1,630 of the 2,036 US death investigation offices are coroner offices where state law commonly requires no medical license, and the accuracy of death investigation findings is not measured

What is happening?

The United States has no federal minimum qualification for the people who determine how a death is recorded. Each state defines the office by statute, each county or city funds and staffs it, and the two forms of office that result answer to different requirements.

In 2018 there were 2,036 medicolegal death investigation offices in the country — 1,630 coroner offices, 384 city, county, district or regional medical examiner offices and 22 state medical examiner offices. Coroner offices are the more numerous by a wide margin and the coverage runs the other way: medical examiner offices serve 65.7 percent of the resident population and coroner offices 34.3 percent, as of 2018.

The volume passing through them is large. In 2018 these offices received 1.32 million deaths as referrals and accepted 605,000 of them for further investigation. Total deaths in the country were 3,090,964 in 2023 on the final national count.

What it takes to hold one of these offices is set by state law and differs by state. For a coroner the conditions are commonly adulthood, residence in the county and the absence of a felony conviction, and a medical license is not among them. A medical examiner is a physician, normally board certified after a residency in anatomic pathology of three to four years and a fellowship in forensic pathology of one to two years. The National Institute of Justice has written that in some small jurisdictions the person who determines cause of death may be a funeral director or a person with no medical experience at all.

The output of all 2,036 offices is unmeasured. The 2025 report of the National Academies of Sciences, Engineering, and Medicine states that the accuracy of cause and manner of death determinations is currently unknown.

Whose problem is this?

The authority to set a qualification, the money to meet it and the duty to do the work sit in three different places, and none of the three can compel the other two.

Nothing in that arrangement is hidden. It is written into the statutes that create the offices, and it is the arrangement the federal reports of 2009 and 2025 both describe.

Where does this problem end?

The boundary here is the procedure that produces a finding, and not any finding it has already produced.

That distinction is what keeps this frame from collapsing into a review of individual cases, which no source opened here attempts and which courts already have a route for.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state: no source opened here carries a target set by a body with authority to set one. No federal statute, no federal agency and no state named in these sources gives a minimum qualification that offices must reach, a share of offices that ought to hold peer-review accreditation, a number of forensic pathologists the country ought to employ, or an accuracy level that determinations ought to meet. The 2025 report contains recommendations and an estimate of the workforce needed, and those are recorded below as an attempt; no agency or statute opened here has adopted any of them as a target.

How big is it?

Between 605,000 and 1.32 million deaths in a year. That is the range between the deaths referred to these offices in 2018 and the subset of them accepted for further investigation in the same year. Every death in that range is recorded by an office whose qualification floor is set by state law and whose determination accuracy is not measured nationally.

The wider figure counts referrals, which is every death the system was asked to look at. The narrower figure counts the cases an office took up and investigated, which is where a cause and manner determination is actually produced. No source opened here says how many of the declined referrals still received a determination from the same office, so the gap between the two numbers is left open rather than resolved.

Both numbers are for 2018, and total deaths in the country reached 3,090,964 by 2023. The current annual figure is therefore likely higher than either end of this range, and no source opened here restates it.

These two figures answer different questions, and neither of them counts the number of findings that were wrong.

Under what conditions does it arise?

1. The office is created by state law and staffed by local government, and no federal body sets a floor. The federal role described in the sources opened here is coordination, research, data collection and partial grant funding. The 2023 forensic science improvement grants named in the federal account come to about 35 million dollars for that year, and no source opened here gives a series for any other year.

2. The qualification required to hold the office and the judgment the office produces are different in kind. A finding of cause and manner of death is a medical determination. In a coroner office the statutory conditions to hold the post are commonly residence, age and the absence of a felony conviction.

3. The workforce that would staff the medical form of the office does not exist at the size the 2025 report estimates is needed. About 800 to 850 forensic pathologists work full time against an estimated need of at least 1,600. The pipeline adds about 37 newly certified pathologists a year and about 21 of them remain in the field full time, so the shortfall does not close on the current intake.

A county that cannot recruit a forensic pathologist does not thereby stop receiving bodies.

4. Peer review is voluntary. As of 2025-03-13, 77 offices held accreditation under the national voluntary peer-review program for medical examiner and coroner offices, which is under 4 percent of about 2,000 offices. The remainder operate with no external review of their casework required by anyone.

5. Nothing measures the result. With no national accuracy measure, an office that determines cause of death badly and an office that determines it well produce records that look identical in the national statistics.

What has been tried?

Six attempts across sixteen years, and every one of them lands in the same place. Each produced a description, a recommendation or a coordinating body, and none of them produced a rule that an office is required to meet.

What was found?

Why is it still unsolved?

Institutional gap — the thing that would close it is a rule nobody in the structure has the authority to write.

The offices are creatures of state law and local budgets. Congress can fund, and it does, in grants counted in the tens of millions. Federal agencies can study, coordinate and collect data, and since 2018 two new bodies have been created to do exactly that. What no federal body in these sources holds is the power to tell a county that its death investigator must meet a qualification. So the recommendation of 2009 was addressed to fifty separate legislatures, and the recommendations of 2025 are addressed to the same fifty.

The second part of the gap is that the remedy needs a workforce the country does not have. Converting a coroner office into a medical examiner office means hiring a board certified forensic pathologist, and there are roughly 800 to 850 of them working full time against an estimated need of at least twice that. A jurisdiction that decided today to convert would be competing for a person who does not exist yet, and the pipeline that would produce that person adds about 21 lasting entrants a year. The reform and the staffing constraint point in opposite directions, and the constraint is the slower of the two to move.

