Anders Hviid1, Henrik Svanström. 1. Department of Epidemiology Research, Statens Serum Institut, Artillerivej 5, 2300 Copenhagen S, Denmark. aii@ssi.dk
Abstract
OBJECTIVES: Antibiotic use in childhood has been hypothesized to increase the risk of intussusception. Antibiotics have well-known effects on gut motility and microflora-possible mechanisms for the hypothesized association. We evaluated the risk of intussusception according to antibiotic use. METHODS: We conducted a nationwide cohort study of all Danish singleton children born between 1995 and 2003 (n = 611 410) with individual-level information on antibiotic prescriptions, intussusception and potential confounding variables. Using Poisson regression, we estimated rate ratios of intussusception according to antibiotic use, including estimating increases in rate ratios per dose of antibiotics and rate ratios for time periods following antibiotic use. RESULTS: Intussusception was diagnosed in 434 children during 1 180 749 person-years of follow-up. The intussusception rate ratio was 1.51 [95% confidence interval (CI), 1.19-1.91] comparing antibiotic users with non-users. In the first week following use of extended-spectrum penicillins the rate ratio was 4.68 (95% CI, 2.93-7.47). In the first week following use of macrolides the rate ratio was 3.82 (95% CI, 1.22-11.90). The proportion of all cases attributable to extended-spectrum penicillins and macrolides was 4%. CONCLUSIONS: This is the first prospective study to show an association between antibiotics and intussusception. The association was strong, temporal and biologically plausible. The magnitude of the observed effect and a number of sensitivity analyses favour a causal relationship. However, the potential for confounding-by-indication cannot be completely discounted and controlled studies of the observed association will be necessary for more definite confirmation.
OBJECTIVES: Antibiotic use in childhood has been hypothesized to increase the risk of intussusception. Antibiotics have well-known effects on gut motility and microflora-possible mechanisms for the hypothesized association. We evaluated the risk of intussusception according to antibiotic use. METHODS: We conducted a nationwide cohort study of all Danish singleton children born between 1995 and 2003 (n = 611 410) with individual-level information on antibiotic prescriptions, intussusception and potential confounding variables. Using Poisson regression, we estimated rate ratios of intussusception according to antibiotic use, including estimating increases in rate ratios per dose of antibiotics and rate ratios for time periods following antibiotic use. RESULTS: Intussusception was diagnosed in 434 children during 1 180 749 person-years of follow-up. The intussusception rate ratio was 1.51 [95% confidence interval (CI), 1.19-1.91] comparing antibiotic users with non-users. In the first week following use of extended-spectrum penicillins the rate ratio was 4.68 (95% CI, 2.93-7.47). In the first week following use of macrolides the rate ratio was 3.82 (95% CI, 1.22-11.90). The proportion of all cases attributable to extended-spectrum penicillins and macrolides was 4%. CONCLUSIONS: This is the first prospective study to show an association between antibiotics and intussusception. The association was strong, temporal and biologically plausible. The magnitude of the observed effect and a number of sensitivity analyses favour a causal relationship. However, the potential for confounding-by-indication cannot be completely discounted and controlled studies of the observed association will be necessary for more definite confirmation.