en/research-and-analysis/articles/2020/06/23/small-but-growing-group-incarcerated-for-a-month-or-more-has-kept-jail-populations-high
- 36.Rockett IR, Wang S, Stack S, et al. Race/ethnicity and potential suicide misclassification: window on a minority suicide paradox? BMC Psychiatry. 2010;10:35. doi: 10.1186/1471-244X-10-35 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Multipliers and Calculations Used to Infer Suicide Counts in Years 1 and 2 After Release From Incarceration vs Counts for Other Periods
eTable 2. Calculation of Crude Mortality Rate (CMR) for the Population Aged 18-68 in Minnesota Demographically Matched to Its Population of People Released From Prison in 2010-2019
eTable 3. Uncertainty Around Suicide Risk Estimates Over the Year and 2 Years After Release From Incarceration Compared to People Without Recent Incarceration
eTable 4. Uncertainty Around Suicide Risk Estimates During Incarceration and Over the Year and 2 Years After Release From Incarceration Compared to People Who Are Incarcerated
Data Sharing Statement
