Should Sex Be Considered an Effect Modifier in the Evaluation of Influenza Vaccine Effectiveness?

Abstract

We investigated sex as a potential modifier of influenza vaccine effectiveness (VE) between 2010-2011 and 2016-2017 in Canada. Overall VE was 49% (95% confidence interval [CI], 43% to 55%) for females and 38% (95% CI, 28% to 46%) for males (absolute difference [AD], 11%; P = .03). Sex differences were greatest for influenza A(H3N2) (AD, 17%; P = .07) and B(Victoria) (AD, 20%; P = .08) compared with A(H1N1)pdm09 (AD, 10%; P = .19) or B(Yamagata) (AD, -3%; P = .68). They were also more pronounced in older adults ≥50 years (AD, 19%; P = .03) compared with those <20 years (AD, 4%; P = .74) or 20-49 years (AD, -1%; P = .90) but with variation by subtype/lineage. More definitive investigations of VE by sex and age are warranted to elucidate these potential interactions.

Keywords: effect modification; gender; influenza vaccine; influenza virus; sex; vaccine effectiveness.

Figure 1.

Vaccine effectiveness estimates by sex for influenza A subtype and influenza B lineage. P values indicate significance of vaccine*sex interaction. ^aCovariates included in the interaction model were vaccination status (no, yes), sex (female, male), age group (1–8, 9–19, 20–49, 50–64, ≥65 years), comorbidity (no, yes), province (AB, BC, ON, QC), collection interval (≤4, 5–7 days), calendar time (week of specimen collection based on cubic B-spline with 3 equal knots), season, and vaccine*sex. ^bA(H1N1)pdm09 analysis excludes 2014–2015 and 2016–2017 due to low A(H1N1)pdm09 circulation those seasons. ^cA(H3N2) analysis excludes 2013–2014 due to low A(H3N2) circulation that season. Excluding both the 2013–2014 and 2014–2015 seasons, VE for A(H3N2) was 49% (95% CI, 37% to 58%) among females and 29% (95% CI, 9% to 44%) among males (AD, 20%; P = .03). ^dB(Yamagata) analysis excludes 2010–2011 due to low B(Yamagata) circulation that season. ^eB(Victoria) analysis excludes 2013–2014, 2014–2015, and 2016–2017 due to low B(Victoria) circulation those seasons. Abbreviations: AD, absolute difference (Δ female – male); CI, confidence interval; VE, vaccine effectiveness.

Figure 2.

Vaccine effectiveness estimates by sex for patient age groups. P values indicate significance of vaccine*sex interaction. ^aCovariates included in the interaction model were vaccination status (no, yes), sex (female, male), comorbidity (no, yes), province (AB, BC, ON, QC), collection interval (≤4, 5–7 days), calendar time (week of specimen collection based on cubic B-spline with 3 equal knots), season, and vaccine*sex. Abbreviations: AD, absolute difference (Δ female – male); CI, confidence interval; VE, vaccine effectiveness.