Definitions

“A disaster may be defined as a relatively acute situation created by man-made, geophysical, weather-related, or biological events that adversely impacts on the health and economic well being of a community to an extent that exceeds the local coping capacity” (Toole and Waldman, 1997: 284). Complex humanitarian emergencies ² are distinguished from acute natural disasters because population displacement and the lack of basic services available to a migrating population result in indirect or secondary health and mortality effects to a degree not usually present in a natural disaster. The disruption of services and life generally can often be addressed with some rapidity, especially if the population remains more or less in place. The difference between a complex emergency and a natural disaster is not necessarily in the mortality rate per se. Natural disasters can result in huge loss of life as a result of earthquake, weather, or other natural causes. Complex emergencies, in addition to being caused by human beings, typically involve large-scale population displacements and the disruption of normal life to an extent that is beyond the means of typical coping mechanisms of a society. People may be displaced either within a country—internally displaced persons (IDPs)—or between one or more countries—refugees. It is the unusual and threatening conditions brought on by the disruption of society that lead to negative health and mortality consequences for such populations.

The concept crude mortality rate (CMR) is discussed frequently throughout the volume, which demographers often refer to as the crude death rate (CDR). The concept denotes the number of deaths in a given period of time divided by an estimate of the population at risk of dying during that period (Shryock and Siegel, 1976). In this chapter, we will refer to the number of deaths per 10,000 population per day as the daily crude mortality rate or CMR, and to the number of deaths per 1,000 population per year as the annual crude death rate or CDR. The two expressions are convertible by multiplying the CDR (more familiar to demographers) by 36.5 to obtain the CMR (more familiar to epidemiologists working in complex emergency situations).

Baseline mortality is the “normal” mortality level in a given population. Epidemiologists often refer to a “return to baseline level,” which indicates a stabilization of the situation and potential end to the mortality crisis. However, with refugee or internally displaced populations, it is often difficult to define the baseline, because the population of comparison may not be clearly defined, populations may have chronically high mortality rates due to ongoing conflict and other problems, and surveillance may have started well into the period of elevated mortality.

Typology

Grouping various complex emergencies into distinct categories may help emergency aid organizations to identify the types of assistance that are most likely to be needed early in a crisis. One such typology distinguishes between five types of crises based on their settings and patterns of population risk (Hansch, 1999).

• Rural Famine or Refugee Paradigm: This is the model on which most relief work has traditionally been based. Populations are expected to be rural, poor, and illiterate, with low vaccination coverage and high chronic malnutrition, and they are generally housed in high-density camps. Mortality is often due to communicable diseases compounded by malnutrition. Deaths generally occur disproportionately among children less than five years of age. Examples of this type of crisis include: Biafra, Nigeria, in 1968; the Sahel in 1973, and Sudan, Ethiopia, and Somalia in the late 1980s and early 1990s.
• Ethnic Cleansing or Genocide: This type is increasingly common and is characterized by armed forces (sometimes assisted by civilians) attacking large numbers of civilians in an effort to kill or displace them. Mortality is due in large part, if not mainly, to physical injury, not communicable diseases or malnutrition. Disability and mental health trauma are other important consequences of this type of emergency. Examples of this type of emergency include: Rwanda in 1994; Bosnia in the early 1990s, and Kosovo in 1999.
• Urban Services Collapse or Urban Depopulation: This type of crisis occurs when generally healthy and well-nourished populations who are dependent on urban services become refugees due to war. Mortality is usually due to chronic diseases and lack of sophisticated health systems (i.e., kidney dialysis machines). This type of crisis has occurred within larger crises in Somalia, Bosnia, and Kosovo.
• Conflict Among Combatants: Most mortality occurs among armed combatants due to battle injuries, landmines, collateral damage, or communicable diseases associated with the effects of war. This type of emergency includes: Cambodia and Angola in the 1980s and 1990s (where landmines were a significant mortality risk) and Chechnya.
• Short-Onset, Short-Duration Natural Disaster: Hurricanes, tornadoes, and earthquakes can create high mortality rates at the beginning of a crisis based on physical trauma or environmental exposure. However, these types of disasters can lead to longer-term problems such as famine and disease if they are not addressed immediately. Examples include: floods in Bangladesh and earthquakes in Mexico and South America. This type of emergency is not discussed in detail in this volume because it is generally caused more by natural than political factors.
  Clearly, these categories are rarely completely distinct and often overlap, but may be useful in a debate about how the nature of complex emergencies are evolving over time (see below).

Extent of the Problem

Although complex emergencies have been occurring for centuries, systematic data on the numbers of forced migrants in the world have only been available for approximately the past 40 years. The number of refugees and IDPs in the world has increased dramatically during the past four decades. As Figure 1-1 shows, by the end of 1998, there were over 11.6 million refugees, almost twice as many as there were 36 years ago. Yet in recent years, since about 1991, the number of refugees has generally declined, despite a brief rise in the late 1990s (United Nations High Commissioner for Refugees, 2000). Meanwhile, the number of IDPs has grown quite rapidly, reaching over 25 million by 1994, although this figure also declined slightly in the late 1990s to about 17 million by 1998. The map that follows page 18 shows that, refugees and internally displaced persons are located around the globe—in Africa, Central and South America, Eastern Europe, the Middle East, and Central and Southern Asia (United States Committee for Refugees, 2000). Due to the political nature of flows of refugees and internally displaced persons, one must acknowledge not only the effect of global and local political events, but also the willingness of states and international organizations to count persons as refugees and IDPs. This varies with circumstances; therefore, interpretations of trends in the number of forced migrants require caution.

FIGURE 1-1

Global trends in refugees and internally displaced persons, 1964-1999. United States Committee for Refugees and United Nations High Commissioner for Refugees, various years.

Although not every refugee or IDP in the world is currently affected by a complex emergency, complex emergencies do produce forced migrants. The number of complex emergencies has also increased over the past decade. In 1989, there were 14 ongoing emergencies; in 1992, there were 17. By 1996, there were 24 ongoing complex emergencies in the world, and there were about 30 by the end of 1999 (Natsios, 1997; United States Committee for Refugees, 2000). However, the realignment of state boundaries and the creation of additional states in places such as the former Soviet Union and former Yugoslavia may have some effect on these statistics. This apparent increase in emergencies has been accompanied by a parallel increase in emergency foreign aid expenditures by the United States. In 1989, the U.S. spent $300 million in bilateral aid for foreign disasters and crises. By 1994, it was spending $1.3 billion. There is a corresponding trend in multilateral expenditures for emergency assistance. Between 1984 and 1989, for example, the World Food Program spent 25 to 40 percent of its annual assistance budget on relief activities. By 1992-1993, this was up to over 60 percent (Natsios, 1997). However, it is important to note that an increase in emergency foreign aid does not necessarily translate into an increase in overall foreign aid.

