4

Methods of Determining Mortality in the Mass Displacement and Return of Emergency-Affected Populations in Kosovo, 1998-1999

Brent Burkholder, Paul Spiegel, and Peter Salama

INTRODUCTION

In early 1998, long-standing tensions between the majority ethnic Albanian and minority Serbian populations in the Federal Republic of Yugoslavia (FRY) province of Kosovo broke out into open hostilities. Fighting intensified throughout the year between an armed resistance movement, the Kosovo Liberation Army (KLA), and local Serbian police and military forces, resulting in the destruction of multiple towns and the internal displacement of thousands of ethnic Albanians. Following the failure of diplomatic initiatives in early 1999, the North Atlantic Treaty Organization (NATO) began an organized bombing campaign in Kosovo on March 24, 1999. During the subsequent three months, additional ethnic violence forced almost 70 percent of the estimated 1.9 million Kosovar Albanians to leave their homes. Approximately 500,000 remained displaced within Kosovo and another 775,000 fled as refugees, including 444,600 to Albania, 244,500 to the Former Yugoslav Republic of Macedonia (FYROM), and 69,900 to the province of Montenegro (see Figure 4-1 ) (del Mundo and Wilkinson, 1999). After the signing of a peace accord on June 10, 1999, the flood of refugees reversed, and within three weeks more than 600,000 people had returned to Kosovo, one of the fastest repatriations in modern times (del Mundo and Wilkinson, 1999).

Tracking the mortality associated with the violence, movements, and conditions of displacement suffered by the emergency-affected populations during this crisis proved extremely problematic. During the height



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 86
FORCED MIGRATION & MORTALITY 4 Methods of Determining Mortality in the Mass Displacement and Return of Emergency-Affected Populations in Kosovo, 1998-1999 Brent Burkholder, Paul Spiegel, and Peter Salama INTRODUCTION In early 1998, long-standing tensions between the majority ethnic Albanian and minority Serbian populations in the Federal Republic of Yugoslavia (FRY) province of Kosovo broke out into open hostilities. Fighting intensified throughout the year between an armed resistance movement, the Kosovo Liberation Army (KLA), and local Serbian police and military forces, resulting in the destruction of multiple towns and the internal displacement of thousands of ethnic Albanians. Following the failure of diplomatic initiatives in early 1999, the North Atlantic Treaty Organization (NATO) began an organized bombing campaign in Kosovo on March 24, 1999. During the subsequent three months, additional ethnic violence forced almost 70 percent of the estimated 1.9 million Kosovar Albanians to leave their homes. Approximately 500,000 remained displaced within Kosovo and another 775,000 fled as refugees, including 444,600 to Albania, 244,500 to the Former Yugoslav Republic of Macedonia (FYROM), and 69,900 to the province of Montenegro (see Figure 4-1 ) (del Mundo and Wilkinson, 1999). After the signing of a peace accord on June 10, 1999, the flood of refugees reversed, and within three weeks more than 600,000 people had returned to Kosovo, one of the fastest repatriations in modern times (del Mundo and Wilkinson, 1999). Tracking the mortality associated with the violence, movements, and conditions of displacement suffered by the emergency-affected populations during this crisis proved extremely problematic. During the height

OCR for page 86
FORCED MIGRATION & MORTALITY FIGURE 4-1 Map of Kosovo region, June 15, 1999. Source: United States Agency for International Development (1999). of the violence from March to June 1999 there were no outside observersin Kosovo. Even within Albania and FYROM, the wide dispersal of the refugees into multiple small collective centers and host families limited access to large portions of this population. United Nations (UN) agencies and nongovernmental organizations (NGOs) did establish emergency

