ORIGINAL ARTICLE

Toxic Epidermal Necrolysis: Performance of

SCORTEN and the Score-Based Comparison of the
Efficacy of Corticosteroid Therapy and Intravenous
Immunoglobulin Combined Therapy in China
Qin-yuan Zhu, MD, Li Ma, MD, PhD, Xiao-qun Luo, MD, PhD,
Hui-yuan Huang, MD

Toxic epidermal necrolysis (TEN) represents the most severe drug-related skin condition
that is potentially life-threatening with no well-established treatments. The application of
corticosteroid therapy is controversial, whereas recently intravenous immunoglobulin (IVIG)
therapy is emerging as a promising new method. A severity-of-illness score for TEN (SCORTEN)
has gained acceptance in some western countries. In this study, our objectives were to assess
the applicability of SCORTEN in Chinese patients with TEN and to evaluate the efficacy
of the combination therapy of IVIG and corticosteroid in these patients. We performed a
retrospective review of data from 61 patients with TEN treated at our intensive care unit from
2000 to 2010 to assess the performance of SCORTEN. In particular, 55 patients between
2002 and 2010 were grouped as a series to compare the therapeutic effects of corticosteroid
therapy and IVIG combined therapy contemporaneously. During this period, 16 patients were
administered with corticosteroid therapy and 39 were treated with the combination therapy.
An initial dose of 1.5 mg/kg/day of methylprednisolone was given to all TEN patients. The
combination therapy was combined with a total dose of 2 g/kg IVIG within 5 days. Areas
under receiver operating characteristic curves and Hosmer-Lemeshow statistic were analyzed to
illustrate the performance of SCORTEN. The comparison of the efficacy of the two therapies
was conducted on the basis of clinical outcomes, standardized mortality ratio (SMR), and
survival analysis. The overall actual mortality of patients between 2000 and 2010 was 16%
(10/61), statistically insignificantly lower than predicted (24%, SMR = 67.98). Excellent
discriminatory power (the areas under the receiver operating characteristic curves: 88.9, 88.2,
90.6%) and good calibration (P = .637, .833, .530) were found in all the groups. In patients
admitted between 2002 and 2010, IVIG combined therapy showed a trend toward reducing
the mortality rate (13%, SMR = 52.35), whereas corticosteroid monotherapy suggested no such
difference (31%, SMR = 123.92). Besides, the cumulative survival rates of the combination
therapy were higher at almost all the levels of SCORTEN (P = .002), especially at the score
of 5 (P = 3.10 × 10−7). Compared with corticosteroid alone, the combination therapy
arrested progression earlier (P = .013), although it did not significantly lead to a tapering of
corticosteroid or a reduction of the time of hospitalization. We concluded that SCORTEN was
generally applicable to Chinese patients with TEN. The comparison of the effect indicated that
the combination therapy might achieve a better therapeutic effect than the administration of
corticosteroid alone, especially in severe TEN patients. (J Burn Care Res 2012;33:e295–e308)

Toxic epidermal necrolysis (TEN) represents reactions, with an incidence of approximately 0.4 to
the most severe form of cutaneous adverse drug 1.2 cases per million.1,2 It is an acute life-threatening

From the Department of Dermatology, Huashan Hospital, Fudan

University, Shanghai, People's Republic of China. 12 Wulumuqi Zhong Rd, Jing’an District, Shanghai,
Supported by the National Natural Science Foundation of China China.
[grant 81071287].
Copyright © 2012 by the American Burn Association.
Drs Zhu and Ma contributed equally to this work.
1559-047X/2012
Address correspondence to Xiao-qun Luo, PhD, MD,
Huashan Hospital affiliated to Fudan University, DOI: 10.1097/BCR.0b013e318254d2ec

e295
Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303
by guest
on 14 March 2018
 Journal of Burn Care & Research
e296   Zhu et al November/December 2012

disorder, characterized by erosions of mucous mem- and 4) clinical documents were reviewed to rule out
branes and detachment of large epidermal sheets on autoimmune bullous diseases, staphylococcal scalded
more than 30% of the BSA.3,4 Although rare, TEN skin syndrome and so forth. Patients were promptly
has a significant effect on public health because of its transferred to the ICU after being clinically diag-
high mortality, which ranges from 20 to 30% even nosed in our emergency room. Data analyzed were
in the intensive care units (ICUs).4,5 Because of sys- those recorded during the first 24 hours in the ward
temic involvement and high mortality, it is important and were abstracted from medical charts.
to have a predictable standard of the severity of TEN
either for prognosis or for treatment.6 Treatment
A severity-of-illness score for TEN (SCORTEN) All patients who received the diagnosis of TEN were
was developed from a logistic regression model in treated with methylprednisolone as early as possible if
2000 in France.5,7,8 Based on seven independent there was no absolute contraindication, with an initial
prognosis factors, it has gained wide acceptance in dose of 1.5 mg/kg/day. In the combination therapy,
western countries.6,9–11 Authors either reported the a dose of 0.4 g/kg/day of IVIG for 5 days, combined
calculated SCORTEN as a clinical parameter or with corticosteroid, was considered for patients
additionally used the score to estimate the efficacy of with progressive disease (eruptions exacerbation,
new treatments.12–18 However, whether SCORTEN laboratory data worsening, or the occurrence of new
is applicable in Asian countries remains uncertain complications) after the administration of 1.5 mg/
and its performance has yet to be evaluated. kg/day of methylprednisolone for 3 to 5 days.
Unfortunately, systemic therapies for TEN continue IVIG was administered by continuous infusion for
to be sources of controversy. Corticosteroid therapy patients without severe renal or heart failure after
has been widely used with uncertain results whereas its routine institution in 2002. Once the condition
recently, intravenous immunoglobulin (IVIG) has was under control (no new eruptions, Nikolsky
stimulated extensive investigation as an adjunctive sign turning negative, and exudation improved)
therapy.19–22 IVIG blocks in vitro interaction of the Fas and reepithelialization had started, with laboratory
receptor with the Fas ligand and prevents keratinocytes tests satisfactorily monitored and recorded on a daily
from death.23 Initial reports of the use of IVIG were basis, the corticosteroid dose was tapered promptly
favorable, with lower mortality.13,24–26 However, repli- to avoid the risk of adverse effects. The whole
cation of those results has been difficult, and several process was judged by a group of dermatologists to
studies have shown no benefit.15,27,28 In addition, the avoid individual bias. All enrolled patients received
efficacy of the combination therapy of IVIG and corti- consecutive treatment in our ICU. This study was
costeroid for TEN has seldom been reported. approved by, and performed according to instructions
The aim of this study was therefore to validate the of the Research Ethics Board of Huashan Hospital
generally predictive ability of SCORTEN in a cohort affiliated to Fudan University. The administration
of Chinese TEN inpatients. On the basis of the score, of corticosteroid or IVIG was also approved by
we therefore retrospectively compared the effects of the ethics committee of Huashan Hospital, and
the solo corticosteroid therapy and the combination informed consent was obtained from each patient.
therapy of IVIG and corticosteroid. A standardized treatment protocol for TEN was
provided to each patient. Systemic antibiotics were
METHODS not used routinely, but only in case of infections
confirmed by microorganism examinations. Fluid
Patients requirement was estimated with Parkland formula
From January 2000 to April 2010, 61 patients from and administered as crystalloids (normal saline
the dermatology ICU of Huashan Hospital affiliated and Ringer’s lactate) as needed. Intravenous fluids
to Fudan University with a diagnosis of TEN were replacement was titrated to obtain a urine output
included in this study. Patients enrolled in the study of 0.5 to 1.0 ml/kg/hr and stable hemodynamics.
fulfilled the following inclusion criteria: 1) diagno- Nutrition requirements were predicted and were
sis of TEN confirmed by skin biopsy pathologically offered to all patients, and enteral nutrition was
showing full-thickness necrosis of the epidermis; 2) instituted in patients who were unable to eat because
the maximal percentage of denuded skin was above of mucosal involvement. Pain was treated with
30% to exclude Stevens-Johnson syndrome or Ste- administration of opiates, analgesics, and anxiolytics
vens-Johnson syndrome/TEN overlap; 3) disease as needed. Daily respiratory monitoring was
progression within the 24 hours preceding admission; performed, on the basis of clinical examination, chest

