FROM THE DERMATOLOGY FOUNDATION

Pediatric Stevens-Johnson syndrome

and toxic epidermal necrolysis in the
United States
Derek Y. Hsu, BA,a Joaquin Brieva, MD,a Nanette B. Silverberg, MD,d Amy S. Paller, MD, MS,a,b
and Jonathan I. Silverberg, MD, PhD, MPHa,c
Chicago, Illinois, and New York, New York

Background: Little is known about the epidemiology of Stevens-Johnson syndrome (SJS) and toxic
epidermal necrolysis (TEN) in children.

Objective: We sought to determine the morbidity, mortality, and comorbid health conditions of SJS and
TEN in US children.

Methods: This was a cross-sectional study of the 2009 to 2012 Nationwide Inpatient Sample, which
contains a representative 20% sample of all US hospitalizations. Sociodemographics, inflation-adjusted cost,
length of stay, comorbidities, and mortality were analyzed using descriptive statistics and multivariate
regression analyses.

Results: The incidences of SJS, SJS-TEN, and TEN were a mean 5.3, 0.8, and 0.4 cases per million children
per year in the US, respectively. Prolonged length of stay and higher costs of care (SJS: 9.4 6 0.6 days,
$24,947 6 $3171; SJS-TEN: 15.7 6 1.5 days, $63,787 6 $8014; TEN: 20.4 6 6.3 days, $102,243 6 $37,588)
were observed compared with all other admissions (4.6 6 0.1 days, $10,496 6 $424). Mortality was 0% for
SJS, 4% for SJS-TEN, and 16% for TEN. In regression models, predictors of mortality included renal failure
(adjusted OR [aOR] 300.28, 95% confidence interval [CI] 48.59-[999.99), malignancy (aOR 54.33, 95% CI
9.40-314.22), septicemia (aOR 30.45, 95% CI 7.91-117.19), bacterial infection (aOR 20.38, 95% CI 5.44-76.36),
and epilepsy (aOR 5.56, 95% CI 1.37-26.2).

Limitations: Data regarding treatment were not available. Date of diagnosis of comorbidities was not
present, precluding temporal analysis.

Conclusions: Pediatric SJS/TEN poses a substantial health burden in the United States. ( J Am Acad
Dermatol http://dx.doi.org/10.1016/j.jaad.2016.12.024.)

Key words: burden of disease; children; epidemiology; hospitalization; pediatric; Stevens-Johnson

syndrome; toxic epidermal necrolysis.

S tevens-Johnson syndrome (SJS) and toxic
epidermal necrolysis (TEN) are potentially
life-threatening disorders characterized by
widespread epidermal necrosis of skin and mucosa.
Classification is based on body surface area (BSA)
affected, with SJS involving less than 10%, SJS-TEN
10% to 30%, and TEN greater than 30% of BSA.1
Incidences of SJS and TEN in the general population
are reported to be 1 to 7 and 1 to 2 cases per million,
respectively.2-5 Little is known about the

From the Departments of Dermatology,a Pediatrics,b and Preven-
tive Medicine and Medical Social Sciences,c Northwestern
University Feinberg School of Medicine, Chicago; and Depart-
ment of Dermatology, Mount-Sinai Icahn School of Medicine,
New York.d
This publication was made possible with support from the Agency
for Healthcare Research and Quality, grant number
K12HS023011, the Dermatology Foundation.
Conflicts of interest: None declared.
Accepted for publication December 15, 2016.
Reprint requests: Jonathan I. Silverberg, MD, PhD, MPH,
Department of Dermatology, Northwestern University
Feinberg School of Medicine, 676 N St Clair St, Suite 1600,
Chicago, IL 60611. E-mail: JonathanISilverberg@gmail.com.
Published online March 9, 2017.
0190-9622/$36.00
Ó 2016 by the American Academy of Dermatology, Inc.
http://dx.doi.org/10.1016/j.jaad.2016.12.024

