U.S. Department of Justice Office of Justice Programs Bureau of Justice Statistics ------------------------------------------------------- This file is text only without graphics and many of the tables. A Zip archive of the tables in this report in spreadsheet format (.csv) and the full report including tables and graphics in .pdf format are available on BJS website at: http://www.bjs.gov/index.cfm?ty=pbdetail&iid=5341 This report is one in a series. More recent editions may be available. To view a list of all in the series go to http://www.bjs.gov/index.cfm?ty=pbse&sid=76 ------------------------------------------------------- Statistical Tables Mortality in Local Jails and State Prisons, 2000–2013 Margaret Noonan, BJS Statistician Harley Rohloff and Scott Ginder, RTI International For the third consecutive year, the number of inmates who died while in the custody of local jails or state prisons increased. A total of 4,446 inmates died in 2013, an increase of 131 deaths since 2012. The number of deaths in local jails and prison decreased an annual average of 2% between 2008 and 2010. In 2013, a total of 967 jail inmates died while in the custody of local jails. The number of deaths increased from 958 deaths in 2012 to 967 in 2013, while the jail population decreased 4%. As a result, the overall mortality rate in local jails increased from 128 per 100,000 jail inmates in 2012 to 135 per 100,000 in 2013. Suicide and heart disease have been the top two causes of death in local jails since 2000 (figure 1). Suicide has been the leading cause of death in jails every year since 2000. In 2013, a third (34%) of jail inmate deaths were due to suicide. The suicide rate increased 14%, from 40 suicides per 100,000 jail inmates in 2012 to 46 per 100,000 in 2013. The number of state prisoner deaths increased 4% in 2013, for a total of 3,479 deaths, or an additional 122 deaths since 2012 (figure 2). From 2001 to 2013, the majority (9 in 10 deaths) of prisoner deaths were due to illness-related causes. Between 2012 and 2013, the overall mortality rate for state prisoners increased 3%, from 265 deaths per 100,000 in 2012 to 274 per 100,000 in 2013. This was the highest mortality rate since the collection began in 2000. The mortality rate for illness increased 4% between 2012 and 2013, from 234 to 243 deaths per 100,000 inmates. In 2013, the mortality rates increased for the leading causes of illness—heart disease (up 11%), cancer (up 4%), and liver disease (up 16%), while the rates decreased for respiratory diseases (down 12%) or AIDS- related deaths (down 30%). The rate of AIDS-related deaths in state prisons has declined 82% since 2001. Data in this report were developed from the Bureau of Justice Statistics’ (BJS) Deaths in Custody Reporting Program (DCRP), an annual data collection about inmate deaths in local jails and state prisons. The DCRP is the only national statistical collection that obtains comprehensive information about deaths in adult correctional facilities. The program began collecting data from jails in 2000 and from prisons in 2001. This report covers deaths occurring in local jails from 2000 to 2013 and in state prisons from 2001 to 2013. Statistical tables provide information about the causes and circumstances of local jail and state prison inmate deaths and present trends by cause of death, selected decedent characteristics, and mortality rates for jail and prison inmates by state. *************************** Mortality in local jails **************************** Cause of death ***************** * Local jail inmate deaths increased 1%, from 958 deaths in 2012 to 967 deaths in 2013 (table 1). * Suicides in local jails increased 9%, from 300 suicides in 2012 to 327 in 2013. * Jail inmate deaths due to liver disease decreased 35%, from 29 deaths in 2012 to 19 in 2013. * Drug or alcohol intoxication deaths in local jails increased 23%, from 57 deaths in 2012 to 70 in 2013. * Illness-related deaths accounted for 50% of all inmate deaths in local jails in 2013. Heart disease continued to be the leading cause of illness-related inmate deaths in local jails in 2013 (table 2). * Suicide was the leading cause of death in local jails in 2013 (34% of all jail deaths) and has been the leading cause of death in local jails each year since 2000. * The suicide rate increased from 40 suicides per 100,000 local jail inmates in 2012 to 46 suicides per 100,000 local jail inmates in 2013 (table 3). Decedent characteristics ************************** * In 2013, a total of 843 male and 124 female inmates died in local jails (table 4). * The typical jail inmate who died in 2013 was male (87%), white (54%), age 35 or older (70%), and in custody for fewer than 7 days (40%) (table 5). * The mortality rate for Hispanics was 135 per 100,000 local jail inmates in 2013. Due to changes in measurement, this rate should not be compared to rates in previous years (table 6). (See Methodology.) Cause of death and demographics *********************************** * More than half (60%) of all suicides in jails from 2000 to 2013 involved inmates who were age 25 to 44 (table 7). * From 2000 to 2013, the number of male jail inmate deaths (12,092) was nearly 8 times the number of female inmate deaths (1,630). * From 2000 to 2013, the average annual mortality rate for male (137 per 100,000) and female (133 per 100,000) jail inmates were nearly equal (table 8). * The mortality