OCR for page 86
FORCED MIGRATION & MORTALITY surveillance systems in these two countries. However, logistical problems and the challenges of integrating these emergency operations into local, ongoing surveillance systems led to difficulties in obtaining comprehensive health data. Nevertheless, it is important to provide a record of mortality associated with this crisis and to highlight the challenges of collecting and analyzing data that confronted epidemiologists in the field. Our best estimates are that the absolute levels of mortality among the Kosovar refugees were not elevated as compared to previous complex emergencies. However, in addition to providing mortality results, we will focus on the methods required to both obtain and analyze this mortality data through surveillance systems in Albania and FYROM, as well as through a retrospective survey in Kosovo itself. In both scenarios, obtaining accurate population figures for denominator data or sample frame determination proved as problematic as obtaining information on deaths. For information on Albania and FYROM, Burkholder collected information from a variety of sources, principally from surveillance systems operational during the crisis. In Kosovo, Spiegel and Salama conducted a retrospective population-based health survey between September 8 and September 17, 1999. The survey was a collaborative effort between the International Rescue Committee (IRC), the Kosovo Institute of Public Health (IPH), the World Health Organization (WHO), and the Centers for Disease Control and Prevention (CDC). METHODS Albania and FYROM Surveillance Systems in Kosovar Refugee Camps Denominators: Obtaining Population Estimates The United Nations High Commissioner for Refugees (UNHCR) had comparatively complete figures for the relatively small number of Kosovar Albanians who fled into neighboring countries prior to the NATO bombing on March 24, 1999. However, from that point until the second week of April, refugee population estimates in both Albania and FYROM varied widely due to the chaotic influx of large masses of refugees crossing the borders daily. UNHCR was gradually able to begin a more systematic enumeration process in cooperation with local governments, the International Organization for Migration (IOM), and the local Red Cross societies. By March 27 in Albania, and April 15 in FYROM, UNHCR issued daily reports on the number of new refugees and esti-

OCR for page 86
FORCED MIGRATION & MORTALITY mates of total refugees residing in the host country. 1 We obtained detailed population demographic information for camp refugees in FYROM from the IOM registration data bank (International Organization for Migration, 1999). Mortality Surveillance In Albania, the government's Institute of Public Health, UN agencies, and NGOs collaborated on the Kosovar Refugee Information System (KRYSIS) which began collecting health data on a weekly basis beginning on April 16 (Instituti Shendetit Publik and World Health Organization, 1999; Ministry of Health of the Republic of Albania et al., 1999). The system attempted to incorporate reporting from local government health clinics in all 37 districts as well as from health posts established by NGOs in refugee camps. While timeliness and completeness varied from week to week, over 86 percent of all health units participated (Instituti Shendetit Publik and World Health Organization, 1999; Valenciano et al., 1999). Because an Epicentre rapid assessment in Kukes (Perea, 1999) and other reports early in the emergency found relatively low mortality rates, the surveillance system purposely focused on outpatient facilities and targeted communicable diseases (Coulombier, 1999). Although KRYSIS did report some deaths which occurred in camps or local clinics, most mortality was tracked through a hospital surveillance system which reported directly to the Ministry of Health (Albanian Ministry of Health Statistic Unit, 1999). Line listings of deaths were not available; however, there does not appear to be duplication between these two sources (Coulombier, 1999). Age and cause of death were provided only for reports from the KRYSIS. Due to the limitations listed above, refugee deaths that occurred in Albania are most likely under-reported to a greater extent than in FYROM. In FYROM, we obtained information on deaths that occurred in the camps during the early phase of the crisis through personal communication with field hospital staff (Alkan, 1999; Beckman, 1999). Later, deaths were reported through the Macedonian Refugee Communicable Disease Surveillance System, which began weekly standardized data collection from NGO camp health posts and field hospitals on April 26, 1999 (World Health Organization Regional Office for Europe et al., 1999). The Ministry of the Interior (MOI) provided an additional list of all refugee deaths that occurred in state hospitals between mid-March and June 30, 1999 (Former Yugoslav Republic of Macedonia Ministry of the Interior, 1999). 1   Additional population figures were obtained from daily refugee population reports produced by UNHCR in Tirana, Albania, and Skopje, FYROM.

