Author: toxlaw

The two eyes can yield different vitreous drug concentrations in the same autopsy. In one MDPHP case, right vitreous contained 1,846.45 ng/mL and left vitreous 2,568.01 ng/mL. The authors proposed partial coagulation as an explanation for the higher left-sided result, but did not experimentally establish the cause of the difference.

What was sampled?

Croce and colleagues investigated one 30-year-old man found dead with a history of cocaine use. The estimated postmortem interval was three to four hours. He had last been seen alive 24 hours before discovery, which did not supply an exact ingestion time.

Autopsy collection included central and peripheral blood, separate right and left vitreous humor, urine, gastric contents and hair. The report did not identify the exact peripheral blood site or provide a controlled serial vitreous collection sequence.

The measured concentrations differed by specimen

Targeted LC–MS/MS measured MDPHP at 1,639.99 ng/mL in central blood and 1,601.90 ng/mL in peripheral blood, alongside the two vitreous values. Urine contained 12,954.13 ng/mL and gastric contents 3,028.54 ng/mL.

These were concentrations in different biological compartments. A higher urine or vitreous result did not represent the same measurand as a blood concentration, and the study did not validate conversion equations between them.

The liquid-specimen procedure used 200 µL aliquots, acetonitrile treatment and LC–MS/MS. The report listed a quantification limit of 0.5 ng/mL, recovery of 78.8% and ion suppression of −20.4%. Those figures describe the reported analytical method; they do not quantify the contribution of coagulation or uneven sampling to the two-eye difference.

What did the coagulation explanation establish?

The authors suggested that partial coagulation of the left vitreous could explain its higher concentration. They did not report an experiment separating clot and fluid fractions, repeated standardized eye collections, or a series of comparable cases measuring that effect.

The explanation should therefore remain a hypothesis linked to this specimen’s condition. The case cannot establish a correction factor for coagulated vitreous or a normal right-to-left range for MDPHP.

Similar blood values did not measure antemortem change

The authors interpreted the similar central and peripheral blood concentrations as evidence against relevant redistribution. That was their assessment of this case. A single collection from two sites, however, did not directly measure the difference between the man’s predeath and postmortem blood concentrations.

There was no paired antemortem sample and no serial study showing how either blood or vitreous concentration changed. The short estimated interval also did not prove that redistribution or other postmortem change was impossible.

Why the result is useful without becoming a universal rule

The case supplies an example of MDPHP detection in both eyes and highlights the importance of recording specimen condition. It supports preserving laterality and keeping the two results separate, rather than silently averaging them into a supposedly definitive blood-equivalent value.

The study did not determine the dose, exact intake time or a threshold for impairment. Its probable poisoning conclusion incorporated the whole investigation; neither one eye’s concentration nor the discrepancy independently established cause of death.

For broader matrix limits, see vitreous drug screening compared with blood and why vitreous detection does not establish a blood concentration.

Primary source

Croce EB, Dimitrova A, Di Milia MG, et al. Postmortem distribution of MDPHP in a fatal intoxication case. Journal of Analytical Toxicology. 2025;49:137–141. doi:10.1093/jat/bkae092.

A mass-spectral library suggestion was not the final drug identity in a reported MDPHP death. General unknown screening returned a 72% match to a differently named cathinone, while subsequent LC–MS/MS identified MDPHP. The score described the screening comparison; it was not a validated 72% probability that the suggested compound was present.

The case required testing beyond the expected drug

Croce and colleagues described a 30-year-old man found dead at home with a history of cocaine use. The report estimated a postmortem interval of three to four hours; being last seen alive 24 hours earlier was a separate observation, not the same interval. Autopsy findings included congestion and pulmonary edema without a uniquely diagnostic drug-poisoning lesion.

Central and peripheral blood, urine, gastric contents, separate right and left vitreous samples, and hair were collected. Common drugs of abuse were not detected in the liquid specimens by the reported GC–MS investigation. That did not establish the absence of every psychoactive substance.

What the screening library suggested

The general unknown screening result suggested “4′-methyl-α-pyrrolidinohexanophenone,” with a reported 72% likelihood against the SWGDRUG library used by the investigators. This was a candidate generated by a screening workflow, not an independently confirmed diagnosis.

