Fax +41 61 306 12 34

E-Mail karger@karger.ch
www.karger.com
Original Paper
Dermatology 2012;225:154162
DOI: 10.1159/000342365
Erythroderma: A Clinical and
Prognostic Study
Jun Li He-Yi Zheng
Department of Dermatology and Venereology, Peking Union Medical College Hospital, Chinese Academy of
Medical Sciences and Peking Union Medical College, Beijing , PR China
riasis was the most common etiology (143/260, 55%). In the
drug-induced group, carbamazepine was the most fre-
quently implicated drug in our study (33.33%). Chinese tra-
ditional herbal medicines are among the causes of drug-in-
duced erythroderma as well. We also found that Langerhans
cell histiocytosis, tongue cancer, hypereosinophilic syn-
drome, bullous pemphigoid and dermatomyositis could be
causes of erythroderma. From our follow-up study, 39 (31.2%)
of the 125 patients from whom information was available
had relapsed. The patients with idiopathic erythroderma
had a higher relapse rate. Of the 5 patients who died, 4
deaths were directly related to erythroderma. Conclusion:
Most of the clinical features of erythroderma are unspecific
with few cause-orienting clues. Although numerous labora-
tory values were abnormal, most findings were nondiagnos-
tic and were related to the inflammatory process, except for
skin biopsy. Our study had a high percentage of erythroder-
ma secondary to pre-existing dermatoses and a low percent-
age of malignancy patients. Repeated evaluations, close fol-
low-up and biopsy are recommended.
Copyright 2012 S. Karger AG, Basel
Introduction
First described by Hebra in 1868, erythroderma, or ex-
foliative dermatitis, is an inflammatory disorder in which
erythema and scaling occur in a generalized distribution
involving more than 90% of the body surface area for
Key Words
Erythroderma Exfoliative dermatitis East Asia Psoriasis
Abstract
Background: Erythroderma is a rare skin disorder and stud-
ies on its causes and prognosis are rare in the literature. Ob-
jectives: We reviewed the clinical, laboratory and biopsy ma-
terials of 260 patients diagnosed with erythroderma who
were treated in our department over an 11-year period. Pa-
tients were followed up to better understand the evolution
of erythroderma. Methods: This study was performed at our
hospital between January 2001 and 2012 and included 260
patients with erythroderma. We recorded epidemio-clinical,
biological and histological data, treatments and outcomes.
Clinical-histological correlation was analyzed. Overall surviv-
al and relapse-free survival for a limited number of etiologies
were described using Kaplan-Meier estimates. Results: The
mean age at onset in this study was 52.57 8 17.94 (SD) years
(range 1387), with a male-to-female ratio of 3: 1. Acute onset
was present in 15.38% of patients. Clinical findings were
dominated by pruritus (87.69%), fever (40%), edema (37.69%),
chills (31.15%), nail changes (29.62%), weakness (19.23%),
lymphadenopathy (19.23%), weight loss (14.62%) and islands
of normal skin (13.46%). Skin biopsies revealed the cause in
55.56% (65/117) of the patients. The most common causative
factors were pre-existing dermatoses (70.77%), followed by
idiopathic causes (14.23%), drug reactions (12.69%) and ma-
lignancies (2.31%). Among the pre-existing dermatoses, pso-
Received: June 19, 2012
Accepted after revision: August 6, 2012
Published online: October 2, 2012
Jun Li, MD
Department of Dermatology and Venereology
Peking Union Medical College Hospital
1 Shuaifuyuan Wangfujing, Beijing 100730 (PR China)
E-Mail lijun35 @ hotmail.com
2012 S. Karger AG, Basel
10188665/12/22520154$38.00/0
Accessible online at:
www.karger.com/drm
Erythroderma: A Clinical and Prognostic
Study
Dermatology 2012;225:154162 155
more than 15 days. The causative factors can be grouped
as previous dermatoses, drug reactions, malignancies, in-
fections and idiopathic disorders. To date, there are few
published studies on the frequency of underlying causes
and prognosis of erythroderma from East Asia. In order
to delineate the salient features of erythroderma in our