1
2 Hsu et al J AM ACAD DERMATOL
n 2017

epidemiology of pediatric SJS/TEN. Previous studies erythema multiforme major (ICD-9-CM: 695.12),
in the United States have been limited to small Kawasaki disease (ICD-9-CM: 446.1), herpes simplex
numbers of cases derived from only a few centers. virus infection (ICD-9-CM: 054.XX), and Mycoplasma
Reports of mortality for SJS-TEN in adults are as pneumoniae infection (ICD-9-CM: 041.81 or 483.0)
high as 34%,6-8 but as low 0% to 3.6%9,10 and 0% to were excluded. No patient had a concomitant diag-
7%11,12 in pediatric SJS and TEN, respectively. The nosis of graft-versus-host disease (GVHD). Newborn
burden of SJS and TEN in children on the health care infants were excluded from the control group.
system in terms of cost and A comorbidity was consid-
length of stay (LOS) in the ered present if listed as a
United States has not been CAPSULE SUMMARY diagnosis when the patient
quantified. We sought to also had a diagnosis of SJS,
determine risk factors, co-
d Stevens-Johnson syndrome and toxic
SJS-TEN, or TEN. Coding of
morbidities, inpatient epidermal necrolysis cause substantial
comorbidities used ICD-9-
burden, and mortality in US morbidity and mortality in children.
CM diagnostic codes.
children with SJS, SJS-TEN, d Body surface area of involvement greater Comorbidities analyzed
and TEN. than 30% and comorbid renal failure, were selected based on pre-
septicemia, epilepsy, bacterial infection, vious associations with SJS-
METHODS and malignancy were associated with an TEN in the literature or to
The 2009 to 2012 increased mortality risk. explore for novel associa-
Nationwide Inpatient d Pediatric Stevens-Johnson syndrome and tions. Odds ratios (OR) and
Sample (NIS) was analyzed. toxic epidermal necrolysis represent a 95% confidence intervals (CI)
The NIS is sponsored by the substantial health burden across the were plotted on forest plots
Healthcare Cost and United States with high costs, prolonged through Open Meta Analyst,
Utilization Project of the length of hospital stay, and potential an open-source software
Agency for Healthcare mortality. supported by Brown
Research and Quality University and the AHRQ.
(AHRQ). NIS contains an ICD-9-CM codes were also
approximately 20% stratified used to determine proced-
representative sample of all US hospitalizations. ures, including mechanical ventilation, skin grafts,
Sample weights were created by NIS to adjust for dialysis, and total parenteral nutrition. Inpatient
the complex sampling design. Cases were multiplied mortality was calculated. Mortality risk and loss of
by weights to provide more accurate estimates for function were determined by the All Patient Refined
overall frequency in the United States. All data were Diagnoses Related Group, which includes severity of
deidentified and no attempts were made to identify illness and risk of mortality subclasses.15
individuals in the database. Patient consent was not
obtained as the databases were received deidenti- Statistical analysis
fied. All parties with access to the Healthcare Cost All data processing and statistical analyses were
and Utilization Project were compliant to its formal performed using software (SAS, Version 9.4, SAS
data use agreement. The study did not require Institute, Cary, NC). Analyses were performed using
approval from the institutional review board. SURVEY procedures that adjusted for survey weight-
ing, sampling clusters, and strata. Weighted inci-
Exposures and outcomes dences of a primary and/or secondary diagnosis of
The databases were searched for a primary and/or SJS, SJS-TEN, and TEN were determined. Calculation
secondary discharge International Classification of of incidences excluded patients who were trans-
Diseases, Ninth Revision, Clinical Modification (ICD- ferred to another hospital. Costs were adjusted for
9-CM) code of 695.13, 695.14, and 695.15 correspond- inflation to the year 2014 according to the Consumer
ing to SJS, SJS-TEN, and TEN, respectively. Based on a Price Index from the US Bureau of Labor Statistics.16
previously validated algorithm, patients who had a Summary statistics were generated for LOS and
LOS less than 3 days were excluded to improve inflation-adjusted cost of care, including sum,
positive predictive value of diagnostic codes.13 mean, SD, minimum, maximum, and median.
Patients with a concomitant diagnosis of any other Associations of SJS-TEN, cost, LOS, and mortality
bullous dermatosis were excluded (ICD-9-CM: 694.0- were examined for the following variables: age in
694.9), because certain bullous dermatoses have years, gender, race/ethnicity, median annual income
rarely been reported to masquerade as TEN.14 In of the hospital ZIP code, health insurance coverage,
addition, patients with a concomitant diagnosis of number of chronic conditions, season of
J AM ACAD DERMATOL
VOLUME jj, NUMBER j
Abbreviations used:
AHRQ: Agency for Healthcare Research and
Quality
aOR: adjusted odds ratio
BSA: body surface area
CI: confidence interval
GVHD: graft-versus-host disease
ICD-9-CM: International Classification of Dis-
eases, Ninth Revision, Clinical
Modification
LOS: length of stay
NIS: Nationwide Inpatient Sample
OR: odds ratio
SJS: Stevens-Johnson syndrome
TEN: toxic epidermal necrolysis
hospitalization, hospital location, teaching status and
bed size (Supplemental Table I; available at http://
www.jaad.org), and patient disposition at discharge.
Multivariate survey logistic regression models were
constructed and the control group included all
pediatric patients without a diagnosis of SJS, SJS-
TEN, or TEN, yielding a representative cohort of
hospitalized pediatric patients. To determine the
predictors of inpatient mortality, multivariate logistic
regression models with stepwise selection were
constructed with mortality as the dependent vari-
able. All sufficiently powered comorbidities were
included in the mortality selection model. Analyses
of mortality only included pediatric patients with SJS,
SJS-TEN, and TEN. Logistic regression was used to
compare incidence and mortality for SJS, SJS-TEN, or
TEN between admission years. Complete case anal-
ysis was performed. A 2-sided P value less than .05
was considered statistically significant.
RESULTS
Overall, 1,687,172 pediatric admissions excluding
healthy newborns were captured in the NIS in 2009
to 2012. Weighted frequencies, which more accu-
rately represent the full spectrum of US hospitaliza-
tions, were 1210 and 369 for a primary and
secondary diagnosis of SJS, 192 and 59 for SJS-TEN,
and 64 and 65 for TEN, respectively (Table I).
Estimated frequencies per million children ranged
from 4.3 to 5.8 for SJS, 0.6 to 1.4 for SJS-TEN, and 0 to
0.7 for TEN. The frequency of SJS and SJS-TEN were
stable between 2009 and 2012 (P [.05). There was a
significant decrease in the frequency of TEN in 2011
(P = .01). Average age was 10.5 6 0.2, 10.4 6 0.7, and
9.3 6 0.7 years for SJS, SJS-TEN, and TEN, respec-
tively. The demographics of children with or without
SJS/TEN are presented in Supplemental Table II
(available at http://www.jaad.org).
The strongest shared predictor of SJS, SJS-TEN,
and TEN was increasing age (linear regression;
Hsu et al 3
P \ .05 for all) (Supplemental Table III; available at
http://www.jaad.org). SJS and SJS-TEN were also
associated with the number of chronic conditions
present.
Comorbidities
In multivariate survey logistic regression models
controlling for sociodemographic factors, both SJS
and SJS-TEN/TEN were associated with ocular infec-
tion or inflammation (SJS: adjusted OR [aOR] 84.31,
95% CI 65.00-109.35; SJS-TEN/TEN: aOR 74.78, 95%
CI 43.49-128.57), affecting 38.7% and 36.8% of the
respective cohorts (Fig 1). Both SJS (aOR 2.99, 95%
CI 1.37-6.51) and SJS-TEN/TEN (aOR 13.90, 95% CI
5.39-35.89) were associated with blindness or vision
defects, affecting 2.3% and 7.9%, respectively.
SJS was associated with any viral infection (aOR
4.42, 95% CI 2.99-6.52), bacterial infection (aOR 2.62,
95% CI 1.84-3.73), fungal infection (aOR 4.32, 95% CI
2.28-7.83), and pneumonia (aOR 3.00, 95% CI 2.17-
4.16). Notably, SJS was not associated with renal
failure or septicemia. SJS-TEN and TEN showed
associations with any fungal, viral, or bacterial
infections, pneumonia and epilepsy. SJS-TEN and
TEN were also associated with septicemia (aOR
11.69, 95% CI 6.34-21.54), renal failure (aOR 7.63,
95% CI 2.71-21.54), and liver damage (aOR 3.47, 95%
CI 1.26-9.58).
Hospital course
The mean 6 SE LOS for patients with SJS, SJS-TEN,
and TEN was 9.4 6 0.6, 15.7 6 1.5, and
20.4 6 6.3 days compared with only 4.6 6 0.1 days
in children without these disorders, respectively.
One in 10 children with SJS, SJS-TEN, and TEN
underwent mechanical ventilation (aOR 10.8%,
95% CI 7.7-13.9) (Table I). Patients were most
frequently routinely discharged to home or other
self-care (aOR 89.5%, 95% CI 86.8-92.2%).
Cost
The mean 6 SE inflation-adjusted cost of hospi-
talization for children with SJS, SJS-TEN, and TEN
was $24,947 6 $3171, $63,787 6 $8014, and
$102,243 6 $37,588, respectively, compared with
$10,496 6 $424 for the control group. The cumula-
tive cost of care from 2009 to 2012 was $36,461,811,
$15,014,602, and $11,695,400 for all patients with SJS,
SJS-TEN, and TEN.
Mortality
Mortality was 0% for SJS, 4% for SJS/TEN, and
16% for TEN. Mortality for SJS-TEN or TEN did not
significantly change over time (P [.05), although in
2011 there were no observed cases of childhood
4 Hsu et al J AM ACAD DERMATOL
n 2017

