Part A_final

13

SUPPORTING STATEMENT

The Bureau of Justice Statistics requests a new clearance for the Mortality in Correctional Institutions (MCI) data collection, formerly known as the Deaths in Custody Reporting Program (DCRP). The current MCI-State Prisons collection is approved under OMB Control Number 1121-0249, which expires March 31, 2019. BJS proposes combining the MCI-State Prisons collection with the MCI-Jails collection, currently approved by OMB under control number 1121-0094 (expiration January 31, 2019). MCI collects data on prisoners and jail inmates who died while in the physical custody of a state prison or local jail.

In an attempt to coordinate the burden of BJS data collections on jail facilities, the MCI-Jails was combined with the Annual Survey of Jails (ASJ) and the Survey of Jails in Indian Country (SJIC) collections under a single OMB clearance (1121-0094) in 2015. The amount of overlap among these three collections, is minimal in content and coverage. The ASJ is a sample of approximately 900 jails while MCI-Jails is a census of all 3,000 jails, and SJIC only collects information from jails in Indian Country, which are not in the scope of the other two collections. Importantly, the reference dates differ between ASJ and MCI-Jails. Due to seasonal fluctuations in jail populations, and fewer inmates held at year-end (December 31), the ASJ uses the last weekday in June as its reference date. MCI-Jails uses December 31 as its reference date, and more importantly, uses average daily population (ADP) from January 1 to December 31 as the denominator in calculating mortality rates, which is consistent with the time period represented by the numerator, the number of deaths in a calendar year.

BJS proposes to separate these three collections and maintain the clearance number 1121-0094 for the ASJ, obtain a new clearance number for SJIC, and combine MCI-Jails with MCI-State Prisons under this request for 1121-0249. The combined mortality collection will collect administrative data from the 50 state departments of corrections (DOCs) and approximately 3,000 local jail jurisdictions on the number and nature of inmate deaths in custody.

The MCI was established in 2000, after the passage of the Death in Custody Reporting Act (DICRA) of 2000 (‘2000 DICRA’, P.L. 106-297, Appendix A), and BJS began collecting data on deaths in state prisons in 2001.

Currently, MCI is the only nationwide, comprehensive source of jail and prison mortality data. Individual-level data about the criminal justice system’s involvement with decedents are not available in other federal mortality data collections. The MCI data are part of BJS’ core correction statistics, as they contribute significantly to BJS’s mission of describing transitions and movements of offenders through the criminal justice system. In the case of the MCI, it describes a subset of releases from correctional populations, (i.e., deaths).

Prisoner death data are submitted annually to MCI by state department of corrections, and jail inmate data are provided by local jail administrators. Under DICRA 2000, states and local jail jurisdictions were required to report deaths to the Department of Justice. Currently BJS’s MCI is voluntary, does not meet the DICRA requirements, and is to be used for statistical and research purposes only. All 50 departments of correction have submitted data to the MCI since 2001, while the response rate for MCI-Jails since 2015 has been 97-98%.

The MCI collects data on the following elements for each decedent:

• Identifying characteristics - name of inmate, name and location of the correctional facility involved;

• Demographics - date of birth, date of death, race and Hispanic origin;

• Correctional characteristics - date of admission to the correctional facility, offenses for which the inmate was imprisoned, whether the prisoner stayed overnight in a mental health facility since admission, whether the death was due to a pre-existing medical condition, and what kind of medical treatment was provided for the underlying pre-existing condition;

• Characteristics and circumstances surrounding the death - cause of death, autopsy status, location of death, location of the incident that contributed to unnatural deaths (e.g., suicides, accidents, and homicides), and time of death;

• Any other relevant information - a ‘notes’ section allows the respondent to record any additional information.

The MCI data collection instruments are included in Appendix B. Each state department of corrections is asked to provide the total number of deaths each calendar year on an annual summary form (CJ-4: Mortality in Correctional Institutions, 2018 Annual Summary of Inmate Deaths in State Prisons), so that BJS can ensure it receives the correct number of individual death forms. MCI-State Prisons uses population data from BJS’s National Prisoner Statistics program (OMB control number 1121-0102) to calculate mortality rates. Individual deaths are reported separately on form CJ-4A: Mortality in Correctional Institutions, 2018 State Prison Inmate Death Report, with one form for each decedent.

Local jails provide the total number of deaths, as well as the facility population count by sex on December 31, the total number of persons held in the jail for other entities (U.S. Immigration and Customs Enforcement, U.S. Marshals Service, state or federal prisoners, Bureau of Indian Affairs, or other jail jurisdictions), ADP of the facility, and the total number of admissions during the calendar year on forms CJ-9A: Mortality in Correctional Institutions, 2018 Death Report on Inmates under Jail Jurisdiction or CJ-10A: Mortality in Correctional Institutions, 2018 Annual Summary on Inmates in Private and Multi-Jurisdictional Jails, depending on the type of jail. BJS needs to collect year-end and ADP counts through MCI-Jails to obtain a denominator for rate calculation. As with prisons, each decedent gets a separate death form: CJ-9: Mortality in Correctional Institutions, 2018 Death Report on Inmates under Jail Jurisdiction or CJ-10: Mortality in Correctional Institutions, 2018 Death Report on Inmates in Private and Multi-Jurisdictional Jails.

A. Justification

1. Necessity of the Information

BJS is authorized to collect these data by the Omnibus Crime Control and Safe Street Act of 1968, as amended (34 U.S.C. § 10132), which established BJS and authorized the collection and analysis of statistical information concerning the operations of the criminal justice system at the Federal, state, tribal and local levels (Appendix C).

BJS began the predecessor to MCI, the Death in Custody Reporting Program, in 2000 in response to the Death in Custody Reporting Act of 2000 (P.L. 106-297), and continued to collect these data when the law expired in 2006. The legislation was motivated by a concern about conditions in prisons and local jails and increased public and Congressional interest in oversight, in light of wrongful death lawsuits and national media attention on deaths in custody. DICRA was reauthorized in 2014 (Appendix D), but the Department of Justice (DOJ) has recently placed more emphasis on the compliance part of the law. As a federal statistical agency, BJS data may not to be used for enforcement purposes. DOJ therefore determined that the Bureau of Justice Assistance (BJA), as opposed to BJS, should manage collection of the data pursuant to the law, since that agency is not under similar requirements to collect data for statistical purposes only (Appendix E: Report of the Attorney General to Congress Pursuant to The Death in Custody Reporting Act, December 16, 2016).

