Creation and validation of a biochemical scoring system and smartphone application to predict 30 day mortality in patients requiring left-sided Impella monotherapy

Abstract

Abstract Funding Acknowledgements Type of funding sources: None. Objectives To create and validate a simple biochemical scoring system and smartphone application that can estimate 30-day mortality in patients requiring left-sided Impella implantation as standalone mechanical circulatory support (MCS). Methods Initially, we retrospectively analysed 79 consecutive patients who required left-sided Impella MCS monotherapy. Regression analysis was used to elucidate significant associations between biochemical markers before Impella implantation and all-cause mortality at 30 days. Using these factors, a simple additive scoring system was create. The scoring system was then validated using prospectively collected data from our sister institution. This validated scoring system was then used to create a smartphone application that was formally evaluated by cardiologists. Results The BALLAR scoring system was created. Patients are assigned points based upon biochemical markers. These are summed and the final points tally provides an estimate of 30-day mortality. The points are assigned as follows: Lactate (mmol/litre): <2: 0 points, 2 – 4.9: 1 Point, >4.9: 4 Points Creatinine Clearance (ml/min): <30: 6 points, 30 - 59.9: 4 points, 60 - 89.9: 1 point, >89.9: 0 points Serum Albumin (mmol/litre): <25: 6 points, 25 - 34.9: 3 points, >34.9: 0 points Base Excess (mmol/L): < -2: 2 points, >= -2: 0 points The total score can be used to estimate the probability of death at 30 days. A score less than 6 predicts a 30-day mortality of under 5%, whereas a score over 11 predicts a greater than 95% chance of death within 30 days. This scoring system was initially validated using prospectively collected data from 10 patients our sister institution, where it predicted 30-day mortality in all of these patients. A smartphone application was then created which received universally positive feedback for ease of use by cardiologists and trainees at our institution. Conclusion The simple heuristic predicted 89% of 30-day deaths in our cohort. All the misclassifications were in the intermediate probability range (scores 5-11). Applied to a small external validation cohort, the predictive accuracy was 100%. This simple scoring system gives an effective estimate of the probability of death at 30 days in our cohort of patients and is now the basis for a novel smartphone application.