Table I. Hospital course and disposition of pediatric patients with Stevens-Johnson syndrome, Stevens-
Johnson syndrome-toxic epidermal necrolysis, or toxic epidermal necrolysis
SJS SJS-TEN TEN
Variable Frequency Prevalence [95% CI] Frequency Prevalence [95% CI] Frequency Prevalence [95% CI]
Ventilation
Yes 117 7.4% [4.5-10.3] 58 23.3% [11.7-34.9] 45 34.9% [23.6-46.1]
No 1461 92.6% [89.7-95.5] 192 76.7% [65.1-88.3] 84 65.1% [53.9-76.4]
Ventilation for \96 h
Yes 54 3.4% [1.4-5.4] 14 5.4% [0-13.6] 10 7.7% [0-20.2]
No 1524 96.6% [94.6-98.6] 237 94.6% [86.4-100.0] 119 92.3% [79.8-100]
Ventilation for [96 h
Yes 67 4.3% [2.1-6.5] 45 17.9% [8.1-27.7] 35 27.2% [7.7-46.9]
No 1511 95.7% [93.5-97.9] 206 82.1% [72.3-91.9] 94 72.8% [53.3-92.3]
Skin graft
Yes 14 0.9% [0.1-1.9] 20 8.1% [0-17.2] 10 7.6% [5.7-9.6]
No 1564 99.1% [98.1-99.9] 230 91.9% [82.8-100.0] 119 92.4% [90.4-94.3]
Skin debridement
Yes 39 2.5% [0.8-4.1] 21 8.6% [1.4-15.8] 20 16.5% [8.8-24.2]
No 1540 97.5% [95.9-99.2] 225 91.4% [84.2-98.6] 103 83.5% [75.8-91.2]
Physical therapy
Yes 30 1.8% [0.4-3.4] 20 8.2% [0-20.1] 0 0%
No 1549 98.2% [96.6-99.6] 230 91.8% [79.9-100.0] 129 100%
Dialysis
Yes 10 0.6% [0-1.5] 5 2.0% [0-6.3] 9 7.1% [0-18.0]
No 1568 99.4% [98.5-100.0] 245 98.0% [93.7-100.0] 119 92.9% [82.0-100]
PEG, NG tube, or TPN
Yes 238 15.5% [11.6-19.3] 82 32.8% [19.5-46.1] 25 19.6% [3.1-36.1]
No 1340 84.5% [80.7-88.4] 168 67.2% [53.9-80.5] 103 80.4% [63.9-96.9]
Mortality risk*
Not specified 11 0.7% [0-1.6] 0 0 0 0
Minor 1221 77.4% [72.8-82.0] 109 43.4% [33.7-53.1] 44 34.0% [20.7-47.2]
Moderate 205 13.0% [9.3-16.6] 84 33.5% [19.3-47.6] 15 11.4% [0-24.9]
Major 98 6.2% [3.5-8.9] 20 8.1% [0.6-15.5] 45 35.2% [24.3-46.1]
Extreme 44 2.8% [1.0-4.6] 38 15.1% [5.3-24.8] 25 19.4% [1.1-37.8]
Loss of function severity*
Not specified 11 0.7% [0-1.6] 0 0 0 0
Minor 415 26.3% [21.5-31.1] 10 4.1% [0-9.0] 25 19.1% [8.0-30.1]
Moderate 717 45.5% [40.0-50.9] 103 41.3% [31.9-50.8] 19 14.8% [0.5-29.1]
Major 332 21.1% [16.6-25.6] 77 30.8% [18.3-43.3] 44 34.4% [11.7-57.1]
Extreme 102 6.5% [3.8-9.2] 60 23.8% [13.5-34.1] 41 31.7% [0-71.6]
Patient disposition at discharge
Routine 1463 92.7% [89.8-95.5] 198 78.9% [69.8-88.0] 85 65.9% [54.7-77.0]
Transfer to short-term 52 3.3% [1.4-5.2] 5 2.0% [0-6.3] 5 3.5% [0-13.4]
hospital
Transfer to SNF, ICF, or other 26 1.6% [0.2-3.1] 15 5.9% [0.9-10.9] 5 4.2% [0-14.8]
facility
Home health care 38 2.4% [0.7-4.1] 23 9.3% [2.2-16.3] 15 11.3% [0-23.5]
Left against medical advice 0 0 0 0 0 0
Died 0 0 10 4.0% [0-9.7] 19 16.0% [12.4-17.9]

