How do advocacy reports and government records classify causes of death in detention, and where can researchers access the underlying data?
Executive summary
Advocacy reports and government records use different classification systems and sources to label causes of death in detention: government data collections (DCRP/MCI/DCRA) rely on reporting forms, ICD coding, and agency submissions, while advocacy groups often reclassify or contextualize deaths using autopsies, legal reviews, and open‑source investigations [1] [2] [3]. Both streams are publicly accessible but government datasets are fragmented and incomplete—GAO and BJS reviews find large gaps—so researchers must combine official files (BJS/MCI, BJA/DCRA) with advocacy databases and the National Death Index to approach a full picture [4] [5] [6].
1. How government programs classify causes of death: forms, ICD codes, and “manner” fields
Federal and state collections capture cause-of-death through structured death records submitted under the Mortality in Correctional Institutions (MCI, formerly DCRP) and the Death in Custody Reporting Act (DCRA) processes, which include fields for cause, autopsy status, and circumstances; illness deaths in BJS datasets are coded to ICD‑10 while some DCRP elements historically did not assign ICD codes to unnatural deaths, creating classification inconsistencies [1] [6]. The DCRA definition of detained/deceased scope is explicit about locations and contractual facilities, and federal agencies submit through BJS surveys or the BJA Performance Measurement Tool, with some records marked “Unavailable, Investigation Pending” when causes await autopsy or review [7] [8].
2. How advocacy reports classify and reclassify causes: context, forensic review, and pattern‑spotting
Advocacy organizations and investigative projects reclassify deaths by re‑examining autopsy reports, medical records, and public records; they emphasize contributing factors such as inadequate medical care, restraint use, and delays in treatment that government cause codes may list only as secondary or omit—examples include ICE detainee death reviews where advocates found inadequate care contributed to many deaths [3]. Nonprofit and media databases (e.g., FatalEncounters-style projects) supplement official records with open‑source searches, obituaries, and FOIA-obtained documents to assign proximate causes or highlight suspected preventable deaths in ways that official ICD labeling may not capture [9] [3].
3. Where researchers can access the underlying government data
Researchers can access BJS/MCI datasets and publications through the Bureau of Justice Statistics website and Data.gov catalogs for the Mortality in Correctional Institutions series, and DCRA submissions are available via the Bureau of Justice Assistance’s reporting portal and the Department of Justice program pages; federal DCRA guidance and FAQs also describe the survey scope and reporting responsibilities [2] [1] [7]. For death certificate coding and multiple‑cause listings, the National Death Index (NDI) is the canonical source for ICD‑10 coded causes and can be matched to DCRP/MCI records, a method used in BJS technical assessments [6] [10].
4. Data quality, gaps, and why classification disagree‑ments persist
Multiple federal reviews document serious completeness and consistency problems: GAO found that many state records submitted to DOJ were missing required elements and that DOJ had not fully studied state data because of incompleteness, and BJA itself cautions that its DCRA portal records may be incomplete or not official statistics [4] [5] [8]. BJS methodological reviews show differing code limits and assignment rules between DCRP and NDI (e.g., different maximum numbers of ICD codes and different rules for choosing an underlying cause), which produces divergent cause labels when the same death appears in two systems [6] [11].
5. Practical steps for researchers assembling cause‑of‑death evidence
Combine sources: download BJS/MCI public files and BJA DCRA extracts, request NDI matches for ICD accuracy, and supplement with advocacy databases and FOIA/autopsy releases for circumstances and contributing factors; be explicit about provenance because official cause codes, investigation‑pending markers, and advocacy reclassifications answer different questions [2] [8] [3]. Treat “unavailable/investigation pending” and missing data as substantive findings—not mere metadata—because GAO and DOJ reviews show that underreporting and incomplete records are systemic and can bias any analysis unless triangulated [8] [4] [5].