Erythema nodosum: etiologic and predictive factors in a defined population

To examine the frequency and features of erythema nodosum (EN), establish disease associations, and identify the optimal set of predictors for the occurrence of secondary EN.

We performed a retrospective study of an unselected population of patients 14 years and older with biopsy-proven EN diagnosed at a referral hospital between 1988 and 1997. Patients were classified as having either idiopathic EN or EN secondary to other diseases if the skin nodules occurred in the context of a well-defined disease, or if there was a precipitating event in close temporal proximity to the onset of EN.

One hundred six patients (82 women) were diagnosed as having biopsy-proven EN. At the time of diagnosis, no precipitating events or underlying diseases were identified in 36.8% of patients. Sarcoidosis and nonstreptococcal upper respiratory tract infections (URI) were the most common conditions associated with secondary EN. Only 1 of 35 patients with an initial diagnosis of idiopathic EN and a followup of at least 1 year was finally diagnosed as having secondary EN. The best predictive model of secondary EN included an abnormal results on a chest radiograph, a previous history of nonstreptococcal URI, and a significant change in antistreptolysin O (ASO) titer in 2 consecutive determinations performed in a 2-4-week interval. Also, the presence of peripheral synovitis, a positive tuberculin skin test, and a history of diarrhea suggested the presence of secondary EN. This model showed high sensitivity and specificity.

Idiopathic EN is common. A basic procedure including careful medical history-taking, a physical examination for peripheral synovitis, 2 consecutive ASO determinations, a tuberculin skin test, and chest radiography may be sufficient to diagnose EN.