OCR for page 86
FORCED MIGRATION & MORTALITY Although deaths occurring in the camps were also required to have FYROM official death certificates, a careful comparison of both camp and MOI lists did not reveal any duplications. Additional deaths were discovered through a listing of funerals provided by a local Albanian NGO (El Hillal, 1999). Kosovo Retrospective Mortality Survey2 To assess mortality rates, major causes of death and the risk factors for mortality during the civil war period of February 1998 to June 1999, we conducted a two-stage cluster survey in the 25 predominantly Albanian municipalities of Kosovo. Assuming a doubling of the baseline mortality rate and a design effect of 4, we calculated a total sample population of 6,440 individuals or 1,200 households. We chose villages or city neighborhoods as our sampling unit; however, determining a proper population sampling frame proved extremely difficult due to the lack of current demographic information. The most recent census data from 1991 was outdated by the crisis-induced population displacement over the last year. Nevertheless, we took this census as the best available baseline and updated the figures for each village and/ or neighborhood based upon population estimates made after July 1999 by UNHCR, the NATO intervention Kosovo force (KFOR), and food distribution lists from various NGOs. The sample was stratified to account for rural/urban and destroyed/non-destroyed status and 50 clusters (of 24 households each) were then assigned based on probability proportional to size. Households within each cluster were selected according to the standard method used by the Expanded Program on Immunization. One member of each household was interviewed and asked to provide a household census during the month of January 1998 and to recount information on the whereabouts of each individual, including any deaths that had occurred since that time. RESULTS Population Estimations: Kosovar Refugees in Albania and FYROM From 1998 through mid-March 1999, approximately 25,000 refugees crossed into Albania fleeing the ethnic violence in Kosovo. In the two 2   Detailed information on survey methodology and results can be found in Spiegel and Salama (2000). This survey includes deaths among all Albanian Kosovars: refugees, internally displaced, and those who never left their homes.

OCR for page 86
FORCED MIGRATION & MORTALITY FIGURE 4-2 Number of Kosovar refugees in Albania and the Former Yugoslav Republic of Macedonia (FYROM), March 24-June 30, 1999. weeks following the NATO bombing on March 24, 1999, an estimated 285,000 refugees entered the Kukes region of northern Albania. A slow, steady stream of additional refugees followed. The refugee population in Albania peaked in mid-June 1999 at a total of 444,500 (see Figure 4-2 ) (del Mundo and Wilkinson, 1999; United States Agency for International Development, 1999). Refugees were eventually dispersed throughout the country. Over two-thirds lived with host families, another 20 percent were housed in 22 camps, and the remainder were accommodated in multiple small collective centers (United States Agency for International Development, 1999). Repatriation occurred almost as quickly as the initial influx. Between June 15 and June 30, 1999, over 287,000 refugees returned home to Kosovo and the vast majority of those remaining returned within the next month. 3 Population flows into FYROM were slightly smaller in terms of numbers, but no less dramatic. Only 16,000 refugees had crossed the border before the NATO bombing, but in the subsequent two weeks another 101,000 Kosovar Albanians fled into FYROM. Over the next months, further immigration varied considerably depending on the level of violence in Kosovo and the ease of border crossing. Eventually, by June 15, 1999, over 245,000 refugees would reach the FYROM (World Health Organization Regional Office for Europe et al., 1999). At that time, 43 per- 3   Additional population figures were obtained from daily refugee population reports produced by UNHCR in Tirana, Albania and Skopje, FYROM.

OCR for page 86
FORCED MIGRATION & MORTALITY cent were living in 8 refugee camps and the remainder were housed with host families, primarily in the 5 regions near the border (Former Yugoslav Republic of Macedonia Ministry of the Interior, 1999). Almost 202,000 returned home by the end of June. Population fluctuations in FYROM were even more pronounced because of the UNHCR/IOM Humanitarian Evacuation Program. Between April 5 and June 25, 1999, 90,189 refugees in the FYROM were evacuated to third countries. 4 This movement led to huge population shifts, particularly in the two camps that were the primary source of evacuees. For example, during the last three weeks in May, 44,417 refugees left the Macedonian camps and 46,492 arrived. Demographic data are available only for the refugee camp population in FYROM. According to IOM registration data on May 27, 1999, the agegender distribution of the camp population was roughly similar to that of the Kosovar population found in the 1991 census (International Organization for Migration, 1999). MORTALITY Crude Mortality Rates Table 4-1 shows reported deaths and crude mortality rates (CMRs) from both surveillance systems in refugee camps in Albania and FYROM, and the retrospective survey. Of the 141 refugee deaths reported in Albania, the KRYSIS (Ministry of Health of the Republic of Albania et al., 1999) reported 34 and the Ministry of Health (Albanian Ministry of Health Statistic Unit, 1999) reported 107 from hospital surveillance. In FYROM, the camp surveillance system (World Health Organization Regional Office for Europe et al., 1999) detected 28 (16 percent) refugee deaths and the Ministry of Interior (Former Yugoslav Republic of Macedonia Ministry of the Interior, 1999) reported 143 (76 percent). Field hospital directors were aware of five additional deaths in the first week of the refugee influx in early April. The NGO funeral list (El Hillal, 1999) included 34 deaths, of which 11 were not found in any other sources. There was a difference in the Albanian and FYROM refugee CMRs (0.11 versus 0.14 and 0.24 versus 0.33, respectively), based on whether the midpoint or average population was used to calculate the denominator. However, the difference was not statistically significant in either case (zstatistic to compare rates). 4   Again, additional population figures were obtained from daily refugee population reports produced by UNHCR in Tirana, Albania and Skopje, FYROM.

