Clinical Classification of Cases of Toxic
Epidermal Necrolysis, Stevens-Johnson
Syndrome, and Erythema Multiforme
Sylvie Bastuji-Garin, MD; Berthold Rzany, MD; Robert S. Stern, MD;
Neil H. Shear, MD, FRCPC; Luigi Naldi, MD; Jean-Claude Roujeau, MD
\s=b\
Background and Design.\p=m-\Toconduct a prospective case\x=req-\
control study about causative factors of severe bullous erythe-
ma multiforme, Stevens-Johnson syndrome, and toxic epider-
mal necrolysis, we needed to define criteria for classifying the
cases and standardize the collection of data so that cases
could be reliably diagnosed according to this classification.
Based on review of case histories and photographs of pa-
tients, a group of experts proposed a classification based on
the pattern of erythema multiforme\p=m-\likelesions (categorized
as typical targets, raised or flat atypical targets, and purpuric
macules) and on the extent of epidermal detachment. An atlas
illustrating this classification that included photographs and
schematic drawings was developed. We compared the eval-
uations of 28 cases by four nonphysicians relying on the atlas
with the evaluations of the same cases by five experts not us-
ing the atlas to determine the usefulness of this atlas for clas-
sifying cases according to our nosologic schema.
Results.\p=m-\Thefollowing consensus classification in five
categories was proposed: bullous erythema multiforme, de-
tachment below 10% of the body surface area plus localized
nosology
The uted drugs
versial.1
to
for severe acute bullous disorders attrib
or infectious agents remains contro
For example, there is disagreement whether
erythema multiforme (EM), Stevens-Johnson syndrome
(SJS), and toxic epidermal necrolysis (TEN) are differ
ent diseases or variants within a continuous spectrum.
Interpretation of the medical literature about EM, SJS,
and TEN is difficult, because no clear criteria have been
established and the proposed definitions are not accepted
worldwide. Therefore, one case diagnosed as SJS might
have been called TEN by another investigator.2
The relative importance of causative factors vary with
the type of illness in the spectrum of EM. Mycoplasma
pneumoniae and Herpes simplex are well-recognized
causes of EM and SJS, but TEN has not been correlated
with these infectious diseases. Erythema multiforme,
SJS, and TE N can all be caused by drug reactions but with
Accepted for publication September 4, 1992.
From the Departments of Dermatology, Henri-Mondor Hospital,
University of Paris XII (Drs Bastuji-Garin and Roujeau), Cr\l=e'\teil,
France; Universit\l=a"\ts-Hautklinik,University of Freiburg (Germany)
(Dr Rzany); Beth Israel Hospital, Harvard Medical School, Boston,
Mass (Dr Stern); Sunnybrook Health Science Centre, University of
Toronto (Ontario) (Dr Shear); General Hospital of Bergamo, Uni-
versity of Milan (Italy) (Dr Naldi).
Reprint requests to Service de Dermatologie, H\l=o^\pitalHenri\x=req-\
Mondor, 94010 Cr\l=e'\teilCedex, France (Dr Roujeau).
Downloaded From: http://archderm.jamanetwork.com/ by a The University of Melbourne Libraries User on 07/26/2015
"typical targets" or "raised atypical targets"; Stevens\x=req-\
Johnson syndrome, detachment below 10% of the body sur-
face area plus widespread erythematous or purpuric macules
or flat atypical targets; overlap Stevens-Johnson syndrome\p=m-\
toxic epidermal necrolysis, detachment between 10% and 30%
of the body surface area plus widespread purpuric macules or
flat atypical targets; toxic epidermal necrolysis with spots, de-
tachment above 30% of the body surface area plus widespread
purpuric macules or flat atypical targets; and toxic epidermal
necrolysis without spots, detachment above 10% of the body
surface area with large epidermal sheets and without any pur-
puric macule or target. Using the atlas, the nonexperts showed
excellent agreement with the experts.
Conclusion.\p=m-\This study suggests that an illustrated atlas is
a useful tool for standardizing the diagnosis of acute severe
bullous disorders that are attibuted to drugs or infectious
agents. Whether the five categories proposed represent dis-
tinct etiopathologic entities will require further epidemiologic
and laboratory investigations.
(Arch Dermatol. 1993;129:92-96)
different frequencies; TEN is mostly related to drugs.
Our group is conducting an international prospective
case-control study of causative factors of severe EM,
SJS, and TEN. For this study, a clear, precise and repro
ducible consensus definition was needed to identify and
classify the cases until more objective markers have been
established.
Thus, the aim of this study was to propose a consensus
definition and classification of severe bullous EM, SJS, and
TEN cases and to evaluate the usefulness as an aid in clas
sification of cases of a photographic atlas that illustrates
the various patterns of the skin lesions and extent of body
