Bullous erysipelas: A retrospective study

of 26 patients
Dror Guberman, MD,a* Leon T. Gilead, MD, PhD,b* Abraham Zlotogorski, MD,b and
Julian Schamroth, MDb Jerusalem, Israel

Background: Erysipelas is a superficial form of cellulitis caused by a variety of microbes, and it responds
to antibiotic treatment. During the past few years we treated several patients with a bullous form of
erysipelas involving the lower legs. We believe their disease had a more protracted course than patients
with nonbullous erysipelas.
Objective: We studied bullous erysipelas by conducting a retrospective analysis of 26 patients with bullous
erysipelas of the legs treated by the authors during a 5-year period.
Methods: We conducted a retrospective review of the records of all patients with a diagnosis of bullous
erysipelas who were treated at the Department of Dermatology, Hadassah Medical Center, Jerusalem,
between the years 1992 and 1996. Data regarding patients with nonbullous erysipelas were obtained from
the medical center’s computerized data pool.
Results: A total of 26 cases of bullous erysipelas were found, comprising 22 women and 4 men whose ages
ranged from 28 to 87 (mean, 58.8) years. The average hospital stay was 20.57 days (range, 12 to 46 days).
The average hospital stay for patients with nonbullous erysipelas and cellulitis treated in the same
department by the authors during the study period was 10.6 days (range, 2 to 54 days).
Conclusion: Bulla formation is a complication of erysipelas, seen in our series in 5.2% of the patients (26
of 498 admissions for erysipelas and cellulitis). The course of the disease is protracted, requiring longer
medical attention. (J Am Acad Dermatol 1999;41:733-7.)

E rysipelas is a superficial form of cellulitis

caused by a variety of microbes; it responds to
antibiotic treatment. The typical clinical pre-
sentation includes leg tenderness, sharply demarcat-
ed erythema, and edema. Lymphangitis and inguinal
lymphadenopathy may be present, and the onset is
usually accompanied by fever and sometimes shiver-
ing.1 During the past few years we have treated sev-
eral patients with a bullous form of erysipelas involv-
ing the lower legs. We believe they had a more pro-
tracted course of the disease than did patients with
nonbullous erysipelas. This is a retrospective study
of 26 patients with bullous erysipelas of the legs
treated by the authors over a 5-year period.
From private practicea and the Department of Dermatology,
Hadassah Medical Center.b
*The first two authors are equal contributors.
Accepted for publication April 27, 1999.
Reprint requests: Dror Guberman, MD, 11 Gordon St, Jerusalem,
Israel 96545.
Copyright © 1999 by the American Academy of Dermatology, Inc.
0190-9622/99/$8.00 + 0 16/1/99552
METHODS
We conducted a retrospective review of all records of
patients with a diagnosis of bullous erysipelas who were
treated at the Department of Dermatology, Hadassah
Medical Center, Jerusalem, between the years 1992 and
1996. Data regarding patients with nonbullous erysipelas
were obtained from the medical center’s computerized
data pool.
RESULTS
The parameters evaluated were summarized in
Tables I and II for prehospitalization course and in-
hospital course, respectively. The 26 patients were 22
women and 4 men ranging in age from 28 to 87 years
(mean, 58.8 years). All patients had erysipelas mani-
fested by fever, leg pain, erythema, edema, increased
local warmth and tenderness, and tender inguinal
lymphadenopathy, with the exception of 6 patients
(1 who was admitted for treatment of post-bullous
erysipelas erosions and 5 who were afebrile on
admission). Predisposing leg conditions, such as pre-
vious erysipelas, thrombophlebitis or bone fracture,
onychomycosis or tinea pedis, venous insufficiency

733
734 Guberman et al J AM ACAD DERMATOL
NOVEMBER 1999

Table I. Prehospitalization course

Illness before
Patient Age (y)/ admission
No. Sex Medical history Predisposing leg factors Treatment (days) Bullae onset

