Responding to the crisis in Congo-Zaire: emergency Feeding of Rwandan Refugees May-July 1997

Author: Susanne Jaspars, is an independent Consultant in Nutrition. This article was based on her experiences while working for CARE UK as a Food Policy Consultant. The emergency described in this article was part of one of the largest and most politically complex crises that humanitarian agencies have ever had to face. It highlights the importance of an analysis of political vulnerability in planning food distribution in a complex political emergency as well as the need for food distributions to be flexible and carefully tailored to the local context. The political background described is not intended to provide a comprehensive analysis but is limited to the immediate period leading up to the crisis in 1997 as this relates to the design and implementation of programmes in that year.

In response to the extreme humanitarian crisis amongst Rwandan refugees remaining in Zaire (now the Democratic Republic of Congo; DRC), CARE implemented and co-ordinated emergency food distribution programmes in Kisangani and Lulingu (Shabunda zone) from May to September 1997. These programmes were designed to have maximum potential to reach the most physiologically and socio-politically vulnerable, whilst at the same time not either reinforcing, or undermining , the control of ex-FAR or Interahamwe elements over the refugee populations. Many of these refugees had been repeatedly displaced from makeshift camps like Tingi-Tingi in Eastern Zaire following the eruption of conflict in November 1996. Whilst attempting to alleviate one of the most severe crises in the history of humanitarianism, conscious efforts were made to implement the lessons learnt from the multi-donor evaluation of the Great Lakes emergency in 1994 and not repeat the same mistakes in food distribution. Internal politics within the refuge population, as well as the hostile and often violent relationship between the ADFL (Alliance of Democratic Forces for the Liberation of Congo-Zaire) and the refugees, was explicitly recognised in the design of the emergency food distribution system.

Both food distributions in Kisangani and Lulingu, were extremely challenging, as the crisis had to be addressed within the context of various political and logistical constraints that limited access. Each situation provided a different set of challenges. South of Kisangani, refugees were crowded in concentrations along the railway line, in extremely poor sanitary conditions, and often without shelter. In Lulingu, about 8000 refugees were dispersed in over 105 locations, many of which were several days walk into the forest. Traditional food distributions consisting of a general ration, supplementary feeding, therapeutic feeding, and conventional practice for selecting target groups, were neither appropriate, nor were they feasible. Whilst adhering to principles of neutrality, equity, and accountability, conventional practice for addressing a nutritional crisis had to be abandoned.

Why was there a health and nutritional crisis?

When the CARE team arrived in Kisangani, the condition of the refugees was almost beyond salvation. The severity of the crisis was indicated by the large numbers of severely malnourished adults (many with oedema), and the extremely high mortality rates. South of Kisangani between 120 to 180 refugees per day were reported to be dying in early April. As the estimated population at the time was 80,000, this represents mortality rates of 15 to 22.5/10,000/day, some of the highest mortality rates ever reported. A large proportion of the population was literally too weak to even move, and were lying down by the side of the road that bisected Biaro. In Lulingu, a team of stretcher bearers was necessary to bring people who were too weak to walk out of the forest into the feeding centres. How did Rwandan refugees come to be in such critical condition? Except for an initial 3-4 month period following the influx into Zaire in July 1994, the nutritional and health status of the refugees had been within acceptable ranges.

By November 1996, after the mass repatriation of some 600,000 refugees, the entire refugee population that remained in Zaire was politically vulnerable. The ADFL considered all as a political and military threat. This was because Hutu militia amongst the refugees had continued their campaign of violence whilst in the refugee camps and had been armed by and fought on, the side of the Zairian army. Also, the international community had failed to bring the genocidaires to justice. To many observers it seemed that direct attack and physical abuse of refugees, as well as denial of access and resources was part of AFDL military strategy. It was thought that this strategy was developed and implemented in order to weaken the support base for the Interahamwe and ex-FAR. Humanitarian assistance to refugees was viewed with great suspicion, as this had the potential to stabilise and strengthen populations opposing the ADFL. Hostility between the local population and refugees was also encouraged

