Emergency risk stratification for early respiratory failure in acute pesticide poisoning: a nomogram prediction model.

Abstract

To develop and internally validate a practical nomogram for predicting early respiratory failure in acute pesticide poisoning and to benchmark its performance against the National Early Warning Score (NEWS). This retrospective cohort study included 205 acute pesticide poisoning patients admitted to a tertiary emergency department (ED) between November 2022 and April 2025. The primary outcome was respiratory failure or invasive ventilation within 48 h. Predictors were selected from initial assessment data (within 6 h) using least absolute shrinkage and selection operator regression to build a multivariable logistic model. Performance was assessed by discrimination [area under the curve (AUC)], calibration, and decision curve analysis, with specific comparison to NEWS using the DeLong test. Respiratory failure occurred in 55 (26.8%) patients. The final model comprised decreased level of consciousness (Glasgow Coma Scale < 15), organophosphate exposure, arterial pH, and serum creatinine. The nomogram demonstrated excellent discrimination (AUC 0.882; 95% confidence interval 0.829-0.934), statistically outperforming NEWS (AUC 0.789; P = .017). Calibration was robust (Hosmer-Lemeshow P = .503). A high-sensitivity screening threshold of 0.25 yielded a negative predictive value of 91.2%. Risk stratification effectively classified patients into low (<0.20; event rate 9.6%), intermediate (0.20-0.50; 28.2%), and high (>0.50; 78.0%) risk tiers. This simple four-variable nomogram provides accurate early risk assessment for respiratory failure in acute pesticide poisoning, offering superior discrimination over standard NEWS. The proposed risk stratification framework can support ED triage decisions regarding monitoring intensity and critical care resource allocation. Key messages What is already known on this topic: Existing severity scores (e.g. PSS, NEWS) are either too complex for rapid triage or lack specificity for the toxicological and metabolic risks of acute pesticide poisoning. What this study adds: This study develops and internally validates a simple four-variable nomogram (decreased level of consciousness, organophosphate poisoning, arterial pH, serum creatinine) that predicts early respiratory failure with significantly better discrimination than the standard NEWS score. How this study might affect research, practice or policy: By providing a practical three-tier risk stratification framework, this bedside tool may support emergency physicians in optimizing monitoring intensity and critical care resource allocation, pending external validation.