rate for suicide among male jail inmates (43) was 1.5 times the rate for female inmates (28) from 2000 to 2013. * The mortality rate for drug or alcohol intoxication for black non-Hispanic (5 per 100,000) and Hispanic (6 per 100,000) jail inmates was nearly equal from 2000 to 2013, but was nearly three times higher for white non-Hispanic jail inmates (16 per 100,000). Facility characteristics ************************** * Most jails (80%) reported no deaths in 2013. Twice as many jails reported a single death (12%) as those reporting multiple deaths (6%)(table 9). * In 2013, approximately a quarter (23%) of the average daily jail population was a hold for another jurisdiction (table 10). * The mortality rate for jail inmates held for another correctional authority (80 per 100,000) was lower than the rate for other jail inmates (152 per 100,000). * In 2013, 23% of all jail inmate deaths occurred in California and Texas (table 12). * The median state-level jail death rate was 140 deaths per 100,000 local jail inmates in 2013 (table 13). Rates by cause and decedent characteristics ********************************************** * The mortality rates for illness among females were higher than the rates for males from 2004 to 2012 (appendix table 2). * The mortality rates for heart disease among males were higher than the rates for females every year from 2010 to 2013 (appendix table 3). * From 2009 to 2013, the suicide rate increased 23% from 35 per 100,000 jail inmates to 43 per 100,000 (appendix table 4). ****************************** Mortality in state prisons ****************************** Cause of death ****************** * Deaths in prison increased from 3,357 in 2012 to 3,479 in 2013, reaching the highest number since the prison data collection began in 2001. The total number of deaths increased 4% between 2012 and 2013 (table 16). * Illness-related deaths accounted for 89% of all deaths in prison in 2013. ** Deaths from liver disease increased 16%, from 304 deaths in 2012 to 354 deaths in 2013. This was the largest percent change observed among illness deaths in prison in 2013. ** Cancer (31%) and heart disease (26%) accounted for more than half of all prisoner deaths (table 17). * The mortality rate increased 3%, from 265 deaths per 100,000 state prisoners in 2012 to 274 per 100,000 in 2013. Although there were some fluctuations in the rates for unnatural deaths (e.g., accidents increased while suicides declined), the change in the rate for illness (up 4%) accounted for the increase in the overall rate (table 18). Decedent characteristics **************************** * The 1,966 prisoners age 55 or older made up more than half (57%) of all deaths in state prisons in 2013 (table 19). * In state prisons, the percentage of decedents age 55 or older increased by an average of 8% annually since 2001 (table 20). * White inmates accounted for more than half (55%) of deaths in state prisons, and nearly all (96%) were male (table 21). * In 2013, an estimated 117,805 (9%) of state prisoners were age 55 or older, up from 47,900 (4%) in 2001. * Combined, deaths due to suicide, drug or alcohol intoxication, accidents, and homicide accounted for fewer than 200 female prisoner deaths between 2001 and 2013 (table 23). * A third of homicides in state prisons involved prisoners age 45 or older between 2001 and 2013. * From 2001 to 2013, male prisoners were twice as likely as female prisoners to die of cancer, heart disease, and liver disease, which were the leading causes of death in state prisons (table 24). * Regardless of race or Hispanic origin, there was little variation in the average annual homicide rate (4 to 6 homicides per 100,000 state prisoners) for state prisoners between 2001 and 2013. * From 2001 to 2013, prisoners age 55 or older were 3 to 9 times more likely to die of an accident than younger prisoners. Deaths by jurisdiction ************************* Every state department of corrections reported at least one prisoner death in 2013 (table 25). * The prisoner death rate by state varied from 115 deaths per 100,000 to 628 deaths per 100,000 prisoners. The median state- level death rate among prisoners was 273 deaths per 100,000 state prisoners (table 26). Cause of death by state **************************** * California’s 4,790 prisoner deaths from 2001 to 2013 represent more than 11% of the total deaths (42,157). During this period, almost a third of all state prisoner deaths from drug or alcohol intoxication were in California (174 of 546 deaths) (table 27). * Overall mortality rates and mortality rates by state and by cause of death may not be directly compared between states due to differences in age, sex, race or Hispanic origin, and other decedent characteristics (table 28). Rates by cause and decedent characteristics ********************************************** * The illness mortality rate increased to 269 per 100,000 prisoners in 2013, up from 262 per 100,000 in 2012 (appendix table 6). * The prisoner cancer mortality rate has increased every year between 2008 and 2013 (appendix table 7). * The prisoner heart disease mortality rate was relatively stable between 2003 and 2013, fluctuating between 64 and 67 deaths per 100,000 prisoner deaths each year (appendix table 8). ************************************************************** ******************* List of tables ********************* TABLE 1 Number of local jail inmate