Articles Association between two mass-gathering outdoor events and incidence of SARS-CoV-2 infections during the fifth wave of COVID-19 in north-east Spain: A population-based control-matched analysis ner,a* Ermengol Coma,b Dan Ouchi,a Eduardo Hermosilla,b,c B Clara Su~  arbara Baro,d Miquel Angel Rodríguez-Arias,a Jordi Puig,a Bonaventura Clotet, a,e,f,g b  Manuel Medina, and Oriol Mitja a,e,f,h a Fight AIDS and Infectious Diseases Foundation, Badalona, Badalona, Spain b Sistemes d'Informacio  dels Serveis d'Atencio Primaria (SISAP), Institut Catala de la Salut (ICS), Gran Via de Les Corts Cata- lanes, 587, 08007, Barcelona, Spain c IDIAP Jordi Gol, Universitat Autonoma de Barcelona, Barcelona, Spain d ISGlobal, Hospital Clínic, Universitat de Barcelona, Barcelona, Spain e Hospital Universitari Germans Trias i Pujol, Badalona, Spain f Universitat de Vic-Universitat Central de Catalunya (UVIC-UCC), Vic, Spain g Institut de Recerca de La Sida, IrsiCaixa AIDS Research Institute, Badalona, Spain h Lihir Medical Centre- InternationalSOS, Lihir Island, Papua New Guinea Summary Background Many countries have resumed mass-gathering events like music festivals, despite the risk of severe The Lancet Regional acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spreading. In this study, we aimed to assess the effect of Health - Europe 2022;15: 100337 two mass-gathering outdoor events, held during a peak of SARS-CoV-2 transmission, on COVID-19 incidence. Published online xxx https://doi.org/10.1016/j. Methods This was a retrospective, population-based control-matched analysis. The study population included lanepe.2022.100337 attendees to two outdoor music festivals held in Catalonia (North-East Spain). The primary objective was to compare the incidence of COVID-19 within the 3-to-10 days following the event between attendees and a population-based control group. Findings The analysis included 18,275 and 27,347 attendees to the first and second festivals, respectively, and their corresponding controls. The post-festival 7-day cumulative COVID-19 incidence among attendees and controls was 4.14% (95% CI 3.86-4.44) vs. 1.69% (1.51-1.88) for the first festival (RR 2.46; 2.16-2.80), and 2.42% (2.35-2.61) and 1.10% (0.99-1.2) for the second festival (RR 2.19; 1.92-2.51). COVID-19 incidence among immunized individuals was also two-fold higher in attendees than in controls. Previous COVID-19 infection, vaccination, and adequate mask-wearing were significantly associated with a lower risk of COVID-19 infection after the events. Interpretation Despite the proven effectiveness of preventive measures such as Ag-RDT screening, mask-wearing and vaccination, caution should be taken when holding these events during a period of high community SARS-CoV- 2 transmission. Funding Crowdfunding campaign YoMeCorono (https://www.yomecorono.com/) and the Generalitat de Catalunya. Copyright Ó 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) Introduction coronavirus disease 2019 (COVID-19) pandemic.1 While The high risk of spreading the severe acute respiratory necessary for the public health control of COVID-19, the syndrome coronavirus 2 (SARS-CoV-2) associated with unprecedented closure of the entertainment industry mass-gathering events has encouraged the banning of has caused a remarkable impact on many economies many cultural and sporting events during the and society wellbeing worldwide.2 In the past year, a deeper understanding of SARS- CoV-2 transmission dynamics has provided various *Corresponding author at: Hospital Germans Trias i Pujol, measures for limiting virus transmission in these Carretera Canyet s/n, 08916 Badalona, Spain events.3 Of them, mask-wearing and point-of-care E-mail address: csuner@flsida.org (C. Su~ ner). www.thelancet.com Vol 15 Month April, 2022 1 Articles for making evidence-based decisions on these types of Research in context events in the rapidly evolving scenario of the COVID-19 pandemic. Besides organizational differences in the Evidence before this study deployment of preventive measures, important We searched PubMed for studies exploring the spread- unknowns exist that are key for the success of these pre- ing of COVID-19 in outdoors mass-gathering events ventive strategies. First, the negative predictive value of with preventive COVID-19 measures. The search was Ag-RDT drops with increasing local incidence of SARS- performed on Aug 1, 2021, and included the keywords CoV-2 infection,6 potentially affecting the performance “COVID-19”, “mass-gathering”, “RDT-screening” and of Ag-RDT-based screenings. Second, preliminary evi- “mask-wearing” with no language restriction. We found dence on this strategy was gathered before the emer- studies providing evidence of the low risk of holding gence of the delta variant of SARS-CoV-2, with these events in low transmission settings, while other significantly higher transmissibility than early circulat- authors reported an increase in COVID-19 cases associ- ated to a spreading event where no measures had been ing variants and currently dominant in many countries implemented. Some authors investigated the risk of worldwide.11 SARS-CoV-2 transmission using simulation studies. We In this study, we retrospectively assessed the found no published works specifically investigating the COVID-19 incidence associated with two large music contribution of mass-gathering events