region, we reviewed 260 cases of erythroderma treated at
our institution between 2001 and 2012 (there were about
a total of 2,000,000 patients admitted to our department
during this period).
Prognostic studies are rare in the literature; we deter-
mined prognostic factors and survival of our patients.
Finally, our data are discussed and compared with other
earlier studies. To our knowledge, ours is the largest in
the literature. In addition, we collected clinical and lab-
oratory data, which have also rarely been reported in
the previous literature. We also found that Langerhans
cell histiocytosis, tongue cancer, hypereosinophilic syn-
drome, bullous pemphigoid, dermatomyositis and Chi-
nese traditional herbal medicines could be causes of
erythroderma.
Methods
260 patients with erythroderma of either sex and any age were
admitted to the Dermatology Department of Peking Union Med-
ical College Hospital in Beijing, China between January 2001 and
2012. Our department is one of the largest dermatologic clinics in
China and often receives patients from distant areas within main-
land China.
Due to the risks that erythroderma implies for the patients life
and to study the cause in each patient, we always treated them as
inpatients. The records of patients who were discharged with a
diagnosis of erythroderma within the period from 2001 to 2012
were carefully reviewed and the following data recorded for all
patients: personal data, history of skin diseases, past medical his-
tory, drug history, previous episodes of erythroderma, onset of
erythroderma (acute or insidious), duration, and clinical data
during the episode. Laboratory investigations including complete
hematological parameters, liver and kidney function tests, serum
protein, serum lipid, urine test, stool exam for occult blood, elec-
trocardiogram, ultrasound and chest radiography were per-
formed for all the patients routinely in the dermatology ward of
our hospital. Disease-specific investigations such as skin biopsy,
lymph node biopsy, bone marrow investigation and CT scan were
performed whenever indicated. We also examined data concern-
ing management, outcome and relapse when available. Data were
compiled electronically into and analyzed using SPSS version 18.
Follow-up time and disease-free survival time were calculated us-
ing Kaplan-Meier estimates.
Results
Epidemiology/Demographics and Clinical Findings
In the 11-year study period, erythroderma was diag-
nosed in 260 patients. The average age of the patients in
this study was 52.57 8 17.94 (SD) years (range 1387);
patient distribution according to age is shown in figure 1 .
Men outnumbered women by 3: 1. Patients were seen in
our department at a median of 1 month (range 1 day to
10 years) after the onset of erythroderma. A shorter dura-
tion was observed in patients with drug-induced eryth-
roderma and a longer one with erythroderma caused by
psoriasis. In 40 (15.38%) of the 260 patients, erythroder-
ma began acutely. 28 of the 33 patients whose erythro-
derma was caused by drugs were included in this acute
onset group. Of the remaining 220 progressive onset pa-
tients, the median length of previous skin disease was
6.18 years.
As shown in table1 , pruritus, the most common com-
plaint, was recorded in 228 (87.69%) patients; 104 of the
260 patients (40%) had a fever during the episode, 98 pa-
tients (37.69%) had pitting edema, and chills were found
in 81 patients (31.15%). Nail changes (including Beaus
lines, onychodystrophy, pits, discoloration, subungual
hyperkeratosis, onycholysis and paronychia) were ob-
served in 77 (29.62%) patients. Nail changes were mostly
found in psoriasis patients (odds ratio [OR]: 5.33, 95%
confidence interval [CI]: 1.9714.47 for psoriasis; OR:
0.91, 95% CI: 0.223.74 for dermatitis; OR: 1.60, 95% CI:
0.337.75 for others compared with idiopathic erythro-
derma). Other main symptoms included lack of sleep in
75 (28.85%) patients, weakness in 50 (19.23%) patients,
lymph node enlargement in 50 (19.23%) patients and
0
20
40
60
Age (years)
N
u
m
b
e
r