The prevailing wisdom amongst the international community was that many of the remaining refugees were Hutu militia in good physical condition. All key political elements involved in the conflict, including key donor countries, saw International Humanitarian Law as an impediment to achieving "conflict resolution". This enabled the ADFL to continue its military strategy. Thus the crisis amongst refugees in Zaire in 1997, was the direct consequence of the denial of humanitarian space, or the lack of political will to create it. Aid agencies could only secure humanitarian space once the desired crisis was well advanced. Even at this late stage, UNHCR, had to negotiate access to refugee populations in Kisangani, by agreeing to repatriate quickly.

Every major refugee concentration had been attacked. Groups assisted by CARE in Kisangani and Lulingu, had been subject to several such attacks. CARE arrived in Kisangani following the latest attack; 80 -100,000 refugees had settled in three sites south of Kisangani, which were attacked between 21st and 24th April, after which only 30,000 returned to one of the sites, Biaro. Refugees in Lulingu had been under attack in Bukavu and later Shabunda, and some of them dispersed into the forest around Lulingu.

Within the refugee population, the unequal power relationships that existed in the refugee camps between '94 and '96, were likely to be magnified. Following mass repatriation, the proportion of exFAR and Interahamwe was likely to be higher. There were clearly individuals or groups who were stronger and able to manipulate the distribution of resources. The extent of their control is indicated by the sheer distance travelled into the interior of Zaire.

The food distribution strategy

Overall strategy - Not only did the team have to address the high degree of malnutrition and mortality amongst the population, but it had to do this within the context of the external and internal political factors that caused the crisis in the first place. In developing the food distribution system, three basic assumptions were made about the refugee population in Eastern Zaire:

• there was a high proportion of malnourished and other vulnerable groups (30-50%)
• refugee populations as a whole are politically vulnerable
• there will be some groups or individuals amongst the refugees who are physically and socially stronger than others.

Overall objectives were to improve access to food, in particular for the most vulnerable and thereby prevent excess mortality. Food aid had to meet the food needs of a malnourished population and effectively reach the physiologically, socially and politically weakest amongst them; In addition, assistance had to be provided in a manner that would not increase the risk of attack or theft, or increase hostility between the local population and the refugees.

Based on this analysis, the distribution of cooked food was considered to be the method of distribution most likely to meet these objectives. This was because:

• cooked food is highly perishable and therefore unattractive and less subject to theft and diversion
• employment in the communal kitchens is less attractive to the politically stronger, reducing potential for control by the allocation of food resources
• recipients can be observed to eat the food
• cooked food does not require possession of cooking utensils and fuel
• it provides food for those too weak to look after themselves.

Dry food distribution on the other hand, is more likely to reach and be controlled by, those who are physically and socially stronger. However, both in Biaro, and Lulingu. a combination of wet and dry distribution was implemented. In Biaro, because the capacity to cook for 30,000 could not be established quickly enough, communal kitchens were set up with a capacity to cook for 17,000. In Lulingu, dry rations were initially distributed to refugees who were not considered to be amongst the most vulnerable; and to discourage refugee concentrations from developing. Any concentration of refugees was at risk of being attacked.

The communal kitchens can be considered as a combination of a general distribution and supplementary feeding. No separate supplementary feeding programmes were established. Therapeutic feeding was not possible in Biaro, mainly because agencies only had access for a maximum of four hours per day. Severely malnourished were eventually preferentially evacuated to the transit site in Kisangani, where therapeutic feeding was possible. In Lulingu, the most severely malnourished were transferred to hospitals, where CARE had therapeutic feeding programmes.

Identifying and reaching the vulnerable

Although the food distribution aimed to feed all refugees, the most vulnerable had to be identified, to make sure they were effectively reached. Rather than using the 'traditional' classification of vulnerable groups, - children under five, pregnant and lactating women, and the elderly, the vulnerable were instead identified as follows:

1. Malnourished of any age or sex group.
2. The sick or those unable to move, or otherwise unable to cook for themselves.
3. Unaccompanied minors or those who are not assisted by relatives.
These groups were identified visually by monitors working in the kitchens in Biaro, and by both monitors and nutritionists in Lulingu.

