Retrospective Mortality Survey Among the Internally Displaced Population , Greater Darfur , Sudan , August 2004

Summary of new report published by WHO, assessing the current health status of the internally displaced people in Darfur, Sudan,


PAGE V Recommendations
1.The survey must be completed in South Darfur as soon as possible; 2. Current humanitarian operations need to be intensified to reduce overall mortality rates; 3. Additional efforts are needed to improve environmental health (access to clean water and latrines); 4. The existing early warning system for disease surveillance needs to be enhanced for purposes of prospective mortality surveillance.This may be done through a combination of activities such as 24-hour surveillance of cemeteries, active mortality reporting at the community level by health visitors (one per 200 families), and monitoring through free provision of burial shrouds; 5. Further study is needed to document patterns of health-seeking behaviours and the quality of curative health care.

Introduction
The Darfur region of Sudan covers 256,000 square kilometres.It is comprised of three states -North, South, and West Darfur -with a total estimated population of between five and six million people.
In February 2003 about 2.2 million people were affected by the escalating conflict in Darfur.
Around 1.2 million of them fled their homes and sought refuge in other towns and villages in Darfur and across the border in Chad.As of August 2004, 127 settlements for internally displaced persons (IDPs) had been identified in Darfur (34 in the North, 45 in the South and 48 in the West) (Figure 1).There are another 15 refugee camps in Chad.
There has been increasing international concern over the health status of the internally displaced population in the region.To monitor morbidity and mortality in this population, the World Health Organization (WHO) and the Sudanese Ministry of Health (MoH) set up an early warning surveillance system.However, this system collects data only from health facilities, and information on mortality rates remains sporadic and incomplete.Data on mortality provide a fundamental measurable indicator of the health status of an affected population.Thus, to better assess the current health status of the internally displaced people in Darfur, WHO and the Sudanese MoH jointly conducted a retrospective mortality survey in August 2004.

Primary objective
To estimate the crude mortality rate in the 62 days from 15 th June to 15 th August 2004, among the IDPs present in the settlements at the time of the survey, in each of the three states of Greater Darfur (North, West and South) region.

Secondary objectives
To describe the demographic characteristics of the study population To identify the major causes of death (fever, respiratory infections, diarrhoea, injury/violence, and other) To describe the availability of basic services availability for IDPs PAGE 2

Methods
The study protocol is attached as Annex 1.

Study setting
The survey covered the IDPs living in accessible IDP settlements in North, West and South Darfur.
The crude mortality rate was estimated in this population for a 62-day period between 15 June and 15 August2004.This reflects the mortality experience of the population regardless of when they arrived in the settlement.

Study design
We used a two-stage cluster sampling design.The sampling frame included IDP settlements identified by WFP or OCHA (in both government and non-government areas).We excluded from the sampling frame settlements that were not accessible because of poor roads or for security reasons at the beginning of the survey.For each state, 50 clusters were randomly selected within the IDP settlements, based on the method described by Henderson and Sanderson (EPI) 1 .We used IDP estimates from WFP, OCHA or NGOs, depending on which was most recent.For each cluster the first household was randomly selected.The third-closest household was subsequently selected.

Sample size calculation
Our study was designed to detect a crude mortality rate (CMR) of 1.5 deaths/10,000/days or higher (95% Confidence Interval 1.0 to 2.0) with a cluster effect of 2. The sample required is 7,500 for each state.Assuming an average household size of five people, we calculated that at least 1,500 households (50 clusters of 30 households) had to be interviewed in each state.

Ethical approval
WHO guidelines do not require ethical review for mortality surveys carried out during humanitarian emergencies.The oral consent of the head of household was obtained before the start of each interview.

PAGE 3 2.5 Data collection
Data were collected using an English/Arabic questionnaire, which we piloted in a suburb of In such cases, local health care workers or health visitors were trained and accompanied by members of the expatriate survey team.
To establish whether a death had occurred during the study period, we developed a calendar of significant events for each settlement, such as vaccination campaigns.In addition, we asked about the occurrence of 'karama' (a remembrance of the dead which is observed during the first three days after death, after 15 days and again after 40 days).

