• Send page
  • Print page

Reuniting children separated from their families after the Rwandan crisis of 1994: the relative value of a centralized database

30-06-2000 Article, International Review of the Red Cross, No. 838, by Maarten Merkelbach

 Maarten Merkelbach   is a delegate of the International Committee of the Red Cross. He took part in the ICRC programme for unaccompanied children in the Great Lakes region of Africa (1994-1997).  

The vulnerability of children during armed conflict is recognized both in general international law [1 ] and in particular in international humanitarian law [2 ] , which stipulate specific protection measures for children. Children separated from their families are more vulnerable still. Apart from caring for them, an important part of their protection is to reunite them with their families. This requires the following: first, identifying these children among the rest of the population; second, registering and interviewing them; third, tracing their parents or relatives during or after the chaos of war and mass migration; and fourth, making the reunification physically possible.

In 1994, war and genocide in Rwanda killed between 500,000 and 800,000 people and an estimated two million Rwandans fled to neighbouring countries. Both in Rwanda and in the refugee camps, the humanitarian aid agencies that had come to the region in response to the emergency saw that many thousands of children were alone, that is, they were apparently unaccompanied by adult family members. However, they could not be assumed to have been orphaned until all possible attempts to identify parents or other close family members had been exhausted - a process that can take years.

 Registering unaccompanied children  

In June 1994 the International Committee of the Red Cross, the International Federation of Red Cross and Red Crescent Societies, the Office of the United Nations High Commissioner for Refugees and the United Nations Children's Fund together agreed that a coordinated approach to the plight of these unaccompanied children was essential. On the assumption that computerized matching would facilitate their reunification with their families, it was decided to centralize on a database the names and other details of unaccompanied children and of parents looking for their children. The responsibility for setting up and running this database was assigned to the ICRC. [3 ]

The agencies agreed on general criteria for the registration and reunification of unaccompanied children. Only children less than 18 years old who were separated from their parents or other close (adult) relatives were registered. The process of reunification with parents or other (adult) members of the family had to take into account that the notion of what constitutes a family is broader in Central Africa than in the industrialized world.

The information recorded for each unaccompanied child was standardized for use. A copy of each paper registration in the field was transferred to the ICRC for entry into its database at the ICRC Tracing Office in Nairobi. A total of 61 different agencies and organizations transmitted registration forms to the ICRC. Similarly, a second standard form was used to register tracing requests, that is, enquiries from parents looking for their children. The information given on such forms was likewise transferred to the central database in Nairobi. Any changes in the respective child's whereabouts were also entered. The third form of standardized data entry related to successful reunifications (when the child was physically handed back into the care of the family) ; data on this form included the reuniting agency and certification by the family of the reun ification. A total of 75 agencies transmitted copies of reunification documents to the ICRC. All registrations, tracing requests and family reunifications were entered in the database. The registrations recorded in the database were effected in Rwanda, in refugee camps in the province of Kivu (Democratic Republic of Congo, formerly Zaire), Tanzania, Uganda and Burundi, and in European countries which had accepted Rwandan children after their evacuation; reunifications took place within Rwanda itself, between Rwanda and refugee camps, between and within refugee camps and between Rwanda and Europe.

The software - called Standard 4 - uses the Clipper programming language and runs on MS-DOS platforms ; its data is stored in dBase files. It is a programme developed by the ICRC in the early 1990s, constantly updated ever since and used in all its operations to manage data on individuals.

In order to coordinate with the many agencies in the field, to register and follow up the information, to centralize the data and to keep track of and reunite these unaccompanied children with their families the ICRC set up offices at ten places in Rwanda, three in Kivu province and two in Tanzania, as well as one each in Burundi and Uganda. At the height of the crisis these offices were staffed by over 40 expatriate and 600 national staff. All required training, coordination and clear procedures for data gathering and data handling. Data entry in Nairobi, Kenya, required standardization not only of procedures for data entry itself but also of seemingly straightforward data such as personal and place names. Some points indicating the particular difficulty of managing a database with names of people and places from Rwanda are given in Table 1. The time lapse between registration of a child and entry in the database varied from two to seven months (see Figure).

