Background: Prediction of hospitalisation costs is crucial for an effective patient journey during TAVI procedures. Decision tree algorithms (DTA) are robust instruments that provide accurate prediction models. We investigated whether a DTA could be used to predict hospitalization costs following TAVI.  

Methods: 284 TAVI patients (147 male; median age = 81 (78-84) years) at our department were prospectively included. The primary endpoint was highest quartile hospitalization costs following TAVI. EuroScore II (ESII = 3.75 (2.30-6.05)) and patient-reported outcome measures (PROMs), namely Clinical Frailty Scale (CFS= 4 (2-5)), EQ-5D-5L (EQ-5D VAS = 53 (38-83)) and Kansas City Cardiomyopathy Questionnaire (KCCQ=39 (29-49) on admission and TAVI complications within 72 hours after implantation were analysed following a dedicated DTA.

 63 (22.18%) patients developed any TAVI-related complications within 72 hours. The median EQ-5D-5L VAS score was 53 (IQR: 38-83). DTA analysis revealed incidental post-TAVI complications (figure 1) as well as baseline EQ-5D-5L VAS score for prediction of highest quartile hospitalization costs (≥ EUR 5.627). This was confirmed by DTA, where patients with incidental acute post-TAVI complications and likewise patients without any acute complications but with a EQ-5D-5L VAS score <= 41,67 could be identified to be the main risk factor for highest hospitalization costs (figure 2). CFS and KCCQ values had not significant predictive value.



(figure 1) 




(figure 2)



Conclusion: Using just two metrics and our novel DTA–analysis, based on a prospectively gathered significant amount of patient data, both clinicians and clinical managers can be provided with information regarding significantly increased hospitalisation costs after TAVI.