Supplemental materials for publication "Validation of Two Predictive Models Designed to Aid Clinicians in Identifying Stevens-Johnson Syndrome/Toxic Epidermal Necrolysis: A Single Institution, Retrospective Review." JAAD. 2021.
This is a supplemental methods section that demonstrates the two models used in our manuscript for external validation. The two models replicate the multivariable models proposed in the manuscript by Weinkle et al., which were formulated via stepwise regression selecting 4 predictor variables. The first model with the presence or absence of Nikolsky sign, atypical target, fever, and lymphopenia (Supplemental Figure 1). The second model with the presence or absence of Nikolsky sign, atypical target, fever, and mucosal involvement (Supplemental Figure 2). In order to externally validate these models, either a value of 1 (if present) or 0 (if absent) was plugged into the equation for each characteristic of each patient in our retrospective cohort. The value of the probability of SJS/TEN was documented and used in the validation efforts. .