【本期导读】

随着接受手术治疗的老年患者和并发症患者人数日益增多，术前鉴别患者肝切除术后发病率和死亡率变得越来越重要。本文旨在开发和验证虚弱指数（rFI），用以预测肝脏手术后的不良预后结果。rFI指数主要包括五个术前变量[美国麻醉学会评分（ASA）、体重指数（BMI）、血清白蛋白、血细胞比容、潜在病理学和肝切除术类型]。研究结果是，rFI具有良好辨别能力和风险分级能力，检测功能优于mFI等早期虚弱指数。

Frailty as a Risk Predictor of Morbidity and Mortality Following Liver Surgery

肝外科手术后发病率和死亡率的风险预测因子：虚弱指数

Abstract

摘要

Background

Given the increasing number of elderly and comorbid patients undergoing surgery, there is increased interest in preoperatively identifying patients at high risk of morbidity and mortality following liver resection. We sought to develop and validate the use of a frailty index (FI) to predict poor postoperative outcomes following liver surgery.

背景

随着接受手术治疗的老年患者和并发症患者人数日益增多，术前鉴别患者肝切除术后发病率和死亡率变得越来越重要。本文旨在开发和验证虚弱指数（FI），用以预测肝脏手术后的不良预后结果。

Methods

Patients undergoing a liver resection were identified using the National Surgical Quality Improvement Program Hepatectomy-targeted database for 2014 and randomized into a training or validation cohort. Multivariable logistic regression analysis was performed to develop a revised frailty index (rFI) to predict adverse postoperative clinical outcomes. Leave one out cross-validation was performed to validate the proposed rFI.

方法

全部患者按照2014年国家外科质量改进计划肝切除术数据库的鉴定结果接受肝脏切除手术，并随机纳入训练队列或验证队列。然后通过多变量逻辑回归分析，制定修订版虚弱指数（rFI），预测术后不良临床结果。最后进行交叉验证，验证rFI结果。

Results

A total of 2714 patients were identified who met the inclusion criteria. Postoperatively, 826 patients (30.4%) developed a postoperative complication, while 39 patients died within 30 days of surgery. Five preoperative variables (ASA class, BMI, serum albumin, hematocrit, underlying pathology, and type of liver resection) were used to develop the rFI. The rFI demonstrated good discrimination (AUROC = 0.68) and outperformed the previously proposed modified frailty index (mFI; AUROC = 0.53, p < 0.001) when evaluated among patients included in the training cohort. On validation, the rFI demonstrated good model discrimination (AUROC = 0.68) and was accurately able to risk-stratify patients within the validation cohort at risk for developing a postoperative complication, prolonged length-of-stay, and postoperative mortality (all p < 0.05).

结果

本研究共纳入了2714例符合入排标准的患者。术后826例（30.4％）出现并发症，其中39例患者术后30d内死亡。rFI指数主要包括五个术前变量[美国麻醉学会评分（ASA）、体重指数（BMI）、血清白蛋白、血细胞比容、潜在病理学和肝切除术类型]。训练队列患者的评估结果是，rFI具有良好辨别能力（AUROC = 0.68），结果优于早前提出的改良版虚弱指数（mFI; AUROC = 0.53，p <0.001）。验证队列患者的评估结果是，rFI是具有良好辨别能力的模型（AUROC = 0.68），同时可以准确地可以对验证队列的患者进行风险分级，包括具有术后并发症、住院时间延长和术后死亡风险的患者（P均 <0.05 ）。

Conclusion

Frailty, as measured by the rFI, was predictive of increased risk of morbidity and mortality following liver surgery and can be used to guide patient decision-making.

结论

rFI测量结果显示，虚弱指数可以预测肝脏手术后增加的发病率和死亡风险，对于患者治疗方案的制定具有指导意义。

Keywords

Hepatic surgery  Liver Frailty  Mortality  Morbidity  Prediction

关键词

肝脏手术 肝脏虚弱指数 死亡率 发病率 预测

This study was previously presented as a poster presentation at the 57th Annual Meeting of the Society for Surgery of the Alimentary Tract held on May 21-24, 2016 in San Diego, CA and as a plenary presentation at the National Surgical Quality Improvement Program Conference held on July 19, 2016 in San Diego, CA.

