Author [citation] | Stated methods for repeated incarcerations | Stated time period examined after prison release | Stated methods of linkage | Stated outcome events or summary measures | Stated sources of data |
---|---|---|---|---|---|
Alex et al. [31] | Not stated | 42 days after release | Probabilistic record linkage | All-cause mortality. No ICD codes | Release records. Bureau of Vital Statistics records. Electronic health records |
Andersson et al. [32] | Not stated | Prison contact during year prior to death | Personal identification number linkage | Prison contact during year before death. Death by intoxication. ICD codes | Swedish National Board of Forensic Medicine, regional health care services. Municipal social services. National Prison and Probation Service |
Barry et al. [33] | None stated | Most recent prison release until death or study completion | Not stated | Death by drug overdose. ICD-10 codes | Veterans Affairs National Patient Care Database. Centers for Medicare and Medicaid Services data (includes prison admission/release dates). Veterans Affairs Suicide Prevention Applications Network. Veterans Affairs National Suicide Data Repository (SDR) (includes cause-specific death information) |
Binswanger 2011 et al. [34] | Person-time at risk in the community; for persons with repeated incarcerations during the study period, the time during a subsequent incarceration was excluded, whereas the time between the next release and death, another incarceration, or the end of the study was included | For early deaths, defined as within 30 days of release from prison | Probabilistic score | All-cause mortality, overdose mortality and early (within 30 days of release) mortality. No ICD codes | Department of Corrections’ records. National Death Index |
Binswanger 2013 et al. [35] | The time at risk included time after release and excluded time in prison during any subsequent incarcerations | First month, months 2 to 12, and subsequent months after release | Identities were linked probabilistically | All-cause mortality, 11 causes of death and their subcauses, substance related causes, and the most common substance combinations. ICD-10 codes | Administrative records of the Washington State Department of Corrections. National Death Index |
Binswanger 2016 et al. [36] | The index release was that closest to death | Not stated | Matched personal identifiers | All-cause mortality and overdose mortality. No ICD codes | Washington State Department of Corrections. National Death Index |
Binswanger 2016 et al. [18] | Excluded data on subsequent person-years in custody for people who were reincarcerated after their first release and deaths in custody | 0–14 days, 15–90 days, 91–180 days, > 180 days and entire observation period after each release | Linked personal identifiers probabilistically | Infectious disease–related mortality. ICD-10 codes | Retrospective cohort studies of people released from prison in Queensland and Washington State. National death index |
Binswanger 2020 et al. [37] | Not stated | Month after prison, parole, and probation release | Linked identifiers | Overdose mortality. ICD-10 codes | Michigan Department of Corrections administrative databases. National Death Index (NDI)-Plus |
Bird 2015 et al. [38] | Calculated person-days at liberty in the first 12 weeks after a qualifying release from the day of release up to the earliest of: date of death, date of re-incarceration for at least 14 days or 12 weeks after the qualifying release date | Risk of DRD in the 12 weeks following release; percentage of these DRDs which occurred during the first 14 days | Not stated | First 2 weeks and 12-week DRD totals. ICD-9 and ICD-10 codes | Linked prisoner-mortality database held at Information Services Division |
Bird 2016 et al. [39] | Most recent prison release date | 4-week after release | Not stated | Opioid-related deaths (ORDs). No ICD codes | National Records of Scotland official statistics on the number of DRDs. Electronically held Scottish prisoner and morbidity records: Scotland’s Privacy Access Committee, Scottish Prison Service and Disclosure Scotland clearances |
Brinkley-Rubinstein 2018 et al. [40] | Not stated | Incarcerated in the year before death | Linked deterministically | Fatal overdose. Fentanyl-related overdose deaths. No ICD codes | RI Office of the Medical Examiner on overdose deaths. Records from RI Department of Corrections |
Brinkley-Rubinstein 2019 et al. [20] | Person-time was censored at reincarceration. Person-time was calculated from the day of release from prison until death, reincarceration, or the end of 2016 | 2 weeks, 1 year and complete follow-up after release | Linkage using last and first names, date of birth, and sex | Opioid overdose death. ICD-10 codes | North Carolina Department of Public Safety (NCDPS). North Carolina death records |
