Information bias of vaccine effectiveness estimation due to informed consent for national registration of COVID-19 vaccination: estimation and correction using a data augmentation model

Abstract

ABSTRACTBackgroundRegistration in the Dutch national COVID-19 vaccination register requires consent from the vaccinee. This causes misclassification of non-consenting vaccinated persons as being unvaccinated. We quantified and corrected the resulting information bias in the estimation of vaccine effectiveness (VE).MethodsNational data were used for the period dominated by the SARS-CoV-2 Delta variant (11 July to 15 November 2021). VE ((1-relative risk)*100%) against COVID-19 hospitalization and ICU admission was estimated for individuals 12-49, 50-69, and ≥70 years of age using negative binomial regression. Anonymous data on vaccinations administered by the Municipal Health Services were used to determine informed consent percentages and estimate corrected VEs by iterative data augmentation. Absolute bias was calculated as the absolute change in VE; relative bias as uncorrected / corrected relative risk.ResultsA total of 8,804 COVID-19 hospitalizations and 1,692 COVID-19 ICU admissions were observed. The bias was largest in the 70+ age group where the non-consent proportion was 7.0% and observed vaccination coverage was 87%: VE of primary vaccination against hospitalization changed from 75.5% (95% CI 73.5-77.4) before to 85.9% (95% CI 84.7-87.1) after correction (absolute bias -10.4 percentage point, relative bias 1.74). VE against ICU admission in this group was 88.7% (95% CI 86.2-90.8) before and 93.7% (95% CI 92.2-94.9) after correction (absolute bias -5.0 percentage point, relative bias 1.79).ConclusionsVE estimates can be substantially biased with modest non-consent percentages for registration of vaccination. Data on covariate specific non-consent percentages should be available to correct this bias.KEY MESSAGES (3-5 bullet points, each a complete sentence)A relatively small degree of misclassification in the determinant (e.g. modest non-consent for registration of vaccination records) can result in substantial bias in effect estimates (e.g. vaccine effectiveness [VE]) in particular when the exposed group is large (high vaccination uptake).In this study, a non-consent percentage of 7.0% for registration of vaccination records in the 70+ years group, in which the observed vaccination uptake was 87%, resulted in an absolute bias of the VE against COVID-19 hospitalization of -10.4 percentage point and a relative bias of the relative risk (true/observed relative risk) of 1.74.Changes over time in vaccination uptake in the context of modest non-consent percentages may result in incorrect conclusions regarding waning of the VE.Similarly, differences in vaccination uptake or non-consent percentages between age groups may result in incorrect conclusions regarding effect modification of the VE by age.Covariate-specific data on non-consent percentages should be available to assess the bias and generate corrected VE estimates under certain assumptions.