LAUSR

vs. 10%) and more frequent need for mechanical ventilation (48% vs. 5%). Besides older patients, with a poor outcome as expected, our study identified two distinct profiles of EN in younger patients. The first was Stevens–Johnson syndrome in young women, and the second was toxic epidermal necrolysis in patients 10 years older than in the first group. Whereas age over 40 years, baseline involved BSA > 10%, cancer, and urea >10 mmol L 1 are known prognostic factors in the SCORTEN and ABCD-10 scores, we identified other baseline biological variables (pharyngolaryngeal involvement, neutropenia, lymphopenia, elevated LDH, hypoalbuminaemia and hypophosphataemia) not previously described to be associated with poorer outcomes. Nonetheless, in this study, we did not incorporate them in new prognostic models. Futures studies should aim to assess their relative weights and their relationships with the pathophysiological mechanisms leading to the worsening of EN in the first days of the acute phase, and thereby to a poorer prognosis. We acknowledge the limitations of our study, mainly its retrospective monocentric design. Moreover, despite the dimensionality reduction, we cannot exclude overfitting of our data. Yet, considering the rarity of the disease, we believe our study provides new insights and delineates different profiles of patients with EN at admission linked to different outcomes.