o
f

p
a
t
i
e
n
t
s
0
0

1
0
15
1
1

2
0
19
2
1

3
0
35
3
1

4
0
40
4
1

5
0
49
5
1

6
0
54
6
1

7
0
41
7
1

8
0
7
8
1

9
0
Fig. 1. Incidence arranged by age groups.
C
o
l
o
r

v
e
r
s
i
o
n

a
v
a
i
l
a
b
l
e

o
n
l
i
n
e
Li/Zheng

Dermatology 2012;225:154162 156
Table 1. E pidemiological, clinical and laboratory features of the 260 patients with erythroderma according to etiology
Etiology Psoriasis
143 (55%)
Idiopathic
37 (14.23%)
Drug reaction
33 (12.69%)
Dermatitis
32 (12.31%)
Others*
15 (5.77%)
Mean age 8 SD, years 46.92816.71 62.00817.02 48.67818.64 65.9189.46 63.33816.00
Male:female ratio 2.67 3.22 2.67 5.4 4.0
Hospital stay, days 33.30819.66 26.73832.75 13.8588.68 23.44810.87 33.47814.83
Duration of previous 11 years 1 year 30.42 days 11.15 months 9.13 months
dermatoses (30 days to 50 years) (60 days to 20 years) (4 days to 17 months) (30 days to 50 years) (4 days to 11 years)
Delay after the onset of 1 month 1 month 10 days 30 days 40 days
erythroderma (2 days to 10 years) (6 days to 19 months) (1 day to 2 months) (7 days to 4 months) (4 days to 1 year)
Pruritus 132 32 28 27 9
Temperatured 37.438C 30 1 3 6 0
38.139C 25 4 3 3 5
39.1C 17 1 4 1 1
Pitting edema 51 7 16 17 7
Chills 45 10 13 8 5
Nail changes 65 5 0 4 3
Lack of sleep 43 6 10 10 6
Weakness 19 13 8 7 3
Lymphadenopathy 21 6 7 10 6
Weight loss 22 3 3 7 3
Appetite decrease 14 1 4 1 3
Pustules 15 4 0 0 0
Arthralgia 12 2 0 0 0
Hypertension 17 3 7 6 1
Diabetes 16 4 1 6 0
Sepsis 18 2 2 2 0
Pneumonia 6 1 1
ALTd 4099 U/l 9 1 6 2 0
100 U/l 3 3 3 0 1
TCd 3 1 2 1 0
TGd 27 5 4 4 0
LDd 3 2 7 7 3
UAd 25 3 4 4 1
CRP >10 mg/l 42 10 7 9 1
EOSd 0.51.49 ! 10
9
/l 19 6 3 22 1
1.5 ! 10
9
/l 3 10 8 7 3
EOS%d 27 15 12 21 4
PLTd 41 5 5 8 0
IgEd 1,000 kU/l 2 5 3 5 3
100999 kU/l 23 8 3 7 0
<99 kU/l 6 2 2 2 0
TPf 5060 g/l 45 4 5 10 3
49 g/l 6 1 5 5 1
ALBf 3035 g/l 34 3 7 13 3
29 g/l 31 7 7 8 4
Caf 22.13 mmol/l 21 3 7 8 5
1.99 mmol/l 19 4 4 5 1
HDLCf 22 3 4 7 0
PAf 100180 mg/l 40 9 9 12 0
99 mg/l 20 3 2 2 2
HGBf 24 2 6 4 1
HLA-B27(+) 10/15 0 0 0 0
*

Idiopathic hypereosinophilic syndrome (3), mycosis fungoides (3), bullous pemphigoid (2), pemphigus foliaceus (2), pityriasis
rubra pilaris (1), dermatomyositis (1), lung cancer (1), tongue cancer (1), Langerhans cell histiocytosis (1).
Erythroderma: A Clinical and Prognostic
Study
Dermatology 2012;225:154162 157
weight loss in 38 (14.62%) patients. Islands of normal skin
were observed in 35 (13.46%) patients. The islands were
not only seen in patients with pityriasis rubra pilaris, but
also in patients with psoriasis (23 patients) and dermatitis
(11 patients).
Laboratory Findings
Laboratory abnormalities are also summarized in ta-
ble1 , including elevated alanine transaminase (ALT), to-
tal cholesterol (TC), triglycerides (TG), lactate dehydro-
genase (LD), uric acid (UA), C-reactive protein (CRP),
eosinophils (EOS), platelets (PLT) and immunoglobulin
E (IgE), and decreased total protein (TP), albumin (ALB),
calcium (Ca), high-density lipoprotein (HDL), prealbu-
min (PA) and hemoglobin (HGB).
Cutaneous biopsy was performed in 117 (45%) pa-