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

by guest
on 14 March 2018
 Journal of Burn Care & Research
Volume 33, Number 6 Zhu et al   e297

x-ray, and arterial blood gas analysis. The decision to between 2002 and 2010, 55 patients were grouped
monitor patients’ respiratory status daily was also in as a series. Clinical characteristics of these patients, as
consideration of the extent of upper airway mucosal well as outcomes in terms of time to the arrest of pro-
involvement, the potential for airway obstruction, gression, time to the tapering of corticosteroid, time of
and the severity of respiratory distress. According hospitalization, and complications were collected and
to the acute respiratory distress syndrome network, summarized.
patients with acute respiratory distress syndrome in
our ICU were treated with approximate 6 ml/kg Statistical Analysis
predicted body weight tidal volumes and positive Descriptive variables were summarized by number
end-expiratory pressure.29 Our ophthalmologist (percentages) or mean ± SD. Parametric data
consultants were consulted for eye evaluations. were compared using Student’s two-tailed t-test.
Wound management was strictly conservative, with Comparisons of the nonparametric data were
evacuation of blister exudates and concomitant performed using χ2 square and Mann–Whitney U
replacement of the detached epidermis on the tests, as well as Fisher’s exact probabilities whenever
dermis underneath. Oozing areas were covered by appropriate. The standardized mortality ratio
nonstick dressings impregnated with isotonic sodium (SMR = Σobserved deaths/Σexpected deaths ×
chloride solution (normal saline) and changed 100) was used to determine whether there was a
at appropriate times based on local conditions. difference between actual and expected mortality.31
To stop the progression of skin symptoms from Likelihood ratios (LRs) were estimated with their
erythema into blisters and epidermal detachment, 95% confidence intervals (CI). Areas under receiver
zinc oxide powder and topical steroid cream were operating characteristic (ROC) curves (area under
applied when there was no exudation or infection. the curve [AUC]) were calculated to evaluate the
Topical antibiotics such as Bactoderm (mupirocin discriminatory power of the model.32 Calibration, that
ointment), neomycin sulfate ointment, and Fucidin is, correspondence between the expected mortality
(fusidic acid cream) were used in accordance with produced by the model and the actual mortality,
daily wound cultures. None of the patients received was evaluated by Hosmer-Lemeshow statistic.33
plasmapheresis or other immunosuppressive drugs. Survival curves in accordance with the SCORTEN
were analyzed by the log-rank test. Differences were
Evaluation of the SCORTEN and Kaplan- considered significant at P < .05. Data were analyzed
Meier Survival Curve by SPSS17.0 (SPSS Inc., Chicago, IL) software.
SCORTEN includes seven clinical variables within
the first 24 hours after admission: 1) age ≥ 40 years,
RESULTS
2) presence of malignancy, 3) skin detachment ≥ 10%
of BSA, 4) heart rate ≥ 120/min, 5) serum glucose Patient Characteristics
level > 14.0 mmol/L (252 mg/dl), 6) serum bicar- The study included 61 TEN patients from 2000 to
bonate level < 20 mmol/L (20 mEq/L), and 7) 2010, 35 male and 26 female patients (Table 1).
serum urea nitrogen level > 10 mmol/L (28 mg/dl). Distribution of SCORTEN in 10 years is shown in
Each variable has an equal weight in the score. The Figure 1 (P = .494). Details of clinical characteristics,
expected mortality rate predicted by the SCORTEN time of admission, previous diseases, and drugs
was calculated using the formula: p(death) = elogit / taken in a 2-month period preceding the onset of
1 + elogit where, logit = −4.448 + 1.237 (SCORTEN).5 TEN are listed in Table 2. The mean age of patients
The outcome at hospital discharge was recorded to was 46.4 ± 19.0 years (range, 18–91); the TBSA
calculate the actual mortality. The cumulative survival percentage of skin slough was 90.3 ± 12.1%, and the
rate over time was determined using Kaplan-Meier delay in admission was 4.9 ± 4.1days. Thirty-nine
survival analysis.30 The point started from admission patients (64%) were treated with IVIG combined
to our hospital. The end point was the outcome at therapy, 22 (36%) with corticosteroid only.
day 120 after admission. All the patients were retro- Me­chanical ventilation was provided to 13 patients
spectively followed up for the outcome within this with respiratory failure. No statistical difference
period by phone or by review of clinical documents. in age (P = .148), sex (P = .458), admission delay
(P = .100), or SCORTEN (P = .842) was found
Evaluation of Therapeutic Effects between the two groups. The most common drug
To compare the therapeutic effects of the ­combined associated with TEN was allopurinol, followed
therapy and corticosteroid therapy con­tem­poraneously, by carbamazepine, antipyretic/analgesic agents,