deaths, by cause of death, 2000–2013 7 TABLE 2 Percent of local jail inmate deaths, by cause of death, 2000–2013 7 TABLE 3 Mortality rate per 100,000 local jail inmates, by cause of death, 2000–2013 8 TABLE 4 Number of local jail inmate deaths, by selected decedent characteristics, 2000–2013 9 TABLE 5 Percent of local jail inmate deaths, by selected decedent characteristics, 2000–2013 10 TABLE 6 Mortality rate per 100,000 local jail inmates, by selected decedent characteristics, 2000–2013 11 TABLE 7 Number of local jail inmate deaths, by cause of death and selected decedent characteristics, 2000–2013 12 TABLE 8 Average annual mortality rate per 100,000 local jail inmates, by cause of death and selected decedent characteristics, 2000–2013 13 TABLE 9 Number and percent of local jail jurisdictions reporting to the Deaths in Custody Program, by number of deaths reported each year, 2000–2013 13 TABLE 10 Number, percent, average daily population, and mortality rate per 100,000 local jail inmate deaths, by hold status, 2013 14 TABLE 11 Number of local jail inmates held on an average day, by state, 2000–2013 15 TABLE 12 Number of local jail deaths, by state, 2000–2013 16 TABLE 13 Mortality rate per 100,000 local jail inmates, by state, 2000–2013 17 TABLE 14 Number of local jail jurisdictions reporting to the Deaths in Custody Reporting Program, by state, 2000–2013 18 TABLE 15 Number of local jail jurisdictions reporting one or more deaths to the Deaths in Custody Program, by state, 2000– 2013 19 TABLE 16 Number of state prisoner deaths, by cause of death, 2001–2013 20 TABLE 17 Percent of state prisoner deaths, by cause of death, 2001–2013 20 TABLE 18 Mortality rate per 100,000 state prisoners, by cause of death, 2001–2013 21 TABLE 19 Number of state prisoner deaths, by selected decedent characteristics, 2001–2013 21 TABLE 20 Percent of state prisoner deaths, by selected decedent characteristics, 2001–2013 22 TABLE 21 Mortality rate per 100,000 state prisoners, by selected decedent characteristics, 2001–2013 22 TABLE 22 Estimated number of state and federal prisoners in custody, by selected inmate characteristics, 2001–2013 23 TABLE 23 Number of state prisoner deaths, by cause of death and selected decedent characteristics, 2001–2013 23 TABLE 24 Average annual mortality rate per 100,000 state prisoners, by cause of death and selected decedent characteristics, 2001–2013 24 TABLE 25 Number of state and federal prisoner deaths, by location, 2001–2013 25 TABLE 26 Mortality rate per 100,000 state and federal prisoners, by location, 2001–2013 26 TABLE 27 Number of state and federal prisoner deaths, by cause of death and location, 2001–2013 27 TABLE 28 Average annual mortality rate per 100,000 state and federal prisoners, by cause of death and location, 2001– 2013 28 ************************************************************** ************************************************************** ************************* List of appendix tables ************************* APPENDIX TABLE 1 Estimated number of local jail inmates in custody on an average day, by selected inmate characteristics, 2000–2013 34 APPENDIX TABLE 2 Illness mortality rate per 100,000 local jail inmates, by selected decedent characteristics, 2001–2013 35 APPENDIX TABLE 3 Heart disease mortality rate per 100,000 local jail inmates, by selected decendent characteristics, 2001–2013 35 APPENDIX TABLE 4 Suicide mortality rate per 100,000 local jail inmates, by selected decedent characteristics, 2001–2013 36 APPENDIX TABLE 5 Mortality rate for all other unnatural deaths per 100,000 local jail inmates, by selected decedent characteristics, 2000–2013 36 APPENDIX TABLE 6 Illness mortality rate per 100,000 state prisoners, by selected decedent characteristics, 2002–2013 37 APPENDIX TABLE 7 Cancer mortality rate per 100,000 state prisoners, by selected decedent characteristics, 2002–2013 37 APPENDIX TABLE 8 Heart disease mortality rate per 100,000 state prisoners, byselected decedent characteristics, 2002–2013 38 APPENDIX TABLE 9 Liver disease mortality rate per 100,000 state prisoners, by selected decedent characteristics, 2002–2013 38 APPENDIX TABLE 10 Respiratory disease mortality rate per 100,000 state prisoners, by selected decedent characteristics, 2002–2013 39 APPENDIX TABLE 11 Mortality rate for all other illnesses per 100,000 state prisoners, by selected decedent characteristics, 2002–2013 39 APPENDIX TABLE 12 Mortality rate for unnatural deaths per 100,000 state prisoners, by selected decedent characteristics, 2002–2013 40 ************************************************************** **************** Methodology **************** Data collection coverage ************************** The Deaths in Custody Reporting Program (DCRP) is an annual Bureau of Justice Statistics (BJS) data collection. The DCRP collects national, state, and incident-level data on persons who died while in the physical custody of the 50 state departments of corrections or the approximately 2,800 local adult jail jurisdictions nationwide. The DCRP began in 2000 under the Death in Custody Reporting Act of 2000 (P.L. 106- 297), and it is the only national statistical collection to obtain comprehensive information about deaths in adult correctional facilities. BJS uses DCRP data to track national trends in the number and causes (or manners) of deaths occurring in state prison or local jail custody. Mortality data measured by the