Erythroderma: Analysis of 247 Cases

The profile of 247 patients with erythroderma during a 23 year period from January, 1962 through March, 1985, with a follow-up period ranging from 1 to 26 years were analysed. The patients presented with diffuse erythema, scaling and pruritus of more than 2 months' duration, and the age ranged from 16 to 60 years. Psoriasis was the most frequent underlying disease with an estimated frequency of 44.9%, the reaction to the use of drugs appeared in 7.3% of total cases and association with reticulosis showed a frequency of 4.1%. The cause of the erythroderma could not be determined in 29.2% of the cases. Sex differences in terms of underlying diseases were not observed. One or more skin biopsies along with the clinical findings were diagnostic or suggestive of the underlying disease in 63.6% of the cases. Repeated skin biopsies are recommended as the best method for etiologic diagnosis of erythroderma. At P=0.05 significance level, masculine/feminine ratio of 2 : 1 was found. The question arises wether causal agent of erythroderma may not be somehow related to different exposure by sex to environmental antigens. Dermatitis, exfoliative; epidemiology incidence.

Montgomery 12 established another classification based on histopathologic findings: a) idio-pathic or primary; and b) secondary to an underlying dermatosis or reticulosis.
According to Rabello et al. 16 , the patients with erythroderma do not show any significant alteration of routine laboratory tests.Montgomery 12 stressed usefulness of multiple skin biopsies as a diagnostic and follow-up method.
Treatment must be directed towards the underlying diseases, although general measures should not be forgotten (Nebenzahl et al) 13 .
The objective of the present work is to profile the etiology of erythroderma and to verify if there was a predominance of the syndrome according to sex.

Subjects and methods
Two hundred and forty seven patients with erythroderma seen at Dermatology Unit of a Hospital of Clinics located at the city of S. Light microscopy study of skin biopsies was done and showed one of the following patterns: a) Subacute or chronic dermatitis; b) Subacute or chronic psoriasiform dermatitis; c) Psoriasis; d) Vasculitis; e) Suggestive of lymphoma; f) Others: lamellar ichthyosis, epidermolytic hyperkeratosis and Darier's disease, and g) Dermatopathic lymphadenitis.
The study has been designed as a case study* (Forattini 5 and Leser et al). 11the subject group was compared to patients with dermatosis withouterythroderma from the same hospital during the same time period.
The Special Program for Social Science (SPSS) was used for the analysis of the data.
In order to compare the ratio between three or more categories in the contingency tables, Goodman's 8 or Kolmogorov-Smirnoffs (apud Costa-Neto) 3 tests, along with Pearson's chisquare statistics (Costa-Neto) 3 were applied, respecting the frequency restrictions.
As for the tendency evaluation, adjustment by the method of minimum square and the regression line was adopted, and the angular coefficient test was applied, according to Student's t distribution, with the necessary adjustments.The comparison between two angular coefficients, with the same adjustment, was possible due to Student's t distribution (Costa-Neto) 3 .The significance level adopted was P=0.05.

Results
Table 1 shows a greater number of feminine patients in the group without erythroderma (t calc =5.051; c.p. (cut point)=2.131),and a greater number of masculine patients in the group with erythroderma (t calc =0.295; c.p.=2.074),however, this difference has no statistical significance (D max =0.140;c.p.=0.163).
Although, when patients were sub-divided into sub-groups of 5 year periods (Table 2), according to the time of the first visit, a significantly higher number of masculine patients with erythroderma was found (Table 2; 1st. 5 year-period: c 2 calc =5.156; 2nd. 5 year-period: c 2 calc =15.751; 3rd. 5 year-period c 2 calc =10.309; c.p.=3.841).Table 3 shows no differences by sex among patients without erythroderma who sought assistance in the outpatient clinic.
The patients with erythroderma (Table 4) showed the following distribution: in the group of from 0 to 15 years old, there was a predominance of feminine patients (Goodman G statis-tics=3.790; c.p.=2.448), and amongst those over 61 years of age, there was a predominance of masculine patients (Goodman G statistics = 2.571; c.p. = 2.448).The highest number of patients was found to be in the group of from 16 to 60 years old.
Amongst the masculine group (Table 5), erythroderma was more frequent in the group of from 60 to 69 years old (Goodman G statis-tics=3.885; c.p.=3.751), while amongst the female group it was more frequent in the 10 to 19 year age group.
The majority of the patients were born in S. Paulo state (44.6%), in spite of the fact that the patients' places of birth have not been recorded in about 40.0% of the cases (Table 6).
Applying the Goodman's test to Table 7 (Goodman's g 2 =2.448; c.p.=12.592),no significant difference was found among the ratio of various causes, by sex.
A group of men between the 3rd and 6th decades, inclusive, of life, and of women between 1st and 5th decades, inclusive, with psoriasis, was identified.Cutaneous eruption caused by the use of drugs was more frequent between the 2nd and 5th decades in both sexes (Table 8).

