Department of General Surgery, University Hospital of Wales, Cardiff & Vale NHS Trust, Cardiff, United Kingdom; Department of General Surgery, Royal Glamorgan Hospital, Cwm Taf University Health Board, Pontyclun, United Kingdom.
Hepatobiliary and Pancreatic Surgery and Liver Transplant Unit, Queen Elizabeth Hospital, Birmingham, United Kingdom.
Int J Surg. 2022 Jun;102:106645. doi: 10.1016/j.ijsu.2022.106645. Epub 2022 May 6.
To compare performance of the Hajibandeh Index (HI) and National Emergency Laparotomy Audit (NELA) score in predicting postoperative mortality in patients undergoing emergency laparotomy.
In compliance with STROCSS guidelines for observational studies a cohort study was conducted. All patients aged over 18 who underwent emergency laparotomy between January 2014 and January 2021 in our centre were considered eligible for inclusion. The HI and NELA indices in predicting 30-day and 90-day postoperative mortality were compared. The discrimination of each test was evaluated using Receiver Operating Characteristic (ROC) curve analysis, classification using the classification table and calibration using a plotted diagram of the expected versus observed mortality rates.
Analysis of 700 patients showed that the predictive performance of the HI and NELA models were comparable (30-day mortality: AUC: 0.86 vs 0.87, P = 0.557; 90-day mortality: AUC: 0.81 vs 0.84, P = 0.0607). In terms of 30-day mortality, HI was significantly better than the NELA model in predicting postoperative mortality in patients aged over 80 (AUC: 0.85 vs 0.72, P = 0.0174); however, the performances of both tools were comparable in patients with ASA status above 3 (AUC: 0.82 vs 0.82, P = 0.9775), patients with intraperitoneal contamination (AUC: 0.77 vs 0.85, P = 0.0728) and patients who needed a bowel resection (AUC: 0.85 vs 0.88, P = 0.2749). In terms of 90-day mortality, HI was significantly better than the NELA model in predicting mortality in patients aged over 80 (AUC: 0.82 vs 0.71, P = 0.0214); however, NELA had better predictive value in patients with intraperitoneal contamination (AUC: 0.76 vs 0.85, P = 0.0268); the performances of both tools were comparable in patients with ASA status above 3 (AUC: 0.77 vs 0.80, P = 0.2582), and patients who needed a bowel resection (AUC: 0.81 vs 0.86, P = 0.05). Both tools were comparable in terms of classification and calibration.
Hajibandeh index was better than the NELA score in predicting postoperative 30-day and 90-day mortality in patients aged over 80 undergoing emergency laparotomy. Its performance in predicting 30-day and 90-day mortality was comparable with NELA score in other subgroups except 90-day mortality in patients with intraperitoneal contamination where the performance of NELA was better. We encourage other researchers to validate HI in predicting mortality following emergency laparotomy.
比较 Hajibandeh 指数(HI)和国家紧急剖腹术审计(NELA)评分在预测接受紧急剖腹术患者术后死亡率方面的性能。
根据 STROCSS 指南进行观察性研究,本研究进行了一项队列研究。我们中心所有年龄在 18 岁以上接受紧急剖腹术的患者均符合纳入标准。比较 HI 和 NELA 指数在预测 30 天和 90 天术后死亡率方面的表现。使用受试者工作特征(ROC)曲线分析评估每个测试的判别能力,使用分类表进行分类,并使用预期与观察死亡率的绘图图表进行校准。
对 700 名患者的分析表明,HI 和 NELA 模型的预测性能相当(30 天死亡率:AUC:0.86 对 0.87,P=0.557;90 天死亡率:AUC:0.81 对 0.84,P=0.0607)。在预测 80 岁以上患者术后死亡率方面,HI 在预测 30 天死亡率方面明显优于 NELA 模型(AUC:0.85 对 0.72,P=0.0174);然而,在 ASA 状态超过 3 分的患者(AUC:0.82 对 0.82,P=0.9775)、腹腔污染患者(AUC:0.77 对 0.85,P=0.0728)和需要肠切除的患者(AUC:0.85 对 0.88,P=0.2749)中,两种工具的性能相当。在预测 90 天死亡率方面,HI 在预测 80 岁以上患者死亡率方面明显优于 NELA 模型(AUC:0.82 对 0.71,P=0.0214);然而,在腹腔污染患者中,NELA 具有更好的预测价值(AUC:0.76 对 0.85,P=0.0268);在 ASA 状态超过 3 分的患者(AUC:0.77 对 0.80,P=0.2582)和需要肠切除的患者(AUC:0.81 对 0.86,P=0.05)中,两种工具的性能相当。在分类和校准方面,两种工具的性能相当。
在接受紧急剖腹术的 80 岁以上患者中,Hajibandeh 指数在预测术后 30 天和 90 天死亡率方面优于 NELA 评分。除了腹腔污染患者的 90 天死亡率外,其在预测 30 天和 90 天死亡率方面的表现与 NELA 评分相当,而 NELA 评分在预测腹腔污染患者的死亡率方面表现更好。我们鼓励其他研究人员验证 HI 在预测紧急剖腹术后死亡率方面的应用。