A library-search output depends on the reference spectra and comparison method. The case supplied no validation converting that particular numerical score into the chance of correct identification. Repeating it as a certainty about the compound would misstate the evidence.

What established the reported MDPHP result?

The investigators used an MDPHP reference standard and a targeted LC–MS/MS procedure. The method reported an MDPHP precursor ion at m/z 290, transitions to m/z 135 and 140, and retention time of 15.6 minutes. Identification and quantification therefore drew on additional analytical evidence beyond the initial library suggestion.

For liquid specimens, the paper reported a quantification limit of 0.5 ng/mL, recovery of 78.8% and ion suppression of −20.4%. These are the performance figures described in this case report; they do not establish that every possible matrix or interfering compound was independently evaluated within the report.

MDPHP was measured at 1,639.99 ng/mL in central blood and 1,601.90 ng/mL in peripheral blood. The exact peripheral anatomical collection site was not specified. Urine, gastric contents and both vitreous specimens also contained MDPHP.

A quantitative finding still required causal interpretation

The authors attributed the death probably to a lethal MDPHP exposure, integrating the high concentrations, examination and available circumstances. Their comparison with previously published cases was descriptive casework, not a statistical experiment establishing a lethal threshold.

They could not determine a precise intake time or exclude previous or repeated consumption. Hair also contained other drugs, which documented a different exposure history rather than proving all those substances were active at death.

What to preserve when reading a laboratory report

Separate the initial candidate, the confirming analytical evidence and the final quantitative result. Record the library version, method scope, specimen and units. A negative conventional screen and a positive broader investigation can coexist without either result being contradictory.

Neither the library score nor MDPHP detection alone establishes impairment or cause of death. For another identification problem, see why accurate mass does not resolve every drug identity.

Primary source

Croce EB, Dimitrova A, Di Milia MG, et al. Postmortem distribution of MDPHP in a fatal intoxication case. Journal of Analytical Toxicology. 2025;49:137–141. doi:10.1093/jat/bkae092.

A “main intoxicant” category in a fatal-poisoning table may partly reflect a classification rule rather than proof that one drug acted alone. In the 2017 Nordic survey, cases without a uniquely identifiable causal drug were assigned using a group hierarchy. The authors explicitly warned that this rule could conceal contributions from other drugs, particularly buprenorphine.

Which deaths were included?

Simonsen and colleagues studied fatal poisonings in a historically defined drug-use population across Denmark, Finland, Iceland, Norway and Sweden during 2017. The country totals were 162, 202, 15, 203 and 408 cases, respectively, totaling 990.

The study’s inclusion definition relied on police or autopsy information about intravenous drug misuse or misuse of substances within specified historical convention schedules. It was not a contemporary clinical diagnostic assessment of every resident. Most cases had medicolegal autopsy and toxicology, with a few medical autopsies included.

Detection and causal assignment were separate steps

Laboratories used screening procedures covering common illicit and medicinal drugs, with additional testing as indicated and routine blood alcohol measurement. Primary findings were confirmed and quantified by specific chromatographic methods. The report did not provide one uniform blood collection site, postmortem interval, storage protocol or analyte-specific detection limit for all five countries.

Cause-of-death assessment incorporated the autopsy report, toxicology, police information and circumstances. A substance’s appearance in the analytical-findings table therefore differed from its assignment as the main intoxicant.

How did the hierarchy operate?

Four groups organized the substances. Group I included heroin/morphine, methadone and several other opioids, as well as cocaine. Group II included various stimulants and other substances. Group III included buprenorphine and benzodiazepines. Group IV covered other drugs and poisons, including ethanol.

Normally, the main intoxicant was the drug judged responsible by the forensic pathologist. If multiple drugs were involved and no single substance could be selected, the drug in the lowest-numbered group was designated the main intoxicant. When multiple drugs belonged to the same group, the substance judged the main contributor was chosen.

Consequently, a Group I drug could receive the primary label in an unresolved mixed-drug death involving a Group III drug. That did not establish that the latter had no causal role. The hierarchy was the study’s historical coding method, not a validated ranking of pharmacological importance for every mixture.

