Study objective After acute intoxication, most patients presenting to the emergency department (ED)—76% of them in the Netherlands—are admitted to hospital. Many will not need medical treatment on the ward. The authors tested two algorithms in the ED, based on vital parameters, ECG findings, and ingested substances, to identify patients who will receive treatment in hospital.

Methods This prospective inception study enrolled patients aged 14 years and older presenting with acute intoxication between January 2006 and April 2008 to a Dutch university hospital. An algorithm was developed based on a previous retrospective study and the medical literature. In a second algorithm the clinical course during the stay in the ED was also taken into account.

Results Of 313 patients presenting with acute intoxication to the ED, 134 (42.8%) were admitted to a ward for somatic care, but only 74 (23.6%) were treated on the ward. Algorithm 1 had 91.9% sensitivity (95% CI 82.6% to 96.7%) and 53.6% specificity (95% CI 47.0% to 60.0%). Algorithm 2 had 90.5% sensitivity (95% CI 80.9% to 95.8%) and 65.3% specificity (95% CI 58.8% to 71.2%). In line with hospital policy, several patients received N-acetylcysteine treatment for subtoxic paracetamol ingestion because they presented outside of office hours, when no measurements of blood paracetamol concentration are performed by the laboratory. When these patients are considered as untreated, both algorithms had 98.5% sensitivity (95% CI 90.6% to 99.9%).

Conclusion The algorithms had good sensitivity and better specificity than current clinical practice in most hospitals. It is too early to advocate their implementation, but results indicate that it is possible to use clinical parameters objectively to reduce unnecessary admissions to the ward.