Analysis
The mortality rate was calculated as the number of deaths per 10,000 persons per day.The numerator included all deaths recorded during the study period.The denominator was the average population over the study period.This included all family members recorded as alive.Individuals recorded as disappeared, absent or dead were assumed to be present for half the study period.For crude mortality, one death per 10,000 people per day was used as the threshold to define an emergency situation 2 3 .For mortality amongst children under the age of five, we used a threshold value of 2 deaths per 10,000 persons per day 2 3 .All analysis was conducted using Epi6.04d and

North Darfur
In North Darfur the number of IDPs was estimated to be 382,626 (Source WFP, 5 August 2004).
We conducted interviews with a sample of 1,290 households representing a total of 9,274 IDPs at the beginning of the study period.Eighty-two deaths were reported during the study period; 26 people were reported as having disappeared and 312 people were absent.At the end of the study period the population alive (excluding deaths, absent and disappeared individuals) was 8,854.The average number of persons per household was seven.Only one household did not consent to the interview.On average, two randomly-selected households per cluster could not be included due to the absence of all household members.These were replaced by other households randomly chosen from the same cluster.One individual had no sex recorded, 12 individuals had no age recorded and no cause was given for one death.

Demographic characteristics of the sample population
Among the individuals in the sample population alive at the end of the study period (15 August) 47% were male and 53% were female.There were more females than males in the 0 to 4 year and 15 to 49 year age groups (Table 1).The age distribution among the sample of IDPs is shown in Figure 1.Children under five years old are under-represented, especially among males.There were fewer males than females between the ages of 15 to 40.Comparative figures from the last census in Sudan in 1993 (Figure 2) suggest that males and females are usually represented equally [1].The crude mortality rate in the IDP sample population in North Darfur between 15 June and 15

PAGE 11
For all ages, diarrhoea was reported as the main cause of death, accounting for almost a quarter of mortalities; a fifth was reported as due to injury or violence; 17% (14/81) due to fever; and 10% (8/81) due to respiratory disease (Table 4).Other causes were reported for about a third of all deaths.Proportional mortality varied by age group, although the numbers were small.Among children under five years old, the most frequently reported cause of death was diarrhoea (44%, 10/23).Among individuals aged five to 14 years old, there was one death reported due to each cause except fever, for which no deaths were reported.Injury or violence was reported in 44% (12/27) of all deaths among individuals aged 15 to 49 years.The largest proportion of deaths among individuals aged 50 years and older was reported as due to other causes (44%, 12/27), with similar numbers of death reported for fever, respiratory disease, diarrhoea and injury or violence.

PAGE 12 3.2 West Darfur
In West Darfur, the number of IDPs was estimated to be 498,528 (source WFP, 5 August 2004).We conducted interviews with a sample of 1,292 households representing a total of 7,995 IDPs at the beginning of the study period.One hundred and forty-two deaths were reported during the study period; seven people were reported as having disappeared, and 187 people were absent.The sample population alive at the end of the study period (excluding deaths, absent and disappeared individuals) was 7,659.The average number of persons per household was six.Three households did not consent to the interview.On average, two randomly-selected households per cluster could not be included due to the absence of the family members.These households were replaced by other households chosen randomly in the same cluster.One individual had no sex recorded, six had no age recorded and no cause was given for one death.

Demographic characteristics of the sample population
Among the individuals in the IDP sample population alive at the end of the study period ( 15August) 47 % were male and 53% were female.There were more females than males in the sample among 15 to 49 years and fifty years and older age groups (Table 5).

