Health and Human Security: The Public Health Impact of Violent Conflict

Abstract

The consequences of violent conflict permeate countless aspects of society, and are not limited to the political and economic institutions of a state. The concept of human security extends traditional, state-centric notions of security to include the security and well-being of people that live within states. Adhering to the human security framework, I examine the effect of militarized conflict on the populations of states by evaluating the relationship between war and public health while taking into account relevant political and economic factors, including democracy and wealth. I argue that interstate and intrastate conflict negatively influences the health achievement of states and, therefore, the human security of their populations. I assess this relationship by analyzing data on summary measures of public health in all states between 1999 and 2001. My analysis suggests that the negative effect of war on health is particularly intense in the short term following the onset of a conflict.

Since the end of the Second World War, there have been numerous interstate and intrastate conflicts resulting in millions of deaths and billions of dollars worth of destruction. Yet scholars have focused very little attention on the consequences of conflict, in particular its social consequences. The World Health Organization's World Report on Violence and Health reveals that 1.6 million people die each year because of violence, including collective violence such as conflicts within or between states. A large number of the people who lose their lives because of militarized conflict are non-combatants. Furthermore, the 25 largest instances of conflict in the twentieth century led to the deaths of approximately 191 million people, and 60% of those deaths occurred among people who were not engaged in fighting (World Health Organization 2002). I assess the costs of conflict by looking at the relationship between conflict and public health.

War leads to direct casualties and deaths during combat; violent conflict also results in widespread death and disability among the civilian populations that get affected either as collateral damage or as deliberate targets. During the Second World War, Russia lost 10.1% of its population; Korea lost 10% of its population in the Korean War; and 13% of the Vietnamese population died in the Vietnam War (Garfield and Neugut 1997). In addition to direct deaths and injuries caused by combat among the military and civilian populations, conflict results in conditions that contribute to the spread of disease and retardation of health care systems. The influenza outbreak during the First World War killed more people than combat-related deaths in the war. States involved in violent conflict are unable to meet the public health needs of their societies as their health care infrastructure gets damaged or destroyed. Moreover, wars are associated with creation of suboptimal health conditions that result in hazards such as famine, epidemics, weapons-induced pollution, lack of clean water, poor sanitation, and general indigence. Consequently, the population is exposed to new health threats without access to proper health care.

Assessing the adverse effects of conflict on public health adds a new dimension to the study of conflict. It shifts the focus of analysis from the systemic or state level to the individual level (King and Martin 2001) as it ponders the influence of conflict on the health of individuals in a society. Studying the effect of war on public health extends the idea of security from the security of states to the security of populations and fits well within the human security framework. Recently, scholars and policymakers have begun to focus attention on the issue of conflict's effects on public health. King and Murray (2001) present a conceptualization of human security that takes into account health and economic factors rather than merely security from armed conflict. Their definition of human security focuses on the expected number of years of future life a person is likely to spend without falling “below the threshold of any key domain of human well-being” (585). This gives rise to the notion that public health is an important indicator of the level of human security enjoyed by the population of a state, and it can be demonstrated that public health is adversely affected by violent conflict. My first task in this paper, therefore, is to explain the concept of human security. I then discuss some linkages between war and health. Next, I outline the hypotheses that I test regarding the effect of violent conflict on levels of health outcome and follow with both a discussion of the measurement of public health and the political and economic influences on levels of health.

Health and Human Security

A number of varying and overlapping definitions of human security exist, but it is appropriate to state that this concept focuses on the well-being and welfare of people. There has been a growing awareness among policy-making and academic circles of evolving notions of human security that are more complex than traditional ideas of state security. Although state security is by no means irrelevant, it must be complemented by other elements, including human rights and public health, to constitute fully the security of people. This calls for a “paradigm of security” (Commission on Human Security 2003) that takes into account threats to the well-being of individuals and communities that may not necessarily threaten state power. This view of security goes beyond the state-centric approach of traditional paradigms, such as realism, and implies the inclusion of actors other than just the state. According to United Nations Secretary-General Kofi Annan, “[h]uman security, in its broadest sense, embraces far more than the absence of violent conflict. It encompasses human rights, good governance, access to education, and health care and ensuring that each individual has opportunities and choices to fulfill his or her potential” (Annan 2000).

The traditional approaches to the study of security focus undue attention on the state-level impact of conflict. The idea of security is generally considered synonymous with protecting the territory and national interests of a state from external aggression or unwelcome interference. Once a state is able to safeguard its military, territorial, and political interests from outside threats, it is perceived as having attained national security. During the Cold War era, realist notions of security dictated foreign policy, and state leaders remained unrelentingly occupied in the pursuit of military superiority. The emphasis of neo-realist theory on states as unitary actors and as the single most important entity in the international system led to the deprecation of people within states. Entities without sovereignty did not warrant attention at the international level, and what occurred within the borders of a sovereign state was to be addressed at the domestic level. Only threats to the security and existence of states were considered detrimental to global security and thus worthy of global action.

The emerging notion of human security, on the other hand, addresses the security of ordinary people. The factors that engender insecurity among the people living within states are not limited to the perpetuation of the state. The security of people is related to their quality of life and, therefore, the threats to their security include a number of social and economic issues. Elements of human security include economic security, political security, access to food and health care, personal and community security, and environmental security (UNDP 1994). The occurrence of violent international conflict can adversely affect any or all of these factors and amplify the insecurity of people in the affected state. However, the absence or cessation of militarized conflict does not guarantee the elimination of all threats to human security. In order to gain an adequate understanding of whether people—and not merely states—are secure, we must address the various components of human security rather than assume that populations are secure when states are not involved in wars. The shift from state security to human security is necessitated by the salience and the global nature of the issues that threaten the security of populations. Problems like environmental degradation and disease proliferation do not just threaten the security of people in a single state, but can easily reach global magnitude.

