Introduction
Mass gatherings are major health risk situations.Reference Memish, Steffen and White 1 The problem is so global that even the World Health Organization (WHO) had to be involved in 2014. 2 Certain religious events lead to the gathering of several dozen, hundreds, or even thousands of pilgrims.Reference Cariappa, Singh and Mahen 3 Nevertheless, on a daily basis, most mass gatherings take place in dedicated enclosures, which are often stadiums. During these events, incidents occur regularly, with, in certain cases, disasters as recently as November 2022, when 125 people died during a football match in Indonesia.Reference Ratcliffe, Connett and Fulton 4 Such mass gatherings are also at-risk situations of terrorist attacks. In 2015, during the terrorist attacks in France, the first explosions occurred on the forecourt of the largest French stadium, the “Stade de France” (Saint-Denis, France), after terrorists were blocked by security guards at the entrance of the stadium (Table 1).Reference Hirsch, Carli and Nizard 5 The attack continued then on terraces of two bars, and finally in a concert hall (Le Bataclan, Paris), with a final death toll of 139 victims. In May 2017, a terrorist bombing attack was perpetrated in Manchester Arena (Manchester, England) during a concert. 23 people were killed (including the attacker), and 1,017 were injured.Reference Craigie, Farrelly and Santos 6
Chronology and casualty management during the terrorist attacks of November 13, 2015 10 - 12
66,866 spectators were attending this international France-Germany football match, including the President of the French Republic, François Hollande, and several ministers of the French government. Three suicide bombers with explosive belts tried, in vain, to enter the stadium enclosure (for undefined reasons) before detonating themselves on the stadium forecourt. As soon as the reality of a terrorist attack was confirmed, it was decided not to interrupt the football match and to keep the 66,866 spectators inside the stadium to ensure their safety, to facilitate emergency traffic, to prepare the evacuation of the wounded spectators if any, and to avoid panic movements. Almost simultaneously, multiple terrorist attacks took place in Paris (in the 10th and 11th arrondissements), where a series of bar and restaurant terraces were targeted (shooting and suicide bombing), as well as the Bataclan theater, where the spectators, attending a Hard-Rock concert, were shot for some and taken hostage for others by three attackers. A total of 1,746 civilian casualties were numbered, among whom 493 were physically wounded and 129 killed (excluding 7 terrorists killed). The chronology of events as well as the casualty management is detailed below.
* OCC, Operational Control Center.
In Western countries, strict safety rules apply to these enclosures with the aim, on the one hand, to limit the risk of accidents, and, on the other hand, to anticipate the organization of emergency services. This concern is all the more significant as the event is international in scope and takes place on several sites, sometimes simultaneously, as is the case for the Olympic Games. In 2024, France, and more precisely the Greater Paris area, will host this event. 7 The opening ceremony, which will take place only in intramural Paris, should alone bring together more than 600,000 spectators. 8 The largest French facility opened to the public is the Stade de France. 9 It was inaugurated in 1998, shortly before the Fédération Internationale de Football Association (FIFA) World Cup. It has been hosting events of all kinds for 25 years, and will be the main venue for the Olympics. The health care plan put in place during each of these events is essentially dictated by security constraints, usually taking little account of health risks that may occur. Thus, a football match between 2 teams with a historical dispute, such as those of Paris (Paris Saint-Germain) and Marseille (Olympique de Marseille), will be considered at very high risk without this clearly predicting medical needs inside the stadium. As the Olympic Games approach, it seemed particularly relevant to us to analyze the 25 past years of medical workload at the Stade de France to help predict future needs by identifying the determinants of medical workload levels.
