Early-life exposure to antibacterials and the subsequent development of hayfever in childhood in the UK: case-control studies using the General Practice Research Database and the Doctors' Independent Network

Background Theoretically, antibacterial agents in early life might influence allergic sensitization in two ways: (i) as an indicator of infectious illness, they might be expected to protect against allergy; (ii) alternatively they might increase the risk through effects on the commensal bowel flora. Epidemiological evidence linking the prescription of antibacterial agents in early life to the subsequent development of hayfever is conflicting. Objective To establish definitively whether an association exists between early-life antibacterial exposure and childhood hayfever diagnosis. Methods Nested case-control studies were based on birth cohorts of children identified within two large UK general practice databases of electronic patient records. One hundred and sixteen thousand and four hundred and ninety-three children from 605 general practices were identified as being continuously registered from birth to at least age 5 years. Seven thousand and ninety-eight cases were diagnosed with hayfever after the age of 2 years. One control per case was matched for practice, birth month, sex and still being registered on case diagnosis date. Odds ratios were derived from conditional logistic regressions within each database followed by pooling using a fixed-effect model. Results The pooled odds ratio for hayfever was 1.11, 95% CI (1.03-1.20) if exposed to antibacterials in the first year of life, 1.35 (1.25-1.46) in year 2 and 1.47 (1.37-1.59) in year 3. Adjusting for consultation frequency reduced these odds ratios to 0.92, 1.05 and 1.10, respectively. There was no evidence that broader spectrum antibacterials, exposure in any specific month of year 1 or in the grass pollen season influenced the risk of hayfever. Conclusion These data exclude any important effect of antibacterial exposure in infancy on subsequent hayfever risk. Associations reported in earlier studies have likely been exaggerated through publication bias and by lack of control for the tendency of some families to consult frequently for a range of conditions.