Toxicology reports in litigation: how to read clinical and postmortem results without losing the case

The word "toxicology" covers 2 different jobs, and the report you are holding came out of 1 of them. Clinical toxicology serves patient care. An emergency physician orders a urine drug screen and a serum alcohol to decide whether a confused patient needs naloxone, a CT scan or observation. Speed wins. The result comes back in an hour as "positive" or "negative" for a drug class, and nobody expects it to survive cross-examination.

Forensic toxicology serves the legal system. A medical examiner, crime laboratory or certified workplace laboratory identifies specific drugs with a more specific technique, measures them, documents custody at every step, and writes for readers who were not there. The American Academy of Family Physicians review of urine drug testing (Kale, American Family Physician, 2019) puts the clinical side plainly: immunoassays "can give false-positive and false-negative results, so all results are considered presumptive until confirmatory testing is performed."

Both kinds land in the same litigation file, use the same drug names and often the same units, so they get read the same way. They should not be.

Why the difference decides cases

A positive urine screen says a drug or a cross-reacting substance was probably present in urine above a cutoff. It does not name the drug with certainty, give a blood level, date the use, or say anything about impairment. The AAFP review adds that urine concentrations "do not reflect serum concentrations." Calling a urine screen proof of intoxication, or proof of drug use, overreads the page.

The reverse error also happens. A precise number from heart blood drawn 3 days after death, or from a specimen with a broken custody record, carries less weight than its decimal places suggest.

Immunoassay

A screening test that uses antibodies to detect a drug class. Fast and cheap. Antibodies can bind to related or unrelated molecules, which produces false positives, and can miss drugs in the class they were not built for.

Cutoff

The concentration at or above which a result is called positive. Below the cutoff the result is reported negative, even if some drug is present.

GC-MS and LC-MS/MS

Gas chromatography or liquid chromatography, which separate compounds, coupled to mass spectrometry, which identifies them by mass. The standard confirmation techniques.

Metabolite

A compound the body makes from a drug, such as benzoylecgonine from cocaine or THCA from marijuana. Metabolites can persist long after the parent drug is gone.

Serum, plasma, whole blood

Whole blood is blood as drawn, cells included. Plasma is the liquid left when cells are removed from anticoagulated blood; serum is the liquid left after clotting. Hospitals usually test serum or plasma. Legal alcohol limits are defined in whole blood.

Federally regulated workplace testing is the clearest model of how a presumptive screen becomes a reportable result, because the rules are written down. The Department of Transportation procedures at 49 CFR Part 40 require laboratories to be certified by the Department of Health and Human Services under the National Laboratory Certification Program for each testing method they perform (49 CFR 40.81). They then set 2 cutoffs for each drug: 1 for the screen and 1 for the confirmation.

The cutoffs themselves are set in 49 CFR 40.85, which covers urine. Section 40.85(b) says that on an initial test the lab "must report a result below the cutoff concentration as negative," and that at or above the cutoff it "must conduct a confirmation test." Section 40.85(c) applies the same logic to the confirmation. Section 40.85(d) requires a quantitative value for morphine or codeine at 15,000 ng/mL or above.

2 table notes and 1 gap matter in a deposition. For grouped analytes, the immunoassay's cross-reactivity to the other drugs in the group "must be 80 percent or greater," or separate immunoassays are required. When a lab screens THCA or benzoylecgonine with an alternate technology, it uses the confirmation cutoff for both steps. And fentanyl, methadone, buprenorphine and benzodiazepines are not in the table, so a clean federal panel says nothing about them.

The medical review officer step

Under 49 CFR 40.137, the medical review officer must verify a confirmed positive for marijuana, cocaine, amphetamines, semi-synthetic opioids or PCP "unless the employee presents a legitimate medical explanation." The employee carries the burden of proof. A valid prescription can be that explanation, and the MRO "must not question whether the prescribing physician should have prescribed the substance." The MRO may allow up to 5 more days to produce evidence. If there is a legitimate explanation, the result is verified negative.

So a workplace file has 2 results, the lab's and the MRO's, and they can legitimately differ. Ask for both, plus the MRO's interview notes.