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

by guest
on 14 March 2018
 Journal of Burn Care & Research
e298   Zhu et al November/December 2012

Table 1. Comparison of the SCORTEN (2000–2010): predicted mortality, actual mortality, and SMR
Predicted Mortality Actual Mortality

No. of
SCORTEN Score Patients (n) % No. of Deaths % Deaths SMR P

All patients
 1 16 2.1 0.336 0 0 0.00
 2 23 12.2 2.806 4.35 1 35.64
 3 12 32.4 3.888 25.00 3 77.16
 4 5 62.3 3.115 40.00 2 64.21
 5 5 91.3 4.565 80.00 4 87.62
Total 61 24.11 14.71 16.39 10 67.98 .637
Corticosteroid therapy
 1 5 2.1 0.105 0 0 0.00
 2 10 12.2 1.220 10.00 1 81.97
 3 3 32.4 0.972 33.33 1 102.88
 4 2 62.3 1.246 50.00 1 80.26
 5 2 91.3 1.826 100.00 2 109.53
Total 22 24.40 5.369 22.73 5 93.13 .833
IVIG therapy
 1 11 2.1 0.231 0 0 0.00
 2 13 12.2 1.586 0 0 0.00
 3 9 32.4 2.916 22.2 2 68.59
 4 3 62.3 1.869 33.3 1 53.50
 5 3 91.3 2.739 66.7 2 73.02
Total 39 23.95 9.341 12.82 5 53.53 .530

SMR, standardized mortality ratio; IVIG, intravenous immunoglobulin.

traditional Chinese medicine, cephalosporins, amo­ (P = .871), skin detachment (P = .145), SCORTEN
xicillin, other antiepileptic agents, and methazolamide (P = .855), or admission delay (P = .358) between
(Table 2). From 2002 to 2010, 55 patients were the two therapy groups between 2002 and 2010.
grouped as a series to evaluate the therapeutic effects.
Table 3 summarizes the demographic and clinical Assessment of the scorten Performance
characteristics of these patients. Similarly, there (2000–2010)
was no statistical difference in age (P = .227), sex All 61 Inpatients. The actual mortality at discharge
was 16% (10/61), insignificantly lower than predicted
(24%; SMR = 67.98; Table 1). So was it at each
SCORTEN level. An overall significant difference in the
outcome of each SCORTEN level was evident (P = .000),
and LR was 20.972 (P = .000; 95% CI, 0.000–0.048).
Compared with a score of 5, the actual mortality of the
SCORTEN of 1 and 2 was statistically decreased (P =
.000, P = .001). Calibration of SCORTEN indicated
good agreement between the expected and observed
deaths (P = .637; Table 1). The ROC curve generated,
revealed a good discriminatory power of the SCORTEN
(AUC = 0.889 ± 0.053, P = .000; Figure 2). Figure
3 shows the estimated probability of 120-day
cumulative survival ranging from 100.0  ± 
0.0% for
a score of 1 to 20.0 ± 17.9% for a score of 5. Log-
rank analysis also exhibited a highly significant
difference in accordance with SCORTEN (P = .000).
Figure 1. Distribution of the SCORTEN from 2000 to Additional analysis revealed that the survival rates of
2010. scores below 3 were statistically higher than those of

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

by guest
on 14 March 2018
by guest
Table 2. Clinical characteristics, treatment, and outcomes of TEN patients from 2000 to 2010

on 14 March 2018
Detachment
Patient Time of (%TBSA)/ Death at Outcome
No./Sex/ Admission, Mucosal Admission Discharge at Day Mechanical IVIG
Age (yr) (yr) Involvement SCORTEN Delay Medical History Previous Therapy Complication (d)* 120 Ventilation Therapy

1/M/32 2000 97/yes 1 3.0 Schizophrenia Carbamazepine No Alive No No

Volume 33, Number 6

2/M/43 2000 97/yes 2 17.0 Hypertension, Amoxicillin Hemorrhage of No Lost No No

hepatitis B the upper
gastrointestinal
Journal of Burn Care & Research

tract
3/M/55 2000 100/yes 2 7.0 — Amoxicillin No Alive No No
4/M/46 2000 94/yes 2 7.0 Gout Allopurinol No Alive No No
5/M/91 2000 97/yes 3 6.0 CI Ciprofloxacin No Lost No No
6/F/62 2001 94/yes 4 11.0 Rheumatic valvulitis, Allopurinol Pneumonia, heart No Alive No No
diabetes, gout failure
7/F/34 2002 48/no 1 0.5 Acute generalized Dipyrone No Alive No No
exanthematous
pustulosis
8/M/23 2002 94/yes 1 2.0 Nephrolithiasis Amoxicillin No Alive No Yes
9/M/58 2002 90/yes 2 5.0 CH Ceftazidime No Alive No No
10/M/78 2002 61/yes 2 12.0 CI, hypertension Cephradine, Respiratory failure No Lost Yes No

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

ibuprofen
11/M/24 2002 94/yes 2 1.0 Epilepsy Phenytoin No Alive No Yes
12/M/81 2002 90/yes 5 0.5 Asthma, chronic Allopurinol MODS Yes, 4 Dead Yes No
gastritis, gout,
CHD, chronic
kidney failure
13/F/23 2003 99/yes 1 1.0 — TCM† Sepsis, respiratory No Alive Yes Yes
failure
14/F/51 2003 99/yes 2 11.0 Epilepsy Carbamazepine No Alive No No
15/F/50 2003 93/yes 2 5.0 Rheumatoid arthritis Cephalosporin No Alive No No
16/M/53 2003 90/yes 2 9.0 Schistosomal hepa- TCM† No Alive No Yes
titis
17/M/60 2003 97/yes 3 4.0 Diabetes TCM† Renal failure No Alive No No
18/M/26 2003 100/yes 3 3.0 Metameros Cephradine, Pneumonia, septic Yes, 31 Dead No No
fibroblastic fleroxacin, shock
sarcoma phenobarbital
19/M/18 2003 69/yes 3 9.0 Epilepsy Phenobarbital MODS, Yes, 20 Dead Yes Yes
pneumonia
(Continued)
Zhu et al   e299
by guest