tients. A skin biopsy was not performed in the rest of the
patients because the cause of erythroderma was clear
from the start (previous dermatosis or treatment with a
drug the days before the appearance of erythroderma).
The biopsy was usually performed during the first three
days after the patient was admitted to the hospital. Mul-
tiple biopsies were performed in some patients: three bi-
opsies in one patient with mycosis fungoides, two biop-
sies in two patients with psoriasis, atopic dermatitis/
eczema and idiopathic erythroderma. The correlation
between histological and clinical diagnosis can be seen in
table2 (the final diagnosis was the result of the evaluation
of the clinical, biochemical and histological findings and
of the evolution of the erythroderma in each individual
patient).
Out of seven lymph node biopsies performed, only one
revealed Langerhans cell histiocytosis. The diagnosis was
confirmed by electron microscopy and immunohisto-
chemical reactivity of histiocytes to CD1a and S-100. His-
topathologic examination of the skin lesions showed an
inflammatory infiltrate consisting of neutrophils, eosin-
ophils, lymphocytes and histiocytes, the latter having
large ovoid and reniform nuclei with abundant ampho-
philic cytoplasm. The atypical cell populations also ex-
pressed CD1a and S-100 protein. The other six showed
dermatopathic lymphadenopathy.
Eleven patients with erythroderma had a bone mar-
row investigation: 5 showed eosinophilia (1 hypereosino-
philic syndrome, 3 atopic dermatitis/eczema and 1 der-
matomyositis), 3 were normal (1 drug reaction, 1 atopic
dermatitis/eczema and 1 psoriasis), 2 were hyperplastic
(1 with atopic dermatitis/eczema, 1 with psoriasis), and
1 showed infection (the patient was a psoriatic who had
a high fever and was diagnosed as having sepsis with
blood culture-proven infection with methicillin-resistant
Staphylococcus aureus ).
Table 2. C orrelation between histological and clinical diagnosis
Histological diagnosis Biop-
sies
Final diagnosis Pa-
tients
Nonspecific dermatitis 51 idiopathic 22
psoriasis 15
drug reaction 3
dermatitis 9
tongue cancer 1
lung cancer 1
Psoriasis 35 psoriasis 34
idiopathic 1
Dermatitis 13 dermatitis 13
Drug reaction 5 drug reaction 5
Mycosis fungoides (MF) 3 MF 3
Dermal infiltrate with
numerous eosinophils
3 idiopathic
hypereosinophilia
3
Bullous pemphigoid (BP) 2 BP 2
Pemphigus foliaceus (PF) 2 PF 2
Pityriasis rubra pilaris (PRP) 1 PRP 1
Langerhans cell
histiocytosis (LCH)
1 LCH 1
D ermatomyositis 1 dermatomyositis 1
Hi stopathological diagnoses of the PRP case in our study: pso-
riasiform acanthosis, alternating parakeratosis and horizontal as
well as vertical hyperkeratosis. A superficial perivascular lym-
phocytic infiltration was present. Histopathological diagnoses of
dermatomyositis: lichenoid inflammation with basal cell vacuo-
lar degeneration, several Civatte bodies, atrophic epidermis and
thickened basement membrane.
56%
14%
13%
12%
2%
3%
Psoriasis Idiopathic Drug reactions
Dermatitis Malignancies Others
Fig. 2. Etiologies of erythroderma.
C
o
l
o
r