Although the literature on the concept of human security offers varying definitions of what constitutes human security, there is clear agreement that health is an important component. Human security entails people's ability to maintain a quality of life that does not fall below the level at which they feel secure. Adequate provision of public health is important in enabling people to achieve an acceptable quality of life and to be functional enough to maintain their lifestyle. I argue that human security is a better framework within which to assess global security than the traditional approach of focusing on the security of states as the best indicator of a stable world. State security is important in that people cannot be safe if the existence of the states in which they live is threatened. However, students of security must go beyond the state level to understand the true nature of people's security. Violent conflict is accepted as a major threat to the stability of states; it is also a formidable threat to the security of people. One way in which conflict decreases the security of people is by causing a decline in provision of public health. Since the health of a population is integral to the well-being of communities and individuals, studying the effect of conflict on health is a contribution to the understanding of human security.

War and Health

For every interstate or civil war, populations of states suffer short-term and long-term effects on their health and well-being. To understand the real cost of violent conflict, it is necessary to take into account the human cost of war. Violent conflict can have economic, social, political, and environmental consequences. A large number of conflict studies focus on causes of conflict, a small body of literature exists about the termination of conflict, and there is a developing literature about the various consequences of conflict. Yet the health consequences of war remain largely unexplored. The effects of conflict on a society continue long after the actual fighting has ceased, and understanding the social consequences of conflict is especially important if the goal is to mitigate the suffering that results from violence. Although scholars have examined some aspects of the economic and political consequences of conflict, far less work has been done on the manner in which conflict undermines public health.

Scholars have long been interested in the cost of waging war and the economics of defense spending (Russett 1969, 1970) and more recently there has been a growing literature on the consequences of conflict. The literature on the political consequences of war explores issues such as democratization, political institutions, leaders and regime type, international power, and human rights (Rasler and Thompson 1985; Bueno de Mesquita, Siverson, and Wollers 1992; Bueno de Mesquita and Siverson 1995, 1997). Work on termination of war addresses issues regarding settlements following conflict, reconstruction, and peacekeeping (Werner 1998, 1999; Walter 1999) as well as the fate of leaders after a conflict (Goemans 2000). The economic effects of conflict have also been examined (Organski and Kugler 1980; Rasler and Thompson 1983, 1985; Collier 1999).

In addition to the political, economic, and environmental consequences of conflict, attempts have been made to study its social consequences. Ghobarah, Huth, and Russett (2003) conduct a cross-national analysis using political and economic variables to assess the relationship between conflict and public health, emphasizing the role of public health in human security. They find that the burden of death and disability in 1999, from the effects of conflicts between 1991 and 1997, was nearly the same as the direct fatal and nonfatal health effects that occurred immediately from all the violent conflicts in 1999. Similarly, Davis and Kuritsky (2002) find that sub-Saharan countries involved in military conflict between 1968 and 1999 experienced lower life expectancies and higher infant mortality rates than sub-Saharan countries not involved in conflict during that period. Ghobarah, Huth, and Russett (2004) extend this nascent literature on war and health by examining the determinants of public expenditures on health and some influences on the health performance of states.

Violent conflict may affect public health through damage to the infrastructure of the society, interruptions in access to basic services such as water and transportation, and lack of availability of health care personnel. In addition to the direct effects of conflict on health, economic resources get diverted from public health to military uses, and the environmental damage resulting from conflict gives rise to disease. Moreover, disruptions in agricultural production can lead to widespread famine, which in turn may trigger disease proliferation or make it difficult for people to recoup from existing health conditions. Thus, strong linkages exist among the more obvious consequences of war—such as the toll on the environmental and economic well-being of a society—and the health consequences of violent conflict. Conflict also generates refugee flows; the living conditions of refugees are generally not conducive to good health, and refugees can transmit disease across borders (Ghobarah et al. 2003). Studying the relationship between conflict and health is particularly important in light of the nature of conflicts in the current international system. Most of the recent and ongoing conflicts in the world are civil or intrastate wars that lead to large-scale devastation of a state's infrastructure since all the fighting occurs on the territory of one state. This amplifies the conditions that deteriorate the health of societies. Civil conflict is also highly likely to result in internal displacement or refugee flows, exposing communities to health menaces.

The crisis in Liberia effectively demonstrates the suffering civil war can inflict on a population. Cholera, among other diseases, has been spreading at alarming rates as internally displaced people with the disease leave Monrovia for camps outside the city. The civil war makes it impossible for either Liberian authorities or international agencies to carry out the extensive process of water chlorination that would halt further spread of cholera. Furthermore, afflicted people are unable to access medical facilities because of the security situation. In September 2003, the World Health Organization reported that only 32% of the Liberian population had access to clean water, no more than 30% of the population had access to latrines, and there had been no regular garbage collection in Monrovia since 1996. The SKD Stadium, the largest camp for internally displaced people in Monrovia, housed about 45,000 people who “cook and sleep in any sheltered spot they can find, in hallways and in tiny slots under the stadium seats,” and there were six nurses in the health center for 400 daily patients (WHO 2003).

In the Sudan, two decades of conflict have exposed the population to diseases (such as yellow fever), malnutrition, displacement of large groups, poverty, and famine. The Iraqi population experienced near destruction of their health care system, previously one of the best in the Middle East, during the first Gulf War. Public health in Iraq continued on a path of steady decline during a decade of sanctions and internal repression, before the general and health infrastructures were subjected to a second war. In 1993, Iraq's water supply was estimated at 50% of pre-war levels (Hoskins 1997), and war-related post-war civilian deaths numbered about 100,000 (Garfield and Neugut 1997).