to COVID-19 festivals (capacity for over 10,000 people), which held during a peak of high SARS-CoV-2 transmission applied similar prevention strategies based on point-of- with preventives measures like Ag-RDT screening, mask care Ag-RDT screening and mask-wearing, and were wearing and vaccination. held during the fifth wave of transmission - predomi- nantly caused by the delta SARS-CoV-2 variant - in Cata- Added value of this study lonia (North-East Spain).12 This is the first analysis to report that, during a peak of community transmission, the COVID-19 incidence among attendees to two music outdoor festivals (18,275 and 27,347 individuals, respectively) was two- Methods fold higher than that of a population-based control Study design and data sources group. The post-festival incidence of new cases per 100 persons-in-7-days among attendees and controls was This was a retrospective control-matched analysis of 4.14 and 1.69 in the first festival, and 2.42 and 1.10 in COVID-19 incidence after two mass-gathering outdoor the second one. Previous COVID-19 infection, vaccina- events, both music festivals, in Barcelona (Catalonia, tion, and adequate mask-wearing significantly reduced North-East Spain). The first event was held on July 3, the risk. 2021, and second on July 8-10, 2021. By the time of the events, Catalonia was experiencing the fifth wave of the Implications of all the available evidence COVID-19 outbreak, and the weekly incidence reached Mass-gathering events may impact COVID-19 incidence, 669 cases per 100,000 inhabitants on July 6-12, 2021 like other social activities, when held during high com- (Figure 1). Public health directives and vaccination cam- munity transmission of SARS-CoV-2. Our results provide paigns in force at that time are included in Figure 1. policymakers with information to make evidence-based Our analysis included all the individuals who decisions on these types of events and on the protec- attended any of the events with demographic and tion conferred by vaccines, Ag-RDT screening and COVID-19-related information (i.e., sex, age, previous mask-wearing. PCR or Ag-RDT results, and vaccine doses received) available in the central health registry of the Catalan Health Service. This registry contains essential demo- screening with antigen-detecting rapid diagnostic tests graphic and clinical data from nearly 7.7 million people (Ag-RDT) have emerged as essential tools for creating a and has been used for relevant COVID-19 research pre- temporal safe environment during circumstances in viously.13−15 The same central health registry was used which physical distance cannot be warranted.4−7 When to build a matched population-based control group for applied adequately, these preventive measures have each attendant of the events. Attendees were paired 1:1 shown the potential capacity to prevent SARS-CoV-2 by exact matching to individuals of the general popula- spread, even in mass-gathering indoor events in a pre- tion according to sex, age (+/- 2 years), healthcare area vaccination scenario.8−10 of residence, vaccination status, and history of a positive Since the emergence of preliminary evidence on the result in Ag-RDT or PCR. Supplementary methods on effectiveness of these measures for creating safe envi- the case-control study are provided in the Supplemen- ronments, various mass-gathering events have been tary Appendix 1. held, mostly during low transmission periods. However, The second festival included a survey-based assess- their epidemiological impact has not been adequately ment of risk behaviours during the event. The survey, assessed, leaving policymakers with limited information which was answered anonymously online, was sent to 2 www.thelancet.com Vol 15 Month April, 2022 Articles Figure 1. Progression of cumulative weekly cases of COVID-19 per 100,000 inhabitants before and after the events in Cata- lonia. Event dates are indicated with yellow bands; the first festival was held on 03 July and the second from 08 July to 10 July, 2021. The progressive role out of vaccination campaigns to younger individuals is indicated. Selected policies regarding the use of the face mask and other COVID-19 preventive measures are indicated. Incidence data was obtained from the public repository https://dadescovid.cat. a The time points indicate only the starting date of vaccination campaigns for each age group. Healthcare and other essential workers of all ages were vaccinated during the first semester of 2021, before the indicated date for their corresponding age group. all individuals who attended the second festival 14 days to 5:00 a.m.), celebrated in a 100,351 m2 area with a after ending the event, and included items such as the maximum capacity for 25,000 people. exhaustivity in the use of face masks and the overall ful- The two events followed a similar approach for pre- filment of the preventive measures (Supplementary venting SARS-CoV-2 spread, based on face mask use, Appendix 2). Attendees had five days to answer the sur- forehead temperature measurement, and Ag-RDT vey, and the organizers raffled festival ticket prizes screening of nasopharyngeal swab for SARS-CoV-2. In among survey participants as an incentive. both festivals, attendees were provided with an FFP2 The study protocol for secondary use of healthcare mask, which they could renew during the event if data and administration