Randomized Controlled Trial
Study population and setting
This was a randomized controlled trial involving 6,678 adults to evaluate the effectiveness of a package of SARS-CoV-2 transmission mitigation strategies in the context of a large indoor concert in Paris, France on May 29, 2021. Infection rates among concert attendees (n=3,917) were compared to those among non-attendees (n=1,947), and the intervention was evaluated for non-inferiority. Participants were recruited from a variety of media and randomized (2:1) to either concert attendance or non-attendance. Both participants and investigators were aware of treatment assignment. Eligible participants were Paris area residents aged 18-45 years with no comorbidities, who were not living with older or at-risk individuals, and who tested negative via a rapid antigen test performed by healthcare workers within the 3 days prior to the concert. Patients reporting symptoms compatible with COVID-19 or who reported contact with a person with a positive SARS-CoV-2 test within the past 14 days were excluded. The intervention consisted of the following elements: 1) presentation of negative rapid antigen tests at the arena gate; 2) distribution of a surgical facemask to all attendees with a mask-wearing mandate for the duration of the event; 3) mandatory hand sanitization at entrance with hand sanitizer stations throughout the arena; 4) a water bottle distributed to all participants with mask removal allowed for drinking; 5) closure of all arena bars and a ban on consumption of alcohol; and 6) ventilation employing 8 units, using only outside air without recirculation, on the arena floor and in all areas open to the public. Only the floor of the arena was open to the public (area=1,900 square meters). The show (one DJ set and one live set) lasted 2.5 hours and the doors were open for 4 hours; all artists and staff members (n=525) were tested for SARS-CoV-2 in the 3 days preceding the concert. All participants returned two self-collected saliva samples for RT-PCR testing (at day 0 and during a window from day 6-15); whole-genome sequencing was performed for SARS-CoV-2 subtyping and transmission cluster analysis. Cameras were used to capture mask-wearing behavior and neural networks were used to classify faces as a) not wearing a mask, b) wearing a mask inadequately (i.e., not covering the nose), or c) wearing a mask adequately; the proportion of adequate mask-wearing was calculated in 5-minute intervals and averaged over the 4 hours. The primary outcome was the proportion of PCR-confirmed SARS-CoV-2 infections at day 7.
Summary of Main Findings
The median age of participants was 28 years, and 59% were female; 43% were vaccinated with one dose and 7% were vaccinated with two doses, with 6% having received the second dose at least 14 days before the concert. Eight concert attendees (0.20%) tested positive on day 7 for SARS-CoV-2 infection, compared with three of the non-attendees (0.15%), for an estimated IRR of 1.33 (95% CI: 0.38 to 4.60). The 95% confidence interval for the absolute difference in incidence was -0.26% to 0.28, and the upper limit of this interval was lower than the prespecified threshold of 0.35% for non-inferiority. Five of the 8 attendees who tested positive at day 7 had also tested positive at day 0 (indicating prior infection), and an additional 5 attendees only tested positive at day 0. It is unclear whether the three remaining cases among attendees at day 7 acquired infection during the concert. Viral subtyping and cluster analyses did not indicate any transmission clusters at the concert, but not all positive results could be sequenced. None of the study participants required hospitalization. A sensitivity analysis using an imputed intention-to-treat dataset also found non-inferiority. In the mask-wearing behavior analysis, 33,349 faces were detected by the algorithm, and 85% were classified. Median mask-wearing compliance was estimated to be 90.0% (95% CI: 76.5 to 94.8) on the arena floor and 97.4% (74.1 to 99.9) in the lobby and staircases.
For an intervention of this size and complexity, there was a fairly low rate of protocol deviation (3.5% in the treatment arm and 12.6% in the control arm, in addition to 8.5% who were assigned to attend the concert but did not). Active follow-up ensured a high rate of sample collection. The camera-based mask compliance analysis was novel and permitted assessment of mask-wearing over multiple time periods.
This study took place when the Alpha variant was dominant in France, and it is unclear how results would generalize to the Delta variant or to other, more transmissible variants. Antigen testing during the 3 days preceding the concert did not detect 10 PCR-positive cases among attendees. The authors excluded 3.5% of attendees and 12.6% of non-attendees from analyses due to protocol violations (e.g., not returning saliva samples during the appropriate time window); estimation of SARS-CoV-2 incidence in either group could therefore be biased. Depending on the severity of the possible bias, the result of non-inferiority observed could be invalid (sensitivity analyses using intention-to-treat imputed datasets also concluded non-inferiority; while this is reassuring, these results depend on the validity of imputation of the missing saliva samples). The study was conducted among young adults; transmission risks may differ in other populations. Participants were aware that cameras were recording their mask-wearing behavior, which may have led to higher rates of mask compliance than would otherwise have occurred.
This is one of only two randomized controlled trials of an intervention to lower SARS-CoV-2 transmission risk at a public event; in fact, it is one of the only randomized controlled trials to date of any non-pharmaceutical intervention aimed at reducing the risk of SARS-CoV-2 transmission.