Interstate and intrastate conflict affect the public health of a population at several levels. The most observable, direct effect of conflict is the number of people killed and wounded. Conflict also affects public health through exposure of populations to hazardous conditions caused by events such as refugee flows and movement of soldiers, giving rise to epidemics as the mobile groups act as vectors for disease. The ability of war-torn societies to deal with new threats to public health is weakened by conflict and, consequently, the negative effect on health outcomes continues to grow. During periods of conflict, resources get diverted toward military purposes, and public health spending goes down, even as the public health needs of the population escalate.¹ This diversion of resources away from public health is often accompanied by a decline in infrastructure that further impedes the ability of a society to handle public health issues. Damage to the general infrastructure combines with damage to or possible destruction of the health care infrastructure to make a society incapable of facing its increasing public health challenges. Public health is affected not only by direct damage to hospitals and other medical facilities, but also by damage to elements of the larger infrastructure such as transportation, the water supply, and power grids. In addition to infrastructural damage, food shortages—and possibly famines—often ensue during and immediately after wars. At a more indirect level, conflict may lead to a general social and economic decline that has dire long-term consequences for the health of a society. In addition to bringing about economic weakness and poverty, conflict may cripple important elements of civil society, such as law and order. A broad spectrum of conflict's social consequences can be linked to decline of public health.

I examine the effect of militarized conflict on the populations of states by evaluating the relationship between war and public health, taking into account relevant political and economic factors, such as democracy and wealth. I assess the effect of war on overall levels of health achievement in states, as indicated by a summary measure of public health. To examine the relationship between violent conflict and health, I test the following hypotheses: (a) the overall level of public health will decline if a state is involved in conflict, and the intensity of the conflict is positively related to the decline in health outcomes; (b) the decline in health achievement is likely to be greater in the short term after the war, but overall levels of health outputs continue to be affected in the long run; (c) states with higher levels of wealth and economic openness enjoy higher levels of health outputs; and (d) higher levels of democracy are associated with higher levels of health outcomes.

I argue that interstate and intrastate conflicts negatively influence the health achievement of states and, therefore, the well-being of their populations. I assess the magnitude of this relationship by analyzing data on summary measures of public health in all states between 1999 and 2001 in light of relevant political and economic factors, including wealth, trade, and regime type. The findings suggest that the negative effect of war on health is significant immediately after a conflict and is also linked to levels of income, trade, and democracy.

Measuring Public Health

Most studies of population health use summary measures of public health that combine information on mortality and non-fatal health outcomes to express the health of a population as a single number, including inputs such as age-specific mortality and the epidemiology of non-fatal health outcomes. Such summary measures are useful for comparing the health of populations across states and time. One of the summary measures of public health used in recent literature is Health-Adjusted Life Expectancy (HALE), provided by the World Health Organization. HALE was previously referred to as Disability-Adjusted Life Expectancy (DALE). The HALE measure subtracts the number of years an individual is expected to spend with a disability as a burden of disease from the total life expectancy at birth. Another measure, derived from HALE, aggregates data on health outcomes by gender and five age groups and is referred to as Disability-Adjusted Life Years (DALY). Ghobarah et al. (2003) use this measure as the dependent variable in their study of conflict and health. Although DALYs provide information about life expectancy in certain population groups and deaths resulting from particular diseases, this measure is only available for 1999. I use HALE as the summary measure of public health in my analyses due to its availability for a longer period of time. Moreover, for the purposes of this analysis, since it aims to gauge the impact of war on the overall levels of health, the population and disease groupings are not necessary. Thus, the added advantage of data availability for multiple years is more important. Ghobarah et al. (2004) also use HALE as their dependent variable; however, they only use HALE data for the year 2000.

The HALE measure is estimated by the Burden of Disease project of the World Health Organization, using a methodology that links incidence and causes of disease to information on “both short-term and long-term health outcomes, including impairments, functional limitations (disability), restrictions in participation in usual roles (handicap), and death” (Mathers et al. 2000). The measure expresses the number of years that individuals in a population are expected to live in “full health,” excluding time spent in states lower than ideal health, as described by an accepted norm for a population. One advantage of using HALE as an expression of levels of health is the ease with which the notion of “healthy life” can be comprehended by audiences who are not well-versed in more technical health measures. HALE is an accessible measure because health is expressed as a single number, made even more intuitive by the use of years of life as the unit of expression. The accessibility of this measure to a wide variety of disciplines as well as policy makers makes it highly attractive and valuable. Mathers et al. (2001) describe HALE as the best Summary Measure of Public Health (SMPH) reflecting the overall level of health in a population. Appendix A contains an excerpt from the WHO World Health Report 2002 that explains the logic of the HALE measure as well as the data-collection and computation methods involved in deriving this measure.

To test the hypotheses posited in this study, I examine aggregate public health levels in all member states of the World Health Organization from 1999 to 2001. To assess the effect of war on health, I also analyze change in HALE by comparing the figures for the current and previous years.

Influences on Health Outcomes

Violent conflict may affect public health through a host of means, including damage to the infrastructure of the society, interruptions in access to basic services such as water and transportation, lack of availability of health care personnel, and diversion of economic resources from public health to military uses. Other indirect effects of conflict on health include food shortages, famine, and economic decline.

One important factor in determining health achievement is the nature of political institutions that govern a state. Democratic states are more likely to be sensitive to the health care needs of their populations since the population possesses channels to voice their grievances, as well as the ability not to reelect the incumbent leadership. The extent to which health achievement declines in the wake of conflict, however, is also related to the level of wealth in the affected society. Broadly speaking, wealthier societies are better able to maintain higher levels of public health. I argue that the higher the level of a state's income, the higher the overall level of health. Similarly, since trade serves as a means to generate revenue, states with high levels of trade will also experience better health care in general as well as during times of conflict. This suggests a positive relationship between economic openness and provision of public health.

I hypothesize, first, that involvement in conflict lowers the overall levels of public health in states. Second, I expect a positive relationship between democracy and levels of health outputs. Third, I argue that populations of states with higher levels of income and trade openness enjoy better health provisions. I also control for population size in my models since a large population can put constraints on a state's public health capacity.