Bukten et al. [41] | The time at risk includes only time outside prison; both for individuals with one or repeated incarcerations in the study period, all the time incarcerated was excluded | First week, second week, 3–4 weeks and 2–6 months after release and by three different time intervals of release (2000–04, 2005–09, 2010–14) | Personal identification numbers | All-cause and cause-specific mortality. ICD-10 codes | Norwegian prison registry. Norwegian Cause of Death Registry |
Calcaterra et al. [42] | For persons with repeated incarcerations during the study period, the time during a subsequent incarceration was excluded, whereas the time between the next release and death, another incarceration, or the end of the study was included | 2-week intervals, weeks 1–2, 3–4, 5–6, 7–8 and all weeks after release | Probabilistic score | Causes-of-death 1) non-cocaine psychostimulants 2) cocaine only and 3) all psychostimulants. ICD-10 codes | Washington State Department of Corrections. National Death Index |
Chang et al. [27] | Not stated | Not stated | Unique personal identification numbers | All-cause and external-cause mortality. ICD-10 codes | National Crime Register. National Patient Register, inpatient psychiatric hospital admissions, and outpatient care. Cause of Death Register. Longitudinal Integration Database for Health Insurance and Labour Market Studies. Multi-Generation Register |
Degenhardt et al. [43] | Included all eligible prison releases. Person-years at risk accrued during time out of prison (time incarcerated was excluded) | First day, first week, first 2 weeks, month and year following release | Probabilistic linkage | Specific causes of death included accidental drug-induced deaths. ICD-10 codes | Pharmaceutical Drugs of Addiction System (PHDAS). The Reoffending Database (ROD) Department of Corrective Services. National Death Index |
Forsyth 2014 et al. [44] | Person-time was calculated from every release during follow-up until death, re-incarceration or the end of study follow-up. Deaths in custody were excluded | Up to 4 weeks, after 4 weeks up to 6 months, after 6 months up to 1 year, all follow-up to 1 year and more than 1 year after a release | Linked probabilistically | Alcohol-related, drug-related, substance-related i.e. drug or alcohol cause of death. ICD-9 and 10 codes | Incarceration data from Corrective Services. National Death Index |
Forsyth 2018 et al. [45] | Person-time starting from the date of the first release after baseline interview and censored on 31 May 2013 or death, with any time in prison removed from follow-up time at risk | Not stated | Probabilistic linkage | Drug-related deaths and alcohol and other drug-related deaths. ICD-10 codes | Baseline survey. Prison medical records. Community health records. Correctional records. National Death Index |
Gan et al. [46] | Cumulative duration of incarcerations during the follow‐up period was excluded from person‐time of follow‐up | 3‐year follow‐up period | Deterministic and probabilistic linkage | Overdose‐related death. ICD-9 or ICD-10 codes | Provincial incarceration records. Linked administrative health data, BC Coroners Service and Vital Statistics Agency. Provincial health insurance data |
Gjersing et al. [47] | Not stated | Release up to 6 months before death | For matching purposes, the data included full name, personal identification number, date of birth, date of death, postal code for region of death, residential postal code and whether the person had a post-mortem examination | Drug-induced deaths. No ICD codes | National Cause of Death Registry. Data on toxicology from the Institute of Forensic Medicine at the University of Oslo. Norwegian Correctional Services. Social and health services. Public social services |
Green et al. [48] | Not stated | Recently incarcerated defined as death within 12 months of release | Not stated | Overdose death attributed to fentanyl. No ICD codes | Office of State Medical Examiners for deaths. Department of Corrections (RIDOC) |
Groot et al. [49] | Not stated | Released within the year before death | Matching names and dates of birth | Drug toxicity deaths. No ICD codes | Deaths from Office of the Chief Coroner. Incarceration records from Correctional Services, both part of the Ontario Ministry of Community Safety and Correctional Services |
Haas et al. [19] | Excluded those reincarcerated within five days of release. Any outcomes occurring after reincarceration were not included in analysis | After release until fatal overdose, reincarcerated or study end date (31 December 2018) | Matching primarily on name and date of birth, supplemented with data on race/ethnicity when available | Fatal and non-fatal opioid overdose. No ICD codes | Department of Correction (DoC) records. Deaths from the Connecticut Office of the Chief Medical Examiner (OCME) |