What a clinical lab does instead

Hospital labs are not bound by Part 40 for patient care, and many do not confirm a screen unless a physician orders it. Their cutoffs are set by the assay manufacturer and the lab, not by regulation, and may differ from the federal numbers. The lab's report may not show the cutoff at all. Ask for the assay name, manufacturer, cutoff for each class and whether any confirmation was ordered.

An immunoassay detects a shape, not a molecule, so a medication that looks enough like the target can turn it positive. A 2014 review (Saitman, Park and Fitzgerald) examined 62 articles and found false positives reported for amphetamines, opiates, benzodiazepines, cannabinoids, tricyclic antidepressants, PCP, LSD and barbiturates. Its conclusion: positives are presumptive until confirmed by a second, independent technique. An update covering 2013 to 2024 (Saitman and colleagues, 2026) (61 studies) repeats that screens remain presumptive and need GC-MS or LC-MS/MS confirmation.

That last point turns the table into a records task. If the chart shows a positive amphetamine screen, pull the medication administration record and the home medication list for the prior days and look for the substances in that row. A patient on bupropion with an unconfirmed amphetamine screen is a different case from a patient with confirmed methamphetamine. Only the records tell you which one you have.

The cocaine row is the opposite problem: after topical cocaine in a procedure, confirmation finds benzoylecgonine correctly, and only the procedure note explains why.

Detection windows: when the drug could have been taken

Urine tests look backward over a window that depends on the drug, the dose and the pattern of use. The AAFP review's table 2 gives typical windows. Treat them as ranges, not clocks. A positive result says the use fell somewhere inside the window; it cannot place use at the time of a crash or a fall.

This is the most common toxicology error in crash and dram shop files. Hospital labs usually measure alcohol in serum or plasma; per se driving limits are defined in whole blood or breath. Serum reads higher than whole blood from the same draw, so a hospital number compared straight to 0.08 overstates the legal measure.

Why serum reads higher

Serum holds more water than whole blood, which includes red cells, so it holds more alcohol. Jones and Tilson (2023, PMID 36317846), put the water content of plasma and serum at about 90% to 92% and whole blood at about 78% to 80%, which predicts a ratio of about 1.15. They recommend dividing a hospital plasma or serum alcohol by 1.2 to get a conservative whole blood estimate, with about a 1 in 2,000 chance of overestimating the true value.

What the measured ratios look like

The studies do not agree on everything. Rainey found the ratio independent of concentration; Barnhill found it concentration dependent. Rainey's 99% range runs to 1.49; Charlebois never saw a ratio above 1.26. They used different patients, different assays and different statistics. A toxicologist choosing a divisor should say which study it comes from and why, and the other side's toxicologist will ask.

The worked example

3 cautions. First, the conversion does not answer the time question. Estimating the level at the time of driving from a draw 90 minutes later, called retrograde extrapolation, rests on assumptions about absorption and elimination; that is expert work, and this guide gives no numbers for it. Second, check units on the page; a misplaced decimal is not rare in transcribed summaries. Third, if the report says only "blood," ask the lab in writing what the analyzer measured. The answer is usually serum or plasma.

Near the line, the side that needs a result over the limit will cite the median conversion; the other side will cite the upper tails of Rainey's data and Barnhill's finding that it took a serum of 103 mg/dL to show whole blood of at least 0.080 g/dL at 95% confidence, a mark this hypothetical serum of 100 does not reach. An expert who shows the full range is harder to cross-examine than 1 who shows 1 number.

After death, drugs stored in organs can diffuse into nearby blood. A 2003 review (Pélissier-Alicot and colleagues) described postmortem concentrations that vary by sampling site and by time, named the gastrointestinal tract, liver, lungs and myocardium as reservoirs, and recommended analyzing samples from more than 1 site. A 2026 review (Borra and colleagues) found cardiac blood prone to early postmortem increases, likely from lung reservoirs, and more diffusion with a longer postmortem interval.

