Of the obstacles to development in the sub-Saharan African (SSA) region, perhaps none has had a more profound impact than the dual burdens of HIV/AIDS and conflict. During the past quarter century, no region of the world has been more acutely affected by large-scale violent conflict than SSA. Almost all SSA countries have directly experienced, or border a country that has directly experienced, violent conflict. The number of states engaged in significant violent conflicts doubled between 1989 and 2000, from 11 to 22 [1, 2]. A "conflict belt" stretches from Angola to the Horn of Africa, while peace in western Africa remains elusive.

The same quarter century has seen the development of an even more pervasive crisis – the HIV/AIDS pandemic. SSA continues to account for the large majority (66%) of the world's HIV/AIDS cases [3], even though its population comprises only 10% of the world's total population. There is great diversity across Africa in the levels and trends of HIV infection. While stabilization of the epidemic appears to have begun in several countries, population prevalence rates across most of the continent continue to increase, including among rural populations and sociodemographic groups not previously considered at elevated risk.

Although much has been written about HIV and conflict individually, surprisingly little has been written about the dynamics of the relationship between the two crises. In fact, an extensive keyword search of the internet and peer-reviewed journal databases turned up fewer than 100 references to the intersection of HIV and conflict. Virtually all studies are descriptive and only one utilized hypothesis-testing analytical strategies [4]. Although constraints on conducting such research are considerable, research on the interface between the two crises is of critical importance. A clearer understanding of the dynamics of the interface between conflict and HIV is crucial for the development of effective and efficient strategies to reduce population risk.

Some literature on conflict-affected countries in SSA underscores mechanisms by which HIV/AIDS and violent conflict may exhibit bi-directional causal associations on the population level. HIV/AIDS has been recognized to play a potential role in creating conditions conducive to violent conflict on the continent, although there is as yet no good empirical evidence to that effect. A policy forum held at the United States Institute for Peace [2] concluded that the pandemic will emerge as a deeply destabilizing force across the social, political, and economic landscapes of Africa. For example, large numbers of AIDS orphans and vulnerable children strain the social support networks in many southern African countries [5, 6]. The psychological trauma and the lack of parental affection and supervision experienced by AIDS orphans put them at risk of involvement with criminal or antisocial activities (e.g. as child soldiers), or activities that serve to further their risk of contracting and further propagating the disease (e.g. as child prostitutes). The extent of the risks faced by AIDS orphans – e.g. risks for anti-social and criminal behavior later in life – needs to be sufficiently researched. By causing elevated AIDS-related mortality among the 15–45 year-old age group, the illness degrades human capital, undermines household capacity and stability, reduces economic productivity, and robs social and political institutions of intellectual resources [7].

On the other hand, violent conflict clearly influences the epidemiology of HIV. To date, the literature emphasizes conflict as a risk factor for HIV transmission. Conflict destroys social and physical infrastructure, resulting in untreated sexually transmitted infections (STIs), poor health and malnutrition and, as a consequence, increased risk of transmission in the event of viral exposure. The migration and poverty created or exacerbated by conflict may result in increased exposure opportunity through: (1) increased prevalence of casual or commercial sexual activity; (2) increased interactions among civilians and combatants/military personnel, known for their high risk behaviors; (3) the development of cultures of violence that promote sexual violence and predation; (4) mass migration, which increases sexual mixing among populations; and (5) the destruction of public health education mechanisms (e.g. mass media, health facilities, and formal education), which negatively affects public health-related knowledge, attitudes, and practices [8, 9].

However, the epidemiologic data despite their limitations², suggest that the relationship between conflict and HIV levels may be more complex than this general picture of joint potentiation implies. With few exceptions, countries that have experienced widespread violent conflict have apparently lower levels of HIV infection compared to those that have experienced relative peace (compare figures 3c and 3d). While many factors operate during conflict to increase vulnerability of affected populations to HIV, exposure opportunities at an aggregate level may actually be lower, therefore decreasing HIV risk. Alternatively, when conflict subsides, exposure opportunity may increase, leading to the potentially explosive spread of HIV. At the end of the Angola conflict in 2002, the country HIV prevalence was relatively lower than the rates in other Southern African countries, which suggests that conflict may have slowed HIV spread in this case [10].