PAGE 14
The age distribution among the sample of IDPs is shown in Figure 4. Children under five years old are under-represented.There were fewer males than females between the ages of 15 to 40 years.
Comparative figures from the last census in Sudan in 1993 (Figure 2) suggest that males and females are usually represented equally.Note: For all ages the threshold for a health emergency is 1.0 /10,000/ day; and for under 5 year olds it is 2.0/10,000/day 2 3 PAGE 17 For all ages, diarrhoea was reported as the main cause of death in over a third of deaths (52/141), 21% (29/141) due to fever; 12% (17/141) was reported as due to injury or violence; and 9% (13/141) due to respiratory disease (Table 8).Other causes were reported for about a fifth of all deaths (30/141).Proportional mortality varied by age group, although the numbers were small.
Among children under five years old, the most frequently reported cause of death was diarrhoea (76%, 19/25).Among individuals aged five to 14 years old, half the reported deaths were reported due to fever (54%, 7/13).Injury or violence was reported in 41% (13/32) of all deaths among individuals aged 15 to 49 years.The largest proportion of deaths among individuals aged 50 years and older was reported as due to diarrhoea (37%, 26/71).Note: The total number of deaths in the table is 141 instead of 142 because cause of death was not recorded for one death.

PAGE 18 3.3 Kalma camp
Kalma camp was the only part of South Darfur that the team was able to survey.The number of IDPs was estimated to be 73,658 (source WFP, 5 August 2004).We conducted interviews in a sample of 558 households (19 clusters) representing a total of 3,506 IDPs at the beginning of the study period.Eighty deaths were reported during the study period; four people were reported as having disappeared and 154 people were absent.The population alive at the end of the study period (excluding deaths, absent and disappeared individuals) was 3,267.The average number of persons per household was six.On average, two randomly-selected households per cluster could not be included due to the absence of the family members.These households were replaced by other households chosen randomly in the same cluster.One individual had no age recorded, one individual had no vital status recorded and no cause was given for three deaths.

Demographic characteristics of the sample population
Among the individuals in the IDP sample population alive at the end of the study period ( 15August) 45 % were male and 55% were female.There were more females than males in all age groups except 5 to 14 years (Table 9).

PAGE 19
Over half the sample population was under 14 years old (Table 10).Children under five years of age represented 16.9 % (95% CI [15.9 -17.9]) of the sample population (552/3267).The distribution of the sample population by age group varied according to sex (Table 10).

PAGE 20
The age distribution among the sample of IDPs is shown in Figure 6.Children under five years old are under-represented.There were fewer males than females between the ages of 15 to 49 years.
Comparative figures from the last census in Sudan in 1993 (Figure 2) suggest that males and females are usually represented equally [1].

PAGE 23
For all ages, diarrhoea was reported as the main cause of death in about 40% (32/77) of deaths, 17% (13/77) due to fever and respiratory disease and 10% (8/81) due to injury or violence (

Summary of results
Monitoring the health status of a population is a priority in complex emergencies.The crude mortality rate is a robust and simple indicator of population health 3 .Despite the implementation of a WHO/MoH early warning surveillance system for communicable disease outbreaks and mortality in health facilities, no reliable community-level mortality data have been available in Greater Darfur.
Our study was designed to estimate baseline mortality data from 15 June to 15 August 2004 for the IDPs in Greater Darfur.This reflects the mortality experience of the population regardless of when they arrived in the settlements.However, these populations are currently living in IDP settlements and need appropriate humanitarian assistance.
The main results of our survey suggest that, despite the relief efforts throughout Greater Darfur, the mortality in the IDP population is not yet controlled.The crude mortality rates in North Darfur and West Darfur are above the emergency threshold (1 death /10,000/day).The same applies to Kalma camp in South Darfur.The main cause of death reported during the survey was diarrhoea, particularly affecting children under five years old.The age distribution among children suggests an important deficit in children under two years old.Among adults under 50 years old, injuries and violence were the main causes of death.
A large proportion of IDP households had access to basic services, food and non-food items.
However, about a third of households still lack access to safe water and sanitation.

Validity of results
We used a two-stage cluster sampling design for the survey.Although this type of study design increases the required sample size, a cluster design was the most practical way of sampling the population in the current context; systematic sampling was not appropriate in the IDP settlements and the absence of nominative lists of IDPs did not allow for random sampling.Our study was designed with a large sample size to account for the anticipated design effect and increase the precision of the results.Although we only completed 43 clusters in North and West Darfur, we exceeded the number of individuals required for our survey, achieving the required precision.In addition, the cluster design effect is below that expected, indicating a similar situation in all clusters.The area covered by our survey in North and West Darfur represents the majority of the IDP population.

