Background and Aim: Mortality risk prediction models following primary PCI suffer from skewed and imbalanced data sets as also involve expensive lab parameters like BNP etc and specialised ECHO techniques and have not been tested in Indians. Our study aimed to establish a simple nomogram predicting in-hospital mortality (upto 14 days) in patients undergoing primary PCI that could be used at the bedside using basic ECG, ECHO and lab parameters that are easily available.

Materials and Methods: We studied 425 patients undergoing primary PCI at two apex centres from Jan 21 2015, Jan 1 2020 undergoing primary PCI. The risk model was generated by logistic regression with the stepwise backward method, while calculating Odds Ratios (OR) and 95% CI for in-hospital mortality. 

Results: Nine patient features were selected to build the nomogram and weightage was given as per attributable hazard to each variate including(weighted using SHAP analysis ): Post PCI TIMI flow grade< 3 ( WF: 3), along with AW- STEMI or LBBB( WF: 3.5), admission Vasotrope-Inotrope Index > 40( WF: 3.5), admission Shock Index > 1.3( WF: 3.5),Uncontrolled insulin-requiring DM( WF: 3.0) anytime in  hospital course, LVEF < 25%( WF: 3), ICU Serum Lactate > 120 anytime in  hospital course (WF: 3), MAPSE < 7mm by ECHO (WF: 2.5) anytime in  hospital course and Admission N/L ratio > 7(WF: 2.0). AUROC was 0.881 (95% CI: 0.63–0.956). On generating tertiles of the score of T1< 16, T2 between 17- 22.5 and T3 > 22.5 of the total 27 points, Low, Intermediate and Highest risk cohorts were identifie. In a forward Validation cohort the score had an excellent predictive accuracy of 85% agreement and outperformed both the PAMI risk score and CADILLAC scores using different variables without losing out on accuracy.

Conclusion: Our Novel 9- parameter risk score in Indians for bedside prediction of in-hospital mortality after Primary PCI uses simple and easy –to- evaluate parameters and will simplify risk stratification in real-world practice.