Weighted frequencies of procedures (ventilation, skin grafts, physical therapy, dialysis, feeding), mortality risk and loss of function estimates,
and patient disposition at discharge by SJS, SJS-TEN, and TEN.
CI, Confidence interval; ICF, intermediate care facility; NG, nasogastric; PEG, percutaneous endoscopic gastrostomy; SJS, Stevens-Johnson
syndrome; SNF, subacute nursing facility; TEN, toxic epidermal necrolysis; TPN, total parenteral nutrition.
*All Patient Refined Diagnosis Related Group Mortality Risk and Loss of Function Severity are calculated based on methodology by 3M
(St. Paul, Minnesota) to allow analysis of severity and mortality risk in large cohorts given discharge diagnoses, pre-existing medical
conditions, and age [http://archive.ahrq.gov/professionals/quality-patient-safety/quality-resources/tools/mortality/Hughessumm.pdf].
J AM ACAD DERMATOL Hsu et al 5
VOLUME jj, NUMBER j

Fig 1. Association between Stevens-Johnson syndrome (SJS ) or SJS-toxic epidermal necrolysis

(TEN )/TEN and health comorbidities. Survey logistic regression models were constructed
with SJS or SJS-TEN/TEN as the independent variable and the respective comorbidity as
the dependent variable. Models included race/ethnicity, age, and sex as covariates.
Adjusted OR and 95% CI were estimated. Forest plots of the adjusted OR and 95% CI are
presented.

TEN. In logistic regression models with stepwise
selection controlling for all sufficiently powered
comorbidities, predictors of mortality included renal
failure (aOR 300.28, 95% CI 48.59-[999.99), malig-
nancy (aOR 54.33, 95% CI 9.40-314.22), septicemia
(aOR 30.45, 95% CI 7.91-117.19), any bacterial
infection (aOR 20.38, 95% CI 5.44-76.36), and epi-
lepsy (aOR 5.56, 95% CI 1.37-26.20) (Table II).
DISCUSSION
This study found that childhood SJS, SJS-TEN, and
TEN caused a significant health care burden, with a
mean cost of hospitalization of [$20,000, [$50,000,
and [$100,000, respectively, and a total 4-year cost
of $63,001,951 (mean $16.3 million per year).
Children with SJS-TEN also had up to a 4-fold longer
duration and 9-fold higher mean cost of hospitaliza-
tion in comparison with other pediatric admissions.
The incidences of SJS, SJS-TEN, and TEN appear to
be lower in US children (5.3, 0.8, and 0.4 cases per
million children per year) than in adults (9.3, 1.9, and
1.6 cases per million adults per year).6 The estimated
incidence of pediatric SJS was similar to previous
reports in the general population (1-7 cases per
million), but slightly lower for TEN (1-2 cases per
million).2-6 A previous study of 37 pediatric patients
with SJS-TEN calculated incidences of erythema
multiforme, SJS, and TEN in patients younger than
20 years to be 7.0 per million person-years.17 A
higher number of chronic health conditions were
associated with SJS and SJS-TEN, suggesting that
underlying comorbidities and/or concomitant poly-
pharmacy contribute to pediatric SJS-TEN. Eye infec-
tion/inflammation manifested in more than a third
of children with SJS/TEN, within the range of previ-
ous reports of ocular involvement in SJS/TEN
(18.8%-84%).18,19 The rate could be higher as ocular
involvement in SJS/TEN may not be coded with
separate International Classification of Diseases
codes. Although we previously observed racial
disparities in the incidence and costs of SJS/TEN in
adults, these disparities were not seen in children.
Mortality in children with SJS (0%), SJS-TEN
(4.0%), and TEN (16.0%) was similar to previous
reports from smaller cohorts (SJS: 0%-3.6%, TEN: 0%-
16.7%).9,11,12,20 A previous study found that children
6 Hsu et al J AM ACAD DERMATOL
n 2017