Peripheral blood, usually femoral, is preferred over heart blood. A 2024 literature review (Abdelaal and colleagues) describes the cardiac to peripheral blood ratio, the C/P ratio, as the common marker: a ratio above 1.0 suggests redistribution. Limits apply: some drugs not known to redistribute, such as salicylate and tramadol, have shown C/P ratios above 1.0, and the liver to peripheral ratio has been proposed as more reliable. Lipophilic, basic drugs with a large volume of distribution are the most susceptible.

These are literature values, not constants; the ratio in a given body depends on time since death, storage and decomposition. Their use is directional: if the only blood in the report is heart blood and the drug is on this list, the number overstates the peripheral level by an unknown amount.

The 12 questions to ask of any postmortem result

The Abdelaal review reproduces 12 questions, attributed to Yarema and Becker, that should accompany the interpretation of a postmortem drug level. Use them as a records checklist. Each has an answer somewhere in the autopsy file, the investigator's report or the lab case file, or it has no answer, and that is a finding too.

• Site of the sampleFemoral, subclavian, heart, or unstated
• If femoral, was the vein ligated or clamped before drawingAn unclamped draw can pull blood from the central vessels
• More than 1 site sampledNeeded to compute any C/P ratio
• Other tissues analyzedVitreous, lung, liver, muscle
• Time after death at samplingFrom the investigator's report and the autopsy start time
• How the body was storedRefrigeration and duration
• DecompositionDegree noted at external exam
• Collection and storage conditions of the specimenTube type, preservative, temperature
• Delay to analysisCollection date against analysis date in the lab file
• Antemortem clinical informationSymptoms, treatment, medications given in hospital
• Antemortem blood availableHospital admission specimens, if the lab kept them
• Drug propertiesLipophilicity, pKa, volume of distribution

0 of 12 checked

The eleventh question is the one lawyers most often miss. If the decedent reached a hospital before dying, the hospital lab may still hold admission blood, and that specimen predates redistribution entirely. Hospital retention can be short, so send a preservation letter early. See how to read an autopsy report.

Postmortem alcohol: drunk before death, or made after it

Alcohol has its own problem: microbes can produce ethanol after death. A 2007 review (Kugelberg and Jones) strongly recommends vitreous humor for postmortem alcohol, notes that unabsorbed alcohol in the stomach can diffuse into nearby blood after death, and flags bodies recovered from water as problematic. It names ethyl glucuronide (EtG), phosphatidylethanol (PEth) and fatty acid ethyl esters as markers that help separate drinking from postmortem production.

A toxicology report is a 1 or 2 page summary of a much larger file. Behind every forensic result sits a case file, often called the litigation packet or lab packet: the chain of custody, the instrument data, the calibration, the quality control results, the analyst's notes and the reviewer's sign-off. The report tells you the answer. The packet tells you whether to believe it.

Chain of custody is the paper trail of who had the specimen, when, and in what condition, from collection to disposal. Each transfer should carry a date, time, the names of the person handing over and receiving, and the seal or specimen identifiers. A gap does not prove contamination or a mix-up. It removes the proof that neither happened, and in a criminal case the defense will say so.

How to audit a custody record

1. Match identifiers end to end. The specimen number on the collection form, the tube label, the seal, the lab accession log and the final report should agree. A single transposed digit is worth a written question.
2. Walk every transfer in time order. Each handoff needs a giver, a receiver, a date and a time. Look for overlaps (2 people holding it at once) and gaps (nobody holding it).
3. Check seal condition at receipt. The lab should record whether the seal was intact. "Not recorded" is different from "intact."
4. Check storage between steps. Refrigeration, freezer logs and the time the specimen sat before analysis. Some drugs degrade, and postmortem alcohol can change in a poorly preserved tube.
5. Compare against the other records. The draw time on the custody form should match the nursing note, the ED record or the autopsy report.
6. Find the end. Is there remaining specimen for retesting? Where is it stored and until when? A retest by an independent lab is sometimes the strongest answer to a disputed result.

A mismatched seal number is often a transcription slip, and exactly the kind of fact a jury remembers. Resolve it with documents, not testimony about habit.

What belongs in the lab packet

Laboratories differ in what they release without a subpoena or court order, so ask for each item by name. The ledger below shows a hypothetical request.