The distribution system had several components to increase the likelihood that the most vulnerable received their entitlements. These included ensuring that the vulnerable received food first, decentralisation and a high degree of monitoring.

All communal kitchens made efforts to ensure that the vulnerable were served first. In Biaro, monitors took buckets of porridge into the blocks (sections) to distribute to those too weak or sick to come to the kitchen. They would then visually identify the thinnest amongst those waiting for food at the kitchen, after which others would be fed. In Lulingu, the location of kitchens was determined by the presence of small concentrations of refugees in extreme conditions, staying in local churches. Initially, only the most vulnerable, as identified by CARE monitors, were brought into the centres from the forest. Teams of stretcher bearers were employed when refugees were too weak to move to the centres themselves.

In Biaro, kitchens were decentralised so that the weakest did not have to move far to get food, or could easily have food brought to them. With each kitchen only responsible for people in a small area, the weakest could also easily be identified. A maximum of 13 kitchens was established in Biaro, each with the capacity to serve 800. In Lulingu, the provision of effective assistance in accessible areas had to be constantly balanced against the risk associated with creating refugee concentrations. The programme was as decentralised as logistically feasible. Four main centres were established, which by mid-July, had been supplemented by a further 4 smaller centres, or transit sites.

Rations provided at the communal kitchens The standard WFP/UNHCR ration was found to be inappropriate for the kitchen programme because it did not meet the requirements of a malnourished population, and was not possible to cook as two meals per day. The ration was altered so that one meal of porridge (120g CSB and 20g oil) and one meal of maize and peas could be provided (15Og maize, 120g peas, 20g oil), with an increased ration of oil and CSB. In the Lulingu centres, the thinnest beneficiaries were also provided with 750mls of therapeutic milk per day.

Ration Composition
WFP/UNHCR ration communal kitchen ration (g/person/day)
Maizemeal	350	150
Peas	120	120
CSB	50	120
Oil	20	40
Salt	5	5
Biscuits	100
Kcals	1990	2116

Monitoring

Each programme had a large number of monitors. In Biaro, there were 10 UNHCR Field Assistants, 45 monitors for the dry distribution, and 26 monitors for the kitchen programme. In Lulingu, the area was divided into 15 divisions, each of which was assigned two monitors. Responsibilities included carrying out a census and regularly updating population figures, monitoring food received at kitchens (or in blocks) and by intended beneficiaries, providing information on the programme to the target population and to CARE about the target population. In Lulingu, the monitors also carried out the dry food distribution.

It was difficult, if not impossible, to monitor the impact of the programme in terms of nutritional status or mortality rates. Population numbers and composition changed on a daily basis. The strongest were repatriated first and hence the proportion of vulnerable increased, which means that nutritional status and mortality data would not be comparable over time. The only means of monitoring was the receipt of food by its intended beneficiaries, which we attempted to do by employing a large number of monitors.

In Lulingu, we are convinced that the programme saved lives. Many of the refugees brought into the centres would undoubtedly have died if they had stayed in the forest. In Kisangani however, the impact would have been more limited due to the extremely poor sanitation in the refugee sites and the lack of a health outreach programme. The greatest impact was possibly on the morale of the refugees.

Constraints to effective implementation

In addition to the constraints explicitly taken into account in planning the programme, the provision of assistance was influenced by, pressure to repatriate, various constraints on access within the area of operation and fear of implementing an unusual food distribution.

Repatriation versus provision of assistance Pressure to repatriate quickly, in return for access, meant that the focus of activities and allocation of resources was highly biased towards repatriation (in terms of allocation of staff, equipment, transport etc.). Initially, many refugees in extremely poor health condition were repatriated. In Lulingu, food assistance came to be linked directly to repatriation. Although CARE initially provided food for all refugees, UNHCR later initiated discussions to devise a system whereby refugees would only be provided with food for a limited number of days. During this time, they would have to decide whether they wanted to return to Rwanda, and if they did not then food assistance would be terminated. CARE insisted however, that refugees should not be repatriated unless they were physically strong. Limited access within the area of operation.