1 53/F Cephalexin 3 During hosp

2 51/F Fracture left femur, sciatic Erythromycin 3 During hosp
nerve injury, previous
erysipelas
3 61/F Diabetes mellitus type 2, Previous erysipelas Doxycycline 1 During hosp
hypertension, atrial
fibrillation
4 42/M Gout Deep vein thrombosis of 3 During hosp
right leg, surgery for foot
malalignment
5 82/F Diabetes mellitus type 2, Amoxicillin, 3 Pre hosp
hypothyroidism dicloxacillin
6 44/F Iron deficiency Erythromycin 4 During hosp
7 70/F Bronchial asthma, Onychomycosis of toe nails 1 During hosp
osteoarthritis
8 87/F Hypothyroidism Stasis dermatitis 1 During hosp
9 49/F Healed fracture of left ankle, Amoxicillin, 8 Pre hosp
deep vein thrombosis, cephalexin
tinea pedis
10 85/M Laryngectomy for Tinea pedis Cephalexin, 1 During hosp
laryngeal carcinoma amoxicillin/
clavulanic acid
11 42/F Obesity Penicillin 3 During hosp
12 78/F Duodenal ulcer Trauma 7 During hosp
13 83/F Carcinoma of vulva, Drainage of inguinal 7 During hosp
diabetes mellitus type 2 abscess
14 34/F Dicloxacillin 5 Pre hosp
15 43/F Dry eyes Poliomyelitis, tinea pedis Amoxicillin, 2 Pre hosp
dicloxacillin
16 75/F Ischemic heart disease, Venous insufficiency 2 Pre hosp
hypertension
17 47/F Anemia Recurrent erysipelas Erythromycin 4 During hosp
18 64/F Diabetes mellitus type 2, Cefuroxime 3 Pre hosp
hypertension, obesity
19 77/F Arthralgia Tinea pedis Doxycycline 7 Pre hosp
20 66/F Thalassemia minor, Recurrent erysipelas, Cefuroxime 3 During hosp
hypertension tinea pedis
21 75/M Diabetes mellitus type 2, Tinea pedis Penicillin, 4 Pre hosp
obstructive lung disease cephalexin
22 63/F Stasis dermatitis, varicose Amoxicillin/ 14 Pre hosp
veins, trauma clavulanic acid
23 28/F Anemia (iron deficiency) Amoxicillin/ 1 During hosp
clavulanic acid,
dicloxacillin
24 39/M Cefazolin 1 During hosp
25 38/F Diabetes mellitus type 2 Trauma Cefazolin 6 During hosp
26 54/F Asthma, carcinoma of Chronic edema Dicloxacillin 30
the uterus

or varicosed veins, or stasis dermatitis, were found in Twenty of the patients were partially treated with
18 patients, and non-insulin–dependent diabetes oral antibiotics before admission. Bullae occurred
mellitus in 6 patients. In only 3 patients were no sys- during hospitalization in 16 patients and the rest
temic or local predisposing factors identified. were hospitalized with the bullous lesions already
J AM ACAD DERMATOL Guberman et al 735
VOLUME 41, NUMBER 5, PART 1

Table II. In-hospital course

Patient Blood Duration
No. culture Bulla culture Treatment* Complications (days)

1 Sterile Sterile Cefazolin 19

2 Not done Mixed enteric flora Dicloxacillin 30
3 Sterile Not done Cefuroxime, metronidazole 16
4 Not done Not done Cefazolin 14
5 Not done Sterile Cefuroxime and metronidazole 17
(diarrhea), cefazolin
6 Sterile Not done Clindamycin (allergy), cefazolin 17
7 Sterile Sterile Cefazolin 46
8 β-Hemolytic Coagulase (-) Cefazolin 17
strep† staph‡
9 Not done Not done Amoxicillin/clavulanic acid 13
10 Sterile§ Sterile Cefazolin, amoxicillin/clavulanic acid
11 Not done Not done Cefazolin 16
12 Sterile Not done Dicloxacillin Leg ulcer 45
13 Sterile Sterile Ceftazidime, vancomycin, cefamezine, 20
metronidazole
14 Sterile Sterile Cefazolin 14
15 Not done Sterile Cefazolin 14
16 Not done Sterile Cefazolin, penicillin, dicloxacillin 14
17 Sterile Not done Clindamycin (rash) 14
18 Sterile Sterile Cefuroxime, metronidazole, clindamycin, 34
ciprofloxacin
19 Sterile Sterile Cefazolin GIT bleeding (NSAID) 25
20 Sterile Sterile→coag Cefazolin, metronidazole, clindamycin Hemolysis, leg ulcer 44
(+) staph
21 Not done Not done Cefuroxime, metronidazole, dicloxacillin 14
22 Not done Sterile Cefazolin 13
23 Sterile Not done Cefazolin 16
24 Sterile Not done Cefazolin 18
25 Not done Not done Cefazolin 16
26 Not done Not done Dicloxacillin 12

