Development and validation of a model to predict acute kidney injury following wasp stings: A multicentre study

Abstract

Objective: To establish and validate a model to predict acute kidney injury (AKI) following wasp stings.

Methods: In this multicentre prospective study, 508 patients with wasp stings from July 2015 to December 2019 were randomly divided into a training set (n = 381) and a validation set (n = 127) for internal and external validation. Risk factors were identified, and a model was established to predict the probability of AKI following multiple wasp stings using an individual nomogram and a predictive formula. The performances of the model were assessed by using the area under the curve (AUC), accuracy (ACC) of the receiver operating characteristic curve and decision curve analysis.

Results: The number of stings, aspartate aminotransferase >147 U/L, lactate dehydrogenase >477 U/L, time from stings to admission >12 h and activated partial thromboplastin time >49 s were demonstrated to be independent risk factors for AKI following wasp stings (all P value < 0.05) and were incorporated into the model. The performances of the model were validated (AUC = 0.950 [95% CI: 0.923 to 0.969], ACC = 0.916 and AUC = 0.953 [95% CI: 0.900 to 0.982], ACC = 0.906 in the training set and validation set, respectively). The predictive formula and the nomogram of the model could be utilized to predict AKI following wasp stings, which have sufficient accuracies, good predictive capabilities and good net benefits.

Conclusion: The predictive formula and the individual nomogram of the model might serve as promising predictive tools to assess the probability of AKI following wasp stings.

Keywords: Acute kidney injury; Nomogram; Predictive formula; Risk factors; Wasp stings.