In any case near a legal limit, the lab's own stated measurement uncertainty decides whether a reported 0.082 g/dL is over 0.08 at the confidence your jurisdiction requires. That is why it goes in the request.

A toxicology report measures how much of a substance was in a specimen. It does not measure what the substance was doing to the person. For some drugs that gap is small enough to argue across; for others it is wide enough to sink a case built on the number alone.

Cannabis

Cannabis is the clearest example. A 2022 systematic review in Neuroscience and Biobehavioral Reviews (McCartney and colleagues) pooled 28 publications reporting 822 driving-related outcomes and looked at how blood and oral fluid THC concentrations related to performance.

"Blood and oral fluid THC concentrations are relatively poor indicators of cannabis/THC-induced impairment."

In regular users (weekly or more often) the review found no significant relationship between these concentrations and performance, though the authors flag those analyses as less certain. Urine adds a second gap: it targets the metabolite THCA, which can stay detectable for weeks in heavy users (chapter 3). A positive urine THCA result in a crash file is evidence of use at some point in a window of days to weeks, not by itself evidence of impairment at the wheel.

That does not end a plaintiff's case. It moves the proof to officer and witness observations, a drug recognition evaluation if 1 was done, field sobriety performance, driving behavior, time of last use and the ED exam. Some states have per se THC limits; whatever the legal rule, the science on concentration and effect is the same.

Opioids and tolerance

Opioids show the problem from another angle. People who take opioids every day develop tolerance, so the same blood concentration can accompany ordinary function in a long-term patient and serious toxicity in someone who has never taken the drug. Redistribution (chapter 5) makes the picture harder still. A level alone rarely answers whether a drug caused a death. The pathologist weighs it against the scene, the history, the autopsy findings and the other drugs present.

So the treatment history is part of the toxicology. Pull prescription and pharmacy fill records, any prescription monitoring report available to you, pain management notes and prior drug tests. A decedent on long-term methadone and a decedent with no opioid history, at the same postmortem level, are very different cases.

Combinations

Many overdose and impaired driving files involve more than 1 substance, and levels that look modest alone may add up, such as an opioid with a benzodiazepine or alcohol. Interpreting combinations is expert work. The records job is to make sure every substance is listed with its specimen, site and time, and none was missed because the screen could not see it.

The same report plays a different role in each kind of case. The table below maps the main case types to the specimens that usually exist, the records that complete them and the error each side most often makes.

Impaired driving

In criminal cases, how the blood was obtained is its own legal question. In Birchfield v. North Dakota (No. 14-1468, 2016), the Supreme Court held that "the Fourth Amendment permits warrantless breath tests incident to arrests for drunk driving but not warrantless blood tests." Collect the warrant, consent form or other documented basis with the record of when and by whom blood was drawn. Whether the basis holds up is for criminal counsel.

Civil crash cases often rely on the hospital chart instead of a police kit, which is where chapters 2 and 4 decide the most arguments.

Overdose wrongful death

These cases turn on postmortem toxicology, and the nursing home wrongful death and hospital death variants add a medication history that the facility controlled. A death certificate listing "acute combined drug toxicity" is a conclusion; the toxicology table, the autopsy findings and the medication records are the basis you need to test it. For deaths after trauma, blunt force trauma records covers how injury findings interact with toxicology in the cause of death.

Medication errors

In hospital medication error cases, the toxicology is often therapeutic drug monitoring: a serum level drawn to adjust a dose. Line up each dose time from the medication administration record, each pump rate change and each draw time on 1 clock; a high level drawn minutes after a dose is a different fact from the same level drawn before the next. See medication error record review.

Workplace

DOT-regulated workplace files follow Part 40 plus the rules of the agency that governs the job. For commercial motor vehicle drivers, 49 CFR 382.201 says no driver shall report for or remain on duty requiring safety-sensitive functions "while having an alcohol concentration of 0.04 or greater." Non-DOT employers may follow their own policies and state law. Ask which program governed the test before comparing it to any cutoff.

Dram shop

Dram shop and social host liability rules vary widely by state, and many turn on what a server knew or should have observed at the time of service. A blood alcohol measured hours later at a hospital or at autopsy does not by itself show what the patron looked like at the bar. Pair the toxicology with the tab, the timeline of service, video and witness accounts.