BJA submitted 60-day (83 FR 27023) and 30-day (83 FR 44064) notices to the Federal Register, with the 30-day notice concluding on September 28, 2018. BJA submitted its collection package to OMB on October 23, 2018 (https://www.reginfo.gov/public/do/PRAViewICR?ref_nbr=201810-1121-004), and announced its intention to start its collection on law enforcement deaths on January 1, 2019, and follow with collection of data on deaths in state prisons and local jails on January 1, 2020. BJS will continue to collect deaths occurring at all levels of the federal system through its Federal Arrest-Related Death program.

BJA has requested that BJS continue to collect MCI data on deaths of prison and jail inmates that occur in the 2018 and 2019 calendar years, to allow for the continuity of the collection and to provide at least one year of comparison data (2019). Once BJA fully implements its DCRA information collection, BJS will discuss with OMB future mortality data collection by BJS. BJS is concerned about unnecessary duplication of burden, but is also concerned that the difference in goals and collection procedures between BJA and BJS, that the BJA collection will not by itself provide accurate national and state-level estimates of deaths in prisons and local jails.

There are no plans by BJA to publish or archive the data for the general public, but only to provide it to the Attorney General. This would be a great loss to the field and diminish our understanding of health and health care in correctional facilities. Currently, the MCI data are protected by BJS’s confidentiality statute (34 U.S.C. § 10231), which encourages high response rates since respondents are assured the data will only be released to the public in the aggregate. This statute does not apply to data collected by BJA.

BJA’s clearance package suggests that it will not verify whether facilities that do not send in death forms, had no deaths. This verification of no deaths is vital to differentiating between facilities with no deaths and those who just failed to report. Consequently, the true total count of deaths will never be known in a collection that does not perform the verification. Finally, BJA has no plan to collect data on year-end or average daily populations from jail facilities to serve as the denominators in calculating mortality rates in jails. Since BJS only obtains these counts through the MCI-Jails collection, cessation of MCI would negate the ability to calculate mortality rates. Over the next two years, BJS hopes to engage OMB in a dialogue to come to an agreement about collection of the data that will satisfy the desire to reduce burden, while continuing to use rigorous statistical methods.

An estimated 6,613,500 persons were under the supervision of U.S. adult correctional systems on December 31, 2016.¹ Of this total, about 1,505,400 persons were incarcerated in state and federal prisons and 740,700 in local jails. The MCI-Prisons and MCI-Jails collections are critical to describing movements of offenders through the criminal justice system. Data from these collections inform policymakers, correctional authorities, researchers, healthcare providers, and the general public on the causes and circumstances of mortality in correctional settings. Inmate mortality data may be used to develop policies and procedures, plan for the provision of physical and behavioral health services, and maintain critical oversight. As a federal statistical agency, BJS provides an important source of mortality data on those who die in custody.

Prior to the MCI, the public’s knowledge of prisoner deaths was limited to annual aggregate counts via the National Prisoner Statistics program (NPS, OMB clearance number 1121-0102) and ASJ, and were typically reported as a type of prisoner release. Furthermore, the data were limited to sex-based summary counts and only NPS provided a limited disaggregation of manners of death (e.g., homicide, suicide, natural causes, AIDS-related).

The MCI collection provides a unique opportunity to study inmate mortality in-depth because it collects individual-level data on deaths in state prison and local jails. BJS can track changes in mortality rates of imprisoned persons over time, and improve the public’s understanding about deaths of persons under the control of the correctional system by reporting on cause and circumstances surrounding the death. Through the MCI, it is known that natural causes (illnesses) account for the majority of both prison and jail deaths, with suicide as the second leading cause of death. Most illness deaths occur as a result of pre-existing conditions that pre-date decedents’ admission to prison or jail. Based on 17 years of mortality data from local jails, BJS has observed that in any given year, about 80% of the roughly 3,000 jails in the U.S. have no deaths in their custody.

Detailed information about prisoner deaths cannot be obtained through other, existing collections. The National Center for Health Statistics (NCHS) mortality files and the National Death Index (NDI), which rely on death certificates files in the U.S., do not identify criminal justice system involvement of the deceased. Other sources of data on mortality in the U.S. prison system also have significant limitations (see item 4 below, Efforts to Identify Duplication).

Under Title 34 of the United States Code, Section 10231 (see appendix C, attachment D) BJS collects MCI data for statistical purposes only and has procedures in place to guard against disclosure of personally identifiable information. MCI data are maintained under the security provisions outlined in U.S. Department of Justice regulation 28 CFR §22.23, which can be reviewed at http://bjs.ojp.usdoj.gov/content/pub/pdf/bjsmpc.pdf, (Appendix F).

2. Needs and Uses

BJS conducts the MCI to measure conditions of confinement and mortality throughout the criminal justice system and to integrate the analysis of mortality into other statistics on other criminal justice system outcomes, such as recidivism. BJS’s most basic need for the data is to enable it to track and report on all-cause and cause-specific mortality and mortality rates during incarceration in jails and prisons. MCI allows BJS to report on mortality outcomes, identify any trends or changes in mortality over time, and assess whether changes are due to changes in the composition of the custody populations or are associated with changes in correctional policy or facility conditions.

Recurring uses of the MCI prisoner collection

BJS publishes annual reports on mortality in state prisons and local jails that include national counts, percentages, and rates of death by demographic characteristics, cause of death, and state. For deaths in local jails, all jails in a single state are aggregated within a year to protect privacy, but since some states may not have many deaths overall, counts of deaths less than 4 are still suppressed in the released state-level statistics. Since 2013, BJS has reported 2 and 3-year moving averages (depending on the number of years in the reference period) by cause of death and inmate characteristics, which led to the finding that the prevalence of prisoner cancer deaths began to increase in 2008.² In contrast, mortality due to liver disease, which is the third leading cause of prisoner deaths, remained flat, with the liver mortality rates being nearly equal between each interval. By decomposing changes in overall mortality rates into compositional and group- and cause-specific rates, BJS can identify sources of changes in mortality rates.

Manner and cause of death are the cornerstone of all mortality studies, and the individual-level demographic information collected in MCI enables BJS to standardize the prisoner mortality rate to the U.S. resident population. The relationship between criminal behavior and mortality can be studied by comparing offense type, conviction status, and cause of death. For instance, BJS can assess any mortality differentials between violent offenders by type of offense (e.g. homicide offenders versus rape offenders) or between convicted and unconvicted jail inmates. Medical treatment variables, including mental health treatment, provide a sense of the health care afforded to prison and jail inmates prior to death.