GIT, Gastrointestinal tract.

*Topical therapy included aluminum acetate or saline compresses, 1% silver sulfadiazine cream or 1% bacitracin ointment dressing, and
mechanical whirlpool debridement.
†Streptococcus.
‡Staphylococcus.
§Anti-streptolysin titer 1:320.

present. The bullae were all flaccid bullae several
centimeters in diameter, and a few of them con-
tained blood (Fig 1).
Despite several attempts to isolate a pathogen,
the causative bacteria was found in the blood of only
1 patient (case 8) out of 15 patients. This patient,
who presented with confusion and severe systemic
signs, was found to be infected with a beta-hemolyt-
ic streptococcus. Of 14 attempts to isolate a
pathogen from the eroded bullae, all early (pre-ero-
sion) cultures were sterile. In subsequent cultures,
only in 1 case was there a finding considered to be
significant: a coagulase-positive Staphylococcus
aureus. The coagulase-negative staphylococci and
mixed enteric flora isolated from the eroded bullae
of 2 other patients were probably the result of cont-
amination.
The initial parenteral in-hospital antibiotic regi-
men included cefazolin in 15 patients, a combination
of cefuroxime and metronidazole in 4 patients, clin-
damycin in 4 patients, and dicloxacillin in 3 patients.
In addition amoxicillin/clavulanic acid were adminis-
tered in 2 patients, and vancomycin, ciprofloxacin,
and ceftazidime were each given to 1 patient. A
change in antibiotic therapy was necessary because
of an allergic reaction (2 cases), drug intolerance (1
case), and an unsatisfactory therapeutic response (6
cases). The average time for defervescence was 5.15
days (range, 2 to 12 days). The average hospital stay
was 20.57 days (range, 12 to 46 days). In all 26 cases,
736 Guberman et al
Fig 1. Leg of patient with bullous erysipelas. Large denuded area on lateral aspect of calf with
remnants of flaccid bullae present at edges.
the course of the disease was protracted without any
other causes influencing its length except for the bul-
lous erysipelas and the concomitant erosions; many
cases required 3 or more weeks of hospitalization. All
patients were reported to have suffered significant
discomfort and pain because of the bullous complica-
tion. Recovery was complete in all patients, although
2 patients were subsequently rehospitalized for recur-
rent (nonbullous) erysipelas, and 2 patients suffered
from postbullous ulcerations. None of the 26 patients
had necrotizing fasciitis or organ failure.
A retrospective analysis of data regarding all
patients admitted for erysipelas and cellulitis to the
department of dermatology during the designated 5-
year period was performed. A total of 472 patients
were admitted to the department of dermatology
during that time. The average age of the patients was
53.4 years (range, 15 to 93 years), and the female to
male ratio was 43.4% (216/498). The average hospital
stay was 10.6 days (range, 2 to 54 days). No signifi-
cant difference was found in the rate of the various
predisposing factors compared with those found in
the bullous erysipelas group. No significant differ-
ences were noted in the extent of skin involvement
in bullous erysipelas as compared with that seen in
cases where bullae did not occur.
DISCUSSION
The bullous form of erysipelas is mentioned only
briefly in the textbooks: “Bullae and vesicles may devel-
op.”2,3 Surprisingly, the medical literature regarding
bullous erysipelas or bullous cellulitis is scant. Of 100
patients in one retrospective study, 30 experienced bul-
J AM ACAD DERMATOL
NOVEMBER 1999
lae.4 Agnholt, Andersen, and Sondergaard5 reported a
fatal case of bullous erysipelas in which culture of
blood and cutaneous swabs revealed the presence of
beta-hemolytic streptococcus. They regarded it as a
