Objective: To investigate the clinical features and prognostic influencing factors of toxic epidermal necrolysis (TEN). Methods: A retrospective observational study was conducted. From January 2008 to March 2019, a total of 46 TEN patients who met the inclusion criteria were admitted to Guangdong Provincial People's Hospital. The gender, age, and hospital admission diagnosis of the 46 patients, the category of department admitted of patients complicated with sepsis, death ratio of the sepsis patients with or without treatment history in intensive care unit (ICU)/department of burns and wound repair, and the cause of death of the deceased patients were recorded. Depending on whether complicated with sepsis, the patients were divided into sepsis group (32 cases) and non-sepsis group (14 cases). According to whether died or not, the patients were divided into death group (9 cases) and survival group (37 cases). The specific conditions of suspected pathogenic agents and combined underlying diseases, the abnormality of transaminase/bilirubin, creatinine, and platelet count in blood on admission, and the detection of pathogenic microorganisms and drug resistance during the course of disease of patients were recorded in both sepsis group and non-sepsis group. The gender, age, lesion area, severity of illness score for TEN (SCORTEN) system score, combined underlying diseases on admission, and blood microbial culture positivity, hormone use, and gamma globulin use during the course of disease of patients between sepsis group and non-sepsis group, death group and survival group were compared respectively. Data were statistically analyzed with chi-square test, Fisher's exact probability test, and Mann-Whitney U test. The factors with statistically significant differences between sepsis group and non-sepsis group, death group and survival group were selected for binary multivariate logistic regression analysis, so as to screen the independent risk factors affecting sepsis and death in TEN patients. Results: Of the 46 TEN patients, 30 were male and 16 were female, aged from 8 months to 92.0 years, with 11 cases (23.91%) of epidermolysis bullosa, 9 cases (19.57%) of exfoliative dermatitis, 9 cases (19.57%) of TEN, 7 cases (15.22%) of epidermolysis bullosa, 6 cases (13.04%) of Stevens-Johnson syndrome, and 4 cases (8.70%) of severe drug rash for hospital admission diagnosis. The patients complicated with sepsis were admitted to 11 departments, and the death ratio of patients with treatment history in ICU/department of burns and wound repair was similar to that of patients without such department treatment history ( P >0.05). All the deceased patients were complicated with sepsis, which was also the main cause of death. On admission, the suspected pathogenic agents of patients in sepsis group were mainly allopurinol (8 cases) and non-steroidal anti-inflammatory drugs (4 cases), while those in non-sepsis group were allopurinol (3 cases) and psychotropic drugs (3 cases). Patients in sepsis group combined as many as 10 underlying diseases, while those in non-sepsis group combined only 4 underlying diseases. The proportions of patients with increased creatinine ( χ 2 =13.349, P <0.01) and decreased platelet count ( P <0.01) in sepsis group were significantly higher than those in non-sepsis group, while the transaminase/bilirubin abnormality was similar to that in non-sepsis group ( P >0.05). A wide variety of pathogens were detected in the blood, respiratory tract secretions, and skin secretions of 21 patients in sepsis group, and 14 patients were infected with drug-resistant bacteria; among the 9 strains cultured from the blood samples, 8 were drug-resistant bacteria and 6 were Gram-positive bacteria. In non-sepsis group, pathogens were detected in blood, respiratory tract secretions, and skin secretions of 8 patients, with fewer species, and 6 patients were infected with drug-resistant bacteria. The gender, age, lesion area, blood microbial culture positivity, hormone use, and gamma globulin use of patients in sepsis group were similar to those in non-sepsis group ( P >0.05). The proportion of patients combined with underlying diseases ( χ 2 =4.493, P <0.05) and the proportion of patients with SCORTEN system score of 4-6 points ( P <0.01) of patients in sepsis group were significantly higher than those in non-sepsis group. The gender, combined underlying diseases, lesion area, blood microbial culture positivity, hormone use, and gamma globulin use of patients were similar between survival group and death group ( P >0.05). The proportion of patients with age≥60 years and the proportion of patients with SCORTEN system score of 4-6 points of patients in death group were significantly higher than those in survival group ( χ 2 =4.412, 11.627, P <0.05 or P <0.01). The SCORTEN system score was an independent risk factor affecting sepsis and death in TEN patients (odds ratio=3.025, 2.757, 95% confidence interval=1.352-6.769, 1.244-6.110, P <0.05 or P <0.01). Conclusions: The diagnosis of TEN is difficult on admission. Male population is susceptible to TEN, and allopurinol is the common pathogenic agent. The proportion of patients combined with underlying diseases is high in TEN patients complicated with sepsis, with mainly drug-resistant bacteria and mostly Gram-positive bacteria in blood-borne infections. The deceased patients are older than the survived, and the main cause of death is sepsis. The SCORTEN system score is an independent risk factor affecting sepsis and death in TEN patients.