Table II. Predictors of mortality in pediatric may have triggered SJS-TEN. Antiepileptics are
patients with Stevens-Johnson syndrome, Stevens- implicated as one of the most common causes of
Johnson syndrome-toxic epidermal necrolysis, or SJS-TEN and TEN,23 although epilepsy and antiepi-
toxic epidermal necrolysis leptics have rarely been associated with mortality.
Variable Adjusted OR [95% CI] P
Strengths of this study include an analysis of a
nationally representative sample of all-payer data
Renal failure 300.28 [48.59-[999.99] \.0001
over a period of 4 years with over 4.5 million records
Septicemia 30.45 [7.91-117.19] \.0001
Epilepsy 5.56 [1.37-26.20] \.0001 that included several hundred cases of SJS, SJS-TEN,
Any bacterial infection 20.38 [5.44-76.36] \.0001 and TEN and using a validated algorithm for case
Malignancy 54.33 [9.40-314.22] \.0001 selection. Virtually all patients present for care and
Mycoplasma - - are admitted to the hospital for evaluation and
Fungal - - management and, thus, this inpatient cohort is likely
Herpes simplex virus - - sufficient to describe the epidemiology of these
Pneumonia - - disorders. Limitations of our study include the
Eye infection or - - inability to perform temporal analysis of comorbid-
inflammation ities, as date of diagnosis was not documented. LOS
Infectious hepatitis - -
may have been overestimated, because we excluded
Liver disease or damage - -
patients with LOS less than 3 days in our case
Influenza - -
Any skin infection - - definition. We were unable to assess medications
Cellulitis - - administered preceding the diagnosis of SJS-TEN or
Bipolar disorder - - for the treatment of SJS-TEN. Thus, we were unable
Chronic kidney disease - - to determine whether SJS-TEN is related to a partic-
ular disease or its treatment. We were also unable to
Predictors of mortality, presented as adjusted odds ratios using analyze whether treatment affects mortality.
logistic regression with stepwise selection controlling for all
Identification of comorbidities was performed using
sufficiently powered comorbidities. Malignancies were combined
into 1 variable. ICD-9-CM codes and not by review of the health
CI, Confidence interval; OR, odds ratio. record. Given that this is a deidentified national
database, we were unable to verify diagnoses of SJS/
TEN. We explored for a concomitant diagnosis of
with 60% to 70% BSA had a mortality of 37.5%.20 hematopoietic stem cell transplantation and SJS/TEN
Thus, it appears that increasing BSA involvement and only found 1 patient with malignancy who
correlates with increased mortality in children. In received a bone-marrow transplant. There were no
contrast, US adults with SJS-TEN and TEN had similar cases of GVHD. Severe cutaneous GVHD can be
mortality,6 in agreement with the SCORTEN (SCORe difficult to distinguish from TEN,16,24 and it is
of Toxic Epidermal Necrosis) where BSA greater than possible that some patients given a diagnosis of a
10% is a predictor of mortality in adults with SJS- malignancy and SJS-TEN/TEN actually had GVHD.
TEN.21 Further studies are needed to confirm The absence of TEN in 2011 could be attributed to
whether BSA greater than 30% confers additional sampling, as the NIS is a 20% sample of all US
mortality risk than BSA of 10% to 29%. hospitalizations, and pediatric TEN is rare. Finally,
Many children with SJS-TEN and TEN died from some procedures may have been coded using
multiple comorbid health conditions and not their Current Procedural Terminology without corre-
skin disease per se. Renal failure, septicemia, bacte- sponding International Classification of Diseases
rial infection, liver damage, epilepsy, and malig- codes. This could result in underestimation of the
nancy were all associated with increased mortality in rates of procedures.
SJS-TEN and TEN. The association between mortality In conclusion, pediatric SJS/TEN is a substantial
and renal failure is consistent with the SCORTEN health burden in the United States with increased
(elevated serum blood urea nitrogen). The lack of costs, LOS, and mortality. Incidences of SJS, SJS-TEN,
association with diabetes and other disorders with and TEN were lower in children than observed in
elevated blood glucose is not consistent with adults. Mortality and procedures such as ventilation
SCORTEN. Septicemia was likely caused by impaired and dialysis were substantially higher in TEN than in
skin-barrier from denuded skin. Although not SJS or SJS-TEN, suggesting that BSA of involvement
included in SCORTEN, infectious complications greater than 30%enot just greater than 10%eis an
were previously found to be the most common additional mortality risk factor. Renal failure, septi-
cause of death in SJS/TEN.22 It is also possible that cemia, and bacterial infections were the strongest
antibiotics used for treatment of preceding infections predictors of mortality.
J AM ACAD DERMATOL
VOLUME jj, NUMBER j
REFERENCES