Levels of Mortality

In complex emergencies, the crude mortality rate (CMR) is often expressed as the number of deaths per 10,000 population per day during the acute phase of an emergency. Calculating a daily rate has been considered to be appropriate since conditions can change dramatically on a daily basis and the large base of 10,000 per day is used to express events in whole numbers. In developing countries, the median crude death rate (CDR) for the total population is 9 deaths per 1,000 per year (Population Reference Bureau, 2000). This translates into a daily rate of 0.25 deaths per 10,000. A threshold of 1.0 per 10,000 per day is widely used as the benchmark of elevated mortality, on the recommendation of the Centers for Disease Control and Prevention (1992). This threshold of 1 per 10,000 per day is equivalent to an annual CDR of 36.5 per 1,000. ³

Although the CMR and CDR are essentially the same concept, there are reasons for preferring to use one rather than the other. Demographers have traditionally favored longer reference periods for demographic rates as they are generally interested in average mortality over a period of time. Epidemiologists working in emergencies, however, are interested in the “instantaneous” rate. Therefore they use the daily rate (CDR) ⁴ to observe rapid changes in the mortality rate which shows whether or not the situation is stabilizing.

Elevated CMRs in complex emergencies vary widely. Table 1-1 , based on data in Toole and Waldman (1997) provides a dozen examples of CMRs and CDRs expressed in terms of daily rates per 10,000 and annual rates per 1,000. The table has the virtue of providing information on emergencies in different parts of the world around the same time period, as well as estimates for some of the same countries at different times and estimates for refugee populations from the same origin country in different asylum countries.

TABLE 1-1

Estimated Daily Crude Mortality Rates (CMRs) and Annual Crude Death Rates (CDRs) in Selected Refugee Populations, 1990-1994.

The data shown in Table 1-1 indicate a daily CMR on a base of 10,000 of over 1 in all of the cases given. The range is between 1.2 in the case of Mozambicans in Malawi in June 1992 to some of the highest levels ever measured—between 19.4 to 30.9 deaths per 10,000 per day at the height of the Rwandan crisis in July 1994. The Rwandan levels, if sustained, would have meant that every refugee would have been dead in less than a year. (The level of 1,127.9 per 1,000 per year means annihilation in less than a year.)

The heavy reliance on data collected from camp populations may distort understanding of the levels and trends of mortality among the total refugee and internally displaced populations. Camp populations may benefit from earlier and more effective assistance interventions that lead to a reversal of the high mortality levels associated with the emergency and result in a more quickly stabilized situation in terms of food, shelter, sanitation, and other basic needs. On the other hand, camp situations may increase the risk of subsequent mortality due to infectious diseases. Although delivering assistance in camps may be more manageable for providers, it may not be more effective for recipients. Under certain circumstances, self-settlement among a host population may be more effective (Van Damme, 1995).

Complex emergencies rarely continue indefinitely. In most cases, international organizations, national governments, nongovernmental organizations, and others intervene to provide some stability for the refugee population, and minimal services are aimed at reduction of mortality, reduction of morbidity, and other threats to life. What the data in Table 1-1 underscore is that the acuteness of the challenge, as indicated by CMRs, varies enormously from situation to situation.

Internally displaced persons, who often face the same difficult survival conditions as refugees who have crossed an international border, also face the prospect of elevated mortality. Because of considerations of sovereignty and the absence of international agreements about the provision of protection and assistance to victims of persecution and war who remain in their own country, internally displaced persons are less likely to receive international assistance that might meet survival needs and provide a modicum of stability. Although mortality data on internally displaced populations are scarce, most of the situations for which data are available display very high mortality rates. As shown in Table 1-2 , the CMR in Baidoa, Somalia, in 1992 was almost 17 deaths per 10,000 per day and in both Sudan in 1992-1993 and Angola in 1995, it was over 7 per 10,000 per day. Crude mortality rates among Muslims in Bosnia during the war in 1993 were about four times the baseline level (Toole et al., 1993).

TABLE 1-2

Estimated Daily Crude Mortality Rates (CMRs) and Annual Crude Death Rates (CDRs) Among Internally Displaced Persons, 1990-1994.

Stages of a Crisis

The data in Table 1-1 highlight the degree to which mortality can rise in crisis situations but reveal nothing about patterns of mortality over the various stages of a particular complex emergency. Each complex emergency is typically different from the last: different logistics, different politics, different social context, etc. However, some generalizations are possible. Figure 1-2 shows the classic rural famine/refugee paradigm pattern, which is a refinement of an inverted U-shaped pattern. Note the sharp increase at the beginning of the crisis (Phase 1), followed by the peak mortality rate (Phase 2) and then a relatively rapid decline (Phase 3), and stabilization (Phase 4). These distinctions should be based not so much on absolute measurements, but on patterns. In other words, in an emergency, population parameters—including mortality—may be quite unstable—either fluctuating or rapidly changing due to interventions or other reasons. The post-emergency phase is usually marked by more stable mortality rates, even though they might remain unacceptably elevated. However, stabilization is what signals the time to shift programming from life-saving interventions to longer-term ones.

FIGURE 1-2

Model of mortality change in a forced migration situation. Source: Reed et al., 1998, Figure 2.

Typically, the period of flight and the time immediately after arrival in a place of asylum are the periods of highest mortality. In 1992, in Chambuta camp, Zimbabwe, for example, Mozambican refugees who had been in the camp for less than one month had a CMR of 8 per 10,000, which was four times that of those who had been in the camp for one to three months and 16 times the baseline (Centers for Disease Control and Prevention, 1993a). In Goma, Zaire, among Rwandan refugees, the average daily CMR from July 14 to August 14, 1994, was between 19.5 and 31.2 per 10,000. This was more than 30 times the baseline rate (Goma Epidemiology Group, 1995).

The rate at which mortality rates decline varies across populations, and the speed of mortality reduction also depends on the rates of mortality and/or out-migration of specific groups at high risk for mortality. For example, the initial high mortality rates of Cambodian refugees in Thailand in 1979 declined to baseline levels in about one month (Toole and Waldman, 1990). In Goma, Zaire, among Rwandan refugees, nearly 2,000 deaths per day were estimated on July 21, 1994, but by July 28, there were over 6,500 deaths per day. By August 4, the number of deaths per day was less than 1,000 (Goma Epidemiology Group, 1995). Although this is still a large number of deaths, the acuteness of the crisis moderated with some rapidity. Other situations, however, are much harder to stabilize, usually because of political factors. In 1988-1989, for example, under-five mortality among Somali refugees in Ethiopia remained high for about 18 months, even increasing during some periods (Toole and Bhatia, 1992).

It is important to note that this model does not hold true for all complex emergencies and it has not been systematically validated. It is simply an approximation of mortality patterns that have been observed in many of these situations in the past. Although it is possible to speculate about the factors that cause a shift in the mortality pattern, it is impossible to generalize and often very difficult to measure.

Sometimes it is quite clear why variations in mortality patterns occur, but generally the relative impact of variables like age and sex composition, proportions and types of vulnerable groups, levels of mortality among vulnerable groups early in an emergency, and other factors is unknown. However, one reason for the variation in the speed of the mortality reduction is obviously the promptness of assistance efforts. How promptly assistance is provided is a function of many factors, including awareness of the situation, political decisions about whether or not to assist, ease of access to the displaced population, vulnerability of the population (because of the conditions and length of their flight), prior health status of the population, and reported mortality rates (among the most vulnerable). Although the general pattern is one of elevated mortality, followed by rapid declines with the arrival of assistance and a modicum of stable and safe living conditions, there is wide variation in the rapidity of mortality declines and improvement in the health and living conditions of refugee populations.

