A finding that clozapine contributed to death in approximately 45% of a selected autopsy series does not mean that 45% of people taking clozapine will die from it. In a 2026 study, the numerator was 20 cases with an attributed causal role and the denominator was 44 clozapine-positive autopsies with paired blood and brain quantification. The study did not follow a population of treated patients.
Who was eligible to enter the study?
Van der Meer and colleagues reviewed Copenhagen forensic cases investigated from late 2014 to early 2026. The overall study included 466 autopsy cases involving one or more of four antipsychotics. Entry required quantification of a selected drug in both blood and brain. Clozapine was quantified in 44 cases.
This selection excludes living recipients and does not enumerate everyone prescribed clozapine in the source population. It also requires suitable paired specimens and measurable drug. Consequently, 20/44 describes attribution within the included autopsies, not the probability of fatal poisoning after a prescription, the risk per dose or the frequency of death among all exposed people.
What did “contributed to death” include?
Class A meant that the antipsychotic was the sole cause of death. Class B meant contribution together with other pharmaceuticals, drugs of abuse, alcohol or disease. Class C meant no attributed contribution. For clozapine, there were 12 class A cases, eight class B cases and 24 class C cases.
The approximately 45% figure combines A and B: 20/44. It should not be relabeled as 20 deaths caused by clozapine alone. Conversely, the 24 class C cases were deaths in which clozapine was detected without an attributed causal role; they were not living, symptom-free control patients.
Two forensic toxicologists evaluated the analytical findings in the original investigations, and the forensic pathologist determined the final cause with case circumstances. A toxicologist assigned the study categories from those conclusions and toxicity remarks. These were case-based judgments incorporating toxicology, not outcomes established independently of the concentrations being compared.
Concentrations helped interpretation without creating a universal cutoff
The combined A+B clozapine group had a median blood concentration of 3.1 mg/kg, with 10th–90th percentiles of 1.4–7.1 mg/kg. Its median brain concentration was 8.5 mg/kg, with percentiles of 3.4–17 mg/kg. In class C, the corresponding medians were 1.3 mg/kg in blood and 2.9 mg/kg in brain, with percentile intervals of 0.16–2.8 and 0.59–8.2 mg/kg.
The higher medians in attributed cases are useful descriptive findings. The overlapping intervals prevent interpreting the group summaries as an infallible individual boundary. They are percentile intervals rather than full ranges, and a class C observation does not prove that the same concentration would be harmless in another person with different co-exposures or disease.
The median brain–blood ratio was 3.0 in A+B and 2.8 in C. Across the four drugs, ratios did not differ significantly between causal categories. Neither a positive finding nor a ratio near the group median supplied an independent cause-of-death test.
Why collection and analytical conditions remain relevant
Blood was collected from the femoral site without iliac-vein ligation, using 4–8 mL tubes with sodium fluoride and potassium oxalate. Frontal cerebral cortex was sampled for brain analysis, with grey matter separated when feasible. Bodies were kept at 5°C after arrival, autopsy generally occurred within four days of arrival, and specimens were stored at −20°C. The actual postmortem intervals were unavailable, preventing assessment of redistribution over time.
UHPLC–MS/MS quantification used blood calibrators and quality controls in both matrices. For clozapine, the reported measurement range was 0.005–0.20 mg/kg. Brain quality-control imprecision at 0.25 mg/kg was 11.6% across 89 measurements. The authors acknowledged uncertainty from tissue composition, mixed tissue in advanced decomposition, calibration and possible drug interactions. These factors limit how directly another specimen can be compared with the reference data.
Can the percentage rank antipsychotics by danger?
The study reported a larger attributed proportion for clozapine than for the other included antipsychotics. That observation is not, on its own, a comparative risk estimate for prescribing. The drug groups differed in size and were selected from deaths; there were no matched exposure denominators accounting for the number, duration and characteristics of treated patients.
A defensible opinion identifies the actual question: how frequently clozapine was judged relevant among these selected autopsies. Estimating a patient’s treatment-related mortality, an individual fatal dose or the probability that a particular result caused death would require additional evidence.
For a broader example of the difference between drug detection and fatal intoxication, see the 2,996-case autopsy screening study.
Primary source
van der Meer MZ, Rasmussen BS, Nedahl M, Nielsen MKK. From blood to brain: postmortem brain concentrations and brain–blood ratios of chlorprothixene, clozapine, olanzapine, and quetiapine. Journal of Analytical Toxicology. 2026;50(7):bkag065. doi:10.1093/jat/bkag065.
Leave a Reply