The antigenic distance hypothesis (ADH) predicts that negative interference from prior season’s influenza vaccine (v1) on current season’s vaccine (v2) protection may occur when the antigenic distance is small between v1 and v2 (v1≈v2) but large between v1 and the current epidemic (e) strain (v1≠e).
Vaccine effectiveness (VE) against medically-attended, laboratory-confirmed influenza A(H3N2) illness was estimated by test-negative design during three A(H3N2) epidemics (2010-11, 2012-13, 2014-15) in Canada. VE was derived with covariate adjustment across v2 and/or v1 categories relative to no vaccine receipt among outpatients ≥9-years-old. Prior vaccination effects were interpreted within the ADH framework.
Prior vaccination effects varied significantly by season, consistent with the ADH. There was no interference by v1 in 2010-11 when v1≠v2 and v1≠e, with comparable VE for v2 alone or v2+v1: 34%(95%CI:-51,71) vs. 34%(95%CI:-5,58). Negative interference by v1 was suggested in 2012-13 with non-significant reduction in VE when v1≈v2 and v1≠e: 49% (95%CI:-47,83) vs. 28% (95%CI:-12,54). Negative effects of prior vaccination were pronounced and statistically significant in 2014-15 when v1≡v2 and v1≠e: 65%(95%CI: 25,83) vs. -33%(95%CI: -78,1).
Effects of repeat influenza vaccination were consistent with the ADH and may have contributed to findings of low VE across recent A(H3N2) epidemics since 2010 in Canada.