Evaluation of multidimensional geriatric assessment as a predictor of mortality and cardiovascular events after transcatheter aortic valve implantation.

Objectives: This study evaluated Multidimensional Geriatric Assessment (MGA) as predictor of mortality and major adverse cardiovascular and cerebral events (MACCE) after transcatheter aortic valve implantation (TAVI).

Background: Currently used global risk scores do not reliably estimate mortality and MACCE in these patients.

Methods: This prospective cohort comprised 100 consecutive patients ≥ 70 years undergoing TAVI. Global risk scores (Society of Thoracic Surgeons [STS] score, EuroSCORE) and MGA-based scores (cognition, nutrition, mobility, activities of daily living [ADL], and frailty index) were evaluated as predictors of all-cause mortality and MACCE 30 days and 1 year after TAVI in regression models.

Results: In univariable analyses, all predictors were significantly associated with mortality and MACCE at 30 days and 1 year, except for the EuroSCORE at 30 days and instrumental ADL at 30 days and 1 year. Associations of cognitive impairment (odds ratio [OR]: 2.98, 95% confidence interval [CI]: 1.07 to 8.31), malnutrition (OR: 6.72, 95% CI: 2.04 to 22.17), mobility impairment (OR: 6.65, 95% CI: 2.15 to 20.52), limitations in basic ADL (OR: 3.63, 95% CI: 1.29 to 10.23), and frailty index (OR: 3.68, 95% CI: 1.21 to 11.19) with 1-year mortality were similar compared with STS score (OR: 5.47, 95% CI: 1.48 to 20.22) and EuroSCORE (OR: 4.02, 95% CI: 0.86 to 18.70). Similar results were found for 30-day mortality and MACCE. Bivariable analyses, including STS score or EuroSCORE suggested independent associations of MGA-based scores (e.g., OR of frailty index: 3.29, 95% CI: 1.06 to 10.15, for 1-year mortality in a model including EuroSCORE).

Conclusions: This study provides evidence that risk prediction can be improved by adding MGA-based information to global risk scores. Larger studies are needed for the development and validation of improved risk prediction models.