1. Bastuji-Garin S, Rzany B, Stern RS, Shear NH, Naldi L,
Roujeau JC. Clinical classification of cases of toxic epidermal
necrolysis, Stevens-Johnson syndrome, and erythema multi-
forme. Arch Dermatol. 1993;129:92-96.
2. Rzany B, Mockenhaupt M, Baur S, et al. Epidemiology of
erythema exsudativum multiforme majus, Stevens-Johnson
syndrome, and toxic epidermal necrolysis in Germany
(1990-1992): structure and results of a population-based
registry. J Clin Epidemiol. 1996;49:769-773.
3. Schopf E, Stuhmer A, Rzany B, Victor N, Zentgraf R, Kapp JF.
Toxic epidermal necrolysis and Stevens-Johnson syndrome.
An epidemiologic study from West Germany. Arch Dermatol.
1991;127:839-842.
4. Strom BL, Carson JL, Halpern AC, et al. A population-based
study of Stevens-Johnson syndrome. Incidence and ante-
cedent drug exposures. Arch Dermatol. 1991;127:831-838.
5. Roujeau JC, Guillaume JC, Fabre JP, Penso D, Flechet ML,
Girre JP. Toxic epidermal necrolysis (Lyell syndrome). Inci-
dence and drug etiology in France, 1981-1985. Arch Dermatol.
1990;126:37-42.
6. Hsu DY, Brieva J, Silverberg NB, Silverberg JI. Morbidity and
mortality of Stevens-Johnson syndrome and toxic epidermal
necrolysis in United States adults. J Invest Dermatol. 2016;136:
1387-1397.
7. Harr T, French LE. Toxic epidermal necrolysis and
Stevens-Johnson syndrome. Orphanet Journal of Rare Diseases.
2010;5:39.
8. Sekula P, Dunant A, Mockenhaupt M, et al. Comprehensive
survival analysis of a cohort of patients with Stevens-Johnson
syndrome and toxic epidermal necrolysis. J Invest Dermatol.
2013;133:1197-1204.
9. Finkelstein Y, Soon GS, Acuna P, et al. Recurrence and
outcomes of Stevens-Johnson syndrome and toxic epidermal
necrolysis in children. Pediatrics. 2011;128:723-728.
10. Lam NS, Yang YH, Wang LC, Lin YT, Chiang BL. Clinical
characteristics of childhood erythema multiforme,
Stevens-Johnson syndrome and toxic epidermal necrolysis in
Taiwanese children. J Microbiol Immunol Infect. 2004;37:366-370.
11. Forman R, Koren G, Shear NH. Erythema multiforme,
Stevens-Johnson syndrome and toxic epidermal necrolysis in
children: a review of 10 years’ experience. Drug Saf. 2002;25:
965-972.
12. Spies M, Sanford AP, Aili Low JF, Wolf SE, Herndon DN.
Treatment of extensive toxic epidermal necrolysis in children.
Pediatrics. 2001;108:1162-1168.
13. Davis RL, Gallagher MA, Asgari MM, et al. Identification of
Stevens-Johnson syndrome and toxic epidermal necrolysis in
Hsu et al 7
electronic health record databases. Pharmacoepidemiol Drug
Saf. 2015;24:684-692.
14. Kakar R, Paugh H, Jaworsky C. Linear IgA bullous disease
presenting as toxic epidermal necrolysis: a case report and
review of the literature. Dermatology (Basel, Switzerland). 2013;
227:209-213.
15. Metry DW, Lahart CJ, Farmer KL, Hebert AA. Stevens-Johnson
syndrome caused by the antiretroviral drug nevirapine. J Am
Acad Dermatol. 2001;44:354-357.
16. Macedo FI, Faris J, Lum LG, et al. Extensive toxic epidermal
necrolysis versus acute graft versus host disease after allo-
genic hematopoietic stem-cell transplantation: challenges in
diagnosis and management. J Burn Care Res. 2014;35:
e431-e435.
17. Chan HL, Stern RS, Arndt KA, et al. The incidence of erythema
multiforme, Stevens-Johnson syndrome, and toxic epidermal
necrolysis. A population-based study with particular reference
to reactions caused by drugs among outpatients. Arch
Dermatol. 1990;126:43-47.
18. Morales ME, Purdue GF, Verity SM, Arnoldo BD, Blomquist PH.
Ophthalmic manifestations of Stevens-Johnson syndrome and
toxic epidermal necrolysis and relation to SCORTEN. Am J
Ophthalmol. 2010;150:505-510.e1.
19. Chang YS, Huang FC, Tseng SH, Hsu CK, Ho CL, Sheu HM.
Erythema multiforme, Stevens-Johnson syndrome, and toxic
epidermal necrolysis: acute ocular manifestations, causes, and
management. Cornea. 2007;26:123-129.
20. Ferrandiz-Pulido C, Garcıa-Fernandez D, Domınguez-Sampedro P,
Garcıa-Patos V. Stevens-Johnson syndrome and toxic epidermal
necrolysis in children: a review of the experience with pediatric
patients in a university hospital. J Eur Acad Dermatol Venereol.
2011;25:1153-1159.
21. Bastuji-Garin S, Fouchard N, Bertocchi M, Roujeau JC,
Revuz J, Wolkenstein P. SCORTEN: a severity-of-illness score
for toxic epidermal necrolysis. J Invest Dermatol. 2000;115:
149-153.
22. Letko E, Papaliodis DN, Papaliodis GN, Daoud YJ, Ahmed AR,
Foster CS. Stevens-Johnson syndrome and toxic epidermal
necrolysis: a review of the literature. Ann Allergy Asthma
Immunol. 2005;94:419-436; quiz 436-438, 456.
23. Levi N, Bastuji-Garin S, Mockenhaupt M, et al. Medications as
risk factors of Stevens-Johnson syndrome and toxic epidermal
necrolysis in children: a pooled analysis. Pediatrics. 2009;123:
e297-e304.
24. Jeanmonod P, Hubbuch M, Grunhage F, et al. Graft-versus-h-
ost disease or toxic epidermal necrolysis: diagnostic dilemma
after liver transplantation. Transplant infectious disease. J
Transplant Soc. 2012;14:422-426.
7.e1 Hsu et al J AM ACAD DERMATOL
n 2017

