Reply to Chung et al

To the Editor—In their correspondence [1], Chung et al respond to 2014–2015 findings from the Canadian Sentinel Practitioner Surveillance Network that showed increased influenza A(H3N2) risk among repeatedly vaccinated compared with consistently unvaccinated individuals across 3 consecutive seasons [2]. Although they credit us with first report of such an association for A(H3N2), Hoskins et al [3] and Keitel et al [4], decades earlier, showed similar effects with repeat vaccination, as outlined in our article [2]. Chung et al provide an update to the 2014–2015 vaccine effectiveness (VE) estimates from the United States FluVE Network previously reported by Zimmerman et al [5]. In their updated analysis, Chung et al consider 3 consecutive seasons’ vaccine history (current and 2 prior), whereas Zimmerman et al considered only 2 consecutive seasons’ vaccine history (current and 1 prior) from the same 2014–2015 dataset. Chung et al restricted vaccination history to participants with medical record documentation in all 3 seasons. Zimmerman et al also relied upon medical record documentation for prior season’s vaccination, but in some states accepted plausible self-report for defining current vaccination. In the reanalysis by Chung et al, VE estimates were highest in those vaccinated in 2014–2015 only, at 26% (95% confidence interval [CI], 0–46%) and slightly more divergent compared to repeat vaccine recipients (–2% [95% CI, −24% to 15%]) than those originally reported by Zimmerman et al (8% [95% CI, –14% to 26%] vs –2% [95% CI, –20% to 13%]). They interpret the updated findings as consistent with VE observations from Canada indicating pronounced negative interference with repeat vs current-season-only vaccination (–47% [95% CI, –101% to –8%] vs 63% [95% CI, 14%–84%]) [1, 2]. However, confidence intervals broadly overlap in both US analyses, and it is only the smaller subset of participants vaccinated in the current-season only that differed meaningfully from those reported by Zimmerman et al. Ultimately, it is unclear whether the updated findings from the United States reflect random variation, varying method(s) of vaccine status ascertainment and its potential misclassification, or negative interference based on the number of consecutive vaccine doses considered. In relying upon medical record documentation for vaccine status ascertainment, Chung et al caution that patients may overreport vaccination by approximately 10% relative to registry data, citing Irving et al [6]. Comparing self-report to registry data for current season’s vaccination, Irving et al reported sensitivity and specificity of 95% and 90%, respectively, across all age groups (Table 2 in [6]). However, among participants aged ≥5 years, Irving et al also showed that registry data missed approximately 10% of patient/guardian-reported vaccinations that were subsequently confirmed by the provider to have been received (ie, registry sensitivity = 90%) (Figure 2 in [6]). When the incompleteness of registry data was corrected, concordance was improved, with overall sensitivity and specificity of 95% and 95%, respectively (Table 3 in [6]). Sensitivity of patient/guardian report in relation to the adjudicated registry would have exceeded 97% if children <5 years old were excluded, which is relevant as repeat VE assessments were restricted to participants ≥9 years old in both Canada and the United States [1, 2]. Recall of prior seasons’ vaccination may be less accurate than for the current season, but is not expected to vary systematically by outcome, and is anticipated to be most reliable for those with habitual vaccination behaviors. In Canada, vaccination status—both current and prior—is consistently based on patient/guardian report and practitioner documentation before either knows the outcome status (ie, influenza test result). This minimizes the likelihood of differential exposure misclassification—in particular because illness severity and the pretest likelihood of influenza are further standardized in Canada by a consistent influenza-like illness syndrome as a study eligibility criterion—requiring a specific combination of respiratory and systemic symptoms. Exposure misclassification that is nondifferential and independent is more likely to obscure than to exaggerate differences between vaccinated and unvaccinated groups [7], attenuating VE estimates from either direction of protective or negative effect. VE estimates are most reliable for the largest subgroup of repeatedly vaccinated participants, whereas sparse data considerations mean that smaller subgroups (such as current-season only recipients) are more susceptible to the impact of measurement error. In that regard, the most contentious finding of negative VE (increased risk) with repeated vaccination is also anticipated to be the most robust among vaccine subgroups in Canada. Nondifferential exposure misclassification may partially explain attenuated protective effects of vaccination in the current season only in the United States compared with Canada (26% vs 63%) [1, 2], or attenuated negative effects of repeated vaccination in the United States compared with Canada (–2% vs –47%) [1, 2], but does not otherwise provide reassurance against the Canadian findings. We agree with Chung et al that the impact of exposure misclassification from both self-report and registry-based documentation warrants further reflection. However, as elaborated in our article, other variability should also be taken into account, including the broader incorporation of other methodological and immunoepidemiological considerations to explain repeat vaccination effects [2]. Potential conflicts of interest. G. D. S. has received grants unrelated to influenza from GSK and Pfizer and travel reimbursement to attend an ad hoc advisory board meeting of GSK also unrelated to influenza; and has provided paid expert testimony in a grievance against a vaccinate-or-mask healthcare worker influenza vaccination policy for the Ontario Nurses’ Association. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.