of a post-event survey was needed. The use of the mask was compulsory, except approved by the independent ethics committee of the for eating or drinking. Eating and drinking was allowed University Hospital Germans Trias i Pujol in Badalona in all areas at the first festival. At the second festival, eat- (Spain). All data were handled according to the General ing and drinking was not allowed in the area in front of Data Protection Regulation 2016/679 on data protec- the stage, where organization staff oversaw the event tion and privacy for all individuals within the European and warned people not wearing the mask. Attendees Union and the local regulatory framework regarding were allowed to move around without restrictions and data protection. no social distancing measures were in place. Regarding Ag-RDT screening, attendees to both festivals had to make an appointment for the same day of the entrance Events characteristics and preventive measures and pay for the Ag-RDT, performed by trained staff. Both events were commercial music festivals with pre- However, the two festivals differed in the organization purchase entrance system. The first festival was a one- of Ag-RDT screening (Supplementary Appendix 1). day event (Saturday from 2:00 pm to 6:30 am) held in a Table S1 (Supplementary appendix 1) provides an esti- 33,000 m2 area with capacity for 25,000 people. The mate of the expected negative predictive value associated second festival was a 3-day event (Thursday from 4:00 with Ag-RDT screening for a range of local prevalence p.m. to 1:00 a.m., Friday and Saturday from 4:00 p.m. values (i.e., pre-test probabilities). www.thelancet.com Vol 15 Month April, 2022 3 Articles Study outcomes number of days attending the event (1 to 3), and self- The primary outcome was the post-event cumulated reported behaviour factors, including mask wearing all incidence of SARS-CoV-2 infections confirmed by a or most of the time (5-point qualitative scale). diagnostic test (PCR or Ag-RDT). The incidence was cal- culated using the central health registry of the Catalan Health Service. Because SARS-CoV-2 is a mandatory Statistical analysis notifiable infection to all public and private laboratories, Categorical and quantitative variables were described as the register is comprehensive and captures all diagnos- frequency and percentages and mean and standard devi- tics of SARS-CoV-2 in patients with symptoms, contact- ation (SD) over available data, respectively. The cumula- tracing, or any other reasons for testing. Based on the tive incidence of COVID-19 infections per hundred high COVID-19 incidence and dominance of the delta people observed within 3 to 10 days after the event was variant in the area during the event, we constrained the compared between the attendees of each festival and its time frame for the primary analysis to 3-to-10 days after matched population-based controls using the relative starting the event, to maximize the likelihood of the risk (RR) and its 95% confidence interval (CI). No data transmission to have occurred during the event. Covid- imputation was made. The influence of baseline and 19 incidence within different post-event periods is behaviour factors was investigated using a logistic shown in Table S3 (Supplementary Appendix 1). regression model for reporting a positive COVID-19 Secondary outcomes included the cumulated inci- result within 3 to 10 days following the event, based on dence according to the immunity status, which was the electronic health records and survey-collected data. stratified into the following categories: fully protected The model included age, gender, vaccination level, and (i.e., had received the complete vaccination regimen or previous infection record in each of the two study popu- one vaccine dose among individuals with a history of lations: the overall sample of attendees for each event natural SARS-CoV-2 infection, more than 14 days before and the sample of respondents to the post-event survey. attending the festival), partially protected (i.e., either a In addition, for the post-event survey sample, we added history of SARS-CoV-2 infection without a vaccine, one the use of the mask, the degree of compliance with pro- dose of a two-dose regimen vaccine, or having received tection measures, and the number of days attending the a complete vaccination regimen within the 14 days event. Assuming a possible interaction between varia- before attending the festival), and unprotected (i.e., bles, we performed pairwise Pearson correlation and unvaccinated with no evidence of previous SARS-CoV-2 compared nested models with the F-test. All analyses infection). Based on the local guidelines in force by the were performed in R v4.1.0.16 time of the events, we considered the "complete vaccina- tion regimen" for all individuals who had received two vaccine doses of Pfizer/BioNTech, Moderna/Lonza, and Results Oxford/AstraZeneca vaccines, or one dose for J&J/Jans- sen, at least 14 days before the event. Characteristics of the study population According to the organizers, the first festival identified 152 individuals with a positive result in the Ag-RDT Risk factor analysis screening and allowed the entrance of 21,012 individu- For the analysis of risk factors for COVID-19 infection als with a negative result. The number of individuals after the event, the time frame considered