Supplemental Table I. Bed size categories as

defined by Healthcare Cost and Utilization Project
Hospital bed size
Location and teaching status Small Medium Large
Northeast region
Rural 1-49 50-99 [100
Urban, nonteaching 1-124 125-199 [200
Urban, teaching 1-249 250-424 [425
Midwest region
Rural 1-29 30-49 [50
Urban, nonteaching 1-74 75-174 [175
Urban, teaching 1-249 250-374 [375
Southern region
Rural 1-39 40-74 [75
Urban, nonteaching 1-99 100-199 [200
Urban, teaching 1-249 250-449 [450
Western region
Rural 1-24 25-44 [45
Urban, nonteaching 1-99 100-174 [175
Urban, teaching 1-199 200-324 [325
Supplemental Table II. Demographics of pediatric patients with Stevens-Johnson syndrome, Stevens-Johnson syndrome-toxic epidermal necrolysis, or

VOLUME jj, NUMBER j

J AM ACAD DERMATOL
toxic epidermal necrolysis
No SJS, SJS-TEN, or TEN SJS SJS-TEN TEN
Variable Frequency Prevalence [95% CI] Frequency Prevalence [95% CI] Frequency Prevalence [95% CI] Frequency Prevalence [95% CI]
Age, y
0 1,945,254 24.3% [23.7-24.9] 15 0.9% [0-2.0] 5 2.0% [0-5.7] 15 11.6% [0-25.0]
1-2 1,286,122 16.1% [15.6-16.6] 120 7.7% [5.0-10.4] 29 11.4% [2.6-20.2] 24 18.8% [5.8-31.8]
3-4 580,666 7.3% [6.9-7.6] 78 5.0% [2.5-7.5] 14 5.6% [0-11.2] 5 3.7% [0-11.0]
5-11 1,595,501 20.0% [19.6-20.4] 572 36.7% [31.4-42.1] 74 29.5% [14.9-44.0] 20 15.5% [1.7-29.2]
12-17 2,583,033 32.3% [31.4-33.3] 774 49.7% [44.1-55.2] 129 51.6% [36.5-66.7] 65 50.5% [33.5-67.5]
Season
Winter 2,162,143 28.6% [28.3-28.9] 384 25.5% [20.4-30.7] 63 26.6% [14.6-38.6] 19 15.0% [0.5-29.6]
Spring 1,797,221 23.8% [23.6-23.9] 377 25.1% [19.9-30.2] 60 25.3% [12.7-38.0] 24 19.0% [3.2-34.7]
Summer 1,732,798 22.9% [22.7-23.1] 307 20.4% [16.1-24.7] 65 27.5% [15.5-39.5] 25 19.5% [6.2-32.8]
Fall 1,868,260 24.7% [24.5-24.9] 437 29.0% [24.2-33.9] 49 20.6% [10.7-30.5] 60 46.5% [27.6-65.3]
Gender
Female 3,906,439 48.9% [48.5-49.3] 620 40.1% [34.7-45.5] 158 63.2% [50.4-76.0] 74 57.3% [39.9-74.7]
Male 4,082,497 51.1% [50.7-51.5] 927 59.9% [54.5-65.3] 92 36.8% [24.0-49.6] 55 42.7% [25.3-60.1]
Race
White 3,510,178 50.2% [48.0-52.4] 780 59.4% [52.4-66.3] 97 42.1% [28.3-55.8] 61 55.5% [32.4-78.6]
Black 1,186,127 17.0% [15.7-18.2] 185 14.1% [9.3-18.9] 74 31.9% [19.2-44.5] 25 22.5% [4.3-40.7]
Hispanic 1,632,441 23.4% [21.3-25.4] 214 16.3% [11.5-21.1] 45 19.5% [8.3-30.6] 15 13.4% [0-28.0]
Asian 195,106 2.8% [2.3-3.3] 50 3.8% [1.4-6.1] 5 2.2% [0-6.6] 5 4.5% [0-12.9]
Native American 79,845 1.1% [0.8-1.5] 14 1.1% [0-2.2] 6 2.4% [0-7.0] 0 0
Other 386,419 5.5% [4.8-6.2] 71 5.4% [2.5-8.3] 5 2.0% [0-5.9] 5 4.1% [0-12.3]
Income quartile
First 2,560,900 32.4% [30.6-34.2] 439 28.1% [22.4-33.7] 53 22.1% [8.5-35.7] 28 22.2% [3.3-41.0]
Second 1,988,793 25.2% [24.2-26.2] 385 24.6% [19.6-29.6] 64 26.5% [9.1-43.8] 39 30.5% [9.0-51.9]
Third 1,814,127 23.0% [22.0-23.9] 390 24.9% [19.9-30.0] 82 34.2% [20.8-47.6] 20 15.4% [1.0-29.7]
Fourth 1,535,194 19.4% [17.6-21.3] 350 22.4% [17.2-27.6] 41 17.2% [5.7-28.7] 41 32.0% [5.6-58.4]
Insurance
Medicare 36,818 0.5% [0.3-0.6] NE 0 0 0 0 0
Medicaid 4,163,513 51.4% [50.1-52.7] 583 36.9% [30.8-43.0] 117 46.6% [32.6-60.6] 64 49.6% [29.7-69.5]
Private insurance 3,307,390 40.8% [39.4-42.2] 902 57.2% [51.1-63.2] 113 45.2% [31.3-59.0] 65 50.4% [30.5-70.3]
Self-pay 256,476 3.2% [2.8-3.6] 38 2.4% [0.8-4.1] 10 3.8% [0-9.1] 0 0
No charge 15,150 0.2% [0.1-0.3] NE 0 NE 0 0 0

Hsu et al 7.e2
Other 324,904 4.0% [3.6-4.5] 55 3.5% [0.9-6.1] 11 4.4% [0-12.1] 0 0
Insured
Yes 7,507,721 92.6% [92.0-93.3] 1485 94.1% [91.3-97.1] 230 91.8% [83.3-100.0] 129 100%
No 256,476 3.2% [2.8-3.6] 38 2.4% [0.8-4.1] 10 3.8% [0-9.1] 0 0
Continued
Supplemental Table II. Cont’d

7.e3 Hsu et al
No SJS, SJS-TEN, or TEN SJS SJS-TEN TEN
Variable Frequency Prevalence [95% CI] Frequency Prevalence [95% CI] Frequency Prevalence [95% CI] Frequency Prevalence [95% CI]
Other or no charge 340,054 4.2% [3.7-4.7] 55 3.5% [0.9-6.1] 11 4.4% [0-12.1] 0 0
No. of chronic conditions
0 3,390,524 41.8% [40.5-43.0] 681 43.1% [37.4-48.9] 62 24.8% [13.6-36.0] 44 33.8% [12.7-54.9]
1-2 3,156,964 38.9% [38.4-39.4] 581 36.8% [31.4-42.2] 114 45.5% [31.4-60.0] 35 27.0% [9.3-44.8]
3-4 999,037 12.3% [11.7-12.9] 131 8.3% [5.3-11.3] 50 20.0% [9.5-30.5] 19 14.7% [0.4-28.9]
$5 574,230 7.1% [6.5-7.6] 186 11.8% [7.9-15.6] 24 9.7% [2.3-17.0] 31 24.5% [0-51.5]
Hospital location
Metropolitan [1 million 2,558,835 31.9% [28.4-35.4] 513 32.5% [26.5-38.6] 85 34.7% [15.1-54.3] 60 47.0% [21.2-72.8]
Fringe/metropolitan \1 4,053,382 50.6% [47.6-53.6] 754 47.8% [41.1-54.5] 127 51.8% [33.0-70.5] 48 37.4% [16.2-58.7]
million
Micropolitan 853,626 10.6% [9.8-11.5] 194 12.3% [8.3-16.3] 24 9.7% [1.4-18.1] 15 11.7% [0-24.2]
Not metropolitan or 551,685 6.9% [6.3-7.5] 117 7.4% [4.3-10.5] 9 2.6% [0-9.0] 5 3.9% [0-11.5]
micropolitan
Region
Northeast 1,354,663 16.7% [13.7-19.7] 194 12.3% [6.6-17.9] 28 11.2% [1.4-21.0] 14 11.0% [0-26.1]
Midwest 1,657,751 20.4% [16.4-24.4] 375 23.7% [15.8-31.7] 14 5.6% [0-11.8] 24 18.7% [2.9-34.5]
South 3,182,798 39.2% [34.9-43.5] 585 37.0% [29.1-45.0] 114 45.4% [27.9-63.0] 34 26.6% [6.1-47.0]
West 1,925,543 23.7% [19.9-27.6] 426 27.0% [19.7-34.2] 95 37.8% [17.8-57.8] 56 43.7% [17.5-69.8]
Teaching status
Nonteaching hospital 1,964,008 32.6% [28.5-26.8] 173 15.3% [9.4-21.1] 10 5.4% [0-12.9] 10 9.3% [0-31.2]
Teaching hospital 4,054,721 67.4% [63.2-71.5] 961 84.7% [78.9-90.6] 170 94.6% [87.1-100.0] 65 9.3% [68.8-100.0]
Hospital bed size
Small 888,075 11.1% [8.8-13.4] 166 10.6% [5.9-15.3] 26 5.6% [0-22.2] 29 24.6% [3.9-45.3]
Medium 2,004,216 25.1% [20.8-29.3] 430 27.5% [19.4-35.6] 38 16.6% [4.4-28.9] 46 38.6% [10.3-67.0]
Large 5,097,204 63.8% [59.4-68.2] 968 61.9% [53.5-70.3] 166 72.1% [56.4-87.8] 44 36.7% [12.8-60.7]

