Dynamics of contact behaviour by self-reported COVID-19 vaccination and infection status during the COVID-19 pandemic in Germany: an analysis of two large population-based studies

Abstract

Background: Contact behaviour is crucial to assess and predict transmission of respiratory pathogens like SARS-CoV-2. Contact behaviour has traditionally been assessed in cross-sectional surveys and not as part of longitudinal population-based studies which simultaneously measure infection frequency and vaccination coverage. During the COVID-19 pandemic, several studies assessed contact behaviour over longer periods and correlated this to data on immunity. This can inform future dynamic modelling. Here, we assess how contact behaviour varied based on SARS-CoV-2 infection or vaccination status in two large population-based studies in Germany during 2021.

Methods: We assessed direct encounters, separated into household and non-household contacts, in participants of MuSPAD (n = 12,641), a population-based cohort study, and COVIMOD (n = 31,260), a longitudinal contact survey. We calculated mean numbers of reported contacts and fitted negative binomial mixed-effects models to estimate the impact of immunity status, defined by vaccination or previous infection, on contact numbers; logistic mixed-effects models were used to examine the relationship between contact behaviour and seropositivity due to infection.

Results: Contact numbers varied over the course of the pandemic from 7.6 to 10.8 per 24 h in MuSPAD and 2.1 to 3.1 per 24 h in COVIMOD. The number of non-household contacts was higher in participants who reported previous infections and vaccinations (contact ratio (CR) MuSPAD: 1.22 (95%CI 0.94-1.60); COVIMOD: 1.35 (CI 1.12-1.62)) compared to unvaccinated and uninfected individuals. Non-household contact numbers were also higher in fully vaccinated participants (MUSPAD: CR 1.15 (CI 1.05-1.26); COVIMOD: 1.43 (CI 1.32-1.56)) compared to unvaccinated individuals. Compared to individuals without household contacts, the odds for seropositivity due to infection were higher among MuSPAD individuals with three or more household contacts (odds ratio (OR) 1.54 (CI 1.12-2.13)) and eleven or more non-household contacts (OR 1.29 (CI 1.01-1.65)).

Conclusions: Different contact behaviours based on infection and/or vaccination status suggest that public health policies targeting immunity status may influence the contact behaviour of those affected. A combined assessment of self-reported contacts, infections, and vaccinations as well as laboratory-confirmed serostatus in the population can support modelling of the spread of infections. This could help target containment policies and evaluate the impact of public health measures.

Keywords: Covid-19; Immunity status; Pandemic; Previous infections; SARS-CoV-2; Seroprevalence study; Social contact behaviour; Social contact survey; Vaccination status.

Fig. 1

Vaccination status over time by age group and study in 2021 in the German population (MuSPAD & COVIMOD). Vaccination status is shown over time in 2021 and stratified by age groups for a MuSPAD and b COVIMOD. Individuals receiving the Janssen COVID-19 vaccine were counted as being fully vaccinated. Booster vaccines were not considered separately. For COVIMOD, the widths of the bars represent the length of each survey wave. Inconsistencies in vaccination rates over time in MuSPAD are likely reflections of study centres changing over the study period

Fig. 2

Contact numbers over time in 2021 in the German population. Temporal course of all, non-household, and household contacts in the MuSPAD and the COVIMOD study samples in 2021. Points mark mean number of contacts per week in MuSPAD and per survey wave in COVIMOD; smoothing to indicate temporal trends was done with the method “loess.” The dashed grey line illustrates the stringency index to quantify containment strategies implemented on the national level

Fig. 3

Mean number of contacts over time stratified by study, immunity status, vaccination status, and serostatus. The plots illustrate the mean number of overall contacts for a MuSPAD stratified by immunity status; b COVIMOD stratified by immunity status; c MuSPAD stratified by vaccination status; d COVIMOD stratified by vaccination status; e MuSPAD stratified by serostatus, based on laboratory results allowing the distinction between infection- or vaccine-acquired antibodies according to the detection of anti-NC, corrected by self-reported vaccination and test history. Points mark measured means; smoothing to indicate temporal trends was done with the method “loess.” The shaded regions indicate the 95% confidence interval of the estimates of the means derived with the loess smoothing method. For MuSPAD, contact behaviour was aggregated as weeks to account for varying contact behaviour on different days of the working week and weekend; for COVIMOD, the mean number of contacts was calculated by survey wave. Note differences in the y-axes between plots. A more detailed version of this plot including the stringency index can be found in Additional file 1: Figure S4.2.1

Fig. 4

Forest plots of regression models, stratified by outcome, exposure, and study. Resulting effect estimates of regression models are displayed as a contact ratios (CRs) using negative binomial regression models for total, household, and non-household contact numbers by immunity status in a1 MuSPAD and a2 COVIMOD; b CRs using negative binomial regression models for contact numbers in different settings by vaccination status in b1 MuSPAD and b2 COVIMOD; and c odds ratios (ORs) for seropositivity due to infection (S>1, RBD>1, NC>1) using logistic regression models in MuSPAD. CRs for contacts and ORs for seropositivity are given with 95% confidence intervals (CI). For a list of adjusted variables for each model, refer to Additional file 1: Table S2.1.1