surface area (BSA) detachment.
METHODS
Consensus Definition of EM, SJS, and TEN
Dermatologists from Canada (N.S.), France (J.C.R. and
S.B.G.), Germany (B.R.), Italy (L.N.), and the United States
(R.S.) reviewed clinical material, histopathologic findings, and
photographs of the first 16 cases included in a case-control study
(EM sufficiently severe to result in hospitalization, SJS, and
TEN). The study group was faced with the task of classifying
these cases. Each member of the group had a unique approach
based on widely used criteria.3'4 Initially, there was a great deal
of confusion on what constituted target lesions, and, unless fur
ther characterized, the presence of lesions labeled as "targets"
was unfruitful in the classification process. Generally, areas of
detachment were also grossly overestimated by the physicians
 Fig 1.Typical targets with erythematous or purpuric central
disk with or without blister, raised edematous intermediate
ring, and erythematous outer ring.
Fig 3.Flat atypical targets
with round lesions reminis
cent of erythema multiforme
but with only two zones and
poorly defined border. The
central zone is often vesicu-
lous or bullous.
who cared for these patients and who had little experience in the
evaluation of these diseases. As a result, reliably classifying a pa
tient as having SJS or TEN was difficult. Subsequently, the sub
group of three European dermatologists (from France and Ger
many) reviewed the photographs of more than 200 cases not in
volved in the case-control study to try to categorize the pattern
of targetlike lesions and to illustrate varying extents of epidermal
detachment. At a further meeting, their conclusions with a review
of cases and an analysis of the original article by Stevens and
Johnson were presented to the whole group of experts. A con
sensus was then reached after reanalyzing the following criteria:
mucous membrane erosions, pattern of targetlike lesions, extent
of epidermal detachment, and skin pathology.
Building and Evaluation of an Atlas
on the 'Consensus' Definition
A photographic atlas including pictures chosen as typical of the
various patterns of EM-like lesions was then developed. Photo
graphs and schematic drawings of cases where the percentage of
BSA detachment had been thoroughly calculated for patients
hospitalized in a specialized unit, with a table used in burn units
(this table differing only slightly from the rule online),5 were used
to illustrate BSA involvement. Because entire body pictures from
severely ill bedridden patients were not available, we selected
schematic drawings to illustrate different BSA involvements
along with typical photographs of selected anatomic sites. The ac-
Downloaded From: http://archderm.jamanetwork.com/ by a The University of Melbourne Libraries User on 07/26/2015
Fig 2.Raised atypical targets with round lesions reminiscent
of papular erythema multiforme but with only two zones and
poorly defined border.
Fig 4.Macules and/or blisters. Nonpalpable erythematous
or purpuric macules with irregular shape and size, often con
fluent. Blisters occur on part of the macules.
curacy of the percentage of involvement represented in each
schematic drawing was confirmed by planimetric measurement
of the drawings. The atlas is summarized in Figs 1 through 9.
Photographic pictures of 28 new cases included in the case-
control study were simultaneously, but independently, reviewed
by three groups. The two expert groups consisted of the three
European experts who developed the atlas and the two North
American experts who participated in the elaboration ofthe def
initions and the classification but who did not choose photo
graphs or schematic diagrams for the atlas. The nonexpert group
comprised four nonphysician epidemiologists from the United
States and France who had no experience in dermatology. For
each case, each assessor independently scored the pattern of
EM-like lesions and the percentage of BSA involved. In contrast
with the experts, the nonexperts referred to the photographic
atlas during their evaluation.
Agreement between assessors was evaluated. The concor
dance on pattern of lesions was evaluated by the percentage of
agreement. Correlation coefficients were calculated to assess the
judges' agreement on BSA detachment.
RESULTS
Consensus Definition of EM, SJS, and TEN
Mucosal erosions were noticed in at least one site in
more than 90% of the patients hospitalized for a disease
 Fig 5.Schematic examples of the usual distribution of detachment in Stevens-Johnson syndrome-toxic epidermal necrolysis.
Below 10%, sparse nonconfluent lesions; 10%, areas of confluence limited to the chest, upper back, arms, and face; 30%, large
areas of epidermal separation on the trunk, arms, and forearms with sparse lesions elsewhere; 40%, most of the trunk is in
volved with minor lesions elsewhere.