In criminal cases, a forensic lab report can be testimony, and the Sixth Amendment gives the accused the right to confront the person behind it. The Supreme Court has built a line of cases on exactly this kind of evidence.

"When an expert conveys an absent analyst's statements in support of the expert's opinion, and the statements provide that support only if true, then the statements come into evidence for their truth."

Smith left open whether the absent analyst's statements in that case were testimonial, and sent that question back to the state court. That question turns on the "primary purpose" of the statement, which is 1 more reason to get the whole lab packet: bench notes, worksheets and review forms are the documents whose purpose gets argued.

What this means for the file

• Identify every person who touched the result. The collector, the accessioning technician, the analyst who ran the batch, the reviewer who certified it, and the expert who will testify. The lab packet names them; the report often names only 1.
• Separate the analyst's statements from the expert's opinion. If the testifying expert did not perform or observe the test, list which facts in the expert's report come only from the absent analyst's notes.

In civil cases the Confrontation Clause does not apply. Admissibility of a lab report or a medical examiner's findings runs through the rules on hearsay, business and public records, and the basis for expert opinions, which vary by jurisdiction. The records work is the same: identify who did what, and get the packet.

Toxicology in a litigation file is scattered: an ED urine screen, a serum alcohol in a chemistry panel, a send-out confirmation scanned in 6 days later, drug levels in the ICU flowsheet, and an autopsy table. AI medical record review is good at the first job, finding and lining up every result, and poor at the second, deciding what they mean.

Where it fails

• Units and decimals. OCR on a scanned lab report can turn 0.08 into 0.8 or drop a unit. Optical character recognition errors on numbers are small on the page and large in court. Low-confidence numeric fields should be flagged for a person, never smoothed over.
• Screen versus confirmation. A large language model (LLM) that reads "positive for opiates" in an ED lab and "morphine 45 ng/mL" in a send-out may merge them into 1 fact. They are 2 facts with different weight. The chronology has to keep them apart.
• Specimen site. A postmortem table with heart and femoral columns is easy to flatten. Every line needs its site.
• Copied history. Once a clinician writes "cocaine positive" in 1 note, copy-forward and templated text, including notes drafted by an ambient AI scribe, can repeat it as history in dozens of later notes. A model that counts mentions will treat 1 unconfirmed screen as a pattern.
• Interpretation. Generative AI will answer if asked whether a level was lethal. Hallucination is 1 risk; the larger 1 is a confident answer to a question that needs a toxicologist.
• Conversion factors. A model asked to convert serum alcohol will pick a divisor without citing a source or showing the range. Conversions belong in the expert's report.

Ground every line in a page-level citation so a person can check it in 1 click. Keep a human-in-the-loop reviewer who checks every toxicology line against the lab report. Lawyers have been sanctioned under Rule 11 for filing fabricated AI-generated citations (Mata v. Avianca, S.D.N.Y. 2023); an invented lab value in an expert disclosure is the same failure. Be wary of agentic tools that act across systems on their own; for a toxicology file, you want a tool that reads only what you uploaded and cites every line to it.

Vendor checklist for legal AI tools on toxicology files

Natural language processing helps with vocabulary: notes write alcohol as ETOH, EtOH or "blood alcohol," and screens as UDS or "tox screen." AI document review for toxicology is mostly that: find, sort, cite and flag, then hand the file to a person.

Copy these into your own forms. Adjust the legal basis line to your jurisdiction and the type of request. Some laboratories release the packet only on subpoena or court order; some charge for it.

TO: [LABORATORY NAME], Custodian of Records
RE: [SUBJECT / DECEDENT NAME], DOB [DATE]
LAB CASE / ACCESSION NO.: [NUMBER]   SPECIMEN DATE: [DATE]
BASIS: [AUTHORIZATION / SUBPOENA / COURT ORDER / DISCOVERY REQUEST]

Please produce the complete case file for the testing above,
including each item below. For any item not produced, state in
writing whether it exists, where it is kept and why it is withheld.