DCRP include decedent characteristics, such as age, sex, race or Hispanic origin, date of admission, conviction status, admission offense, and the location and type of facility (prison or jail) where the inmate died. The DCRP also collects data about circumstances surrounding the death, including the cause, time and location where the death occurred, and information on whether an autopsy was conducted and the availability of results to the respondent. Data on executions are excluded from this report but are accessible on the BJS website along with the DCRP mortality data. Statistics presented in this report are current as of March 1, 2015. For more information on mortality in correctional settings, see Mortality in Local Jails, 2000– 2007 (NCJ 222988, BJS web, July 2010); Medical Causes of Death in State Prisons, 2001–2004 (NCJ 216340, BJS web, January 2007); and Suicide and Homicide in State Prisons and Local Jails (NCJ 210036, BJS web, August 2005). The DCRP data collection instruments are administered annually to both state prisons and local jails. Respondents provide an aggregate count of the number of deaths that occurred during the referenced calendar year. The jail (CJ-9) and prison (NPS- 4A) survey instruments used to obtain data on each prison and jail death are available on the BJS website at www.bjs.gov. In addition to the death count, BJS requests that jails provide summary statistics about their population and admissions. All jails, including those with no deaths to report (which includes about 80% of jails in any given year) are asked to complete the annual summary survey form. The 2013 DCRP jail annual summary form was combined with the Census of Jails data collection, which gathered facility-level data on the jail reporting units. Items such as each facility’s count of inmates confined at yearend by race, operating expenditures, rated capacity, and number of employed staff were new for respondents this year. The instrument used for the Census of Jails included 18 questions, which was a significant increase compared to the 5-question instrument used in the past. BJS obtains a separate report describing the decedent’s characteristics and the circumstances surrounding the death for each death that occurred in a state prison or local jail. State prison and local jail respondents can submit individual records on decedents at any time during a collection cycle through a BJS web-based collection system. BJS has modified the survey instruments slightly over time, including changes to clarify questions and ease the burden on respondents. Several questions were added to capture information on any medical treatment that the inmate received prior to death. Changes also allowed respondents to elaborate on cause of death by adding text boxes to the intoxication, suicide, and inmate-involved homicide death questions. Previously, these fields had only been available for deaths due to illness, accident, homicides not caused by other inmates, and other unspecified causes. In addition, BJS clarified respondent instructions on the prison forms and removed a question collecting the conviction status of an inmate after an analysis showed that more than 99% of prison inmates were convicted at the time of death. BJS also streamlined the survey layout to simplify the progression through questions related to autopsies. Rather than initially submitting an incomplete instrument that indicated pending autopsy results, respondents were only permitted to complete the entire survey after autopsy results were available. This change reduced follow-up contact and the burden on respondents. Starting in 2001 and annually thereafter, BJS has collected DCRP data directly from state prison systems, maintaining a 100% response rate. Nonresponse *************** The jail universe includes all jails currently operating and jails that have been contacted for the DCRP but have closed, consolidated, or otherwise eliminated operations. This universe allows BJS to determine jail participation in the DCRP. The most recent jail universe, constructed in January 2015, identified 2,870 jurisdictions representing 3,170 jail facilities. Of these, 2,703 jurisdictions (94.2%) participated in the DCRP. A jail jurisdiction is a legal entity that has responsibility for managing jail facilities. Jail jurisdictions typically operate at the county level, in which a sheriff’s office or jail administrator manages the local facilities. The DCRP data identify the jail facility in which a jail inmate dies, but the data are arrayed at the jail jurisdiction level. BJS defines a jail as a locally operated correctional facility that confines persons before or after adjudication for more than 72 hours, excluding temporary lockups. Typically, there is one facility per jail jurisdiction, but the 2013 Census of Jails found that 6.7% of jail jurisdictions had multiple facilities under a central authority *** Footnote 1 Approximately 15% of all facilities were part of a multifacility jurisdiction.