Reasons for Elevated Risk of Mortality in Complex Emergencies

It is too easy to overlook what are usually the initial direct causes of mortality and the underlying causes for all other mortality in a complex emergency. Violence from war and starvation due to famine kill many civilians directly and are often the reasons for flight which results in refugees and internally displaced persons. The root of most complex humanitarian emergencies is that governments and other combatants use violence and deprivation to seek solutions for political problems.

Violence is a major cause of mortality in complex emergencies. Armed conflicts, both civil wars and transnational conflicts, have increasingly targeted civilians. High numbers of civilian dead, human rights abuses, forced migration, and socioeconomic breakdown have been the result. In addition, injuries from war and landmines are common, particularly among IDPs (Toole and Waldman, 1997). For example, between April 1992 and January 1993 in Sarajevo, Bosnia, 57 percent of all mortality was due to war trauma (Centers for Disease Control and Prevention, 1993b).

Food scarcity, especially if experienced by a population with already elevated levels of malnutrition, can also lead to elevated mortality in complex emergencies. The same is true of lack of access to water. Malnutrition and dehydration can quickly increase mortality rates in a population, especially one in an already weakened state. Data on 42 different refugee populations between 1984 and 1988 showed a strong positive correlation between the acute protein energy malnutrition (PEM) prevalence and crude mortality rates. Populations with low PEM prevalence rates (less than 5 percent) had a low average monthly CDR (0.9 per 1,000 per month). But those populations with PEM prevalence rates of at least 40 percent had an average CDR of 37 per 1,000 (Person-Karell, 1989). In 1988-1989, among Somali refugees in Eastern Ethiopia, malnutrition prevalence and the CDR were also found to be positively correlated (Centers for Disease Control and Prevention, 1990).

There are many correlates to these causes of mortality, however. Displacement itself, because of the often-harsh conditions and long duration of flight, may be related to mortality. Deprivations during the ordeal, additional dangers encountered along the way, and lengthy disruption of ordinary life put great direct physical stress on people and also indirectly affect health status through physical and psychological stress that may increase their vulnerability to health problems and the levels of mortality. Thus refugees are often at the highest risk for mortality immediately after they arrive in a host country (Toole and Waldman, 1997).

Weakened populations are also more vulnerable to disease. During the early phases of an emergency, diarrheal diseases (e.g., cholera, dysentery), measles, acute respiratory infections, and malaria are the most common causes of death (Toole and Waldman, 1997). Among Rwandan refugees in Zaire in 1994, over 90 percent of deaths within the first month of the crisis were attributable to a severe cholera epidemic followed by a dysentery outbreak (Goma Epidemiology Group, 1995). Before 1990, measles epidemics were quite common in many refugee settings and led to large numbers of deaths in Somalia, Bangladesh, Sudan, and Ethiopia (Toole et al., 1989). However, since that time, immunization campaigns have reduced this threat somewhat. Malaria is often a problem in tropical areas such as Southeast Asia and sub-Saharan Africa. Other diseases that frequently attack refugees include acute respiratory infections, meningitis, hepatitis, tuberculosis, and HIV and other sexually transmitted diseases (Toole and Waldman, 1997).

A collapse of or lack of health services can also contribute to increased mortality. The breakdown of health services, particularly preventive services such as immunization and prenatal care, is often due to a combination of infrastructure collapse, economic failure, and lack of resources for public services. Personnel and equipment shortages exacerbated by the challenge of treating countless war casualties can overwhelm health systems (Toole and Waldman, 1997). Again, the multiplier effect of more than one element is evident in the particular risk of increases in communicable diseases if there is a combination of bad living conditions and a collapse of health services (Noji, 1997).

The Age Pattern of Mortality in Complex Emergencies

Another factor that contributes to elevated mortality in complex emergencies is the presence of vulnerable groups in the population. Those who are already at highest risk are going to be even more vulnerable during times of displacement and deprivation. The leading causes of death in refugee situations (with the exception of direct violence leading to death) are the same killers encountered in ordinary situations, and those who are most vulnerable in refugee situations are generally the same persons who are vulnerable under normal circumstances.

Risk clearly varies by age. For example, in 1980, in one camp in Somalia for Ethiopian refugees, daily mortality rates for those younger than five during the emergency phase were 28 per 10,000, much higher than for adults (Toole and Waldman, 1988). In 1985, under-age-five mortality rates among Ethiopian and Eritrean refugees in some Sudanese camps were one and a half times the CMR (Toole and Waldman, 1988). And among displaced Iraqis on the Turkey-Iraq border in March to May 1991, children under five years of age made up over 63 percent of all deaths (Centers for Disease Control and Prevention, 1991). Again, this pattern of mortality is quite similar to the normal circumstances in many developing countries.

In more developed regions, however, the elderly are often at greater risk. For example, in Sarajevo, Bosnia, from December 1993 to March 1994, elderly residents and refugees were found to be a particularly vulnerable group to malnutrition and its effects. The percentage of adults 60 years and older who were undernourished during these four months was over 15 percent, which was over three times the percentage of undernourishment in the rest of the population (Watson et al., 1995).

A more sophisticated question, however, is to ask how does the shape of the mortality pattern in complex emergencies differ from the underlying mortality pattern pre-existing in a stable population. By comparing age-specific mortality rates in emergencies to those occurring in a hypothetical stable population with a similar life expectancy, researchers can determine how the relative mortality risk differs for different age groups during an emergency. Few data on age-specific mortality in emergencies exist, but it is possible to analyze the data that can be found. We have analyzed data for three different emergencies by comparing them to age-specific mortality rates from relevant Coale-Demeny West life tables. In each of these graphs, one can observe the typical “J-shaped” mortality pattern of the Coale-Demeny curve, with the highest death rates occurring in the youngest and oldest ages (Coale et al., 1983).

In Figure 1-3 , the age-specific mortality pattern for a Coale-Demeny “West” Level 12 life table with an average life expectancy of about 45.5 years ⁵ is compared to the mortality pattern among Rwandan refugees living in Katale Camp, Zaire, in the summer of 1994 (Davis, 1996). Although only data for broad age groups were collected in the camp (and therefore the curve is incomplete and not very smooth), the general pattern is similar to the Coale-Demeny curve. Mortality is extremely elevated in the youngest and oldest age groups, compared to the middle ages. Mortality at most ages appears to be significantly higher than for the stable population.

FIGURE 1-3

Age-specific death rates for Rwandan refugees in Katale Camp, Zaire, July 17-August 5, 1994 and for Coale-Demeny life table “West” level 12.

Figure 1-4 shows the relative risk of dying in the Rwandan refugee camp compared to the risk of dying for a stable population with a life expectancy of 45.5 years at birth (approximately the same as that for the population of Rwanda before the 1994 crisis). All age groups have an elevated risk of mortality, but some are enormously high. Children aged 1 to 4 are 6000 percent more likely to die in the refugee camp compared to the stable population. Adults aged 45-49 also had high risk of mortality; they were about 3000 percent more likely to die compared to the same group in the stable population. It is known that cholera and shigella were the main causes of disease in this camp. It is not surprising that young children were quite vulnerable to these epidemics. The large risk of death for adults may also be related to the waves of disease (Davis, 1996).

