Toward Optimizing Risk Adjustment in the Dutch Surgical Aneurysm Audit

Area of Science:

• Vascular Surgery
• Health Services Research
• Medical Informatics

Background:

• Accurate hospital outcome comparisons for aortic aneurysm surgery necessitate casemix correction.
• Mortality risk prediction models are key to identifying essential preoperative variables.

Purpose of the Study:

• To determine the optimal set of preoperative variables for mortality risk prediction.
• To establish a relevant and efficient casemix model for aortic aneurysm surgery.

Main Methods:

• Analysis of 12,401 patients from the Dutch Surgical Aneurysm Audit (2013-2016).
• Utilized multivariable logistic regression and compared with existing models (GAS, VBHOM, DAS).
• Assessed model performance using C-statistic (>0.8) and Hosmer-Lemeshow (H-L) P-value (>0.05).

Main Results:

• Overall mortality was 6.5%, varying by surgical indication (EAAA, SAAA, RAAA).
• The optimal mortality prediction model included gender, age, pulmonary comorbidity, operative setting, creatinine, aneurysm size, hemoglobin, Glasgow coma scale, ECG, and systolic blood pressure (C-statistic 0.871).
• External validation showed varying performance for existing models (VBHOM: 0.836, DAS: 0.782, GAS: 0.761).

Conclusions:

• The optimal casemix model for the Dutch Surgical Aneurysm Audit requires both clinical and laboratory parameters.
• These variables are readily available in electronic patient files.
• This approach facilitates an efficient and accurate casemix model for surgical outcome assessment.