Once access to a particular area was gained, access to refugees within that area was still restricted due to poor infrastructure, made worse by political objectives of denying assistance. For example, to access refugees from Kisangani, the Congo river had to be crossed by barge, after which all aid vehicles had to be accompanied by a military escort to the refugee site. Often the barge did not work and military escorts could be late. Access to the refugees was limited to at most 4 hours a day, but the entire process often took from 7 in the morning to 11 at night or later.

In Lulingu, those roads that existed were initially impassable by car, and most of the refugee concentrations could only be reached on foot. Much of the food aid had to be delivered on foot, requiring the employment of over 300 porters, who carried 25 kgs each. The distribution of just 1 kg each of food to 5618 refugees, took 2 weeks.

Fear of the unusual

Many of the agencies involved in the programme in Kisangani initially feared implementing a food distribution strategy that was unlike the distribution systems they were familiar with. To our surprise, it was even suggested to carry out distribution on a commune basis, even though this method of distribution had clearly been problematic in the past and resulted in inequitable food allocations.

Discussion and lessons learnt

First, it must be categorically stated, that although the CARE team tried to establish a programme along basic humanitarian principles, programmes in Eastern Zaire were to a large extent determined by the denial of humanitarian space. Aid agencies were required to accommodate the political aims of the ADFL and its supporters. From a nutritional point of view, there are two major lessons we learned from this experience:

1. An analysis of political vulnerability is essential in planning food distribution in a complex political emergency. The programme was in fact planned almost entirely from an analysis of political vulnerability. An assessment .of nutritional status, or even food economy, was not feasible in this context and in any case, the severity of the crisis was clear. Even if we had been able to carry out such assessments, this would have allowed us to plan rations, but not how we were going to distribute food so that it reached those who needed it most.
2. Food distribution systems need to be flexible, and carefully tailored to the local context. This extends to the identification of vulnerable groups and the methods for reaching them and the planning of rations to be distributed. Clear objectives, based on our situation analysis and basic principles, were more important in planning the distribution strategy, than standard guidelines on general distribution, supplementary and therapeutic feeding. Politically charged situations like those in Zaire, need a decision making framework for food distribution, which considers the potentially harmful as well as the positive effects of food distribution¹.

Finally, the crisis in Eastern Zaire is not over. UNHCR and the other agencies have now pulled out of Lulingu, but only around 1000 refugees have been repatriated, some of whom were Burundians. This means that at least 7000 refugees remain in the forests around Lulingu. Presumably these decided not to return to Rwanda, because either, at this moment in time they were safer in Zaire than in Rwanda, because they would be implicated in the genocide or are being held captive by their leaders. Who now, will bring the perpetrators of genocide to justice, and who will provide protection and humanitarian assistance to those who are not genocidaires? The refugees we found around Lulingu are almost certainly not alone; similar small groups of refugees are likely to be scattered throughout the forests of Eastern Zaire. In addition, unless these issues are addressed, the continued presence of refugees in Eastern Zaire, is likely to continue to destabilise the balance of power. An uprising of local Hutu, together with ex-FAR and Interahamwe in Massisi, has already resulted in the displacement of several thousand Tutsi.

It is difficult to forget the condition of some of the refugees as we first found them several months ago. In the face of abuses of international humanitarian principles and human rights law, such crises will continue to occur, albeit silently, hidden in the vast forests of the new Congo.

See also the Post Script to this article.

---

¹I first proposed such a framework in 1995; see The Management of Food Distribution to Large Refugee Populations, Jaspars, 1995, July; and General Food Distribution in Emergencies; from Nutritional Needs to Political Priorities, by Susanne Jaspars and Helen Young, RRN, ODI, December 1995.