Outcome of patients with toxic epidermal necrolysis syndrome revisited

Toxic epidermal necrolysis syndrome is an uncommon, acute, life-threatening, medication-induced disorder with a reported mortality rate of 20 to 60 percent. Different variables have been identified as risk factors. The extent to which these variables, when combined, affect the mortality and outcome in toxic epidermal necrolysis syndrome patients has not yet been reliably defined. Because of the high mortality rate, the logistic analysis of studied variables was performed to see whether a prognostic algorithm could be developed to aid the management of these patients. Thus, a retrospective review of 56 consecutive toxic epidermal necrolysis syndrome patients treated over a period of 13 years was undertaken in the authors' burn center. The demographics included age, sex, race, and total body surface area involved. The other variables studied were comorbidities, sepsis, steroid administration, and the interval between onset of rash and burn center admission. Data were subjected to Fisher's exact test and logistic analysis. Thirty-six patients (64.3 percent) were alive and 20 (35.7 percent) died. Univariate analysis indicated that the male/female ratio was 12:24 for survivors and 9:11 for nonsurvivors (p = 0.4). The white/nonwhite ratio was 80 percent for survivors and 54 percent for nonsurvivors (p = 0.58). The median age was 48.4 +/- 22.8 years (survivors, 41.7 +/- 22.0; nonsurvivors, 60.5 +/- 19.5; p = 0.002). Total body surface area involvement for survivors was 56.9 +/- 32 and 77.7 +/- 21 for nonsurvivors (p = 0.005). The presence of one or more comorbidities between the two groups differed (53 percent survivors and 90 percent nonsurvivors, p = 0.007), indicating eight times higher odds of dying in their presence. The average time between the onset of symptoms and admission to the burn unit was 5.25 +/- 3.4 days for survivors and 7.15 +/- 4.5 days for nonsurvivors (p = 0.08). The presence of sepsis (19.4 percent survivors, 95 percent nonsurvivors, p < 0.001) decreased odds for survival by a factor of 79. Steroids given as a single dose or multiple doses before the patient's transfer to the burn unit were not significantly associated with death (44 percent survivors, 65 percent nonsurvivors, p = 0.14). A multivariate logistic regression model yielded odds ratios of 1.11 (95 percent confidence interval, 1.03 to 1.19) for age in years, 304 (95 percent confidence interval, 8.83 to 10,400) for the presence of sepsis, and 1.03 (95 percent confidence interval, 0.99 to 1.08) for body surface area in percent. All those entering the burn unit with sepsis died. Equivalently, no survivors had sepsis before admission to the burn unit, whereas 55 percent of nonsurvivors had sepsis before admission and 40 percent developed sepsis after admission. When investigating the effect of age and sepsis, no patients over age 60 ever having sepsis survived, whereas all those who were under 60 and without sepsis survived. Likewise, all patients whose age was over 60 and whose total body surface area involved was over 60 percent died. The main factors contributing to the mortality from toxic epidermal necrolysis syndrome, when considering covariates separately, are the presence of sepsis at any time (odds ratio, 79), the presence of comorbidities (odds ratio, 8.05), age, and total body surface area, whereas multivariate models suggested age (odds ratio per year of additional age, 1.11), total body surface area (odds ratio per additional percent of body surface area, 1.03), and the presence of sepsis (odds ratio, 304). By using the actual coefficients in the logistic model, the log odds that the patient will die as the result of his or her condition can be summarized in the following formula: -11.5 + (10 percent of the patient's age + 3 percent of total body surface area + 5.75 if sepsis is present). The awareness of the importance of these covariates, and their early recognition as risk factors, should offer a focused approach to the patients' management and improve their outcome.