The performance of COBRA, a decision rule to predict the need for intensive care interventions in intentional drug overdose.

Background: COBRA was developed as a decision rule to predict which patients visiting the emergency department (ED) following intentional drug overdose will not require intensive care unit (ICU) interventions. COBRA uses parameters from five vital systems (cardiac conduction, oxygenation, blood pressure, respiration, and awareness) that are readily available in the ED. COBRA recommends against ICU admission when all these parameters are normal.
Objective: The primary aim of this study was to determine the negative predictive value (NPV) of COBRA in predicting ICU interventions. Secondary outcomes were the sensitivity, specificity and positive predictive value (PPV), and the observation time required for a reliable prediction.
Design: Observational cohort study.
Settings and Participants: Patients with a reported intentional overdose with drugs having potential acute effects on neurological, circulatory or ventilatory function were included, and data necessary to complete the decision rule was collected. The attending physician in the ED made the actual admission decision, on the basis of clinical judgement. COBRA was measured 0, 3 and 6 h after arrival at the ED.
Outcome Measures: Need for ICU interventions (treatment of convulsion; defibrillation; mechanical or noninvasive ventilation; intravenous administration of vasopressive agents, antiarrhythmics, atropine, calcium, magnesium or sedation; continuous hemofiltration or administration of antagonist/antidote and fluid resuscitation).
Main Results: Of 230 new cases (144 unique patients), 59 were immediately referred to the psychiatric services and/or sent home by the attending physician, 27 went to a regular ward, and 144 were admitted to the ICU. Of these 144 cases, 40 required one or more ICU interventions. By the time the first parameters were collected, the NPV of COBRA was 95.6%. After 3 h of observation, NPV was 100%, while sensitivity, specificity and PPV were 100, 61.1 and 35.1%, respectively. None of these values improved by prolonging the observation time to 6 h.
Conclusion: In patients with a reported intentional overdose with drugs having potential acute effects on neurological, circulatory or ventilatory function, the COBRA decision rule showed good performances in predicting the need for intensive care interventions, with a NPV of 100% after 3 h of observation.
(Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved.)