Fig 6.Photographic example

of involvement of less than
10% of the body surface area.
The lesions are sparse in the
body; confluence could be ob
served on very few areas but
without significant epidermal
sheets.

Fig 7.Photographic example

of involvement of about 10% of
the body surface area. The le
sions are. sparse with limited
areas of confluence on the
chest, arms, and face (forehead
or chins).

Fig 8.Photographic example

of involvement of about 30% of
the body surface area. Large
areas of confluence of epider
mal separation are observed on
the trunk, arms, and forearms
with large epidermal sheet.
Conversely, lesions are sparse
on the lower parts of the body
(not shown).

Fig 9.Photographic example

of involvement of about 40% of
the body surface area. Most of
the trunk is involved by the de
tachment with minor lesions on
the legs (not shown).

Downloaded From: http://archderm.jamanetwork.com/ by a The University of Melbourne Libraries User on 07/26/2015

 considered part of the so-called EM spectrum. There
was no correlation between the extent and severity of
mucous membrane erosions and the extent of epider
mal detachment. The presence and distribution of these
mucosal lesions were considered useful for including a
disease within that spectrum but not for classifying it
as severe EM, SJS, or TEN.
Skin pathology was also considered important for rec
ognizing these disorders and excluding differential diag
nosis but not for classifying the disease. According to the
site and the timing of the biopsy, a mild interface derma
titis was sometimes obtained from early lesions of typical
TEN, and a pattern of necrosis of full-thickness epidermis
was seen from the biopsy specimen ofthe blistering center
of a typical target of EM.
The experts agreed that, at present, the most reliable
means to classify cases must rely on the pattern of indi
vidual lesions of EM and on the estimation of the BSA in
volved at the worst stage of the disease.
In the retrospective analysis, the clinical pattern of E M-
like lesions appeared as four different types.
Typical targets were defined as individual lesions less
than 3 cm in diameter with a regular round shape, well-
defined border, and at least three different zones, ie, two
concentric rings around a central disk. One ring consisted
of palpable edema, paler than the center disk (Fig 1 ). In the
review of more than 200 cases by the three European ex
perts, typical targets, as defined above, were never ob
served in patients with widespread epidermal detach
ment. They were usually acraly located and characteris
tically seen in postherpetic cases.
Raised atypical targets were defined as round, edem-
atous, palpable lesions, reminiscent of EM but with only
two zones and/or a poorly defined border (Fig 2).
Flat atypical targets were defined as round lesions,
reminiscent of E M but with only two zones and/or a poorly
defined border, nonpalpable with the exception of poten
tial central blister (Fig 3).
Macules with or without blisters were defined as non-
palpable, erythematous or purpuric macules with irreg
ular shape and size, often confluent. Blisters often oc
curred on all or part of the macule. This kind of lesion
was characteristically seen in cases attributed to drugs,
with widespread epidermal detachment (Fig 4).
Some patients had several of the above patterns of le
sions simultaneously that appeared to fit in the following
two different categories: typical and raised atypical tar
gets on the one hand and flat atypical targets and macules
on the other.
As for the BSA involved at the worst stage of the dis
ease, all experts agreed that it should measure the extent
ofdetached and detachable epidermis (which is often much
less than area of erythema), because detachment is a ma
jor prognostic factor. In accordance with the original pub
lication of Stevens and Johnson, SJS was defined as cases
with limited areas of epidermal detachment (<10%) (Fig
6). In keeping with the original paper by Lyell, we re
quired TEN cases to have large areas of detachment
(>30%) (Fig 9). To overcome the problem of the border
between TEN and SJS, we adopted an "overlap" category
for patients with areas of epidermal detachment between
10% (Fig 7) and 30% (Fig 8) of the BSA.
Downloaded From: http://archderm.jamanetwork.com/ by a The University of Melbourne Libraries User on 07/26/2015
Proposed Classification of Cases in the
Spectrum of Severe Bullous EM*
TEN
I-1
Bullous Overlap With Without
Classification EM SJS SJS-TEN Spots Spots
Detachment <10% <10% 10%-30% >30% >10%
Typical
targets Yes
Atypical
targets Raised Flat Flat Flat
Spots Yes Yes Yes
. ..
*EM indicates erythema multiforme; SJS, Stevens-Johnson syn
drome; and TEN, toxic epidermal necrolysis.
In many cases, the epidermal detachment progressed
over a period of a few days. These cases were categorized