*** (See the BJS website for more information.) Determining eligibility for reporting to the DCRP *************************************************** In the DCRP, custody refers to the holding of an inmate in a facility or to the period during which a correctional authority maintains a chain of custody over an inmate. For instance, if a jail transports an ill inmate to a hospital for medical services and that inmate dies while in the chain of custody of the jail, then that death is counted as a death in custody. A death that occurs when an inmate is not in the custody of a correctional authority is considered beyond the scope of the DCRP. Out-of-scope deaths include inmates on escape status or under the supervision of community corrections on probation, parole, or home-electronic monitoring. BJS instructs both state prison and local jail officials to determine whether the inmate was in the physical custody of the jurisdiction at the time of death, regardless of the reason an inmate was being held. For state prisons responding to the survey, inmates in physical custody include those held in any private prison facility under contract to the responding state’s department of corrections or in any of their state-operated facilities, including halfway houses, prison camps or farms, training or treatment centers, and prison hospitals. BJS instructs state prison officials to exclude deaths of inmates who were transferred to local jails while still serving a prison term because the DCRP obtains information about such deaths through the jail reports. Jail inmate custody includes inmates who are temporarily out of the jail facility but are within the chain of custody of the jail. For example, a death in custody would include a jail inmate who died after being transferred to an offsite facility that cares for critically ill persons. From 2000 to 2013, the DCRP data have shown that about 84% of jail deaths occurring in medical facilities have occurred outside of the jail facility. Custody is further complicated by the dual law enforcement and jail administration functions of some sheriffs’ offices. As a result, some deaths reported as jail deaths actually occurred before the jail had custody of the decedent. BJS identifies and excludes from the DCRP these deaths that occurred in the process of arrest by using information about the circumstances surrounding the death. Identifying and excluding duplicate records ********************************************** Duplicate death records may occur in the DCRP due to overlapping correctional populations and overlapping duties within correctional facilities. For example, a jail jurisdiction may have more than one unit responsible for reporting data to BJS, a jail may be holding a state prison inmate, or state prison systems may report the death of an inmate who was transferred to a local jail but was serving a prison sentence at the time. To identify duplicate death records, BJS reconciles the aggregate summary counts of deaths occurring during a calendar year with the number of individual death records obtained from a reporting jurisdiction. When discrepancies are identified, BJS contacts reporting jurisdictions to clarify matters. However, even if summary counts and individual reports are reconciled, duplicate death records may exist if multiple reporting units within a jurisdiction provide reconciled data. Duplicate records occur primarily in reports from jail jurisdictions that have multiple reporting entities. To identify duplicate records, BJS performs record-matching based on inmate name and date of birth, date of death, and date of admission into a correctional facility. After the aggregate count review, BJS searches for deaths reported to both the jail and prison DCRP collections, which most commonly occur when a local jail is housing an inmate for the state department of corrections. The death is considered part of the count of the facility that had custody of the inmate at the time of death, and the duplicate record is deleted. Duplicate and out-of-scope records are excluded from the analyses. Cause-of-death information ******************************** The instructions for completing the DCRP tell respondents to report death information as determined by an autopsy or other official medical death investigation. For this collection, intoxication deaths, accidents, suicides, and homicides are considered discrete causes of death. Although there is a distinction between manner and cause of death from a medico- legal standpoint, no such distinction is made in the DCRP. When reporting a death due to illness, accident, suicide, intoxication, or homicide, BJS requests that respondents describe the events surrounding these deaths. Clinical data specialists convert illness-related death text entries into standard medical codes according to the World Health Organization’s International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD- 10). Homicides include all types of intentional homicide and involuntary manslaughter as ruled by a medical examiner or pathologist at autopsy. For example, an inmate may die of positional asphyxia (suffocation caused by the position of the inmate’s body) while the inmate is being removed from a cell. A legal-intervention homicide committed while the inmate is trying to escape would also be included. In addition, homicides include cases that are ruled a homicide at autopsy when events that led to the death occurred prior to incarceration. For example, an inmate who was shot in the community years prior to incarceration died from complications of the gunshot wound while incarcerated. Other BJS sources of correctional mortality data ************************************************** BJS collects other data reported to the DCRP on correctional mortality. These other collections