The present paper evolved out of the observation that available evidence for conflict-affected countries in SSA suggests a complex set of possible net effects of conflict on the HIV epidemic, with multiple and quite variable outcomes. The implications of the analysis and findings have great significance for developing policies and programs to confront HIV in the context of conflict. These are discussed along with recommendations for improved policies and programs to address this important constraint to African health development.

We also develop the notion that conflict is a complex social phenomenon and that its effects are highly contextualized.

To articulate a framework for understanding the interface between the two crises, the concepts of vulnerability, hazard exposure opportunity and risk are useful. These are borrowed from the disaster literature [11] but are germane for understanding the evolution of HIV/AIDS as a crisis. Vulnerability is the ability of a population to withstand hazards or shocks to the system when they are present. Classical vulnerability factors include poverty, low education/knowledge, poor social infrastructure and attitudinal factors. Additional factors specific to conflict and HIV include levels of economically-motivated sex, levels and types of civil-military interactions, migration patterns, sex-related knowledge and attitudes, and population health status (particularly STIs and nutrition). Vulnerable groups in conflict settings may include child-headed households, child soldiers, unaccompanied children, women, demobilized soldiers, and repatriating refugees. Vulnerability means that, when exposed to a hazard, an individual or group is more likely to experience adverse effects (risk).

Hazards or shocks to human communities such as conflict affect the HIV exposure opportunity of members of these affected communities. HIV exposure opportunity can be mediated by war or other ways in which these communities interact with other communities (regional human ecology), which ultimately affects disassortive mixing (i.e. mixing of population groups of differing levels of HIV prevalence). In Southern Africa, the regional ecology is one that favors extraordinarily high HIV risk through, for example, high population density, good physical infrastructure, economically-motivated mobility, high poverty levels, and large wealth disparities within and between countries. By lowering exposure opportunity, war may lead to isolation from the general regional ecology and, therefore, to lower HIV risk at the population level. Alternatively, war leads to a changed and often more intense mixing of mobile military populations and civilians, which can increase HIV risk through disassortive mixing.

Figure 1 illustrates how conflict shocks may affect vulnerability and exposure opportunity to affect population HIV risk. Conflict shocks can weaken a community's ability to avoid HIV exposure/infection (vulnerability), or it can influence exposure opportunity itself. Vulnerability and exposure are the basic determinants of population-level HIV risk. Underlying determinants here include violent conflict and the regional ecology of HIV, which exhibit measurable effects on both vulnerability and exposure opportunity (and interact with each other). Finally, HIV infection goes on to influence the progression of conflict and the social ecology of HIV in a feedback loop. The model has utility for structured inquiry into the effects of conflict on HIV. It draws attention to mechanisms by which vulnerability and risk of exposure, and therefore risk of transmission, may be increased or decreased during conflict. The relative importance of each of the components, and therefore the ultimate effect of conflict (i.e. augmenting or reducing risk), will be highly context specific.

Figure 2 illustrates that there is an important time dimension to consider as well. The most basic distinction of interest is that of conflict and post conflict, although these are rarely clear-cut distinctions. What is important to note here is that vulnerability increases during conflict and rarely decreases rapidly after conflict subsides because of the profound societal changes that generally occur during conflict. On the other hand, opportunity for exposure to HIV may change dramatically post conflict due to decreased population isolation and rapid improvement in freedom of movement. These changes will likely be most acute in countries with high economic potential, such as Angola, Côte d'Ivoire and the Democratic Republic of Congo. The juxtaposition of high vulnerability and increased exposure opportunities post-conflict can lead to explosive growth in the epidemic. This already has occurred to some extent in Mozambique [12].

