Predicting mortality from acetaminophen poisoning shortly after hospital presentation

Aims: Early identification of patients likely to die after acetaminophen (APAP) poisoning remains challenging. We sought to compare the sensitivity and time to fulfilment (latency) of established prognostic criteria. Methods: Three physician toxicologists independently classified every in-hospital death associated with APAP overdose from eight large Canadian cities over three decades using the Relative Contribution to Fatality scale from the American Association of Poison Control Centres. The sensitivity and latency were calculated for each of the following criteria: King's College Hospital (KCH), Model for End Stage Liver Disease (MELD) ≥33, lactate ≥3.5 mmol/L, phosphate ≥1.2 mmol/L 48+ hours post-ingestion, as well as combinations thereof. Results: A total of 162 in-hospital deaths were classified with respect to APAP as follows: 26 Undoubtedly, 40 Probably, 27 Contributory, 14 Probably not, 25 Clearly not, and 30 Unknown. Cases from the first three classes (combined into n = 93 “APAP deaths”) typically presented with supratherapeutic APAP concentrations, hepatotoxicity, acidaemia, coagulopathy and/or encephalopathy, and began antidotal treatment a median of 12 hours (IQR 3.4–30 h) from the end of ingestion. Among all patients deemed “APAP deaths”, meeting either KCH or lactate criteria demonstrated the highest sensitivity (94%; 95% CI 86–98%), and the shortest latency from hospital arrival to criterion fulfilment (median 4.2 h; IQR 1.0–16 h). In comparison, the MELD criterion demonstrated a substantially lower sensitivity (55%; 43–66%) and longer latency (52 h; 4.4–∞ h, where “∞” denotes death prior to criterion becoming positive). Conclusions: Meeting either KCH or serum lactate criteria identifies most patients who die from acetaminophen poisoning at or shortly after hospital presentation.