according to the area ofmaximal involvement. In all ofthe
50 cases reviewed that progressed to large areas of de
tachment, the discrete lesions were always erythematous
or purpuric macules or atypical targets, never typical tar
gets. In rare instances, extensive epidermal necrosis oc
curred without any discrete EM-like lesions. These cases
were classified as TEN without spots.
As a consensus definition of EM, SJS, and TEN, the fol
lowing classification in five categories was proposed (Ta
ble): bullous EM, detachment below 10% of the BSA plus
localized typical targets or raised atypical targets; SJS,
detachment below 10% of the BSA plus widespread ma
cules or flat atypical targets; overlap SJS-TEN, detach
ment between 10% and 30% of the BSA plus widespread
macules or flat atypical targets; TEN with spots with or
without blisters, detachment above 30% of the BSA plus
widespread macules or flat atypical targets; and TEN
without spots, detachment above 10% of the BSA with
large epidermal sheets and without any macule or target.
Evaluation of the Atlas
For the pattern of lesions, the three European experts
who developed the atlas agreed on 22 (78.6%) of 28 cases,
and the agreement reached 100% for the decision of two
or more experts. The two North American experts agreed
on 68% and 75% of these decisions. With the help of the
atlas, the four nonexperts agreed on 71% to 75% of these
decisions.
The agreement on the area of detachment was quite
good among European experts on one hand (correlation
coefficients, .84 to .87) and North American experts on
the other (correlation coefficient, .71), with mean intra-
group differences below 10%. Nevertheless, those mean
differences in BSA detachment evaluation reached 22%
when comparing the experts of the two groups. The es
timations of the North American experts were system
atically over those of European experts.
The estimations of BSA detachment by three of four
nonexperts were in quite good correlation (correlation co
efficients, .54 to .74) with the decision of the three Euro
pean experts, with an overestimation of less than 10%. The
decisions ofthree of four nonexperts were in better agree
ment with those of the three experts who developed the
atlas than with the decisions of other experts.
 COMMENT
The classification of severe forms of EM is still a matter
of debate and confusion. The first description of EM was
attributed to von Hebra,6 who described a self-limited,
mild skin disease characterized by symetrically distribut
ed skin lesions, located primarily on the extremities and
by a tendency for recurrences. The primary lesions were
papules evolving with a concentric color change (erythe
ma iris) or blisters (herpes iris). In 1916, Rendu7 described
an acute febrile illness (later named ectodermosis erosiva
plunorificialis), characterized by severe erosions of all
mucous membranes and a vesicular skin eruption. In 1922,
Stevens and Johnson8 described two boys who were fe
brile with erosive stomatitis, severe purulent conjunctivi
tis, and a disseminated cutaneous eruption. This eruption
"consisted of oval, dark to purplish macules separated by
normal areas of skin_a few of the largest spots showed
a yellow, dry, necrotic center." The disorders described by
Rendu and Stevens and Johnson were probably very
close, if not identical. All of these authors believed that
they had described a new disease, distinct from EM. Nev
ertheless, the eponym SJS was later adopted as a syn
onym of "EM major," the more severe mucocutaneous va
rieties with mucosal involvement.1 The first description of
TEN in 1956 by LyelF made no reference to EM or SJS.
However, it soon became evident that severe forms ofSJS
could lead to TEN and that the same drugs could induce
both disorders. Furthermore, in most cases ofTEN poorly
defined macules with darker centers are present around
the areas of epidermal sloughing. Many considered SJS to
be a form of TEN. By analogy, many persons also came to
regard TEN as within the "EM spectrum."1 In our expe
rience, the individual discrete lesions of TEN invariably
differed from typical targets of EM by their less regular
shapes and the absence of an edematous ring.
After reviewing hundreds of clinical slides, we devel
oped what we believe is a logical and reproducible system
for categorizing the lesions seen in these disorders. For
the consensus definitions adopted in the present work, we
chose to adhere as closely as possible to the original de
scriptions with the hypothesis that SJS and TEN could be,
in fact, different from classic EM. We then proposed what
we believe is a reasonable nosology for these disorders
based on the character of the lesions and extent ofinvolve
ment. To overcome the problem of the border between
TEN and SJS, we adopted an "overlap" category for pa
