Case Study 37.1 — The Untested-Kit Backlog: Detroit's 11,000 Kits and the Reckoning That Followed

A note on sourcing and tone. The facts below are drawn from the well-documented public record of the Detroit (Wayne County, Michigan) untested-sexual-assault-kit discovery and testing initiative, with the parallel Cuyahoga County (Cleveland), Ohio effort noted as corroborating public record. This case study concerns thousands of real victims; we treat the subject soberly, confine ourselves to documented facts and figures reported publicly, and use the case to teach an institutional forensic-science failure — not to sensationalize the harm done. Where a figure is reported in ranges or "approximately," we say so.

Background

In 2009, during a tour of a Detroit Police Department property-storage facility, officials discovered something that would become a national emblem of institutional failure: row upon row of sexual-assault evidence kits — ultimately counted at roughly 11,000 — that had been collected from victims and then stored, untested, some for more than a decade. Each kit represented a person who had reported a sexual assault and then submitted to the lengthy, invasive medical-forensic examination this chapter describes (§37.2–37.3), enduring it in the reasonable belief that the evidence would be used. For thousands of them, it never had been.

The discovery prompted one of the most studied criminal-justice reform efforts in recent American history. Under the leadership of the Wayne County Prosecutor's Office, in partnership with researchers (including a federally funded action-research project that documented the effort in detail), forensic laboratories, and outside funders, Detroit undertook to inventory, test, and — where leads emerged — investigate and prosecute the cases the kits represented. The initiative became a template, and a cautionary tale, for jurisdictions across the country.

Detroit was not unique; it was simply early and large. Cuyahoga County, Ohio (Cleveland) mounted a comparably ambitious and well-documented effort after the Ohio attorney general called on law-enforcement agencies statewide to submit their previously untested kits for testing. Cuyahoga County sent thousands of shelved kits to the state laboratory and stood up a dedicated Sexual Assault Kit Task Force to investigate the resulting CODIS leads and bring charges. Together, Detroit and Cuyahoga County form the documented core of what a national accounting — driven by investigative journalism, advocacy organizations, and federal programs such as the Sexual Assault Kit Initiative (SAKI) — revealed to be a backlog of hundreds of thousands of untested kits nationwide.

The forensic evidence

The forensic story here is, pointedly, not one of scientific difficulty. It is a story about evidence that worked as soon as anyone used it.

  • The kits were largely testable. A central, sobering finding of the testing initiatives was that decades-old kits, when they had been stored adequately (dry, as §37.3's collection principles require), frequently yielded interpretable DNA profiles. DNA is a relatively stable molecule under good storage conditions, and the passage of years did not, in large numbers of cases, defeat analysis. This deepens rather than excuses the failure: the evidence had been usable the whole time it sat unused.

  • The technical caveats were the familiar ones. Not every kit produced a clean single-source profile. As Chapter 8 would predict, older kits more often presented as mixtures or low-template samples, requiring the careful interpretation and honest statement of uncertainty that Chapter 9 demands. Some kits, poorly stored, yielded partial or degraded results. But these were ordinary forensic-DNA challenges, not a reason the kits had gone untested.

  • CODIS turned profiles into leads. Profiles developed from the kits were uploaded to CODIS (Chapter 7), and the database returned large numbers of hits. Crucially, many hits were case-to-case matches — linking different assaults to the same unknown or known offender. This is the affirmative power of the database made visible: a pattern of serial offending that had been invisible precisely because the kits that connected the crimes had never been tested.

  • The leads exposed serial offenders. Both the Detroit and Cuyahoga efforts reported substantial numbers of identified suspects and, strikingly, large numbers of serial-perpetrator linkages — offenders connected by DNA to multiple separate assaults. The testing produced new prosecutions and convictions of dangerous repeat offenders, some of whom had remained free, and in some cases continued offending, during the years the kits sat on shelves.

What the evidence did — and didn't — establish

The lesson must be stated at its true strength, in both directions.

What the testing established: that the backlog was overwhelmingly a failure to use the science, not a failure of the science. The kits, when tested, identified offenders, linked serial crimes, and supported convictions — exactly what they were collected to do. The evidence vindicated the victims who had endured the examinations.