OCR for page 86
FORCED MIGRATION & MORTALITY TABLE 4-1 Crude Mortality Rates (CMRs) in Kosovo, February 1998-June 1999, and among Kosovar Refugees in Albania and the Former Yugoslav Republic of Macedonia (FYROM), March-June 1999   Time Period Reported Deaths Midpoint Population CMR (Deaths/1,000/Month) Average Population CMR(Deaths/1,000/Month) Kosovo Baseline 1996 N/A N/A N/A N/A 0.31 Refugees in FYROM 3/20/99–6/30/99 187 232,900 0.24 170,063 0.33 Refugees in Albania a 4/1/99–6/30/99 141 432,267 0.11 334,778 0.14 Kosovo Survey b 2/1/98–6/30/99       Initial Population All causes 105 N/A N/A 8,553 0.72 War-related 67 N/A N/A 8,553 0.46 Sources: Ministry of Health of the Republic of Albania et al. (1999);Albanian Ministry of Health Statistic Unit (1999); World Health OrganizationRegional Office for Europe et al. (1999); Former Yugoslav Republicof Macedonia Ministry of the Interior (1999); and El Hillal (1999). a There was significant underreporting of Kosovo refugee deaths in Albania. b Displaced and nondisplaced.

OCR for page 86
FORCED MIGRATION & MORTALITY FIGURE 4-3 Crude mortality rate in Kosovo, January 1998-September 1999, and among Kosovar refugees in the Former Yugoslav Republic of Macedonia (FYROM), March-June 1999. The retrospective mortality survey in Kosovo that covered the complete 17-month period of the crisis found a much higher overall CMR (0.72 per 1,000 per month). Note that the data sources for the surveillance and the retrospective survey are not mutually exclusive. The respondents to the survey in Kosovo included refugees who had returned from FYROM (as well as other countries) and therefore may be reporting deaths that were also included in the surveillance system. Figure 4-3 shows that mortality during the Kosovo crisis peaked in April 1999. In FYROM, the death rate in the refugee camps was highest during April in the initial phase of the refugee exodus and then steadily declined. Even during April, the CMR only reached 0.56 per 1,000 per month (6.7 per 1,000 per year) in the refugee camps. The Kosovo data show a similar pattern but with a much higher peak CMR of 3.25 per 1,000 per month (39 per 1,000 per year). This major increase in mortality and smaller peaks throughout 1998 closely correspond to flare-ups in fighting. As stated below, the majority of the elevated mortality was due to warrelated trauma.