Conflict

Involvement in violent conflict negatively affects the health achievement of a state because of the direct effects of war in terms of military and civilian casualties and the destruction of infrastructure and medical facilities. The magnitude of the health impact of wars varies immensely, but it is rare for a war—interstate or intrastate—to incur no costs in terms of human well-being. According to the U.K. Department for International Development, the war in Bosnia caused serious repercussions for the Bosnian population:

The civil war was disastrous, the toll on the country and its citizens a catalogue of misery. It left an estimated 250,000 people dead, 240,000 wounded, and 25,000 permanently disabled. Some 50,000 children were wounded. About 50,000 people still require rehabilitation and 15 percent of the population suffer from post traumatic stress disorder. There are estimated 800,000 externally displaced people still refugees abroad. In addition there are about 1 million internally displaced people, living within BH, but not in their home communities (UK Department of International Development 2003).

The Liberian crisis has caused the spread of diseases such as cholera and diarrhea because of lack of clean drinking water and widespread malnutrition. Although some of the hospitals continue to function in Monrovia, the chaotic conditions of the city make it impossible for most people to reach the hospitals. Moreover, medical supplies are scarce because of transportation issues; “Due to the ongoing fighting in Monrovia, vaccines estimated at USD 250,000 stored in National Drugs Services (NDS) are at risk to be lost for lack of fuel. The port where fuel is available is controlled by the LURD group and not accessible from Central Monrovia” (World Health Organization 2003).

The first independent variable in this study is conflict. Health achievement is likely to decline during and after a conflict because of the direct and indirect ways in which violent conflict affects health of populations. Ghobarah et al. (2003, 2004) operationalize civil wars as the number of civil war deaths, which raises the issue of endogeneity since the HALE measure incorporates life expectancy and deaths. My conflict variables, on the other hand, measure both the presence of conflict at various levels of intensity and the frequency of conflict. Moreover, I include both civil and interstate conflict. To assess the presence of conflict, I consult data from the Peace Research Institute of Oslo (PRIO) Dataset on Armed Conflict (Gleditsch et al. 2002). These data measure conflict according to both its intensity and its type, including domestic and international conflict. Following the PRIO guidelines, I categorize conflict as “major war” and “intermediate war.” Major war refers to militarized conflicts that result in at least one thousand battle deaths in a given year; intermediate war refers to conflicts in which there have been at least 25 deaths in a given year and more than 1,000 deaths in the history of the conflict. During the three-year period under consideration, there were a total of 80 armed conflicts, 36 of which were major wars with at least 1,000 battle deaths in a given year. Out of these conflicts, 63 were intrastate. I also include a variable for the number of conflicts in which a state was involved in a given year. The conflict variables have been lagged to account for temporal effects. I assess the effects of the intensity and the number of conflicts separately. It should be noted that most of the conflicts during the period examined were internal conflicts or civil wars. In general, I expect violent conflict to have a negative effect on public health, as measured by HALE.

Democracy

I expect to find higher levels of health achievement in states that possess highly democratic regimes. It is widely accepted that democratic regimes are more responsive to the needs of their populations, including those that are health related. The perpetuation of the leadership is closely related to the well-being of the public in democratic systems, which suggests that democratic governments would be more likely to allocate greater resources to the public health infrastructure. Ghobarah et al. (2004) find that democracies allocate more public spending to health than non-democracies; they do not, however, evaluate the direct effect of democracy on health. Interestingly, many highly autocratic societies enjoy levels of public health that are very close to industrialized democracies of Western Europe and North America. During the three-year period studied, Cuba, Qatar, and Saudi Arabia had average health achievement scores of 67.2, 61.97, and 61.43, respectively. These states supported some of the most autocratic regimes in the international system, according to the POLITY IV democracy scores for these years. The health achievement levels in Norway, the United Kingdom, and the United States were 71.07, 70.17, and 68.33, respectively. Unlike the first set of countries, these states maintained the highest possible democracy score (10 on the POLITY scale). The differences in the health outcomes of these groups of highly autocratic and highly democratic states are by no means as great as one would expect if a linear relationship existed between democracy and health outcomes. On the other hand, public health scores in Zambia, Liberia, and the Democratic Republic of the Congo—as measured by HALE—range from 30 to 37, significantly lower than the autocratic and the democratic regimes mentioned earlier. The POLITY scores of this last group of countries were between −1 and 1 during the years being examined. These states, therefore, were neither completely autocratic nor highly democratic.

This hints at a curvilinear relationship between the regime type of a state and the level of public health. But I argue that such a curvilinear effect would exist only in the absence of the variable for wealth. The highly autocratic states that maintain high levels of health outputs also enjoy per capita Gross Domestic Products (GDPs) close to the wealthiest western democracies. Accordingly, I suggest that it is proliferation of wealth in society rather than the policies of the regime that accounts for better public health. Democracies, on the other hand, are conscientious about their public health provisions because of the norms and values embedded in their institutional structures and the ability of the population to oust the incumbent regime in the next election. Governments often have to make difficult decisions regarding the allocation of scarce resources among competing uses, such as expenditures on public health (Russett 1969; Mintz 1989), and conflict further complicates these decisions by increasing the security needs of the state. The governmental systems in democratic states are more likely to prioritize in favor of health expenditures than those of autocratic regimes. The latter are not accountable to their populations for their resource allocation decisions and are more likely to divert resources toward military expenditures to enhance their power. Therefore, I argue that levels of democracy are positively related to public health.

I hypothesize that the level of democracy in a state will have a positive effect on the well-being of a society. States that are highly democratic experience higher levels of health achievement than those with autocratic regimes or lower levels of democracy. To measure the level of democracy, I use the POLITY IV scores. These scores range from −10 to 10, with higher scores denoting higher levels of democracy. Thus, a state with a score of 10 is considered completely democratic, and one with a score of −10 is an autocracy. A squared term for POLITY scores is also included to test for a possible curvilinear relationship between the covariates and the dependent variable.

Economic Factors

It is generally accepted that poverty and disease are closely related and that poorer states and communities do not enjoy the same levels of health as richer states. Inequality in health achievement across states is a major concern for scholars of development (Leon and Walt 2001; Sen 2001; Zwi 2001). At higher levels of wealth, states are able to provide better preventive and curative health care. Leon and Walt (2001), however, warn against ignoring the effect of higher levels of health on the economic well-being of states: “In much of what is written about the link between health and wealth, it is often implicitly assumed that the direction of causality is from wealth (or poverty) to health (or disease). However, the possibility that either at the individual or population level, there can be a causal link running from health to wealth needs to be considered” (5). A healthier population is more productive and contributes to economic activity more than an ailing population. Nonetheless, economic resources are necessary for the provision of health care and the effect of wealth on health is potent in the short run and in the long term. I argue that levels of wealth are positively related to health achievement, with the causal arrow running from wealth to health.

