755 State-wide One-year Mortality for Stevens-Johnson/toxic Epidermal Necrolysis Spectrum Disorders Is Greater at Burn Facilities: Expected, or Sobering Wake-up Call?

Abstract

Abstract Introduction Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) exist on a spectrum of disease and are rare dermatologic emergencies with a significant risk of morbidity and mortality in the post-onset period. Retrospective studies attribute differing impacts of specialized treatment facilities on outcomes. However, the role of comorbidities and hospital-acquired conditions is unknown. We hypothesized that additional factors determine post-discharge mortality. Methods We queried a dataset from a statewide mandatory-reporting administrative database. We evaluated for initial presentations to acute care facilities of unique patients with discharge diagnoses of SJS, SJS/TEN overlap syndrome, and TEN over 16 years. We extracted all available pre- and post-onset records for these patients. Burn centers were identified. We accounted for all transfers. We evaluated demographic, comorbidity, and management data for the index presentation and subsequent transfers, evaluating the outcome of mortality and secondary outcomes. Exclusion criteria focused on detecting incomplete records. Comorbidities were captured by the Elixhauser comorbidity index. Univariate and multivariate logistic analyses were performed for risk of mortality. Results There were a total of 1903 unique initial patient encounters. Of these, 847 satisfied all exclusion criteria. 579 patients (68.4%) had a discharge (or pretransfer) diagnosis of SJS, 117 patients (13.8%) had SJS/TEN overlap syndrome, and 151 patients (17.8%) had TEN. Application of exclusion criteria caused no significant change in distribution of the occurrence of spectrum diseases (p = 0.31) nor in the frequency of care at a burn care facility (p = 0.60), suggesting no bias from exclusion. Of the 847 included patients, 22.8% of patients were treated at burn centers, of whom 32.1% were transferred from external facilities. Crude 90-day and 1-year mortality were identical, 21.7% at non-burn centers versus 32.1% at burn centers (p < 0.01). Pre-existing comorbidities did not significantly impact mortality (p = 0.86). Final logistic model details are shown below. The model’s predictive capability was acceptable (c-statistic = 0.88) and there was good fit (Hosmer-Lemeshow p = 0.42). Conclusions These results suggest crude evaluation of outcomes by treatment facility may be confounded by analysis that does not account for complex comorbidities and hospital diagnoses. Furthermore, severity of illness is not a independent predictor. Further research is necessary. Applicability of Research to Practice SJS/TEN spectrum disorders have a complex and prolonged impact on patient health. The role of burn centers is complex.