Introduction
Terrorist attacks are a multimodal threat that can include the use of firearms, explosives, incendiary devices, and readily accessible weapons such as knives, blunt instruments, and even bare hands. While mass casualty events involving firearms, explosives, and incendiary devices tend to dominate public discourse and policy making, a growing body of evidence highlights the rising prevalence and mortality of melee terror attacks. Data show that each attack varied in the number of casualties, the target population, and the country in which it occurred. In the United States only, terrorist organizations elected to conduct attacks most commonly with incendiary-only weapons, followed by firearms and explosives.Reference Kapur, Hutson, Davis and Rice 1 The impact in terms of the rate of fatalities and injuries is also dependent on the chosen type of weapon. For example, chemical, biological, radiation, and nuclear weapons have the potential to inflict the greatest number of casualties.Reference Hamele, Poss and Sweney 2 Nevertheless, the rate of its implication is low. However, all the above-mentioned weapons require significant financial expenditures, lawful barriers to acquire, and barriers to carry.
Melee weapons are widely accessible due to their cheap cost and lack of required skills to use them effectively. While other weapons may be restricted in public spaces through legislation, it is not feasible to prevent attacks from pocket knives, fists, and knees.Reference Vernick 3 The kinetic strength of melee-related attacks may be lower compared with explosive and firearm weapons; however, melee-type injuries would still cause similar injury severity and in-hospital mortality, which may still present challenges to hospital resources and create an additional burden to the medical staff.Reference Hwang and Park 4
Data analysis of melee-type terrorist attacks has the potential to provide an understanding of injury patterns, geographical distribution, commonly used weapons, and casualty count per single attack. This data can serve as a valuable tool for informing hospital preparedness, guiding resource allocation, and improving emergency response strategies for victims of melee-related mass casualty events. In this study, we aimed to analyze the prevalence, injury burden, and mortality associated with melee-based terrorist attacks.
Methods
Study Design and Setting
We conducted a retrospective review of the Global Terrorism Database (GTD) from 1970 to 2020 to analyze the worldwide prevalence of melee-related attacks against civilians and non-militant groups. The GTD is an open-source collaborative publication produced by the National Consortium for the Study of Terrorism and Responses to Terrorism (START; College Park, Maryland, USA). 5 This database catalogs intentional, violent acts perpetrated by non-governmental actors. For an incident to be included in the GTD, it must meet two of the following three criteria: a) there must be evidence of an intention to coerce, intimidate, or convey a message to a larger audience besides the immediate victims, b) the act must aim to achieve a political, economic, religious, or social goal; and c) the action occur outside the context of legitimate warfare activities.
For this study, the GTD was searched for attacks using the search term “melee.” The GTD database defines melee as “a weapon targeting people rather than property that does not involve a projectile in which the user and target are in contact with it simultaneously.”
Melee incidents were identified using GTD, by filtering for weapon subtypes and exclusively identifying “melee” attacks. All cases were exported into an Excel sheet for further analysis (Microsoft Corporation; Redmond, Washington). Incidents were independently reviewed by two authors to determine if they were appropriate for inclusion in this study. If the two reviewers disagreed regarding whether a case should be included, they would revise each conflicting case to come to a consensus on whether it should be considered for further analysis or excluded from an analysis. Incidents were included in our study if the primary weapon type was categorized as a melee attack. Furthermore, all included incidents occurred worldwide. Cases were excluded if they lacked melee weapons, involved unknown weapons, if an attack was targeted against property, if melee related weapon was used secondary to other types of weapons, such as firearms or explosives, and if the incident resulted in no injuries or death. Initially, 3900 events were reviewed as melee attacks. After assessing them based on the exclusion criteria, 2395 cases were excluded. Therefore, 1505 events were considered applicable to this study and were analyzed. All incidents were approached as individual events, data from each case was extracted, and multiple outcome and predictor variables were analyzed. This study was conducted according to the STROBE guideline (Strengthening the Reporting of Observational Studies in Epidemiology).Reference Cuschieri 6
Participants
All patients in the GTD database who sustained an injury from melee-related attacks were included. These patients were from all geographical regions, of unknown ethnicity, age, and sex since this piece of data was not reported in the GTD database. Our sample size was determined based on the number of available cases that met our inclusion criteria in the GTD database.
