Does the ICU Requirement Score allow the poisoned patient to be safely managed without admission to the intensive care unit? - a validation cohort study

Background The Intensive care unit (ICU) Requirement Score (IRS) has been defined as identifying poisoned patients on hospital admission who do not require ICU referral, in an effort to reduce health expenses. However, this score has been poorly validated. We aimed to evaluate the IRS in a large cohort of poisoned patients. Methods We performed a single-center retrospective cohort study. IRS was calculated using clinical parameters obtained on admission including age, systolic blood pressure, heart rate, Glasgow coma score, intoxication type, co-morbidities (i.e., arrhythmia, cirrhosis, and respiratory insufficiency), and the combination of the intoxication with another reason for ICU admission. We evaluated the ability of IRS < 6 determined on admission to predict the lack of need for ICU treatment, defined as the need for mechanical ventilation, vasopressors, and/or renal replacement therapy in the first 24 h post-admission and/or death during the hospital stay. This score was compared to the usual prognostic scores, i.e., SAPS II and III, SOFA score, and PSS. Results During the 10-year study period, 2,514 poisoned patients were admitted, 1,011 excluded as requiring ICU treatment on admission, and 1,503 included. Among these patients, 232 met the endpoint whereas only 23/510 patients with IRS < 6 (4.5%) presented the endpoint and one patient died. The area under the curve of the IRS ROC curve was 0.736 (95% confidence interval (CI), 0.702-0.770). The negative predictive value of IRS < 6 was 95% (95% CI, 93-97), sensitivity 89% (95% CI, 85-93), specificity 38% (95% CI, 36-41), and positive predictive value 21% (95% CI, 18-24). IRS performance was similar to those of the other tested scores, which are however not readily available on admission. Conclusion Our data demonstrate the excellent negative predictive value of the IRS, allowing the exclusion of ICU requirements for poisoned patients with IRS < 6. IRS usefulness should be confirmed based on a prospective multicenter cohort study before extensive routine use.