FIGURE 1-4

Relative risk of mortality for Rwandan refugees in Katale Camp, Zaire, July 17-August 5, 1994 compared to risk of mortality for stable population based on Coale-Demeny life table “West” level 12.

Figure 1-5 and Figure 1-6 compare male age-specific mortality in a long-term Cambodian refugee camp in Thailand to a Coale-Demeny West Level 14 male life table with a life expectancy of about 49.5 years ⁶ (Elias et al., 1990). In Figure 1-5 , mortality appears to be only slightly elevated among the older and younger age groups. As Figure 1-6 shows, however, the relative risk of dying is over 80 percent greater for male refugees under one year of age compared to the same age group in the stable population. Mortality risk for the rest of the male refugee population is lower than for the stable population. These data demonstrate the stark difference between the crisis of the Rwandan refugee camp and the relatively reduced mortality risk found in a long-term stable refugee camp that was in existence for over 10 years in Thailand.

FIGURE 1-5

Age-specific death rates for male Cambodian refugees in Thailand, June 1987-May 1988 and for Coale-Demeny life table “West” level 14, males.

FIGURE 1-6

Relative risk of mortality for male Cambodian refugees in Thailand, June 1987-May 1988 compared to risk of mortality for stable population based on Coale-Demeny life table “West” level 14, male.

Figure 1-7 and Figure 1-8 compare female age-specific mortality in the same Cambodian refugee camp to a Coale-Demeny West Level 14 female life table with a life expectancy of about 52.5 years. In Figure 1-7 , again mortality appears to be only slightly elevated among the older and younger age groups. In Figure 1-8 , however, again the mortality risk for those under one year of age is elevated and almost 20 percent greater compared to the same age group in the stable population. Mortality risk for the rest of the female refugee population is also lower than for the stable population.

FIGURE 1-7

Age-specific death rates for female Cambodian refugees in Thailand, June 1987-May 1988 and for Coale-Demeny life table “West” level 14, females.

FIGURE 1-8

Relative risk of mortality for female Cambodian refugees in Thailand, June 1987-May 1988 compared to risk of mortality for stable population based on Coale-Demeny life table “West” level 14, female.

Within this volume, both Robinson et al. ( Chapter 3 ) and Heuveline ( Chapter 5 ) take a closer look at age-specific mortality rates in emergency settings. In the North Korea study by Robinson et al. (see Figure 3-1 ), mortality rates for their sample are somewhat similar to those from the Coale-Demeny “West” Level 4 life table, although elevated in the younger and older ages and reduced in the middle ages. Both of these curves show mortality levels much higher than those estimated by the 1993 North Korean census.

In Cambodia (see Figure 5-4 ), Heuveline found that deaths from natural causes followed the typical J-shaped curve for both men and women. Deaths from violent causes, however, had a completely different age and sex pattern. Men were much more likely to be killed in the younger age groups, from about age 5 to age 24 years, but has a lower probability of dying thereafter. Women, on the other hand, had a much lower probability of death due to violent causes, but the groups that were most vulnerable were very young women (under age 10) and women between the ages of 40 and 64.

How Can Excess Mortality Be Reduced in a Complex Emergency?

The basic mechanisms to reduce excess mortality in complex emergencies are widely agreed upon in the emergency assistance community and are based on many years of experience. The first requirements are to provide a stable situation that allows the displaced persons to stop moving; to be free from violence; and to have access to nutritious food, clean water, adequate sanitation and shelter, and basic health services, of which the important aspect is often measles immunizations (Sphere Project, 2000). Note that the requirements to reduce mortality include only one primarily medical intervention: measles vaccination. What refugees need is a reduction in the physical demands of flight and then access to the ordinary means of human survival. The provision of food, water, shelter, and sanitation allows people to live, to regain strength, and to be protected from common communicable diseases. Protection from measles is also important because children who are in weakened condition are quite vulnerable to measles.

In addition to this basic list of provisions, some situations may require attention to specific environmental hazards. A frequent example is malaria. If malaria is present, a refugee population is likely to be more than usually vulnerable. In this case, it is important to implement environmental prevention methods, such as removing any stagnant water that attracts mosquitoes, distributing bed nets to the refugees, and possibly spraying the area with mosquito repellants. Other environmental hazards, including other insects and rodents, may require different interventions.

A third component that is necessary in refugee situations is the early establishment of a surveillance system to monitor the health of the population and detect early signs of emerging problems. Although in some instances surveillance systems are already in existence (early warning mechanisms, for example), if displaced populations are in remote or undeveloped areas, it is usually necessary to implement a system immediately. Surveillance systems can utilize community health workers from the affected refugee population. This requires training, but it may have positive externalities such as community involvement and increased knowledge about health conditions among the population. ⁷

Finally, efforts to create new health services or supplement existing services can be a useful tool to address ordinary health issues as they arise in refugee situations, such as accidents, pregnancy, or other issues. In addition to basic survival needs, providing primary care services, addressing specific environmental risks (e.g., malaria), and establishing a working surveillance system should bring mortality rates back to base levels relatively quickly.

Knowing what to do in an emergency, however, does not minimize the difficulty of achieving these goals under field conditions. At times, the challenges of logistics are enormous, sometimes exacerbated by combatants' attempts to disrupt assistance. What is clear from a review of the literature is the confidence, based on repeated experiences, that the international assistance community knows how to reduce mortality to preflight baseline levels and maintain these levels (Waldman and Martone, 1999). Persistent excess mortality in a complex emergency is not the result of ignorance about effective procedures, but usually the result of extraordinary difficulties in the logistics of access to the refugees, interference of combatants, or the failure of the international community to provide sufficient resources.

The Shifting Geopolitical Map

The end of the Cold War marked the beginning of a new geopolitical context for complex emergencies. In some cases, the withdrawal of U.S. or Soviet support has led to a destabilization of developing country governments. In addition, many conflicts that were previously exacerbated by the involvement of the United States and the Soviet Union are now regional or internal conflicts. On the other hand, however, there are no longer two distinct ideological camps that create barriers to resolving conflicts or prevent assistance interventions for fear of reprisal.

In the past, intervention on behalf of forced migrants was often due to ideological considerations, rather than simply humanitarian concern. The Eastern Bloc countries were not part of the international refugee regime, but produced many refugees of their own who were given automatic asylum in the West. Now these countries are not necessarily producing refugees (with the exception of the Balkan region), but they are hosting them. And migrants who do leave Eastern Europe and the former Soviet Union are much less likely to gain asylum in Western Europe or the United States (Rogers and Copeland, 1993).

