Human stampedes, which occur around the world, are an important and unique yet incompletely understood type of disaster resulting from complex and often lethal dynamics that occur at large crowd gatherings. Given that human stampedes are a recurring phenomenon that can trigger mass casualties ranging into the thousands, they must be considered alongside other significant causes of disaster events. By definition, a stampede may be characterized as an impulsive mass movement of a crowd that often results in injuries and deaths. Another commonly associated term is trampling that leads to infliction of crushing casualties. Although human stampedes have widely been reported in the media, they have attracted scant scientific attention or investigation, limiting basic understanding of the pathophysiology and behavioral aspects of these events. This article is a review of elements within the context of historically significant human stampedes and specific events described in the literature. Strategies for preincident preparedness informed by improved systematic categorization and identification of human stampede determinants including pathophysiology, mechanisms, settings, and triggers are explored.

HUMAN PATHOPHYSIOLOGY OF STAMPEDES

The vast majority of human stampede casualties result from traumatic asphyxia caused by external compression of the thorax and/or upper abdomen, resulting in complete or partial cessation of respiration.Reference Gill and Landi1–Reference Byard, Wick and Simpson8 Significant compression forces can be present with even moderate crowds; forces of up to 4500 N (1000 lb) can be generated by just 6 to 7 people pushing in a single direction with forces large enough to bend steel railings.Reference Elliott and Smith9, Reference Smith and Dickie10

The earliest stampede victims with facial petechiae were characterized as “masque ecchymotique.”Reference Fred and Chandler5–Reference Byard, Wick and Simpson8 Subpleural petechiae termed “Tardieu spots” and Perthes's clinical description formed the original definition of traumatic asphyxia or crush asphyxia.Reference Fred and Chandler5,Reference Williams, Minken and Adams6,Reference Byard, Wick and Simpson8,Reference Lowe, Rapini and Johnson11,Reference Byard12

Clinical signs include conjunctival petechiae, facial petechiae, cutaneous blue–purple discoloration and congestion of the face and neck, and seizures, confusion, and unconsciousness.Reference Fred and Chandler5–Reference Byard, Wick and Simpson8 In forensic studies, up to 95% of cases presented with at least 1 of the 3 classical signs—conjunctival petechiae, facial petechiae, and congestion of the face and neck attributed to venous congestion from external pressure hindering venous flow.Reference Sklar, Baack and McFeeley7, Reference Byard, Wick and Simpson8 Neurological findings, such as loss of consciousness and confusion due to cerebral anoxia, were believed to be transient and no specific pathological brain tissue findings were reported. Reference Jongewaard, Cogbill and Landercasper7,13,14

Although survivors of human stampedes and autopsy reports suggest traumatic asphyxia as the primary cause of death, other mechanisms have been considered, including myocardial infarction, direct crushing injury to intrathoracic or intraabdominal organs, head injury, and neck compression.Reference Simpson15 All of these mechanisms are possible; however, little actual supportive evidence exists.

Gill and LandiReference Gill and Landi1 concluded from autopsy findings that “people who succumb in these scenarios typically die (standing up) in a vertical position” due to compression force and “do not collapse to the floor until after the crowd density and pressure have been relieved.” Compressive forces applied front to back or vice versa resulted in ventilatory failure, whereas those experiencing compressive forces from side to side were spared, presumably because chest expansion was not compromised to the same extent.

The full spectrum of injuries—including fractures, dislocations, and other mechanical injuries—may be expected. Among survivors, many may suffer from posttraumatic stress, grief, or survivor guilt and require psychological counseling or intervention.Reference Thompson16

STAMPEDE MECHANISMS

Models have been developed to depict behavior with pedestrian flow.Reference Yu, Chen and Dong17 To date, only a few studies take into account the social–psychological behavior of escape panicReference Helbing, Farkas and Vicsek18, Reference Helbing, Johansson and Al-Abideen19 (Table 1). In situations leading to stampedes, crowds do not stop accumulating even with local densities up to 10 people per square meter. Quantitative measurement taking into account crowd density and speed of movement have been suggested as predictive warning signs of a stampede.Reference Helbing, Farkas and Vicsek18, Reference Helbing, Johansson and Al-Abideen19 Classic escape panic is described by a “headlong rush away from something” while a craze is “a rush toward something believed to be gratifying.”Reference Johnson20, Reference Smelser21 In either situation, psychological elements play a central role in stampede events, which effectively always involve mass behavior.

TABLE 1

Features of Escape Panic

Human stampedes may be characterized as unidirectional or turbulent. Unidirectional stampede events may occur when a crowd moving in the same direction encounters a sudden positive or negative change in force, thus altering the movement of the crowd. Examples of positive forces include a “sudden stop” situation, such as the closing of a door or gate or a fallen bystander. Reference Dickie3,20,26 Positive force can also result from narrow passages or a bottleneck phenomenon.Reference Dickie26 A sudden decrease in force, such as in the case of a barrier collapse or gate opening, may also trigger a unidirectional stampede event.Reference Gill and Landi1, Reference DeAngeles, Schurr and Birnbaum2 In contrast, turbulent stampede events can occur in situations with uncontrolled crowds, induced panic, or crowds merging from numerous directions.Reference Helbing, Johansson and Al-Abideen19, Reference Madzimbamuto and Madamombe27