Why buprenorphine exposed the limitation

Buprenorphine is an opioid but appeared in Group III rather than Group I. The authors stated that the hierarchy could camouflage its influence on cause-of-death classification and suggested reorganizing the groups in future work. They separately included buprenorphine when comparing opioid deaths.

The survey found a median of four to six detected drugs per case, excluding ethanol. Such co-detection made a single main category useful for tabulation but potentially incomplete as a description of the toxicological evidence.

What the historical statistics cannot decide

The study did not experimentally measure the toxicity of every detected combination. Nor could one selected category establish an individual’s impairment, dose or isolated mechanism of death. Its 2017 findings also should not be presented as today’s drug-market distribution.

When reading surveillance tables, preserve the distinction between drugs detected, drugs considered contributory and the single category selected for counting. Different coding rules can change that last count without changing the underlying specimens. Related record-based distinctions appear in death certificates versus integrated forensic poisoning assessments.

Primary source

Simonsen KW, Kriikku P, Thelander G, et al. Fatal poisoning in drug addicts in the Nordic countries in 2017. Forensic Science International. 2020;313:110343. doi:10.1016/j.forsciint.2020.110343.

Standard addition can reduce matrix-related bias by building calibration within the specimen being measured. It cannot reverse decomposition, redistribute a drug back to its earlier location, or reconstruct an antemortem concentration. A controlled cannabinoid study illustrates both the purpose and the limits of this analytical approach.

Why the researchers used individual-specimen calibration

Schaefer and colleagues studied JWH-210, RCS-4, THC and their metabolites in twelve male pigs. Each received 200 µg/kg of each parent drug by pulmonary administration. Initial samples were collected at euthanasia eight hours later; six carcasses were then stored at room temperature and six at 4°C, with collections at 24, 48 and 72 hours.

Tissues, bile fluid and duodenal contents differed substantially from one another. Postmortem material also varied between individuals. A calibration prepared in one blank matrix might not reproduce every specimen’s effect on extraction and measurement.

How standard addition worked in this study

The researchers homogenized the material and prepared four portions of each specimen. One received no added calibrator; three received increasing known amounts of the target analytes. The preparation included enzymatic cleavage, solid-phase extraction and LC–MS/MS measurement.

Initial calibrator levels were described as 20, 40 and 60 ng/g or ng/mL, with concentrations adjusted according to preliminary estimates. The calibration related signal to added concentration using a straight-line equation. Extrapolating that line to the negative concentration axis supplied the estimate of the amount originally present.

Because each curve used portions of the same specimen, calibration was matched to that specimen’s matrix. The reported regression coefficients ranged from 0.95 to 0.99. That fit supported the calculation but was not, by itself, proof of zero analytical error.

Blood used a separately described method

The tissue standard-addition procedure should not be confused with the blood method. Blood underwent a previously validated extraction and LC–MS/MS workflow with isotope-labeled internal standards.

Reported blood lower quantification limits were 0.5 ng/mL for JWH-210, HO-JWH-210, JWH-210-COOH, RCS-4 and HO-RCS-4, and 2.0 ng/mL for RCS-4-COOH, THC and its measured metabolites. Those blood limits cannot be assigned automatically to every tissue matrix or to an unrelated laboratory method.

Calibration addressed the sample at analysis

After correction for a matrix effect, the result still described the material collected after death. If the drug moved into that tissue before collection, calibration did not remove that increase. If a compound degraded, the analytical procedure did not recreate the lost amount.

Likewise, enzymatic cleavage made sample preparation part of the measurand. A concentration obtained after that processing should retain its method context, rather than being described automatically as a free, unconjugated concentration in the original tissue.

Other sources of variation remained

The authors observed substantial between-animal differences. They did not standardize sampling among different areas within each organ, and acknowledged that uneven drug distribution could contribute to changing results. Opening the abdomen for repeated collections could also alter decomposition.

Standard addition did not repair those sampling limitations. A well-fitted calibration curve cannot make a small biopsy representative of an entire organ if the analyte is distributed unevenly.

