Skip to main content

Vaccine Effectiveness against Infection and Onwards Transmission of COVID-19: Analysis of Belgian Contact Tracing data, January-June 2021

Our take —

This study linked contact tracing data, vaccination status, and SARS-CoV-2 testing results in Belgium from January to June 2021 to estimate the effectiveness of vaccines in a) preventing a vaccinated person from becoming infected, and b) preventing transmission from an infected vaccinated person to susceptible contacts. The authors found that the mRNA vaccines from Pfizer and Moderna were highly effective in preventing both infection (74% and 85%, respectively) and onward transmission (62% and 52%, respectively). The estimated effectiveness of the “viral vector” vaccines from AstraZeneca and Johnson & Johnson was lower, but too few people received these vaccines to permit precise estimation. There are two primary caveats with these results: first, the behavior of vaccinated and unvaccinated contacts may have differed from one another during exposures (e.g., mask use). Second, the study was conducted before the widespread emergence of the Delta variant and may not reflect vaccine effectiveness under the current mix of circulating strains. Despite these limitations, this study was able to estimate the effectiveness of vaccines to prevent both infection and onward transmission by leveraging data from a national contact tracing program.

Study design

Retrospective Cohort

Study population and setting

This was a study of SARS-CoV-2 vaccine effectiveness in Belgium from January 25 to June 24, 2021. The study population (52% female, mean age 33 years) was identified through the national contact tracing program and included all recorded contacts between infected and susceptible individuals within 1.5 meters lasting longer than 15 minutes. Susceptibility was defined by the absence of a positive PCR or antigen test result within the prior 90 days. There were 301,741 contact events arising from 131,283 index cases after excluding individuals with missing test results and those who were vaccinated within the previous 14 days. Vaccination status was classified by type of vaccine (AstraZeneca, Pfizer, Moderna, or Johnson & Johnson) and number of doses given (i.e., partial vs. full). In 91% of contact events, both the index case and the susceptible contact were unvaccinated. Among the remaining events, 2.7% of index cases were partially vaccinated and 0.8% were fully vaccinated, while 4% of susceptible contacts were partially vaccinated and 2.6% were fully vaccinated. The vast majority of vaccinated individuals received either the AstraZeneca or Pfizer vaccines. PCR testing of susceptible contacts was carried out as soon as possible and repeated 7 days after exposure (or sooner if symptoms developed). The authors fit Bayesian logistic regression models to test results, adjusting for previous infection, calendar week, and whether the exposure occurred in the household. Vaccine effectiveness was calculated with respect to preventing infection (from an index case to a vaccinated contact) and preventing onward transmission (from a vaccinated index case to a contact).

Summary of Main Findings

Vaccine effectiveness (VE) at preventing infection (from an unvaccinated index case to a fully vaccinated susceptible contact) was estimated at 74% (95% CI: 72 to 76) for the Pfizer vaccine and 85% (95% CI: 80 to 90) for the Moderna vaccine. For the viral-vector vaccines, estimated VE was lower with wider confidence intervals: 53% (95% CI: 12 to 84) for the AstraZeneca vaccine and 61% (95% CI: 29 to 84) for the Johnson & Johnson vaccine. There was no statistically significant difference between estimated VE and prior infection. For preventing onward transmission from a fully vaccinated index case to an unvaccinated susceptible contact, estimated VE was 62% (95% CI: 57 to 67) for the Pfizer vaccine and 52% (95% CI: 33 to 69) for the Moderna vaccine. The estimated VE for onward transmission for the viral-vector vaccines was not significantly different from zero.

Study Strengths

The study population was derived from the national contact tracing program, which meant the data collection and testing procedures were relatively standardized. A high proportion of identified contacts were tested for SARS-CoV-2 infection.


A small proportion of the study population was fully vaccinated, which makes the model estimates relatively imprecise. Very few individuals received full doses of the AstraZeneca and Johnson & Johnson vaccines. Although there was a standard definition of a close contact, the nature of this contact could have systematically differed by vaccination status and this could have biased results. For example, if unvaccinated contacts were more likely to wear masks than vaccinated contacts, and if masks are partially effective in preventing infection, then the results would be biased against vaccine effectiveness. Actual numbers of positive test results among susceptible contacts were not reported, making interpretation of model results difficult. Data were not provided regarding the timing of testing relative to the date of exposure; if infections were missed due to late testing, results might be biased toward the null. The data also do not distinguish between symptomatic and asymptomatic infection. Finally, the emergence of more transmissible SARS-CoV-2 strains means that the results presented here may not be applicable to other calendar periods.

Value added

Because this study used data from a national contact tracing program, it was able to estimate vaccine effectiveness in both preventing infection and preventing onward transmission.

This review was posted on: 11 September 2021