Trade openness is associated with economic prosperity, and it may be argued that states with higher levels of trade are likely to enjoy higher standards of health achievement. In the current international atmosphere of globalization and interdependence, open economies fare better than closed economies and are able to provide higher standards of public services, including health care. Trade acts as an economic stimulus and has a similar effect on health outputs as GDP. I expect trade openness to have a positive effect on health achievement.

The covariates in this analysis include wealth, trade openness, and population. Data on wealth and economic openness were obtained from the Penn World Table (Heston, Summers, and Aten 2002). The income or wealth of states is measured by per capita Gross National Product expressed in constant prices and in U.S. dollars. Trade openness is measured by the total volume of exports and imports as a proportion of GDP. The larger this percentage, the more open the economy since a larger proportion of its income is being generated through trade with other states. Population statistics come from the World Bank's World Development Index and are expressed in millions. I also control for education; the level of education of a population is measured by an education index used in the United Nations Human Development Index. This education index is based on the adult literacy rate and the combined gross enrollment ratio for primary, secondary, and tertiary schools (United Nations Development Program 2004).²

I argue that both wealth and openness have a positive effect on public health, and I include a variable that measures the interaction effect between wealth and openness to evaluate whether the influence of openness is different in wealthy versus poor states. Taking into account the wealth—or poverty—of states when examining the effect of economic openness on health will address speculation about the benefits of globalization for lesser developed states. Education is expected to be positively related to health since educated populations are better capable of making decisions, such as taking advantage of public health programs, that would contribute to higher health levels (Ghobarah et al. 2004). Summary statistics for these variables are shown in  Appendices B and C.³

Analysis and Results

The fact that these data comprise repeated observations on a large number of cross-sectional units raises the specter of both temporal correlation and non-constant error variance. To address these issues, I adopt the method of generalized estimating equations (GEE) (Liang and Zeger 1986; Zorn 2001). This approach has been shown to provide consistent and asymptotically efficient estimates of the parameters of interest. In particular, I estimate a GEE model with a first-order autoregressive temporal covariance structure.

All independent variables (except education) have been lagged to take into account the delayed effect of the levels of covariates on HALE. As levels of factors such as GDP and democracy increase or decrease for a given country, effects on health outcomes as measured by HALE take some time to materialize. Lagged independent variables capture temporal effects and, therefore, more adequately reflect the relationship between the covariates and the dependent variable. Moreover, the variables of national income, trade, and population have been logged due to diminishing marginal returns. For instance, if a country's GDP per capita increases from $100 to $200, the marginal effect on public health would be much greater than if a country's GDP per capita increases from $3,000 to $3,100. Taking the log of the figures for these variables takes into account this issue of diminishing marginal returns and offers more interpretable results.

I estimate two models to test my hypotheses. The first model measures the effect of the independent variables on the overall level of public health as expressed by HALE for a given year (Y_it=X_itβ+u_it). The second model uses the change in public health, taking the difference between the HALE for the current year and the previous year (ΔY_it=ΔX_itβ+ɛ_it). This model assesses the immediate to short-term effect of conflict on health. In order to evaluate adequately the relationship between war and health, it is important to examine both the level of health outputs and the change in health outputs. The effect on the level of health in the first model is a better indicator of how the state would maintain its public health in the long run after the conflict; the change in health achievement due to conflict indicates the direct effect of conflict on health in the year following the violence. For each model, I present separate results for the intensity and the count of conflicts. The variables for the intensity of conflict and the number of conflicts in which a state was involved during the year are not included in the same equation because of multicollinearity.⁴

Levels of Health Achievement

The effects of conflict and the other covariates on the level of health achievement are shown in Table 1. This model assesses the overall levels of HALE and is a reflection on health output trends in the long run. I find in this model that the overall level of health declines because of involvement in war, but only the coefficient for the number of conflicts is significant. Thus, although the mere presence of conflict may not significantly affect overall long-term levels of public health, as the number of conflicts in which states are involved increases, the level of health decreases.⁵ This finding, illustrated in Figure 1, reveals that every additional (lagged) conflict takes about seven months off a nation's HALE. This is consistent with the earlier discussion about the impact of violence on public health through factors such as destruction of infrastructure, generation of refugee flows, and diversion of resources away from health expenditures.⁶ These factors reduce the ability of a state to provide public health at a time when the health needs of the population have escalated in the aftermath of militarized conflict. Thus, violent conflict does not just impact the populations of states through creating casualties of war, but also by reducing the capability of states to provide adequate health care and public health.

1

Determinants of Public Health: Levels and Long-Term Effects, 1999–2001

Variables	Intensity	Count
(Constant)	−80.30	−73.77
	(21.68)	(21.44)
Lagged Conflict (major war)	−0.56	—
	(1.11)
Lagged Conflict (intermediate war)	0.58	—
	(0.81)
Lagged Conflict (count)	—	−0.59*
		(0.29)
Lagged POLITY score	0.18	0.19
	(0.12)	(0.12)
Lagged POLITY score squared	−0.01	−0.01
	(0.02)	(0.02)
Lagged ln(Per Capita GDP)	12.00**	11.32**
	(2.79)	(2.76)
Lagged ln(Trade Openness)	13.92**	12.37*
	(4.80)	(4.75)
Lagged ln(GDP) × Lagged ln(Openness)	−1.31**	−1.16**
	(0.55)	(0.55)
Lagged ln(Population)	1.13**	1.19**
	(0.42)	(0.42)
Education	15.91**	15.83**
	(3.90)	(3.90)
Wald test	885.95	857.76
(p-value)	(<0.001)	(<0.001)
N	367	367

Note: Cell entries are coefficient estimates; numbers in parentheses are robust standard errors, clustered by nation.

