Validation of the BALLAR score for predicting 30-day mortality in patients requiring Impella monotherapy

Abstract

Abstract Background The BALLAR score was published in 2021 with the aim of allowing clinicians to calculate 30-day mortality quickly and easily for patients requiring left-sided Impella monotherapy. The points are assigned as follows: Base Excess (mmol/L): < -2: 2 points, ≥ -2: 0 points ALbumin (mmol/litre): <25: 6 points, 25 - 34.9: 3 points, >34.9: 0 points LActate (mmol/litre): <2: 0 points, 2 – 4.9: 1 Point, >4.9: 4 Points Renal function - Creatinine Clearance (ml/min): <30: 6 points, 30 - 59.9: 4 points, 60 - 89.9: 1 point, >89.9: 0 points The scoring system performed well on its initial training cohort but had not previously been validated. Purpose To validate the BALLAR score using a predominantly external validation cohort requiring left-sided Impella therapy only, for any indication. Methods The data required to calculate the BALLAR score was retrospectively acquired for 89 patients. 39 patients were from our own UK tertiary centre, 7 from our sister hospital, and 43 from centres across the United States from the Impella Quality (IQ) assurance program. Results For the entire validation cohort of 89 patients, the BALLAR score showed both excellent discrimination and calibration statistics. The area under the receiver-operator characteristic (ROC) curve was 0.91 (0.84–0.98) (Figure 1), the Brier score was 0.11 versus a Naïve classifier of 0.23, the calibration slope was 0.92 calibration with Spiegelhalter’s Z-statistic of 0.98 (p=0.323, Figure 2). For the external validation cohort of patients, the discrimination remained excellent with the area under the ROC curve of 0.88 (0.79-0.98). The Brier score remained better than a naïve classifier at 0.15, but the calibration slope was unsurprisingly worse at 0.73 with a Spiegelhalter’s Z statistic of 2.07 (p=0.04). The negative predictive value of a BALLAR score ≤5 was 94%, and the positive predictive value was 100% of a BALLAR score > 12, highlighting the rapid escalation of mortality risk in patients in the mid-zone of the shock spectrum. Conclusions The BALLAR score maintains high discriminatory ability when applied to a validation cohort of patients requiring left-sided Impella monotherapy for any indication. Further prospective work is underway, including app development, to assess the clinical utility of the BALLAR score as a means of categorising patient risk.Image 1Image 2