# Predictive value of a severity-of-illness score for toxic epidermal necrolysis (SCORTEN) factors for in-hospital mortality in Stevens–Johnson syndrome/toxic epidermal necrolysis

## TL;DR

This study evaluates how well a severity score (SCORTEN) predicts death in patients with Stevens–Johnson syndrome or toxic epidermal necrolysis.

## Contribution

The study identifies specific SCORTEN components that are strong predictors of mortality in real-world clinical settings.

## Key findings

- Age >40 years, malignancy, heart rate >120 bpm, and high serum urea were strongly linked to in-hospital death.
- Urea >28 mg/dL, malignancy, and tachycardia remained independent predictors of mortality in multivariate analysis.
- Some SCORTEN parameters like initial epidermal detachment and age lost significance in adjusted models.

## Abstract

Stevens–Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) are life-threatening mucocutaneous reactions with high mortality. The A severity-of-illness score for toxic epidermal necrolysis (SCORTEN) scale remains the principal tool for early estimation of in-hospital death risk. The study was aimed to assess the predictive value of individual SCORTEN components for hospital mortality in a real-world Moscow cohort and to characterize demographic and etiologic factors associated with outcome.

We retrospectively reviewed 150 adult patients (median age 50 years; IQR: 31–63; 59.3% female) admitted with SJS (45.3%) or TEN (54.7%) between 2019 and 2024. SCORTEN parameters were recorded within 24 h of admission. Univariate analyses (chi-square or Fisher's exact test) identified associations between each SCORTEN variable and in-hospital death. Kaplan–Meier survival curves and Cox proportional-hazards models quantified time-to-event outcomes and adjusted hazard ratios (HR).

Overall mortality was 18.7% (n = 28). In univariate analysis, age >40 years (OR: 3.53; p = 0.01), associated malignancy (OR: 3.35; p = 0.03), heart rate > 120 bpm (OR: 9.42; p < 0.001), serum urea >28 mg/dL (OR: 14.25; p < 0.001), and bicarbonate <20 mmol/L (OR: 7.25; p < 0.001) were significantly linked to death. In multivariate Cox regression, malignancy (HR: 3.57; p = 0.05), urea > 28 mg/dL (HR: 4.33; p = 0.03), and tachycardia (HR: 2.77; p = 0.04) remained independent predictors. Initial epidermal detachment, serum glucose, and age did not retain significance.

These findings support continued use of SCORTEN while highlighting the need to recalibrate or augment its parameters—particularly renal and oncologic variables—to improve risk stratification in current therapeutic contexts.