What a quantitative result can support

The method supplied concentrations in specified processed specimens for comparing experimental conditions. It did not provide a human dose estimate, impairment threshold or independent cause-of-death determination. Keep specimen identity, units, preparation and collection timing attached to the number.

For the separate distinction between a signal and a validated concentration, see postmortem metabolite peaks versus quantitation.

Primary source

Schaefer N, Kröll AK, Körbel C, et al. Time- and temperature-dependent postmortem concentration changes of the (synthetic) cannabinoids JWH-210, RCS-4, as well as Δ9-tetrahydrocannabinol following pulmonary administration to pigs. Archives of Toxicology. 2020;94:1585–1599. doi:10.1007/s00204-020-02707-4.

A metabolite detected in postmortem blood need not have been detectable in that blood at death. In a controlled pig experiment, HO-RCS-4 was not detected in perimortem blood but appeared in some blood specimens after room-temperature storage. The result demonstrated a postmortem change in detectability, without proving one mechanism or establishing when a human used the parent drug.

The experiment controlled administration and death timing

Schaefer and colleagues gave twelve male pigs pulmonary doses of JWH-210, RCS-4 and THC, each at 200 µg/kg body weight, using an ultrasonic nebulizer. The animals were anesthetized and mechanically ventilated. Eight hours after administration, they were euthanized and initial specimens were collected.

Six carcasses were stored at room temperature and six at 4°C. Further collections occurred at 24, 48 and 72 hours after death. These were known experimental intervals, unlike the frequently uncertain timelines in human autopsy cases.

The peripheral specimen site changed

The initial peripheral blood came from the jugular vein. Later peripheral collections came from the femoral or brachiocephalic vein. Central blood and numerous tissues and fluids were also sampled. Collected specimens were stored at −20°C pending analysis.

Changing peripheral collection sites limited interpretation as a pure same-site time course. The abdominal cavity was opened for sampling and sutured afterward; the authors acknowledged that opening could alter microbial conditions and decomposition, especially at room temperature.

What was observed for HO-RCS-4?

HO-RCS-4, a hydroxylated RCS-4 metabolite, was not detected in peripheral or central blood at the initial collection. In the room-temperature group it was subsequently detected in peripheral blood from some animals, with the timing of detection varying between pigs, and in central blood from four pigs.

In the positive peripheral specimens, concentrations increased from 24 to 72 hours. Central-blood concentrations increased to 48 hours and then decreased. In contrast, HO-RCS-4 was not detected in postmortem peripheral or central blood in the 4°C group. The result was not universal across animals, time points or storage conditions.

Analytical sensitivity belongs with the finding

The blood LC–MS/MS method had a reported HO-RCS-4 detection limit of 0.15 ng/mL and lower quantification limit of 0.5 ng/mL. A nondetection was bounded by that method’s sensitivity; it did not establish that absolutely no metabolite was present.

The study separately quantified tissues and bile/duodenal material using standard addition after enzymatic cleavage and extraction. Concentrations from those processed matrices should not be silently equated with free-metabolite concentrations in blood.

What could explain the appearance?

The authors proposed redistribution from neighboring tissues or release of free HO-RCS-4 from glucuronide through postmortem hydrolysis. Those were possible explanations. The experiment did not directly apportion the observed blood increase between those mechanisms.

HO-RCS-4 was the most prevalent metabolite in the investigated postmortem material, but that did not make its blood concentration a validated intake clock. The study also showed substantial variability and warned against overinterpreting concentration changes.

Implications for human casework

The experiment supports caution when treating a postmortem parent/metabolite pattern as unchanged from life. It did not establish a human concentration cutoff, dose conversion, impairment measure or cause-of-death criterion. The controlled co-administration, anesthesia, species and sampling procedures all restrict transfer to an individual human death.

For another metabolite-attribution issue, see why clonazepam metabolites do not uniquely establish clonazafone exposure.

Primary source

Schaefer N, Kröll AK, Körbel C, et al. Time- and temperature-dependent postmortem concentration changes of the (synthetic) cannabinoids JWH-210, RCS-4, as well as Δ9-tetrahydrocannabinol following pulmonary administration to pigs. Archives of Toxicology. 2020;94:1585–1599. doi:10.1007/s00204-020-02707-4.