new cutaneous manifestation of streptococcal infec-
tion. Other reports of invasive streptococcal infections
have been described, including 1 case of bullous cel-
lulitis in the leg of a surgeon who was exposed to peri-
toneal fluid of a streptococcus-infected patient.6
Streptococci and staphylococci are the most com-
mon causes of erysipelas in immunocompetent
patients. However, other bacteria may cause bullous
cellulitis. In a study of 7 patients with cirrhosis and
cellulitis, Corredoira et al7 described 5 events of bul-
lous erysipelas. Various gram-negative enteric organ-
isms were cultured from needle aspiration of cuta-
neous lesions. Bacteremia was found in 6 patients
and the disease was fatal in 4. Bacteremic bullous cel-
lulitis may also be caused by the marine bacteria
Vibrio vulnificus,8 and has been caused by strepto-
coccus pneumonia.9 Hemorrhagic cellulitis was
referred to as a separate entity with significant mor-
tality, and glucocorticosteroids were believed to be
needed.10 Only mild complications of bullous
erysipelas such as milia11 were described to date.
The mechanism of bulla induction remains
unclear. Although a biopsy of a bulla could probably
contribute to the understanding of the mechanism
and etiology, skin biopsies were contraindicated
because of the significant probability of a secondary
ulceration in an already severely ill limb. However, all
bullae were flaccid and therefore were clinically con-
sidered intraepidermal, and almost all of them cleared
J AM ACAD DERMATOL
VOLUME 41, NUMBER 5, PART 1
without scarring. We can therefore only speculate on
the basis of the present data that the bulla is pro-
duced by severe intraepidermal spongiosis caused by
the superficial dermal and epidermal infectious and
inflammatory process. It is possible that the extent of
skin involvement in the basic infection is significant in
the occurrence of the bullae; however, our clinical
impression is that this was not a significant factor. The
lymphangitis typically seen in erysipelas may also be a
factor, because lymphatic drainage is impaired. Bullae
may also be formed as a result of direct bacterial inva-
sion. However, this possibility is probably ruled out
because blister fluid cultures are usually sterile or
grow irrelevant local contaminating bacteria. Toxin-
mediated blister formation for bullous infections,
including bullous impetigo and scalded skin syn-
drome, is another possible mechanism. In these dis-
eases an exotoxin activity results in a subcorneal sep-
aration within the epidermis. Bacterial elements may
also function as superantigens, and thus activate
inflammatory cascades.12 Mechanical pressure may be
a contributory factor in bullae production, such as
that occurring in friction blisters.
Our failure to culture pathogens from intact bul-
lae is not surprising. It is known that the yield of cul-
tures in erysipelas is very low. Although blood cul-
tures were reported to identify the responsible
pathogen in only 5%,13 needle aspiration cultures
give about 5% to 12.5% positive and relevant
results.13 Lesional skin biopsy cultures yield 10% to
20% bacterial isolations.12 Although the yield is low,
cultures of both blood and bullae fluid should prob-
ably be performed in all such patients.
The retrospective nature of our study and the lack
of consistency in initial antibiotic therapy through-
out the reviewed years prevents us from assessing
the efficacy and significance of the various antibiotics
and topical medications that were used. However,
because the treatments in the control population
were the same, no relationship between the type of
systemic treatment and final outcome was identified.
Topical treatment seems important in bullous
erysipelas. Our therapeutic approach included main-
