Abstract: 【Abstract】 Objective To analyze the clinical features of Stevens-Johnson syndrome （SJS） and toxic epidermal necrolysis （TEN）, and to evaluate the accuracy of the TEN-specific severity-of-illness score （SCORTEN） in predicting death. Methods A retrospective analysis was conducted on 104 patients with SJS/TEN at the First Affiliated Hospital of Army Medical University between December 2003 and December 2023. Clinical data, such as sensitizing drugs, clinical manifestations, laboratory tests, SCORTEN scores, and treatment regimens, were collected and retrospectively analyzed. The efficacy of different treatment modalities for TEN was analyzed. The receiver operating characteristic （ROC） curve was used to assess the accuracy of SCORTEN in predicting death in TEN patients. Results Among the 104 patients, 57 were males and 47 were females, with ages ranging from 12 to 93 years （45.45 ± 19.76 years）. There were 52 cases of SJS and 52 cases of TEN （including 1 case of SJS-TEN overlap）. The ages of SJS patients （40.42 ± 17.06 years） were significantly lower than those of TEN patients （50.48 ± 21.13 years; t = 2.67, P = 0.009）; the total hospital stay was 14.47 ± 7.24 days, and the TEN patients had significantly longer hospital stays （16.65 ± 7.82 days） compared with the SJS patients （12.29 ± 5.92 days; t = 3.21, P = 0.002）. Definite sensitizing drugs were identified in 82 patients; combined drugs were the most common sensitizing cause （26 cases, 25.00%）, and 20 patients reported combination treatment with antibiotics; 56 patients （53.85%） were treated with single drugs, and the common sensitizing drugs included carbamazepine （16 cases, 15.38%）, nonsteroidal anti-inflammatory drugs （13 cases, 12.50%）, and allopurinol （9 cases, 8.65%）. Laboratory test results showed that the proportions of patients with decreased lymphocyte counts, elevated alanine aminotransferase and/or aspartate aminotransferase levels, decreased albumin levels, and with increased creatinine levels were significantly higher in the TEN patients than in the SJS patients （all P < 0.05）, while the proportion of patients with increased monocyte counts was significantly lower in the TEN patients than in the SJS patients （P = 0.006）. Among the 52 TEN patients, 33 had SCORTEN scores < 3 points and 19 had scores ≥ 3 points, with 12.32 expected deaths and 4 actual deaths; the ROC curve analysis indicated that the area under the curve for SCORTEN in predicting death in TEN patients was 0.784 （95% CI: 0.558 - 1.00）. Of the 104 patients, 57 （54.81%） received a monotherapy regimen （glucocorticoids only）, and 47 （45.19%） received combination therapies, including glucocorticoids combined with intravenous immunoglobulin （IVIG） in 42 cases. Among the TEN patients, 30 started combination therapy with IVIG within 7 days after glucocorticoid treatment （early combination therapy group）, and 7 patients started IVIG after 7 days of glucocorticoid treatment （late combination therapy group）. The early combination therapy group showed a shorter time to lesion stabilization （10.82 ± 3.35 days） compared with the late combination therapy group （15.50 ± 4.04 days; LSD-t = 2.87, P = 0.006）. Conclusions The main sensitizing cause in SJS/TEN patients was the combination of antibiotics. Early combination of glucocorticoids and IVIG could shorten the disease course in TEN patients to some extent. SCORTEN still holds a certain clinical value in predicting the prognosis of TEN patients.

Key words: Stevens-Johnson syndrome, Epidermal necrolysis, toxic, Drug eruptions, Clinical features, SCORTEN, Death, Therapy, Prognosis