*

p<.05;

**

p<.01 (one-tailed).

GDP, Gross Domestic Product.

1

The Effect of Conflict on Levels of HALE

Contrary to expectation, the level of democracy does not have a significantly positive effect on the long-term level of health. As expected, however, there is no evidence of a curvilinear relationship between democracy and health that would suggest that autocracies also enjoy higher levels of health outcomes. It is interesting to note, however, that in a bivariate analysis of the relationship between democracy and health, a curvilinear relationship does exist, but this effect disappears once the model controls for wealth. This explains why some highly repressive but wealthy states, such as Qatar, maintain high levels of public health. The positive effect of democracy on health is also insignificant when looking at the change in HALE.⁷ Population size is positively related to health outcomes. The education level of a population has a highly positive effect on health achievement, which supports the expectation that educated populations are better able to maintain higher levels of health due to the ability to make informed choices about health care.

As expected, wealth has a positive effect on levels of health; the higher the GDP of a state, the higher will be the level of its public health. Clearly, wealthier states have the resources to perpetuate better public health systems and maintain higher levels of health outputs. A similar effect exists for trade openness. Open economies have an additional source of income through trade and thus more means for better health provisions. However, when the effects of wealth and openness are interacted, there is a positive effect of openness on health only in poor states; there is no real effect on health in wealthy states: {Y_it=…+16.7(GDP)+19.8(Openness)−1.9[GDP×Openness]} represents the interacted effect of wealth and economic openness and reveals that wealth is positively related to health achievement. However, openness increases health only in poor states; the positive effects of openness are diminished to zero in open and/or wealthy states. This suggests that overall levels of health outcomes over the long run benefit more significantly from trade openness in poor states than in wealthy states.

Changes in Health Achievement

The second model assesses the changes in the independent variables and HALE, thus giving an indication of the short-term relationship between the explanatory factors and health achievement. The changes in HALE are calculated by taking the difference in the values of HALE for the current and the previous year. Similar calculations are made on the covariates to measure yearly changes. As in the previous model, which assesses the levels of HALE, the conflict variable is measured along the dimensions of intensity and count.

The results for this model are shown in Table 2, revealing a significant decline in health because of war. This suggests that when a country is involved in a war, its health achievement is likely to go down in the next year or the short term. This is particularly significant when states are involved in higher intensity conflicts. Clearly, all the direct effects of war—including combat casualties—are going to take their toll on a society within the short term. The immediate devastation caused by war cannot be compensated in the short term through international assistance and post-conflict reconstruction. In the long run, states recover from the impact of conflict through internal efforts, economic improvement, and external help from other states as well as international organizations. Although non-governmental organizations—such as Doctors Without Borders (MSF)—have been commendable in entering conflict-ridden areas to provide essential medical care (Taipale 2002), there can be a considerable lag between the onset of conflict and an inflow of international humanitarian and medical assistance. Intervention by the United Nations and other international organizations has been instrumental in post-conflict social and political recovery in states such as Mozambique (see Russett and Oneal 2001). However, international intervention and peacekeeping often occurs after ceasefires or peace agreements have been established and the warring parties are amenable to an international presence (Goulding 1993). International intervention can help ameliorate the health impact of violent conflict, but such intervention is unlikely to be effective immediately following a conflict. Consequently, the effects of war on population health are felt most drastically in the short term after conflict begins. Not surprisingly, levels of democracy do not have a significant effect on changes in health outcomes immediately after a war.

2

Determinants of Public Health: Changes and Short-term Effects, 1999–2001

Variables	Intensity	Count
(Constant)	−1.71	−1.74
	(0.17)	(0.18)
Lagged Conflict (major war)	−1.03**	—
	(0.27)
Lagged Conflict (intermediate war)	−0.10	—
	(0.47)
Lagged Conflict (count)	—	−0.15
		(0.11)
Lagged ΔPOLITY score	0.09	0.12*
	(0.09)	(0.09)
Lagged ΔPOLITY score squared	0.00	−0.00
	(0.01)	(0.01)
Lagged Δln(Per Capita GDP)	7.32**	7.72**
	(1.78)	(1.74)
Lagged Δln(Trade Openness)	7.41**	7.51**
	(1.30)	(1.31)
Lagged Δln(GDP) × Lagged Δln(Openness)	4.51**	5.21**
	(1.64)	(2.0)
Lagged Δln(Population)	56.66	57.70
	(9.01)	(8.73)
Wald test	92.41	73.52
(p-value)	(<0.001)	(<0.001)
N	234	234

Note: Cell entries are coefficient estimates; numbers in parentheses are robust standard errors, clustered by nation.

*

p<.05;

**

p<.01 (one-tailed).

GDP, Gross Domestic Product.

As in the model looking at levels of health, both national income and economic openness are positively related to health outcomes. However, when interacted with national income, trade openness has different effects for wealthy and poor states. In the model evaluating overall levels of health outputs, both wealth and openness were positively related to public health, but the positive effect of openness on health was limited to poor states. Openness was not positively related to health in wealthier states, suggesting that higher income levels, even in the absence of economic openness, lead to better health achievement. In the short-term model, however, these results are reversed. In states with higher levels of income, there is a much greater positive relationship between openness and short-term changes in public health than in states with lower levels of wealth. This implies that free trade does not improve the standard of public health in developing states in the short term to the same extent as in richer countries. A move toward economic openness may be related to an increase in health outputs within the immediate to short term in states that enjoy higher levels of income. A similar shift toward openness does not bring about the same level of improvement in the health outcomes of states with lower income levels.