Methods
Setting
The Stade de France is located in the town of Saint-Denis in the department of Seine-Saint-Denis, 8 kilometers north of Paris. It was inaugurated on January 28, 1998. It is 320 meters long, 280 meters wide, and 35 meters high (Figure 1). It has been designed to host both non-sporting and sporting events of all kinds, including motorsport. It can accommodate 80,000 seated spectators in a football or rugby match configuration. This capacity is reduced for athletics, the stands being moved back to clear the track. Conversely, it can be increased in certain events, for example, shows with spectators on the lawn. No theoretical limit was known in 1998 as it strongly depended on the configuration chosen for a show. The stadium is organized into 9 floors, including 3 floors of stands (Figure 2). It is divided into 4 zones: east, west, north, and south, isolated from each other. An operational control center (OCC) is located on level 5 of the western sector (Figure 2). There are 14 infirmaries located at strategic locations in the stadium: 1 dedicated to players on level 0, + 8 on level 3 including one dedicated to officials, + 2 on level 5 + 2 on level 6 + 1 main infirmary is located on level zero, equipped with three “shock rooms.”
Air photography (©Sylvie Montois) and global map of the Stade de France.
Detailed plan of the Stade de France with infirmaries locations (red crosses).
Meteorological conditions in Paris area are as follow (data extracted from the French Organization for Weather Forecast database, Météo-France): mean annual temperature: 12°C, ranging from 5°C in January to 20°C in July; mean annual rain volume: 637 mm/m2, ranging from 41 mm/m2 in February to 63 mm/m2 in May; mean annual hygrometry: 75%, ranging from 65% in July to 86% in December.
Specific Health Care System
Each event taking place at the Stade de France leads to the implementation of specialized emergency plans. According to French law, a preparatory meeting with the police commissioner of the department makes it possible to define and validate the provisional on-site health care system. The OCC is activated as soon as an event brings together more than half of the stadium’s capacity, namely 40,000 people, or at the request of the police commissioner. It gathers representatives of police and gendarmerie forces, of the Stade de France consortium, and of Saint-Denis city, under the authority of the police commissioner (or his representative). In addition, managers of metro and train companies are also called upon for a major concern: ensuring a secure outflow of tens of thousands of spectators leaving the stadium and reaching the nearby metro or train platforms in a few tens of minutes. The on-site health care system is run by a private health care provider ensuring medical coverage of the event. It is led by a global manager, an emergency physician dispatcher and a dispatcher assistant. Other health care representatives are present in the OCC: a first-aid responders’ coordinator, an emergency physician and the Emergency Operations Commander of the Paris Fire Brigade, and finally, an emergency dispatcher physician and a dispatcher assistant detached from the EMS of the Department 93 (so-called “Service d’Aide Médicale d’urgence,” SAMU 93). The management of patients in the infirmaries within the Stade de France is the responsibility of the teams of the association of first aid responders and the private health care provider. The coordination of the mobile rescue and medical teams is ensured by the physician and his assistant of the private health care provider. They send the teams to the patient, receive the medical reports and decide, if necessary, on the best means and the best conditions for his transport toward the on-site medical center. Transport to hospital decisions (as well as patient’s orientation) are made by the emergency physician in charge of the patient in the infirmary, in agreement with the 2 dispatchers located at the OCC. In case of a need for evacuation outside the stadium, the intervention is overtaken by the emergency physician and the assistant dispatchers of the SAMU-EMS. Their role is to liaise with local hospitals, to provide transportation means, including Mobile Intensive Care Units (MICUs, staffed by an emergency physician) if necessary. In the event of an Exceptional Healthcare Situation (e.g., mass casualty), the representatives of the public services present at the OCC take over all health care tasks, including those of the private health care provider, under the responsibility of the police commissioner.
The number of rescuers required for a large public event is regulated by law (decree of November 7, 2006), setting the national reference system related to Provisional Healthcare Systems. Police services assess the risk associated with the event, taking into account the nature of the event, the number of spectators expected, and, for example, the existence of a dispute between supporters. They determine the number of infirmaries to be activated, as well as the number of first aid responders needed, emergency physicians and nurses to be engaged in the system, apart from the dispatcher physician and the assistant of the SAMU-EMS. In a standard configuration for an international match, as for the Union of European Football Associations (UEFA) Champions League final between Liverpool FC (England) and Real Madrid FC (Spain) in May 2022, the system included the opening of all infirmaries, mobilizing 70 first-aid responders, 6 nurses, 11 doctors and about 1,600 security guards.