Ethanol in a child’s autopsy specimen could reflect ingestion or microbial production. In a seven-child diphenhydramine series, three ethanol-positive cases had different supporting evidence: one was interpreted as likely ingestion and two as likely microbial activity. The measured concentration alone did not decide the origin.

The cases came from a selected drug-positive series

Wallage and Elliot described seven Ontario deaths in children aged four months to eight years. The source cases were selected for diphenhydramine detection, principally through a 2016–2022 search of deaths under five, with selected older children added.

This design did not estimate the frequency of ethanol ingestion or postmortem formation among all pediatric deaths. The three ethanol findings were case examples within a selected series, not a representative three-person ethanol experiment.

The analytical result identified ethanol, not its source

Specimens were collected at autopsy, with blood and urine placed in tubes containing sodium fluoride and sodium citrate. Targeted headspace GC-FID testing addressed ethanol and other volatile substances. Screening and drug-identification procedures varied across the years represented.

A result from that method established ethanol in the tested sample. It did not distinguish a swallowed ethanol molecule from one produced by microorganisms. The report did not provide serial ethanol measurements from these children to demonstrate the direction or magnitude of change over time.

Case 2: scene evidence supported ingestion

A six-year-old child had ethanol at 18 mg/dL in femoral blood and diphenhydramine at 1.2 mg/L. Scene information described alcohol and adult sleeping medication. No decomposition was observed at autopsy.

The authors considered ingestion the likely ethanol origin, and the certified cause included combined diphenhydramine and ethanol toxicity. That was an integrated case interpretation using scene, pathology and toxicology. It was not a rule that any pediatric ethanol concentration of 18 mg/dL proves ingestion or fatal synergy.

Case 4: decomposition and another volatile supported production

A three-year-old child had ethanol at 49 mg/dL in heart blood. Widespread decomposition was noted, and n-propanol was detected during headspace analysis. Diphenhydramine was 0.17 mg/L and was considered incidental to death from complications of a viral infection.

The authors interpreted the ethanol as likely microbial production. They did not derive a numerical production correction from the n-propanol result. Its presence supported that interpretation in context, rather than measuring how much ethanol had formed.

Case 6: infection complicated the origin question

A two-year-old child had ethanol at 11 mg/dL in heart blood and 13 mg/dL in urine. Cultures showed heavy growth of Klebsiella pneumoniae, and the cause of death was septicemia complicating acute gastroenteritis. Diphenhydramine was 0.051 mg/L and considered incidental.

The authors again favored microbial activity as the ethanol explanation. A documented organism capable of contributing to fermentation made that interpretation plausible; it did not establish the precise time or amount of ethanol production in this child. The paired urine result was not converted to an earlier blood alcohol concentration.

Interpretation requires preserving uncertainty

These cases support examining specimen site, decomposition, microbiology, accompanying volatiles and scene information together. They do not establish a pediatric ethanol cutoff separating ingestion from production, and they do not show that every infection produces ethanol.

For related limits, see why absent 1-propanol cannot exclude microbial ethanol and why assay uncertainty does not quantify postmortem formation.

Primary source

Wallage HR, Elliot M. Postmortem diphenhydramine blood concentrations in children. Journal of Analytical Toxicology. 2025;49:129–133. doi:10.1093/jat/bkae084.

Detecting diphenhydramine in a child’s postmortem blood did not automatically establish fatal poisoning. A seven-case series included two deaths attributed wholly or partly to diphenhydramine and five deaths in which the finding was considered incidental. The concentrations informed the investigations but did not supply a universal pediatric lethal cutoff.

How the seven cases were selected

Wallage and Elliot reviewed Ontario Centre of Forensic Sciences cases from 2016–2022. They searched deaths under five years of age with diphenhydramine detected and added selected older children through database searching and author involvement in investigations. The resulting children were four months to eight years old: three males and four females.

This was a selected positive-case series, not a population survey or a controlled dosing study. It could describe findings in these deaths but could not calculate a child’s probability of dying at a particular concentration.