OCR for page 86
FORCED MIGRATION & MORTALITY FIGURE 4-4 Age-specific mortality of Kosovar Albanians in Kosovo, February 1, 1998-June 30, 1999 (N=105) and in the Former Yugoslav Republic of Macedonia (FYROM), March 20-June 30, 1999 (N=141). Age, Sex, and Cause-Specific Mortality Specific information on mortality in Albania was limited (Valenciano et al., 1999). Between April 16 and June 6, 1999, this surveillance system reported 34 deaths; 11 in children under 5 years old and 23 among those over 5 years old. Cause of death was clearly specified in only 17 cases. Among children under 5 years, death was attributed to acute respiratory infection in 3 (36 percent) cases; and among those over 5 years, 3 (13 percent) were also due to acute respiratory infection and 11 (48 percent) deaths were labeled as “cardiac.” Males accounted for 54 percent of the deaths in FYROM. Specific age data was available for only 141 (75 percent) of the 187 deaths (see Figure 4-4 ). The majority (66 percent) of deaths were in adults 60 years and over. Of the 19 deaths in children under 5 years, 15 occurred among neonates and were ascribed to either “prematurity” or “aspiration.” Over 55 percent of all deaths were attributed to either “natural causes” or chronic disease (see Figure 4-5 ). There were few infectious disease deaths. The “other” category included two deaths due to hypothermia, two to gunshots, and one due to a motor vehicle accident. Seventy-one (40 percent) deaths occurred among refugees who lived in camps and 112 (60 percent) occurred among host-family refugees (Ministry of Health of the Republic of Albania et al., 1999). There was no statistical significance in the CMR based on residence status (data not shown).

OCR for page 86
FORCED MIGRATION & MORTALITY FIGURE 4-5 Proportional mortality of Kosovar Albanians in Kosovo, February 1, 1998-June 30, 1999 (N=105) and in the Former Yugoslav Republic of Macedonia (FYROM), March 20-June 30, 1999 (N=187). In the Kosovo survey, males accounted for 67 percent of all reported deaths, a higher proportion than in FYROM. Unlike FYROM, a smaller percentage of deaths occurred among the 60 years and older age group (44 percent), while a much larger proportion occurred among the 15 to 59 year olds (51 percent). A smaller percentage of deaths under 5 years (5 percent) occurred in the Kosovo survey, possibly due to methodological problems in capturing infant deaths in a retrospective survey (see Figure 4-4 ). “War-related trauma,” which constituted the largest cause of death (63 percent), was defined as any death occurring directly or as a result of an injury sustained during the conflict (see Figure 4-5 ). These war-related traumas included summary and arbitrary killings with bullets, shrapnel or other munitions, or by the burning or collapsing of buildings, bridges,

OCR for page 86
FORCED MIGRATION & MORTALITY and other structures. In data not shown here, we found that 24 (36 percent) of the 67 deaths due to war-related trauma were among the elderly (60 years and older), and of this group, 92 percent were male. The elderly were 5.9 times (95 percent confidence interval: 3.6-6.6) more likely to die from this cause than any other age group. Among the non-war-related causes of death, chronic disease and “ natural” causes accounted for 70 percent of the deaths. As in FYROM, infectious diseases only accounted for a relatively small percentage of deaths. LIMITATIONS Albania and FYROM Surveillance Systems in Kosovar Refugee Camps The relative lack of mortality data from Albania precludes making any conclusions about the status of refugees in that country or drawing any comparisons to those in FYROM. The wide geographic disbursement of the refugees, their access to multiple health facilities, and the understandable focus of the surveillance system on outpatient clinics seem to have complicated the process of tracking mortality in this setting. Even in FYROM, where the refugees were more concentrated and had access to only a limited number of health centers and hospitals, it is difficult to gauge accurately the completeness of reporting. Deaths occurring during the early chaos of the influx may not have been reported. After that period, reporting for camp refugees should have been more complete since these refugees were either treated in camp field hospitals or were directly transported to one of a limited number of local tertiary care centers. However, only 28 (39 percent) of 71 camp refugee deaths were reported by the camp surveillance system; the remainder were reported months later through routine channels. Additionally, due to FYROM government restrictions on travel and access to health facilities, host family refugees may have been less likely to seek care. Underreporting of deaths may be likely among this group, particularly for those in rural areas. The local NGO funeral list was impossible to verify; however 11 of the 34 deaths they reported were not on the official MOI record. There were similar limitations in obtaining accurate denominator data. The rapid mass movement of the refugees at multiple border crossings, the refugees' wide geographic dispersal, and their distribution into multiple types of accommodations all presented challenges to obtaining accurate estimations of the number of people involved. Refugees arriving before early April routinely made their own arrangements for housing with host families. Only refugees who arrived later were sent to camps or, in the case of Albania, to camps or collective centers. Registra-