tients with areas of epidermal detachment, which are of
ten much less than areas of erythema, between 10% and
30% of the BSA. We emphasized the percentage of BSA
detachment in our classification because that is a major
prognosis factor. That was an attempt to avoid the use of
the same denomination for conditions with a totally differ
entprognosis.
Ultimately, our classification comprises five groups.
These categories are not inclusive of all of the potential
combinations of the nature of the discrete lesions and the
percents of detachment. For example, a patient with typ
ical targets and 20% detachment could not be classified. In
the retrospective review of more than 200 cases, we did
not find a single patient who met these criteria. The pro
posed classification is currently used in classifying cases
reported to a German registry and cases included in our
Downloaded From: http://archderm.jamanetwork.com/ by a The University of Melbourne Libraries User on 07/26/2015
ongoing case-control study. Using it, we are able to clas
sify more than 95% of the properly documented cases in
one of the categories.
This classification leads to a practical algorithm in the
classification of these diseases. The first question the cli
nician needs to ask is this: What is the percent of detach
ment? The second question is the following: What is the
nature of the discrete lesions? It is quite different from
the usual question on how many mucous membranes are
involved.
This classification is purely descriptive and may appear
quite arbitrary. Its validity is yet unproven. It would be
supported by the finding ofdifferent risks of various caus
ative agents in the above subgroups. It is the clinical ex
perience of the authors that SJS, TEN, and overlap cat
egories are usually drug-induced, with cases evolving
from one category to another, while postherpetic cases fit
in the EM group and never evolve to widespread skin de
tachment. From this clinical experience, the authors think
that bullous EM and SJS are probably two distinct disor
ders with different patterns of cutaneous lesions and dif
ferent causative factors. This remains to be demonstrated,
and the determination of the relative risks linked to caus
ative factors is the aim of an ongoing case-control study.
Given the consistency in diagnoses between experts
and nonexperts who used the atlas, we believe this atlas
should help collaborating investigators in our present
study. The atlas will also be used by primary care physi
cians to communicate with experts and to standardize
communication between experts.
The realization of the photographic atlas was made possible by a
grant from Cilag Co, Levallois-Perret, France. The meetings of inves
tigators, as a part of a prospective international case-control study on
severe cutaneous adverse drug reactions, were supported by grants
from INSERM, Paris, France; Sunnybrook Research Fund, Toronto,
Ontario; Canadian Dermatology Foundation, Toronto; Smith Kline
Beecham, Nanterre, France; Bristol, Paris; Ciba-Geigy, Reuil-
Malmaison, France; Hoechst, Puteaux, France; Pfizer, Orsay, France;
Roche, Neuilly sur Seine, France; Roussel-UCLAF, Paris; Sandoz,
Reuil-Malmaison; SPECIA, Paris; Sterling-Winthrop, Clichy, France;
UPSA, Reuil-Malmaison; and Wellcome Laboratories, Paris.
Dr Shear is a Career Scientist of the Ontario (Canada) Ministry of
Health, Toronto.
The authors thank Ariane Auquier, MA, David Kaufman, ScD,
Judy Kelly, MS, and Catherine Paoletti for their participation as
"nonexperts" in the evaluation of the atlas.
References
1. Huff JC, Weston WL, Tonnesen MG. Erythema multiforme: a critical re-
view of characteristics, diagnostic criteria, and causes. J Am Acad Dermatol.
1983;8:763-775.
2. Goldstein SM, Wintroub BW, Elias PE, Wuepper KD. Toxic epidermal
necrolysis: unmuddying the waters. Arch Dermatol. 1987;123:1153-1156.
3. 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.
4. Roujeau JC, Chosidow O, Saiag P, Guillaume JC. Toxic epidermal necrolysis
(Lyell syndrome). J Am Acad Dermatol. 1990;23:1039-1058.
5. Lund CC, Browder NC. The estimation of areas of burns. Surg Gynecol Ob-
stet. 1944;79:352-359.
6. von Hebra F. Atlas der Hautkrankheiten. Vienna, Austria: Kaiserliche
Akademie der Wissenschaften; 1866.
7. Rendu R. Sur un syndrome caract\l=e'\ris\l=e'\par l'inflammation simultan\l=e'\ede
toutes les muqueuses externes (conjonctivale, nasale, linguale, bucco-pharyng\l=e'\e,
anale et balano-pr\l=e'\puciale)coexistant avec une \l=e'\ruptionvaricelliforme puis pur-
purique des quatre membres. J Prat. 1916;30:351.
8. Stevens AM, Johnson FC. A new eruptive fever associated with stomatitis
and ophthalmia: report of two cases in children. AJDC. 1922;24:526-533.
9. Lyell A. Toxic epidermal necrolysis: an eruption resembling scaling of the
skin. Br J Dermatol. 1956;68:355-361.