Blood site and preservation mattered

Specimens were collected at autopsy. Blood and urine were placed in tubes containing sodium fluoride and sodium citrate. Femoral blood was available in three older children; only heart blood was submitted in four younger children. The report did not establish a uniform death-to-autopsy interval for all seven.

Diphenhydramine ranged from 0.051 to 2.6 mg/L. In the two six-year-old children whose deaths were attributed to toxicity, femoral concentrations were 1.2 and 2.6 mg/L. The first death involved combined diphenhydramine and ethanol toxicity. The second case had obvious decomposition and limited specimen volume.

The five incidental detections were 0.11 mg/L in femoral blood and 0.051–0.46 mg/L in heart blood. Their causes of death were natural. These heart and femoral observations should not be pooled into a validated concentration boundary separating safe from fatal exposure.

Not every case had the same analytical evidence

The principal targeted LC–MS/MS method used a deuterated internal standard, two diphenhydramine transitions and detection and quantification limits of 0.0125 and 0.025 mg/L, respectively. One older analytical method had corresponding limits of 0.0625 and 0.125 mg/L.

Identification was confirmed using LC-QTOF-MS or GC-NPD-MS in six cases. Limited volume in case 3 allowed only duplicate diphenhydramine quantification, reported as the average; orthogonal identity confirmation was not performed. Duplicate quantitative measurements were not equivalent to the broader confirmation and screening available in the other cases.

Product identity did not reveal the administered dose

Scene or treatment information identified diphenhydramine products in three cases and dimenhydrinate products in three. Dimenhydrinate can yield detectable diphenhydramine, so the measured analyte did not uniquely identify a diphenhydramine-only preparation.

The two toxicity deaths involved adult sleep-aid products, but quantities ingested were unknown. The report did not reconstruct doses from the blood values. Nor did it establish the extent of redistribution in either individual.

What can be concluded in a new case?

Preserve the concentration, anatomical collection site, decomposition status, analytical scope and documented product history. Consider natural disease and co-detected substances alongside the drug finding. A higher value is evidence to investigate, not an automatic mechanism-of-death diagnosis.

The series adds pediatric case observations without proving an adult threshold can be transferred to children or supplying a new universal pediatric threshold. The same distinction between concentration and recovered dose is discussed in postmortem pregabalin interpretation.

Primary source

Wallage HR, Elliot M. Postmortem diphenhydramine blood concentrations in children. Journal of Analytical Toxicology. 2025;49:129–133. doi:10.1093/jat/bkae084.

Midazolam detected after a fatal crash does not, by itself, establish that the person took it before the collision. In a Milan autopsy review, investigators attributed midazolam findings in two crash cases to hospital treatment; one also had ketamine detected. The timing and source of exposure had to be interpreted through treatment records and circumstances.

What the crash cases actually showed

Di Candia and colleagues reviewed 136 cases referred for toxicological analysis among 1,323 autopsies during 2018–2019. Thirteen people in the reviewed series died following car accidents. Seven had negative toxicological findings, and four had ethanol concentrations reported between 0.83 and 1.38 g/L.

The remaining two cases involved midazolam. One was a male in the 21–30-year age category, identified as case 7. Another, case 123, had both midazolam and ketamine. The authors explained those findings by hospital administration for sedation and severe pain in the polytrauma setting.

This was a retrospective interpretation of two cases within a broader series, not a prospective trial of emergency treatment. It did not estimate how frequently all fatal-crash toxicology findings arise from medical administration.

The specimen did not timestamp administration

Biological samples were collected at autopsy and stored at −20°C. Femoral and cardiac blood, urine and gastric contents were the principal matrices across the study, with selected organs analyzed when needed. The investigators used HPLC–MS/MS and GC–MS, as well as headspace methods for alcohol and volatile substances.

Table 1 reports one car-accident midazolam finding at 0.09 µg/mL in cardiac blood and 0.08 µg/mL in femoral blood. Those are measurements in two postmortem specimens. Their similarity did not establish whether administration preceded the crash, followed rescue, or occurred during subsequent hospital treatment.