comprised with a positive Ag-RDT (excluded) and total attendees in the 3-to-10 days after the event or the first day of atten- the second festival was 292 and 34,518 individuals, dance in the case of the survey. Attendees that did not respectively. Of these, the second festival received report a COVID-19 infection during this time frame 11,134; 16,655 and 15,650 attendees Thursday, Friday were used as controls. For the survey analysis, cases and Saturday, respectively. The two study samples, reported outside this time frame were excluded from including attendees with enough data for control- the analysis. As potential risk factors, when analysing matched pairing with the general population, consisted the data from the electronic health records, the follow- of 18,275 (87.0% of total attendees) and 27,347 (79.2% ing variables were considered: sex, age, history of previ- of total attendees) for the first and second festival, ous COVID-19 infection, and being partially or fully respectively (Figure S1, Supplementary appendix 1). vaccinated. Full vaccination comprised individuals who Table 1 summarizes the demographic and COVID-19 had received a complete vaccination regimen >14 days immunity status of the individuals included in the anal- before the event, and partial vaccination comprised indi- ysis. The first event had a lower mean age (25.5 (SD 9.4) viduals who had received an incomplete vaccination reg- vs 33.0 (SD 8.8) years), and a higher percentage of imen or a complete vaccination regimen <14 days females (11,923 (65.2%) vs. 14,236 (52.1%) women) than before the event. In the subset of attendees who the second. Regarding the previous SARS-CoV-2 infec- responded to the post-event survey after the second tions, 1,737 (9.5%) and 2,862 (10.5%) of the attendees event, the incidence was also stratified according to the had a history of COVID-19 confirmed diagnosis (either 4 www.thelancet.com Vol 15 Month April, 2022 Articles Population-based analysis Survey Second Festival (N=13,137) First Festival (N =18,275) Second Festival (N =27,347) Age (SD) 25.5 (9.43) 33.0 (8.84) 34.6 (8.54) Sex/Gendera Female 11,923 (65.2%) 14,236 (52.1%) 7,582 (57.7%) Male 6,352 (34.8%) 13,111 (47.9%) 5,471 (41.6%) Non-binary - - 84 (0.6%) Previous COVID-19 Infection No 16,538 (90.5%) 24,485 (89.5%) 11,520 (87.7%) Yes 1,737 (9.50%) 2,862 (10.5%) 1,617 (12.3%) COVID-19 immunity Statusb Unprotected 10,323 (56.5%) 8,983 (32.8%) 2,996 (22.8%) Partially Protected 4,280 (23.4%) 11,991 (43.8%) 5,554 (42.3%) Fully Protected 3,672 (20.1%) 6,373 (23.3%) 4,587 (34.9%) Table 1: Main characteristics of individuals included in the two investigated cohorts: population-based analysis (individuals who attended the event and had enough data in central healthcare registries for a population-based matched control group) and survey (individuals who responded the 14-day post-event survey after the second festival). a For Population-based analysis, sex was obtained from central health registries. For Survey analysis, gender was self-reported. b Categories of the COVID-19 immunity status were as follows: fully protected (i.e., had received the complete vaccination regimen or one vaccine dose among individuals with a history of natural SARS-CoV-2 infection), partially protected (i.e., either a history of SARS-CoV-2 infection without a vaccine, one dose of a two-dose regimen vaccine, or a complete vaccination regimen <14 days before the event), and unprotected (i.e., unvaccinated with no evidence of pre- vious SARS-CoV-2 infection).SD: Standard deviation. PCR or Ag-RDT), in the first and second event, respec- individuals attending the event the three days tively. The immunity status of attendees in the first and (54.5%). Regarding the immunity status of survey second event, respectively, was as follows: 3,672 (20.1%) respondents, 4,587 (34.9%) were fully protected, and 6,373 (23.3%) were fully protected (i.e., full vaccina- 5,554 (42.3%) partially protected, and 2,996 (22.8%) tion regimen or one vaccine dose and a history of unprotected. The time since last vaccine dose in days COVID-19), 4,280 (23.4%) and 11,991 (43.8%) were par- [Median (IQR)], for partially vaccinated and fully vac- tially protected (i.e., one dose of a two-dose vaccine regi- cinated survey responders was 37 (8-38) and 39 (37- men or two doses <14 days before the event, or non- 129), respectively. vaccinated with COVID-19 history), and 10,323 (56.5%) and 8,983 (32.8%) were unprotected (i.e., without any vaccine dose and no history of COVID-19). Among vac- cinated individuals, partially vaccinated individuals had COVID-19 incidence received their last vaccine dose within less days than The number of attendees who had a positive test for fully vaccinated individual (time since last vaccine dose SARS-CoV-2 reported to the central health registry of in days [Median (IQR)], partially vaccinated vs fully vac- the Catalan Health System within the 3 to 10 days fol- cinated attendees: 12 [2-25] vs 33 (27-114), and 11 [6- lowing the event was 757 (4.14%; 95% CI 3.86% - 22] vs 40 (30-142), for the first and second festivals, 4.44%) and 662 (2.42%; 2.32% - 2.61%) for the first respectively). The characteristics of the attendees that and the second festival, respectively. The risk ratio (RR) were not included because no control-match pairing for reporting a COVID-19 case within the 3-10 days fol- was found