Comments
In the literature, there are few notes analyzing erythroderma under its multiple aspects (Abrahams et al 1 , Adam 2 , Gatti et al 6 , Gentele et al 7 , Hasan & Jansen 9 , Montgomery 12 , Nicolis & Helwig 14 , Nigam et al 15 and Rabello et al). 16nitially, the group of patients with erythroderma and the group of patients with other dermatosis were compared, in an attempt to find a sampling bias that might have resulted in a higher number of masculine patients in the group with erythroderma.
As no difference by sex was found when the population without erythroderma was studied (Table 1 and 3), it was concluded that the predominance of masculines in the group with erythroderma was not due to any sampling bias.This result is in agreement with those of other authors, ranging from 2: 1 (Adam 2 ) to 11: 1 (Nicolis & Helwig). 14n more than 59.0% of the total cases there had been a prior manifestation of some dermatosis, of these cases, 44.9% had had psoriasis, 4.0% reticulosis/leukemia, and about 30.0%unknown causes (Table 7), with no difference among the ratio of the vaious causes according to sex.Therefore, in this population, the underlying disease was not responsible for the difference observed between the sexes in the group of patients with erythroderma (2 masculine: 1 feminine).
The literature shows that there is a predominance of some dermatosis as a prior manifestation in the group of patients with erythroderma (Abrahams et al 1 ; Adam 2 ; Gatti et al 6 , Gentele et al 7 and Hasan & Jansen). 9In literature, psoriasis is the most frequent dermatosis (25.0 to 60.0%), although no sensu strictu comparisons between the literature and our data could be made, since the diagnosis and criteria of inclusion varied so much.This is perhaps the cause of the predominance of cases of drug reaction in Nicolis & Helwig's study 14 and the great number of cases of reticulosis in Montgomery's study 12 .
The frequency of unknown causes with erythroderma varied from 8.0% (Montgomery) 14 to 55.0% (Rabello), 16 against 29.0% in this present study.It is interesting to note that Montgomery 12 had undertaken more skin biopsies than any other author.
It is not easy to compare the data of cases related to the use of drug in the present study with those recorded in the literature because of the great number of drugs involved and the difference in the criteria of inclusion.There are many available drugs and they vary from one country to another and from one time to another, as observed by Nigam et al 15 .Moreover, some authors consider any drug used before the appearance of erythroderma as the causal agent, whereas others do not do so when a primary dermatosis is present.
Only 19.4% of the cases gave any information regarding to the use of drugs prior to the development of erythroderma, and even in these cases, no information about the means of administration or dosage was available.The history of the use of corticosteroid just before the manifestation of erythroderma raised the question as to a possible cause in the inadequate treatment of a previous dermatosis.
All six patients involved with agrotoxic products were men (Tables 9 and 10) of from 30 to 60 years old and came from the countryside.In spite of the small sample, it was asked wether environmental antigens might not be causally related to erythroderma, and whether the predominance of masculine patients amongst the group with erythroderma could be due to their work-related exposure to the environmental antigens.
The data from this research corroborate the previous findings of Montgomery's 1933 study 12 that skin biopsies, when performed in sufficient numbers, and under appropriate conditions, are still the best mean of diagnosis and follow-up for patients with erythroderma.Skin biopsy is specially useful in the diagnosis or exclusion of an early reticulosis, since the latter process shows earlier changes in the skin than in other organs (Montgomery). 12nother biopsy is recommended when a patient shows a poor therapeutic response and when the histologic feature is not cleared-up, as may happen the differential diagnosis of Sézary's erythroderma versus pre-Sézary's erythroderma, mycosis fungoides versus persistent reaction to insect bites, and nodular scabies versus mycosis fungoides or Hodgkin's lymphoma (Montgomery). 12n striving for an ideal biopsy, the following advice from Montgomery 12 should always be kept up in mind: whenever possible, biopsies should be done on an untreated active area, away from the abdomen or seborrheic areas.