By 31 December 1997 the database contained 271,297 names, of which 119,577 were of child ren registered as unaccompanied. It should be noted that this figure pertains to the number of children and not to the number of registrations carried out, for many children were in fact registered more than once and by more than one organization. Table 2 shows a breakdown of the main outcome of this caseload.

 Chronology: three phases of activity  

The joint interagency approach was not decided upon until late June 1994; however, the database does include a limited number of registrations and tracing requests collected by the ICRC as of April 1994. The data therefore cover a period from April 1994 to December 1997. The way that the database was used varied during this 45-month period. Three phases can be distinguished, during which new registrations and reunifications continually took place (Figure 1):

  •  Phase 1 (April 1994 to August 1995) covers the period of the initial emergency and the spectacular migration of 850,000 people in four days to Goma in northern Kivu, subsequent data gathering and entry into the database. It ends when the bulk of the data had been entered and started to be shared with other organizations. Some agencies did nevertheless reunite children during this period; they must have done so without the assistance of the database. However, reunifications brought about by the ICRC at that time, including those in which a tracing request was made by the family, very probably did benefit from the database in some way. For example, families had access to the ICRC database to check whether their children had been registered somewhere, and in many cases were thus able to find their children without relying on an organization to reunite them. Foll ow-up in the field allowed these reunifications to be documented and subsequently entered as such in the database.

  •  Phase 2 (September 1995 to October 1996) covers the period when the refugee populations were more or less stable in the camps and when most of the registration data had been entered in the database. During this phase, the database was shared with other agencies in a variety of ways, for example, by transmitting extracts from the database on individual children registered and by listing children according to their whereabouts in a camp or children's centre. Especially useful in this phase was the ability to group hundreds of children according to their place of origin ; communities, villages and refugee camps could then be approached with lists of children with a likelihood of finding the relatives there. In view of the availability and use of the collected data, the reunifications in this phase can be assumed to have been attributable to the database.

  •  Phase 3 (November 1996 to December 1997) corresponds to a new period of migration in the form of the mass return of Rwandans to Rwanda in late 1996 and early 1997 from the refugee camps in Kivu and Tanzania; this was followed by the return of those who had fled further into the Democratic Republic of Congo. There were also new separations as a result of this population movement and hence an increase in new registrations. Reunifications in this phase were rapid and unassisted by the database; adults had either remained at or returned to their place of origin, and newly registered unaccompanied children were taken to their place of origin to be identified by family members and thus reunited with them on the spot. Registrations and reunifications were nevertheless documented and subsequently entered in the database. During this phase many previous registrations of unaccompanied children were of necessity " suspended " because they were no longer in the camps; these cases were labelled as such in the database. Some of them were gradually reactivated as children reappeared in a transit camp and were then registered again.

 In appropriate registrations  

In Phase 1 several agencies reported that registration criteria were not being applied systematically and that some registrations should have been discounted. Follow-up in the field revealed that some " unaccompanied children " were in fact accompanied by other adult relatives, or even a parent; these cases were closed on the database. Others clearly knew the whereabouts of their parents, and vice versa, and had been voluntarily placed by parents in the centres run by relief agencies to benefit from food and other assistance given there. These cases were also closed in the database, but often continued to be handled by specialized agencies in view of the social or economic problems faced by the families that had resorted to voluntary placement. In Phase 2, follow-up of all children enabled application of registration criteria to be reviewed on a case-by-case basis. It was then also found that some registered children simply could not be retraced and were unknown at the place of registration. These cases were labelled in the database as " not seen since 1994 " and subsequently as " closed " , after a check on the identities at the given place of origin revealed that they were not recognized by any of the communities'inhabitants. The conclusion was that the names or addresses were fictitious, for in a close-knit, community-based social structure such as that in Rwanda, inhabitants know each and every individual of a community. Failure to recognize a child's stated identity in such a community is highly unlikely. In brief, cases were closed in the database for reasons other than family reunification .


Given the quantity of data collected, the geographical spread, the complexity of the data handling, the extraordinary circumstances, the cost and the multiple agencies present in the field, the ICRC wanted to evaluate the efficacy of the central database in case such a crisis ever arose again. The following aspects were examined: first, the project's overall success; second, how, why and which reunifications were attributable, in the different phases described, to the centralization of data; third, which proportion of the workload was inappropriate; and fourth, whether the approach adopted was an appropriate response to the emergency.