Table 2.  Continued

on 14 March 2018
Detachment
Patient Time of (%TBSA)/ Death at Outcome
No./Sex/ Admission, Mucosal Admission Discharge at Day Mechanical IVIG
Age (yr) (yr) Involvement SCORTEN Delay Medical History Previous Therapy Complication (d)* 120 Ventilation Therapy
e300   Zhu et al

20/F/43 2003 90/yes 4 15.0 Hypertension, CHD, Aspirin, ofloxacin, MODS, DIC, Yes, 22 Dead Yes No
hysteromyoma roxithromycin. septic shock,
norvancomy- hemorrhage of
cin, amikacin, gastrointestinal
lincomycink, tract, electro-
cephalosporin lyte distur-
bances
21/M/74 2003 99/yes 5 3.0 Hypertension, CHD, Allopurinol, Heart failure, Yes, 8 Dead No No
chronic gastritis paracetamol septic shock,
pneumonia
22/F/19 2004 97/yes 3 1.0 — Cephradine, dipy- MODS, pneumo- Yes, 6 Dead Yes Yes
rone nia
23/M/51 2004 97/yes 3 9.0 — Paracetamol, Pneumonia No Alive No Yes
amantadine
hydrochloride
24/M/74 2004 97/yes 4 3.0 Chronic bronchitis, Allopurinol MODS, pneumo- Yes, 16 Dead Yes Yes

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

hypertension nia
25/M/75 2004 97/no 5 3.0 Hypertension, gout Allopurinol MODS, pneumo- Yes, 16 Dead Yes Yes
nia
26/F/35 2005 86/yes 1 10.0 — Cephalosporin No Alive No Yes
27/M/40 2005 97/yes 2 1.0 — TCM† No Alive No Yes
28/F/72 2005 97/yes 2 7.0 Pulmonary TB, CI, Allopurinol, cefu- Heart failure, No Lost No Yes
hypertension roxime pneumonia
29/M/48 2005 97/yes 3 4.0 Hypertension, Amoxicillin, Respiratory failure No Alive Yes Yes
diabetes cefotaxime,
dipyrone
30/M/40 2006 48/no 1 2.0 — Cephalosporin, No Alive No No
31/F/19 2006 97/yes 1 5.0 — Dipyrone No Alive No Yes
32/M/42 2006 97/yes 2 2.0 Pulmonary TB Allopurinol, No Alive No Yes
amoxicillin
33/M/59 2006 97/yes 2 10.0 Epilepsy, hyperten- Phenytoin Sepsis No Alive No Yes
sion, diabetes
34/F/61 2006 79/yes 3 4.0 Brain tumor Phenytoin, No Alive No Yes
cephradine
(Continued)
November/December 2012
Journal of Burn Care & Research
by guest
Table 2.  Continued

on 14 March 2018
Detachment
Patient Time of (%TBSA)/ Death at Outcome
No./Sex/ Admission, Mucosal Admission Discharge at Day Mechanical IVIG
Age (yr) (yr) Involvement SCORTEN Delay Medical History Previous Therapy Complication (d)* 120 Ventilation Therapy

35/M/60 2006 79/yes 5 1.0 — Paracetamol, ami- MODS, pneumo- Yes, 13 Dead Yes Yes
Volume 33, Number 6

nophenazone, nia
chlorphena,
aspirin, TCM†
36/M/34 2007 94/yes 1 3.0 — Aminophenazone- No Alive No No
Journal of Burn Care & Research

co, cefopera-
zone, cefaclor
37/F/39 2007 79/yes 1 1.0 — Carbamazepine No Alive No No
38/F/37 2007 94/yes 1 6.0 — Amoxicillin, No Alive No Yes
ofloxacin
39/F/18 2007 94/yes 1 2.0 — Amoxicillin, No Alive No Yes
ofloxacin,
dipyrone
40/F/24 2007 79/yes 1 15.0 — Cephalosporin, No Alive No Yes
paracetamol,
ofloxacin
41/M/47 2007 73/no 2 6.0 CI TCM† No Alive No No

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

42/F/42 2007 56/yes 2 5.0 Hepatitis C, medul- Carbamazepine No Alive No Yes
litis
43/F/86 2007 100/yes 3 2.0 Gout TCM† No Lost No Yes
44/F/71 2007 91/yes 4 2.0 Hypertension Allopurinol No Lost No Yes
45/M/22 2008 97/yes 1 6.0 — Dipyrone Liver failure No Alive No Yes
46/M/62 2008 93/yes 2 5.0 Hypertension, CHD Cefuroxime Sepsis, pneumonia No Alive No Yes
47/F/58 2008 86/yes 2 2.0 — Methazolamide Respiratory failure, No Alive Yes Yes
pneumonia
48/M/65 2008 97/yes 2 6.0 Hypertension, Methazolamide No Lost No Yes
CHD, retinal
detachment,
cataract
49/F/28 2009 82/yes 1 4.0 Neuromyelitis optica Carbamazepine No Alive No Yes
50/M/15 2009 99/yes 1 16.0 Viral encephalitis, Levetiracetam, Pneumonia No Alive No Yes
epilepsy phenobarbital
51/F/46 2009 97/yes 2 16.0 Hepatitis B Cefuroxime, mez- MODS,pneumonia Yes, 33 Dead Yes No
locillin, TCM
52/F/56 2009 96/yes 2 5.0 — Cefoperazone No Alive No No
sodium
(Continued)
Zhu et al   e301
by guest

Table 2.  Continued

on 14 March 2018
Detachment
Patient Time of (%TBSA)/ Death at Outcome
No./Sex/ Admission, Mucosal Admission Discharge at Day Mechanical IVIG
Age (yr) (yr) Involvement SCORTEN Delay Medical History Previous Therapy Complication (d)* 120 Ventilation Therapy
e302   Zhu et al