Along with this change, there is a continuing transformation of the concept of sovereignty. There appears to be a more rapid and more controversial change process than occurred in the past. Many see an erosion of the concept, implying a negative shift in global political structure and practice. Others see greater transparency and accountability demanded of states and thus a positive development for the human, civic, and social rights of people. Globalization and the integration of economies, the media, the environment, and human rights have increasingly encroached upon traditional understandings of sovereignty in today's world. Many of these forces, in turn, have expanded the concept of security to include non-military issues. One consequence of this expanded notion of security is that refugee flows are viewed as a security threat (Abiri, 2000; Wæever et al., 1993). Some national governments feel so threatened by these flows that they try to prevent refugees from crossing into their territory or they may force refugees to repatriate (Rogers and Copeland, 1993). Thus it has become more difficult to cross a border and become an official refugee; the number of internally displaced persons (IDPs) has increased at least partly due to policies such as these (see Figure 1-1 ).

In addition, legal barriers to refugee flows have increased, mainly in the form of restrictions on immigration and citizenship (Kushner and Knox, 1999). The increase in the number of IDPs makes the job of international organizations and nongovernmental organizations (NGOs) more difficult, since IDPs are often outside of their reach and governments and other factions may create obstacles to humanitarian intervention (Newland, 1999; Cohen, 1998). Therefore NGOs must be even more innovative in their efforts to protect and assist these groups and lower morbidity and mortality.

As noted, the human rights movement is one force that has been chipping away at the notion of sovereignty. Although the United Nations continues to uphold sovereignty in most instances, the human rights movement, along with other social, economic, and political interests, has pushed the world community to act in ways that violate the traditional notion of sovereignty in recent years: Iraq in 1991, Somalia in 1992, and Bosnia in 1992-96 (Jean, 1993). This increase in the use of collective action for the enforcement of human rights still retains a political nature, however, and is not applied universally and impartially. Instead of the East-West conflict of the Cold War era, interventions increasingly appear to have a North-South dynamic, with Northern developed countries intervening into regional and internal conflicts in the broadly-defined “South” (Rogers and Copeland, 1993).

The increase in collective action also raises the potential for greater collaboration between human rights groups, the military, and NGOs. It forces organizations like the International Committee for the Red Cross, humanitarian NGOs, and even UN agencies to reevaluate their position and often work in areas that are not safe or well protected by the military of a sovereign state or acting under the mandate of accepted international law. This has made the NGOs' job of assistance and protection even more dangerous and difficult and in turn can lead to increases in mortality among refugees and relief workers.

Sometimes, however, the changing political context can mean that NGOs have “unprecedented access” to refugees and IDPs. When governments are weak and therefore unable to limit access to populations within their borders, then major powers or coalitions of states are able to intervene rather easily. And because of their presence, ability, and resources, NGOs are the natural choice for humanitarian intervention (Stein, 2000).

In this context, the United Nations High Commissioner for Refugees, Sadako Ogata, perhaps ought to be singled out for using her office to call attention to IDPs. While certainly not alone in identifying IDPs as requiring international attention, her efforts lent authority and legitimacy to this issue, requiring at least an acknowledgement by states of this issue and its importance.

A Changing Epidemiological Context

Complex emergencies have generally been operationally defined as situations of war or civil strife, food insecurity, and/or population displacement that result in an excess mortality rate of more than 1 death per 10,000 population per day. However, many new emergencies, such as Bosnia and Kosovo, which are occurring in more developed regions of the world, do not fit this definition. The epidemiological context in developed countries is different from traditional refugee settings, such as sub-Saharan Africa and Southeast Asia. The populations are generally healthier and better nourished. Often chronic diseases, rather than communicable diseases, are an important part of the morbidity profile in developed countries. This is one reason why it is important to think about including measures of morbidity as well as mortality when assessing new emergencies (Waldman and Martone, 1999).

Health conditions are also changing; no longer are malnutrition and communicable diseases always the most pervasive threat during an emergency. Although these continue to have major impacts in many complex emergencies, physical trauma, psychosocial problems, and chronic illnesses are new issues that need attention. Measuring only mortality during an emergency says nothing about sequelae of a complex emergency that may have profound effects on the population. The psychosocial effects of trauma and disability resulting from injuries suffered during the emergency are two examples of indicators that may signify a severe emergency, even if mortality was low. For example, mortality levels in the Bosnian and Kosovar cases were lower than generally experienced in emergencies in developing countries (Waldman and Martone, 1999). Yet judgments about the severity of emergencies based on the single criterion of the number of deaths miss the suffering and human tragedy of facts like the tactical use of rape as a weapon of war, or, as in Sierra Leone, the practice of intentional mutilation that did not always result in death.

In some of today's complex emergencies, morbidity may be a better indicator of population health than mortality, because it may be easier to react to a broad range of issues as they appear, including health problems that may not be directly related to mortality, such as psychosocial issues. Measuring morbidity might also help to change the general assumption among some relief workers that their objectives are basic subsistence, followed by mortality reduction, without any other goals. Although basic needs and mortality reduction should be addressed, in some cases, morbidity maybe an important indicator (Waldman and Martone, 1999). Such a change, however, carries the danger of overlooking the fact that mortality rates are tragically higher in developing country complex emergencies compared to those in developed countries. Discounting mortality rates could be used as a rationale for disproportionate expenditures in developed country crises. The ethical and policy implications are not simply solved, but need further reflection and serious discussion.

There is already a growing appreciation among the assistance and medical communities of the importance of morbidity in addition to mortality as a measure of severity and progress in emergency situations. Aid workers are now beginning to focus on care and counseling to war victims, particularly those who have been victims of rape. In addition, the issue of reproductive health has received increased attention. Finally, the experiences of refugees in the Balkans have introduced greater sensitivity to the issue of chronic illness among refugees and the problems of responding to the chronically ill in a complex emergency. This expanded mission is based on not only preventing mortality, but also “protecting life with dignity” (Waldman and Martone, 1999:1484).

New Actors and New Roles for Older Actors

Another recent development is the emergence of new actors and the creation of new roles for those responding to refugee flows and complex emergencies. The United Nations has emerged as an important actor in initiatives involving military assets in the 1990s (after years of general inactivity during the Cold War). Interventions in Angola, El Salvador, Cambodia, Bosnia, Somalia, Iraq, and Indonesia have demonstrated that the UN is now not only focusing on traditional peacekeeping, but also on “the restoration of law and order and the protection of humanitarian aid operations” (Jean, 1993; Newland, 1999). This means that the UN is intrinsically involved in protecting NGOs, and often in a de facto position of non-neutrality. Therefore, the very presence of the UN can draw severe criticism and even fire from opposing groups.

Other new roles include the presence of the military in the delivery of humanitarian assistance and a greater role for external military forces in peacekeeping, refugee protection, and operations under an international authority. The logistical capacity of many military units is unparalleled. This was recognized most notably in the case of Kurdish refugees fleeing from northern Iraq in 1991 (Centers for Disease Control and Prevention, 1991). In other situations, the ability to deliver materials efficiently and rapidly has meant that the military was called on to deliver assistance, despite the typically high cost of the military. Furthermore, there is increasing interaction between military and civilian aid workers in situations where each has more or less different roles but some overlap and a need to coordinate. Coordination is required between military and both the NGOs and international (multilateral) organizations (IOs), such as the UN High Commissioner for Refugees.

