Background: Critical limb ischemia (CLI) represents the most severe form of peripheral artery disease (PAD) and has an immense impact on quality of life, morbidity and mortality. Interventions aim at improving tissue perfusion and averting amputation with an optimal risk-benefit ratio. In order to improve clinical decision-making several prediction models regarding post-intervention outcomes have been developed in the last few years. Validation of these models improves assessment of future treatment options for CLI patients.
Methods: This study aims to validate the BASIL, FINNVASC and PREVENT models in data sets from a registry study (Athero-Express) and two cohorts of patients treated for CLI in the Netherlands; the JUVENTAS and PADI. Receiver Operating Characteristics (ROC) curve analysis was used to calculate their predictive capacity. The outcome measures where all-cause mortality, amputation free survival (AFS) and amputation at 12 months after intervention.
Results: The BASIL shows high predictive values regarding mortality in the cohort with the highest cardiovascular risk burden (JUVENTAS) with an area under the ROC-curve (AUC) of 80.9%. Similar figures are observed for the PREVENT (77.7%). Prediction of AFS is rather poor for all models in each population, with the best predictive value of 70.9% for the BASIL in the JUVENTAS population. The FINNVASC performs best in the prediction of amputation in the intermediate (PADI) population with 78.1% at 12 months.
Conclusions: All models perform similarly to their initial cohort validation. In general all models perform poor to fair on predicting amputation and AFS and we consider the FINNVASC to be the least suitable for predicting mortality. A possible explanation for the better performance of the BASIL and PREVENT on predicting mortality is the fact that they do include kidney function/ dialysis and age in their model. Especially the BASIL survival prediction model can be valuable in patients with (C)LI and might be a useful addition to the decision making process in clinical practice.