Abstract Body (Do not enter title and authors here): Background: Cardiogenic shock (CS) is a lethal condition and mechanical circulatory support is indicated if refractory to drug treatment. A recent report showed using Impella with standard care led to a lower risk of death in patients with acute myocardial infarction and CS. This finding may increase the clinical use of Impella, but prognostic models for patients after Impella are lacking.
Aims: To identify factors that predict in-hospital mortality in CS requiring Impella and to develop a new prognostic model.
Methods: We used the dataset of the Japanese registry for Percutaneous Ventricular Assist Device (J-PVAD), which included all the cases treated with Impella in Japan. Two-thirds of the patients treated with Impella due to CS were randomly assigned to a derivation cohort, and the other third was reserved as a validation cohort. Using the derivation cohort, a backward stepwise logistic regression model was built to identify the factor associated with in-hospital mortality.
Results: Of the total 1,715 patients in the derivation cohort, 989 patients were discharged alive and 726 patients (42.3%) died during hospitalization. From 28 variables available in the J-PVAD registry, 12 variables were independently associated with in-hospital mortality and applied for a component of the risk model; age, sex, body mass index, etiology of fulminant myocarditis, cardiac arrest in hospital, extracorporeal membrane oxygenation use, mean arterial pressure, lactate level, lactate dehydrogenase level, total bilirubin level, creatinine level, and albumin level (Table 1). The J-PVAD risk score was created by assigning scores to these components based on the coefficients (Table 1). An Example of predicted in-hospital mortality according to the J-PVAD risk score is shown in Table 2. The area under the receiver operating characteristics curve (AUC) of the J-PVAD risk score was 0.76 (95% confidence interval [CI] 0.73–0.78). The comparison of predicted and observed in-hospital mortality according to the 7th quantiles by the J-PVAD risk score showed good calibration (Figure A). In the validation cohort (n=726), the J-PVAD risk score showed good discrimination ability (AUC 0.77 [95%CI 0.74–0.81]) and calibration (Figure B).
Conclusions: The J-PVAD risk score can be calculated using variables easily obtained in routine clinical practice. It helps the accurate stratification of the risk for mortality and facilitates clinical decision-making.