International terrorism, or aspects of it, have been argued to be a reaction to globalization and/or to be aided by many of its features 12. In the last twenty or more years, there has been a substantial focus on terrorism-related policies in many jurisdictions, particularly since the attacks of 11 September 2001 in the United States. This focus has included spending and legislation, and has included public health measures relating to bioterrorism protection 34. The focus is understandable, considering the political significance of attacks by non-state organisations, and the economic and psychological effects on the societies which may consider themselves attacked 567. However, it is important for policy makers to know of the opportunity costs of the response to international terrorism, relative to addressing other causes of premature death, and to better understand how differences in risk perception influence policy making. Therefore, we contrasted the mortality impacts of international terrorism with another major cause of preventable death – tobacco use 8 (which is also exacerbated by globalization 910). This work is part of a wider attempt to put international terrorism into a public health context 1112.

As part of a study to describe the epidemiology of international terrorism 11 we extracted data for 1994–2003 on international terrorist attacks involving any deaths among non-perpetrators from United States (US) Department of State reports. The definition of terrorism used by the Department is: 'Premeditated, politically motivated violence perpetrated against noncombatant targets by subnational groups or clandestine agents', with international terrorism meant as 'terrorism involving citizens or the territory of more than one country'. These data were supplemented with findings from more detailed published studies (see: 11). Countries included were 21 'established market economy' countries and 16 'former socialist economies of Europe' (as per the classification in an international mortality study) 13. These two groups of countries were selected because there was better quality data available for both terrorism and tobacco. From these data, an average annual mortality burden was calculated for each country.

Data on tobacco mortality was based on the updated estimates for the year 2000 by Peto et al 14. This method involves country-specific rates of lung cancer mortality together with corresponding rates from the American Cancer Society's Cancer Prevention Study II to derive 'smoking impact ratios' by age and sex. The burden includes tobacco-related: respiratory diseases, vascular diseases and other tobacco-related cancers. This methodology has been shown to be a robust indicator of the accumulated hazards of smoking 15.

Rates were calculated using the most recent population data for each country from the World Health Organization website http://www.who.int/country/en/.

For the selected countries collectively, the annual mortality burden from tobacco was approximately 5700 times that of the average annual mortality burden from international terrorism (Table 1). For 26 of the countries, there were no deaths from international terrorism. Within the other 11 countries, the ratio of annual tobacco to international terrorism deaths was lowest for the US at 1700 times, followed by Russia at 12,900 times.

The absolute annual burden from tobacco was highest for the US at 514,000 deaths per year in 2000 (Table 1). This is equivalent to the impact of an 11 September type terrorist attack every 2.1 days. For all of these 37 countries collectively, the tobacco mortality burden was equivalent to the impact of an 11 September type terrorist attack every 14 hours.

Definitions of terrorism are highly contended 161718. Furthermore, we have identified some limitations with the US State Department dataset, including with the definition used 11. Indeed, if a tighter definition of international terrorism was used, then this would substantially reduce the number of deaths categorised in this way (eg, relative to domestic terrorism or other types of homicide 11). Therefore this analysis may over-represent the mortality burden from international terrorism to some degree.

In contrast, the tobacco mortality estimates may be underestimates of the true mortality burden. This is because the estimates by Peto et al ignore all deaths in those aged under 35 years (including neonatal deaths and deaths from sudden infant death syndrome attributable to smoking), and the methodology was one of 'conservative underestimation of tobacco hazards' 19. More recent data also suggests that the long-term hazards of smoking on health are probably higher than previously thought 20. Nevertheless, methodologies for assessing the tobacco-related mortality burden differ and for the US a more recent analysis 21 indicates a lower mortality burden attributable to tobacco (ie, 438,000 versus the 514,000 calculated by Peto et al and used in this analysis).

Despite these various limitations, the findings of this analysis suggest that the mortality burden from tobacco use is at present vastly greater than from international terrorism in all the selected countries studied. This is even the case for the US, which has suffered the worst mortality burden from international terrorism out of these countries in the last decade.