Hacker et al. [50] | Defined incarceration as ≥ 1 episode of incarceration ever and in the year before death | Incarceration in the year before death | Matching algorithm, including first and last name, date of birth, social security number, and demographic information | Opioid-related overdose death | Allegheny County Medical Examiner autopsy data. Allegheny County Department of Human Services (ACDHS) Data Warehouse |
Hakansson et al. [51] | No access to re-incarcerations and releases | Not stated | Not stated | Causes of death. ICD-10 codes | Database of criminal justice clients with substance use problems. National Causes of Death Register |
Huang et al. [22] | In repeat incarcerations during the study period, used the date of release from the last incarceration for the starting point to measure the period from prison release to death | First week after release compared to following 4 weeks after release | Unique ID linkage | All-cause mortality and overdose mortality. ICD-9 codes | National Death Registry. Methadone Maintenance Treatment (MMT) database |
Kinner et al. [30] | The date of first release from custody was determined; follow-up periods of 4 weeks and 1 year were used regardless of reimprisonment within these time frames | Four weeks and 1 year | Not stated | Cause of death by drug-related, natural and all other causes. ICD-9 and 10 codes | Australian Bureau of Statistics. Data from two recent Australian record-linkage studies conducted in Western Australia and New South Wales were used. WA cohort: all prisoners released from custody. WA Registrar General’s record of deaths. NSW cohort: all prisoners released from custody. National Death Index |
Kouyoumdjian et al. [52] | Not stated | Not stated | Deterministic linkage and probabilistic linkage | Cause of death, deaths due to specific preventable diseases of interest, and certain risk factors. ICD-9 codes | Ontario Ministry of Community Safety and Correctional Services. Registered Persons Database. Mortality data Registrar General Death database |
Krawczyk et al. [26] | Not stated | Not stated | Probabilistic matching | Opioid overdose death. No ICD codes for death | Maryland statewide criminal justice records. All-payer hospitalization database. Overdose death records |
Larochelle et al. [53] | Not stated | Past 12 months, 0–3 months, 4–12 months and not 0–3 months, 13–24 months and not 0–12 months, 25–36 months and not 0–24 months | Multistage deterministic linkage | Opioid overdose death. No ICD codes | APCD. Registry of Vital Records and Statistics (RVRS). Prescription Monitoring Program (PMP). Acute Care Hospital Case Mix (Case Mix). Massachusetts Ambulance Trip Record Information System (MATRIS). Bureau of Substance Addiction Services’ (BSAS) licensed treatment encounters. Department of Corrections (DOC) and Houses of Corrections (HOC) |
Lim et al. [54] | Person-years defined as the number of days during which they were not incarcerated from 2001–2005, including days between each discharge and the subsequent incarceration | 1–2 weeks, 3–4 weeks, 5–6 weeks, 7–8 weeks, ≥ 9 weeks after release | Probabilistic matching | Underlying cause of death, drug-related death. ICD-10 codes | Jail records. Death and single-adult homeless registries |
Loeliger et al. [55] | Incorporated data across multiple incarcerations during follow-up | Not stated | Not stated | All-cause mortality and drug overdose. ICD-10 codes | Linked pharmacy, custodial, death, case management, and HIV surveillance data from Connecticut Departments of Corrections and Public Health |
Marsden et al. [28] | Participants could be recruited on each occasion of incarceration during the recruitment period | First year of release: 1–28 days, 29–121 days and 122–365 days | Not stated | All-cause mortality and drug-related poisoning deaths. ICD-10 codes | Prison National Offender Management Information Service (P-NOMIS). Prison IDTS healthcare provider. Justice Statistics Analytical Services (JSAS database). Office for National Statistics, national deaths register, accessed from the Health and Social Care Information Centre (HSCIC). English National Drug Treatment Monitoring System (NDTMS) |
Pizzicato et al. [56] | For multiple incarcerations, the time between subsequent incarcerations was excluded from person time at risk | 0–2 weeks, 3–4 weeks and ≥ 5 weeks after release | Matched on name, date of birth, and gender | All-cause, overdose, and non-overdose mortality. No ICD codes | Incarceration records. Philadelphia Department of Prisons (PDP). Medical Examiner’s Office and death certificate records. Death records from the Pennsylvania Department of Health’s Bureau of Vital Records |
