Scorpion Envenomation: An Intensive Care Unit Transfer Prediction Score.

Abstract

Background: Scorpion envenomation frequently requires critical care intervention. Early determination of cases with a higher likelihood of clinical deterioration may allow for resource optimization and improvement of clinical endpoints. In this study, patients from a tertiary hospital with similar clinical characteristics but diverse outcomes were analyzed to propose a clinical deterioration prediction score.

Methods: Demographic, clinical, laboratory, and imaging findings of patients with scorpion envenomation were retrieved from a database used in a toxicology center based on in-person attendance data from 2019 to 2024. Multivariate models were defined and evaluated using receiver operating characteristic curves. A nomogram was created to estimate the probability of ICU transfer. Optimal cutoff points, sensitivity, and specificity were computed. Multiple imputation was performed using chained equations. Normality of errors in linear regressions was assessed via QQ plots.

Results: Overall, 146 patients were eligible for enrolment. Sex (p = 0.06), weight (p = 0.30), time-to-serum (p = 0.89), and exposure area (p = 0.49) were comparable between patients admitted to the intensive care unit (ICU) and other inpatients, whereas age (p = 0.006) and initial severity (p < 0.001) displayed significant differences. Logistic regression models displayed area-under-the-curve metrics ranging from 65% to 93%, with high specificity. Hyperamylasemia (p = 0.004), vomiting (p = 0.006), and cardiac dysfunction on echocardiography or point-of-care ultrasound (p < 0.001) were strongly associated with critical care.

Conclusion: Prognostication of factors leading to ICU transfer from initial clinical and laboratory findings is feasible. The regression model achieved high specificity and moderate sensitivity; being a potentially useful confirmatory test in cases with clinical suspicion.