以CUSUM方法分析达芬奇机器人肺叶切除术的学习曲线  

The learning curve of da Vinci robot-assisted pulmonary lobectomy by using cumulative sum analysis

作　　者：赖湘敏 刘博[1] 孙灿 许达强 史彦鹏 胡博潇 刘星池[1] 许世广[1] 王述民[1] Lai Xiangmin;Liu Bo;Sun Can;Xu Daqiang;Shi Yanpeng;Hu Boxiao;Liu Xingchi;Xu Shiguang;Wang Shumin(Department of Thoracic Surgery,General Hospital of Shenyang Military Region,Shenyang 110016,China)

机构地区：[1]沈阳军区总医院胸外科,110016

出　　处：《中华腔镜外科杂志（电子版）》2018年第4期220-223,共4页Chinese Journal of Laparoscopic Surgery(Electronic Edition)

基　　金：辽宁省科学技术计划项目(2015020431)

摘　　要：目的探讨109例达芬奇机器人肺叶切除术的学习曲线。方法以累积和分析法(cumulative sum analysis,CUSUM)分析沈阳军区总医院2012年3月至2015年2月完成的第1例至第109例达芬奇机器人肺叶切除术的手术时间。采用多元线性回归排除影响手术时间的因素。对CUSUM学习曲线进行拟合,以R^2判断拟合优度。比较学习曲线不同阶段的手术时间、术中出血量、淋巴结清扫、术后并发症、术后带管时间和住院时间的差异。结果手术时间随手术例数的累积呈逐渐下降趋势。学习曲线最佳拟合为三次方曲线,拟合优度系数R^2=0. 847,拟合方程:CUSUM(n)=0. 002×n^3-0. 323×n^2+13. 360×n-18. 195(n为手术例数)。拟合曲线在手术例数累积至第28例时达到顶点,以此为分界将学习曲线划分为两个阶段,其手术时间(P=0. 022)、术中出血量(P=0. 014)、清扫淋巴结数目(P=0. 022)、术后并发症(P=0. 015)、术后带管时间(P=0. 025)和住院时间(P=0. 005)均存在显著的统计学差异。结论通过CUSUM分析法对达芬奇机器人肺叶切除术的学习曲线进行精确分析,表明掌握该技术须累积的手术例数为28例。Objective This study aims to investigate the learning curve of da Vinci robot- assisted pulmonary lobectomy. Methods A cumulative sum analysis （CUSUM） was performed on the operation time （OT） of the 1st to the 109 th patients who underwent da Vinci robot- assisted pulmonary lobectomy from Mar. 2012 to Feb. 2015 in our hospital. A multivariate linear regression was performed, in order to exclude the bias factors on OT. The CUSUM learning curve was modeled by curve fitting and R 2 was used to testify the goodness. The different phases of the learing curve was compared on OT, estimated blood loss （EBL）, lymph nodes dissecting, complications, drainage time and hospital stay. Results The OT decreased with the accumulation of cases. The CUSUM learning curve was best modeled as a cubic curve with the equation： CUSUM（n）=0.002×n 3-0.323×n 2＋13.360×n-18.195, which had a higher R 2 value of 0.847. The fitting curve reached the top at the 28 th case. As a cut- off point, the 28 th case di Vided the learning curve into two phases. There were statistical differences in OT （ P =0.022）, EBL （ P = 0.014） , numbers of lymph nodes dissected （ P =0.022）, complications （ P =0.015）, drainage time （ P =0.025） and hospital stay （ P = 0.005）. Conclusions The learning curve of da Vinci robot- assisted pulmonary lobectomy can be defined by CUSUM analysis, which suggested from our data that the surgeons need finish about 28 cases to master the technique.

关 键 词：机器人肺叶切除术 CUSUM分析 学习曲线 多元线性回归 

分 类 号：R655.3[医药卫生—外科学]