What the testing did not establish, and what honesty requires us to add: a CODIS hit is an investigative lead, not a proof of guilt or even of a crime (§37.3, Chapter 7). A database match identifies whose biological material is present; it does not, by itself, establish non-consent or adjudicate a case. Each lead still required investigation, and each case still had to be proven. Moreover, testing a kit could not undo the other damage the delay had done: witnesses' memories had faded, some statutes of limitation had run, some victims could not be located or understandably declined to revisit the trauma years later, and some offenders had died or could not be found. The reform proved the kits' value and, in the same motion, measured the cost of having ignored them.

One exoneration thread deserves particular note, because it shows the backlog's harm cut in both directions of this book's first theme. The same testing that identified guilty serial offenders also held the power to exclude — and nationally, the broader movement to test old biological evidence has been entangled with post-conviction DNA work that has freed wrongly convicted people (the Innocence Project record, Chapter 6). Untested or belatedly tested biological evidence is a double failure: it leaves the guilty unidentified and the innocent uncleared. Exclusion, the book's surest forensic voice (Chapter 1, §1.6), cannot speak from a kit no one tests.

Outcome

Detroit's initiative resulted in the testing of its backlog, thousands of CODIS uploads, hundreds of identified suspects, the identification of numerous serial offenders, and a stream of prosecutions and convictions that continued for years; it also generated detailed research on why the kits had gone untested and how to prevent a recurrence. Cuyahoga County's parallel effort produced similar results — thousands of kits tested, a dedicated task force, and numerous prosecutions of serial offenders.

The deeper outcome was structural. The exposure of the backlog drove legislative reform in many states: laws mandating that collected kits be submitted to a laboratory within set timeframes, requiring agencies to inventory their stored kits, and — in a number of jurisdictions — establishing kit-tracking systems that let a victim follow the status of their own kit online. The point of these laws is precisely to remove the discretion that had allowed thousands of kits to be shelved one decision at a time.

The lesson

The lesson of this case is the chapter's §37.5 made real: the backlog was an institutional and bias failure, not a scientific one. The science worked. The kits were testable. The offenders were identifiable. What failed was the system around the evidence — and beneath the genuine resource constraints lay something less excusable: discretionary decisions not to submit kits, frequently shaped by the very misconceptions this chapter exists to dismantle (about how "real" victims behave, about which cases and which victims "warranted" the science). That is cognitive bias (Theme 3) operating at the scale of an institution, repeated thousands of times.

The case also teaches the book's first theme, exclusion over proof (Theme 1), from an unusual angle. We usually meet that theme as a caution against overstating an inclusion. Here it appears as a cost of inaction: evidence that could have identified offenders and excluded the innocent did neither while it sat untested. The remedy required both halves of reform — the technical (test the kits, interpret the mixtures honestly) and the policy (mandate submission and tracking, strip out the bias-laden discretion) — because the failure itself was both technical and human. The most important forensic instrument in this story was not a sequencer. It was the decision to use one.

Discussion questions

  1. The kits were largely testable decades later. Explain why this fact makes the failure to test them harder, not easier, to defend — and connect it to the "dry, not wet" storage principle of §37.3.

  2. Distinguish the backlog from an ordinary laboratory queue (§37.5). Why did advocates and journalists insist on this distinction, and how does conflating the two obscure responsibility?

  3. Many CODIS hits were case-to-case matches exposing serial offenders. Using Chapter 7, explain why those linkages had been invisible before testing, and what that says about the systemic cost of an untested kit.

  4. A commentator says, "A CODIS hit from a backlog kit means the case is solved." Using §37.3 and Chapter 7, explain precisely what a hit does and does not establish, and what work still remains.

  5. The reforms that followed combined a technical fix (testing) with a policy fix (mandatory submission and tracking). Using §37.5's claim that the backlog was "both technical and human," explain why neither fix alone would have been sufficient.

  6. Connect this case to the cold case's Case File for this chapter, in which a prior domestic-violence report is treated as context, not accusation. What common discipline — about not letting assumptions substitute for evidence — links the institutional failure here with the careful handling of Dana Whitfield's old report?