v
e
r
s
i
o
n

a
v
a
i
l
a
b
l
e

o
n
l
i
n
e
Li/Zheng

Dermatology 2012;225:154162 158
Etiologically, the 260 patients were categorized into
four groups ( fig. 2 ): (1) Pre-existing dermatoses (184
patients, 70.77%): psoriasis 143 (55.0%), dermatitis 32
(12.31%) (27 atopic dermatitis/eczema, 3 contact derma-
titis, 1 seborrheic dermatitis, 1 chronic actinic dermatitis,
3 idiopathic hypereosinophilic syndrome, 2 bullous pem-
phigoid, 2 pemphigus foliaceus, 1 pityriasis rubra pila-
ris, 1 dermatomyositis). (2) Drug reactions (33 patients,
12.69%): carbamazepine (11), allopurinol (7), aminopy-
rine (3), Chinese traditional herbal medicines (3), cepha-
losporins (2); in the remaining 7 patients, multiple drugs
were suspected but no single drug could definitely be in-
criminated. (3) Malignancies (6 patients, 2.31%): mycosis
fungoides (3), lung cancer (1), tongue cancer (1), Langer-
hans cell histiocytosis (1). (4) Idiopathic (37 patients,
14.23%).
Treatment and Prognosis
Therapy was topical and/or systemic medication ac-
cording to etiology: (1) Psoriasis: 143 patients (acitretin
81, topical therapy 20, methotrexate 8, PUVA/UVB 5, cy-
closporine 1, etanercept 1, Tripterygium 1, oral steroids 1,
combination therapy 25). (2) Idiopathic: 37 patients (top-
ical therapy 34, Tripterygium 2, oral steroids 1). (3) Drug
reaction: 33 patients (oral steroids 25, topical therapy 4,
Tripterygium 1, combination therapy 3). (4) Dermatitis:
32 patients (Tripterygium 13, topical therapy 7, oral ste-
roids 3, acitretin 3, cyclosporine 2, combination therapy
4). (5) Others: 15 patients (oral steroids 7, Tripterygium 3,
acitretin 2, topical therapy 2, chemotherapy 1). For the
patient with tongue cancer, the skin lesion was under
control gradually after treatment with oral steroids at
first. However, the lesion was recurrent at follow-up and
readmission for another 2 times. 4 years later, he was di-
agnosed as having tongue cancer. During the 6 months
of follow-up after resection of the cancer, the skin lesion
was not improved. The patient still needed oral steroids
to control the skin lesion. There was evidence of recur-
rence of tongue cancer 12 months after surgery, and
1 month after the second operation, his clinical symp-
toms improved dramatically.
Independently of the etiology, the mean follow-up pe-
riod was about 1 year (range 7 days to 11 years). Follow-up
information was obtained from only 125 patients ( ta-
ble3 ). We were unable to contact 135 of the 260 patients.
Eighty-six (68.8%) of the 125 patients from whom infor-
mation was available were in good condition (i.e. free of
cutaneous lesions or in those patients with dermatosis the
latter was under control, allowing the patients to lead
normal lives, or relapse-free survival). In 39 (31.2%) of the
125 patients, erythroderma had relapsed. As shown in
figure 3 and 4 (overall relapse-free survival and relapse-
free survival according to etiology, respectively), the pa-
tients with idiopathic erythroderma relapsed much more
quickly than those with psoriasis.
Five patients died. In 4 patients, the causes of death were
directly related to erythroderma during hospitalization
(2 patients with drug reactions died of cardiac failure dur-
ing the erythroderma episode, 1 patient with drug reaction
died of infectious shock, and 1 patient with dermatitis died
Table 3. T herapeutic effect and prognosis of 125 patients
Etiology Psoriasis
143 (55%)
Idiopathic
37 (14.23%)
Drug reaction
33 (12.69%)
Dermatitis
32 (12.31%)
Others
15 (5.77%)
Outcome during admission
Cleared 0 2 0 0 0
Improved 141 34 28 30 14
Unchanged 2 1 2 1 1
Death 0 0 3 1 0
Follow-up 75 25 7 16 2
Prognosis
Relapse-free survival 50 21 5 8 2
Relapsed 25 4 2* 8 0
Death 0 0 0 1 0
* Two patients with drug-induced erythroderma in this study relapsed, one resumed the same medication
and the other recurred 15 days after discharge; the dose of prednisone was increased and the symptoms finally
disappeared.
Erythroderma: A Clinical and Prognostic
Study
Dermatology 2012;225:154162 159
of acute myocardial infarction). Only 1 patient with der-
matitis whose cause of death was unrelated to erythroder-
ma died of severe pneumonia 39 months after discharge.
Discussion
The approach to patients with erythroderma depends
on their previous dermatologic background. Patients
with dermatologic disorders recalcitrant to therapy may
develop erythroderma during a dermatosis flare. In such
patients, the etiologic diagnosis is easy to establish. Oth-
erwise, erythroderma remains a diagnostic challenge, es-
pecially in those patients without history of dermatolog-
ic diseases and who deny having recently taken any med-
ications.
In our study, the average age of patients was 52.57 8
17.94 years and most of the patients with erythroderma
were above 40 years old. Men outnumbered women by
3: 1. Such findings are in accordance with many other
studies [14] . The onset of erythroderma is usually grad-
ual and insidious [2, 5] except in drug-induced cases,
where it is typically sudden and florid and the resolution
faster than in the other causes [5, 6] . Similarly, in our
study, 28 of the 33 patients whose erythroderma was
caused by drugs were included in the acute onset group.
As in other studies, the majority of clinical features did
not correlate with the etiology [3, 57] . However, in the
present study, nail changes were shown to be predictive
clinical signs of psoriasis. Thus, in the absence of history
of psoriasis, these cutaneous modifications may direct
clinicians to psoriasis. In well-known psoriatic patients,
they probably develop before the onset of erythroderma
and are only exacerbated. Islands of normal skin are con-
sidered to among the diagnostic features in pityriasis ru-
bra pilaris. In our study, in addition to pityriasis rubra
0
0 20 40 60 80 100 120 140
0.2
0.4
0.6
0.8
1.0
C
u
m
u
l
a
t
e
d

s
u
r
v
i
v
a
l

p
r
o
b
a
b
i
l
i
t
y
Time to relapse (months)
Fig. 3. Overall relapse-free survival (Kap-
lan-Meier method).
0
0 20 40 60 80 100 120 140
0.2
0.4
0.6
0.8
1.0
C
u
m
u
l
a
t
e
d

s
u
r
v
i
v
a
l

p
r
o
b
a
b
i
l
i
t
y
Time to relapse (months)
Eczema
Idiopathic
Psoriasis
Fig. 4. Relapse-free survival for psoriasis,
dermatitis and idiopathic (Kaplan-Meier
method).
C
o
l
o
r