This finding is critical in that it explains in part why, in some chronic conflict settings, apathy has developed towards HIV because seroprevalence levels may be low. However, the potential for rapid progression of the infection post conflict due to high vulnerability argues for a much more aggressive and deliberate approach to HIV in post-conflict settings.

Considerable research has been conducted to date to construct comprehensive typologies of conflict. Several well-established research initiatives, such as the Correlates of War Project [13], have developed typologies of conflict for application to conflict early warning and prevention efforts. These typologies categorize conflicts based upon variables associated with either the determinants of conflict or the manifestations of conflict. Typologies that focus on the determinants of conflict categorize conflicts primarily based upon characteristics of the political, legal, economic, social, or cultural environments that are associated with greater risk of conflict. Examples of such factors include [14]:

In contrast, some approaches focus on the manifestations of conflict, defining categories in terms of measurable aspects of the conflict and its effects on populations. Examples include:

A determinant-based typology has more direct application to the critical areas of conflict early warning and prevention, despite the analytical complexities derived from the high degree of interdependence (i.e. interaction) among these determinants [15]. However, most existing classification schemes fail to incorporate an emphasis on the public health impacts of conflicts other than battle-related fatalities. For example, other categories of war-related mortality, such as civilian deaths, deaths following famine or disease and deaths that are allowed to occur for political reasons, are often omitted. Similarly, morbidity, malnutrition and displacement among civilians, highly characteristic of post-Cold War conflicts in SSA, have been largely overlooked [14]. These population-level results of conflict appear, in many cases, to help explain the links between conflict and HIV/AIDS progression. Therefore, the present framework draws from both types of typologies for maximum utility regarding the problem of HIV and conflict, recognizing the complex nature of conflict as an epidemiologic study factor.

Despite large cross-cultural and historical variation in the determinants and manifestations of conflict, and variation in the systems that have been developed to analyze conflict, a model can be developed specific to inquiry about the public health impacts of conflict on HIV epidemiology. Conflict can be analyzed in terms of the following dimensions, all of which may help to shape its influence on HIV risk.

The models and mechanisms identified above help to develop an explanatory model built on a foundation of social ecology. A social ecology approach argues for models of explanation that explore the interactions of multiple causes (environmental, social, and biological) working through different mechanisms and at differing levels of scale (individual, household, community etc.) with emphasis on dynamic change over time. The social ecology of the epidemic is defined in terms of sociodemographic, socioeconomic, structural, contextual, biologic and behavioral variables that facilitate or inhibit progression of the epidemic [68–70]. The ecological perspective explains how conflict might impede the progress of the epidemic and it provides clues as to regional strategic factors that may exacerbate risk in the post-conflict setting. While the epidemiologic literature postulates a number of factors that we do not elaborate here (such as comorbidity and religious practice) these factors also are likely to be important. For the sake of illustrative simplicity, we examine four key variables not commonly cited that are more closely linked to the conflict-HIV axis: (1) population density, (2) level and geographic distribution of economic growth, (3) the prevalence of poverty, and (4) the existence of physical infrastructure, particularly transport infrastructure.

One key feature of mapped data is the relative correspondence of population density, road infrastructure, economic productivity, and HIV risk in Southern Africa (see Figures 3a,3b,3c). Central Africa is marked by more clustered populations and sparse road infrastructure, especially in Congo and Angola – those areas that have experienced chronic conflict. Coastal West Africa has high population density and road density but comparatively low HIV seroprevalence. This may be due in part to poor road quality and lack of road connectedness to areas of high HIV risk. Also the economic productivity of western Africa is much lower than that of southern Africa. Therefore, economic migration patterns may not be as intense. As mentioned earlier, other factors, such as the moderating influence of religion and biological co-factors (e.g. circumcision, and STI rates), may also be important. It is also possible that conflict in central Africa may have moderated HIV risk by buffering the west coast of Africa from movement of HIV up the coast from southern Africa.