53/M/46 2009 100/yes 2 1.0 — Methazolamide No Alive No Yes

54/F/30 2009 97/yes 2 3.0 — Miconazole nitrate Respiratory failure No Alive Yes Yes
supositories,
cephalosporin
55/M/54 2009 93/yes 2 2.0 Gout Allopurinol No Alive No Yes
56/M/42 2009 92/yes 3 6.0 CH, hypertension Sodium valproate, No Alive No Yes
phenobarbital,
cefoperazone,
vancomycin,
chlorpromazine
57/F/75 2009 96/yes 3 5.0 Hypertension Cefprozil, cephalo- Pneumonia No Lost No Yes
thin, tylenol
58/F/34 2009 93/yes 5 3.0 — Cephalothin, Pneumonia, renal No Lost No Yes
pazufloxa- failure, kidney
cin mesilate, failure
paracetamol,
59/M/24 2010 100/yes 1 1.0 Hepatitis B Amoxicillin, Pneumonia No Alive No Yes

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

cephalosporin,
paracetamol,
pseudoephed-
rine hydrochlo-
ride
60/F/34 2010 89/yes 3 2.0 — Methazolamide, No Alive No Yes
brinzolamide
61/M/60 2010 90/yes 4 4.0 Gout Allopurinol No Alive No yes

MODS, multiple organ dysfunction syndrome; CHD, coronary heart disease; CI, cerebral infarction; CH, cerebral hemorrhage; TB, tuberculosis; TCM, traditional Chinese medicine.
* The duration of days between the admission to our hospital and the deaths of the patients.
† The ingredients of traditional Chinese medicine (TCM) used in three of the seven patients were unknown. All four patients were administered a different type of TCM. Pugongying Pian (Coptis chinensis
Franch ‘黄连’, Gardenia jasminoides Ellis ‘栀子’, Forsythia suspensa (Thunb.) Vahl ‘连翘’, Vitex trifolia L. ‘蔓荆子’, Radix Saposhnikoviae ‘防风’, Herba Schizonepetae ‘荆芥穗’, Radix Angelicae Dahuricae ‘白芷’,
Scutellaria baicalensis Georgi ‘黄芩’, Herba Menthae ‘薄荷’, Radix et Rhizoma Rhei ‘大黄’, Phellodendron chinense Schneid. ‘黄柏’, Radix Platycodonis ‘桔梗’) and Qingrejiedu Chongji (Gypsum Fibrosum ‘石膏’,
Flos Lonicerae Japonicas ‘金银花’, Radix Scrophulariae ‘玄参’, Rehmannia glutinosa ‘地黄’, Forsythia suspensa ‘连翘’, Gardenia jasminoides Ellis ‘栀子’, Viola mandsurica ‘甜地丁’, Scutellaria baicalensis Georgi ‘黄芩’,
Gentiana scabra Bge ‘龙胆’, Radix Isatidis ‘板蓝根’, Rhizoma Anemarrhenae ‘知母’, Ophiopogon Japonicus ‘麦冬’) were used in two patients for the common cold respectively, and Tongfengding Pian (Gentiana
macrophylla Pall. ‘秦艽’, Cortex Phellodendri ‘黄柏’, Rhizoma Corydalis ‘延胡索’, Radix Paeoniae Rubra ‘赤芍’, Radix Cyathulae ‘川牛膝’, Rhizoma Alismatis ‘泽泻’, Semen Plantaginis ‘车前子’, Rhizoma Smilacis
Glabrae ‘土茯苓’)was used for another patient for gout. The fourth patient was treated with Xingnaojing Pian (Moschus berezovskii Flerov ‘麝香’, Dryobalanopsaromatica Gwaertn.f. ‘冰片’, Gardenia jasminoides
Ellis ‘栀子’, Curcuma wenyujin ‘郁金’) for cerebral infarction.
November/December 2012
Journal of Burn Care & Research
 Journal of Burn Care & Research
Volume 33, Number 6 Zhu et al   e303

Table 3. Clinical characteristics and outcomes of TEN

patients from 2002 to 2010
Corticosteroid Combination
Therapy Therapy

No. of patients 16 39
Age, yr, mean ± SD 51 ± 16 (26–81) 44 ± 20 (15–86)
(range)*
Male/female* 9/7 21/18
Detachment 84.6 ± 17.8 91.7 ± 9.1
(% TBSA)*
SCORTEN (No.
of deaths/
No./SMR)*
 1 0/4/0.00 0/11/0.00
 2 1/7/117.10 0/13/0.00
 3 1/2/154.32 2/9/68.59
 4 1/1/160.51 1/3/53.50
 5 2/2/109.53 2/3/73.02
Admission 5.8 ± 5.0 (3.1–8.4) 4.6 ± 3.7 (3.4–5.8)
delay (d)* Figure 2. Receiver operating characteristic curve of all
Hospitalization 28.6 ± 29.9 22.6 ± 13.7 toxic epidermal necrolysis (TEN) patients from 2000 to
time (d)* (12.7–44.5) (18.2–27.0) 2010. Area under receiver operating characteristic curve
was 0.889 ± 0.053 (P = .000).
Time to tapering of 12.9 ± 7.3 10.5 ± 3.4
corticosteroid (d)* (8.6–17.1) (9.4–11.7) of survivors significantly ranged from 100.0 ± 0.0 to
Time to arrest of 12.3 ± 10.1 7.6 ± 2.7
0.0 ± 0.0% (P = .000; Figure 5). Additional analysis
progression since (6.2–18.4) (6.7–8.6)
showed that the survival rates of the scores below 4
admission, (d)†
Mortality* 31% (5/16) 13% (5/39)
were statistically higher than those of a score of 5
(P = .008, P = .000, P = .039).
TEN, toxic epidermal necrolysis; SMR, standardized mortality ratio. Patients Treated With IVIG Combined Therapy.
Descriptive data are presented as mean ± SD along with 95% confidence
The actual mortality of this group was 13% (5/39),
intervals if not stated otherwise.
* P value for all comparisons >.10. much lower than predicted (24%, SMR = 53.53). No
† P = .013. significant difference was found between the expected
and actual deaths according to the SCORTEN (P >
scores of 3 or higher (P = .024, P = .004, P = .000; P .2). The overall outcome was statistically different
= .036, P = .006, P = .000; Figure 3). The survival rate (P = .011), and the LR was 12.70 (P = .013; 95%
of a score of 3 was also statistically higher than that of a CI, 0.000–0.048). Compared with a score of 5, the
score of 5 (P = .014). In all, six patients were lost in the mortality rate of the scores below 3 was statistically
survival analysis. decreased (P = .033; P = .025). Good calibration (P
Patients Treated With Corticosteroid Therapy = .530) and excellent discri­minatory power (AUC =
Alone. The actual mortality was 23% (5/22), in 0.906 ± 0.054, P = .004) of the SCORTEN could also
agreement with the predicted one (24%, SMR = be found (Table 1 and Figure 6). Cumulative survival
93.13). Similarly, there was no statistical difference rates largely ranged from 100.0 ± 0.0 to 33.3 ± 27.2%
between the actual and the predicted number of (P = .004; Figure 7). The survival rates of the scores
deaths at each SCORTEN level (all P = 1). The high below 3 were statistically higher than the rate of a
P value (P = .833) indicated excellent calibration score of 5 (P = .003, P = .001). The probability of
(Table 1). The ROC curve presented good discri­ survival of a score of 2 was also statistically higher than
minatory power of the SCORTEN (AUC = that of a score of 4 (P = .037).
0.882 ± 0.091, P = .011; Figure 4). The outcome
of each score was statistically different (P = .037), Comparison of the Effect of Corticosteroid
and the overall LR was 10.489 (P = .033; 95% CI, and IVIG Combined Therapy (2002–2010)
0.000–0.077). Compared with a score of 5, the During the period from 2002 to 2010, IVIG combined
mortality rate was statistically lower for the scores of therapy showed a trend toward reducing the mortal-
1 and 2 (P = .048; P = .045). Cumulative proportions ity rate (13%, SMR = 53.53), whereas corticosteroid