v
e
r
s
i
o
n

a
v
a
i
l
a
b
l
e

o
n
l
i
n
e
Li/Zheng

Dermatology 2012;225:154162 160
pilaris, islands were also seen in psoriasis and dermatitis,
which was only previously reported by Pal and Haroon
[1] . The islands may simply reflect less responsive skin in
these particular areas; they cannot be depended upon for
the diagnosis of pityriasis rubra pilaris unless accompa-
nied by other features of the disease.
Though several clinical and laboratory abnormalities
were summarized in our study that have not been men-
tioned in any previous studies, nothing remarkable was
seen in the laboratory tests except the skin biopsy. This
finding was similarly reported by Pal and Haroon [1] . In
previous studies [56, 8] , skin biopsy has been reported
to be of variable (4366%) usefulness in making the cor-
rect histological diagnosis. In our study, skin biopsies
were helpful in establishing a final diagnosis in 65
(55.56%) of the 117 patients in whom cutaneous biopsies
were performed. We therefore regard skin biopsy as a use-
ful investigatory tool in making the diagnosis. They are
recommended to be carried out as soon as possible and
sometimes should be repeated in a different area and a
different stage of erythroderma for a final diagnosis [1, 9] .
In this and previous studies, lymph node biopsy usually
demonstrated dermatopathic changes in patients with
erythroderma. In only one patient in our study did a
lymph node biopsy show a Langerhans cell histiocytosis
that had not previously been diagnosed, suggesting that
lymph node biopsy is not indicated in the initial evalua-
tion of most patients with erythroderma.
Our study had a relatively high percentage (70.77%) of
erythroderma secondary to pre-existing dermatoses, in
accordance with Pal and Haroon (74.4%) [1] . In our study,
psoriasis was the most common (55%) underlying etiol-
ogy of pre-existing dermatosis, which was much higher
than all the previous reports. As a result, we concluded
that there might be a higher frequency of psoriasis or a
greater tendency for its generalization in our district [10] ;
furthermore, unlike in western reports [11] , injudicious
use of medications, either by the individual or prescribed
by physicians, may partly be responsible for the higher
rate of psoriasis (the most frequently suspected trigger
factors were upper respiratory tract infections and with-
drawal of oral medicines including systemic corticoste-
roids or Chinese traditional herbal medicines). In our
study, we also found bullous pemphigoid, hypereosino-
philic syndrome and dermatomyositis as causes of eryth-
roderma, which have never or rarely been mentioned
in previous reports. Hypereosinophilic syndrome was
shown to be a causative factor of erythroderma, which is
very rare [12] . Erythroderma linked to dermatomyositis
is also very unusual and only a few cases have been re-
ported in the literature [13, 14] . Although the incidence is
limited, we did not find other causes for erythroderma in
this patient. Further studies will need to confirm this cor-
relation.
In our study, we found a similar percentage of drug-
induced patients compared to previous reports [23, 5] ,
and the agents of greatest erythroderma-inducing po-
tential are carbamazepine and allopurinol. Previously,
Akhyani et al. [15] reported similar findings. Compared
with a high percentage (64.3%; 9/14) of Chinese tradi-
tional herbal medicines responsible for erythroderma re-
ported in another study from China [12] , we only found
3 patients with erythroderma caused by Chinese tradi-
tional herbal medicines. Penicillin has been mentioned as
one of the most common causes of drugs inducing eryth-
roderma in some other recent studies [4, 8] . Surprisingly,
in spite of the fact that penicillin is prescribed frequently
in our country, we observed no erythroderma related to
this drug.
The percentage of malignancies as a cause for eryth-
roderma is still relatively low in our study and previous
reports [8, 16] . Mycosis fungoides and Szary syndrome
were the most frequent malignant causes of erythroder-
ma, but occasionally, erythroderma was associated with
internal malignancies. (In our study, we did not find a
single case of Szary syndrome; though 17 patients were
suspected of having Szary syndrome at first, Szary cells
could not be detected in peripheral blood samples and
histopathological examination did not show dermal and
epidermal infiltration of atypical lymphocytes either.) So,
for patients with a previous history of known dermatoses
whose clinicopathological features are inconclusive, we
recommend careful investigation to rule out any underly-
ing malignant causes. In our study, after thorough ex-
amination, we found one patient with lung cancer and
one patient with tongue cancer during follow-up, which
has never been reported in the previous literature. We
