Effectiveness of COVID-19 Vaccines: A Vaccinated-Only Approach

We used a modified screening method that ensures quasi-exchangeability of comparison groups to estimate COVID-19 vaccine effectiveness in people resident in the Federal State of Hessen, Germany. COVID-19 vaccination history of vaccinated subjects with reported symptomatic SARS-CoV-2 infection was used to determine vaccination status. Subjects with their first COVID-19 vaccination within 7 days before the imputed date of infection were considered unvaccinated. Vaccination is assumed not to have a relevant effect on outcome risk for the first seven days and to be fully developed after between 14 and 21 days. The immunization profile of the source population was estimated from the number of subjects vaccinated by dose, date and age group as recorded in the Hessian COVID-19 vaccination registry. Effect estimates were obtained using logistic regression, fitted by a Bayesian approach. The first dose of COVID-19 vaccines had a measurable effect during the predominance of the Alpha and Delta variants of SARS-CoV-2, but a smaller effect during Omicron predominance. Only during Alpha and Delta predominance did the second dose provide an added benefit. During Omicron predominance, the third dose provided additional protection, but that effect was smaller than for the Delta period. Comparison of our estimates with estimates using a conventional, not quasi-exchangeable, approach revealed substantial differences in some cases, without any recognizable pattern.

PACS

0000, 1111

MSC

0000, 1111

Graphical Abstract

Highlights

• We propose a vaccine effectiveness (VE) study design for COVID-19 that is based on a modification of the screening method. The modification ensures quasi-exchangeable: Only vaccinated subjects are considered, COVID-19 cases on the one hand and subjects registered in the COVID-19 vaccination registry. Risk comparisons are calculated for individuals who received their n -th vaccination one to two weeks apart, assuming no relevant vaccine effect within a week.

• The first dose of COVID-19 vaccines had a measurable effect during the predominance of the Alpha and Delta variants of SARS-CoV-2, but a smaller effect during Omicron predominance. Only during Alpha and Delta predominance did the second dose provide an added benefit. During Omicron predominance, the third dose provided additional protection, but that effect was smaller than for the Delta period.

• Comparison with a conventional approach revealed substantial differences in some cases, that did not follow a clear pattern.