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

by guest
on 14 March 2018
 Journal of Burn Care & Research
e304   Zhu et al November/December 2012

Figure 3.  Kaplan-Meier 120-day survival curves of all toxic epidermal necrolysis (TEN) patients according to the SCORTEN
from 2000 to 2010.

therapy suggested no such difference (31%, SMR = In addition, an increased trend of the overall cumula-
123.92) to patients. At each SCORTEN level of the tive survival rate was determined in patients treated
two groups, our results drew the same conclusion with IVIG combined therapy (P = .002; Figure 8).
although no significant difference was found (Table 1). Further analysis of the score of 5 revealed a marked
increase of the survival rate in the combination therapy
group (P = 3.10 × 10−7). Table 3 reports the clinical
outcomes of the two therapy groups. In the combina-
tion therapy group, some decrease was observed both
in the hospitalization time (P = .310) and in the time
to tapering of corticosteroid (P = .131). Moreover,
disease progression was arrested significantly more
quickly in patients treated with the combination ther-
apy (P = .013).

DISCUSSION
The SCORTEN plays an important role in predict-
ing the prognosis of TEN patients. It can be easily
and quickly calculated on the basis of seven param-
eters within the first 24 hours after admission.5,6 As it
was developed in France, it is still in debate whether
the score is applicable to populations in other coun-
tries. Assessments in the United States, Korea, Can-
Figure 4. Receiver operating characteristic curve of the ada, United Kingdom, and Taiwan revealed positive
toxic epidermal necrolysis (TEN) patients receiving corti- results.9,34,10,35,36 In contrast, some authors observed
costeroid therapy from 2000 to 2010. Area under receiver considerable inaccuracy especially in severely affected
operating characteristic curve was 0.882 ± 0.091 (P = .011). patients.37–39 In addition, they doubted its general

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

by guest
on 14 March 2018
 Journal of Burn Care & Research
Volume 33, Number 6 Zhu et al   e305

Figure 5.  Kaplan-Meier 120-day survival curve of the toxic epidermal necrolysis (TEN) patients receiving corticosteroid
therapy according to the SCORTEN from 2000 to 2010. Cumulative proportions of survivors significantly ranged from
1.000 ± 0.000 to 0.000 ± 0.000% (P = .000). Cumulative survival rates of the scores below 4 were statistically higher than
those of a score of 5 (P = .008, P = .000, P = .039).

applicability in other countries for presumably differ- ability of SCORTEN in a section of the Asian popu-
ent treatment protocols.14 As information about the lation. No matter what therapies patients received,
validation of the score in Asia is inadequate, one aim the result indicated that this score could offer accu-
of this analysis was therefore to assess the predictive rate information about prognosis consistent with the
high discriminatory power (more than 80%) and the
good calibration (P > .5, especially in the corticoste-
roid therapy group) that was observed. Moreover,
the SCORTEN retained its predictive power even in
cases of late deaths (after more than 15 days of hos-
pitalization) in terms of the survival analysis. Review
of the results suggested that the outcomes of the
scores below 3 were much better than of the score
of 3 or higher, in particular when compared with a
score of 5. We conclude that although the gap in the
therapies exists, the score’s validity is not altered in
Chinese patients with TEN. The score could provide
clinicians with an objective assessment of mortal-
ity that would help them when discussing patients’
prognosis with family members or medical staff.5,6 It
is also useful in formulating therapeutic proposals in
the countries lack of medical resources such as IVIG.
In addition, evaluation of new treatment is more
reliable if investigators are able to define the patients’
Figure 6. Receiver operating characteristic curve of the severity-of-illness. The low incidence of TEN limits
toxic epidermal necrolysis (TEN) patients receiving intra- the collection of very large databases. To obtain suf-
venous immunoglobulin (IVIG) combined therapy from ficient numbers and proceed with the analysis, we
2000 to 2010. Area under receiver operating characteristic included 61 patients during a period of 10 years.
curve was 0.906 ± 0.054 (P = .004). During this long period, however, systemic therapy

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

by guest
on 14 March 2018
 Journal of Burn Care & Research
e306   Zhu et al November/December 2012

Figure 7.  Kaplan-Meier 120-day survival curve of the toxic epidermal necrolysis (TEN) patients receiving combination ther-
apy according to the SCORTEN from 2000 to 2010. Cumulative proportions of survivors largely ranged from 1.000 ± 0.000
to 0.333 ± 0.272% (P = .004). Cumulative survival rates of the scores below 3 were statistically higher than the rate of a score
of 5 (P = .003, P = .001). The rate of a score of 2 was also statistically higher than that of a score of 4 (P = .037).

has changed. The SCORTEN has thus been used attention as it blocks in vitro interaction of the Fas
to evaluate the efficacy of new therapeutic agents receptor with the Fas ligand and prevents kerati-
such as IVIG and plasmapheresis.13,15 With clarifica- nocyte death.23 Initial reports of the use of IVIG
tion of the mechanism, IVIG has gradually attracted were favorable with lower mortality than reported

Figure 8.  Kaplan-Meier 120-day survival curves of the patients receiving corticosteroid therapy and intravenous immuno-
globulin (IVIG) combined therapy from 2002 to 2010.