are summarized in Table S2 (Supplemen- lowing the event regards to the population-based control tary Appendix 1). group was 2.46 (2.16 - 2.80) and 2.19 (1.92 - 2.51) for the Of the 34,518 attendees of the second event, 13,137 first and second festivals, respectively (Table 2). The RR (38.1%) responded to the survey sent 14 days after the for a COVID-19 case was similar, regardless of the event. The mean age (34.6 years, SD 8.5) and the pro- immunization status, although incidence was signifi- portion of females (57.7%) were slightly higher in sur- cantly higher in non-immunized individuals among vey respondents compared to the overall cohort both, attendees and population-based control individu- population (Table 1, Table S2). Individuals who attended als (Table 2). Table S3 (Supplementary appendix 1) sum- the event the first day responded less frequently than marizes the incidences and risk estimated for different those who attended the other two days (response post-event periods. The overall RR for COVID-19 within rates for Thursday, Friday and Saturday attendees the first 3 days after the events were 0.74 (0.53 - 1.03) were 22.4%, 40.7% and 37.1%, respectively). The and 0.21 (0.14 - 0.31) for the first event and second festi- highest response rate was observed among vals, respectively (Table S3). www.thelancet.com Vol 15 Month April, 2022 5 6 Articles Attendees Controls Relative 95% CI Risk Cohort COVID-19 Cases % 95% CI COVID-19 Cases % 95% CI FIRST FESTIVAL All attendees 18,275 757 4.14% (3.86% - 4.44%) 308 1.69% (1.51% - 1.88%) 2.46 (2.16 - 2.80) According to immunity status Fully protected 3,672 67 1.82% (1.44% - 2.31%) 31 0.84% (0.60% - 1.20%) 2.16 (1.42 - 3.30) Partially protected 4,280 79 1.85% (1.48% - 2.29%) 39 0.91% (0.67% - 1.24%) 2.03 (1.38 - 2.97) Unprotected 10,323 611 5.92% (5.48% - 6.39%) 238 2.31% (2.03% - 2.61%) 2.57 (2.22 - 2.97) SECOND FESTIVAL All attendees 27,347 662 2.42% (2.35% - 2.61%) 302 1.10% (0.99% - 1.2%) 2.19 (1.92 -2.51) According to immunity status Fully protected 6,373 96 1.51% (1.24% - 1.84%) 42 0.66% (0.49% - 0.89%) 2.29 (1.6 - 3.32) Partially protected 11,991 182 1.52% (1.31% - 1.75%) 112 0.93% (0.78% - 1.12%) 1.63 (1.3 - 2.05) Unprotected 8,983 384 4.27% (3.88% - 4.71%) 148 1.65% (1.4% - 1.93%) 2.59 (2.15 - 3.13) Table 2: Incidence of COVID-19 3-to-10 days after the events and relative risk of attendees compared with a matched population-based control group. a Immunity status categories: fully protected (i.e., had received the full vaccination regimen or one vaccine dose among individuals with history of natural SARS-CoV-2 infection), partially protected (i.e., either a history of SARS-CoV- 2 infection without vaccine, an incomplete regimen of vaccination, or a complete vaccination regimen <14 days before the event), and unprotected (i.e., unvaccinated with no evidence of previous SARS-CoV-2 infection). CI: confi- dence interval. RR: relative risk. www.thelancet.com Vol 15 Month April, 2022 Articles Figure 2. Odds ratio for COVID-19 infection after attending the events. Risk factor multivariate analysis for COVID-19 infection after attending the events. Time frame considered: 3-to-10 days. (A) Analysis including all attendees and variables recorded in the central health registries [n(cases)/n(total): 757/18,275 for the first festival; 662/27,347 for the second festival). (B) Sub analysis of attendees who responded the 14-day survey regarding behavioural factors after attending the second festival [n(cases)/n(total): 570/13,097]. Definitions: Previous COVID-19 Infection (with or without receiving any vaccine); Full vaccination (i.e., had received the full vaccination regimen >14 days before the event), partial vaccination (i.e., had received an incomplete vaccination regimen or a complete vaccination regimen <14 days before the event); Preventive Measures Compliance and Mask-wearing (OR for answering "all or most of the time"); Age (OR for an increase in one unit); Days of Attendance (OR for each extra day of assistance). Dashed ver- tical lines indicate Odds Ratios of 0.5 and 2. Alternative models including the interaction between age and gender showed no contri- butions to the model (ORs of the interaction age*gender were 0.99 [0.96-1.01] for the first festival, 1.00 [0.98-1.02] for the second, and 1.00 [0.98-1.02] for the survey). The values for OR and 95% CI are presented in Table S4. Risk factors for COVID-19 during the events 3-10 days following the event. The incidence was lower The logistic regression analyses for the two events con- among fully protected individuals (118/4,587; 2.57%) sistently identified age, vaccination (either partial or full and fully protected individuals who stated they had used regimen), and history of COVID-19 as contributors to the face mask all the time (67/3,452; 1.94%). reducing the likelihood of experiencing a COVID-19 epi- The multivariate analysis among survey respondents sode within the 3-10 days following the event (Figure 2a, confirmed age and immunity (either complete or par- Table S4). The effect size of each of the factors was simi- tial, natural or vaccine-driven) as factors with significant lar in the two events, with a history of previous COVID- contribution to reducing the odds of COVID-19 19 being associated with the lowest in either of the (Figure 2b, Table S4). Additionally, behavioural factors events. such as