Study Design
Since the first event at the Stade de France, information on medical workload has been collected prospectively by the SAMU-EMS emergency physician dispatcher present at the OCC. We therefore included all the events that took place at the Stade de France, giving rise to the OCC activation from January 1, 1998 to December 31, 2022. Events which did not need the OCC activation or which took place behind closed doors—during the period of the COVID epidemic—were excluded.
The parameters recorded were as follows:
-
- Date and Time: For the analysis, we have distinguished the events taking place on weekdays (Monday to Friday) and weekends (Saturday to Sunday) as well as events taking place during the day (doors opening before 6 p.m.) or in the evening (doors opening after 6 p.m.). The duration of the event corresponded to the actual duration of the event itself, such as for a football match, from the first to the final whistle. The times from the opening of the stadium gates until the start of the event and the time it took for spectators to leave the stadium at the end of the event were ignored. The events were distributed into quartiles from 1998 to 2022 in order to test the hypothesis of a variation in medical workload over time.
-
- Nature of the Event: Artistic performances (concerts, one-man shows, etc.), football matches, rugby matches, athletics competitions, motor sports, and other events.
-
- Number of spectators.
-
- Level of Event: National or international. For artistic performances, the level was considered national if the artist was French, international in other cases.
-
- Issue: The finals of sports competitions were identified independently of the level of the event.
-
- Weather: The temperature at the start of the event and the presence of rain or sun were specifically recorded.
-
- Medical Workload: The patient presentation, defined as a spectator seen in the infirmaries, was noted as well as the transport to a hospital and the medicalized transport to a hospital when done by a Mobile Intensive Care Unit.Reference Lund, Turris and Bowles 13 In accordance with what has been proposed in previous publications, a standardization of rates for 10,000 spectators has been used, defining patient presentation rate (PPR), transport to hospital rate (TTHR), and medicalized transport to hospital rate (MTTHR).Reference Lund, Turris and Bowles 13 , Reference Arbon, Bridgewater and Smith 14
End-point: The patient presentation rate was retained as the primary end-point to perform multivariate analysis.
Analysis
The 75th percentile of the patient presentation rate has been arbitrarily chosen to define a high level of medical workload. The impact of the various parameters recorded on this level of medical workload was tested in univariate analysis (Chi-square and Mann-Whitney tests—as variables were expected not to be normally distributed). For the nature of the events, events representing less than 6% of the total have been combined into a single category. End-points with a P-value < 0.2 were included in the multivariate analysis. High medical workload was included in the analysis as a variable to be explained. Qualitative variables are expressed in numbers (percentage) and quantitative variables in median [25-75% IQR]. A value of P < 0.05 was retained as the significance threshold.
Results
Of 480 events, 459 (96%) were included in the study. 21 (4%) events were excluded from the analysis, 12 during the COVID-19 epidemic period, 9 because OCC was not open, all with less than 55,000 spectators, and all before 2006. 167 (36%) were football matches, 142 (31%) rugby matches, 111 (24%) artistic performances, 26 (6%) athletics competitions, 11 (2%) motor sports competitions, and 2 (0.5%) other types of events (windsurfing competitions and “forgotten holidays” for children) (Figure 3).
Selection of events analyzed, 1998 to 2022 (Flow chart). OCC: operational control center
30,122,042 spectators attended these events, i.e., a median of 72,057 [56,825-78,500] spectators per event. Peak attendance was close to 95,000 at a U2 concert in summer 2009 (Table 2).
Events at the Stade de France taking place at the most spectators, patient presentations in the stadium infirmaries, transports to hospital and medicalized transports to hospital
* See Table 1.
20,672 patient presentations were recorded, corresponding to a median of 29 (15-59) per event or a PPR of 5 (2-9) per 10,000 spectators. The maximum was 448 patient presentations during a Rolling Stones concert in summer 1998 (Table 2).
