Human Stampede: An Unexamined Threat

1 Comment

From the incomprehensible disaster in Cambodia to senseless injuries during Black Friday shopping, human stampedes are a present reality in emergency care that can be studied, predicted and ultimately avoided.

From the incomprehensible disaster in Cambodia to senseless injuries during Black Friday shopping, human stampedes are a present reality in emergency care that can be studied, predicted and ultimately avoided.


On November 22, 2010, millions of people gathered in Phnom Penh, Cambodia in celebration of the three day long Water Festival, known as Bon Om Touk, traditionally commemorating the end of the annual rainy season. On the last day of the festival, thousands of participants crossed back and forth on a narrow suspension bridge connecting Koh Pich, or Diamond Island, with the city. They packed the bridge, moving chaotically as they merged from both directions. The crowds became incredibly dense, creating critical chokepoints for foot traffic. And then, before officials could recognize the warning signs, the crowd turned into a powerful, deadly force.

While the inciting cause of the stampede remains speculative, some indicated that it began when some individuals passed out on the bridge. Others cited panicked rumors that the swaying bridge was about to collapse. However it started, the tragedy resulted in 347 deaths and hundreds of injuries 1 – primarily due to asphyxiation or crushing. Bodies were piled up five or six deep as trampled victims struggled to extricate themselves amidst the screams and chaos. The stampede was the worst disaster in Cambodia since the fall of the Khmer Rouge more than thirty years ago. By all accounts, this marked the event as one of the deadliest stampedes in the world over the past century and the third deadliest in the last decade. 2 (Table 1)



Lest you think this tragedy was some rare fluke, note that the Water Festival stampede was the fourth major human stampede to have occurred this year alone. There was a stampede at the Ram Janki Temple in Kunda, India which killed at least 63, and there was the Love Parade disaster in Duisberg, Germany which resulted in 21 deaths and over five hundred injuries. 3,4

Although stampedes more commonly occur in developing countries, developed countries are far from immune. In the United States, stampedes have occurred time and again during Black Friday sales, including the trampling death of a Walmart employee in November 2008 and another crowd surge that nearly turned deadly at a Target store in Buffalo this year. In March 2009, a crowd of thousands of hopefuls who had gathered in New York City to audition for America’s Next Top Model stampeded due to rumors of a car bomb.  

Direct forces generated by a crowd surge are staggering and can easily break down gates or even bend metal structures. Most of the deaths in the Cambodian stampede resulted from suffocation and crushing internal injuries. Others drowned in the river below and still others were electrocuted by exposed electrical cables when use of water cannons (in a botched attempt at restoring order by the military) were aimed at the fleeing crowd.

The nature of this particular stampede in many ways follows a known pattern.


  • It occurred in southeast Asia, a part of the world that has had among the highest frequency of stampedes.
  • Local resources were completely overwhelmed, and since no emergency access routes were available, it took upwards of thirty minutes for rescuers to reach victims to begin to deliver medical aid.
  • Those who succumbed to life-threatening injuries because they were not reached in time may have survived under different conditions.
  • Most victims were women, consistent with observations that those less readily able to protect themselves are disproportionately affected. 
  • The event was not a sport, religious, music or political event (classified as an NSRMP event). This classification carries an exceptionally high relative fatality rate, perhaps due to poorly organized or nonexistent planning measures when judged against other events of comparable crowd size. 5

What can we learn from prior stampedes and crowd disasters? The conditions are present for a lethal stampede at any gathering of significant size, particularly when crowd control precautions are lax or suboptimal.  During disaster conditions, such as in post-quake Haiti, where hundreds of victims may simultaneously seek treatment at local hospitals, victims could easily overwhelm local resources, distribution operations and facilities. This is why it is particularly important for prehospital and emergency medicine providers to learn the lessons from crowd disasters that have occurred. Perhaps the most important lesson to understand is that stampedes are neither unforeseeable nor inevitable; they follow predictable patterns and conditions. While not perfect, strict traffic control at bottlenecks, well-marked directions, use of public announcement systems with clear messaging, and suitable ingress and egress routes for emergency personnel can avert the calamitous consequences of a crowd surge.

Due to recurrent lethal crowd disasters, extensive planning measures have been taken by some countries, like Saudi Arabia, that have been successful to date. For example, to accommodate the millions of pilgrims during the Hajj – an annual Islamic pilgrimage that has been the site of several of the deadliest stampedes – Saudi Arabia constructed a $1.2 billion dollar state-of-the-art Jamarat bridge that features automated foot traffic controls, video monitoring and surveillance technology. Organizers firmly believe that this has greatly reduced the risk of stampedes to those in attendance. Indeed, none have occurred since 2006. For developing countries which do not have widespread resources at their disposal, other relatively simple and cost-effective solutions, such as increasing the number of access points and instituting stricter traffic control at bottlenecks, may still be implemented at events where large mass gatherings are anticipated. Emergency physicians may take a proactive role by working with local event planners in their community to address potential medical needs for mass gatherings. There is no substitute for thorough and careful planning.

