摘要
目的探讨基于分位数回归模型分析血流动力学不稳骨盆骨折患者临床小样本数据的应用价值。方法回顾性分析自2018-07—2021-06诊治的27例血流动力学不稳骨盆骨折,基于分位数回归模型进行统计推断,考虑了1/4、1/2、3/4分位数,研究协变量(患者年龄、性别、致伤原因、损伤严重程度、休克程度)和自变量(血管介入止血术、外科损伤控制术)与应变量(存活及住院时间)的关系。结果23例存活(存活率85.2%),经治疗后好转或治愈出院,住院时间13~241 d。统计学分析结果显示,在1/4分位数上,是否进行血管介入止血术对患者住院时间没有影响,差异无统计学意义(P=0.192);年龄、损伤严重程度、休克程度、外科损伤控制术对患者住院时间有影响,差异有统计学意义(P<0.05)。在中位数上,损伤严重程度对患者住院时间有影响,差异有统计学意义(P<0.05);年龄、休克程度、血管介入止血术、外科损伤控制术对患者住院时间没有影响,差异无统计学意义(P>0.05)。在3/4分位数上时,年龄、损伤严重程度、休克程度、血管介入止血术、外科损伤控制术对患者住院时间均有明显影响,差异有统计学意义(P<0.001)。结论危重创伤患者临床数据协变量信息经常缺失或相对不准确,基于分位数回归模型分析预后是一种值得借鉴的方法。
Objective To explore the application value of quantile regression model in the analysis of small clinical sample data of patients with hemodynamically unstable pelvic fractures.Methods A retrospective analysis for 27 cases of hemodynamically unstable pelvic fractures diagnosed and treated from July 2018 to June 2021 was conducted.Statistical inference was made based on quantile regression model,considering 1/4,1/2 and 3/4 quantiles,respectively.The relationship between covariates(age,gender,cause of injury,severity of injury,degree of shock)together with medical variables(vascular interventional hemostasis,surgical injury control)and dependent variables(survival status and hospital day)was analyzed by quantile regression model.Results Twenty-three cases survived(survival rate 85.2%),and were improved or cured after treatment,and the length of stay in hospital was between 13 and 241 days.The results of statistical analysis showed that for the 1/4 quantile,vascular interventional hemostasis had no effect on the length of hospital stay with significant level(P=0.192);the length of hospital stay was significant affected by age,injury severity,shock degree and surgical injury control(P<0.05).The severity of injury had an effect on the median for the length of hospital stay,and the difference was statistically significant(P<0.05);while age,shock degree,vascular interventional hemostasis and surgical injury control had no effect on the length of hospital stay(P>0.05).Finally,the research showed that age,injury severity,shock degree,vascular interventional hemostasis and surgical injury control had significant effects on the length of hospital stay for the 3/4 quantile(P<0.001).Conclusion When the covariate information of severe traumatic clinical data is often missing or incomplete,the quantile regression model is a feasible method to analyze prognosis.
作者
欧阳林
胡天琳
彭新建
施建华
OUYANG Lin;HU Tian-lin;PENG Xin-jian;SHI Jian-hua(Department of Diagnostic Radiology and Interventional Therapy,the 909th Hospital of Joint Logistic Support Force(Dongnan Hospital of Xiamen University),Zhangzhou,Fujian 363000,China;不详)
出处
《中国骨与关节损伤杂志》
2022年第4期342-345,共4页
Chinese Journal of Bone and Joint Injury
基金
军队后勤科研计划项目(CLB18J060,CLB21J017)
福建省科技计划项目(2020J01794,2019Y3007)
漳州市科技拥军项目(ZZ2018KD01)。