Population density may directly influence the occurrence, severity, and spread of violent conflict and HIV. Where population density is lower, population mixing associated with conflict may be more sporadic (as opposed to consistent and sustained). Population density paired with excellent road infrastructure may have a synergistic effect on HIV risk. This may, in part, explain the pattern of risk seen in southern and eastern Africa.

Economics and poverty are important drivers. Although higher economic output is associated with higher levels of HIV in general, the poor are increasingly affected for a variety of well-accepted reasons. From this perspective, Southern Africa remains the most problematic with respect to the potential for rapid northward spread of HIV. Angola and Mozambique are characterized by high poverty, economic potential, and in the case of Angola, the potential to be a gateway to West Africa. One factor that may be driving the epidemic is the interaction between poor women and rich men (or men with some income, such as the military and truck drivers) as exemplified by Drain et al. [68] who found that income inequality was independently associated with HIV seroprevalence.

The ecology of HIV and conflict also results in greater micro-level variation in the determinants of HIV risk. Mock and Drapcho [37] showed that regional variability and the design effects (sample variation across clusters/villages) associated with measures of nutrition and mortality were far greater among post-conflict countries surveyed by DHS than are typically found in more stable settings. This is plausible given that conflict does not have a consistent effect in all areas of countries. Also, it would be expected that functioning social systems found in stable settings might have an equalizing effect on health status.

Sadly, the international community has compartmentalized its responses to HIV and conflict – as well as its relief, recovery, and development programming instruments – which precludes an integrated and aggressive attack on the epidemic. As a result, very little has been learned in terms of designing and implementing programs to address HIV among conflicted-affected populations, with some exceptions [71].

The current structure of development assistance results in poor availability of resources to address HIV in conflict settings, as these are traditionally the domain of humanitarian assistance. To date, though much lip service is paid to the concept of developmental relief, the bulk of all humanitarian assistance is supply-side delivery of immediate survival commodities and services. This approach continues to dominate even though humanitarian assistance may be provided to conflict-affected population for years and decades, and even when displacement may provide unique opportunities to reach populations that might otherwise be inaccessible.

Similarly, the immediate post-conflict period is typically serviced by transition programs that emphasize demobilization, reinsertion, reintegration and development of the foundations of governance. While these are clearly important considerations, a problem as severe as HIV cannot be compartmentalized as a development problem (i.e. not a humanitarian or transition-period concern). And indeed, problem-focused activities such as HIV prevention might be a motivating cause for the stimulation of civil society groups.

Even internal to the development community, the problem of HIV is not linked to the problem of conflict, but rather these two issues are seen and treated as unrelated concerns. Different strategies and partners are managed by separate bureaucratic units that have rare interactions. Such compartmentalization of both conflict and HIV along humanitarian/development lines and internal to development management has resulted in fragmented approaches to addresses these important and interactive influences on the health development of SSA.

Similarly, the transitional period should be assessed for opportunities to creatively integrate HIV-related interventions into priority program strategies to promote economic recovery and social rehabilitation. Examples of such integrated programs include integrating HIV education into micro-credit or economic development programs and incorporating HIV risk reduction strategies into demobilization and social integration programs. Demobilized ex-combatants might also be incorporated in educational and health extension programs such that they become part of the solution instead of the problem.

Humanitarian assistance provides numerous opportunities for the cross-sectoral integration of effective HIV-directed initiatives, including the opportunity for synergistic amelioration of adverse health outcomes.

Even though conflicts typically last for years or decades, countries affected by conflict continue to receive assistance, primarily for immediate survival needs, through humanitarian assistance efforts. In post-accord transition programs, the emphasis is on establishing democratic rule, governance, and re-integration. Sequential application of traditional humanitarian and transition interventions ignores the regional context of high HIV risk. Since HIV risk may be lower in conflict settings while vulnerability to HIV is increasing, this programming approach results in a major constraint to timely, prevention-oriented intervention.