PAGE 25
To estimate the current mortality, we used a two-month recall period.Field teams were specifically trained to avoid including deaths that occurred before the study period.For this we used event calendars created for each settlements, based on significant events such as vaccination campaigns.
In addition, we enquired about 'karama', a remembrance of the dead that is observed during the first three days after death, then 15 days and again 40 days after death.
Interviewers were trained and accompanied during the interview by WHO or MoH team members.
Absent households were re-visited several times during the same day and when possible visited on subsequent trips to the settlement.Regular debriefing with the field teams was conducted to check the quality of data collected.

Possible limitations
Not all IDP settlements were included in the survey due to accessibility.It is possible that mortality rates in these populations are higher as they have less access to humanitarian aid.Therefore, we cannot exclude the possibility that the mortality rates estimated for the accessible areas in our survey are lower than for the whole state.In addition, our survey could not include households in which all members died.This may underestimate the mortality rate.As food distribution is related to the size of the family, we cannot exclude that in some cases, the interviewee exaggerated the number of individuals in the household and under-reported the number of deaths.This would also have led to an underestimation of mortality rates.
Conversely, if for any reason interviewees over-reported the number of deaths, this would have increased the mortality estimate.However, our results suggest that prevarication is unlikely to have occurred or was limited since our data show high consistency between death rates (even with small numbers when analysed by cause), age distribution and cause of death in the three surveys.
The survey collected information from respondents on their perceptions of the principal causes of deaths, selected from a menu of conditions that normally occur during emergency situations.The answer categories were selected as those which are meaningful to respondents and easily recorded during an interview.Hence information was not sought on acute malnutrition as a possible direct cause of death .

PAGE 26 4.4 Kalma Camp
In the South the survey was interrupted and limited to Kalma camp.The 19 clusters from Kalma camp included 3,506 persons.This allows inference about mortality rates in the camp with good precision.The results, however, cannot be generalized to other settlements in South Darfur.The lower confidence level of the crude mortality rate in Kalma camp is 2.9 /10,000/day.This does not allow us to rule out a severe situation in Kalma camp.

Mortality
is not yet under control in IDP populations currently living in North Darfur, West Darfur and Kalma camp in South Darfur, and those now living in IDP settlements need appropriate humanitarian assistance.In North Darfur, our estimate of the crude mortality rate is about three times the expected rates under normal conditions in Africa 6 (0.5 deaths /10,000/day).In West Darfur, our estimates suggest that the crude mortality rate is six times the expected level for Africa.
In Kalma camp the estimated mortality rate was over seven times the expected level for Africa.
These estimates suggest that the humanitarian situation is still in the emergency phase.
Children under five years old were more likely to die from communicable diseases * .Deaths due to fever are likely to reflect death by malaria.Deaths due to diarrhoea are likely to reflect poor environmental sanitation.This is supported by the proportion of IDPs with no access to safe water and latrines in our survey as well as direct observation by the survey teams of the conditions in IDP settlements.Deaths by malaria, diarrhoea and respiratory infections can also reflect poor access to, or poor quality of, health curative care.Almost all households reported to have sought medication from a health centre.However, the survey did not enquire into the health-seeking behaviours of the IDPs, nor was information obtained on the quality of any health care that they received.This is a critical area that needs further study.Mortality figures are of paramount importance in complex emergency situations.They provide an obvious but very crude indicator of the severity of the health status of the affected population 1 .One death per 10,000 people per day is generally accepted as the threshold value to determine an emergency situation 2 .
Since the beginning of 2003 a humanitarian crisis has been unfolding in the greater Darfur region of Sudan, with 1.2 million affected people 3 .
Despite the implementation of an early warning surveillance system for this crisis, which includes number of deaths per week, mortality information remains sporadic and limited due to the accessibility of settlements, intermittent reporting and imprecise denominators.Given the importance of having a baseline measurement of the current mortality a retrospective mortality survey will be conducted in the three states of Darfur.

