The risk of indoor sports and culture events for the transmission of COVID-19

Governments rapidly banned mass gathering events (MGE) early in the COVID-19 pandemic due to suspected high transmission risk. MGEs are economically and societally important, motivating evidence-based assessment of their risks. SARS-CoV-2 transmission can occur via droplets, aerosols, and fomites; while the relative contributions remain debated, proximity, exposure duration, activity (e.g., singing), and ventilation are key determinants. This study asks: Under what hygiene practices and ventilation conditions do indoor MGEs meaningfully contribute to SARS-CoV-2 spread? The authors conducted an experimental indoor concert to measure close contacts, simulated aerosol exposure under different ventilation strategies, and integrated these data into an individual-based epidemiological model to estimate excess infections attributable to MGEs under controlled epidemic conditions.

Study design: RESTART-19 comprised (1) an experimental indoor pop concert to quantify close contacts using contact tracing devices (CTDs) under three hygiene scenarios; (2) computational fluid dynamics (CFD) simulations to estimate aerosol exposure under two ventilation variants (VV1, VV2); and (3) an individual-based epidemiological model integrating contact and aerosol exposure data to estimate population-level excess infections. Experimental event and participants: Conducted August 22, 2020 in Quarterback Immobilien Arena (Leipzig, Germany). Targeted 4000 participants aged 18–50; 1212 ultimately attended (Scenario 1: 1192; Scenario 2: 1158; Scenario 3: 1054). All participants/staff had negative PCR tests within 48 h pre-event; on-site exclusion included fever ≥37.5°C, symptoms, recent high-risk contact, or travel to risk areas. All attendees wore N95 masks throughout the event except in designated, distanced catering areas outside. Hygiene scenarios and schedule: Each scenario followed Entry (60 min), 1st half (20 min), halftime (20 min) with simulated catering, 2nd half (20 min), Exit (15 min). Scenario 1: pre-pandemic baseline (two main entrances, no distancing, full seating). Scenario 2: moderate restrictions (arena divided into four quadrants; entry/exit via quadrant-specific doors; checkerboard seating pattern; restricted mixing). Scenario 3: strong restrictions (pairwise seating with 1.5 m between pairs; eight entrances/exits; quadrant separation). Hygiene stewards supported compliance; distancing required in Scenarios 2 and 3. Contact measurement: Participants wore UWB/BLE CTDs (ICDWpro quad 164643) with ±20 cm accuracy. Contacts defined by combined distance-duration thresholds: <50 cm ≥3 s; <1 m ≥6 s; <1.5 m ≥10 s. A 2 s gap ended a contact; 10 s reset allowed a new contact. Data gaps were reconciled; analysis used the 1.5 m threshold. Contacts were summarized for >10 s, >5 min, and critical contacts >15 min within 1.5 m, consistent with Robert Koch Institute definitions. Sitting halves were scaled to 45 min for realistic accumulation. Analyses conducted in R 4.0.2. CFD aerosol simulations: Arena modeled 1:1 in PHOENICS with FLAIR (drift-flux) and GENTRA for particle tracking. 4000 virtual spectators arranged in checkerboard seating (Scenario 2 baseline); 24 infectious persons distributed across stalls and grandstands. Breathing modeled as ideal gas emitting 20 L CO2 per hour; breathing volume 12 L/min. Infectious emit aerosols at 0.5, 5, and 10 µm with rates adjusted to singing; viral load assumed 10^6 RNA copies/mL; estimated 4×10^5 virus copies emitted over 90 min of singing vs 4×10^3 at rest. Exposure measured at the virtual mouth opening. Exposure threshold set pragmatically at 1% of emitter output (≈40–4000 copies per concert), reflecting plausible minimal infectious dose estimates. Ventilation variants: VV1 (actual system): lateral jet nozzles plus under-seat swirl diffusers; exhaust via corner towers; 100% fresh air; air supply 198,000 m^3/h; ACH 1.46 h^-1. VV2 (modified): jet nozzles and exhaust towers off; exhaust via roof pipes; supply 115,000 m^3/h; ACH 0.85 h^-1, aiming for displacement flow but resulting in weaker buoyancy and stationary eddies, especially above grandstands. Epidemiological model: Extended SEIR individual-based model in R 4.0.2 with age structure reflecting Leipzig demographics (~601k residents). Contact networks from POLYMOD: household, school/work, other (transport/leisure/other); only ≥15 min contacts considered for transmission in networks. Event attendees drawn randomly daily (excluding test-positive and <15 yrs). Control measures mirrored Germany in summer 2020: reduced contacts (~50%), symptomatic testing, and contact tracing with specified probabilities and delays (household 100% same/next day; school/work 80% with 2-day delay; other 50% with 4-day delay). Isolation/quarantine 14 days for detected cases and household members; assumes adequate tracing capacity. Per-contact transmission calibrated to R≈3 pre-pandemic; with reductions and controls, R≈1. Transmission probability per ≥15 min contact set to ~7%. Aerosol-exposed individuals were assigned the same 7% infection risk per exposure event. Incidence scenarios: 10, 50, 100 per 100,000/week; event volume: 100,000 or 200,000 participants per 30 days; masks vs no masks compared. Outcomes evaluated over 30 days in 1000 simulation runs per setting. Acceptance survey: Online questionnaire 3 weeks post-event assessed feasibility perceptions and mask acceptability (960 respondents; 79% response).