The difference in the short-term effects of openness on health for rich and poor states is consistent with some of the arguments in development economics about the short-term benefits of maintaining closed economies for lesser-developed countries (Balassa 1971). The decades of 1950s, 1960s, and 1970s witnessed strong advocacy for protection as a viable approach for developing economies (Edwards 1993). Rodrik (1998) poses the question, “can nations pursue independent monetary and fiscal policies in the presence of high levels of capital mobility?” (5). Increased dependence on trade for revenue-generation may not allow poorer states to take necessary actions to increase public health outputs in the short run, especially during times of conflict. Consequently, the positive effect of openness is significantly lower in poorer states. In the long term, however, trade leads to a stable increase in income that enables poorer states to increase their levels of health achievement.

Education is positively related to health even in the short term. My results, therefore, demonstrate that the negative effects of conflict on public health can best be explained by taking into account economic and political influences as well as characteristics of states. Levels of wealth, education, and population are positively related to health achievement. Trade openness generally has a positive effect on public health, but to a much greater extent in wealthier states than in poorer countries.

Conclusion

The findings of the analyses in this paper reflect that conflict can undermine health, demonstrate that a number of political and economic factors are important in assessing the effect of conflict on health, and show how the influence of these factors varies across countries with differing characteristics. For instance, the health outcomes of states vary at different levels of wealth, trade, and democracy; the effect of economic openness on health achievement is different for wealthier and poorer states, and these effects further vary in the short term and the long run. Taking temporal dynamics and state characteristics into account contributes to a nuanced understanding of the determinants of health outcomes, reflecting the need for different approaches to health challenges in different parts of the world.

Work on consequences of conflict has mainly focused on states and institutions, including the political, economic, and environmental effects of violence. However, it is more important to study the effect of conflict on the quality of life of states' populations. One of the most significant factors in one's personal well-being is health; similarly, public health reflects the quality of life and well-being of a society. The real costs of violent conflict cannot completely be understood without a clearer comprehension of the mechanisms through which war affects the individuals in a society. There are always normative concerns surrounding war and the use of force, and rhetoric based on these concerns is often used in decisions to avoid violent conflict or to intervene in wars. Empirical evaluation of the effect of war on public health allows us to assess systematically the negative influences of war and enables us to support our normative concerns with empirical evidence. In illuminating the human cost of war, this line of research offers significant implications for the expected utility of violent conflict and provides incentives for conflict prevention. Policy makers invariably take into account the monetary costs of war before becoming involved in a conflict; understanding how state populations get affected by war gives them the opportunity to factor in the cost of war in terms of human security in addition to economic resources and national security.

However, for a more complete understanding of the effect of conflict on health, a time period longer than three years must be studied. Unfortunately, because of the complex nature of the data required for calculating HALE, this measure is only available from 1999 to 2001. The HALE was chosen as the dependent variable for this study as it is the best measure that summarizes health as a single number for each country and year. Analysis of a longer period of time could evaluate aspects of the relationship between war and health that have not been addressed here due to limitations in data availability, including examination of particular causal mechanisms through which violent conflict undermines public health.

Appendices

Appendix A World Health Organization—The World Health Report 2002: Statistical Annex Explanatory Notes

Annex Table 4 reports the average level of population health for WHO Member States in terms of healthy life expectancy. On the basis of more than 15 years of work, WHO introduced disability-adjusted life expectancy (DALE) as a summary measure of the level of health attained by populations in The World Health Report 2000. To better reflect the inclusion of all states of health in the calculation of healthy life expectancy, the name of the indicator used to measure healthy life expectancy has been changed from disability-adjusted life expectancy (DALE) to health-adjusted life expectancy (HALE). HALE is based on life expectancy at birth but includes an adjustment for time spent in poor health. It is most easily understood as the equivalent number of years in full health that a newborn can expect to live based on current rates of ill-health and mortality.

The measurement of time spent in poor health is based on combining condition-specific estimates from the Global Burden of Disease 2000 study with estimates of the prevalence of different health states by age and sex derived from health surveys. As noted above, for this year's World Health Report, burden of disease estimates of prevalences for specific diseases, injuries, and their sequelae have been updated for many of the cause categories included in the Global Burden of Disease (GBD) 2000 study. Analyses of over 50 national health surveys for the calculation of healthy life expectancy in The World Health Report 2000 identified severe limitations in the comparability of self-reported health status data from different populations, even when identical survey instruments and methods are used. The WHO Household Survey Study carried out 69 representative household surveys in 60 Member States in 2000 and 2001 using a new health status instrument based on the International Classification of Functioning, Disability and Health, which seeks information from a representative sample of respondents on their current states of health according to six core domains. These domains were identified from an extensive review of the currently available health status measurement instruments. To overcome the problem of comparability of self-report health data, the WHO survey instrument used performance tests and vignettes to calibrate self-reported health on selected domains such as cognition, mobility and vision. WHO has developed several statistical methods for correcting biases in self-reported health using these data, based on the hierarchical ordered probit (HOPIT) model. The calibrated responses are used to estimate the true prevalence of different states of health by age and sex. Annex Table 4 reports average HALE at birth for Member States for 2000 and 2001, and for 2001 the following additional information: HALE at age 60, expected lost healthy years (LHE) at birth, percent of total life expectancy lost, and 95 percent uncertainty intervals. LHE is calculated as life expectancy (LE) minus HALE and is the expected equivalent number of years of full health lost through living in health states of less than full health. The percentage of total LE lost is LHE expressed as a percentage of total LE and represents the proportion of total LE that is lost through living in health states of less than full health. HALEs for 2000 differ from those published in The World Health Report 2001 for many Member States, as they incorporate new epidemiological information, new data from health surveys, and new information on mortality rates, as well as improvements in survey analysis methods. The uncertainty ranges for healthy LE given in Annex Table 4 are based on the 2.5th percentile and 97.5th percentile of the relevant uncertainty distributions. The ranges thus define 95 percent uncertainty intervals around the estimates. HALE uncertainty is a function of the uncertainty in age-specific mortality measurement for each country, of the uncertainty in burden of disease based estimates of country-level disability prevalence, and of uncertainty in the health state prevalences derived from health surveys.

Source: World Health Report 2002. Geneva: World Health Organization.