Why does tobacco mortality not receive a proportionate response? Some may find comparisons between 'catastrophic' and 'normal' deaths misplaced 22. We recognise the subjectivity of risk perception 2324, and the tendencies of populations to: (i) overestimate risks stemming from visible, well publicised sudden violence with collective results, particularly where the cause is not well understood, compared to risks with results dispersed over place and time; and (ii) to overestimate risks from causes were there is little apparent control by the individual, compared to risks from causes which appear to many to be voluntarily undertaken 252627.

This tendency may be exacerbated by disproportionate media coverage of certain causes of mortality which involve low risk at the individual level 2829. There is also the political problem of giving priority to long-term issues, compared to dealing with emotive immediate concerns 3031. However, we have also demonstrated elsewhere that even for another cause of mortality which results in visible, well publicised sudden death (road crashes), policymaking does not appear to take into account the disproportionate mortality burden, compared to that from international terrorism 12.

International terrorism and the harm from tobacco use have similarities, in that they both involve discrete perpetrators – international terrorist groups and the globalized tobacco industry – against which governments can take action. Also, many tobacco deaths globally are due to the actions of foreigners – policymakers and company officials in tobacco manufacturing and exporting countries. Both international terrorism and tobacco use can substantially harm national economies and the international economic fabric in many ways 3233. Similarly, both can have widespread impacts on the way society functions and on its institutions eg, terrorism on security arrangements, and tobacco via the tobacco industry on the functioning of political processes 3435. The costs from both are largely or totally preventable, and investment in long-term prevention for both, as opposed to containment, may not necessarily be mutually exclusive (eg, if military budgets are diverted to terrorism prevention).

Despite these similarities, there are substantive differences. One is that the tobacco industry, unlike terrorists, is generally described as 'a legal industry' ie, an industry taking part in legal activity. This is despite the fact that the deliberate sale of a highly addictive, commonly lethal substance, and the routine denial of some harms (eg, of secondhand smoke) may be considered reckless criminal behaviour under the laws of some countries 36. This presumed 'legality' contributes to the societal acceptance and political strength of the tobacco industry in developed countries, relative to international terrorist groups.

Secondly, there is considerable evidence about the preventability of tobacco-related harm using current methods, and of their cost-effectiveness 37383940, compared to the high uncertainty about the effectiveness of particular measures to prevent international terrorism or its health impacts 4142. From a public health perspective, anti-terrorism efforts tend to focus on immediate containment, rather than addressing the possible root causes of terrorism 43444546. The cost-effectiveness of public health measures related to potential terrorism impacts has had little conclusive research 4748.

A further difference, as this analysis indicates, is the vastly different scale of the consequent mortality burdens. The policy implications of this include the relative extent, effectiveness and cost-effectiveness of the resources used to address the two problems 49505152. A public health and evidence-based approach may suggest a greater relative emphasis on tobacco control both nationally and internationally. While public health budgeting will always have to take into account public concerns that are not based on the evidence of relative risks, we argue that such policy moves should be as rigorously examined, as is the budgeting for tobacco control. A further possible implication is to learn from the response to international terrorism, so as to inform the way that tobacco marketing can be reframed as a serious threat to the social and economic well-being of individual countries and to international social and economic development.

This analysis suggests a very large mortality differential between these two problems exacerbated by globalization, international terrorism and tobacco use. Different perceptions of risk may contribute to the relative lack of a policy response to tobacco mortality, despite its greater scale. The lack of an appropriate response is also despite tobacco control having a stronger evidence base for the prevention measures used. National and international policy makers need to consider these issues if they are to make more rational use of resources to prevent premature mortality.

Both authors have undertaken contract work for tobacco control-related non-government agencies, and NW has undertaken contract work in tobacco control for the New Zealand Ministry of Health.

Both authors contributed to the design of the study, the data collection and the drafting and final write up of the manuscript. NW undertook the data analysis.