GDC Release-by-Death: a single-methodology in-custody death series for Georgia (CY2015-CY2025, FY2016-FY2026)

Georgia's in-custody death figures have until now come from two different counting methods stitched together at 2020: the UCLA Law Behind Bars roster for 2015-2019 and GPS's own registry for 2020 onward. Any rise measured ACROSS that seam is contestable as an artifact of method — better counting rather than more dying.The GDC release-by-death series has no such seam. Every figure in it is produced by the same office (GDC Office of Information Technology, Data Management Section), in the same report series ("Inmate Statistical Profile — Inmates Released During "), from the same field (actual release type code 62, Death), in the same table, on roughly the same page, every year from CY2015 to CY2025. Male, female and total are printed separately and male+female sums to the stated total in all 22 reports examined — no figure in this collection is inferred.That single-methodology property is the entire point. A trend computed within this series is a trend in deaths. A trend computed by splicing the UCLA roster to the GPS registry is a trend in deaths plus a trend in record-keeping, and the two cannot be separated after the fact.GDC also has a statutory duty to count deaths in its custody. Concealment is therefore possible but bounded: the six deaths GDC refused to NAME in 2025 were still inside the 301 it COUNTED. This is an administrative RELEASE-REASON count produced by GDC's Data Management Section — the number of people whose release from GDC custody was recorded as release type 62, Death. It is NOT an autopsy finding, NOT a cause-of-death determination and NOT a manner-of-death classification.

The Georgia Department of Corrections publishes its "Profile of Inmate Releases" in two separate editions: a calendar-year edition covering 1 January to 31 December, and a state fiscal-year edition covering 1 July to 30 June. Because the two editions share six months, a calendar-year figure and a fiscal-year figure must never be added together, placed in one time series or compared with each other.

From 2020 through 2025, the Georgia Department of Corrections' calendar-year count of releases by death agreed with GPS's own name-level registry of in-custody deaths to within five in every year and exactly in two, with six-year totals of 1,694 for the agency and 1,702 for the registry. The agency's count for 2025, 301, included the six decedents whose names it refused to release, so the gap in its disclosure is a failure to name the dead, not a failure to count them. These figures count releases recorded as deaths, not causes of death.

For 2015 through 2019, the Georgia Department of Corrections counted 863 releases by death, while the UCLA Law Behind Bars roster names 791 deaths: the roster is short by 72, or 8.3% of the agency's count, and was short in every one of the five years. The agency's figure should be used for any Georgia in-custody death total before 2020, and the UCLA roster for names, facilities and individual circumstances. The UCLA roster must never be added to or compared with GPS's own registry, which covers separate years by a separate method.

On the Georgia Department of Corrections' own calendar-year count of releases by death, the annual average rose 63.6%, from 172.6 a year in 2015-2019 to 282.3 a year in 2020-2025, and the highest year before 2020 (196, in 2018) was still below the lowest year from 2020 on (252, in 2022). The count comes from the same office, report and field every year, so the rise does not come from a change in who was counting or how. These are calendar-year counts of releases recorded as deaths, not causes of death, and must not be extended with fiscal-year figures.