also found one case of Langerhans cell histiocytosis
through skin and lymph node biopsies, which has never
been mentioned in previous reports either.
The true reason for 37 patients with erythroderma re-
mained elusive. This group is also called idiopathic eryth-
roderma, or red man syndrome [17] , but we prefer the
name undetermined erythroderma because the etiology
will be identified in time. Some authors think these pa-
tients may represent a premalignant phase [1] . Close fol-
low-up might allow doctors to identify the causes and to
initiate earlier treatment.
Comparison of the etiologic groups among recent
studies and our own is given in table4 . This table reveals
Erythroderma: A Clinical and Prognostic
Study
Dermatology 2012;225:154162 161
some differences in relative incidence of the different eti-
ologic subgroups of erythroderma. This may be partly
related to a genetic, geographical and social disparity.
We obtained follow-up information from 125 pa-
tients. Other patients could not be contacted. The high
number of patients lost to follow-up and the short period
of follow-up constitute some bias for the present study.
However, despite these limitations, it provides us with
information related to prognosis and survival. Indeed,
we recorded a high overall relapse rate (31.2%) indepen-
dently of etiology. The patients with idiopathic erythro-
derma relapsed much more quickly than those caused
by psoriasis. The explanation as to why the patients with
specific causes of erythroderma relapsed more quickly
than those induced by psoriasis may lie in the treatment
(unlike psoriatic erythroderma, which can be well con-
trolled by acitretin, idiopathic erythroderma remains
mostly undiagnosed in follow-up in our study; specific
treatment occurs in too small fractions of patients and
the disease reoccurred easily). In initial documented
studies, the recorded death rate due to erythroderma
varied from 18 to 64% [23] . Up to now, we found that
only 5 patients with erythroderma died. Most of our pa-
tients were cured or improved. Our findings therefore
support Hasan and Jansens [2] and Akhyani et al.s [15]
view that erythroderma, although often bringing dis-
tress to patients, and sometimes even being life-threat-
ening, does not pose a significant risk to the patients
life.
We acknowledge several strengths but also some lim-
itations of our study. First, the data originated from a
retrospective study design, which is generally consid-
ered inferior to prospective study design. Second, varia-
tions in the use of medical therapies before admission to
hospital and after discharge may have influenced the
results of our study. Third, a short follow-up period and
the high number of patients lost to follow-up constitute
some bias for our study. Over the 11-year period study,
of the 260 patients, 135 could not be contacted. How-
ever, despite these limitations, the study provides us
with information related to clinical features and prog-
nosis of erythroderma. Long-term prognostic assess-
ment, close follow-up and better study design are need-
ed in future studies.
In conclusion, erythroderma is a fairly uniform clini-
cal syndrome of generalized erythema and desquama-
tion. Most of the clinical features of erythroderma are un-
specific, with few cause-orienting clues. Although numer-
ous laboratory values were abnormal, most findings were
nondiagnostic and were related to the inflammatory pro-
cess except for skin biopsy. Though the processes that ini-
tiate erythroderma in different patients may be highly
variable, most of the patients with erythroderma have a
pre-existing skin disease. Patients with previously normal
skin often have drug-induced erythroderma or malignan-
cy. No underlying cause of erythroderma was found in up
to 14.23% of patients in our study. Repeated evaluations,
close follow-up and repeat biopsy are recommended.
Table 4. C auses of erythroderma in previous studies compared with the present study
Reference, year Patients Pre-existing
dermatoses, %
Drug
reaction, %
CTCL,
%
Paraneo -
plastic, %
Miscella -
neous, %
Idio pathic,
%
Vasconcellos et al. [18] (1995) 247 59 7.3 4.1 0 0.4 29.2
Morar et al. [19] (1999) 138 65.3 22.5 2.2 0 0 10
Nicolis and Helwig [16] (1973) 135 27 40 8 3 10 12
Sigurdsson et al. [20] (1996) 102 53 5 13 2 1 26
Akhyani et al. [15] (2005) 97 59.8 21.6 10.3 1.1 0 7.2
Pal and Haroon [1] (1998) 90 74.4 5.5 5.5 0 0 14.6
King et al. [8] (1986) 82 31 34 18 0 1 16
Khaled et al. [21] (2010) 82 47.56 21.95 4.88 0 0 25.61
Yuan et al. [12] (2010) 82 72 17 3.67 1.23 0 6.1
Rym et al. [3] (2005) 80 67.5 11.25 8.75 0 5 7.5
Sehgal et al. [4] (2004) 80 58 20 0 0 0 22
Botella-Estradas et al. [5] (1994) 56 66 12.5 12.5 0 0 9
Hasan and Jansen [2] (1983) 50 54 10 4 0 0 32
Leenutaphong et al. [22] (1999) 49 26.5 38.77 0 2.06 0 32.67
Present study (2012) 260 70.77 12.69 1.15 1.15 0 14.23
Li/Zheng