OCR for page 86
FORCED MIGRATION & MORTALITY tion in the confines of camps or collective centers was difficult due to the constant influx of new arrivals and, in FYROM, departures for third countries, but nevertheless proved easier than dealing with refugees in host families. Enumeration was particularly delayed in FYROM where host family refugees had to go through a multistage registration process before inclusion on the official Macedonian Red Cross list. Kosovo Retrospective Mortality Survey This survey has some of the same limitations that have been pointed out in previous retrospective mortality studies (Ascherio et al., 1992). The 17-month recall period is relatively long and could potentially introduce recall bias. Survivor bias can be present in any such surveys since households in which all members have died or remain refugees could not have been selected. Additionally, limitations related to verbal autopsies can make cause of death determinations problematic. Epidemiologists in Kosovo had additional constraints, particularly in estimating local population figures for sample determination. Inaccurate or outdated census data and lack of current population figures can introduce selection bias in any conflict situation, but these factors may have introduced particular challenges in this case due to the exceptionally high percent of household displacement. Furthermore, there may be inherent limitations to using cluster sampling for mortality surveys in general. DISCUSSION Albania and FYROM Surveillance Systems in Kosovar Refugee Camps While the validity of the data cannot be confirmed due to the possible exclusion of Kosovar Albanians from the surveillance system, the Kosovo Institute of Public Health reported an average CMR of 4.0 per 1,000 per year (0.33 per 1,000 per month) for 1989-1996 (Kosovo Institute of Public Health, 1997). Compared to this baseline, the CMR available from the KRYSIS and Albanian MOH for refugees in Albania is exceedingly low. These surveillance systems did not concentrate on mortality so the low rates most likely represent underreporting. The aggregated mortality data on refugees in FYROM appears to be more complete; however, the CMR is still quite low by any comparison to emergency standards. Certainly the mortality rate did not come close to the previously accepted complex emergency threshold of greater than 1 death per 10,000 per day, or 3 per 1,000 per month (Sphere Project Steering Committee for Humanitarian Response and Interaction, 1998). The

OCR for page 86
FORCED MIGRATION & MORTALITY average CMR for the three-month crisis was markedly similar to the prewar baseline and surpassed it only during the height of the crisis in April. Even allowing for substantial underreporting, it appears that the Kosovar Albanians were a generally healthy and relatively young population whose health status was maintained during their refuge in the FYROM. Kosovo Retrospective Mortality Survey The Kosovo survey did report an average CMR between February 1998 and June 1999 that was 2.3 times higher than the preconflict baseline. However, only during the month of April 1999 did the CMR cross the conventional threshold for emergencies. Nevertheless, contrary to most previous emergencies where children have been the most vulnerable (Toole, 1996), data here suggest that the elderly may have been most at risk for both war-related trauma and overall increased mortality. Questions and Challenges for Collecting Mortality Data in Future Humanitarian Emergencies Recent crises in Kosovo, Bosnia and elsewhere outside of Africa have led to a call to change the mortality definition of an emergency and to recognize the altered epidemiological profile that has been demonstrated in these circumstances (Waldman and Martone, 1999). The experience of collecting mortality data for the Kosovo crisis provided useful information but raised questions of how to respond to this call. Furthermore, it raises several methodological issues that need to be addressed if we are to improve our understanding and techniques of health surveillance. Should mortality still be considered the most sensitive indicator of an emergency-affected population's health status? The CMR in FYROM (and probably also in Albania) did not appear to be elevated for the refugees, but few humanitarian workers would deny that they faced a public health crisis. This issue raises the further question of whether emergency surveillance systems should focus primarily on morbidity and detecting epidemics in certain situations where mortality appears low. Similarly, in low CMR situations, what is the appropriate threshold to define significant excess mortality? Obviously, the previous standard of 1 death per 10,000 per day is not sensitive enough to detect major increases in mortality among populations with comparatively low preemergency rates. We need to consider whether a doubling of baseline rates may be an appropriate definition in these circumstances. How do we monitor mortality in non-camp settings? Most refugee deaths in Albania and FYROM occurred in hospitals, not in camps where

