Mortality and Malnutrition Among Populations Living in South Darfur, Sudan:  Results of 3 Surveys, September 2004 FREE

In early 2003, 2 rebel movements (the Sudanese Liberation Army and the Justice and Equality Movement) launched an insurgency against the rule of Khartoum and the government of Sudan, and Janjaweed tribal militias have responded with decisive counteroffensives. Militia attacks have been blamed by refugees and internally displaced persons (IDPs) for indiscriminate killings, rape, abduction, cattle and property looting, and razing of villages. More than 1.5 million people (25% of the population of the region) are now scattered in 127 encampments in Darfur and 15 in neighboring Chad.

Humanitarian assistance arrived late, having been limited initially by severe government restrictions and low institutional donor interest; and also by insecurity, which is ongoing. Médecins Sans Frontières (MSF) began work in West Darfur in December 2003, and operations in South Darfur, including feeding centers and primary care clinics, opened in May 2004 and have been centered on 3 sites (Figure 1). Kass is a busy market town where 50 000 IDPs arrived mainly in November 2003 and now live both within the compounds of the 30 000 residents and in distinct encampments in school grounds and waste areas. Kalma, which in January 2004 was a small village on the main railway line running east of the regional capital Nyala, is the largest single IDP settlement in South Darfur and in August 2004 had 66 000 inhabitants; it has continued to grow. Muhajiria is in relatively inaccessible rebel-held territory east of Nyala and has had 2 major influxes of new arrivals, one in October 2003 and another in August 2004. The 3 surveys reported herein were part of site-specific assessments of health and nutritional status designed to inform the delivery of relief activities in the region.

We performed 2-stage household-based cluster surveys in 3 sites in South Darfur (Figure 2). In 2 of these (Kass and Muhajiria), residents were considered by aid agencies to be as vulnerable as IDPs; therefore, both groups were surveyed. Sectors were defined by the Sudanese authorities for local resident populations; for IDP camps, the sector boundaries used were those recognized by IDPs (populations) and their community leaders and were mainly distinct geographical landmarks. Sector-specific populations were estimated at each site using a combination of local government census (for residents), the reports of community leaders (for IDPs), health worker censuses, World Food Program data, and shelter counts.

Households, defined as groups of people eating together, were selected on the basis of the standard World Health Organization/Expanded Program of Immunization method.¹ For the first stage of sample selection, we allocated numbers of clusters to camp or town sectors proportionally to the population.² For the second stage, the first dwelling within each cluster was selected either by walking in a random direction from the center to the perimeter, counting the number of households (n) and randomly selecting a number between 1 and n, or, where dwellings were arranged on a grid, by randomly selecting a global positioning system coordinate. Subsequent households were selected by proximity.

At each site, general household status, crude and under 5-year mortality, and nutritional and vaccination status of children aged 6 to 59 months were assessed simultaneously within the same survey design. Within any 1 cluster, the general household questionnaire was administered to the first 7 households and the mortality questionnaire to the first 30 households. Anthropometric data were collected on all children residing within selected households and aged 6 to 59 months until data had been obtained on a total of 30 children. Sample size calculations were performed prior to the survey. Nine hundred households (30 clusters × 30) or 4500 people (assuming a mean household size of 5 people) would have been sufficient to estimate a crude mortality rate of 2.0 per 10 000 per day with a 95% confidence interval (CI) of 1.0 to 3.0, a recall period of 30 days, and a design effect (loss of variance due to intracluster homogeneity) of 2. Similarly, with an expected prevalence of malnutrition of 10%, 900 children would give a 95% CI of 7.5% to 13.3%. For access to items and services, 210 households give adequate precision, assuming a prevalence of 50% for the main outcomes.

Each survey team included a community health worker, a local person who spoke Arabic and English, and a member of the expatriate or Khartoum-based MSF staff who acted as supervisor. Training at each site was performed over 3 days by 2 authors (F.G. and F.S.) and included pilot testing of the questionnaires. Questionnaires were forward- and back-translated into Arabic before they were issued, and were delivered orally. The majority of questions required simple yes or no answers or a choice from a list of up to 7 items. We asked the help of neighbors to trace absentees and revisited empty households and households in which children were temporarily absent. Each cluster was completed in 1 working day.