In the database, the following data entered between April 1994 and the end of 1997 were examined by month:

  • the number of children registered as unaccompanied and the number of family tracing requests in the field;

  • the number of children registered and the number of family tracing requests entered into the database;

  • the number of reunifications of children with their families and the agency responsible for the reunification.

The data covering the 45-month period were then analysed in relation to the three phases described above.

With knowledge of how the database was used in the three phases and of which agency - or the family itself - was instrumental in the reunification, a calculation was made as to which reunifications were assisted by the database and which were not. In Phase 1, only family reunifications carried out by the ICRC and by the families themselves can be assumed to have been facilitated by the database ; all others were necessarily carried out without its assistance. During Phase 2, family reunifications can be considered as having been facilitated by the database. During Phase 3, family reunifications can be considered as having been carried out without assistance from it.

The proportion of inappropriate registrations was examined in relation to Phases 1 and 2 only because this was not a primary consideration in Phase 3. The number of inappropriate registrations was calculated by combining the number of children inappropriately registered as unaccompanied with those classified as " not seen since 1994 " ; the latter cases can be assumed to have been subsequently closed after follow-up in the field.

The number of children who refused in Phases 1 and 2 to return to their families despite successful location of their relatives in Rwanda was recorded. Likewise recorded was the number of children whose case was " suspended " at the end of Phase 2, as well as the number of " suspended " cases that were subsequently reactivated in Phase 3, i.e. those relating to children who returned to Rwanda and were re-registered upon arrival in transit camps.

Figures for reunifications of children registered in the Goma area between April to August 1994 were examined separately.

 Results of the audit  

Table 2 shows the overall number of registrations and reunifications.

Of the 56,984 children eventually reunited with their families, 9,547 (16.7 %) were reunited as a result of parents having consulted the database in search of their child with or without having lodged a formal tracing request ; 3,495 (6.1 %) reunifications resulted from the computerized matching system, i.e. the children concerned were both registered themselves and sought as a result of a tracing request sub mitted by the family. As there may be some overlap between these two categories, the total number of reunifications as a result of mere consultation of the centralized database lies between 9,547 (16.7 %) and 13,042 (22.7 %).

The graph in Figure 1 shows, by month, the number of registrations and family tracing requests in the field, their entry into the database, and the number of reunifications of children with their families.

Table 3 shows the proportion of reunifications in each phase that were definitely not assisted by the database, and those that were, or most likely were, assisted by it.

Table 4 shows registrations, reunifications, inappropriate registrations, cases " suspended " and those that had remained active at the end of Phase 2. The " suspended " cases also included 6,771 children in the camps who had refused to return although their families had been traced in Rwanda.

In Phase 3, 17 % of the " suspended " cases were reactivated.

Of the 7,638 children registered in the Goma area between April and August 1994, 4,532 (59.3 %) were reunited with their families; of these 3,591 (79.2 % of those reunited) were reunited in the Goma area.

 Evaluation of the programme  

Reuniting unaccompanied Rwandan children with their families during the period under consideration was fraught with

difficulties. There were considerable cultural and social complexities influencing data gathering and tracing, as alluded to in Table 1. Furthermore, each reunification after registration and tracing entailed physical relocation of at least some of the people concerned to communities which had suffered a tremendous social upheaval and severe trauma. Political sensitivities betwe en and within the dispersed communities exerted strong and complex influences on the process.

The central database assisted in 39.7 % of the reunifications carried out over the entire period, including 16.7 to 22.7 % of those which resulted from computerized matching of a registration and a tracing request, the initially intended purpose. However, the database was most useful for the more difficult reunifications, for example in Phase 2, when the displaced population had ceased to move around and unaccompanied children were found in completely different areas to their families. Although the ICRC may initially have spent a lot of time and resources in establishing the database, it was subsequently able to concentrate its efforts on reuniting unaccompanied children located in the refugee camps with family members traced in Rwanda. But these so-called " cross-border " reunifications involved the difficulties of, and resistance to, repatriation. Children were subject to strong social and political pressure in the refugee camps not to return to Rwanda, and 6,771 repeatedly postponed their decision to return or refused at the last moment. Had these reunifications taken place successfully, the overall number of reunifications attributable to the database would have been larger.