OCR for page 86
FORCED MIGRATION & MORTALITY emergency surveillance systems were operational. In FYROM the government required that mortality be reported through the routine surveillance system, which resulted in lengthy delays in obtaining potentially critical health data. Separate registration and reporting systems for refugees admitted to hospitals should be instituted but may be problematic in many situations. Agencies in charge of surveillance in emergency settings must be willing to commit additional resources to bolster routine systems and provide expedited reporting. How do we calculate denominators in situations of massive population flows, particularly when many refugees may be housed with local populations? When populations shift over short periods of time, should the midpoint or average population be used as a denominator in calculating CMR? Do we need to account for situations of rapid in- and outmigration where the total population remains the same, but the population turnover is high? Epidemiologists involved in emergencies have developed some techniques for cross-sectional population estimations, but we need to discuss with demographers how to further develop these tools and address calculating denominators. Cluster sampling, which was developed to measure immunization coverage, has routinely been used for retrospective mortality surveys. Can both the sampling and analysis of this methodology be refined to reflect the non-homogenous pattern of deaths in crisis situations? Beyond the questions inherent in estimating populations and determining clusters, does this method offer the precision needed to document relatively rare events such as deaths? Epidemiologists, demographers, and survey statisticians need to collaborate on answering these and other questions to improve practical methods of mortality surveillance in emergencies. REFERENCES Albanian Ministry of Health Statistic Unit 1999 Number of deaths among refugees in Albanian hospitals (4/1/99 to 6/3/99) . Alkan, M. 1999 Personal communication . Israeli Military Field Hospital, Brazda camp . Ascherio, A. , R. Chase , T. Cote , G. Dehaes , E. Hoskins , J. Laaouej , M. Passey , S. Qaderi , S. Shuqaidef , and M.C. Smith 1992 Effect of the Gulf War on infant and child mortality in Iraq . New England Journal of Medicine 327(13) : 931-936 . Beckman, H. 1999 Personal communication . German Red Cross Field Hospital, Brazda camp . Coulombier, D. 1999 Personal communication . Institut de Veille Sanitaire, November 22, 1999 . del Mundo, F. , and R. Wilkinson 1999 A race against time . Refugees 3(116) : 4-15 .

OCR for page 86
FORCED MIGRATION & MORTALITY El Hillal 1999 Request for funeral expenses submitted to UNHCR, June 1, 1999 . Former Yugoslav Republic of Macedonia Ministry of the Interior 1999 Line listing of refugee deaths, March-June, 1999 . Instituti Shendetit Publik and World Health Organization 1999 Surveillance system among Kosovar refugee population in Albania: Final report . World Health Organization/European Regional Office. International Organization for Migration 1999 IOM/UNHCR registration database accessed on May 27, 1999 . Kosovo Institute of Public Health 1997 Statistical Report, 1997 . Ministry of Health of the Republic of Albania, World Health Organization Humanitarian Mission, Tirana, and Institut de Veille Sanitaire 1999 Kosovar Refugee Information System (KRYSIS) weekly reports from April 16-June 6, 1999 . Perea W. 1999 Report on rapid needs assessment among Kosovar refugees hosted by Albanian families, and assessment of human rights violations committed in Kosovo . Paris : Epicentre . Sphere Project Steering Committee for Humanitarian Response and Interaction 1998 Sphere Project Report . Geneva, 1998 . Spiegel, P.B. , and P. Salama 2000 War and mortality in Kosovo, 1998-99: An epidemiological testimony . Lancet 355 : 2204-2209 . Toole, M.J. 1996 Vulnerability in emergency situations . Lancet 348(9031) : 840 . United States Agency for International Development 1999 USAID Fact Sheet #72, June 15, 1999 . [Online]. Available: http://www.info.usaid.gov [Accessed: June 17, 1999.] Valenciano, M. , A. Pinto , D. Coulombier , E. Hashorva , and M. Murthi 1999 Surveillance of communicable diseases among Kosovar refugees in Albania . Eurosurveillance, Sept 1999 . [Online]. Available: http://www.cese.org/eurosurv_eng.htm [Accessed: November 1, 1999]. Waldman, R. , and G. Martone 1999 Public health and complex emergencies: New issues, new conditions . American Journal of Public Health 89(10) : 1483-1485 . World Health Organization Regional Office for Europe , Office for Humanitarian Assistance of the FYROM , and the Health Information Network for Advanced Planning 1999 Health Bulletins from April 26-July 2, 1999 (Nos. 5-14) .