刘孟双, 金克峙, 王亚, 应佳丽, 杨琛. 道路交通伤害死亡风险的影响因素分析及预测[J]. 环境与职业医学, 2021, 38(11): 1224-1230. DOI: 10.13213/j.cnki.jeom.2021.21146
引用本文: 刘孟双, 金克峙, 王亚, 应佳丽, 杨琛. 道路交通伤害死亡风险的影响因素分析及预测[J]. 环境与职业医学, 2021, 38(11): 1224-1230. DOI: 10.13213/j.cnki.jeom.2021.21146
LIU Mengshuang, JIN Kezhi, WANG Ya, YING Jiali, YANG Chen. Analysis of factors affecting fatality risk in road traffic injury[J]. Journal of Environmental and Occupational Medicine, 2021, 38(11): 1224-1230. DOI: 10.13213/j.cnki.jeom.2021.21146
Citation: LIU Mengshuang, JIN Kezhi, WANG Ya, YING Jiali, YANG Chen. Analysis of factors affecting fatality risk in road traffic injury[J]. Journal of Environmental and Occupational Medicine, 2021, 38(11): 1224-1230. DOI: 10.13213/j.cnki.jeom.2021.21146

道路交通伤害死亡风险的影响因素分析及预测

Analysis of factors affecting fatality risk in road traffic injury

  • 摘要: 背景

    近几年,道路交通伤害(RTI)已成为中国严重的公共卫生问题,RTI死亡风险的影响因素也较为复杂。

    目的

    寻找RTI死亡风险的影响因素,建立风险预测模型。

    方法

    回顾性收集2010—2016年间上海市浦东新区交通事故伤亡资料,并收集道路交通安全相关影响因素。采用logistic回归筛选RTI死亡风险的影响因素。建立RTI死亡风险列线图模型,用C-index评价模型的一致性和准确性,使用自抽样法对模型进行内部验证,并进行敏感性分析。

    结果

    研究共纳入3521名交通事故中伤亡的人员。logistic回归结果显示伤亡人员的年龄、医疗救援距离、道路类型、交通方式、受伤部位、事故发生时间、是否为工作日均对RTI死亡风险的影响有统计学意义(P<0.05)。以此建立RTI死亡风险列线图,模型显示影响最大的因素是受伤部位(尤其是头颈部受伤),其次是年龄、交通方式、医疗救援距离、道路类型、事故发生时间、是否为工作日。模型的C-index为0.790,说明模型预测结果准确度良好,模型拟合良好。建立头颈部受伤的RTI死亡风险列线图模型,结果显示各纳入因素的评分标尺均有膨胀,最突出的是年龄,即影响最大的因素;不同道路类型对RTI死亡影响的风险改变,城市公路成为风险最大的道路类型;步行成为头颈部伤RTI死亡风险最大的交通方式。对不同伤亡人数的事故进行敏感性分析,结果显示所建立的模型具有一定的稳健性。

    结论

    RTI死亡风险受到诸多因素的影响。基于logistic回归建立的列线图作为预测RTI死亡风险的简易工具,对道路交通安全具有一定的参考意义。

     

    Abstract: Background

    In recent years, road traffic injury (RTI) has become a serious public health problem in China, and the factors affecting deaths caused by RTI are also complicated.

    Objective

    This study is designed to identify factors of RTI fatality risk and establish a road user fatality risk prediction model.

    Methods

    The data of traffic accident casualties in Pudong New Area of Shanghai from 2010 to 2016 were collected retrospectively, and the related impact factors of RTI were collected. Logistic regression was used to screen the selected factors of RTI fatality risk. A nomogram of RTI fatality risk was established, the consistency and accuracy of the model was evaluated by C-index and bootstrap internal verification, and a sensitivity analysis was also conducted.

    Results

    A total of 3521 casualties in traffic accidents were included in the study. The logistic regression results showed that age of victims, medical rescue distance, road type, transport means, injured body part, time of accident, and weekday/weekend affected RTI death risk (P<0.05). The nomogram model for RTI death risk showed that the most affecting factors were injured body part (especially head and neck injury), followed by age, transportation means, medical rescue distance, road type, time of accident, and weekday/weekend. The C-index of the model was 0.790, indicating high prediction accuracy and good fitness. The nomogram model for RTI death risk of head and neck injury showed that the score scales of all included factors expanded, the most prominent (most affecting) one was age; the RTI fatality risk of different road types changed, where urban road became the most dangerous road type; in addition, walking was the transportation means with the greatest risk of RTI fatality from head and neck injury. The results of the sensitivity analysis on accidents with varied casualties confirmed the robustness of the model.

    Conclusion

    The road user fatality risk of RTI is affected by many factors. As a simple tool to predict fatality risk of RTI, the nomogram based on logistic regression has certain reference significance for road traffic safety.

     

/

返回文章
返回