During Phase 1, data entry played no part in the majority of reunifications ; most of them were effected rapidly and directly by aid agencies working in the field who found family members in the same area as the child. Centralization of data on these children did take place, but in retrospect was unnecessary to accomplish these reunifications. Therefore, in the event of a similar emergency the centralization of data should be deferred until it becomes clear which cases would genuinely benefit.

During Phase 2, data were shared with agencies and enabled geographically widely dispersed families to be reunited. The second benefit of the database was that inappropriate registrations could be identified, thereby allowing the overall workload to be reduced and priority given to children who were genuinely unaccompanied. This likewise argues for a delay in data centralization.

 Impact of the circumstances of separation  


The decision to centralize all identities of and data on unaccompanied children was taken in June 1994 in the midst of the genocide and the war. A first reason for registering and keeping a record of children separated from their parents and families was to preserve their identities; the situation was chaotic and there was no way of knowing whether the activities undertaken at that time to protect and care for those children would be successful. The second reason was to lay the groundwork for tracing their parents and/or other family members.

Displacement of the population was seemingly haphazard and widespread; it was assumed that members of separated families would be randomly dispersed. Despite the mass flight of 850,000 people across the border into Goma between 14 and 18 July, the assumption of random dispersal remained unchallenged. It led to a basic reasoning in favour of centralization which argued that by registering both unaccompanied children and families looking for children, the data could be cross-matched. The database reveals, in retrospect, that the assumption of random separation was in part incorrect : first, the number of reunifications carried out in Phase 1 without reference to the database was high; secondly, most of the children registered in the Goma area during the first months of the crisis were reunited in the same area. These facts indicate that although there was large-scale population displacement, whole families, and to a certain extent social structures, moved to roughly the same area; many separat ions were not random but had occurred in a specific geographical area and could be resolved there. This explains the early and rapid success of the direct approach of reunification without reference to the database.

The success of the direct approach in Phase 3 is evident from Figure 1. The numbers of registrations and reunifications in the field reach a peak in the same month. The predetermined destination for all returning refugees was their place of origin ; knowing this facilitated the direct approach because agencies could simply take unaccompanied children to those predetermined places for rapid reunification. Figure 1 shows that data entry occurred two months after registration and reunification. These reunifications therefore could not have been facilitated by the database.

 Assistance programmes as a cause of separations ?  

Of the children entered in the database, 29 % were registered inappropriately. This indicates either manipulation of the agencies or voluntary separation - or both - as well as a lack of adherence by agencies to criteria.

The considerable media attention given to the issue of unaccompanied children early on in the emergency generated donor interest, which in turn encouraged humanitarian organizations to provide specific medical, food and other material assistance. The only way that the larger refugee population could benefit from this focused assistance was to ensure that children were identified and registered as unaccompanied [4 ] . A spiral of humanitarian supply and demand was thus created which, on occasion, was also consciously exploited by organizations wishing to manifest their operational presence. This cycle could have been broken by reorienting the assistance to the general refugee population. Instead, however, both aid agencies and authorities reduced the number of centres dedicated to caring for un accompanied children, thereby reducing the opportunity for exploitation of assistance available for these children.

The database does not provide clear data on how many children were ultimately left without any adult relatives. At the end of 1997, the fact that tracing efforts were continuing for 13,878 children, including those cases which were reactivated in Phase 3, provides a reasonable indication. This is important because it shows that adoption or placement in institutional care, for which there were popular calls, would not have been an appropriate emergency response. Adoption might eventually be appropriate for only a small minority of children initially identified as unaccompanied, and such a decision can be taken only after careful, time- consuming and labour-intensive tracing.

 Concluding remarks and recommendations  

There are clear advantages to having one centralized database rather than several or a collection of partial databases. It enables the overall problem to be assessed and a comparison made of the various agencies'responses, both during the process and as a general overall audit, while applying the same criteria to do so.