Variables and Measurement
The primary outcome variables were the number of deaths per attack (defined by the variable “nkill”) and the number of injuries per attack (defined by the variable “nwound”). Both variables of interest were available for each attack instance. The main independent variables included geographic region based on the World Health Organization (WHO), year of attack (1970–2020), attack type (armed assault, assassination, etc.), target type (private citizens, government, business, etc.), and weapon type (knives/sharp objects, blunt objects, other melee weapons). To allow for statistically valid categories, “Abortion Related”, “Maritime”, “NGO”, “Telecommunication”, and “Violent Political Party” from the “target type” variable were combined into a category labeled “Other”. Secondary variables included annual mean death and injuries per region.
Statistical Analysis
Results for categorical data were summarized as counts and percentages of the overall population and stratified by WHO Region. Results for continuous variables were summarized as mean values and standard deviation of the overall population and stratified by WHO Region. To determine the relationship of the number of melee attacks over time, a trend analysis was conducted using the Jonckheere-Terpstra Test due to the non-parametric, continuous nature of the data.
To determine predictors associated with increased death (defined as the number of deaths per attack), a negative binomial regression analysis of the number of individuals killed per attack was conducted. For both the “nkill” and “nwound” variables, the variance is greater than the mean, justifying a negative binomial regression. Incidence Rate Ratios (IRRs) were calculated from the negative binomial regression results. IRRs for categorical predictors were based on comparison to pre-determined reference categories. Trends in number of deaths and number of individuals wounded in each period and stratified by WHO region were analyzed and visually represented as a heatmap. Data cleaning and analysis were conducted using SAS 9.4.
Results
A total of 1,505 melee-related terrorist attacks were identified in the GTD between 1970 and 2020. These incidents resulted in 5,110 fatalities and 5,528 injuries, highlighting the severe consequences of melee-based violence despite the low-cost and easily accessibility of such weapons (Table 1). On average, each attack resulted in 3.67 (SD ± 11.60) injuries and 3.40 (SD ± 24.43) deaths. Marked regional variation was observed in both the frequency and fatality of attacks. Europe accounted for the largest proportion of events, with 386 events (25.65%), reflecting the widespread use of melee weapon across that continent. However, these attacks in Europe tended to be less fatal when compared to most other regions, resulting in an average of 0.52 (SD ± 1.48) deaths per attack, and 3.21 (SD ± 8.28) injured, respectively. By contrast, African Region experienced fewer total incidents than European Region but had a disproportionately higher average number of deaths per attack. The African Region recorded the highest fatality rate, averaging 14.54 (SD ± 55.48) deaths and 6.07 (SD± 21.07) of wounded, respectively, establishing melee violence in African Region to have a significant, high-rate ratio of 18.33 (95% CI 10.18–33.01). The Western Pacific Region demonstrated the lowest overall burden with 5.85% of melee attacks, but substantial death rate with an average of 5.82 (SD ± 21.66) deaths per attack. These findings highlight the geographic disparities in both the incidence and fatality of melee violence. Knives and sharp objects were the predominant weapons used, responsible for 863 incidents (57% of all attacks) and had the highest IRR of death compared to when hands, feet, and fists were used (IRR = 10.32, 95% CI 6.07–17.54). Blunt objects ranked second, accounting for 279 events (18.5%). Together, these two categories comprised more than three-quarters of all recorded melee-based incidents, reinforcing the accessibility and effectiveness of these weapon types in terrorist-related violence. Other weapons, such as hands and feet, contributed a smaller but still notable proportion of events, 185 (12.6%). Most attacks were directed at private citizens (57.5%). Regression analysis evaluating the relationship between the number of injuries and the number of deaths per attack revealed a positive association. Specifically, for each additional injury sustained in an attack, the expected count of death increased on average by 3.1%. Trend analysis revealed a positive trend of melee-based attacks from 1970 to 2020 (Z-score: 6.23 P-value: <0.0001). This illustrates a significant, consistent upward pattern in the number of melee attacks over the time period analyzed. Results from the trend analysis were visualized in a linear plot and a fit curve (Figure 1).