complying with the overall preventive measures The anonymous survey launched 14 days after the and mask-wearing (all the time or most of the time, cor- second event was answered by 13,137 individuals. Of responding to the two highest scores on the 5-point them, 570 (4.34%) reported having been diagnosed of scale) also reduced the odds of COVID-19. Although the COVID-19 case to the central health registry within the variables “mask-wearing” and “compliance with the www.thelancet.com Vol 15 Month April, 2022 7 Articles overall preventive measures” showed a positive correla- One of the benefits from the Ag-RDT screening strat- tion (rho=0.23), the nested version excluding any of the egy was the identification of a remarkable number of two variables did not improve the model fit (Figure S2). asymptomatic SARS-CoV-2-infected individuals, which Another variable pair with important correlation was close to the absolute excess cases in one of the festi- included the vaccination status and age (rho=0.35), pre- vals. The screening effectiveness was clearly manifested sumably due to the age-driven rollout of vaccination by a lower risk of experiencing COVID-19 within the campaigns (Figure 1). Each additional attended day sig- first 3 days following the event among attendees com- nificantly increased risk of experiencing COVID-19 by pared controls. Therefore, the identification and possi- 1.14. Considering that all participants attended the event bility of contact-tracing and quarantine of these cases at least one day, a RR of 1.14 and 1.30 were associated that would be otherwise unnoticed should also be con- with attending the festival two days (i.e., one additional sidered when appraising the overall impact of these day) and the entire festival (i.e., two additional days), events in the pandemic course. Likewise, other factors respectively. such as clinical severity (likely associated with the indi- vidual profile), psychological, social, and economic well- being of these events should be considered. Our population-based analysis is strengthened by the Discussion analytical approach and the building of a population- Attendance to an outdoors mass-gathering event held based control group matched by age, sex, healthcare during a COVID-19 transmission peak was indepen- area of residence and immunity status for SARS-CoV-2. dently associated with a two-fold risk of post-event Although positive experiences with these types of events COVID-19 compared to a population-based matched have been conducted elsewhere, incidence data after the control group of non-attendees. These findings were event is often reported in press releases, with little cer- consistent in two independent music festivals with a tainty of accuracy,22−24 or descriptive studies without similar approach for preventing SARS-CoV-2 spread, formal comparison with non-attendees that allow esti- which included compulsory mask-wearing and point-of- mating the actual risk associated with the event.9,25 On care screening with Ag-RDT. We also found that previ- the one hand, the population-based control-matched ous COVID-19 infection, vaccination, adequate mask- approach constrained our study sample to individuals wearing, and following general recommendations with available data on the vaccination status and basic for preventing SARS-CoV-2 spread were significantly demographic information in administrative health associated with a lower risk of COVID-19. However, records, warranting high exhaustivity and representa- COVID-19 incidence among immunized festival attend- tiveness of the datasets. Although the high number of ees was also two-fold higher than controls. individuals included in the analysis (45,622 overall) pro- The increased risk observed in our cohort of attend- vides confidence in our risk estimates, this approach ees was in line with that reported for other social activi- limited the analysis sample, which excluded attendees ties, such as going to restaurants (OR 1.95 - 2.8) and without information in the central Healthcare Registry bars (OR 1.95 - 3.9).17-20 The effectiveness of face mask- (presumably non-Catalan individuals) and those with- wearing and Ag-RDT testing for preventing SARS-CoV- out individuals in the general population meeting the 2 spread in this type of events had been suggested in matching criteria. Another limitation of our population- previous experiences, which could not show an based analysis was the unfeasibility of considering increased risk of SARS-CoV-2 transmission.8-10 How- behavioural or socioeconomic characteristics for the ever, these previous studies were conducted during peri- pairing. Although the immunity status is expected to be ods of low COVID-19 incidence in the background the most relevant factor for COVID-19, virus transmis- population, thus resulting in an overall limited number sion may also be influenced by social behaviours, health of cases. The festivals included in our study were held awareness, fulfilling of public health recommendations, in a moment of notably higher background incidence in and factors related to the socioeconomic status, includ- the area: 620-to-658 7-day cumulative COVID-19 inci- ing the place of residence, profession, or household dence per 100,000 inhabitants by the time of the out- occupation.26,27 door events investigated vs. the 96-to-132 range Besides the population-based control-matched