The time and resources devoted to establishing a centralized database are considerable. Whereas the direct value of such a database in facilitating family reunifications during an emergency is limited, its value becomes apparent later on for families found to be more widely dispersed, when movement of the population has ceased and when families separated but in the same geographical area have been reunited by a direct approach. In addition, the database may indirectly assist and facilitate family reunifications by permitting easy identification of groups and trends.

All agencies must coordinate their efforts and adhere to clear and strict criteria for registration of unaccompanied children.

Alongside measures to identify and assist unaccompanied children while tracing their parents or other adult relatives, appropriate assistance must be provided for the whole population.

 Table 1  


Specific to people in Rwanda:

  • a family name common to family members is extremely rare;

  • most people at best know their year of birth; only 2.5% were able to state their exact date of birth, i.e. including month and day;

  • a Rwandan’s identity (name) is usually made up of one or two first names of either Christian or Islamic origin, followed by a

Kinyarwanda term or name; the fact of having the same name or names as someone else does not imply kinship.

To distinguish between individuals and establish a possible family relationship, the database includes, inter alia:  

  • not only the person’s name but also those of both parents as well as the place of origin;

  • a definition of the place of origin of each individual in terms of four hierarchical administrative units (prefecture, sector, community, and cellule – a subdivision of a community) which are significant components of identity in Rwandan society;

  • a standardized spelling of names and places; a dictionary of names was developed which at present contains more than 138,000 possible names of people and an index of 1,980 possible place names in the region.



Table showing some of the complications encountered in setting up and using a database with 271,297 identities for reuniting unaccompanied Rwandan children.

 Table 2  

 Children registered as unaccompanied  




 Children reunited with their families  




  • with assistance of database




without assistance of database




 Children registered and not reunited  




including cases closed/cases suspended




including cases for which tracing efforts are continuing




Table showing the total number and percentage of children reunited with their families according to use or not of the centralized database, as well as the number an d percentage of reunifications for each category per phase and the percentage per phase and category of the total number of reunifications.

 Table 3  



% of total

% by phase

Total: Phases 1,2 & 3




- most likely with database




- definitely without database








 Phase 1  




- most likely with database




- definitely without database








 Phase 2  




- most likely with database




- definitely without database








 Phase 3  




- most likely with database




- definitely without database




Table showing the total number and percentage of children reunited with their families according to use or not of the centralized database, as well as the number and percentage of reunifications for each category per phase and the percentage per phase and category of the total number of reunifications.

 Table 4  

Children registered



Family reunifications



Inappropriate registrations



- of which: closed/out of criteria



- of which: " not seen since 1994 "



Cases suspended



Cases still active



Table showing the number and percentages of children registered as unaccompanied, reunited with family or inappropriately registered, and of cases suspended or still the subject of active enquiries at the end of Phase 2 (December 1997).


Graph showing month-by-month development over a 45-month period (April 1994 - December 1997) of :

- the number of children registered in the field as unaccompanied;

- the number of tracing requests formally submitted in the field by families searching for a child;

- the number of registrations and tracing requests entered into the database;

- the number of reunifications in the field of children with their families.

Phases 1, 2 and 3 are separately indicated in the graph (for additional explanations see text).


1. Convention on the Rights of the Child, of 20 November 1989.

2. Fourth Geneva Convention relative to the Protection of Civilian Persons in Time of War, 12 August 1949, in particular Article 50 ; Protocol additional to the Geneva Conventions of 12 August 1949, and relating to the Protection of Victims of International Armed Conflicts (Protocol I), 8 June 1977, in particular Articles 77 and 78 ; Protocol additional to the Geneva Conventions of 12 August 1949, and relating to the Protection of Victims of Non- International Armed Conflicts (Protocol II), 8 June 1977, in particular Article 4, para. 3.

3. Joint Statement of the International Committee of the Red Cross (ICRC), the Office of the United Nations High Commissioner for Refugees (UNHCR), the United Nations Children's Fund (UNICEF) and the International Federation of Red Cross and Red Crescent Societies (International Federation) on the Evacuation of Unaccompanied Children from Rwanda, Geneva, 27 June 1994.

4. Sarah Uppard and Celia Petty, Working with Separated Children: A Field Guide, Save the Children, London, 1998.

Abstract in French