[ ] Final report(s), all versions, with dates of issue and amendment
[ ] External chain of custody forms, collection to receipt
[ ] Internal chain of custody, receipt to disposal or storage
[ ] Accession record: seal condition, specimen type, tube type,
    preservative, volume received
[ ] For each test: method name, instrument, and batch identifier
[ ] Chromatograms and mass spectra (or equivalent raw data) for
    the case sample, calibrators, controls and blanks in each batch
[ ] Calibration curve and acceptance criteria for each batch
[ ] Quality control results for each batch, with acceptance limits
[ ] Analyst bench notes, worksheets, reruns and deviations
[ ] Technical and administrative review records, with reviewer names
[ ] Method validation summary: limits of detection and
    quantitation, interferences studied
[ ] Measurement uncertainty for each reported quantitative result
[ ] Accreditation certificate and scope in force on the analysis date
[ ] Immunoassay cutoffs and assay names, if a screen was performed
[ ] Current storage location and planned disposal date of any
    remaining specimen

PRESERVATION: Please preserve all remaining specimen and the
materials above pending further notice.

CASE: [NAME]   EVENT TIME: [DATE HH:MM] per [SOURCE, PAGE]
DEATH TIME (if any): [DATE HH:MM]   AUTOPSY START: [DATE HH:MM]

RESULT | SUBSTANCE | VALUE + UNITS | SPECIMEN | SITE | DRAW TIME | LAB | METHOD | SCREEN OR CONFIRMED | CUTOFF | PAGE
-------|-----------|---------------|----------|------|-----------|-----|--------|---------------------|--------|-----
[ ]    | Ethanol   |               | serum / whole blood / vitreous / urine
[ ]    | Drug      |               | urine / serum / femoral / heart / liver
[ ]    | ...

ALCOHOL CONVERSION (serum or plasma only):
Serum value [ ] mg/dL; divisor [ ] per [STUDY]; estimate [ ] mg/dL
Range shown: divisor [ ] to [ ] gives [ ] to [ ] mg/dL
Time from event to draw: [ ] min (extrapolation is for the expert)

SCREEN CHECKS:
Cross-reactants in MAR or home meds for each positive: [ ]
Drugs the assay could not detect: [ ]
Confirmation ordered? [Y/N]  Result page: [ ]

POSTMORTEM CHECKS:
Sites sampled [ ]; femoral vein clamped? [ ]; C/P ratio [ ]
Vitreous alcohol [ ]; decomposition [ ]; antemortem specimen [ ]

HISTORY: prescriptions and fills [ ]; prior drug tests [ ]

CUSTODY GAPS AND MISSING ITEMS: item | evidence it exists (page) | date requested | answer

Medrecords AI is medical chronology software. It works on the records you upload; it does not request or retrieve records from hospitals, laboratories or medical examiners. On a file with toxicology in it, it does 4 things:

• Builds a cited chronology that puts every drug screen, alcohol level, send-out result and drug level on the timeline with the medication record and the clinical notes around it, with a citation on every line back to the source page.
• Flags missing records the file implies, such as a note saying a specimen was sent for confirmation with no result in the production, or an autopsy report that references a toxicology report nobody produced. Each flag cites the line that implies it.
• Routes scanned lab reports and handwritten forms through OCR and flags low-confidence pages for a human read.
• Lets you search the whole record for the variant terms clinicians use for the same test, so an "EtOH" in a nursing note is not missed.

It does not convert serum alcohol, interpret levels, or decide impairment or cause of death. Flags are signals, not verdicts. It runs under SOC 2 and HIPAA with a signed BAA (see security and HIPAA). Self-Service bills 10 cents a deduplicated page, down to 5 cents at volume, duplicates free; Enterprise On-Prem is an annual license. More on the use cases: wrongful death record review and medical examiner case record review.

What is the difference between a drug screen and a confirmed toxicology result?

A screen is usually a urine immunoassay that detects drug classes and is presumptive. A confirmed result identifies the specific drug with a second, more specific method such as GC-MS or LC-MS/MS. Many hospital screens are never confirmed.

How do you convert a hospital serum alcohol to whole blood alcohol?

