A validated predictive model for mid- and long-term mortality risk assessment after elective endovascular repair in abdominal aortic aneurysm patients.

Background: Risk scoring systems for open surgical repair of abdominal aortic aneurysm (AAA) may overestimate mortality after endovascular aneurysm repair (EVAR). A model for mid-term and long-term mortality after EVAR is still lacking.

Material and method: MEDLINE, Embase and WOS were searched from January 1, 2000 to December 31, 2022. Hazard ratios and 95% confidence intervals (CI) for mortality-related risk factors were extracted and synthesized in a meta-analysis. The C-statistics, IDI, NRI and DCA were used to assess the stability. A predictive model incorporating independent meta-analytic variables was developed, validated in a clinical cohort and compared with the Giles model.

Results: 35 studies containing 49272 patients were analyzed. A prediction model was established, including age, gender, aneurysm diameter, American Society of Anesthetists score, chronic obstructive pulmonary disease, cardiac disease, renal disease, cerebrovascular disease, diabetes, peripheral vascular disease, statins, aspirin, and smoker. The model had a C-statistic of 0.738 (95%CI:0.697, 0.779) in validation cohort, comprising 537 patients after EVAR. The sensitivities were 0.765, 0.796 and 0.756, and the specificities were 0.744, 0.652 and 0.668 at 1/3/5 years. In contrast, Giles model had a C-statistic of 0.657 (95%CI:0.608, 0.706). Integrated discrimination improvement (0.03, p < 0.001; 0.045, p = 0.01; 0.062, p < 0.001) and net reclassification index (0.342, p < 0.001; 0.306, p < 0.001; 0.356, p < 0.001) indicated improved predictive performance by the new model over Giles model.

Conclusion: This meta-analysis-derived AAA long-term mortality prediction model employs precision risk stratification to enhance clinical decision-making and implement personalized follow-up protocols, thereby delivering evidence-based support for clinical practice.