Médecins Sans Frontières and Epicentre do not routinely subject individual standard rapid assessment surveys to formal ethical scrutiny; however, both organizations subscribe to the ethical principles outlined in the Declaration of Helsinki.³ In addition, the conduct of the study and the content of the questionnaires were reviewed and approved by the local Sudanese authorities (Humanitarian Aid Commission) and were described to IDP community leaders who agreed to and in many cases facilitated the survey. No incentives were offered to participants or officials. The interviewee was the most senior adult household member, who gave oral informed consent to participate in the study. For children aged 6 to 59 months, consent to anthropometric measurement was obtained from a parent or guardian. No names were obtained or recorded except when respondents agreed to the referral of malnourished children or sick individuals to the relevant MSF clinics.

Ages of children were not recalled reliably, and thus the target age range of 6 to 59 months was defined by a height of at least 65 cm and less than 110 cm. A standard United Nations Children's Fund (UNICEF) height board was used and children with a height of less than 85 cm were measured lying down. Weight was determined with a 25-kg Salter scale (UNICEF kit) that was calibrated daily. Acute malnutrition was said to be present in a child with a weight-for-height ratio z score of less than −2 compared with the reference population median or if there was pedal edema.

Measles vaccination cards were scarce and vaccination status was based on verbal report by the parents. Access to clean water was defined as the last collection of drinking water having come from a protected well, tap, water bladder, or other chlorinated container. There was a reluctance to show food distribution cards and ownership was assessed by verbal report. Respondents were asked whether they generally used a latrine for defecation. Soap was reported as present by the interviewee or observed by the survey team.

It is customary for mortality data to be collected using a well-defined date to mark the beginning of the recall period. For Kass, the celebration of the birth of the prophet Mohammed on April 30, 2004 (a national holiday in Sudan), was chosen and gave a recall period of 121 days. For Kalma and Muhajiria, a shorter recall period (30 days) was dictated by the priorities of the operational teams. Following a death in Sudan, the custom is to mourn the deceased for a period of 40 days. This period is punctuated by ceremonies of remembrance and the days following a death are reliably counted by the family of the deceased. For each death, we noted the age of the person who died and coded the reported cause as violence, diarrhea, or other (includes multiple other causes of death, such as pneumonia, measles, injuries, and tumors). For diarrhea, we used the standard definition of 3 or more loose stools per 24 hours. Interviewees were asked to report household size and the number of children younger than 5 years; this information was used to determine the denominator for mortality calculations.

We entered data into EpiData version 3.0 (EpiData Association, Odense, Denmark). Analyses were performed using EpiNut (EpiInfo version 6.04; Centers for Disease Control and Prevention, Atlanta, Ga) and Stata version 7.0 (StataCorp, College Station, Tex). Ninety-five percent CIs were calculated and adjusted for the design effect.

Surveys were performed from August 26-30, 2004, in Kass; September 2-6, 2004, in Kalma; and September 11-15, 2004, in Muhajiria. There were no recorded cases of refusal to take part in the survey. The main characteristics of the surveyed population are described in Table 1. Two hundred seventeen deaths were reported over the previous 121 days in Kass; and in Kalma and Muhajiria, there were 30 and 36 deaths over the 30 days prior to the survey. The crude mortality rates at all 3 sites were considerably higher than the 1 per 10 000 per day that is recognized internationally as defining an emergency situation^4- 5 and 4 to 6 times the expected rate in sub-Saharan populations (crude mortality rate, 0.5). In both Kass and Kalma, the under 5-year mortality rates exceeded the 2 per 10 000 per day used as the emergency benchmark.

Deaths from medical causes predominated in Kass and Kalma (80% and 90% of deaths, respectively). Diarrheal diseases were responsible for 25% and 47% of all deaths in these 2 sites, affecting mainly children younger than 5 years and adults older than 50 years. Violence was the major cause of death in Muhajiria (72%), with all but 1 of the 25 violent deaths in men. Most violent deaths had occurred during the fighting prior to displacement and were the major factor in determining the excess mortality. In Kass, 18% of deaths were due to violence, and violence was ongoing; in Kalma, 2 (7%) of 30 deaths were violent.

The prevalence of acute malnutrition was high, particularly in Kalma where nearly 24% of children younger than 5 years were affected (Table 2). Reported measles vaccination coverage ranged from 46% to 70%. Between 25% and 71% of people reported regularly using latrines; only in Kalma did a high proportion have access to clean drinking water; soap was scarce. In Muhajiria, where 13 000 displaced people had arrived in the previous month, there had as yet been no distribution of food or essential household items, such as plastic sheeting.