本研究曾经为2016年5月21日至24日在加利福尼亚州圣地亚哥举行的消化道手术协会第57届年会提供海报展示，并纳入于2016年7月19日在加利福利亚州圣地亚哥举行的国家外科质量改进计划大会的全体会议日程。

The American College of Surgeons National Surgical Quality Improvement Program and the hospitals participating in the ACS NSQIP are the source of the data used herein; they have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors

本研究使用的数据来自美国外科医师学院国家外科质量改进计划（ACS NSQIP）和参与ACS NSQIP各大医院的数据库；ACS NSQIP及参与ACS NSQIP的各大医院对于本研究数据分析的有效性和数据结果不承担任何责任。

Electronic supplementary material

The online version of this article (doi:10.1007/s11605-017-3373-6) contains supplementary material, which is available to authorized users.

电子辅助材料

本研究的在线版本（doi：10.1007 / s11605-017-3373-6）包含电子辅助材料，可供授权用户使用。

Introduction

Given the recent improvements in surgical technique, patient selection, and perioperative care pathways, an increasing number of patients are considered candidates for liver resection.1–4 Despite an increasing number of liver resections performed each year, the morbidity associated with liver surgery remains high with an estimated 25–40% of patients developing some type of postoperative complication following surgery.5,6 Furthermore, given that an increasing proportion of elderly patients, as well as patients with underlying comorbidities will undergo surgery, the preoperative identification and subsequent risk stratification of patients at risk for developing adverse postoperative outcomes has taken on increased importance.7,8

简介

近年来，外科手术、患者选择和围手术期护理等技术不断进步，能够接受肝切除术的患者人数也不断增加。1-4 虽然每年接受肝切除术的患者人数有所上升，但是与肝脏手术相关的发病率一直居高不下，大约有25%—40％的患者手术后会产生一系列各种类型的术后并发症。5,6此外，由于接受肝脏手术的老年患者以及潜在合并症患者人数越来越多，术前鉴别以及对于可能出现不良术后结果的患者进行风险分级变得越来越重要。7,8

Frailty, a syndrome characterized by a decreased physiological reserve, has been identified as an important metric to measure and assess preoperative risk.7,9–11 Traditionally, frailty has been measured by combining a patient’s medical history, physical examination, and the assessment of physical and functional status. These proposed composite measures are, however, time consuming, subjective, and often not applicable to nationally representative research databases.12–15 Given the need to study population-based outcomes, Velanovich and colleagues mapped the 70 variables included within the frailty index (FI) proposed by the Canadian Study of Health and Aging (CSHA) onto the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database to develop a modified frailty index (mFI) consisting of 11 preoperative variables measuring patient frailty.16 While several previous studies have demonstrated an association between the mFI and adverse postoperative clinical outcomes including postoperative complications, increasing length-of-stay (LOS), and postoperative mortality, data used to calculate the mFI are not routinely collected in recent iterations/versions of the ACS-NSQIP databases.17–20

虚弱是人体老化的特征之一，也是衡量和评估术前风险的重要指标之一。7,9-11 一般来说，虚弱指数的测评包括患者病史、身体检查、身体和功能状态等评估数据。但是这种早期的综合测评方法即耗时又耗力，通常不适于成为全国代表性研究数据。12-15 由于虚弱指数是以人群为基础进行的研究分析，Velanovich及其同事将70个变量（由加拿大健康与老龄研究项目[CSHA]提交给美国外科医师学院国家外科质量改进计划[ACS-NSQIP]数据库）纳入虚弱指数研究，开发出由11个测量患者术前虚弱变量组成的升级版虚弱指数（mFI）。16早期的多项研究已经证明，mFI与术后不良临床结果之间存在关联性，其中包括术后并发症、医院停留时间（LOS）和术后死亡率，但是本研究用于计算mFI的数据并不是在最近的ACS-NSQIP迭代/版本中常规收集的数据。

Specifically, beginning in 2011, the reporting of five variables included within the mFI was made optional by the ACS-NSQIP, with data following 2012 not reporting these data altogether.21 As such, there is a need to develop a robust and clinically applicable preoperative frailty model that incorporates procedure-specific information to risk stratify patients.13,22–26 Given this, the objective of the current study was to develop and validate a parsimonious, clinically relevant frailty index using a nationally representative dataset of patient undergoing a liver resection. Additionally, we sought to compare the predictive power of the proposed revised frailty index (rFI) to existing frailty indices including the mFI.