- Contact patterns: Mean direct contacts per visitor were about 9 overall. Considering only critical contacts (>15 min within 1.5 m), median counts fell below 10 across scenarios. High numbers of brief contacts occurred during entry, halftime, and exit, but few lasted >15 min; no >15 min contacts during exit. During concert halves, contacts were fewer but mostly >15 min (due to fixed seating). Hygiene scenarios reduced contacts of all durations; in Scenario 1, new longer contacts accumulated throughout, whereas in Scenarios 2 and 3 most contacts occurred during entry with little subsequent increase. No gender or age effects on contact counts were observed. - Aerosol exposure and ventilation: With 24 infectious persons present, VV1 (ACH 1.46) led to 85 exposed individuals in total versus 612 under VV2 (ACH 0.85). Mean exposed per infectious person in Scenario 1: 3.5±2.9 (VV1) vs 25.5±27.8 (VV2), maxima 10 vs 108. Area-specific mean exposures (VV1 vs VV2): stalls 6.75 vs 24.25 (≈4×), mobile grandstands 10.25 vs 59.75 (≈6×), solid grandstands 4.25 vs 69.0 (≈16×). Hygiene practices further reduced mean exposures per infectious person: Scenario 2 to 1.9±1.5 (VV1) and 11.8±13.5 (VV2); Scenario 3 to 0.7±1.0 (VV1) and 5.3±6.4 (VV2). - Expected infectious attendees: For monthly participation of 200,000, average infectious attendees per event month rise with incidence: 7.8 (10/100k/week), 37.8 (50/100k/week), 75 (100/100k/week), assuming controlled epidemic and exclusions of known cases. - Excess infections from MGEs (masks on): At 10/100k/week, mean excess detected cases ranged from 5.1 (Scenario 3, VV1) to 22.0 (Scenario 1, VV2). At 100/100k/week, from 11.7 (Scenario 3, VV1) to 196.8 (Scenario 1, VV2). Poor ventilation and lack of hygiene measures markedly increased excess cases; mask absence further increased them. Hospitalizations and deaths became plausible at higher incidences, particularly under VV2. - Population-level impact (% increase in positives for 200,000/month at 100/100k/week): Without masks in VV2, increases were 23.6% (Scenario 1), 10.8% (Scenario 2), 4.5% (Scenario 3); with masks and VV1, these dropped to 4.8%, 2.8%, 1.2%. Smaller event sizes further reduce impact. - Acceptance: 88% of respondents were willing to attend under Scenario 2 conditions; 82% under Scenario 3. Mask acceptability was high: 89% reported N95 masks were unproblematic or only slightly restrictive; 90% would wear surgical/cloth masks and 78% would wear N95 masks if required.

The study shows that indoor MGEs do not inherently cause widespread transmission when appropriate mitigation is in place. Physical distancing and organizational measures substantially reduce prolonged close contacts, while ventilation quality predominantly determines aerosol exposure risk. CFD results highlight an order-of-magnitude increase in exposed individuals under poorer ventilation (lower ACH, unfavorable airflow with eddies). Integrating these data into an individual-based model under controlled epidemic conditions indicates that, with effective ventilation and mask use, indoor seated MGEs contribute minimally to community transmission, even at moderately high incidence. Conversely, inadequate ventilation and absence of hygiene measures can produce sizable excess infections. The phases with the most contacts (entry, halftime, exit) mainly involve short-duration encounters, suggesting that seat-phase aerosol control (via masks and ventilation) is critical. The findings support mandatory masking (especially while seated) and strict ventilation requirements for MGEs. They also suggest that while stochastic variability can produce occasional clusters, systematic contributions to overall incidence remain small when precautions are enforced. The generalizability to standing events or very large crowds may be limited, as such settings can increase contact density and mobility, elevating risks.

An experimental concert combined with CFD and epidemiological modeling demonstrates that seated indoor MGEs, conducted with effective ventilation, masking, and moderate distancing, have little impact on COVID-19 spread. Ventilation quality is the dominant factor for aerosol-mediated risk; hygiene practices further reduce both close contacts and exposures. High public acceptance of masking and structured seating supports feasibility. Policymakers can enable MGEs under strict ventilation standards and hygiene protocols, with mask mandates, and by adjusting event sizes to prevailing incidence. Future research should refine infectious dose thresholds for aerosols, evaluate diverse venue geometries and ventilation systems, assess standing/large-scale events, and quantify real-world adherence and contact tracing capacity constraints.

- Sample size below target (1212 vs planned 4000) may have reduced contact density despite adjustments (closing ranks/areas). - Aerosol modeling required assumptions on emission rates, viral loads, and minimal infectious doses, which remain uncertain; exposure thresholds were pragmatic approximations. - Venue-specific ventilation exceeded ASHRAE recommendations; results may not generalize to poorly ventilated or differently configured venues without validation. - Potential additional contacts linked to MGEs (e.g., transit, pre/post-event gatherings) were not modeled; assumptions relied on other settings maintaining their own hygiene practices. - Population structure effects (e.g., repeated attendance by the same subgroup) and network clustering could increase MGE impact beyond model assumptions. - Assumed adequate contact tracing capacity; overload would increase MGE-attributable transmission. - The study focused on seated events; standing concerts or large-scale sporting events may entail higher contact rates and different risk profiles.