Divide the serum or plasma value by a ratio from the peer-reviewed literature. Studies put the median near 1.15, and Jones and Tilson (2023) recommend 1.2 for a conservative estimate.

What is postmortem redistribution?

The movement of drugs after death from organs such as the lungs and liver into nearby blood. It can raise heart blood levels above femoral blood levels.

Why is femoral blood preferred over heart blood at autopsy?

Femoral blood is farther from the drug reservoirs in the chest and abdomen, so it is less affected by redistribution. A heart to peripheral blood ratio above 1.0 suggests redistribution has occurred.

Does a positive THC test prove impairment?

No. A 2022 systematic review of 28 studies found blood and oral fluid THC concentrations to be poor indicators of impairment, and urine THCA can stay positive for weeks.

What should a toxicology lab packet include?

Chain of custody, chromatograms and spectra, calibration and quality control for the batch, analyst and review notes, method validation, measurement uncertainty, accreditation, and the status of remaining specimen.

What does a toxicology expert witness do?

A forensic toxicologist interprets results: whether a finding is reliable, what a concentration can and cannot show, alcohol conversions and extrapolations, and how drugs interact. Cause and manner of death remain the pathologist's opinion.

Can AI read a toxicology report accurately?

AI can find and list results across a large file and cite each 1 to its page. It can misread units and decimals on scanned reports and blur screens with confirmations, so every line needs a human check against the lab report.

Is it HIPAA compliant to upload toxicology and autopsy records to AI software?

It can be, with a vendor that signs a business associate agreement, holds SOC 2, does not train on your data and controls access. Consumer chatbots without a BAA are the wrong place for protected health information.

Can ChatGPT convert a blood alcohol level for court?

A general chatbot will do the arithmetic, but it may pick a divisor without citing a source or showing the range. Conversions offered in litigation belong in a toxicologist's report, with the study named and the uncertainty stated.

Every number and cutoff in this guide was checked in September 2026 against the source linked below: full text where it was open (the AAFP review, the Abdelaal review on PubMed Central, the eCFR and the Supreme Court opinions) and the PubMed abstract otherwise. The worked conversion, the custody log, the lab packet ledger and the case examples are hypothetical. Nothing here is legal or medical advice.

• Rainey PM, "Relation between serum and whole-blood ethanol concentrations," Clinical Chemistry, 1993.
• Charlebois RC and colleagues, Journal of Analytical Toxicology, 1996.
• Barnhill MT and colleagues, Journal of Analytical Toxicology, 2007.
• Jones AW and Tilson C, Journal of Forensic Sciences, 2023: water content, divisor of 1.2.
• Kale N, "Urine Drug Tests: Ordering and Interpretation," American Family Physician, 2019: presumptive screens, assay blind spots, tables 2, 3 and 5.
• Saitman A, Park HD and Fitzgerald RL, Journal of Analytical Toxicology, 2014, and the 2026 update: immunoassay false positives.
• Moeller KE and colleagues, Mayo Clinic Proceedings, 2008.
• Abdelaal GMM and colleagues, "Postmortem redistribution of drugs: a literature review," Forensic Science, Medicine and Pathology, 2024: C/P ratios, table 2, and the 12 Yarema and Becker questions.
• Pélissier-Alicot AL and colleagues, Journal of Analytical Toxicology, 2003, and Borra LCP and colleagues, Forensic Science International, 2026: mechanisms and timing of redistribution.
• Kugelberg FC and Jones AW, Forensic Science International, 2007: postmortem alcohol, vitreous humor, markers of drinking.
• McCartney D and colleagues, Neuroscience and Biobehavioral Reviews, 2022: THC concentration and driving impairment.
• 49 CFR 40.81, 40.85, 40.137 and 382.201, current eCFR.
• Melendez-Diaz v. Massachusetts, 557 U.S. 305 (2009); Bullcoming v. New Mexico, 564 U.S. 647 (2011); Birchfield v. North Dakota (2016); Smith v. Arizona (2024).
• Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023): Rule 11 sanctions for fabricated AI-generated citations.

Related guides: how to read an autopsy report, nursing home wrongful death records, blunt force trauma medical records, how to get medical records for a lawsuit, and record review for expert witnesses.