也就是说，本研究中mFI纳入报告的五个变量选自ACS-NSQIP2011——2012年的研究数据，2012年以后的数据不包括在报告中。21因此，研究开发的术前虚弱模型必须拥有强大功能，能够应用于临床，同时可以将程序特异性信息用于患者风险分级。13,22-26 总之，本研究旨在使用接受肝切除术患者的国家代表性数据集，设计开发出一个简便的可供临床使用的虚弱指数。此外，还试图对比分析修订版虚弱指数（rFI）与现有虚弱指数（包括mFI）的临床检测预估能力。

Methods

方法

Study Population and Data Sources

The current analysis was performed using data from the ACS-NSQIP Hepatectomy Targeted Participant Use Data File (PUF) for 2014. The ACS-NSQIP PUF represents a quality improvement initiative by the ACS that collects data from over 500 institutions in patients undergoing surgery in North America. All data are prospectively reviewed by trained surgical reviewers/nurses and undergo a systematic audit to ensure inter-reviewer reliability and quality assurance. For each patient record, the ACS-NSQIP Hepatectomy Targeted PUF includes data pertaining to baseline demographics including patient age, sex, and race, as well as detailed clinical variables describing preoperative comorbidity, the primary indication for surgery, intra- and postoperative clinical details, 30-day postoperative morbidity, and 30-day postoperative mortality. All patient records are de-identified and compliant with the Health Insurance Portability and Accountability Act (HIPAA) of 1996. This study was approved by the Johns Hopkins University Institutional Review Board.

研究对象和数据来源

本研究的分析数据来自2014年ACS-NSQIP肝切除术参与者使用数据手册（PUF）。ACS-NSQIP PUF是ACS的一项质量改进计划，收集了美国北部500多家机构中接受手术的患者数据。所有数据由经过培训的外科评估员/护士进行前瞻性审查，并进行系统性审核，确保了审核数据的可靠性和质量。ACS-NSQIP肝切除术PUF的每一份患者记录均是与基线人口统计学有关的数据，其中包括患者年龄、性别和种族，以及描述术前并发症的详细临床变量、手术的主要指征、术中和术后临床的详细信息、 30d术后发病率、30d术后死亡率等等。所有患者记录均取消标识，并符合1996年“健康保险便携性和责任法案”（HIPAA）。本研究由约翰霍普金斯大学机构审查委员会批准。

Patients undergoing an elective major hepatectomy (Current Procedural Terminology [CPT] codes “47122,” “47125,” and “47130”) or an elective partial lobectomy (CPT code “47120”) between January 01, 2014 and December 31, 2014 were identified and included within the study population (Supplemental Figure 1). Patients who underwent surgery on an emergent basis or patients who were transferred in from another medical center were excluded from the final analysis. Using a previously described methodology, patients included within the final study population were randomized to either a training (70% of all patients) or validation cohort (30% of all patients).27

本研究纳入并评估了2014年1月1日至2014年12月31日期间接受选择性肝大部分切除术（现行程序术语[CPT]代码“47122”、“47125”和“47130”）或选择性部分肺叶切除术（CPT代码“47120”）的患者（补充图1）。本研究分析最终排除在紧急情况下进行手术的患者或从其他医疗中心转入的患者。本研究按照早期描述的方法，将全部入选患者随机分配到训练（患者总人数的70％）队列或验证队列（患者总人数的30％）。27

Preoperative patient body mass index (BMI) was classified according the World Health Organization (WHO) classification for BMI, while preoperative performance status was assessed using the American Society of Anesthesiologists (ASA) physical status classification score.28,29 Similarly, as previously defined, standardized values were used to define preoperative serum albumin (decreased serum albumin: albumin <3.5 g/L) and preoperative hematocrit (elevated hematocrit: ≥50.0% for men and ≥45.0% for women). An mFI, as proposed by Velanovich et al., was calculated for each patient using all variables with available data.16 For ease of analysis, patients were categorized into three groups based on their mFI score: mFI = 0, mFI = 1, and mFI ≥ 2 (out of a possible maximum score of 11). The primary outcome of interest was the development of either a postoperative complication or postoperative mortality within 30 days of surgery. Postoperative morbidity was defined using a composite measure for postoperative complications and included surgical site infections, pneumonia, need for intubation, ventilator dependence, venous thromboembolism (pulmonary embolism or deep venous thromboembolism), acute renal failure, urinary tract infections, myocardial infarction, bleeding, sepsis, bile leakage, liver failure, and postoperative coma.

