Suicide Mortality Among Formerly Incarcerated People Compared With the General Population in North Carolina, 2000–2020

Abstract

We aimed to compare rates and characteristics of suicide mortality in formerly incarcerated people to the general population in North Carolina (NC). We conducted a retrospective cohort study of 266,400 people released from NC state prisons between January 1, 2000 and March 1, 2020. Using direct and indirect standardization by age, sex, and calendar year, we calculated standardized suicide mortality rates and standardized mortality ratios comparing formerly incarcerated people to the NC general population. We evaluated effect modification by race-ethnicity, sex, age, and firearm involvement. Formerly incarcerated people had approximately twice the overall suicide mortality of the general population for three years after release, with the highest rate of suicide mortality being in the two-week period after release. In contrast to patterns in the general population, formerly incarcerated people had higher non-firearm-involved suicide mortality rates than firearm-involved suicide mortality rates. Formerly incarcerated female, White and Hispanic/Latino, and emerging adult people had greater elevation of suicide mortality compared to their general population peers than did other groups. These findings suggest a need for long-term support for formerly incarcerated people as they return to community living and identify opportunities for interventions that reduce the harms of incarceration for especially vulnerable groups.

Share of Adult Suicides After Recent Jail Release

“…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, 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 studies identified in the 2023 meta-analysis by Janca et al plus 3 studies we identified . All studies were dominated by decades-old data.…”

“…We built 10 jurisdictional estimates from 8 adult incarceration studies 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 identified in the 2023 meta-analysis by Janca et al 5 plus 3 studies we identified. 22 , 23 , 24 All studies were dominated by decades-old data. All except 3 studies 14 , 20 , 24 gave partial information on suicide timing in the 2 years after release.…”

“…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, Canada 17 ; (3) the SMR for releases in England and Wales 23 ; (4) a consolidated estimate from separate computations by sex; and (5) the meta-analysis by Janca et al 5 …”

“…Variance for CMRs and SMRs were computed using standard formulas. 12 Six of these high-quality studies 16 , 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.…”

“…We matched this estimate to the US release distribution by sex. Five studies 16 , 18 , 20 , 21 , 22 included people who were released unconvicted, all studies were adjusted to cover 1 year, but only 1 study 23 was from North America.…”