Taylor Jordan, Arias-Espinosa Luis, McGeoch Catherine, Shah Vaishali, Shyu Ethan, Shahi Niti, Rodier Simon, Kaplan Brian, Malcher Flavio, Damani Tanuja
Division of General Surgery, New York University Langone Health, 530 1Th Ave, New York City, NY, 10016, USA.
Surg Endosc. 2025 Jun 27. doi: 10.1007/s00464-025-11886-z.
The National Surgical Quality Improvement Program (NSQIP) American College of Surgeons (ACS) risk calculator is a validated method of predicting postoperative complications that was recently updated to a machine-learning structure. The objective of this study was to measure the accuracy of this calculator in our institution on paraoesophageal hernia (PEH) repair.
Procedures performed between 2019 and 2023 were retrospectively collected regarding demographics, operative variables, and outcomes with a 30-day follow-up. Thirteen outcomes measured by NSQIP-ACS calculator were measured. Observed and predicted rates were compared by receiver operating curves (ROC) and length of stay was compared by Wilcoxon signed rank test.
A total of 203 paraoesophageal hernia repairs on patients with a median age of 68 (IQR 61-75) and 70.9% (n = 144) predominantly female. The size of the paraoesophageal hernia (PEH) was large or giant in 59.1% (n = 120) and mesh was placed in 70.4% (n = 143). The predicted risk was consistently higher than observed events on all but discharge destinations. Eight outcomes had no event to measure; however, the calculator accurately predicted a risk of ≤ 1% on all of these. The area under the curve (AUC) was fair (0.6-0.79) on discharge to nursing or rehabilitation facilities and failed in the rest of the measurable outcomes.
The ACS-NSQIP risk calculator correctly predicted a low occurrence of postoperative outcomes in patients undergoing robotic paraoesophageal hernia repair.
美国外科医师学会（ACS）国家外科质量改进计划（NSQIP）风险计算器是一种经过验证的预测术后并发症的方法，最近已更新为机器学习结构。本研究的目的是评估该计算器在我们机构中对食管旁疝（PEH）修复的预测准确性。
回顾性收集2019年至2023年间进行的手术的人口统计学、手术变量和30天随访结果。测量了NSQIP-ACS计算器所测量的13项结果。通过受试者工作曲线（ROC）比较观察到的和预测的发生率，并通过Wilcoxon符号秩检验比较住院时间。
共对203例食管旁疝进行了修复，患者中位年龄为68岁（四分位间距61-75岁），70.9%（n = 144）为女性。59.1%（n = 120）的食管旁疝（PEH）为大疝或巨大疝，70.4%（n = 143）放置了补片。除出院目的地外，预测风险始终高于观察到的事件。八项结果无事件可测量；然而，计算器准确预测所有这些结果的风险均≤1%。在出院到护理或康复机构方面，曲线下面积（AUC）为中等（0.6-0.79），在其余可测量结果中未达预期。
ACS-NSQIP风险计算器正确预测了接受机器人食管旁疝修复术患者术后低发生率的结果。