Objectives A) Main Objective:
To estimate crude mortality rate in the 62 days prior to the survey, among the Internally Displaced Population (IDP) present in the settlements at the time of the survey, in each of the three states of Greater Darfur (North, West and East) region.For the purpose of this study an IDP is defined as a person who, at the date of survey, is living in an IDP settlement and not living in their place of permanent resident.
An accessible area is defined according to UN security criteria and road access.IDP settlements not accessible for UN staff but accessible to NGO field staff, who are trained by the WHO study team, will be included.

Sampling Method
We will conduct a two-stage cluster sample survey with the individual as the basic sampling unit.
The sampling frame is based on the World Food Program (WFP) IDP population estimates for all known camps in the greater Darfur Region.These figures are dated 5 August 2004 and will be updated by the WHO study team when they reach the field.The WFP estimates provide an indication of the relative size of camps, required for the selection of clusters.

Sample size Calculation
The Crude Mortality Rate will be calculated based on a recall period of 62 days.This recall period will give an acceptable indicator of the current mortality within the refugee settlements.
In order to detect a Crude Mortality Rate (CMR) of 1.5 deaths/10,000/daz with a 95% CI of 1.0 to 2.0 deaths/10,000/daz (relative precision of 33%) assuming a cluster effect of 2, we will require 7500 study participants.Based on an average household size of five people, 1500 households will be included in the study.These will be sampled using 50 clusters of 30 households.Sample size calculations were carried out using Epitable, Epi info 6.04d (CDC Atlanta).
A household is defined as a group of people living together (sharing the same meals and sleeping area) on the 15 th August 2004.

PAGE 32 4. Study period
The study period for the calculation of the CMR is from June 15 th to August 15 th 2004.
The survey will be carried out in August 2004.It is estimated that the data collection in the field will take 15 days.

Data collection
Data will be collected using a standardized and piloted questionnaire, administered at the household level by a team of trained interviewers.The head of each household will be approached and asked to participate in the study.If he/she refuses to participate the nearest household will be selected.
The questionnaire will cover: Basic service availability in the households: water and sanitation, food and non food items Demographic composition of the household at the start and end of the recall period, information will also be recorded for people joining (newborn, reunification) or leaving (deaths, disappearances, prolonged absences) the household during the study period.
Status of the household members: alive, absent, disappeared and death.Main cause of death: fever, diarrhoea, respiratory illness, violent death and other.
Five teams of 2 people will cover each State.A team will consist of an interpreter who speaks English and Arabic and a Sudanese member of staff or an expatriate.In each State a training day will be organised covering methods, sampling techniques, use of questionnaire, and logistical aspects of the survey.All team members will attend this training.In each State one member of the WHO study team will act as supervisor.Five cars and 5 drivers will be needed in each State.

Data Validation and data entry
To ensure data quality daily meetings will be held among field workers and supervisors to review the data collection process, to check data completeness and to resolve any logistical or methodological issues.
Team members will enter data every evening.Epiinfo 3.2.2 will be used for that purpose.
The database data entry form will include quality checks, such as mandatory fields for completion and data format.Data validation will be performed before the analysis.

Data Analysis
Analysis will be carried out using Epiinfo 3.2.2.
The response rate will be calculated • Proportion of households with non-responders and refusals.
The description of the sample will include • Age and sex distribution of the IDP population (age sex pyramid) PAGE 33 • Proportion of absent and disappeared household members • Proportion of households with access to basic services The main outcome by State will be: • Crude mortality rate (number of deaths/10,000/day).Mid population sizes will serve as denominator.• Under 5 years age mortality rate (number of deaths/10,000/day) • Design effect for the CMR and under 5 years CMR will be computed and applied to calculate 95% Confidence Intervals.• Proportional mortality for the main causes of death (fever, respiratory diseases, diarrhoea and injury/violence, other)

Ethical Considerations
The survey will be presented to the head of each household and their oral consent will be obtained before the start of each interview.The questionnaires will be anonymous.