ly leg elevation. Whirlpool jets were used to debride
some of the lesions, and dressings with either 1%
bacitracin or 1% silver sulfadiazine were applied top-
ically. Subcutaneous enoxaparin was administered
for deep vein thrombosis prophylaxis for patients
with previous thrombosis or morbid obesity. We
have treated macerated interdigital tinea pedis, pre-
scribed elastic support stockings where not con-
traindicated, and used antibiotic prophylaxis in
patients suffering from multiple recurrent erysipelas.
The control population, although not matched,
serves as a statistically significant control group
Guberman et al 737
because of its size and intrinsic properties. The differ-
ence in the length of hospitalization (20.57 days in bul-
lous erysipelas patients as compared with 10.6 in the
control population is of high statistical significance (P
< .001). The mean age difference between the 2
groups is not statistically significant, and no other rel-
evant significant differences were found between
groups. We believe the striking 84.6% female predom-
inance (22/26 patients with bullous erysipelas) in the
bullous erysipelas group as compared with the 43.4%
seen in the control population is coincidental.
Despite the retrospective nature of this study,
which prevents the drawing of significant conclu-
sions for many of the questions it may raise, we
believe that bullous erysipelas is a significant compli-
cation of severe erysipelas induced by a yet
unknown mechanism that, as this series of cases
indicates, results in a protracted hospitalization and
significant discomfort to the patients. The lack of
publications dealing with this complication is sur-
prising especially in view of its relative frequency
(5.2% of the patients [26/498 admissions for
erysipelas and cellulitis] in our series).
REFERENCES
1. Braverman IM. Skin signs of systemic disease. Philadelphia:
Saunders; 1998. p. 577-81.
2. Swartz MN, Weinberg AN. Infections due to gram-positive bac-
teria. In: Fitzpatrick TB, Eisen AZ, Wolff K, Freedberg IM, Austen
KF, editors. Dermatology in general medicine. New York:
McGraw-Hill Inc; 1993. p. 2313-5.
3. Stevens DL. Infections of the skin, muscle, and soft tissues. In:
Isselbacher KJ, Braunwald E, Wilson JD, Martin JB, Fauci AS,
Kasper DL, editors. Harrison’s principles of internal medicine.
New York: McGraw-Hill Inc; 1994. p. 562.
4. Crickx B, Chevron F, Sigal-Nahum M, Bilet S, Faucher F, Picard C,
et al. Erysipelas: epidemiological, clinical, and therapeutic data
(111 cases). Ann Dermatol Venereol 1991;118:11-6.
5. Agnholt J, Andersen I, Sondergaard G. Necrotic bullous
erysipelas. Acta Med Scand 1988;223:191-2.
6. Schwartz B, Elliott JA, Butler JC, Simon PA, Jameson BL, Welch
GE, et al. Clusters of invasive group A streptococcal infections in
family, hospital, and nursing home settings. Clin Infect Dis
1992;15:277-84.
7. Corredoira JM, Ariza J, Pallares R, Carratala J, Viladrich PF, Ruff J,
et al. Gram-negative bacillary cellulitis in patients with hepatic
cirrhosis. Eur J Clin Microbiol Infect Dis 1994;13:19-24.
8. Penman AD, Lanier DC, Avara WT, Canant KE, DeGroote JW,
Brackin BT, et al. Vibrio vulnificus wound infection from the
Mississippi gulf coastal waters: June to August 1993. South Med
J 1995;88:531-3.
9. House NS, Helm KF, Marks JG Jr. Acute onset of bilateral hemor-
rhagic leg lesions. Arch Dermatol 1996;132:81-6.
10. Heng MC, Khoo M, Cooperman A, Fallon-Friedlander S.
Haemorrhagic cellulitis: a syndrome associated with tumor
necrosis factor-alpha. Br J Dermatol 1994;130:65-74.
11. Lapidoth M, Hodak E, Segal R, Sandbank M. Secondary milia fol-
lowing bullous erysipelas. Cutis 1994;54:403-4.
12. Sachs MK. Cutaneous cellulitis. Arch Dermatol 1991;127:493-6.
13. Goldgeier MH. The microbial evaluation of acute cellulitis. Cutis
1983;31:649-56.