Information about the location of deaths, particularly facility name and place of death in the facility, permits comparisons among facility types and linkage to the censuses of prison and jail facilities (OMB Control Nos. 1121-0147 and 1121-0100, respectively). Since the facility censuses obtain information on staffing levels, BJS can determine whether jails or prisons with a small staff to inmate ratio are less likely to experience a death than facilities that have staff supervising more inmates. Date of admission to the facility is used to calculate time served before death and can be used to calculate hazard rates, which address the problem of exposure time in prisons and local jails. The location and time of death allow BJS to describe the circumstances surrounding unnatural deaths (e.g., suicides, accidents and homicides) in prisons and jails by location and the time of day.

The pre-existing medical conditions and medical examiner questions allow BJS to examine differences in the types of deaths that occur in prison from those in jail. Since 90% of prison deaths are the result of an illness and therefore somewhat expected, fewer autopsies are performed of decedents in prisons. Conversely, the higher proportion of unnatural jail deaths results in nearly all jail inmates (95%) being autopsied³ , since jails assume they may be sued by family members.

In addition to an annual series on trends in correctional mortality, BJS has published special topical reports on medical causes of death in state prisons and local jails, including on suicides and homicides in correctional facilities, and natural deaths. The MCI data also inform BJS’s biennial report on HIV/AIDS in prisons. While data on prevalence, testing, and treatment of HIV/AIDS by state departments of corrections are obtained through the NPS collection, HIV-related death data come from MCI. These data have allowed BJS to chart a significant decrease in HIV-related deaths among the prison population over time as anti-retroviral drug treatments for the disease have become more widely used and effective.

Finally, the MCI-Jails population data are used annually in BJS’s Corrections Populations in the United States bulletins to provide year-end counts for local jail populations by state. These are combined with prison, probation, and parole counts for states to yield state-level estimates of the total number of people under correctional authority for a given year. These reports are available to the public through the BJS website (https://www.bjs.gov/index.cfm?ty=pbdc&dcid=243&iid=1).

Use of the MCI by other entities

Other entities rely on the MCI data for research, planning, and programmatic purposes. MCI data and statistical reports are used by the U.S. Department of Justice, U.S. Congress, local, and state correctional administrators, public health officials and practitioners, researchers, and special interest groups.

The Office of Justice Programs (OJP), within which BJS is located, has devoted effort to coordinate the work of all of its bureaus on a suicide prevention effort. The Federal Partners in Suicide Prevention, a federal working group sponsored by the Center for Disease Control and Prevention and Health and Human Services, reports on suicide awareness and prevention in resident and incarcerated populations. The group is particularly interested in incarcerated populations because the crude suicide rates for prisoners is higher than the crude suicide rate for the U.S. non-incarcerated resident population. This difference, however, disappears once the rates are adjusted to account for age, sex and race difference. This adjustment would not have been possible without data from the MCI.⁴

Congress used BJS reports as part of the justification for the passage of the 2013 DICRA. The legislation’s main sponsor cited BJS statistics on the decline in mortality in prisons, arguing that with detailed statistical data, policymakers at the local, State, and Federal levels can make informed judgments about the appropriate treatment of prisoners and develop ways to lower the prisoner death rate.

Additional examples of groups and individuals who have used MCI to address policy issues related to mortality include -

Public Health, Legal and Justice Policy Researchers:

• Dr. Ingrid Binswanger of the University of Colorado-Boulder used MCI data to publish an article on smoking cessation policies and cancer deaths in prisons.⁵ Former BJS Visiting Fellow Dr. Christopher Wildeman of Cornell University used MCI and NCRP data to analyze state-level differences for prisoners and their resident population counterparts after controlling for sex, age and race-ethnic origin differences.

• Three doctoral candidates have received enclave licenses from the University of Michigan’s National Archive of Criminal Justice Data (NACJD) to use MCI prisoner data in their dissertation work.

• The American Psychiatric Associations’ American Psychiatric Publishing Textbook of Suicide Assessment and Management, Second Edition (2012) used MCI data in a section on suicides in special populations.

• Psychiatrists from several teaching hospitals, such as Bellevue, St. Vincent’s and St. Luke’s Hospitals in New York City, have requested data on suicides in prisons (and jails).

• Staff at NCHS reported that the MCI collection is important and useful for understanding certain types of mortality that they cannot address with their national mortality files. (See item 4 below, “Efforts to Identify Duplication.”)

• The National Association of Medical Examiners (NAME) has endorsed the MCI as being the sole source of identifying inmate deaths. NAME used data from the MCI to urge members to revise state death certificates to indicate whether the deceased had recently been in law enforcement or correctional custody. Washington, DC is planning to revise its death certificate accordingly, with the expectation that other states will follow suit.⁶

• MCI data, both those published by BJS in the annual Mortality reports, and those accessed through NACJD, have been used MCI for media and academic articles. A bibliography for publications from the past 3 years is located in Appendix G.

Correctional administrators

• Three departments of corrections (DOCs) requested MCI data regarding inmate homicides (Georgia) and suicides (Wisconsin and Mississippi) to inform annual reporting.

• The Washington State Department of Health routinely incorporate MCI tables in its own annual reports.

• The Oklahoma State Department of Health Injury Prevention Service uses MCI tables in regular “Injury Updates” as part of its Injury Surveillance Program.

• In September 2018, a number of jails requested copies of their death records from 2009-2016 to respond to FOIA requests from media organizations. While BJS has successfully fought FOIA requests based on our pledge of confidentiality (34 USC § 10231), some states and local governments have laws that require the provision of data when it is requested by a member of the public. BJS provides a jail’s or DOC’s death records back to them as a courtesy; the jail or DOC must then decide whether it has to release its data to the public.

Efforts to improve the usefulness of the MCI data

In the past three years, BJS has worked to improve the MCI in ways that will enhance the collection’s usefulness as a research database. These efforts include:

Linking MCI records to the National Death Index (NDI)

Underlying causes are often missing in MCI cases because the MCI respondents are requested to list only the final cause of death⁷. In 2012, BJS linked MCI 2007, 2008, 2009, and 2010 death files to the NDI to determine how many MCI deaths had a matching record in the NDI and to assess cause of death as reported in both collections. Nearly all (94%) MCI records had a match in the NDI, and the final cause of death was accurately reported in most cases. The mean number of ICD-10 codes assigned to MCI records based on causes of death reported by correctional facilities was 1.1, while the matched records yielded a mean of 2.7 ICD-10 codes in the NDI. This information provides a more accurate picture of perimortem health conditions that contributed to the individual’s death.