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

by guest
on 14 March 2018
 Journal of Burn Care & Research
Volume 33, Number 6 Zhu et al   e307

in the previous literature.13,24–26 However, replica- loss of follow-up in our database is 9.84%. Around
tion of those results has been difficult, and several half of the loss occurred 60 days after admission. We
studies have shown no benefit for the course or the deduced that it would not affect the results signifi-
prognosis.15,27,28 Two European studies drew con- cantly on account of the acute course of TEN. The
tradictory conclusions, although they were both survival curve also suggested that the incidence of
performed in the burn care units.10,14 In China also, death approximately occurred within 40 days and
combination therapy (IVIG combined with cortico- reflected a steady trend after that.
steroid) is replacing systemic corticosteroid therapy One limitation of this study is the small sample
alone, which has been routinely used for TEN.18,40 size, which may have led to a statistically insignifi-
Our hospital has the experience of treating TEN for cant difference in the mortality rate. As a retrospec-
more than 50 years although the treatments and care tive study, the conditions of the patients may be
protocols were not standardized until 1993. IVIG heterogeneous because of the relatively long time.
was introduced to our TEN patients after its rou- Without SCORTEN 6 or 7 cases, the applicability of
tine institution in 2002. Based on our many years our results to seriously ill patients is still unknown.
of experience, the role of systemic corticosteroids Additional, large, randomized, placebo-controlled
in halting the progression of TEN cannot be easily trials are required to verify and extend these results.
replaced by other medications, such as cyclophos-
phamide or cyclosporine. We are fully aware of the
ACKNOWLEDGMENTS
possible adverse effects of systemic corticosteroids
and therefore taper the doses promptly whenever We thank the Department of Dermatology of
appropriate.28 Thus, another focus of our study was Huashan Hospital for offering clinical data.
to explore whether combination therapy is superior
to corticosteroid therapy alone. Despite no signifi-
REFERENCES
cant difference of the SCORTEN being found in
the two groups, the indication for combination 1. Wolkenstein P, Roujeau JC. Epidemiology of toxic epider-
mal necrolysis. European J Dermatol 1994;4:175–80.
therapy determined that this group included more 2. Rzany B, Mockenhaupt M, Baur S, et al. Epidemiology of
patients with severely progressive diseases than the erythema exsudativum multiforme majus, Stevens-John-
group treated with corticosteroid alone. Probably son syndrome, and toxic epidermal necrolysis in Germany
(1990–1992): structure and results of a population-based
because of the limited population, neither of the two registry. J Clin Epidemiol 1996;49:769–73.
therapies showed statistical benefit on actual mortal- 3. Bastuji-Garin S, Rzany B, Stern RS, Shear NH, Naldi L,
ity, as expected. Nevertheless, combination therapy Roujeau JC. Clinical classification of cases of toxic epidermal
necrolysis, Stevens-Johnson syndrome, and erythema multi-
appeared to exhibit a tendency to reduce the mor- forme. Arch Dermatol 1993;129:92–6.
tality rate, whereas corticosteroid therapy suggested 4. Roujeau JC. Clinical heterogeneity of drug hypersensitivity.
no such difference. Referring to survival analysis, Toxicology 2005;209:123–9.
5. Bastuji-Garin S, Fouchard N, Bertocchi M, Roujeau JC,
the cumulative survival rate of each score was higher Revuz J, Wolkenstein P. SCORTEN: a severity-of-illness
in combination therapy (P = .002), especially to the score for toxic epidermal necrolysis. J Invest Dermatol
score of 5 (P = 3.10 × 10−7). We additionally evalu- 2000;115:149–53.
6. Guegan S, Bastuji-Garin S, Poszepczynska-Guigne E, et al.
ated the parameters reflecting the clinical outcomes Performances of the SCORTEN during the first five days of
of the two therapies. Compared with corticosteroid hospitalisation. J Invest Dermatol 2004;123:272–6.
alone, combination therapy arrested progression 7. Collin B, Marshall T, Heagerty A. Scoring with SCORTEN.
Arch Dermatol 2003;139:1215; author reply 1215–16.
significantly earlier (P = .013). This implied that 8. Trent JT, Kirsner RS, Romanelli P, et al. Scoring with
combination therapy might achieve a better thera- SCORTEN—Reply. Arch Dermatol 2003;139:1215–16.
peutic effect than the administration of corticoste- 9. Trent JT, Kirsner RS, Romanelli P, Kerdel FA. Use of
SCORTEN to accurately predict mortality in patients with
roid alone, especially in critically ill TEN patients. toxic epidermal necrolysis in the United States. Arch Derma-
Considering the high expense and limited supply of tol 2004;140:890–2.
IVIG in developing countries, this evidence might 10. Cartotto R, Mayich M, Nickerson D, Gomez M. SCORTEN
accurately predicts mortality among toxic epidermal necrol-
be useful in early administration of IVIG in critically ysis patients treated in a burn center. J Burn Care Res
ill TEN patients. 2008;29:141–6.
Attention should be given to the observed mor- 11. Sekula P, Liss Y, Davidovici B, et al. Evaluation of SCORTEN
on a cohort of patients with Stevens-Johnson syndrome and
tality of our study (16%). As the actual mortality of toxic epidermal necrolysis included in the RegiSCAR study.
the corticosteroid group was consistent with those J Burn Care Res 2011;32:237–45.
previously reported, the low mortality may prob- 12. Lissia M, Figus A, Rubino C. Intravenous immunoglobulins
and plasmapheresis combined treatment in patients with se-
ably be attributed to markedly decreased mortality vere toxic epidermal necrolysis: preliminary report. Br J Plast
of combination therapy (13%). The incidence of Surg 2005;58:504–10.