Descriptive statistics of melee incidents globally and by WHO Region, N = 1,505
Table 1. Long description
The table consists of eight columns. The first column lists variables: number killed (mean and standard deviation), number wounded (mean and standard deviation), attack type, target type, and weapon type. The next seven columns present data for Global (N equals 1,505), Region of the Americas (N equals 118), European Region (N equals 386), African Region (N equals 262), Eastern Mediterranean Region (N equals 362), South-East Asian Region (N equals 289), and Western Pacific Region (N equals 88). For number killed, the global mean is 3.40 (SD 24.43), highest in the African Region at 14.54 (SD 55.48), lowest in the Eastern Mediterranean at 0.38 (SD 1.05). For number wounded, the global mean is 3.67 (SD 11.60), highest in the African Region at 6.07 (SD 21.07), lowest in the Eastern Mediterranean at 2.49 (SD 4.92). Attack types are listed with counts and percentages: armed assault is most common globally (47.91 percent), especially in the Western Pacific (62.50 percent) and African Region (56.11 percent). Hostage taking is highest in the African Region (32.45 percent) and Eastern Mediterranean (26.52 percent). Unarmed assault is prominent in the Americas (36.44 percent) and South-East Asian Region (38.06 percent). Target types show private citizens as the most frequent target globally (57.54 percent), peaking in the African Region (74.81 percent). Police are targeted in 11.30 percent of global incidents, highest in the Western Pacific (15.91 percent). Weapon types indicate knife or other sharp object is most common globally (58.63 percent), especially in the African Region (79.31 percent) and Western Pacific (73.86 percent). Blunt objects are used in 18.95 percent globally, highest in the Americas (27.35 percent). Hands, feet, and fists are used in 12.6 percent globally. Data for some categories are marked as N/A in certain regions.
Fit curve trend of count of melee attack incidents from 1970 to 2020.
Figure 1. Long description
The line graph plots year on the x axis, ranging from 1970 to 2020, and count of melee attacks on the y axis, ranging from 0 to 150. Each year is marked by a blue dot representing the annual count. A red regression line runs through the data, showing a gradual increase from near zero in the 1970s to over 100 by 2020. A shaded blue band around the regression line marks the 95 percent confidence limits, which widen after 2000, indicating greater variability. The legend at the bottom identifies the blue band as 95 percent confidence limits, the red line as regression, and blue dots as count.
Discussion
This study presents an extensive analysis of melee-related terrorist attacks, highlighting a trend of increasing death rate and frequency of attacks. From a disaster medicine and public health preparedness perspective, these findings underscore the evolving nature of intentional mass violence, particularly the growing use of low-cost, easily accessible weapons that require minimal planning and technical expertise. Because melee-based attacks are difficult to anticipate and often occur in densely populated civilian settings, they also require no funding, and they present unique challenges for prevention, emergency response, and health system preparedness.Reference Merin, Sonkin and Yitzhak 7
Our findings indicate that sharp weapons, particularly knives and other bladed instruments, were the most commonly used weapons and presumably resulted in thoracic, abdominal, and head injuries. These injury patterns have important implications for prehospital emergency medical services (EMS) and trauma systems, as rapid hemorrhage control, airway management, and expedited transport to definitive care remain critical determinants of survival.Reference Ferrada, Shafique and Brenner 8 Regional differences in death rates may reflect disparities in EMS response times, trauma system capacity, access to trauma centers, and availability of trained first responders.Reference Newgard, Fu and Bulger 9 Strengthening EMS infrastructure, expanding trauma system accessibility, and improving coordinated regional disaster response planning may therefore play a key role in reducing mortality associated with these events.