Coordination between NGOs and IOs has been and continues to be a nettlesome issue. Recently, in places like Bosnia, the multiplication of NGOs and their impact on the local economy and labor force (especially the professional labor force) has raised deep concerns. It is often the case that aid workers from more developed countries flood the local labor market and local specialists are not used to their full potential. This may mean that relief is more costly than it needs to be and also violates the current development paradigm of building local capacity.

In other cases, the targeting of aid workers by military or paramilitary groups raises the need for protection for NGOs in many locations. Guerrilla movements and their increasingly dangerous tactics—often including a lack of respect for human rights, humanitarian principles, and the Red Cross and NGOs—have made it impossible for NGOs to operate on their own without protection (Jean, 1993). This makes it even more difficult to provide high-quality relief aid. In this new era of “disengagement and privatization” by the world's governments, NGOs are encountering new responsibilities and risks. They must take control of emergency situations, because they are often the only organizations who are on the ground, but they are apt to be ill-equipped to contend with a dangerous conflict setting (Stein, 2000).

Finally, situations in which a choice must be made between two goods (or evils) requires some sort of coordinated response. For example, in the Rwandan case, Hutu forces controlled camps in Zaire, which raised the possibility that assistance might go to military forces who had committed genocide, controlled the camps, and planned an armed return to Rwanda. However, withholding aid meant that civilians in camps controlled by the military forces would suffer and perhaps die. The question of whether to give or withhold aid was a difficult one, compounded by the fact that there were many different NGOs assisting in this area (Goma Epidemiology Group, 1995).

While issues like these have no simple answer, often there is greater need of field coordination and coordinated response. If there are difficulties in coordination, it can have devastating consequences for refugees and IDPs, and even for aid workers themselves.

Increasing Attention to the Quality of Relief

Another change in the context of complex emergencies has to do with setting standards for assistance and protection. International organizations and NGOs have been working to create and implement regulations for relief aid, especially under the Sphere Project. ⁸ Many NGOs have joined this project, which focuses on setting minimum standards for aid, training workers to implement these standards, evaluating assistance programs, and creating accountability. These new trends are helping to ensure that the level and quality of assistance provided in emergency settings is monitored (International Federation of the Red Cross and Red Crescent Societies, 1998). And standards are critical in continued work to reduce morbidity and mortality in crisis settings.

These steps are crucial in the new climate of reduced foreign aid funding. Emergency aid is still high—almost three times its 1990 level— but within a context of an overall decrease in development aid, crises continue to flourish (International Federation of the Red Cross and Red Crescent Societies, 1998). Furthermore, NGOs are under continuous pressure to prove that their funds are being put to work in an efficient and effective manner to save lives.

A Growing Appreciation of Information Needs

The need for further research and information on complex emergencies is now becoming quite clear to many NGOs, international agencies, states, donors, and scholars. Several universities around the world have established centers for research and training on how to deal with crisis situations. NGOs are forming partnerships with these centers to create standards, evaluate their own work, and learn new ways to implement relief efforts more effectively.

Much is already known about mortality in complex emergencies, but that knowledge is not complete. Much remains unknown about complex emergencies and issues surrounding appropriate responses, including ethical issues, management of specific diseases, and understanding how to treat specific populations. Reproductive health and mental health are two of the most important areas in need of further study (Waldman and Martone, 1999). In order to improve understanding, response, and assistance to forced migrants, research and collaboration must continue. The case studies in this volume are an example of a step towards more and better knowledge of mortality in complex emergencies.

Uniform Protocols for Data Collection

Relief workers, particularly medical personnel, seek the baseline information needed to respond to the most pressing health problems and develop monitoring systems. However, in the chaotic situation of a complex emergency, these systems are often poorly coordinated and sometimes even duplicate information. Many times, there is no general agreement or protocol on what data to collect or the appropriate methods to follow to ensure quality, interpretation, and comparability in order to assess the severity of problems and to provide markers for assessing progress over time. Field personnel need better systems of data collection to generate the information they need to plan, even in rudimentary ways, their response to the specific problems of a complex emergency. It is not that data collection as it is currently done is without merit. Much knowledge of mortality in complex emergencies results from such data collection by the medical community. However, questions of quality and comparability retard efforts to accumulate a body of knowledge that would facilitate sophisticated analysis of the determinants and pace of mortality change under stressful situations.

Denominators

The estimation of a population at risk in the construction of any demographic rate seems deceptively simple, but unfortunately it is often wrong and/or the result of compromise. In emergency situations, the basic estimation of the total number of refugees, which is needed to construct even a crude mortality rate, is elusive. The difficult conditions of emergency situations can make producing even rudimentary estimates an extreme challenge (see appendix). In addition, there are several specific reasons for population overestimation in crises. The leaders of displaced persons may try to hide those who are not legitimate refugees (those who have been involved in war crimes or military operations) in with the rest of the displaced population and thus inflate the numbers. Refugees may also try to register themselves more than once in order to gain more food rations. Out-migrations and deaths may also be underreported for the same reasons. When refugees are located within a host country community, local residents may register as refugees in order to obtain food and medical aid. Finally, refugee events are quite fluid and change rapidly.

On the other hand, refugee and displaced populations may be underestimated for a variety of reasons. Refugees who are self-settled among local populations may be difficult to count because they are hidden or continue to be on the move. If relief workers are not permitted access to the populations, then they are likely to misestimate their numbers. Those who are sick, impoverished, or malnourished may be hidden or cut off from the rest of the group and therefore not counted.

Thus, estimates of the same events taken from different sources often vary greatly. Perhaps the most familiar example of this is the different population estimates of the Rwandan refugees in Goma, Zaire, in 1994. Estimates from different agencies and NGOs ranged from 500,000 to 800,000, making it impossible to determine the mortality rate with any accuracy (Goma Epidemiology Group, 1995). In many situations, therefore, even if there is confidence in the estimated number of deaths, it is not a foregone conclusion that one can estimate the mortality rate with any confidence in the result (see also the case studies on Afghanistan and North Korea in the appendix).

Composition of Denominators

Even more demanding than estimating the total population is obtaining information on the composition of a given population. Of special interest in the study of mortality is age composition because of the vulnerability of children less than five years of age in developing countries. This age group is vulnerable even under normal circumstances, but much more so in situations of conflict, violence, and displacement (Davis, 1996). The age composition of a refugee population can have very important effects on the crude mortality rate. A population with a higher proportion of young children and elderly (like many developing country populations) will probably have a higher crude mortality rate than a population with a middle-aged distribution, because children under five years of age will probably experience higher mortality rates. Whether mortality is “excess” or not and why mortality is “elevated” are both a function of a population's age composition. This is a problem in many emergencies because only crude mortality rates are collected and therefore nothing is known about age- and sex-specific mortality rates. Even when age-specific mortality rates are known, they are generally only broken down into two categories: children under five years of age and others, which does not permit careful analysis.

Thus, age composition may explain some “excess” mortality. However, as noted above, if large proportions of children under the age of five, who may be over-represented in the refugee population, die in the early stages of a complex emergency, then the converse of excess mortality may occur. It is possible that the remaining population might appear to have lower than normal mortality because of the age composition of the surviving population. Heavy loss of vulnerable populations in an acute phase of an emergency, followed by the availability of assistance (including some assistance elements, such as vaccinations, that may not normally be available), may result in mortality levels for survivors that are significantly lower than those in the pre-emergency situation. There are various scenarios about the effects of early deaths on different groups within various populations that may affect subsequent mortality patterns. What is lacking is systematic data on these situations and analysis of mortality patterns within different populations.