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

by guest
on 14 March 2018
 Journal of Burn Care & Research
e308   Zhu et al November/December 2012

13. Trent JT, Kirsner RS, Romanelli P, Kerdel FA. Analysis of immunoglobulins: our experience. Dermatology (Basel)
intravenous immunoglobulin for the treatment of toxic 2001;203:45–9.
epidermal necrolysis using SCORTEN: The University of 27. Brown KM, Silver GM, Halerz M, Walaszek P, Sandroni A,
Miami Experience. Arch Dermatol 2003;139:39–43. Gamelli RL. Toxic epidermal necrolysis: does immunoglob-
14. Imahara SD, Holmes JH 4th, Heimbach DM, et al.
ulin make a difference? J Burn Care Rehabil 2004;25:81–8.
SCORTEN overestimates mortality in the setting of 28. Sekula P, Schneck J, Fagot JP, et al. Effects of treatments on
a standardized treatment protocol. J Burn Care Res the mortality of Stevens-Johnson syndrome and toxic epi-
2006;27:270–5. dermal necrolysis: a retrospective study on patients included
15. Bachot N, Revuz J, Roujeau JC. Intravenous immuno-
in the prospective EuroSCAR study. J Am Acad Dermatol
globulin treatment for Stevens-Johnson syndrome and toxic 2008;58:33–40.
epidermal necrolysis: a prospective noncomparative study 29. The Acute Respiratory Distress Syndrome Network. Ventila-
showing no benefit on mortality or progression. Arch Der- tion with lower tidal volumes as compared with traditional
matol 2003;139:33–6. tidal volumes for acute lung injury and the acute respiratory
16. Quinn AM, Brown K, Bonish BK, et al. Uncovering histo- distress syndrome. N Engl J Med 2000;342:1301–8.
logic criteria with prognostic significance in toxic epidermal 30. Kaplan EL, Meier P. Nonparametric estimation from incom-
necrolysis. Arch Dermatol 2005;141:683–7. plete observations. J Am Stat Assoc 1958;53:457–81.
17. Oplatek A, Brown K, Sen S, Halerz M, Supple K, Gamelli 31. Liddell FD. Simple exact analysis of the standardised mortal-
RL. Long-term follow-up of patients treated for toxic epi- ity ratio. J Epidemiol Community Health 1984;38:85–8.
dermal necrolysis. J Burn Care Res 2006;27:26–33. 32. Hanley JA, McNeil BJ. The meaning and use of the area
18. Xu JH, Yang YS, Li F, et al. Combination therapy of intrave- under a receiver operating characteristic (ROC) curve. Radi-
nous immunoglobulin and corticosteroid in the treatment of ology 1982;143:29–36.
toxic epidermal necrolysis and Stevens-Johnson syndrome: 33. Lemeshow S, Hosmer DW Jr. A review of goodness of fit
a retrospective comparative study in China. Int J Dermatol statistics for use in the development of logistic regression
2009;48:1122–8. models. Am J Epidemiol 1982;115:92–106.
19. Halebian PH, Corder VJ, Madden MR, Finklestein JL,
34. Kim KJ, Lee DP, Suh HS, et al. Toxic epidermal necrolysis:
Shires GT. Improved burn center survival of patients with analysis of clinical course and SCORTEN-based comparison
toxic epidermal necrolysis managed without corticosteroids. of mortality rate and treatment modalities in Korean pa-
Ann Surg 1986;204:503–12. tients. Acta Derm Venereol 2005;85:497–502.
20. Schlingman J, Mockenhaupt M, Fagot JP, et al. Evaluation 35. Wu SC, Ho YL, Chang YT, et al. Performance of the

of treatment efficacy in a cohort of 281 patients with Ste- SCORTEN in Taiwanese patients with Stevens-Johnson syn-
vens-Johnson syndrome (SJS) or toxic epidermal necrolysis drome and toxic epidermal necrolysis. Dermatologica Sinica
(TEN). J Invest Dermatol 2004;123:367. 2010;28:15–20.
21. Criton S, Devi K, Sridevi PK, Asokan PU. Toxic epi-
36. George SM, Harrison DA, Welch CA, Nolan KM, Fried-
dermal necrolysis–a retrospective study. Int J Dermatol mann PS. Dermatological conditions in intensive care: a
1997;36:923–5. secondary analysis of the Intensive Care National Audit and
22. Wolff K, Tappeiner G. Treatment of toxic epidermal necroly- Research Centre (ICNARC) Case Mix Programme database.
sis: the uncertainty persists but the fog is dispersing. Arch Crit Care 2008;12 (Suppl 1):S1.
Dermatol 2003;139:85–6. 37. Spornraft-Ragaller P, Theilen H, Gottschlich G, et al. Treat-
23. Viard I, Wehrli P, Bullani R, et al. Inhibition of toxic epi- ment of toxic epidermal necrolyis. Experience with 9 pa-
dermal necrolysis by blockade of CD95 with human intrave- tients with consideration of intravenous immunoglobulin.
nous immunoglobulin. Science 1998;282:490–3. Hautarzt 2006;57:185–194.
24. Magina S, Lisboa C, Gonçalves E, Conceição F, Leal V, 38. Vaishampayan SS, Das AL, Verma R. SCORTEN: does it
Mesquita-Guimarães J. A case of toxic epidermal necroly- need modification? Indian J Dermatol Venereol Leprol
sis treated with intravenous immunoglobin. Br J Dermatol 2008;74:35–7.
2000;142:191–2. 39. Hague JS, Goulding JM, Long TM, Gee BC. Respiratory
25. Prins C, Kerdel FA, Padilla RS, et al.; TEN-IVIG Study involvement in toxic epidermal necrolysis portends a poor
Group. Toxic epidermal necrolysis-intravenous immuno- prognosis that may not be reflected in SCORTEN. Br J Der-
globulin. Treatment of toxic epidermal necrolysis with matol 2007;157:1294–6.
high-dose intravenous immunoglobulins: multicenter ret- 40. Chen J, Wang B, Zeng Y, Xu H. High-dose intravenous
rospective analysis of 48 consecutive cases. Arch Dermatol immunoglobulins in the treatment of Stevens-Johnson
2003;139:26–32. syndrome and toxic epidermal necrolysis in Chinese pa-
26. Stella M, Cassano P, Bollero D, Clemente A, Giorio G. Toxic tients: a retrospective study of 82 cases. Eur J Dermatol
epidermal necrolysis treated with intravenous high-dose 2010;20:743–7.

Downloaded from https://academic.oup.com/jbcr/article-abstract/33/6/e295/4588303

by guest
on 14 March 2018