Though widely reported in the press, stampedes have received little scientific attention and remain incompletely understood. Despite the frequency of events and the number of fatalities, information from human stampede events are not currently monitored by the WHO Collaborating Centre for Research on the Epidemiology of Disasters nor other official agencies.6 Casualty figures, including reported deaths and injuries, are available through the internet primarily based upon anecdotal news reports which may vary considerably and contain inaccuracies. Given the unique nature of such events, it is imperative to better understand crowd disasters. As a starting point, formal tracking of stampedes is required to collect better  data that can be rigorously studied and analyzed by physicians working closely with disaster planners and researchers. Additional epidemiological research on human stampedes is urgently needed to improve our collective understanding of the causes of crowd disasters and how to best prevent them.

Dr. Ed Hsu is the Director of Training at the Johns Hopkins Office of Critical Event Preparedness and Response (CEPAR) and an Associate Professor in the Dept. of EM. A focus of his recent research has been on the epidemiology and dynamics of crowd disasters.

1.     Cheang, S. Fear of bridge collapse set off Cambodia stampede. Availab
le at: (accessed Nov 24, 2010).

2.     Ngai KM, Burkle FM Jr., Hsu A, et al. Human stampedes: a systematic review of historical and peer-reviewed sources. Disaster Med Public Health Prep 2009; 3:191-95.

3.     Burkle FM Jr, Hsu EB. Ram Janki Temple: understanding human stampedes. Lancet. 2010 Jul 22. [Epub ahead of print].

4.     Wikipedia: List of accidents and disasters by death toll. Available at:  (accessed Nov 30, 2010).

5.     Hsieh YH, Ngai KM, Burkle FM Jr, et al. Epidemiological characteristics of human stampedes. Disaster Med Public Health Prep 2009; 3:217-23.

6.     Centre for research on the Epidemiology of Disasters (CRED). Available at  (accessed Nov 30, 2010).

1 Comment

  1. There is an underlying assumption here that the Cambodian tragedy (and other crowd disasters) was incited by something. The use of the word “stampede” is, I think, problematic when it is applied to progressive collapses like this. It implies a triggering event and a sudden charge of people, when in all likelihood people were not able to charge anywhere. Events such as this are often caused by large crowds of people attempting to occupy a space that cannot accommodate them. People continue to pour in from a distance, unaware that those ahead of them are dying. I think a statement that “the inciting cause remains speculative” assumes there actually was, or needed to be, an inciting cause. People walking calmly to a bridge that cannot accommodate them are not stampeding, nor are they incited by anything. There is a great deal of footage available showing how the progressive crowd collapse happened at The Love Parade in Duisberg, 2010. This was also called a stampede when it happened (as well as in this paper by Hsu), but clearly it is a matter of people being unable to choose their own body movements due to excess crowd pressure. Nobody suddenly began to charge in any direction, nor was there anything impulsive about it. There was no inciting cause. I have seen no evidence, nor does this paper offer any, that the Cambodian tragedy was any different, and I would discourage the use of the loaded word “stampede” unless that was the type of even that actually occurred.

    Calling something a stampede directs attention toward the participants (the victims) and away from those charged with protecting them with effective crowd management. For example, the people who burst through the door at the Wal Mart in Long Island did so because they were being pushed from behind by tremendous crowd pressure. The people up front had no choice. They did not break the doors down so the people behind them could get to the “door-buster” deals. How was that situation allowed to build when basic crowd safety principles in use at the time would have structured the crowd approaches very differently, preventing such crowd pressure from building up at all? Call it a stampede, and it becomes the result of an unruly crowd. Call it a crush and it becomes a crowd management issue, which is what it is.

    I also take issue with the description of NSRMP events (a non-category) as being particularly deadly. In order to find a deadlier NSRMP event, one needs to go back to 1896 for the Khodynka Tragedy, when 1,389 were killed during festivities celebrating the crowning of Czar Nicholas II in Moscow. Crowd disasters, including crushes, collapses, panics, and yes, stampedes, are far deadlier and far more frequent in religious settings. Both of the events shown at the top of this page as the first and second deadliest such events occurred at mass gathering events that were religious in nature. Since publication of this paper, the crowd disaster at the 2015 Hajj has eclipsed all three of these events. The Hajj disasters in 2015 (~2,300 dead) and 1990 (1,426 dead), the 2005 Al-Aaimmah bridge disaster (above), and the 1954 Kumbh Mela disaster (~800 dead) all eclipse this disaster. Along with the Khodynka Tragedy, they make up the 5 deadliest crowd disasters, excluding escape panics associated with fire and war. (Arguably the deadliest escape panic associated with a fire was also at a religious event, the Church of the Company fire in Santiago, Chile, in 1863.) Four of the five deadliest crowd disasters occurred during religious mass gathering events, a trend that is not changing. The history of disasters during the modern Hajj and during religious events in India are particularly notorious for their frequency and for the number of fatalities. The Cambodian tragedy is a very sad outlier that is instructive in how it happened, not because of the category of the event during which it happened.

    Understand the nature of mass gathering events in your area and be prepared, but please do not perpetuate the myth that crowd disasters are all stampedes caused by unruly participants. They are well studied and easily preventable disasters. Organizers of mass gathering events might look to the medical community for advice on crowd safety. Don’t forget to discuss crowd management, and establish roles and responsibilities for all participants beforehand. Make sure there is a system in place for recognizing an impending collapse or crush, and a means to respond very quickly. Emergency physicians prefer an empty emergency department, right?

Leave A Reply