Death certificates, forensic files and integrated case review did not produce identical pharmaceutical-poisoning counts in a Swedish population study. Twelve deaths were classified from certificates, ten from medico-legal files and nine after multidisciplinary review of all available records. Adding those counts would double-count overlapping assessments rather than identify 31 separate deaths.
The population was sampled beyond the autopsy room
Jönsson and colleagues studied deaths during 2001 in Östergötland, Jönköping and Kalmar counties, with a mean population of 974,884. They randomly selected one in every seven of the 11,015 recorded deaths, yielding 1,574 study subjects.
Certificates were obtained for all 1,574. Medical records were available for 1,503, prescription information for 1,553 and forensic autopsy records for 83. Thus, this was not a toxicology-screened series of 1,574 autopsies. Most subjects were assessed through records rather than a uniform postmortem laboratory protocol.
What counted as pharmaceutical poisoning?
The study concerned pharmaceutical agents, not every possible illicit-drug or chemical poisoning. Review focused particularly on drug histories during the 14 days before death. Two pharmacists, two clinical pharmacologists and a forensic pathologist evaluated possible cases.
Classification required consensus and at least a possible relationship between drug toxicity and death, with known toxic potential and reasonable timing. Other contributing causes could coexist. The study’s classification was therefore a structured causality judgment, not a rule equating a positive drug result with poisoning.
The three counts answered different questions
Death certificates identified 12 possible pharmaceutical-poisoning fatalities and medico-legal files identified ten. Integrated review retained nine. All nine were represented on death certificates, which also listed three additional cases. The study did not show that consulting more records necessarily increases the final count; fuller information can also lead reviewers to exclude a suspected case.
Cases with more likely alternative explanations or unresolved inconsistencies were excluded under the authors’ conservative assessment. This approach reduced some classification uncertainty but did not establish an infallible reference standard for the true cause of every death.
Of the nine retained cases, four were intentional, three unintentional and two of unknown intent. Only seven had forensic autopsies. Reviewing every tenth of the 1,565 remaining deaths found no additional pharmaceutical-poisoning cases; that check was not exhaustive re-examination of all negatives.
What toxicology contributed when available
Forensic autopsies included femoral blood, urine and vitreous screening for drugs and ethanol. Drug concentrations were interpreted alongside medical history and circumstances. The report did not provide a uniform collection interval, storage history or numerical analytical-performance dataset covering all subjects.
Table 2 illustrates why analyte identity matters: one nitrazepam-associated result was actually measured as its metabolite 7-aminonitrazepam, reported at 0.7 µg/g in femoral blood. That should not be relabeled as a measured parent-nitrazepam concentration. Some retained poisoning cases lacked a postmortem femoral concentration altogether.
Two different denominators produced two different measures
The nine cases represented 0.57% of sampled deaths, with a 95% confidence interval of 0.20–0.94%. The estimated population incidence was 6.5 per 100,000 person-years, with a 95% interval of 2.3–10.7. A proportion of deaths and an incidence rate in the general population are not interchangeable quantities.
These are historical, regional estimates published in 2009 from 2001 data. They do not establish today’s poisoning incidence or an individual’s risk from a prescription. Only nine cases contributed to the final estimate, limiting precision and comparisons among drugs.
What the study supports in case review
Toxicological measurements, diagnoses on certificates and final causal assessments should be kept distinct and reconciled through the underlying records. The study supports integrating those sources while preserving missing evidence and alternative causes. It does not validate a concentration threshold, impairment test or automatic cause-of-death rule.
For another counting distinction, see unconfirmed findings in postmortem surveillance totals.
Primary source
Jönsson AK, Spigset O, Tjäderborn M, Druid H, Hägg S. Fatal drug poisonings in a Swedish general population. BMC Clinical Pharmacology. 2009;9:7. doi:10.1186/1472-6904-9-7.
Leave a Reply