include -- * Capital Punishment, which provides data on legal executions. Further discussion on executions is available on the BJS website. * The National Prisoner Statistics (NPS), which annually collected aggregate counts of deaths in state and federal prisons prior to the establishment of the DCRP. Prior to 2007, the NPS also collected counts of deaths by cause of death, including deaths due to execution, illness, AIDS, suicide, accident, homicide, and other causes. After 2006, the Federal Bureau of Prisons continued to submit the counts of deaths by cause of death using the DCRP, but no longer provided counts of deaths using the NPS. Further discussion of the NPS is available on the BJS website. * The Census of Jails, conducted every 5 to 6 years, provides counts of inmate deaths in local jails. Further discussion of the Census of Jails is available on the BJS website. * The Survey of Jails in Indian Country (SJIC), which provides aggregate counts of the number of deaths occurring in all known Indian country correctional facilities operated by tribal authorities or the U.S. Department of the Interior’s Bureau of Indian Affairs. Further discussion of the SJIC is available on the BJS website. Reported statistics ************************ Mortality data are shown in statistical tables by type of correctional institution (state prisons and local jails) and include the number of deaths and mortality rates by year, cause of death, selected decedent characteristics, and state. Mortality rates are calculated per 100,000 inmates, with the denominators providing estimates of the number of person-years of exposure in custody in institutional corrections. Until 2010, the mortality rate for state prisons was calculated as the number of deaths per year divided by the midyear state prison population in custody multiplied by 100,000. Starting in 2011, the rate was calculated using yearend custody counts because midyear populations were no longer available. Custody counts for state prisons provide estimates of person-years for prison populations. BJS uses data from the NPS to provide midyear and yearend custody counts of prisoners. For more information on the NPS, see the BJS website. The mortality rate in local jails is calculated as the number of deaths per year divided by the jail inmate average daily population (ADP) multiplied by 100,000. The ADP for local jails is defined as the average daily number of jail inmates held in a jail jurisdiction during a calendar year, from January 1 through December 31. The ADP is used as the denominator for jail mortality rates to accommodate the high turnover and daily fluctuation in local jail populations. Also, the ADP better reflects the number of inmate days per year than a 1-day count. Jail populations have a higher turnover than prison populations; mean length of stay in local jails is about 21 days, compared to 2 years in state prisons. The jail ADP also reflects the annual number of admissions and mean length of stay, and can be expressed as the product of these two values. When mean length of stay is expressed in years, the ADP is equivalent to the number of person-years spent by jail inmates during a given year. BJS obtains the jail ADP data directly from jails through the DCRP using the summary form CJ-9A. Starting in 2002, BJS collected the ADP directly from respondents. Prior to 2002, BJS calculated the jail ADP by taking the average of the January 1 count from the prior year and the December 31 count from the reference year, which is an appropriate proxy measure for the ADP. Both denominators provide for annualizing mortality rates, which are calculated separately by group or by characteristic. The annualized mortality rates in state prisons and local jails are comparable to annual crude mortality rates reported by the National Center for Health Statistics (NCHS). The NCHS calculates crude mortality rates as the number of events for a period (e.g., a year) divided by the population estimate at the midpoint of the period. For general population mortality statistics, the NCHS employs the midyear population as an approximation to the average population exposed to risk of death during any given year*** Footnote 2 See Siegal, J. & Swanson, D. (2004). The Methods and Materials of Demography, Second Edition. San Diego, CA: Elsevier Academic Press, 269***. The crude mortality rates reported in the DCRP annual statistical tables are not directly comparable to the crude mortality rates within the (nonincarcerated) general population, and the crude mortality rates in state prisons are not directly comparable to those of local jails. The composition of the general population (i.e., age, sex, and race or Hispanic origin) differs from the population in state prisons and local jails. Because mortality is correlated with age, sex, and race or Hispanic origin, the crude mortality rates in state prisons, local jails, and the general population should not be compared. Individual inmate death records collected annually in the death file are included in the national death count. Independent jail-specific summary death counts are collected in the annual summary form (CJ-9A or CJ-10A) and serve as control death totals. If the death count in the summary form file is greater than the count in the individual inmate death file, the summary file count is used as the total in calculating a jail mortality rate. For 2013, individual