In 2018, BJS funded an interagency agreement (IAA) with the Department of Health and Human Services (HHS) for $750,000 to link all retrospective MCI records to the CDC’s NDI. This new IAA will fund the linkage of the two databases to obtain NDI ICD-10 codes for all MCI deaths since 2000 (for jails) and 2001 (for prisons).

Appending U.S. Census Bureau facility codes

Over the last three years, BJS has continued to assign U.S. Census Bureau government facility identification numbers (GID) to the death records in MCI, to allow for linkage of these data to BJS’s Census of State and Federal Adult Correctional Facilities (OMB Clearance No. 1121-0147). The goal is to match prisoner deaths to the facilities where the inmates were housed prior to death, so that BJS can examine whether there are facility characteristics correlated with mortality rates, including staffing levels, existing consent decrees, programming options, and security levels. The process has been significantly complicated because the assignment of the GID was contingent upon the facility name in MCI. Since facility name is a descriptive literal entered by the DOC and jail respondents, there is a variation in how facility name is reported across the years. As a result, BJS has been unable to assign a GID code for a number of historical MCI deaths. Currently, if the MCI data collection agent cannot assign a GID to a facility name on a submitted death form, they flag this item for immediate data quality follow-up.

3. Use of Technology

The instruments used to collect the MCI data include the following forms (see Appendix A):

• CJ-4: Mortality in Correctional Institutions, 2018 Annual Summary of Inmate Deaths in State Prisons

• CJ-4A: Mortality in Correctional Institutions, 2018 State Prison Inmate Death Report

• CJ-9: Mortality in Correctional Institutions, 2018 Death Report on Inmates under Jail Jurisdiction

• CJ-9A: Mortality in Correctional Institutions, 2018 Annual Summary on Inmates under Jail Jurisdiction

• CJ-10: Mortality in Correctional Institutions, 2018 Death Report on Inmates in Private and Multi-Jurisdictional Jails

• CJ-10A: Mortality in Correctional Institutions, 2018 Annual Summary on Inmates in Private and Multi-Jurisdictional Jails

The majority of respondents submit MCI data to BJS through a web data entry form (see screenshots for 2017 data collection, since 2018 collection is not yet live, Appendix H), or, in the case of jurisdictions with large numbers of deaths, as a SAS dataset or an Excel spreadsheet. The current MCI web reporting system includes a set of online edit checks to identify data entry errors by respondents. BJS has also implemented computer methods for reviewing and editing the data in more detail. A set of decision rules are coded, and this code is run against the data to identify out-of-range or erroneous values and to assess the impacts of out-of-range values on quantities to be estimated. These methods are used to make decisions about priorities for follow-up contact with respondents.

Based on analytics and comments obtained from respondents during follow-up, the bulleted items that follow are some of the ways in which BJS has used information technology to reduce burden and improve data quality, agency efficiency and responsiveness to the public.

• Real time, “always on” data collection. With the start of a calendar year, respondents will have access to the web collection instrument and can report their data as soon as the data collection year opens on the MCI website. Historically, the web collection has a ‘soft’ opening in December, so respondents can begin entering records before the official opening of the collection year.

• Improvements in identifying the reporting year. Because the MCI collections are “always on”, the Web site will include buttons that allow MCI respondents to select the relevant year and access multiple years (including previous years) if needed.

• Pre-filled forms. Forms are prepopulated with respondent contact information, eliminating the need for respondents to enter this information unless there has been an update to contact information.

• Improvements in survey flow. The online data collection tool follows the paper forms, but enhancements to the web tool facilitate navigation through the form. Respondents will be led through the items in smaller segments rather than scrolling through the items on a single screen, minimizing the possibility of them inadvertently skipping an item. This simultaneously enhances data quality while reducing the burden of future data quality follow-up.

• Enhanced capacity to add death reports. Respondents are able to create new blank death records for data entry simply by choosing an “Add a Death Report” button. This is especially helpful for prisons, who often submit multiple deaths at a time.

• Enhanced capacity to review existing death reports. Respondents are able to easily locate existing records based on a combination of identification criteria (e.g., date of death, date of birth, etc.) or by a list of prisoner names. This can be useful if they need to provide mortality information for their jurisdiction to some other entity.

• Timeout warning so that important data are not lost. Warnings are sent to respondents if a web session is about to automatically timeout due to inactivity. This warning prevents data loss and eliminates re-entry of data.

• Real-time prompts that alert respondents of potential errors. The functionality of the Web forms alerts agencies to potential data problems. This includes soft prompts when respondents report improbable values on the prison death forms, ex. a birth year greater than admission to prison year. This system reduces data errors and item non-response.

• On-screen reporting capabilities. Upon completion of each form, respondents receive an on-screen report that summarizes their responses. This enhancement was designed in response to respondents’ interest in wanting to assess the completeness of their submission while allowing them to review and edit their entries prior to final submission.

• Explicit confirmation of form completion following online form submission. Upon submitting their data, respondents receive confirmation that their submission is complete for the relevant reference year.

• Paper versions of completed submissions for web respondents. Many respondents have expressed that, while they prefer to use the Web-based tool to enter their data, they also need paper copies for documentation. As result, the Web-based tool automatically generates .pdfs of completed survey forms that can be printed or saved.

• Real-time reporting to data collection agent of errors encountered by respondents. In 2009, the MCI data collection agent introduced an error log, which captures errors encountered by respondents as they maneuver within the MCI web site. This allows the data collection agent to identify and correct systemic issues, which in turn, has resulted in increased user satisfaction with web reporting. (This is a behind-the-scenes enhancement that does not affect the content of the instrument.)

• Continual additions to the frequently asked questions (FAQs) document. The FAQ is available on the public-facing MCI Web site (https://bjsmci.rti.org) and can be accessed without user credentials. BJS and the data collection agent modify the content in response to evolving needs (Appendix I).

BJS has been phasing out the use of paper forms for collecting MCI data. Prior to 2012, respondents received a mailing packet that included a cover letter, reporting instructions, and copies of the survey forms. Paper surveys were not included in the annual mailings announcing the start of the 2012 data collection, to encourage respondents to report their data through the MCI website. Web-based data submissions increased from about 40% to 64% in 2013. As a result, paperless data submission became a permanent feature of MCI.