From a broader public health systems perspective, melee-based attacks represent a form of intentional injury that places substantial strain on emergency departments, surgical services, and intensive care resources, particularly during multi-casualty incidents due to significant acuity of these patients, often requiring blood transfusions and interventions. Because these events often occur without warning, hospitals must maintain sustained readiness through routine mass casualty incident (MCI) training, surge capacity planning, and interagency coordination exercises. Public preparedness initiatives, including community hemorrhage-control programs, public access bleeding-control kits, and rapid public notification systems, can improve early lifesaving interventions prior to EMS arrival and support a whole-of-community preparedness framework aimed at reducing morbidity and mortality from intentional mass violence events.Reference Ugelvik, Thomassen, Braut, Geisner, Sjøvold and Montán 10
More broadly, to move beyond descriptive reporting toward a meaningful contribution to disaster medicine, these findings should be interpreted within the context of evolving health system preparedness. The observed increase in attack frequency and lethality, supported by the Jonckheere–Terpstra trend analysis, underscores a growing need for adaptive surge-capacity planning within emergency departments and trauma systems. This includes ensuring rapid activation of mass casualty protocols, adequate blood product availability, operating room readiness, and critical care resource allocation. Given the unpredictable and low-resource nature of melee-based attacks, healthcare systems must adopt flexible and scalable preparedness models that can respond to sudden, high-acuity patient influxes without reliance on advanced warning. Integrating these epidemiologic trends into operational planning highlights the importance of coordinated regional response systems, continuous simulation training, and interdisciplinary collaboration to mitigate the impact of such events on patient outcomes and system performance.
Limitations
This study has several limitations. First, reliance on the GTD, although comprehensive, introduces the possibility of reporting bias and under documentation, particularly in low-income regions or areas with limited media coverage. The GTD is dependent on open-source intelligence, including media reports, which vary substantially across regions due to differences in surveillance infrastructure, and political climate. As a result, regions with limited media coverage or restricted information may have substantial underreporting of lower-severity events, while more severe or fatal incidents are more likely to be captured. This may artificially inflate observed case-fatality rates in underreported regions and exaggerate apparent regional disparities. Therefore, geographic comparisons in this study should be interpreted cautiously, as differences may reflect reporting density as much as true variation in attack characteristics or healthcare system performance. In addition, case identification may be influenced by factors such as differential media reporting, subjective classification of terrorism, and regional differences in the prevalence of certain attack types, which could introduce residual confounding. Finally, narrative data review may introduce selection bias, especially in ambiguous cases.
Conclusion
This study underscores several critical implications for trauma preparedness and public safety. Trauma systems worldwide must recognize that melee-based terrorism is not a low-risk or low-incidence phenomenon and should be incorporated into existing emergency preparedness and response frameworks. Future research should coordinate efforts on early detection, informed planning, and effective mitigation strategies for the rising threat of melee-based mass-casualty events.
Author contribution
AP, QK Tran, VS, and NA contributed to study design. VS, NA, BB, and MG contributed to data acquisition. IL and QKT conducted data analysis. VS, NA, BB, and MG conducted the original draft. AP, QK Tran, and VS contributed further to data interpretation. All authors contributed to revisions of the draft. AP supervised the research. All authors approved the final version of the manuscript. All authors contributed substantially to this research.
Funding statement
This is a non-funded study, with no compensation or honoraria for conducting the study.
Competing interests
The authors do not have a financial interest or relationship to disclose regarding this research project.
Ethical approval
The Global Terrorism Database (GTD) is an open-source, publicly accessible dataset compiled by the National Consortium for the Study of Terrorism and Responses to Terrorism (START; College Park, Maryland, USA). Because the GTD contains de-identified, publicly available data and does not involve human subjects, this study did not require institutional review board (IRB) approval.
Declaration of AI-assisted technologies
This manuscript represents the authors’ own original work, and no generative AI or AI-assisted technologies were used in the writing or preparation of the content.