records for jail inmate deaths were adjusted to match independent counts of deaths occurring in each jail, and the mortality rates were adjusted accordingly. Estimating inmate population characteristics to calculate mortality rates by demographic subgroups ********************************************************* BJS does not obtain annual data for all demographic characteristics of prison and jail inmates in all jurisdictions. BJS uses data from sources other than the DCRP to estimate the nationwide composition of state prison and local jail inmate populations by age, sex, and race or Hispanic origin. These data sources consist primarily of periodic surveys of inmates in custody in prisons and jails. The inmate characteristic distributions obtained from these surveys are applied to denominators (i.e., counts of inmates or ADP) to estimate the number of inmates in each demographic subgroup. BJS estimated the demographic distribution of the state prison population data from the NPS and National Corrections Reporting Program (NCRP) collections. For a discussion on the methodology for obtaining state prisoner distribution estimates by age, sex, and race or Hispanic origin, see Prisoners in 2012 (NCJ 243920, BJS web, December 2013). Prior to using the NPS and NCRP to estimate demographic distributions, reports of mortality rates for state prison inmates used demographic distributions derived from BJS’s 2004 Survey of Inmates in State and Federal Correctional Facilities (SISFCF). As a result, the state prison mortality rates shown in these tables may differ from previously published rates. A rate comparison between the two sources showed very little difference in the resultant mortality rates. In most instances, the rates either matched or nearly matched. The rates differed in only three instances: Hispanics in 2001, and females and inmates age 55 or older in 2002. In each instance, the rates calculated using population data from the NCRP and NPS were slightly higher (less than a 1% increase) than rates calculated using population data from the SISFCF. To estimate inmate demographic characteristic distributions of the ADP, BJS used data from several surveys to generate distributions of age, sex, and race or Hispanic origin and applied these distributions to the ADP. BJS’s Annual Survey of Jails (ASJ) provided estimates of the sex distribution of inmates for each year of the DCRP collection and applied these to each year’s ADP from the DCRP to estimate the ADP of male and female jail inmates. Jail Inmates at Midyear 2010 - Statistical Tables (NCJ 233431, BJS web, April 2011) documents that the distributions of inmate characteristics have changed slowly over time. For the most recent information on the ASJ, see the BJS website. To estimate the distribution of race and Hispanic origin of adult jail inmates, BJS used data from the Survey of Inmates in Local Jails (SILJ), the National Inmate Survey (NIS), and the Census of Jails to estimate the relative distribution of adults by race and Hispanic origin for different periods. Because the SILJ (2002), the NIS (2007 to 2009), and the Census of Jails (2013) are not fielded annually, the population estimates were smoothed before being applied to DCRP data. BJS used the SILJ estimates to cover the period from 2000 to 2004, the NIS estimates to cover the period from 2005 to 2012, and the Census of Jails to cover 2013. In all cases, the percentages associated with the distribution of race and Hispanic origin were applied to the adult jail ADP. For more information about jail inmates in 2002, see Profile of Jail Inmates, 2002 (NCJ 201932, BJS web, July 2004). To estimate the age distribution of the jail inmate population, BJS first obtained an estimate of the number of jail inmates age 17 or younger from the 2013 Census of Jails data collection. By applying the annual percentage of jail inmates age 17 or younger to the annual ADP, BJS obtained an estimate of the ADP of jail inmates age 17 or younger. To estimate the age distribution of adult jail inmates, BJS used data from the SILJ for 2000 to 2006 and the NIS for 2007 to 2013 to estimate the relative distribution of adults by age category for different periods. The age estimates were smoothed to account for gaps in reference years when age estimates were available, specifically for 2003, 2006, 2010, and 2011. Moving averages ****************** Moving averages were used to smooth short-term irregularities and to estimate long-term trends. For instance, moving averages were computed to examine data trends for certain causes of death in jails and prisons while smoothing short- term fluctuations. The data were cut into several 3-year overlapping periods spanning 13 years of prison data and 14 years of jail data. The moving averages in this report describe some changes in cause-specific mortality rates over time (e.g., whether the overall decline in the AIDS-related mortality rate was steady, or whether the increase of suicides in jails was recent). Moving averages were not computed for all causes of death in custody because the resultant rates would have been unstable and therefore statistically meaningless due to small cell sizes. Random error and suppression ****************************** The DCRP data on deaths in state prisons and local jails are not subject to sampling error. However, mortality data from a complete or near-complete enumeration may be subject to random error. Following the methodology of Brillinger ***Footnote 3 See Brillinger, D.R. (1986). The Natural Variability of Vital Rates and Associated Statistics. Biometrics, 42, 693– 734.