The management of these people gave rise to 1,060 (non-medicalized) transports to hospital, i.e., a median of 2 (1-3) per event, or a TTHR of 0.3 [0.1-0.5] per 10,000 spectators. Finally, 112 medicalized transports to hospital were carried out, i.e., a median of 0 [0-0] per event, or an MTTHR of 0 [0-0] for 10,000 spectators. The maximum number of (non-medicalized) transports to hospital was 33 during a France-England football match in winter 2015, and the maximum number of medicalized transports to hospital was 4 during a French Paris-Bordeaux football match in spring 1998. The maximum number of (non-medicalized) transports to hospital took place in a particular context of terrorist attacks on the evening of November 13, 2015. Data on the impact of the terrorist attack at the Stade de France are detailed in Table 1.
A high level of medical workload (above the 75th percentile) was set at 70 patient presentations. In univariate analysis, the nature of the event (in descending order, artistic performances, athletics competitions, football and rugby matches), the number of spectators, the temperature, the time of the day, the stake of the event, its level, duration, and weather, were significantly associated with a high level of medical workload. Conversely, the day of the week and the period since 1998 were not significantly associated with a high level of medical workload. The results are detailed in Table 3.
Parameters associated with a high level of medical workload at the Stade de France—univariate analysis
* The statistical test (Chi-square) focused on entertainment events, football, rugby, athletics representing 97% of events.
The results are expressed in N (%) and medians (IQ).
These criteria were included in the multivariate analysis. However, the collinearity between the day vs evening, sun, rain, and temperature parameters being proven, the day vs evening variable was discarded. The nature of the event (rugby), level of the event (international), and temperature were associated with a greater frequency of high medical workload in multivariate analysis. The results are detailed in Table 4.
Parameters associated with a high level of medical workload at the Stade de France—multivariate analysis
* For 10 more minutes of event duration.
Discussion
More than 20,000 people—among more than 30 million spectators—have been cared for in the infirmaries during the 459 events that have taken place at the Stade de France since 1999. The nature of the event was a major independent determinant of medical workload. The importance of the stake (international vs national) and the ambient temperature also favored a high level of medical workload. Conversely, the number of spectators and the duration of the event had no influence.
Most of mass gathering studies have been carried out on series of homogeneous events over a given period, e.g., pilgrimages, series of concerts, football or rugby matches during a World Cup or the Olympic Games. In the case of the Stade de France, the events were extremely variable, ranging from sports events and concerts to car races on ice, a windsurfing event, and various music concerts from Hard-Rock to classical music and opera. The interest here was to identify “at-risk” events among all these. Rugby matches were the only risk event category after multivariate analysis (with an OR = 8). Similar results have already been published. However, the number of patient presentations remained higher than that reported during the 2019 Rugby World Cup in Japan (respectively 5.0 vs 2.3 per 10,0000 spectators).Reference Tajima, Takazawa and Yamada 15 Alcohol consumption has been considered the main cause of excess medical workload during rugby matches.Reference Tajima, Takazawa and Yamada 15 –Reference Delany, Crilly and Ranse 17 Yet, banning the sale of alcohol inside the stadium did not seem to have a favorable result.Reference Spaite, Meislin and Valenzuela 18 Furthermore, alcohol availability before the event and around the stadium is a major confusing factor to analyze its impact. This is particularly true for the Stade de France, where this is almost always the case, regardless of the rules applied inside the stadium, alcohol available or not. In addition to alcohol consumption, the authors reported increased medical workload when the ambient temperatures were high. We also report a significant interaction between ambient temperature and the level of medical workload, especially for temperatures above 25°C, but such situations were rarely (49 events) reached in our series. Indeed, the median ambient temperature was 16°C and even below 20°C in 75% of cases. Rain, sun, and duration of exposure to weather conditions were not related to the level of medical workload. The existence of a certain level of interaction between these parameters and the ambient temperature is quite likely.Reference Perron, Brady and Custalow 19 Meteorological conditions impact is probably dependent on mean local conditions. The heat index, associating ambient hygrometry and temperature, has been recently proposed to assess this impact.Reference Perron, Brady and Custalow 19 , Reference Locoh-Donou, Yan and Berry 20
In the event of a sports competition and more particularly a match between 2 collective sports teams, the outcome of the match was considered to be able to influence the level of medical workload.Reference Moore, Shepherd and Eden 21 Such information is difficult to interpret. It assumes a “relation of interest” between the public and the result, which is not always the case.Reference Slabbert and Ukpere 22 This information was not collected in our series. Nevertheless, the level of medical workload was significantly related to the outcome, characterized here by the national or international importance of the event. By itself, the level of competition is poorly documented but seems a factor in medical workload.Reference Van Remoortel, Scheers and De Buck 23 This can be explained by the fact that the public concerned is not the same.