Dermatology 2012;225:154162 162
Acknowledgement
We thank the patients, physicians (staff and investigators), and
study coordinators for their hard work. Hong-Zhong Jin, MD,
Qiu-Ning Sun, MD, Yue-Hua Liu, MD, Dong-Lai Ma, MD, from
Department of Dermatology, Peking Union Medical College Hos-
pital, Chinese Academy of Medical Sciences and Peking Union
Medical College, Beijing, China provided insightful suggestions
at several stages of this study.
Disclosure Statement
The authors have no conflicts of interest to disclose.

References
1 Pal S, Haroon TS: Erythroderma: a clinico-
etiologic study of 90 patients. Int J Dermatol
1998; 37: 104107.
2 Hasan T, Jansen CT: Erythroderma: a follow-
up of fifty cases. J Am Acad Dermatol 1983;
8: 836840.
3 Rym BM, Mourad M, Bechir Z, et al: Eryth-
roderma in adults: a report of 80 patients. Int
J Dermatol 2005; 44: 731735.
4 Sehgal VN, Srivastava G, Sardana K, et al:
Erythroderma/exfoliative dermatitis: a syn-
opsis. Int J Dermatol 2004; 43: 3947.
5 Botella-Estradas R, Sanmartin O, Oeiver V,
et al: Erythroderma: a clinicopathological
study of 56 patients. Arch Dermatol 1994;
130: 15031507.
6 Zip C, Murray S, Walsh NMG: The specific-
ity of histopathology in erythroderma. J Cu-
tan Pathol 1993; 20: 393398.
7 Rothe MJ, Bialy TL, Grant-Kels JM: Erythro-
derma. Dermatol Clin 2000; 18: 405415.
8 King LE Jr, Dufresne RG Jr, Lowelt GL, et al:
Erythroderma review of 82 patients. South
Med J 1986; 79: 12101215.
9 Walsh NM, Prokopetz R, Tron VA, et al: His-
topathology in erythroderma: review of a se-
ries of patients by multiple observers. J Cutan
Pathol 1994; 21: 419423.
10 Chen XY, Jin LW, Chen YW, et al: The asso-
ciation between the IL-20 1723C ] G allele
on the 1q chromosome and psoriasis trig-
gered or exacerbated by an upper respiratory
tract infection in the Chinese Han popula-
tion. Dermatology 2011; 222: 2430.
11 Ormerod AD, Augustin M, Baker C, et al:
Challenges for synthesising data in a net-
work of registries for systemic psoriasis ther-
apies. Dermatology 2012; 224: 236243.
12 Yuan XY, Guo JY, Dang YP, et al: Erythro-
derma: a clinical-etiological study of 82 pa-
tients. Eur J Dermatol 2010; 20: 373377.
13 Maruani A, Armingaud P, Nseir A, et al:
Erythroderma and multiple cutaneous ne-
crosis revealing a dermatomyositis. Ann
Dermatol Venereol 2003; 130: 353356.
14 Miyagawa S, Okazaki A, Minowa R, et al:
Dermatomyositis presenting as erythroder-
ma. J Am Acad Dermatol 1992; 26: 489490.
15 Akhyani M, Ghodsi ZS, Toosi S, et al: Eryth-
roderma: a clinical study of 97 patients. BMC
Dermatol 2005; 9: 5.
16 Nicolis GD, Helwig EB: Exfoliative dermati-
tis. A clinicopathologic study of 135 patients.
Arch Dermatol 1973; 108: 788797.
17 Thestrup-Pedersen K, Halkier-Srensen L,
Sgaard H, et al: The red man syndrome/ex-
foliative dermatitis of unknown etiology: a
description and a follow up of 38 patients. J
Am Acad Dermatol 1988; 18: 13071312.
18 Vasconcellos C, Domingues PP, Aoki V, et al:
Erythroderma: analysis of 247 cases. Rev
Saude Publica 1995; 29: 177182.
19 Morar N, Dlova N, Gupta AK, et al: Erythro-
derma: a comparison between HIV positive
and negative patients. Int J Dermatol 1999;
38: 895900.
20 Sigurdsson V, Toonstra J, Hazemans-Boer
M, et al: Erythroderma a clinical and fol-
low-up study of 102 patients with special em-
phasis on survival. J Am Acad Dermatol
1996; 35: 5357.
21 Khaled A, Sellami A, Fazaa B: Acquired
erythroderma in adults: a clinical and prog-
nostic study. J Eur Acad Dermatol 2010; 24:
781788.
22 Leenutaphong V, Kulthanan K, Pohboon C,
et al: Erythroderma in Thai patients. J Med
Assoc Thai 1999; 82: 743748.
23 Sehgal VN, Srivastava G: Exfoliative derma-
titis. A prospective study of 80 patients. Der-
matologica 1986; 173: 278284.

Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.