The 3 surveys, all performed under difficult and insecure field conditions, represent 190 000 people, including 47% of the 289 000 IDPs estimated to be residents in South Darfur in September 2004.⁶ Despite relief efforts in Kass and Kalma, both crude and under 5-year mortality rates were well above the benchmark emergency thresholds and a high proportion of deaths were reported to be from diarrhea. In Muhajiria, adult mortality was high and the majority of the deaths were violent. Acute childhood malnutrition at all 3 sites was equal to or more than the 10% defined as “serious”.⁵ A large proportion of households had access to food and nonfood items but many still lacked access to safe water and sanitation.

We used a 2-stage cluster design for the surveys. Although this increases the sample size required, it is the most practical solution in the absence of a nominative list of refugees that would have allowed for random sampling. The studies were designed with a large sample size to account for the anticipated heterogeneity between clusters (design effect of 2) and met the 5 criteria for validity and precision previously described for nutritional assessments conducted during complex emergencies.⁷

Surveys such as these have important and well-defined limitations.⁸ Our survey could not include households in which all members died, which may underestimate the mortality rate. As food distribution relates to the size of the family, we cannot exclude that some interviewees may have exaggerated the size of the household or underreported the number of deaths. However, field teams were specifically trained not to count deaths that occurred before the study period, and the short recall periods used for both Kalma and Muhajiria should have minimized recall bias.

Mortality rates in refugee situations are commonly compared with the emergency threshold, which is calculated as twice the normal mortality rate for the region (in this case sub-Saharan Africa); hence, the reason for using 1 per 10 000 per day in our analyses. An alternative might be to compare our rates with background rates in Darfur. A region-wide survey undertaken by Save the Children Fund-United Kingdom in 2001⁹ estimated the crude mortality rate of residents and IDPs to be 1.0 and 1.4, respectively. Compared with these data, our estimates are high.

Children younger than 5 years are more likely to die of communicable diseases, and these deaths are inextricably linked with their underlying nutritional status. In Muhajiria, childhood mortality and the incidence of malnutrition were both relatively low and likely reflect the recent displacement of many of the families interviewed, with childhood health still close to predisplacement levels. In Kalma, malnutrition was well above even the “critical” threshold of 15%.⁵ In this case, IDPs had no access to cultivated land or to food other than that provided in the general food distributions performed by international aid agencies. Although 98% of residents had registration cards at the time of the survey, distributions in the area had been sporadic.

Deaths from diarrhea are likely to reflect inadequate environmental sanitation. This is supported by the reports of poor access to safe water and latrines and low ownership of soap in our surveys as well as direct observation by the field teams of conditions in IDP settlements. In Kass, a measles outbreak documented by MSF in June 2004 may have contributed to the high under 5-year mortality rate at this site and prompted a measles campaign run by UNICEF. Low background levels of measles vaccination have been documented in Darfur⁹; inadequate coverage in the June 2004 measles campaign in Kass and the huge growth of Kalma camp since the June 2004 campaign may account for the low vaccination coverage reported herein. Muhajiria is in rebel-held territory and was not included in the measles campaign. Inadequate primary health care, which was not directly evaluated by our survey, is also likely to contribute to deaths from medical causes.

In September 2004 in 3 of the major IDP settlements in South Darfur, living conditions for people affected by the conflict were still precarious and clearly met the criteria for a humanitarian emergency. Very similar situations were documented in West Darfur in June and July 2004.¹⁰ Although food distributions in the main centers and at least in 2 sites (Kass and Kalma) were improved as a result of this and other surveys, the recent news from South Darfur is not encouraging. Regular measles catch-up campaigns are not favored by local authorities who advocate that regular Expanded Program of Immunization schedules should be sufficient, and there is serious risk of measles outbreaks in IDP settlements. Security is deteriorating and there are constant new arrivals of IDPs to the camps. As of February 2005, the World Food Program food registration data for the population of Kalma stood at 150 000. After recent violence in the region, the population of Muhajiria has scattered throughout the surrounding area. Violent attacks are reported frequently by those individuals who stray from the camps to farm or collect fire wood. Additional efforts from humanitarian and governmental actors are urgently needed to guarantee acceptable living standards for these populations.