Our analysis suggests that the relationship between conflict and HIV is complex and contextualized; however, important general conclusions can serve as a basis for action. The analysis demonstrates that while vulnerability to HIV is heightened as a result of conflict, exposure opportunities may be significantly reduced. This may lead to important opportunities for interventions to keep HIV prevalence rates low in the affected settings (i.e. to prevent HIV in relatively lower-prevalence countries). More urgently, post-conflict changes in exposure opportunities could result in explosive epidemic waves. The case of Angola is particularly troublesome given its strategic location and economic potential.

Conflict and HIV are inter-related problems that demand a clear strategy and coordinated use of humanitarian and development assets. This means that objectives for addressing these problems should drive the response, and conflict should be viewed as a key determinant of HIV risk. The HIV/AIDS epidemic is a major constraint to development in Africa. Programs and bureaucracies should more clearly align along a consistent vision of HIV prevention and mitigation in both conflict and post-conflict settings.

A portion of the Declaration of Commitment on HIV/AIDS, adopted at the United Nations General Assembly Special Session on HIV/AIDS (UNGASS) on June 27, 2001, articulates strategies and goals to address HIV/AIDS in conflict and disaster-affected regions. It calls for the development and implementation of national strategies that incorporate HIV/AIDS awareness, prevention, care, and treatment elements into programs or actions that respond to emergency situations. The declaration recognizes that populations destabilized by armed conflict, humanitarian emergencies, and natural disasters – including refugees, internally displaced persons and, in particular, women and children – are at increased risk of exposure to HIV infection, and calls for HIV/AIDS components to be factored into international assistance programs where appropriate [72].

As international, regional and national agencies strive to abide by the Declaration of Commitment on HIV/AIDS, the framework provided here will be useful in the identification of determinants of HIV risk. We recommend that factors affecting HIV risk in conflict settings be systematically assessed as a basis for strategic planning to address HIV in conflict. The assessment should include vulnerability profiling, exposure risk assessment, and characterization of mechanisms through which conflict affects HIV risk. It is particularly important to pre-empt disassortive mixing associated with post-conflict improvements in mobility and resettlement. The assessment should be conducted at the different planning levels (i.e. regional, national and sub-national).

Conflict risk assessment should also be a key component so that synergistic programming between conflict and HIV initiatives can be achieved. Examples of these include:

Our analysis stresses the importance of maintaining a global perspective, while at the same time recognizing that micro-planning may be even more important in conflicted-affected settings than in stable settings. This is because local ecologies may be more diverse due to the varying spatial and temporal effects of conflicts. Indeed, most countries that have experienced large-scale conflict have had quiet zones where life was relatively stable, even during the course of large-scale war. These stable areas could absorb major non-emergency initiatives. In post-conflict settings, these micro-level differences persist, while others, such as differing levels of infrastructure destruction, ethnic tensions and pockets of high HIV prevalence, may require highly specific approaches.

These findings also suggest somewhat different approaches to HIV surveillance in conflict-affected settings. First, a more deliberate attempt should be made to support surveillance in countries and areas affected by conflict so that a better evidence base is developed. Unfortunately, a review of the UNAIDS surveillance database revealed that conflict-affected countries have little or no systematic surveillance, despite the existence of relatively stable areas [73]. This may be due to the fact that lab testing and quality control for HIV surveillance is generally centralized. Perhaps a more de-centralized approach would better ensure continued data collected in conflict-setting such that at least some regions of a country would continue to collect data during conflict. Higher micro-level variation and greater social change in the post-conflict setting argues for more finely graded surveillance. Also, surveillance should take into account the differing risk groups resulting from conflict, and the dynamics of exposure opportunity that occur as a result of opening international borders, rapidly evolving trade corridors and improved internal mobility.