***and NCHS, “the number of deaths that actually occurred may be considered as one of a large series of possible results that could have arisen under the same set of circumstances***Footnote 4 See Xu, J. et al. (2010). Deaths: Final Data for 2007. National Vital Statistics Report, 58(19). Retrieved from http://www.cdc.gov/nchs/data/nvsr/nvsr58/ nvsr58_19.pdf.*** The random variation can be large when the number of deaths is small. Therefore, caution is warranted when interpreting statistics based on small numbers of deaths. According to NCHS standards, mortality rates based on fewer than 100 deaths per year should be interpreted with caution. Continuing to use the NCHS and Brillinger methods, BJS quantified random variation by assuming that the appropriate underlying probability distribution for the number of deaths is a Poisson distribution. This provides a computationally simple and reasonable approach for estimating variances in mortality statistics when the probability of dying is low. BJS calculated variances based on the assumption of a Poisson process. From these variances, BJS calculated relative random error estimates, which are comparable to relative standard error in that the relative random error is the ratio of random error derived from the Poisson variance to the number of deaths. Following NCHS, when the relative random error exceeded 30%, BJS flagged estimated mortality rates due to the instability of the rate with a “!” symbol (Interpret with caution; too few sample cases to provide a reliable rate). Survey performance issues ************************** Survey administration and modifications to the survey form generated data from 2007 to 2009 that may not be wholly compatible with prior DCRP data. In 2012, BJS identified a previously undetected data error that mischaracterized more than 600 illness-related deaths as illness-related deaths missing detailed cause-of-death information. The data error led to an inflated frequency of prison deaths in reference year 2007 being characterized as all other illnesses while simultaneously depressing frequencies of known illnesses (e.g., heart disease, cancer, and liver disease). The error was identified and resolved, correcting the distributions. In 2008, local jail officials were unable to provide causes of death for 21.4% of jail inmate deaths. During 2009, BJS made modifications to the collection instruments to improve reporting and reduce burden. * Item nonresponse in 2008 jail data and unknown cause of death: An abnormally large number of cases were missing a response for cause of death in the 2008 jail file (n = 203; 21.4% of all jail deaths in 2008). This coincided with the final year the U.S. Census Bureau acted as the data collection agent for the DCRP. In prior years of the DCRP jail data collection, an average of 6% of all causes of death were classified as other or unknown. For this report, BJS categorized all 203 jail deaths from 2008 as missing data about cause of death. * 2009 data collection: Prior to fielding the 2009 DCRP collection, BJS reviewed the data collection instrument and data submission procedures and assessed the communications with DCRP respondents to reduce the burden on respondents. The survey was modified to facilitate navigation and to encourage online response. The modifications led to delays in implementing data collection. As a result, death reports were not collected in the year the deaths occurred, but were instead collected retrospectively during 2010 for the first time. Data collection for subsequent years resumed on a normal schedule. * Hispanic origin: In 2013, the death record form was altered to collect a two-question format regarding ethnicity and race. If Hispanic was selected, the decedent was classified as Hispanic regardless of what was selected for race. ************************************************************** The Bureau of Justice Statistics of the U.S. Department of Justice is the principal federal agency responsible for measuring crime, criminal victimization, criminal offenders, victims of crime, correlates of crime, and the operation of criminal and civil justice systems at the federal, state, tribal, and local levels. BJS collects, analyzes, and disseminates reliable and valid statistics on crime and justice systems in the United States; supports improvements to state and local criminal justice information systems; and participates with national and international organizations to develop and recommend national standards for justice statistics. William J. Sabol is director. This report was written by Margaret Noonan, Harley Rohloff, and Scott Ginder. Danielle Kaeble, Tracy Snell, and Margaret Noonan verified the report. Kim Aspinwall carried out data collection and processing with assistance from Alissa Chambers, under the supervision of Chris Ellis, RTI International. Scott Ginder provided statistical assistance, and Elizabeth Robbins provided technical assistance. Irene Cooperman and Jill Thomas edited the report. Tina Dorsey and Barbara Quinn produced the report. August 2015, NCJ 248756 ************************************************************** *************************** 7/27/2015 4:15pm JER ***************************