In short, mortality rates in all emergencies should be standardized for age and sex, which requires some ability to decompose the population by these characteristics. If this is not done, then reliance on crude mortality rates as the major indicator of the severity of a complex emergency can lead to incorrect conclusions.

Collecting Mortality Data

There are many ways to collect mortality data in refugee settings, including burial site surveillance, collecting information from hospital and burial records, community-based reporting systems, and population surveys. However, none of these methods is flawless. Some of the reasons why data may be inaccurate are:

• Poorly representative population sample surveys;
• Failure of families to report all deaths for fear of losing food ration entitlements;
• Inaccurate estimates of affected populations for the purpose of calculating mortality rates; and
• Lack of standard reporting procedures (Toole and Waldman, 1997: 287).

Mortality rates are often underestimated because of deaths being undercounted and populations being overestimated. Secure and well-organized refugee camps seem to have generally produced the best estimates, while it is very difficult to get good mortality data on scattered populations and internally displaced persons. Mortality may be skewed in one direction or another because those with the highest risk of death are drawn to camps where there is food and medical attention or because those with the highest risk of death are in areas with the least access to the relief aid (Toole and Waldman, 1997). It is very difficult to compare mortality survey results from different settings because of the huge variation in methods. In Somalia between 1991 and 1993, 23 field surveys were found to have extreme differences in populations, sampling methods, units of analysis, computation of rates, and analysis techniques (Boss et al., 1994). However, it is hoped that efforts like the Sphere Project will increase the comparability between data from different settings.

Sampling

Sampling is the process whereby researchers determine a subset of the population under study from which to collect data that will hopefully be representative of the entire population. If the sample is properly drawn, then one should be able to make inferences about a population based on the characteristics of a sample. Although sampling is already a challenging enterprise under normal circumstances, in complex emergencies and refugee settings, it becomes even more difficult. In addition to “normal” issues that may bias the sample, because the total population is often unknown and unreachable, it is very difficult to obtain a representative and unbiased sample in an emergency setting.

Again, the major bias of current knowledge of demographic processes among refugee and internally displaced populations is the heavy reliance on information gathered from populations in camps. This is because of the relative ease of sampling and collecting data in a camp setting, where the sampling frame, or total population, is known, or area samples of a confined population are used. However, some scholars have argued that over 60 percent of Africa's refugees do not reside in camps; they live among the population in host countries (Harrell-Bond, 1994; Van Damme, 1995). What is known about refugee mortality may not hold true for noncamp populations. The reality is that the potential differences between these two populations are unknown because most information comes from camp situations where refugees are collectively aided by relief and protection agencies.

Furthermore, because estimates of the total size of a refugee population are so difficult to obtain, any attempts to sample from this more or less unknown universe become problematic. Sampling can move from a concern with population parameters to sampling geographically or spatially. In camp settings, such approaches have been implemented by dividing space into coordinated blocks and collecting data within specific blocks or sampling areas (Médecins Sans Frontières, 1997). In non-camp settings, such techniques are less useful unless one has knowledge of the spatial distribution of refugees among the host population. Therefore, other nonrandom sampling methods, such as snowball sampling, where one finds one refugee who then identifies other refugees to be included in the sample, must be used. However, these types of sampling techniques often mean that some refugees and demographic events, like deaths, may be missed.

Recall

The effect of the experience of a complex emergency on people's ability to recall events, and whether it is more of an issue than in normal situations, is unknown. Depending on specific cultural beliefs about death, psychosocial trauma, and other issues, it may be quite difficult to get an accurate estimate of mortality based on a population survey in an emergency setting. The impact of recall on monitoring and surveillance is not trivial because it is important in trying to develop baseline parameters.

Issues of data quality, interpretation, and methodology are not limited to those mentioned above. These are examples of issues that come up again and again in discussions of mortality patterns in complex emergencies. In many published papers, there is only a brief allusion in the form of a caveat for interpretation of data. Progress in understanding levels, trends, patterns, determinants, and consequences of mortality in complex emergencies requires attention to these technical issues from demographers, epidemiologists, and statisticians. Although the issues may seem sterile and esoteric, they have a large impact on what is known and consequently how relief workers are likely to respond to crisis situations.

AFGHANISTAN

Afghanistan has suffered cyclical conflict, displacement, massacres, food insecurity, epidemics, collapsed health services, and earthquakes since the 1970s. During this period, virtually no international aid organizations have been able to observe conditions in much of the country, although groups like International Medical Corps (IMC), CARE, and Save the Children have had periodic access to Kabul and the eastern districts. Afghanistan's reported population of 24.8 million is therefore very difficult to confirm, and in any case, millions of Afghan citizens continue to live in Iran or in refugee camps in Pakistan, where many of them fled during the conflict with the Soviet Union that began in 1979.

BOSNIA-HERZEGOVINA

In spring 1992, Serbian forces attacked Sarajevo, and thus began a war for Bosnia-Herzegovina, which had a population of approximately 4.5 million. During the war, roughly 3 million people became refugees; estimates of internally displaced persons inside Bosnia were around 1.2 million. The Serbs laid siege to Sarajevo, cut it off from outside contact, and began bombing and sniping at civilians in 1993.

NORTH KOREA

North Korea (the Democratic People's Republic of Korea) has been isolated since the fall of the Soviet Union and therefore very vulnerable during times of crisis. Between 1992 and 1995, government food ration distributions were drastically curtailed to citizens in the northeastern provinces. In 1995, after 23 inches of rain fell during 10 days in July and August, North Korea declared a disaster and appealed for international food aid while it repaired its damaged agriculture and infrastructure. This was an unusual shift for the government, which had previously resisted admission of need or failure. A year later, a drought hit the country, leading to an even greater need for foreign aid.

The peak of North Korea's famine appears to have been in late 1996 and early 1997, and international food aid grew during those years, peaking in 1998. It appeared to save large numbers of lives. International aid, including over a million tons of food from China, and several million tons of food from the World Food Program and the United States, permitted new observers to enter North Korea for the first time in decades.

RWANDA

After years of simmering tensions between Hutu and Tutsi ethnic groups, Rwanda erupted in the early 1990s, when civil conflict flared after Tutsi army incursions from Uganda, leading to the displacement of 900,000 people due to the 1993 fighting. In 1994, the worst genocide in recent times took place, followed by retribution killings of civilians, by excess mortality in refugee camps related to poor health, and ongoing battles with internally displaced persons inside Rwanda. The largest share of excess mortality, however, was due to the systematic campaign of ethnic cleansing by the ruling Hutu government prior to April 1994.

SIERRA LEONE

Civil conflict began in Sierra Leone in 1991 and has been heavily influenced by spillover from the ongoing conflict in Liberia. In general, the war has pitted the democratic government against Revolutionary Unity Front (RUF) rebels, backed by Charles Taylor's militia in Liberia. The course of recent events has been greatly determined by the military intervention of West African peacekeeping troops (ECOMOG). However, despite a supposed peace agreement that the elected government of Sierra Leone and rebels signed in 1999, fighting continues and the situation has not improved.