NOTABLE STAMPEDE EVENTS

Much of what is known about human stampedes is derived from anecdotal news reports and observations.28–34 In recent decades, the toll from human stampedes has been increasingly well documented. Although there appears to be an uptick in the occurrence of deadly stampedes mirroring risks of increasing overall population densities, especially in the developing world, the number of reports may in reality simply reflect greater sharing of information via mass media. Numerous other events before the 1990s, particularly in less-developed countries, may have resulted in sizeable stampedes that were unreported. In general, injuries are estimated or are not recorded, whereas direct deaths (not including deaths following injuries) are more precisely reported and readily obtained. Although we recognize the subjective nature of categorizing severity based upon resultant mortality, we propose the following logarithmic stampede scale to encompass the entire observed casualty range: class I (mild): injuries, 0 deaths; class II (moderate): 1 to 10 deaths; class III (severe): 11 to 100 deaths; class IV (devastating): 101 to 1000 deaths; class V (catastrophic): >1000 deaths. This scale could be used during operational response to readily convey the estimated magnitude of an event and mobilize necessary resources. Classification according to this scale may also offer a unified nomenclature to facilitate research in identifying characteristics associated with different reported stampede events. Further study is necessary to determine how injuries alone affect the scale of an event and whether any relation exists between injuries and deaths during human stampedes. Table 2 lists the deadliest documented human stampede events.

TABLE 2

Deadliest Documented Human Stampede Events

Mass gatherings during political, sporting, and religious events historically have been important backdrops against which human stampedes occur. Typically, human stampedes are then triggered by a real or perceived inciting event. One of the first documented human crowd disasters occurred on the eve of Russian Tsar Nicholas II's coronation ceremony in 1896 at Khodynka Field near Moscow.35 Greater than 1000 people were crushed or trampled to death when the crowd surged based on rumors that souvenirs were in short supply. Fearing cancellation of the ceremony would result in further civil unrest, the strewn corpses were quickly removed during the night and the event proceeded as scheduled the following day.

Another infamous incident occurred in 1913 during the Italian Hall disaster in Calumet, Michigan, when the false warning of “Fire!” was shouted during a holiday gathering of about 700 striking copper mine workers and their families.Reference Lehto36 During the ensuing panic, people stumbled and fell on the stairs, leading to a pileup of bodies and 73 deaths, including women and children. This incident along with other smaller-scale events served as the impetus for the US Supreme Court in 1919 to place certain limitations on the right to free speech, prohibiting acts such as “falsely shouting fire in a theatre and causing a panic.”37

Sporting events, particularly international soccer matches, have been marred by numerous deadly human stampedes. Two of history's deadliest stampedes occurred in the context of soccer matches. In 1964, a Peruvian stampede leading to 318 deaths was touched off by a controversial referee call during the final minutes of the game.29 Following another highly contested match in Russia in 1982, 340 fans were killed when some were pushed down a narrow, icy staircase.30

Most notably, many of the deadliest stampedes have occurred during various religious events. As the largest annual pilgrimage in the world, the Hajj requires the participation of all Muslims in good health at least once during their lifetime. With availability of modern transportation, more pilgrims are taking advantage of this opportunity from far-off regions of the world. Among the religious rituals, pilgrims circle counter-clockwise 7 times around a building called the Kaaba and during another ritual, known as the stoning of the devil, pilgrims throw a series of pebbles at 3 long walls along a path. Once at their destination, the crowds that converge during these rituals can number more than 2 million, raising major safety concerns. Despite increasing Hajj safety regulations, these events have led to tragic major human stampedes in 1987, 1990, 1994, 2004, and 2006, which account for half of the 10 deadliest recorded human stampedes. 31,32,34 In a separate religious festival in 2005, 965 Shia pilgrims perished in a stampede over a river bridge in northern Baghdad, Iraq, when rumors of a bombing spread.28

REVIEW OF EVIDENCE-BASED PUBLISHED STUDIES

RESULTS

The literature review identified relatively few published articles pertaining to human stampedes. The PubMed search identified only 8 articles in English of relevance among 137 citations.Reference DeAngeles, Schurr and Birnbaum2,Reference Begum3,Reference Helbing, Farkas and Vicsek18,Reference Helbing, Johansson and Al-Abideen19,Reference Madzimbamuto and Madamombe27,Reference Madzimbamuto38–Reference Memish, Venkatesh and Ahmed40 Hand searches of published references identified an additional 12 relevant articles.Reference Gill and Landi1,Reference Wardrope, Ryan and Clark4,Reference Elliott and Smith9,Reference Johnson20,Reference Dickie26,Reference Wardrope, Hockey and Crosby41–Reference Ahmed, Arabi and Memish47 Sixteen citations were returned from ERIC, none of which met inclusion criteria. No citations that met inclusion criteria were returned from the search of EMBASE.

Since 1970, only 8 human stampedes have been described in detail in the peer-reviewed literature (Table 3), the majority of which took place in the United Kingdom or the United States (5 of 8).Reference DeAngeles, Schurr and Birnbaum2–Reference Wardrope, Ryan and Clark4,Reference Elliott and Smith9,Reference Johnson20,Reference Dickie26,Reference Madzimbamuto and Madamombe27,Reference Madzimbamuto38,Reference Wardrope, Hockey and Crosby41–Reference Ahmed, Arabi and Memish47 More than half of the stampedes reported (5 of 8) were in the setting of sporting events, notably soccer games. Other mass gatherings such as concerts and religious occasions were associated with the other stampedes (3 of 7). Three events were triggered by falls, 2 events were triggered by locked doors or fixed barricades, and another 2 events resulted from sudden gate opening or barrier collapse. Events ranged from those with no fatalities to 380 dead. In general, the events described were smaller in scale with references to 1 class I event, 1 class II event, and 5 class III events. There was a single class IV event and no mention of class V events, the deadliest of recorded human stampedes. Although we sought to identify patterns, the sparseness of published literature on the topic of human stampedes combined with a complete absence of reports on greatest impact events significantly limits the conclusions that can be drawn.

TABLE 3

Human Stampedes Referenced in the Published Literature

Authors' Disclosures

The authors report no conflicts of interest.