The main report did not provide the precise dosing and treatment-to-death timeline for these crash examples. It therefore supplies no validated concentration-to-administration-time conversion or estimate of a precrash dose.

Analytical identity and exposure provenance are different questions

The laboratory workflow moved from screening to further qualitative identification and, where confirmed, calibration-based quantification. Identifying a substance answers whether the analytical evidence supports its presence in that specimen.

Determining why it was present requires a separate timeline. Medication charts, emergency records, documented procedures and specimen collection times can place a detected drug within treatment. A concentration alone cannot replace that history.

Similarly, detecting both a sedative and another drug does not prove a precrash recreational combination. In these cases the authors considered hospital administration the explanation; their attribution should not be rewritten as independent proof of earlier self-administration.

What cannot be inferred about driving or causation

The series did not measure driving performance before the collision or demonstrate that midazolam caused either crash. A drug can be detected after treatment without having contributed to the event that required treatment.

The study’s interpretation also does not make every postmortem midazolam result automatically therapeutic. The relevant conclusion depends on evidence from the individual case. Treatment attribution in these two cases cannot be generalized to a person with different records or circumstances.

A useful way to read the report

Keep the analytical result, specimen site and unit alongside the exposure timeline. Identify which findings are measured and which are inferred from records. Preserve any missing administration details rather than filling them with a guessed dose or time.

These distinctions are relevant whenever hospital care precedes death. For another specimen-related complication in that setting, see hospital ethanol specimens and differences between serum and whole blood.

Primary source

Di Candia D, Giordano G, Boracchi M, Zoja R. Postmortem forensic toxicology cases: A retrospective review from Milan, Italy. Journal of Forensic Sciences. 2022;67:1640–1650. doi:10.1111/1556-4029.15050.

Dividing average cardiac drug concentration by average femoral concentration does not necessarily produce the average paired ratio, especially when the averages come from different people. A Milan autopsy review provides a concrete example: its cardiac and femoral cocaine means used different numbers of available specimens.

The source was a selected casework series

Di Candia and colleagues reviewed 136 cases with requested toxicological analyses among 1,323 autopsies performed during 2018–2019. Samples were collected at autopsy and stored at −20°C. Testing primarily involved femoral blood, cardiac blood, urine and gastric contents, with organs used selectively.

The report used HPLC–MS/MS, GC–MS and headspace analysis as appropriate. Its workflow separated library-based screening, qualitative identification and subsequent quantification using calibration standards. It was not a controlled serial-sampling experiment measuring the same person’s concentration before and after death.

The cocaine averages did not have identical denominators

Table 1 described 15 cocaine acute-intoxication cases in the single-substance category. Cardiac blood concentrations ranged from 0.33 to 19.8 µg/mL, with a reported average of 5.37 µg/mL. Femoral blood ranged from 0.49 to 18.78 µg/mL, with an average of 5.28 µg/mL.

The footnotes specify that the cardiac mean used 13 cases, whereas the femoral mean used 14. The ratio of these published averages is approximately 1.02, displayed as 1.0 in the table. The available-specimen sets were therefore not identical.

This is a materially different calculation from obtaining each decedent’s cardiac/femoral ratio and then summarizing those individual ratios. A ratio of means also weights observations differently from a mean of ratios, even when every person has both specimens.

Similar averages do not establish matching specimens

The closeness of 5.37 and 5.28 µg/mL cannot establish that all paired samples agreed. Aggregate values can conceal individual differences, and missing specimens can alter the comparison further. The table does not provide a validated prediction interval for estimating an individual’s femoral concentration from cardiac blood.

Results for different analytes also should not be combined into one universal conversion factor. The table included a ketamine case with cardiac blood at 5.69 µg/mL and femoral blood at 3.36 µg/mL, a ratio of approximately 1.7. One ketamine pair is descriptive evidence from one person, not a population calibration.

A site ratio is not a time-course measurement

Cardiac and femoral specimens sample different anatomical locations. A comparison between them describes site-dependent concentrations at collection. It does not directly measure the change from antemortem concentration to autopsy concentration.