The third part is the one that keeps the pattern stable. There is no national measure of whether these determinations are accurate, so there is no number that can get worse. A backlog produces a queue, a shortage produces a vacancy and a missed deadline produces a date, but an unmeasured output produces nothing at all. Every year the offices file their records, the records enter the national statistics, and the statistics carry no field for whether the finding was right. A failure that generates no measurement generates no occasion to act on it either.

What observation would mean it is solved?

Candidates — (a) a national measure of cause and manner determination accuracy exists and is published on a recurring basis (b) a minimum qualification for death investigators is in force and applies across jurisdictions (c) the share of offices under external peer-review accreditation rises well above the 4 percent recorded in 2025 (d) the full-time forensic pathologist workforce approaches the estimated need.

(a) is the one that has to come first, and it is the hardest. Accuracy in this field has no obvious denominator, because the fact a determination is compared against is usually not available separately from the determination itself. A published measure could therefore be an audit of process rather than of outcome, and a process audit can be passed by an office that gets the answers wrong.

(b) alone counts statutes. A minimum qualification can be set low enough to change nothing, or set high enough that jurisdictions without a candidate simply leave the office vacant. The condition would then have been met on paper by places that already met it.

(c) alone counts memberships. Accreditation is voluntary and offices that seek it are more likely to be the ones already running well, so a rising share can reflect self-selection rather than a rising floor. It would still be worth reading, because a jump from 4 percent would be hard to produce by self-selection alone.

(d) is the slowest and the most necessary. At about 21 lasting entrants a year against a shortfall of several hundred, this indicator moves over decades. The four have to be read together, and none of them means much while (a) is missing, because without an accuracy measure the other three describe inputs to a process whose output nobody is checking.

What is it connected to?

Fills with researchfederal reporting compliance for deaths in custody, forensic laboratory capacity and casework backlogs, public health mortality surveillance and the quality of the underlying cause of death data, the overdose mortality count and how it is produced, and the same qualification question in the coroner systems of other countries. Relation type and evidence grade were not established in this round.

What these sources do not say

• A national error rate. No such figure exists in any source opened here, and the 2025 report states directly that the accuracy of cause and manner determinations is currently unknown. That is nearer to a statement that the measurement is not made than to a statement that the data are missing.
• The state by state training requirements for coroners. The 2025 report gives the common statutory conditions and does not list them by state. A 2013 broadcast piece that reports on those requirements is listed below with its URL unconfirmed, so no state level list is carried in this document.
• Which jurisdictions converted after 2009. No source opened here counts the offices that changed form between the 2009 recommendation and the 2018 census, so the 2,036 figure cannot be read as a before or an after.
• A funding series. The federal account names about 35 million dollars in forensic science improvement grants for 2023 and no other year. No source opened here gives federal spending on death investigation over time.
• An exact pathologist headcount. Two sources both put the full-time workforce in the low 800s and they do not give the same integer, and neither states its counting basis.
• What replaced the 2013 forensic science commission. No source opened here describes what followed it. A Federal Register notice on advancing forensic science dated to the same week is listed below with its URL unconfirmed, and nothing in this document rests on it.
• A response from the offices or from the states. The federal account published by the National Institute of Justice is the closest thing here to a statement of position by a body being described, and it sets out the coordination bodies and grant programs rather than asserting an authority to set qualifications. No statement from a state government, a county association or a national association of these offices responding to the fragmentation finding was opened in this round.
• Whether the 2025 recommendations were acted on. The report is dated 2025-11 and no source opened here reports legislation, a budget line or an agency decision responding to it as of 2026-08-18.
• How much of the report is specific to deaths in custody. The 2025 report was commissioned around deaths in custody, and the figures used here are taken from the chapters describing the national system as a whole. No source opened here restates those system-wide counts independently of that report.

See the evidence

No primary statute or state code was read. The evidentiary spine of this document is one report of the National Academies of Sciences, Engineering, and Medicine published in 2025, opened here in five separate chapters, together with one federal agency article and one federal mortality brief. Where a figure could be checked against a second source it was: the forensic pathologist workforce appears in the report and in a news account of it, and the two agree on the low 800s without giving the same integer, which is why that row is graded medium rather than high. Where no second source exists the grade says so. The absence of a federal minimum qualification is established by the silence of every source opened here rather than by a document declaring the absence, and that is a weaker form of evidence than a positive finding, so it is graded medium. Four listed rows could not be opened in this round and their URL cells are left blank with the reason attached; the claims those rows would carry are not made anywhere in this document. Two chapters of the report were opened and are listed without being used for any claim. The 2025 report was commissioned around deaths in custody and the counts used here are drawn from the chapters that describe the national system as a whole, which is a different frame from federal reporting compliance for custodial deaths. This is a Path A output, a research based definition, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 605,000–1,320,000 As of 2018

Sensitivity The width of the range is the difference between a death being referred and a death being accepted for investigation. No source opened here says how many of the declined referrals still received a determination from the same office, so the gap is left open rather than resolved. Both figures are for 2018, while total deaths in the country reached 3,090,964 in 2023, so the current annual figure is likely higher than either end and no opened source restates it. The range counts deaths in one year and not people: surviving families, the criminal and civil proceedings that rest on a determination, and the national mortality statistics that absorb it are all outside the count. In the other direction, the range overstates harm if it is read as a count of wrong determinations, because no national measure of determination accuracy exists and these numbers say nothing about how many findings are correct.

Regional breakdown No source opened here breaks referrals or accepted cases down by state or county. The only geographic split available is the share of the resident population served by each office type, which is a population share rather than a count of referred deaths, and converting the one into the other would be a proportional allocation.