术前患者体重指数（BMI）按照世界卫生组织（WHO）的BMI分类标准进行分类，而术前活动状态（PS）则采用美国麻醉医师协会（ASA）的身体状况分级标准。28,29同理，按照如前所述使用标准化值，定义术前血清白蛋白（血清白蛋白降低：白蛋白<3.5 g / L）和手术前血细胞比容（血细胞比容升高，男性≥50.0％，女性≥45.0％）。按照Velanovich等提出的mFI指数，使用具有可用数据的所有变量计算每个患者的mFI指数。16为了便于数据分析，根据mFI评分将患者分为三组：mFI = 0、mFI = 1和mFI≥2（最大可能得分为11）。感兴趣的主要结果是手术后30d内出现术后并发症或术后死亡。术后发病率定义为一种术后并发症的综合指标，包括手术部位感染、肺炎、插管需求、呼吸机依赖、静脉血栓栓塞（肺栓塞或深静脉血栓栓塞）、急性肾功能衰竭、尿路感染、心肌梗塞、出血、败血症、胆漏、肝功能衰竭、术后昏迷等等。

Statistical Analysis

Continuous data were reported as medians with interquartile range (IQR), while categorical data were reported as whole numbers and percentages. The Kruskal-Wallis test was used to compare continuous data, while categorical data were compared using Pearson’s chi-squared test. To generate our prediction model, univariable logistic regression analyses were performed to assess the association between preoperative characteristics and the development of either a postoperative complication or postoperative mortality among patients included within the training cohort, while receiver operative characteristic (ROC) curve analyses were performed to test the discriminatory (predictive power) ability of preoperative characteristics. Multivariable logistic regression was then performed to develop a parsimonious preoperative risk score including all preoperative variables demonstrating appropriate discrimination on univariable analysis.

统计分析

连续数据报告为具有四分位距（IQR）的中位数，而分类数据报告为整数和百分比。 Kruskal-Wallis检验用于比较连续数据，而Pearson卡方检验用于比较分类数据。本研究进行单因素Logistic回归分析，以评估术前特征与训练队列中患者术后并发症或术后死亡率发展的关系，从而产生预测模型。同时，进行接受者操作特性（ROC）曲线分析，测试术前特征的辨别能力（预测能力）。然后，进行多变量逻辑回归分析，开发简便的术前风险评分，包括所有术前变量，与单变量分析相比有一定的区别。

Final model selection was based on area under the receiver operative characteristic curve (AUROC) statistics, Akaike Information Criterion (AIC), and variance inflation factor analysis for colinearity. Results from the final multivariable logistic regression model were used to develop a clinically relevant nomogram to identify patients at risk for postoperative complications or postoperative mortality following a liver resection. Validation of the proposed nomogram was performed using a leave one out cross-validation methodology.30 Calibration of the proposed nomogram was evaluated among patients included within the validation cohort by comparing predicted probabilities to observed events. Statistical significance was defined by a p value of <0.05. All statistical analyses were performed using STATA version 14.0 (StataCorp, College Station, TX).

最终模型选择以接收操作特征曲线下的面积（AUROC）、Akaike信息标准（AIC）和共线性方差膨胀因子分析为准。最终多变量逻辑回归模型得出的结果用于开发临床相关的标准图，以确定患者肝切除术后出现术后并发症或术后死亡的风险。本研究采用留一法交叉验证方法，进行诺模图的验证。30通过将预测概率与观察到的事件进行对比分析，对验证队列内的患者进行诺模图分级标准的评估。p值<0.05定义为具有统计学意义。所有统计分析采用STATA版本14.0（StataCorp，College Station，TX）。