The Milan series lacked a uniform reported postmortem interval and a paired predeath concentration dataset. A ratio near one therefore did not prove the absence of redistribution, while a ratio above one did not isolate redistribution from every other influence on the specimens.

How to use the table responsibly

Preserve each drug, specimen site, unit and available-sample denominator. Establish whether a published ratio is calculated from individual pairs or aggregate means before applying it to a case. Do not silently treat a missing specimen as zero or as agreement.

The measurements can inform a broader interpretation, but neither an average ratio nor an individual site ratio independently establishes dose, impairment or cause of death. For a separate drug-specific example, see fentanyl ratios near one and the limits of redistribution inference.

Primary source

Di Candia D, Giordano G, Boracchi M, Zoja R. Postmortem forensic toxicology cases: A retrospective review from Milan, Italy. Journal of Forensic Sciences. 2022;67:1640–1650. doi:10.1111/1556-4029.15050.

A Milan review found positive toxicology in 101 of 136 selected autopsies, or 74.3%. That did not mean 74.3% of all autopsied people used drugs, and it did not mean every positive result caused death. The series illustrates how referral for testing and causal assessment determine what an autopsy percentage means.

Only part of the autopsy population was tested

Di Candia and colleagues reviewed 2018–2019 cases at the University of Milan’s Bureau of Legal Medicine. There were 1,323 autopsies, but the Judicial Authority requested toxicological analysis in 136, approximately 10.3%.

The requests sought clarification of cause and manner of death. This selection process matters: a tested subgroup assembled for additional investigation is not a random sample of the general population, or necessarily of all autopsies. The other 1,187 cases were not 1,187 laboratory-negative observations.

What did a positive finding represent?

Samples were collected during autopsy and frozen at −20°C. Femoral and cardiac blood, urine and gastric contents were the principal matrices. Organs were examined in selected cases when conventional specimens were unavailable or further investigation was needed. The report does not provide a uniform death-to-collection interval for the series.

Testing used HPLC–MS/MS and GC–MS, with headspace testing for ethanol and volatile substances. The workflow distinguished initial library screening, further qualitative analysis and quantification using calibration standards. A reported positive was an analytical finding; assigning its role in death required another interpretive step.

Three denominators described three different results

Of the 136 tested cases, 101 were positive and 35 negative. The investigators classified 54 deaths as fatal acute intoxication using toxicology together with circumstances and autopsy findings.

Those 54 cases represented 39.7% of the tested series and 53.5% of the positive cases. They also represented approximately 4.1% of all 1,323 autopsies, although the paper rounds this overall proportion to 4.0%. These statements answer different questions and cannot be substituted for one another.

Positive findings in the remaining cases could still inform the investigation without constituting the primary poisoning diagnosis. Conversely, an untested autopsy cannot be counted as toxicologically negative simply to turn the overall proportion into an unbiased prevalence estimate.

A cross-country comparison needs comparable ascertainment

The authors discussed a lower overall intoxication proportion than in Finnish data, explicitly assuming that autopsies without requested toxicology could be treated as effectively negative. That assumption is a limitation, not a measured laboratory result.

Different testing policies, case referral, population coverage and definitions of drug-related death can alter apparent rates. This two-year, single-bureau retrospective review did not standardize those features across countries. It therefore cannot independently establish that one country’s population had a lower risk of fatal poisoning.

What the series supports

The findings demonstrate the usefulness of toxicology in selected medicolegal investigations. They do not establish a contemporary Milan-wide drug-use prevalence, a universal probability of poisoning after a positive test, or a drug concentration threshold for impairment.

When citing the percentages, preserve the testing decision, the matrix and the causal classification. In practical review, ask whether the numerator counts detections, positive cases or deaths attributed to intoxication, and whether the denominator is tested cases, positive cases or all autopsies. The distinction is also central to reconciling poisoning counts from death certificates and forensic files.

Primary source

Di Candia D, Giordano G, Boracchi M, Zoja R. Postmortem forensic toxicology cases: A retrospective review from Milan, Italy. Journal of Forensic Sciences. 2022;67:1640–1650. doi:10.1111/1556-4029.15050.