The online system allows BJS’s data collection agent to generate weekly progress reports, so BJS can assess response rates in real time and determine the progress of each jail reporting unit. For reporting year 2016, only 3% of respondents to the MCI submitted data via U.S. mail or over the telephone, with 95.3% responding on the web, and the remaining 1.7% submitting forms over encrypted e-mail.

4. Efforts to Identify Duplication

The MCI is not duplicated by any other federal agency or program at this time. While there are other sources of mortality data related to the topic of prisoner and jail inmate deaths, none are as comprehensive as the MCI. Since the beginning of the MCI, BJS has undertaken efforts to identify other national data collections that could be redundant with the MCI. While states like California, Texas, and Washington have state-level laws that mandate the collection of this information and the Arizona DOC issues a press release in the event of a prisoner death, the MCI is the only national database that collects prisoner deaths for all state and local correctional systems in the United States.

BJS has identified three national databases with death data, but there are significant differences between these systems and the MCI. The national collection systems that capture death data are the NCHS’s NDI (as mentioned in section 2, under ‘Necessity of the Information’), the National Violent Death Reporting System (NVDRS), and the Social Security Administration’s (SSA) Death Master File (DMF).

The NDI is a subcomponent of the National Vital Statistics System, and has total coverage of known deaths in the United States. The NDI is based on death certificates submitted by the states. However, with a few locality-based exceptions, such as Washington, D.C., death certificates do not indicate whether the deceased had been in correctional custody and in turn, the NDI is unable to independently identify persons who were in custody at the time of their death. BJS has had discussions with both NCHS and NAME about the possibility of adding such an identifier to the U.S. Standard Death Certificate, and while states may plan to incorporate such a flag on an ad hoc basis, any proposed national changes have to be approved by the World Health Organization, which proposes and votes on changes on a decennial schedule. Per NCHS, the next opportunity will not present itself for several years.

SSA collects information on deaths through the DMF to remove decedents from the Social Security benefits rolls, but these data do not include either a cause of death (COD) or an indicator of criminal justice involvement.

The Center for Disease Control and Prevention’s NVDRS tracks homicides and suicides in 40 states and the District of Columbia. The NVDRS is a state-based surveillance system that triangulates data from death certificates, medical examiner/coroner reports and police reports to create a database on violent deaths. The NVDRS excludes deaths by suicide or homicide in correctional settings in its reporting, but even if these deaths were included, they would not capture the majority of deaths in prisons or jails, since 51% of jail deaths and 88% of prison deaths are due to illness.

BJS will assess the overlap and quality of the data once BJA begins its collection of prison and jail deaths in 2020, and the need for BJS to continue MCI, revise the collection, or discontinue it in its entirety. This decision will be based on whether BJA central state respondents are able to obtain participation from all local jails on a quarterly basis. One option BJS is investigating is to keep the MCI an active collection, but only request identification variables of decedents to document fact of death from jails and state departments of corrections, and then link those data to NDI to obtain cause of death. This will decrease the burden associated with data quality follow-up for MCI, since much of that involves waiting for death certificates and autopsies to be completed and sent to the respondents. In that case, BJS would have a break in series introduced by the different collection mode for cause of death. The longitudinal database of all deaths with NDI ICD-10 codes from 2000 forward that BJS will create through its IAA with NDI will solve the problem of a break in series, in addition to providing more detailed information about the perimortem health of decedents.

5. Impact on Small Businesses

Not applicable. This statistical collection does not involve small businesses or other small entities.

6. Consequences of Less Frequent Collection

If the collection is done on less than an annual basis there would be a loss of information. MCI respondents have relayed that medical records and death certificates are often shipped off site within a comparatively short period of time, usually within a year of the death. If the data were collected on less than an annual basis, some respondents would no longer be able to access this critical piece of data. Other respondents would be required to go to off-site storage to obtain records, typically at an additional cost to the respondent. This would likely result in a negative effect on participation in the collection.

Collecting death records on a less than an annual basis would also compromise BJS’s capacity to report in a timely manner on trends in deaths in custody. It would also pose challenges for data collection because of how respondents store older data. It would impose additional costs associated with restarting the collection at various intervals. Less than annual collection would delay publication of mortality data and collection on other than an annual cycle and would make it difficult for BJS to maintain high levels of participation. Respondents know that the collection is annual and over the years have developed internal procedures to facilitate responding to the MCI. Every year since collection began, BJS has collected data from all 50 state Departments of Corrections and the vast majority of local jails.

Turnover among respondents to the collection may also potentially negatively impact response rates and increase follow-up costs if the collection were fielded less frequently. With annual collection, BJS learns about pending turnover during routine data collection and verification calls (see Part B, section 2 for more information) and plans for this.

7. Special Circumstances Influencing Collection

Not applicable. These data will be collected in a manner consistent with the guidelines in 5 CFR 1320.6.

8. Consultation Outside the Agency

The research under this clearance is consistent with the guidelines in 5 CFR 1320.6. The 60- and 30-day notices for public commentary have been published in the Federal Register. In renewing the data collection procedures, BJS has consulted with various experts to obtain their views on the survey instruments. BJS consulted the following correctional officials and public health experts:

• Dr. Ingrid Binswanger, MD, University of Colorado

• David Espey, M.D. Division of Cancer Prevention and Control, Acting Director. Centers for Disease Control and Prevention

• Dr. Roger Mitchell, Chief Medical Examiner, Washington, D.C.

• Dr. Margaret Warner, National Center for Health Statistics

• National Association of Medical Examiners

BJS maintains frequent contact with MCI data providers and data users in an effort to improve data collection, reporting procedures, data analysis, and data presentation.

BJS received one comment in response to the 60-day Federal Register Notice (83 FRN 46190), (Appendix J). The comment, from the president of Citizens United for the Rehabilitation of Errants (CURE), requested that BJS expand the MCI to cover deaths of all persons under the Civil Rights Institutionalized Persons Act (CRIPA), including persons in correctional facilities, nursing homes, mental health facilities, and facilities for intellectually or developmentally disabled persons. In the past, BJS has met with CURE to discuss the feasibility of collecting data on persons held under civil commitments by the states. At this time, however, BJS does not have the resources to extend collection of deaths to these institutions.

9. Payment or Gift to Respondents

Participation is without direct payment or compensation.