Ranapurwala et al. [21] | For multiple incarcerations, excluded time in prison | 2 weeks, 1 year and complete follow-up after release | Linked using last and first names, date of birth, and sex | Cause of death, opioid overdose death. ICD-10 codes | Prison release data from the NC Department of Public Safety. NC death records from the NC Division of Public Health |
Rosen et al. [57] | Release from index incarceration to death, another incarceration or December 31, 201 (which ever occurred first) | Not stated. Until death, censored by reincarceration or study end | Deterministic matching algorithms with probabilistic matching routines | Cause of death. ICD-10 code | Records from the prison system. Death records from the NC State Center for Health Statistics |
Saloner et al. [29] | Not stated | Not stated. Released in 2015 and outcomes occurring in 2016 | Probabilistic matching | Fatal opioid overdose. ICD-9 and ICD-10 codes nonfatal opioid overdose | All-payer hospital discharges. Prescription drug monitoring program (PDMP). Public-sector specialty behavioral treatment criminal justice records |
Spaulding 2011 et al. [23] | Person-time included all time (between incarcerations and following the final incarceration) outside of prison during the study period | 0– < 1 month, 1– < 6 months, and 6–12 months after released | Matched on name, Social Security number, age, home address, and known aliases | Cause-specific mortality. No ICD codes | Georgia Department of Corrections (GDC). Georgia Death Registry |
Spaulding 2015 et al. [24] | Person-years of follow-up for the total cohort, as well as for each period of observation inside and outside prison. Inside prison considered either during the index incarceration or subsequent reincarceration | In prison (either during the index incarceration or subsequent reincarceration) and during first 2 weeks, second 2 weeks and more than 1 month after released | Probabilistic algorithms | Mortality from liver disease HIV and overdose. ICD-9 or ICD-10 codes | Georgia Department of Corrections Planning and Strategic Management Section. Georgia Death Registry, National Death Index |
Spittal 2014 et al. [58] | For repeated incarcerations, subsequent time in prison was excluded | First six months and complete follow-up after release | Probabilistic method and manual review | Cause-specific mortality and drug-related deaths. ICD-9 and ICD-10 codes | Queensland Corrective Services (QCS). National Death Index |
Spittal 2019 et al. [59] | Not stated | Not stated | Probabilistic matching and clerical review | Death from external causes, defined as drug overdose, suicide, transport accidents or violence. ICD-9 and ICD-10 codes | Queensland Corrective Services (QCS), Queensland Health. National Death Index |
Van Dooren et al. [60] | For subsequent incarcerations, time in prison was deducted from time at risk and deaths in prison were excluded | Censored at death or 365 days after release | Probabilistic matching | Drug‐related deaths, other substance abuse and opioid‐related deaths. ICD-9 and ICD-10 codes | Correctional facilities data. Australian Bureau of Statistics |
Victor et al. [25] | Coded each reincarceration between the initial 2017 release date and the 2 years following the initial release date as an ‘additional post-release booking’ to determine the potential effect of re-booking(s) on the hazard rate | First 2 weeks, up to 1 year and 2 years after released | Probabilistic linkage | Accidental fatal overdose. No ICD codes | Administrative records from the Marion County Sheriff’s Office (MCSO). Marion County Coroner’s Office (MCCO) |
Webb et al. [61] | Not stated | Contact with the criminal justice system from 1 January 1980. Controls were selected during 1994–2006 | Unique Central Person Registration number | Cause-specific mortality including self-poisoning by narcotics & hallucinogens. ICD-10 codes | National Causes of Death Register. National Criminal Register. Psychiatric Central Research Register. Central Population Register and the Integrated Database for Labour Market Research (IDA). IDA database |
Wortzel et al. [62] | Person-time at risk in the community excluded time in prison during subsequent incarcerations | Not stated | Matched by first name, last name, sex, birth date (month, day, and year, within one year), and eight of the nine digits in the social security number | All-cause deaths. Deaths from injury by self or others, medical deaths, suicide, alcohol or drug overdose, homicide, cardiovascular disease and cancer. ICD-10 codes | Washington State DOC. Veterans Benefit Administration (VBA) database |