To determine whether recent corticosteroid use was associated with an increased risk of complicated varicella-zoster virus infection in otherwise immunocompetent children.
A case-control study design was used because the outcome of interest, complicated varicellazoster virus infection, is rare.
Cases and controls were selected from the population of children aged 2 months to 18 years admitted to two hospitals, between January 1979 and July 1994 in one and between January 1974 and July 1994 in the other, with diagnosis codes that indicated chickenpox.
Cases were defined as children with invasive varicella-zoster virus infection or associated invasive bacterial infection. Controls were defined as children with uncomplicated varicella admitted for elective surgery, fracture or burn management, psychiatric or social evaluation, treatment of simple dehydration, or evaluation of fever or rash not yet diagnosed. Exclusions included varicella-zoster virus infection in neonates and immunocompromised children.
A priori criteria were formulated on the basis of a comprehensive literature review to define complicated varicella-zoster virus infection. Recent corticosteroid exposure was defined as corticosteroid use of any sort within 30 days of onset of the chickenpox rash. Data were abstracted by medical chart review.
In total, 167 cases and 134 controls were identified. Only three children (two cases and one control) had a history of recent corticosteroid therapy. Recent corticosteroid exposure was therefore not statistically associated with an increased risk of complicated varicella-zoster virus infection (odds ratio, 1.6; 95% confidence interval, 0.2 to 16.9). No differences between cases and controls were found in sex, history of asthma, or length of hospital stay. The mean age of cases was greater than that of controls (6.0 vs 4.7 years; P<.01).
Recent corticosteroid therapy in otherwise immunocompetent children does not appear to be associated with a statistically increased risk of complicated varicella. A conservative estimate of risk, using the upper limit of the 95% confidence interval, is markedly lower than previously published risk estimates.(Arch Pediatr Adolesc Med. 1996;150:409-414)