10. Assurance of Confidentiality

Pursuant to 34 U.S.C. § 10134, data collected by BJS shall be used only for statistical or research purposes and shall be gathered in a manner that precludes their use for law enforcement or any purpose relating to a private person or public agency other than statistical or research purposes. Further, as required by the confidentiality provisions of 34 U.S.C. § 10231, no officer or employee of the Federal Government, including BJS employees or its contractors, may use or reveal any research or statistical information furnished in connection with a BJS data collection, including data identifiable to any specific private person, by any person for any purpose other than the purpose for which it was obtained.

BJS will not publish any data identifiable to a private person (including respondents and decedents). In addition BJS does not report data by individual facility in which deaths occur. Mortality records are held in an enclave at the National Archive of Criminal Justice Data (NACJD), located at the University of Michigan. Researchers requesting use of the MCI death records must comply with strict requirements. Requests for private information through the Freedom of Information Act will be forwarded to the Office of Justice Programs’ General Counsel for determination of data to be released.

BJS provides the following confidentiality pledge to MCI respondents:

The Bureau of Justice Statistics (BJS) is authorized to conduct this data collection under 34 U.S.C. § 10132. BJS will protect and maintain the confidentiality of your personally identifiable information (PII) to the fullest extent under federal law. BJS, its employees, and its contractors will use the information you provide for statistical or research purposes only pursuant to 34 U.S.C. § 10134, and will not disclose information in identifiable form to anyone outside of the BJS project team without your consent. All personally identifiable information collected under BJS’s authority is protected under the confidentiality provisions of 34 U.S.C. § 10231. Any person who violates these provisions may be punished by a fine up to $10,000, in addition to any other penalties imposed by law. Further, per the Cybersecurity Enhancement Act of 2015 (6 U.S.C. § 151), federal information systems are protected from malicious activities through cybersecurity screening of transmitted data. For more information on how BJS and its contractors will use and protect your information, go to https://www.bjs.gov/content/pub/pdf/BJS_Data_Protection_Guidelines.pdf.

11. Justification for Sensitive Questions

Items regarding cause and circumstances of death are essential to understanding mortality in the criminal justice system. Such items may be considered sensitive to correctional and law enforcement administrators; however, this information is a matter of public record, as part of reports by medical examiners and coroners. BJS guards access to these data closely. Researchers wishing to use the MCI death records must comply with the standards of the data enclave at the NACJD, including travel to the University of Michigan and disclosure review of all analysis results, as well as all materials brought into and taken from the enclave room.

12. Estimate of Respondent Burden

MCI collects data from prison administrators in all 50 states and from administrators of approximately 3,000 jails. Estimates of the annual burden on respondents are based on the number of hours needed to review the instructions associated with the instruments, search existing data sources, obtain the information necessary to complete the data collection instruments, and provide data quality follow-up responses and verification.

Burden estimates are based on data reported by prison respondents for prior years. A general summary of how burden estimates were calculated is provided in Table 1, with more detailed text below.

Table 1. Summary of Total Respondent Burden for MCI Data Collection

Respondent	Type of contact	Number of total data respondents	Number of responses	Average reporting time (minutes)	Total burden (hours)
State prisons
	Annual summary form (NPS-4) submission – online and mail	50	50	5 minutes	4 hours
	Death record form (NPS-4A) submission – online and mail	50	3,750	30 minutes per death	1,875 hours
	Data quality follow-up – email and telephone	50	46	15 minutes	12 hours
	Initial verification call – telephone	50	50	8 minutes	7 hours
Total for state prisons		50	3,800		1,898 hours
Local jails
	Annual summary form (CJ9A, CJ-10A) submission – online and mail	3,000	3,000	15 minutes	750 hours
	Death record form (CJ-9, CJ-10) submission – online and mail	3,000	1,100	30 minutes per death	550 hours
	Data quality follow-up – email and telephone	3,000	900	15 minutes	225 hours
	Initial verification call – telephone	3,000	3,000	8 minutes	400 hours
Total for local jails		3,000	4,100		1,925 hours
Total hours for MCI (state prison and local jail)		3,050	7,900		3,823 hours

State Prison Respondent Burden. Reporting time estimates from state prison authorities are based on previous MCI data collection cycles. Based on 2016 data, 50 DOCs submitted data on 3,750 inmate deaths. The average response time for the death report form (NPS-4A) is 30 minutes per death. Given these assumptions, we expect that in any data collection year:

• 50 state prison respondents will submit 3,750 death reports, with an average response time of 30 minutes per death report form, for a total burden of 1,875 hours.

Each state prison respondent will need to fill out a one-item annual summary form (NPS-4) with an estimated reporting time of 5 minutes. We expect the response time to remain constant:

• 50 state prison respondents will each submit an annual summary form, with an average response time of 5 minutes per annual summary form, for a total burden of 4 hours.

The 50 state prison respondents will be called to participate in the verification call (with an estimated reporting time of 8 minutes) prior to the start of the data collection. We expect that in any data collection year:

• 50 state prison respondents will have an average response time of 8 minutes per verification call, for a total burden of 7 hours.

Data quality follow-up occurs between May and July of each collection year and involves confirming, clarifying or correcting data reported on the NPS-4 and NPS-4A forms submitted by the respondent. Not all DOC records submitted require data quality follow-up. An average of 46 respondents require some level of follow-up annually, which averages about 15 minutes per respondent, for a total burden of 12 hours.

Thus, the total burden of hours associated with the prison data collection is 1,898 hours, with an average burden of 38 hours per year across the 50 state prison respondents. This is an increase in burden from the 2016 MCI clearance package, due to an increase in the number of deaths reported.

Local Jail Respondent Burden. Reporting time estimates from local jail authorities are based on previous MCI data collection cycles. Based 2016 data, approximately 3,000 DOCs submitted data on 1,100 inmate deaths. The average response time for the death report form (CJ-9 or CJ-10) is 30 minutes per death. Given these assumptions, we expect that in any data collection year:

• 3,000 local jail respondents will submit 1,100 death reports, with an average response time of 30 minutes per death report form, for a total burden of 550 hours.

Each local jail respondent will need to fill out a five-item annual summary form (CJ-9A or CJ-10A) with an estimated reporting time of 15 minutes. We expect the response time to remain constant:

• 3,000 local jail respondents will each submit an annual summary form, with an average response time of 15 minutes per annual summary form, for a total burden of 750 hours.