analy- reported in previous assessments of the impact of mass- sis, we collected behavioural information in one of the gathering indoor events.8-10 The higher transmission in two investigated events. The post-event survey showed the outdoor events included in our study highlights the that wearing a face mask and complying with protective relevance of the background COVID-19 incidence measures most of the time was independently associ- regarding the safety of mass-gathering events, which ated with a lower risk of COVID-19. Attending two or could be compromised by the sharp increase on the three days to a multiday festival also increased the risk number of false negative results of Ag-RDTs with of COVID-19 by 14% and 30%, respectively. The results increasing local prevalence (i.e., pre-test probability) from the post-event anonymous survey must be read in (Table S1, Supplementary Appendix 1).21 the context of the limited and not randomized sample 8 www.thelancet.com Vol 15 Month April, 2022 Articles of respondents, who were biased towards older-, female- E.C., D.O., E.H., B.B., M.A.R., M.M. and O.M. analysed and 3-days attendees. The positivity rate among survey data. C.S., E.C., D.O., M.M. and O.M. interpreted data. C. responders also indicated that positive cases were more S., E.C., D.O., M.M. and O.M. wrote the manuscript. All likely to answer. Thus, results from this analysis could authors reviewed and approved the final version of the be affected by these confounding factors. Furthermore, manuscript. C.S. and O.M. verified the underlying data. we cannot rule out the recall bias (i.e., inaccurate recol- lection of events by respondents) and social desirability bias (i.e., a tendency to give a favourable view of one’s Supplementary materials self).28 Nevertheless, anonymity, the high number of Supplementary material associated with this article can respondents (over 13,000 individuals), and the consis- be found in the online version at doi:10.1016/j. tency of the results with previous evidence regarding lanepe.2022.100337. the use of masks in crowded outdoor environments29 support the importance of this measure in preventing SARS-CoV-2 spread in mass-gathering events. References In summary, our results raise awareness on the limi- 1 Althouse BM, Wenger EA, Miller JC, et al. Superspreading events in the transmission dynamics of SARS-CoV-2: Opportunities for tations of Ag-RDT screening and mask-wearing for con- interventions and control. PLoS Biol. 2020;18:(11) e3000897. taining SARS-CoV-2 transmission in mass-gathering https://doi.org/10.1371/journal.pbio.3000897. events held during high COVID-19 incidence. Although 2 Thomas D. Music festivals face devastating blow from coronavirus. Financial Times. Published 2020. Accessed October 6, 2021. COVID-19 incidence was lower among vaccinated https://www.ft.com/content/59541249-f8c1-479f-a1b6. attendees, the higher risk compared with controls was 3 Centers for Disease Control and Prevention. Guidance for Organiz- ing Large Events and Gatherings. COVID-19. Published 2021. maintained, supporting the use of additional measures Accessed October 6, 2021. https://www.cdc.gov/coronavirus/2019- besides vaccination certificates. Therefore, these find- ncov/community/large-events/considerations-for-events-gather ings are relevant for countries with high vaccination ings.html. 4 Centers for Disease Control and Prevention (CDC). Community rates that are experiencing an increase of COVID-19 Use of Cloth Masks to Control the Spread of SARS-CoV-2. Science incidence and ¡more importantly¡ low- and middle- Brief. Accessed October 20, 2021. https://www.cdc.gov/coronavi income countries with low vaccination rates, where the rus/2019-ncov/science/science-briefs/masking-science-sars-cov2. html?CDC_AA_refVal=https%3A%2F%2Fwww.cdc. pandemic still wreaks havoc. gov%2Fcoronavirus%2F2019-ncov%2Fmore%2Fmasking-science- Availability of data sources: Data that support the sars-cov2.html. 5 World Health Organization. Mask use in the context of COVID-19. findings of this study are openly available on the Sup- Interim Guidance. Published 2020. Accessed October 20, 2021. plementary Appendix 3. https://apps.who.int/iris/bitstream/handle/10665/337199/WHO- 2019-nCov-IPC_Masks-2020.5-eng.pdf?sequence=1&isAllowed=y. 6 Alemany A, Baro B, Ouchi D, et al. Analytical and clinical perfor- mance of the panbio COVID-19 antigen-detecting rapid diagnostic Declaration of interests test. J Infect. 2021;18(0):16. https://doi.org/10.1016/j. The authors declare no competing interests. jinf.2020.12.033. 7 Baro B, Rodo P, Ouchi D, et al. Performance characteristics of five antigen-detecting rapid diagnostic test (Ag-RDT) for SARS-CoV-2 asymptomatic infection: a head-to-head benchmark comparison. J Infect. 