Performance of ABCD-10 and SCORTEN mortality prediction models in a cohort of patients with Stevens-Johnson syndrome/toxic epidermal necrolysis

Background: Age, bicarbonate, cancer, dialysis, 10% body surface area risk model (ABCD-10) has recently been proposed as an alternative to the SCORe of toxic epidermal necrolysis (SCORTEN) model for predicting in-hospital mortality in patients with Stevens-Johnson syndrome and toxic epidermal necrolysis (SJS/TEN). In contrast to SCORTEN, ABCD-10 incorporates prior dialysis and upweights the impact of cancer. Objective: To determine the performance of ABCD-10 compared with that of SCORTEN in mortality prediction at a large, tertiary burn center. Methods: A retrospective analysis of 192 patients with SJS/TEN admitted to the North Carolina Jaycee Burn Center from January 1, 2009, to December 31, 2019, was conducted. Data on these patients were collected using the burn registry and a manual chart review. The performance of both the mortality prediction models was assessed using univariate logistic regression and the Hosmer-Lemeshow test. Results: The overall mortality was 22% (n = 43). Nine (5%) patients had cancer, and 7 (4%) had undergone prior dialysis; neither factor was associated with mortality (P = .11 and P = .62, respectively). SCORTEN was well calibrated to predict inpatient mortality (P = .82), whereas ABCD-10 appeared to have a poorer fit (P < .001) in these patients. Both the models showed good discrimination. Limitations: Small sample size. Conclusion: SCORTEN was a better predictor of inpatient mortality than ABCD-10 in a North American cohort of patients treated at the tertiary burn center.

Automated summary

The ABCD-10 model was proposed as an alternative to SCORTEN for predicting in-hospital mortality in patients with SJS/TEN, but in a study at a large tertiary burn center, SCORTEN was found to be a better predictor of inpatient mortality. Both models showed good discrimination, but SCORTEN had better calibration for predicting mortality in this patient cohort.