The 3,000 local jail respondents will be called to participate in the verification call (with an estimated reporting time of 8 minutes) prior to the start of the data collection. We expect that in any data collection year:

• 3,000 local jail respondents will have an average response time of 8 minutes per verification call, for a total burden of 400 hours.

Data quality follow-up occurs between May and July of each collection year and involves confirming, clarifying, or correcting data reported on the CJ-9/CJ-9A or CJ-10/CJ-10A forms submitted by the respondent. Not all local jail records submitted require data quality follow-up. An average of 900 local jail respondents require some level of follow-up annually, which averages about 15 minutes per respondent, for a total burden of 225 hours.

The total burden of hours associated with the local jail data collection is 1,925 hours, with an average burden of 39 minutes per year across the 3,000 local jail respondents. This is an increase in burden from the 2016 MCI-Jails clearance package (OMB control number 1121-0094), due to an increase in the number of deaths reported and in the number of local jail respondents who will need to be contacted for data quality follow-up, based on recent years’ data collections.

Combined State Prison/Local Jail Burden estimate. The total burden estimate for MCI across all 3,050 state prisons and local jails is 3,823 hours per year.

13. Estimate of Cost Burden

The costs to respondents incurred as a result of participating in this data collection are costs that would be incurred in the normal course of daily operations. Assuming a pay rate approximately equivalent to the GS-12 / 01 level ($72,168 per year), the estimated agency cost of employee time would be approximately $34.58 per hour.

Fifty departments of corrections and 3,000 local jails will be asked to participate in this activity each year. For the departments of corrections, the total cost is estimated at $65,633 ($34.58*1,898 hours), or $1,313 per state. For local jails, the total cost is estimated at $66,567 ($34.58*1,925 hours), or $22 per local jail.

14. Estimated Cost to Federal Government

The estimated costs for collection, processing, and dissemination of the MCI data in 2019 is $867,141, including:

$740,677—RTI, Inc. (MCI data collection agent)

$679,148 for data collection, data processing, computer programming, providing data, furnishing publication-ready tables, indirect personnel costs, and program management

$43,340 for nosology coding of MCI deaths

$18,189 for charges -- costs related to postage, telephone calls, printing, etc.

$126,464 -- Bureau of Justice Statistics

60% GS-13, Statistician ($60,000)

3% GS-15, Supervisory Statistician ($4,000)

5% GS-13 Editor ($5,000)

Other editorial staff ($5,000)

Front office staff (GS-15, SES, Director) ($2,000)

Fringe benefits (@28% of salaries -- $21,280)

Other administrative costs (@30% of salary & fringe $29,184)

15. Reason for Change in Burden

The total respondent burden has increased by 2,100 hours from the previous OMB request for clearance for this collection, due to the addition of the local jail MCI data collection, an increase in the number of overall deaths in state prisons and local jails, and an increase in the number of local jail respondents who require data quality follow-up after submission.

16. Project Schedule and Publication Plan

BJS does not require that respondents wait until the end of the calendar year in which the death occurred to submit a death form – respondents can do so on a rolling basis once they have enough information to complete the form. Collection of the MCI death forms generally begins in February of the year of death, and concludes in August of the following year. Annually, MCI verification calls are scheduled to begin either in late October or early November of the year of death (for deaths occurring in 2018, verification calls would begin in October or November of 2018, Appendix K). In January of the year following the year of death (2019 in the previous example), local jails and departments of corrections will receive email (Appendix L) or postal (Appendix M) invitations to complete the death forms and Annual Summary Forms (ASF) for the previous year. Email or postcard reminders (Appendix N) will be sent to jurisdictions that do not complete the online forms by March. A second reminder (Appendix O) along with printed forms will be mailed in April.

Data quality follow-up is conducted through email and phone calls (Appendix P) throughout the data collection period until July of the year following the year of death. Reminder mailing and data quality follow-up for MCI death forms will span from March through August 2019. The field work for death forms requires more time than the ASF, because respondents must locate autopsy or medical examiner reports to ensure complete cause of death information on the death forms. Final analysis data files and associated documentation are delivered to BJS by September 30 of the year following the year of death, and the final tables of mortality statistics, by November 30 of that next year.

A summary of the data collection for MCI, using deaths that occurred in 2018 as an example, is located in Table 2.

Table 2. MCI project schedule for collection of 2018 deaths

Year of death

Year of death + 1

Year

2018

2019

Task

Month

1

2

3

4

5

6

7

8

9

10

11

12

1

2

3

4

5

6

7

8

9

10

11

12

Submission of 2018 death forms to RTI

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

Coding of cause of death to ICD-10

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

Verification calls for 2018 facility frame

X

X

X

Email invitations for 2018 ASF

X

Email/postcard reminders for 2018 ASF

X

2nd reminder/printed form for 2018 ASF

X

Data quality follow-up for 2018 deaths

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

Reminder for DQ for 2018 deaths

X

X

X

X

X

X

Deliver final 2018 analysis files to BJS

X

Deliver 2018 mortality tables to BJS

X

BJS’s plans for products and publications from MCI data over the next 3 years fall into three categories: BJS bulletins, BJS special topic reports, and technical reports. The products include the following (a calendar of proposed publications is located at the conclusion of this section):

BJS Bulletins

Annually, BJS releases statistical tables that report on trends in deaths in local jails and deaths. These data provide a “first cut” from an annual collection, and focus on the effects of changes in the composition of criminal justice populations on the overall change in mortality rates. More specifically, the bulletins/statistical tables will report, annually, on changes in the overall mortality rate and number of deaths, as well as mortality rates by characteristics such as age, sex, race, and jurisdiction in which deaths occurred. BJS will continue to use the MCI data in the production of annual statistical tables. The 2016 tables have been delayed due to other BJS priorities, but are expected to be published in early 2019:

• Mortality in state prisons, 2001-2016 (expected release spring 2019)

• Mortality in local jails, 2000-2016 (expected release spring 2019)

• Mortality in state prisons, 2000-2017 (expected release winter 2020)

• Mortality in local jails, 2000-2017 (expected release winter 2020)

• Mortality in state prisons, 2000-2018 (expected release winter 2021)

• Mortality in local jails, 2000-2018 (expected release winter 2021)

• Mortality in state prisons, 2000-2019 (expected release winter 2022)

• Mortality in local jails, 2000-2019 (expected release winter 2022)

The annual bulletins also provide updates to previous years’ statistics by incorporating delayed data submissions. These statistics are consistent with the Committee on National Statistics (CNSTAT) recommendation to produce mortality data on a timelier schedule.⁸

BJS special topic reports

Periodically, BJS produces special topic reports from its data collections. These reports address a specific issue in more depth than can be addressed in the bulletins. BJS plans special topic report:

• Intoxication deaths in state prisons and local jails, 2000-2016 – This paper will examine the effects of drug and alcohol abuse and mortality in prisons and jails. Using data from the NDI-MCI match, BJS will identify causes of death that are have a strong association with a history of drug or alcohol abuse. The analysis will use the NDI-MCI linked files, because the linked death file has richer death data than the unlinked MCI files. The analysis will select International Classification of Disease, 10th revision (ICD-10) codes that are indicate deaths associated with acute intoxication (ex. Cocaine toxicity) as well as deaths associated with chronic disease or conditions associated with a history of substance abuse (ex., Cirrhosis)⁹ (Expected release 2020).