Limitations
This study will provide an average estimation of the crude mortality of IDPs in each of the States of Darfur, but the methods used do not allow the calculation of the CMR for each of the IDPs settlements included in the survey.
The study will provide an average CMR over a period of 2 months.Dividing the periods by month or week in order to observe trends in mortality decreases the power of the study and is likely to provide inaccurate estimations because of the sample size limit.
In order to calculate the CMR the study requires household data covering a very precise and short period (62 days).Deaths occurring around the beginning of the period may be subject to misclassification as inaccuracies in the recall of dates can lead to the incorrect inclusion or exclusion of a death in the study period.To limit this possible bias, a calendar of events for each of the camps will be created to assist recall of the precise date of death.
Information bias due to the provision of inaccurate death data and/ or current household size can be limited by ensuring that the interviewees are aware that all the information they provide is anonymous and that the study is not part of a registration process for the distribution of aid.This information forms part of the introduction to the study provided to each interviewee.
The camps excluded for security and/or climatic conditions may represent the most vulnerable populations and they will not be represented in the study.
Households in which all members have died cannot be included.This could lead to an underestimation of the CMR.

Figure 3 : 2
Figure 3: Age distribution among the sample of IDPs under 5 years of age, North Darfur, Sudan, 15 August 2004

Figure 4 :
Figure 4: Age distribution among the sample of IDPs, West Darfur, Sudan, 15 August 2004

Figure 6 :
Figure 6: Age distribution among the sample of IDPs, Kalma camp, South Darfur, Sudan, 15 August 2004 total number of deaths in the table is 77 instead of 80 because cause of death was not recorded for three deaths

3 Results
Forty-three clusters in North Darfur and 43 clusters in West Darfur were completed.Field work was suspended on 2 September due to the deteriorating security situation.In South Darfur, the movement of field teams initially was limited because of the lack of United Nations (UN) vehicles.

Table 5 : Age-specific sex ratio among the sample population of IDPs, West Darfur, Sudan, 15 August 2004 Age group Number of Males Number of Females Sex ratio 0 to 4 years
Over half the sample population was under 14 years old (Table6).Children under five years old represented 16.6 % (95% CI [15.7 -17.5]) of the sample population (1270/7651).The distribution of the sample population by age groups varied according to sex.

Table 6 : IDP sample population by age and sex, West Darfur, Sudan, 15 August 2004
Note: One individual had no sex recorded, three males and three females had no age recorded

Female Male PAGE 16 3.2.2 Retrospective mortality in the sample of internally displaced persons, West Darfur, 15 June to 15 August 2004
Among the 7,996 persons included in the sample, 142 deaths were reported during the study period, including 25 deaths among children under 5 years of age and 7 deaths among children less than 1 year of age.The crude mortality rate in the sample of IDPs in West Darfur between 15 June and 15 August was

Table 12 : Distribution of reported cause of death by age group, in a sample of IDPs, Kalma Camp, South Darfur, Sudan, 15 June -15 August 2004
Table 12).

PAGE 30 Annex 1: Survey Protocol Protocol for Retrospective Mortality Survey Among Internally Displaced Populations, Greater Darfur, Sudan, August 2004 Ministry of Heath, Republic of Sudan World Health Organization 1. Study Context and Justification
The age structure of the IDPs in our study samples is similar in all three surveys.It reflects a deficit of children under five years old, especially among children under two years of age.This means that * The synergistic effect of malnutrition, infection and immunity is well established.The current available information on acute malnutrition in North Darfur ranges between 7.8% -39%, and in West Darfur between 11.3% -22.5% (WHO compilation of NGOs and SMOH findings, September 2004) PAGE 27 these children have died during the last two years, reflecting the difficult living conditions of the population.Children under two years of age are the most vulnerable and as such are the first group to be adversely affected during humanitarian emergencies.The age structure also reflects a deficit of adult men, an observation also made in previous conflict situations.

Methods as applied in each state of Greater Darfur Region 1. Study population
The study population is the IDPs within the accessible IDP settlements in the three states of Greater Darfur (estimated by WFP for West, North and South Darfur as 460997, 322483, and 288539 respectively).IDP populations living with the resident populations will not be included in the survey (due to logistical constraints)