REFERENCES

• Andersson, N., C. Palha da Sousa, and S. Paredes1995Social cost of land mines in four countries: Afghanistan, Bosnia, Cambodia and Mozambique. British Medical Journal311: 718-721. [PMC free article: PMC2550718] [PubMed: 7549685]
• Bennett, J.1999Personal communication.
• Borden, A., and R. Caplan1996The former Yugoslavia: The war and the peace process. P. 203 in SIPRI Yearbook 1996.Oxford, UK: Oxford University Press.
• Boss, L., E.W. Brink, and T. Dondero1986Infant mortality and childhood nutritional status among Afghan refugees in Pakistan. International Journal of Epidemiology16(4): 556-560. [PubMed: 3440665]
• Brown, V., B. Reilley, M. Ferrir, and S. Manoncourt1996Cholera outbreak during massive influx of Rwandan returnees in November, 1996. Lancet349(9046): 212 . [PubMed: 9111572]
• Centers for Disease Control and Prevention1993Status of public health in Bosnia and Herzegovina. Morbidity and Mortality Weekly Report973: 979-982. [PubMed: 8259110]
• Choudhry, V., I. Fazal, G. Aram, M. Choudhry, L.S. Arya, and M.S. Torpeki1989Pattern of preventable diseases in Afghanistan: Suggestions to reduce the morbidity and mortality at IGICH. Indian Pediatrics26(7): 654-659. [PubMed: 2511141]
• Cohen, M.1996Afghanistan: Abandoned to Violence, Drugs, Hunger, Disease and Death.Maryland: Bread for the World.
• Cohen, R.1998Hearts Grown Brutal: Sagas of Sarajevo.New York: Random House.
• Coupland, R.1991Injuries from antipersonnel mines: The experience of the International Committee of the Red Cross. British Medical Journal303: 1509-1512. [PMC free article: PMC1671849] [PubMed: 1838290]
• Cuny, F.C.1996Refugees, displaced persons and the United Nations system. Pp. 187-211 in R. Caynes, and R. Williamson, US Foreign Policy and the United Nations System.New York: WW Norton.
• Des Forges, A.1999Leave None to Tell the Story: Genocide in Rwanda.New York: Human Rights Watch.
• Dowell, S., A. Toko, C. Sita, R. Piarroux, A. Duerr, and B. Woodruff1995Health and nutrition in centers for unaccompanied children: Experience from the 1994 Rwandan refugee crisis. Journal of the American Medical Association273: 1802-1806. [PubMed: 7769778]
• European Union, World Food Programme, and United Nations Children's Fund1998Nutrition Survey of the Democratic People's Republic of Korea.Geneva: European Union, World Food Programme, and UNICEF.
• Gessner, B.1994Mortality rates, causes of death, and health status among displaced and resident populations of Kabul, Afghanistan. Journal of the American Medical Association272: 382-385. [PubMed: 8028170]
• Goma Epidemiology Group1995Public health impact of Rwandan refugee crisis: What happened in Goma, Zaire in July 1994?Lancet345: 339-343. [PubMed: 7646638]
• Hodges, M., and R. Williams1998Registered infant and under-five deaths in Freetown, Sierra Leone from 1987-1991 and a comparison with 1969-1979. West African Journal of Medicine17(2): 95-98. [PubMed: 9715114]
• Horton, R.1999Croatia and Bosnia: The imprints of war. Lancet353(9170): 2139-2144. [PubMed: 10382711]
• International Federation of the Red Cross and Red Crescent Societies1999World Disasters Report 1999.Geneva: International Federation of Red Cross and Red Crescent Societies.
• Ivker, R.1990United Nations launches aid appeal as Afghanistan faces further disaster. Lancet351(9101): 508 .
• Kenney, G.1995The Bosnian calculation. New York Times Magazine. April 23: 42-43.
• Khalidi, N.A.1991Afghanistan: demographic consequences of war, 1978-1987. Central Asian Survey10: 101-126. [PubMed: 12317412]
• Lindenberg, M.1999Complex emergencies and NGOs: The example of CARE. Pp. 211-246 in J. Leaning, S.M. Briggs, and L.C. Chen, eds. Humanitarian Crises: The Medical and Public Health Response.Cambridge, MA: Harvard University Press.
• Maricevic, A., and M. Erceg1997War injuries to the extremities. Military Medicine162(12): 808-811. [PubMed: 9433087]
• McDiarmid, J.1995Deaths and injuries caused by landmines. Lancet346(8983): 1167 . [PubMed: 7475637]
• Médecins Sans Frontières1995Bosnia. In Populations in Danger 1995.Paris: Médecins Sans Frontières.
• Minear, L., J. Clark, R. Cohen, D. Gallagher, I. Guest, and T. Weiss1994Humanitarian Action in the Former Yugoslavia: The UN's Role 1991-1993. Occasional Paper #18. Providence, RI and Washington, DC: Humanitarianism and War Project and the Refugee Policy Group.
• Natsios, A.1999The Politics of Famine in North Korea. Washington, DC: U.S. Institute of Peace.
• Office of Foreign Disaster Assistance1999Sierra Leone Complex Emergency Fact Sheet 17.Washington, DC: U.S. Agency for International Development, Office of Foreign Disaster Assistance.
• Omaar, R., and A. de Waal1994Rwanda: Death, Despair and Defiance.London: Africa Rights.
• Pilsczek, F.1996Visiting doctor's perspective in Afghanistan. Lancet348(9041): 1566-1568. [PubMed: 8950886]
• Pretto, E.A., M. Begovic, and M. Begovic1994Emergency medical services during the siege of Sarajevo, Bosnia and Herzegovina: A preliminary report. Pre-hospital and Disaster Medicine 9(2 Suppl 1): S39-S45. [PubMed: 10155517]
• Prunier, G.1995The Rwanda Crisis: History of a Genocide.New York: Columbia University Press.
• Reno, W.1998Warlord Politics and African States.Boulder, Colorado: Lynne Rienner Publishers.
• Robinson, C., M. Lee, K. Hill, and G. Burnham1999Mortality in North Korean migrant households: A retrospective study. Lancet354: 291-295. [PubMed: 10440306]
• Sliwinski, M.1988The decimation of Afghanistan. Orbis 33(Winter): 39-56. [PubMed: 11617850]
• United Nations Office for the Coordination of Humanitarian Affairs. 1999 Mid-Term Review and Revision of United Nations Consolidated Inter-Agency Appeal for Sierra Leone.New York: Office for the Coordination of Humanitarian Affairs.
• 1999b Integrated Regional Information Network for West Africa.News bulletin.
• Wallensteen, P., and M. Sollenberg1998Armed conflicts and regional conflict complexes, 1989-97. Journal of Peace Research35(5): 621-634.
• Weiss, T.1999Military-Civilian Interactions: Intervening in Humanitarian Crises.Lanham, MD: Rowman and Littlefield Publishers, Inc.