Kong Cherng Huei, Guest Glenn D, Stupart Douglas A, Faragher Ian G, Chan Steven T F, Watters David A
Department of Surgery, Barwon Health, Geelong, Victoria, Australia.
North West Academic Centre, The University of Melbourne, Melbourne, Victoria, Australia.
ANZ J Surg. 2015 Jun;85(6):403-7. doi: 10.1111/ans.13066. Epub 2015 Mar 30.
Colorectal surgery carries a significant mortality risk, with reported rates of 1-6% for elective surgery and up to 22% in the emergency setting. Both clinicians and patients will benefit from being able to predict the likelihood of death before surgery. Recently, we have described and validated two risk stratification models for colorectal surgery, the Barwon Health 2012 and Association Française de Chirurgie models. However, these models are not suitable for assessment at patient's bedside. The purpose of this study is to develop a simplified preoperative model capable of predicting mortality following colorectal surgery.
The new model is termed Colorectal preOperative Surgical Score (CrOSS). The development and internal validation of CrOSS was performed using a prospectively maintained colorectal database. External validation was performed using retrospective data. Univariate and multivariate analyses were performed in model development. Calibration and discrimination were used for model validation.
There were 474 and 389 consecutive colorectal surgeries at Geelong Hospital and Western Hospital. Overall mortality rates were 5.16% and 1.03%, respectively. Significant predictors for mortality were as follows: age ≥70, urgent operation, albumin ≤30 g/L and congestive heart failure (receiver operating characteristic (ROC) = 0.870, calibration P-value = 0.937). The predicted risk of mortality was stratified according to the risk profile of 0.39-66.51%. When validated externally, CrOSS predicted mortality accurately (ROC = 0.847, calibration P-value = 0.199).
A robust and simple preoperative model has been created to risk-stratify patients for colorectal surgery. This was successfully validated at another tertiary hospital.
结直肠手术具有显著的死亡风险，择期手术的报告死亡率为1%-6%，急诊情况下高达22%。临床医生和患者都将受益于术前能够预测死亡可能性。最近，我们描述并验证了两种结直肠手术风险分层模型，即Barwon Health 2012模型和法国外科协会模型。然而，这些模型不适合在患者床边进行评估。本研究的目的是开发一种能够预测结直肠手术后死亡率的简化术前模型。
新模型称为结直肠术前手术评分（CrOSS）。CrOSS的开发和内部验证使用了前瞻性维护的结直肠数据库。外部验证使用回顾性数据。在模型开发中进行单变量和多变量分析。使用校准和辨别力进行模型验证。
吉朗医院和西部医院分别有474例和389例连续的结直肠手术。总体死亡率分别为5.16%和1.03%。死亡率的显著预测因素如下：年龄≥70岁、急诊手术、白蛋白≤30 g/L和充血性心力衰竭（受试者操作特征曲线（ROC）=0.870，校准P值=0.937）。预测的死亡风险根据0.39%-66.51%的风险概况进行分层。在外部验证时，CrOSS准确预测了死亡率（ROC=0.847，校准P值=0.199）。
已创建一个强大且简单的术前模型，用于对结直肠手术患者进行风险分层。该模型在另一家三级医院成功得到验证。