Survey weighted frequencies of patients with no diagnosis of SJS-TEN/TEN, SJS, SJS-TEN, or TEN examining sociodemographics and racial/ethnic breakdown. Missing values were encountered in
238,103 (14.1%) for race/ethnicity, 27,358 (1.6%) for sex, 44,519 (2.6%) for median household income, 27,260 (1.6%) for age, 3401 (0.2%) for insurance status, 20,712 (1.2%) for urban/rural setting of
hospital, 117,248 (6.9%) for month/season of admission, 0 (0%) for number of chronic conditions, 0 (0%) for year, 422,203 (25.0%) for teaching status, 28,042 (1.7%) for hospital bed size, and 0 (0%)
for hospital region.
CI, Confidence interval; NE, not estimable; SJS, Stevens-Johnson syndrome; TEN, toxic epidermal necrolysis.

J AM ACAD DERMATOL
n 2017
J AM ACAD DERMATOL Hsu et al 7.e4
VOLUME jj, NUMBER j

Supplemental Table III. Adjusted predictors of Stevens-Johnson syndrome, Stevens-Johnson syndrome-

toxic epidermal necrolysis, or toxic epidermal necrolysis in pediatric patients
SJS SJS-TEN TEN
Variable OR [95% CI] P OR [95% CI] P OR [95% CI] P
Age, y
0-2 Reference - Reference - Reference -
3-4 2.96 [1.47-5.95] <.0001 4.47 [1.64-12.19] .002 1.18 [0.34-4.06] .8
5-11 7.98 [5.06-12.59] .006 2.34 [0.57-9.67] .2 0.78 [0.10-6.35] .8
12-17 7.24 [4.64-11.30] <.0001 4.83 [1.82-12.84] .003 2.38 [1.07-5.27] .03
Season
Winter 0.83 [0.59-1.18] .5 0.88 [0.40-1.90] .7 0.66 [0.18-2.47] .5
Spring Reference Reference - Reference -
Summer 0.79 [0.55-1.14] .3 1.13 [0.51-2.51] .8 0.86 [0.23-3.21] .8
Fall 1.03 [0.73-1.44] .3 0.79 [0.36-1.74] .6 2.36 [0.79-7.07] .1
Gender
Female 0.64 [0.50-0.83] .0006 1.79 [1.04-3.10] .04 1.54 [0.75-3.15] .2
Male Reference - Reference - Reference -
Race
White Reference - Reference - Reference -
Black 0.78 [0.52-1.17] .06 2.25 [1.21-4.15] .01 1.20 [0.42-3.43] .7
Hispanic 0.62 [0.42-0.91] .002 1.00 [0.45-2.20] .9 0.52 [0.13-2.02] .3
Asian 1.18 [0.60-2.33] .6 0.96 [0.12-7.37] .9 1.45 [0.17-12.11] .7
Native American 1.02 [0.33-3.20] .6 2.49 [0.35-17.82] .4 NE -
Other 0.77 [0.40-1.46] .8 0.43 [0.06-3.16] .4 0.68 [0.085-5.45] .7
Income quartile
First 1.11 [0.75-1.65] .06 0.77 [0.30-1.96] .6 0.42 [0.10-1.72] .2
Second 1.15 [0.79-1.68] .4 1.19 [0.39-3.60] .8 0.73 [0.19-2.85] .7
Third 1.13 [0.78-1.64] .9 1.69 [0.74-3.85] .2 0.31 [0.067-1.39] .1
Fourth Reference - Reference - Reference -
Insurance
Medicare NE - NE - NE -
Medicaid 0.64 [0.49-0.84] <.0001 0.82 [0.46-1.46] .5 0.85 [0.37-1.91] .7
Private insurance Reference - Reference - Reference -
Self-pay 0.73 [0.34-1.56] .1 1.09 [0.25-4.70] .9 NE -
No charge NE - NE - NE -
Other 0.84 [0.44-1.60] .3 0.99 [0.16-6.21] .9 NE -
Insured
Yes Reference Reference - - -
No 0.80 [0.42-1.53] .5 1.22 [0.29-5.14] .8 NE -
Other 0.91 [0.46-1.81] .8 1.06 [0.17-6.50] .9 NE -
No. of chronic conditions
0 Reference - Reference - Reference -
1-2 0.64 [0.48-0.85] .002 1.97 [1.01-3.86] .048 0.74 [0.27-2.02] .6
3-4 0.37 [0.23-0.60] <.0001 2.73 [1.28-5.84] .01 1.47 [0.44-2.93] .5
$5 0.91 [0.59-1.41] .7 2.30 [0.87-6.06] .1 4.27 [0.90-20.37] .07
Hospital location
Metropolitan $1 million Reference - Reference - Reference -
Fringe/metropolitan \1 0.83 [0.61-1.12] .7 0.94 [0.39-2.29] .9 0.45 [0.15-1.33] .1
million
Micropolitan 1.06 [0.69-1.64] .5 0.84 [0.27-2.64] .8 0.74 [0.18-3.07] .7
Not metropolitan or 0.85 [0.49-1.47] .9 0.51 [0.11-2.37] .4 0.38 [0.045-3.26] .4
micropolitan

Predictors of SJS, SJS-TEN, or TEN in children, presented as odds ratios based on logistic regression models controlling for
sociodemographics and race and adjusting for survey weighting, sampling clusters and strata. Bold denotes a statistically significant finding.
CI, Confidence interval; NE, not estimable; OR, odds ratio; SJS, Stevens-Johnson syndrome; TEN, toxic epidermal necrolysis.