The SCORTEN scale consists of seven independent risk factors for death, including the extent of lesions at the onset [18,19]. The SCORTEN value at admission was 3.29, with an estimated mortality of 41,9%. The actual mortality rate was 12,5%. Our protocol improved the survival, OR=26,57, RR=6,34, p=0,022. These results suggest that our applied protocol significantly reduced the prognosed outcome. We use a combination of plasmaphereses and intravenous immunoglobulins. Our observation is not the first to show the observed decrease in mortality risk in a group of patients treated with immunomodulation . Differences between predicted and actual mortality, as well as a tendency for SCORTEN to overestimate the risk, were also observed by Imahara et.al .
In our observation SCORTEN showed 100% sensitivity, and only 23,81% specificity. Independent risk factors did not reach statistical significance in our observation.
The extent of lesions is one of the factors distinguishing TEN from SJS and SJS/TEN [2,18], included in SCORTEN, where values in excess of 10% of the affected body surface area are an independent mortality risk factor . Our cohort consisted of 24 patients with confirmed TEN, whereas in the Bastuji-Garin study TEN patients amounted to 41.8% of the group, therefore the study group was not homogenic. This might be one of the causes of the lower accuracy of the scale for the most severe cases. In a study by Zavala et al., TBSA and heart rate were not statistically significant predictors of death on days 1 and 3 .
Torres-Navaro et al. evaluated the current accuracy of SCORTEN in a meta-analysis of 64 papers . SCORTEN scores lower than 3 overestimated mortality, whereas ones higher than 3, underestimated it. It is not only certain variables that are controversial, but the methodology used to determine SCORTEN is questionable as well . The scale was created on the basis of a database from the 1970s and 1980s, validated with data from the 1990s. Significant advances were made in supportive treatment and immunomodulatory therapy over the subsequent decades. With these developments and better sepsis control , different factors emerged. One of them is acute renal failure within the initial days of hospitalization , or a history of renal failure and dialysis prior to admission . Renal insufficiency is not a novel risk factor, as it was described by Revuz in 1987 . Bronchial necrosis and respiratory failure worsen the outcome as well . Watanabe recognized late initiation of treatment at the specialty hospital as one of the mortality prognostic tools . There is some evidence, however, that SCORTEN assessment can remain a valuable prognostic tool within the first 5 days after admission . What is more, some authors use it later, to evaluate the efficacy of immunomodulatory treatment .
Age is another known death risk factor. In our cohort, survivors were younger than non-survivors (49 years vs 55 years respectively). In our observation, the cut-off for death was 82 years, with the data not reaching statistical significance. In SCORTEN study, the mean age of the study group was 42.3 years, with the age cut-off established at over 40 years. In the ABCD-10 score, a prognostic scale created on the basis of the outcomes of 370 patients in a multi-institutional study conducted in the US, the cut-off age was 50 . Thakur did not observe any impact of age on survival in SJS/TEN or TEN patients . Watanabe determined age over 70 to be a death risk factor .
The accuracy of laboratory results evaluated in SCORTEN is also questionable [7, 26]. Hyperglycemia is one of the risk factors for sepsis and death in TEN. However, there are many variables that can influence serum glycemia . In our ICU, all severe patients received parenteral nutrition and often required insulin intake. In such cases, the serum levels of glycemia, even within first 5 days of observation, are in our opinion an unreliable indicator. Our analysis did not prove serum glycemia to be a valuable death prediction factor in our cohort.
Blood urea nitrogen (BUN) is an independent mortality risk factor in critically ill patients . BUN levels represent renal function but also neurohumoral activity in fluid homeostasis . TEN pathogenesis is associated with dermal infiltration with dermoepidermal cytotoxic T-lymphocytes [1,4]. Keratinocyte apoptosis and local inflammation  lead to subepidermal bullae, or epidermal necrolysis without destruction of dermal layers [2,17]. The destruction of the external skin layer causes massive fluid loss and, as a consequence, hypovolemic and redistributive shock, similar to burn shock . Non-inflammatory early acute renal failure is a common complication in severe burns affecting over 20% of the TBSA [29-31]. In TEN, however, early renal failure might be caused not only by loss of fluid, but also by circulating pro-inflammatory cytokines and complement . Renal failure requiring renal replacement therapy before the onset of TEN symptoms is the most important factor burdening survival in the ABCD-10 score . Renal function insufficiency at admission to the burn unit was corelated with a poorer outcome.
Low serum bicarbonate level is associated with higher mortality risk in chronic kidney disease. The NHANES survey, analysing cause-specific mortality, revealed that serum bicarbonate below 22 mEq/l was associated with a significantly worse survival. Low serum bicarbonate indicated a 46% higher risk of death among neoplastic patients . Low bicarbonate levels in ICU patients are also a valuable indicator for predicting early mortality and risk of acute kidney injury . Low bicarbonate levels showed to be the most important predicting factor for TEN patients . Also in our cohort, low serum bicarbonate was associated with higher mortality.
The mortality rates among patients with SJS/TEN and a history of malignancy are worse . According to SCORTEN, history of neoplasm is a mortality risk factor . Wu et al. analysed malignancies in SJS/TEN patients. They discovered that phenytoin was more commonly used in the neoplastic group. The types of neoplasms that correlated with higher mortality risk were hepatocellular carcinoma and colorectal cancer . However, it is not only the malignancy itself that should be seen as a risk factor, but also systemic treatment. The study showed that patients who underwent chemotherapy 3 months before the onset of SJS/TEN had worse survival . Malignancies are often connected with malnutrition and pancytopenia, which are independent risk factors in SJS/TEN. Wound healing is also affected unfavourably . We did not find a statistically significant relationship between history of neoplasm and higher mortality risk, but this might be due to limited data.