Our study took place in the perspective of the preparation of the emergency health care system for the Paris 2024 Olympic Games. Athletics and football were the most represented Olympic sports in this series. They did not seem to be associated with an increased risk of medical workload. It would be questionable to extrapolate the results observed with rugby 15-a-side to rugby as played at the Olympic Games (7-a-side). The latter is infinitely less popular in France and in most countries. In other words, the nature of the sports does not seem to be such as to significantly increase the level of medical workload. Moreover, the World Athletics Championship, which took place at the Stade de France in 2003, did not give rise (results not specified) to a remarkable level of medical workload. Still, the ambient temperature (which will be summer) and the importance of the event (which will be international) may modulate this workload. With, moreover, an audience whose profile is certainly different from the one usually encountered at the Stade de France. Different models have been proposed for sizing the health care system.Reference Smith, Cosgrove and Driscoll 24 The number of spectators, as we have reported here, is not a major determinant.Reference De Lorenzo, Gray and Bennett 25 Moreover, the models generally consider sporting and non-sporting events separately.Reference Smith, Tuffin and Stratton 26 , Reference Spaepen, Haenen and Kaufman 27 The authors emphasize the importance of not undersizing the system. Drawing inspiration from past experience is complex. During the last Olympic Games (Tokyo, 2021), attention was focused on the COVID-19 epidemic.Reference Morimura, Mizobata and Sugita 28 The same was true for Rio (Brazil) in 2016 with Zika.Reference Zumla, McCloskey and Bin Saeed 29 Finally, the impact on the emergency departments close to the event seems modest.Reference Todkill, Hughes and Elliot 30
Limitations
Interactions between the potential determinants of medical workload studied cannot be ruled out. Music concerts taking place, for example, in the summer, in the evening, sport finals being most often played in the evening, etc. The multivariate analysis including a series of more than 450 events certainly limits the bias. The risk of infection is the subject of specific attention in mass-gatherings management. It gives rise to numerous works and recommendations.Reference Memish, Steffen and White 1 , Reference Ebrahim, Memish and Uyeki 31 , Reference Memish, Zumla and McCloskey 32 It has not been considered in this work. We are just reporting that the events organized during the period of the pandemic were not included in our analysis, because they were held behind closed doors. Recent studies have underlined the interest of “heat index” in such analysis.Reference Perron, Brady and Custalow 19 Unfortunately, it was not routinely used in 1988 when we started our prospective data collection. It was impossible to obtain these data exhaustively retrospectively—including from the database of the French organization for weather forecast (Météo-France). Nevertheless, data on presence/absence of rain during the events allowed us to mitigate the potential impact of other missing data, including in the multivariate analysis during the event. Finally, the causes of medical consultations were not collected prospectively in our register. However, only 6% of patients required transport to hospital, by non-medicalized transportation in 90% of cases. Assessment by an experienced emergency physician in the infirmary and, if required, discussion with 2 other emergency medicine physicians in the OCC probably contributed to limit transport to hospital. Therefore, the remaining 94% of consultations can be considered as medically benign.
Conclusion
The nature of the event was a major independent determinant of medical workload in this analysis of 459 events that have taken place at the Stade de France since 1999. More than 20,000 people, among more than 30 million spectators, have been cared for in the infirmaries. The importance of the stake (international vs national) and the ambient temperature also favored a high level of workload. Conversely, the number of spectators and the duration of the event had no influence. These results could contribute to optimizing the safety and health system for the next 2024 Olympic Games in Paris.
Conflict of interest
