Impact of the Euro 2020 championship on the spread of COVID-19

The COVID-19 pandemic presented significant challenges in managing large-scale events. While some events under strict restrictions showed minimal impact on transmission, large gatherings like the Euro 2020, with substantial media coverage, stadium attendance, increased travel, and viewing parties, posed a higher risk, especially in less restrictive settings. This study uses the Euro 2020 championship as a natural experiment to quantify the impact of such events on COVID-19 spread. The staggered timing of matches across different countries, resembling a randomized study, and the expected gender imbalance in attendance at football-related events provide unique opportunities for analysis. The research aims to quantify the impact of the Euro 2020 on COVID-19 cases and provide policymakers with data-driven insights to inform mitigation strategies for future large-scale events.

Previous research on the impact of sporting events on COVID-19 transmission yielded mixed results. Studies on events with significant contact restrictions reported only minor effects on transmission. However, the potential for large, globally-televised events to drive significant outbreaks, particularly when held in less-restrictive settings, was acknowledged. Observational studies hinted at the potential for considerable impact, motivating this quantitative investigation of the Euro 2020's effect.

The researchers developed a Bayesian model to quantify the effect of the Euro 2020 on COVID-19 spread in 12 participating countries. The model simulates COVID-19 transmission for each country separately, distinguishing between male and female cases. The model incorporates a discrete renewal process to capture the spread dynamics for each gender, considering the gender imbalance in football-related gatherings (estimated at 33% female participation). The effective reproduction number (R_eff) is modeled as the sum of three components: a base reproduction number (R_base) representing the underlying infection dynamics, a match-induced boost (ΔR_football) concentrated on match days, and a noise term (ΔR_noise) accounting for other influences. The model accounts for a reporting delay using a Gamma distribution to capture the latent period and reporting time. A weekday-dependent delay component accounts for potential variations in testing behaviour across days of the week. Data on daily gender-resolved case numbers were collected from governmental institutions and the COVerAGE-DB. The model uses a hierarchical Bayesian structure, incorporating uncertainties in individual country estimations to derive average effects. The model's robustness was assessed through sensitivity analyses, modifying various parameters and assessing the impact on the results. The model outputs the number of primary cases directly linked to match days, and subsequent cases stemming from these primary infections. Additional analyses explored correlations between the championship’s impact and other variables such as pre-championship infection rates, reproduction numbers, mobility data, and governmental responses.

The analysis revealed that the Euro 2020 matches significantly impacted COVID-19 spread in most countries, except for the Netherlands where simultaneous policy changes confounded the results. On average, each match increased the reproduction number (R) by 0.46 (95% CI: [0.18, 0.75]) on the match day itself. This translated into approximately 2200 (95% CI: [986, 3308]) primary and subsequent cases per million inhabitants across the 12 countries. This amounts to an estimated 840,000 (95% CI: [0.39M, 1.26M]) additional COVID-19 cases attributed to the championship. England and Scotland showed the strongest effects, with England experiencing a substantial surge in R towards the end of the championship. Scotland experienced a dramatic spike in R (+3) during a single match. The analysis indicated that the overall impact of the championship depended non-linearly on pre-existing incidence rates, the base reproduction number (R_base), and the number of matches played. The low incidence and lower reproduction number in the Czech Republic, despite similar match-day gatherings as England, resulted in far fewer infections. Counterfactual scenarios demonstrated the significant impact of both the team's success in the tournament and the pre-existing pandemic conditions on overall infection numbers. While mobility data did not show a clear correlation with the championship's impact, the number of matches played and hosted showed a trend towards higher impact, although not statistically significant. The study also notes that hosting matches themselves seemed to have minimal impact.

The findings highlight the significant public health impact of large-scale sporting events, even in the face of existing COVID-19 awareness and mitigation measures. The study demonstrates how specific contextual factors, such as pre-existing infection levels and the intensity of gatherings, can heavily influence the scale of such impacts. The non-linear relationship between pre-event conditions and event impact emphasizes the importance of comprehensive pre-event risk assessments that account for epidemiological context. The results provide crucial insights into the effectiveness of different interventions, suggesting that simultaneous strategies such as vaccination, mask mandates, limiting gathering sizes, testing, and requiring COVID passports might be the most effective. The authors note that future events, even with potential improved immunity, could still have substantial impacts.

This research offers a quantitative assessment of the impact of a major sporting event on COVID-19 transmission, demonstrating a substantial contribution to case numbers and highlighting the influence of pre-existing conditions and gathering intensity. The findings underscore the need for careful consideration of public health implications when planning and implementing large-scale events and emphasize the importance of multifaceted mitigation strategies. Further studies are warranted to investigate similar events under differing pandemic contexts and to examine the long-term effects of such events on infection patterns and healthcare burdens.

The model relies on reported case numbers, which might underrepresent the true number of infections. The analysis focuses on 12 participating countries, and the results may not be generalizable to all settings. The confounding effect of policy changes, particularly in the Netherlands, highlights the complexities of isolating the championship's impact. The assumption of equal age distribution across football-related activities might underestimate deaths associated with the spread.