2021;82(6):269–275. https://doi.org/10.1016/j. Acknowledgments jinf.2021.04.009. We thank Gerard Carot-Sans for providing medical writ- 8 Revollo B, Blanco I, Soler P, et al. Same-day SARS-CoV-2 antigen ing support with manuscript preparation and Roser test screening in an indoor mass-gathering live music event: a randomised controlled trial. Lancet Infect Dis. 2021;0(0). https:// Escrig for her support in the study design and medical doi.org/10.1016/s1473-3099(21)00268-1. writing assistance with the study documentation. We 9 Llibre JM, Videla S, Clotet B, Revollo B. Screening for SARS-CoV-2 also thank Laia Bertran for her support on the project antigen before a live indoor music concert: an observational study. Ann Intern Med. 2021. https://doi.org/10.7326/m21-2278. Pub- management. We also thank the festival organizers for lished online July 20. their collaboration. Funding sources were the crowd- 10 Delaugerre C, Foissac F, Abdoul H, et al. Prevention of SARS-CoV- 2 transmission during a large, live, indoor gathering (SPRING): a funding campaign YoMeCorono (https://www.yomecor non-inferiority, randomised, controlled trial. Lancet Infect Dis. ono.com/) and the Generalitat de Catalunya. The 2021. doi.org/10.1016/S1473-3099(21)00673-3. funders had no role in study design, data collection and 11 Ong SWX, Chiew CJ, Ang LW, et al. Clinical and virological fea- tures of SARS-CoV-2 variants of concern: a retrospective cohort analysis, decision to publish or preparation of the man- study comparing B.1.1.7 (Alpha), B.1.315 (Beta), and B.1.617.2 uscript. BB is a Beatriu de Pinos postdoctoral fellow (Delta). SSRN Electron J. 2021. https://doi.org/10.2139/ granted by the Government of Catalonia's Secretariat ssrn.3861566. 12 Catala-Moll F, Noguera-Julian M. CovidTag. http://covidtag.paseq. for Universities and Research, and by Marie Sklodow- org/ (accessed Jan 20, 2022). ska-Curie Actions COFUND Programme (BP3, 801370) 13 Morales DR, Conover MM, You SC, et al. Renin−angiotensin sys- tem blockers and susceptibility to COVID-19: an international, open science, cohort analysis. Lancet Digit Heal. 2021;3(2):e98– e114. https://doi.org/10.1016/S2589-7500(20)30289-2. Author contributions 14 Burn E, Tebe C, Fernandez-Bertolin S, et al. The natural history of symptomatic COVID-19 during the first wave in Catalonia. Nat C.S., E.C., B.C., M.M. and O.M. designed the study. C. Commun. 2021;12(1):1–12. https://doi.org/10.1038/s41467-021- S., E.C., D.O., E.H., J.P. and M.M. collected data. C.S., 21100-y. www.thelancet.com Vol 15 Month April, 2022 9 Articles 15 Cabezas C, Coma E, Mora-Fernandez N, et al. Associations of Accessed October 20, 2021. https://www.nbcchicago.com/news/ BNT162b2 vaccination with SARS-CoV-2 infection and hospital coronavirus/lollapalooza-has-no-connection-to-increases-in-covid- admission and death with covid-19 in nursing homes and health- cases-in-chicago-arwady/2592576/. care workers in Catalonia: prospective cohort study. BMJ. 23 Ellyatt H. Music festival in the Netherlands leads to over 1,000 2021;374:n1868. https://doi.org/10.1136/bmj.n1868. Covid infections. CNBC. Accessed October 20, 2021. https://www. 16 R Core Team. R: A language and environment for statistical com- cnbc.com/2021/07/15/music-festival-in-holland-leads-to-over- puting. R Found. Stat Comput Vienna, Austria.. 2017.. https:// 1000-covid-infections.html. www.r-project.org. (accessed Dec 20, 2021). 24 Youngs I. Covid: Packed pilot festival brings the good times back - 17 Fisher KA, Tenforde MW, Feldstein LR, et al. Community and for one night. BBC. Accessed October 20, 2021. https://www.bbc. Close Contact Exposures Associated with COVID-19 Among Symp- com/news/entertainment-arts-56962231. tomatic Adults ≥18 Years in 11 Outpatient Health Care Facilities — 25 Domenech-Montoliu S, Pac-Sa MR, Vidal-Utrillas P, et al. Mass United States, July 2020. CDC - Morbidity and Mortality Weekly gathering events and COVID-19 transmission in Borriana (Spain): Report. doi:10.15585/mmwr.mm6936a5. A retrospective cohort study. PLoS One. 2021;16(8 August):1–29. 18 Tenforde MW, Fisher KA, Patel MM. Identifying COVID-19 risk https://doi.org/10.1371/journal.pone.0256747. through observational studies to inform control measures. JAMA - 26 Amengual-Moreno M, Calafat-Caules M, Carot A, RC AR, Rıo- J Am Med Assoc. 2021;325(14):1464–1465. https://doi.org/10.1001/ Berge C, Ventura-Gabarro C. Social determinants of the incidence jama.2021.1995. of Covid-19 in Barcelona: a preliminary ecological study using pub- 19 Lin A, Vittinghoff E, Olgin J, et al. Predictors of incident SARS- lic data. Rev Esp Salud Publica. 2020;94:(94) e202009101. CoV-2 infections in an international prospective cohort study. BMJ 27 Patel JA, Nielsen FBH, Badiani AA, et al. Poverty, inequality and Open. 2021;11:(9) e052025. https://doi.org/10.1136/bmjopen-2021- COVID-19: the forgotten vulnerable. Public Health. 2020;183:110– 052025. 111. https://doi.org/10.1016/j.puhe.2020.05.006. 20 Galmiche S, Charmet T, Schaeffer L, et al. Exposures associated 28 Chang EM, Gillespie EF, Shaverdian N. Truthfulness in patient- with SARS-CoV-2 infection in France: a nationwide online case- reported outcomes: factors affecting patients’ responses and impact control study. Lancet Reg Heal - Eur. 2021;7: 100148. https://doi. on data quality. Patient Relat Outcome Meas. 2019;10:171. https:// org/10.1016/j.lanepe.2021.100148. doi.org/10.2147/PROM.S178344. 21 Watson J, Whiting PF, Brush JE. Interpreting a covid-19 test result. 29 Chu DK, Akl EA, Duda S, et al. Physical distancing, face masks, BMJ. 2020;369. https://doi.org/10.1136/bmj.m1808. 10.1136/bmj. and eye protection to prevent person-to-person transmission of m1808. SARS-CoV-2 and COVID-19: a systematic review and meta-analy- 22 Arwady A. Lollapalooza Has ‘No Connection’’ to Increases in sis. Lancet. 2020;395(10242):1973–1987. https://doi.org/10.1016/ COVID Cases in Chicago: Arwady.’ 5chicago. Published 2021. S0140-6736(20)31142-9. 10 www.thelancet.com Vol 15 Month April, 2022