Technical reports

• BJS plans to undertake methodological work to expand BJS’s understanding of inmate mortality by linking MCI to other data sources. In particular BJS’s facility censuses for state prisons and local jails, will be fielded in 2019 and BJS proposes to examine issues related to facility characteristics, such as staffing composition, programs offered to inmates, inmate characteristics (e.g. race/ethnicity, number of juveniles held and custody classification), age of facility and whether the facility has a geriatric wing in relation deaths occurring at the facility. The report will also examine what percentage of facilities did not report a death in recent years and how the characteristics of these facilities differ from facilities that consistently report deaths to the MCI (Expected release: 2021).

Online dissemination tools

Due to the nature of the individual-level records in MCI and the sensitivity of the topic, the National Archive of Criminal Justice Data (NACJD) has classified the archived MCI data as a restricted dataset, available only through the enclave. This classification means that any member of the public who wishes to get access to the data must obtain approval or a waiver from an institutional review board (IRB), and write a description of their project to justify use of the data. Once a user license is granted, the researcher must travel to Ann Arbor, Michigan, and analyze the data onsite. Data sets cannot be removed from the enclave, and all output undergoes a disclosure avoidance review by NACJD staff prior to results being removed from the enclave. While this set-up protects the MCI data from disclosure, it also significantly limits the number of users -- only three students have accessed them since the data have been made available.

Over the next 3 years, BJS will work with NACJD to create restricted (requiring IRB approval or waiver and an application to NACJD to gain access) and public-use (freely-available) versions of the data files that will allow users who are not able to access enclave-level data access to masked MCI data files. This will be accomplished by stripping all personally identifiable information (PII) from the file, limiting the variables on the dataset, and aggregating variables to the state or national level to protect privacy. For example, while the restricted file may contain the exact year of birth for age at death calculation, the public file will have age at death calculated and aggregated into age ranges. Details on prior medical treatment and ICD-10 codes for COD may be suppressed from both the restricted and public-use archive files, with only a general COD indicator variable included. Geographic level of aggregation is still to be determined, but it is likely that the restricted will be aggregated at the state level, while the public-use file may only be at the national level. This would mean that only those users gaining access to the enclave file would be able to get facility-level data. These suppression techniques may limit the research uses of the MCI data, but should allow for more persons to access it.

Table 3. BJS Calendar for MCI Publications/Products

Type of BJS publication	Title/topic of publication/product	Estimated publication date
Annual bulletin	Mortality in state prisons, 2001-2016	Spring 2019
Annual bulletin	Mortality in local jails, 2000-2016	Spring 2019
Annual bulletin	Mortality in state prisons, 2001-2017	Winter 2019
Annual bulletin	Mortality in local jails, 2000-2017	Winter 2019
Annual bulletin	Mortality in state prisons, 2001-2018	Winter 2020
Annual bulletin	Mortality in local jails, 2000-2018	Winter 2020
Annual bulletin	Mortality in state prisons, 2001-2019	Winter 2021
Annual bulletin	Mortality in local jails, 2000-2019	Winter 2021
Special topic report	Intoxication-related inmate deaths in state prisons and local jails, 2000-2016	Winter 2020
Technical report	Facility-base mortality: linking the MCI and 2019 state prison and jail facility censuses	Winter 2021

BJS will continue to archive MCI data at NACJD on an annual basis, submitting the data files for a given year concurrent with its publication of statistical tables update for that year. Statistical tables for a given calendar year are published in the fall of the following calendar year, given the roughly 18-month period to collect MCI data. (See Part B, Item 2, Procedures for Information Collection).

17. Expiration Date Approval

The OMB Control Number and the expiration date will be published on all forms given to respondents and on the web portal used for electronic submissions of prison death data.

18. Exceptions to the Certification

There are no exceptions to the Certification Statement. The Collection is consistent with the guidelines in 5 CFR 1320.9.

1 Correctional Populations in the United States, 2016 (BJS web, NCJ 251211, April 2018)

2 Noonan, M.E. and Ginder, S. (2015). Mortality in Local Jails and State Prisons, 2000-2013. U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics. Washington, DC. NCJ 248756

3 Zeng, Z., et al (forthcoming). Assessing Inmate Cause of Death: Deaths in Custody Reporting Program and National Death Index. U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics, Washington, D.C.

4 Noonan, M.E. and Ginder, S. (2015). Understanding Mortality in State Prison: Do Male Prisoners Have an Elevated Risk of Death? Justice and Public Policy. Vol. 16. P.65-80.

5 Binswanger, I.A., Carson, E.A., Krueger, P.M., Mueller, S.R., Steiner, J.F., Sabol, W.J. (2014) Prison tobacco control policies and deaths from smoking in United States prisons: population based retrospective analysis. British Medical Journal. Aug 5;349:g4542.

6 Dr. Roger Mitchell, Chief Medical Examiner, Office of the Chief Medical Examiner, Washington, DC. Personal conversation, July 2015.

7 Zeng, Z. et al. (2016) Assessing Inmate Cause of Death: Deaths in Custody Reporting Program and National Death Index. NCJ 249568, U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics. Washington, DC.

8 Panel to Review the Programs of the Bureau of Justice Statistics, National Research Council. "Abstract." Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics. Washington, DC: The National Academies Press, 2009.

9 Shield, K.D., Parry, C. and Rehm, J. (2013). Focus On: Chronic Diseases and Conditions Related to Alcohol Use. Alcohol Research: Current Reviews. 35(2), pp. 154-173.