Department of Emergency, Taihe Hospital, Hubei University of Medicine, Shiyan, 442000, Hubei, People's Republic of China.
The Intensive Care Unit, The First Dongguan Affiliated Hospital, Guangdong Medical University, Dongguan, 523710, Guangdong, People's Republic of China.
Sci Rep. 2024 Oct 31;14(1):26242. doi: 10.1038/s41598-024-77152-9.
Numerous studies have shown that wasp stings can lead to serious, sometimes fatal, health outcomes. Predicting deaths associated with wasp stings remains challenging yet is of critical importance. This study was conducted to identify predictors and develop a visual model for predicting mortality following wasp stings. Clinical data from 486 patients were analyzed, dividing them into two groups: survival group (N = 435) and death group (N = 51). Various statistical methods were used to create a prognostic model, including one-way analysis, the least absolute shrinkage and selection operator (LASSO) regression, and binary logistic regression. The model's accuracy was evaluated through ROC curves, calibration plots, and decision curve analysis (DCA). The study identified four key predictors of mortality: receiving more than 50 stings, having serum lactate dehydrogenase (LDH) levels of ≥ 2200 U/L, activated partial thromboplastin time (APTT) of ≥ 90 s, and the requirement for invasive mechanical ventilation within 24 h. These factors contributed to a model with an area under the ROC curve of 0.980 (95% CI: [0.968-0.992]), indicating high calibration and applicability. The decision curve analysis confirmed the model's substantial net clinical benefit. Thus, the number of stings, serum LDH, APTT, and the need for early invasive mechanical ventilation are reliable, independent predictors of death among patients experiencing wasp stings. The developed predictive model exhibits high levels of accuracy, sensitivity, consistency, and practical use.
大量研究表明，黄蜂蜇伤可导致严重甚至致命的健康后果。预测与黄蜂蜇伤相关的死亡仍然具有挑战性，但至关重要。本研究旨在确定预测指标，并建立一个预测黄蜂蜇伤后死亡率的可视化模型。对 486 例患者的临床数据进行分析，将其分为存活组（N=435）和死亡组（N=51）。采用单因素分析、最小绝对收缩和选择算子（LASSO）回归和二项逻辑回归等多种统计方法建立预后模型。通过 ROC 曲线、校准图和决策曲线分析（DCA）评估模型的准确性。研究确定了死亡的四个关键预测指标：被蜇超过 50 次、血清乳酸脱氢酶（LDH）水平≥2200 U/L、活化部分凝血活酶时间（APTT）≥90 s、24 小时内需要有创机械通气。这些因素导致 ROC 曲线下面积为 0.980（95%CI：[0.968-0.992]）的模型，提示校准度和适用性高。决策曲线分析证实了该模型具有显著的净临床获益。因此，蜇伤次数、血清 LDH、APTT 和早期有创机械通气的需求是黄蜂蜇伤患者死亡的可靠、独立预测指标。所建立的预测模型具有高准确性、高灵敏度、高一致性和高实用性。