Appendix B

B1

Summary Statistics: Levels and Long-Term Effects

Variables	Mean	Standard Deviation	Minimum	Maximum
Dependent Variable
HALE	56.33	12.00	29.1	74.5
Independent Variables
Lagged Conflict (major war)	0.049	0.2	0	1
Lagged Conflict (intermediate war)	0.08	0.27	0	1
Lagged Conflict (count)	0.31	0.73	0	6
Lagged POLITY score	4.52	5.78	−9	10
Lagged POLITY score squared	53.79	35.28	0	100
Lagged ln(Per Capita GDP)	8.47	1.11	6.14	10.69
Lagged ln(Trade Openness)	4.27	0.52	2.84	6.09
Lagged ln(Population)	9.32	1.52	5.61	14.05
Education	0.78	0.20	0.16	0.99

Note:N=367.

GDP, Gross Domestic Product; HALE, Health-Adjusted Life Expectancy.

Appendix C

C1

Summary Statistics: Changes and Short-Term Effects

Variables	Mean	Standard Deviation	Minimum	Maximum
Dependent Variable
ΔHALE	−0.74	2.22	−12	7.8
Independent Variables
Lagged Conflict (major war)	0.06	0.23	0	1
Lagged Conflict (intermediate war)	0.07	0.26	0	1
Lagged Conflict (count)	0.33	0.75	0	6
Lagged ΔPOLITY score	0.25	1.78	−13	12
Lagged ΔPOLITY score squared	3.22	17.30	0	169
Lagged Δln (Per Capita GDP)	0.02	0.07	−0.44	0.55
Lagged Δln(Trade Openness)	0.02	0.12	−0.46	0.63
Lagged Δln(Population)	0.01	0.01	−0.08	−0.05

Note: N=367.

GDP, Gross Domestic Product; HALE, Health-Adjusted Life Expectancy.

Footnotes

1

For a discussion of allocation of public and private resources to health expenditures, see Ghobarah, Huth, and Russett (2004)

2

The education variable is not included in the second model, which assesses the relationship between short-term changes in the covariates and HALE, because the education measure used here does not change on a yearly basis.

3

As in a large number of studies in international relations, missing data is an issue in these models. Although HALE is available for nearly all the countries in the sample (170 of 189), a number of states are dropped from the models due to missing data on the covariates, particularly the economic variables. For instance, trade figures are missing for the entire period being studied for several African states, including the Sudan and Congo; economic data are also sparse for some Central Asian and Middle Eastern states. This resulted in the exclusion of 33%, 35%, and 38% of the countries for which HALE is available for 1999, 2000, and 2001, respectively. However, the group of countries that has been dropped from the analysis due to missing data on independent variables does not significantly differ in health output levels from states that are included in the models; in the first model, the mean HALE is 56.33 for the countries in the analysis and 54.37 for those not included (t=−1.82, p=.07, two-tailed). The two sets of states are also very similar in economic and social factors; the GDP means of states included in the model and those excluded are 8.47 and 8.60, respectively (t=0.72, p=.47). The mean education levels of the two groups are 0.78 and 0.74 (t=−2.07, p=.04). Thus, the groups of countries included and excluded from the analysis have no significant differences along the economic and social dimensions considered. However, the states that were dropped from the model due to missing data on independent variables have an average democracy score (−3.62) that is generally lower than the mean democracy level of the states in the model (4.52) (t=−12.02, p<.001). Moreover, the excluded countries had higher levels of intense conflict than the included group; the intense conflict mean for the former group during the period studied was 0.12, as opposed to 0.04 for the latter (t=2.89, p=.004).

4

A possible influence on health achievement that has not been included in these models is a state's health care capacity. States with larger health care capacities—including hospitals, health care personnel, and medical supplies—are likely to have higher HALEs. However, health care infrastructure—as well as general infrastructure—can be damaged or even destroyed due to violent conflict (Ghobarah et al. 2004). In fact, it is not uncommon for health care facilities to be targeted during combat. As a result, controlling for this variable in the models in Tables 1 and 2 would confound the influence of the primary covariate of interest, conflict. Ray (2003) recently warned against controlling for intervening variables or including independent variables that are related to each other. I did, however, estimate alternative models using immunization rates as a proxy for health care capacity, but this variable did not have any significant effect.

5

As discussed in footnote 3, approximately 35% of the states for which HALE is available were not included in the analysis due to missing data, mainly on economic variables. The states that were excluded from the analysis experienced significantly higher levels of intense conflict than the countries that were included in the model, and bivariate analyses reveal that the negative relationship between conflict and health is considerably stronger in the group of states excluded from the analysis than in the included group. Thus, the exclusion of states with high levels of conflict likely mitigated the effect of conflict on HALE as revealed by the results in Table 1, and the negative relationship between conflict and health would be stronger if those states were included in the analysis.

6

It should be noted, as in Ghobarah et al. (2004), that some of the states with observed health outcome levels that are significantly lower than their expected HALE have high HIV prevalence rates. These outliers include South Africa, Namibia, Botswana, Swaziland, Zimbabwe, and Zambia. High levels of HIV/AIDS prevalence could have clear repercussions for health achievement. Moreover, arguments have been made about the contribution of conditions created by civil wars in Sub-Saharan Africa to an increase in HIV/AIDS rates.

7

Both this model and the model assessing short-term changes in health were also estimated using a dichotomized measure of democracy, in which states with a POLITY IV score of 5 or higher were considered democracies and those with a score lower than 5 were considered non-democracies. The results with this dichotomous measure of democracy were very similar in significance to the models shown, with a higher coefficient and a higher standard error.

Author notes

Author's note: A previous version of this paper won the Malcolm Jewell Award for the best paper presented by a graduate student at the Annual Convention of the Southern Political Science Association, New Orleans, LA, January 7–10, 2004. Thanks to David Davis, David Lektzian, Dan Reiter, Suzanne Werner, Christopher Zorn, and three anonymous reviewers for valuable comments. The data and procedures necessary for replicating the analyses in this article are available at http://www.isanet.org/data_archive.html.