Regression model for predicting the severity of acute poisoning cases by estimating PSS in the Emergency Department

Abstract

Acute poisonings represent an emergency pathology that associates an increased risk of unfavorable outcomes or death. The mortality rate can be improved by the fast recognition of its severity, which in turn allows a prompt intervention of the medical team. In the practical approach of the case, a standardized measure of predicting the patient's evolution would be necessary, which could be applied quickly to the patient's bed and easy to calculate and apply irrespective of the evaluator. Currently, for acute poisoning cases, Poisoning Severity Score (PSS) is used, a complex and dependable tool that accurately stages the severity of the case but has the disadvantage of being quantified retrospectively. This study presents the development and validation of a linear regression model that can be applied right in the emergency department (ED) and predicts the severity of the case by estimating PSS with an accuracy of 75%. The proposed model uses ten objective and quantifiable variables representing anamnestic, clinical, and biological parameters evaluated in the early stages of the poisoning. The regression was developed in the study group consisted of 62 pediatric patients diagnosed with severe acute poisoning with cardiotoxic agents complicated by cardiogenic shock.