A 5-year war that ravaged the Democratic Republic of Congo (DRC) officially ended with the endorsement of peace agreements and withdrawal of troops in 2003. The war had a devastating impact; millions of civilians died, many due to lack of access to health care[1–5]. Although a national peace process has been held and elections were conducted in 2006, the eastern regions of the country have yet to see the end of hostilities and in 2009, for the 11^th consecutive year, DRC was featured on a list of the Top Ten Humanitarian Crises[6].

In the province of North Kivu, in eastern DRC, conflict has never ceased. Intermittent fighting continues between the national army and multiple, shifting rebel groups. Frontlines move, leading to continual instability and population displacement. In August 2008, after months of relative calm, offensives restarted, resulting in mass population displacement.

The non-governmental organisation (NGO) Médecins Sans Frontières (MSF) has run health-care programmes in this region for many years. Services provided in collaboration with the Ministry of Health vary between sites, and can include primary and secondary health care, emergency surgery, therapeutic treatment for malnourished children, response to epidemics (eg, cholera), mental health care, and care for individuals subjected to sexual violence. These services are provided for both internally displaced persons (IDPs), mainly living in camps, and for resident populations. Additional health services and support, such as the provision of water and latrines, and the distribution of essential non-food items such as jerry cans and plastic sheeting, is provided by a range of actors (Solidarité, Merlin, MSF, UNICEF, Save the Children).

In May 2009, to understand the impact of the renewed conflict on selected civilian populations, and to assess their access to health care in order to orient pre-existing humanitarian aid, surveys were carried out in the catchment areas of three MSF programmes in North Kivu. They covered the period from the last major offences in September 2008 to the time of the surveys in May 2009.

Three cross-sectional surveys were undertaken, one in each of three geographic areas supported by MSF. Both IDP camps and local villages were included.

The results of these surveys are representative of an estimated population of 200 000 inhabitants. The surveys were conducted between May 4 and May 23 2009. In Masisi, one cluster could not be completed because fighting broke out in the village where the survey was being conducted and the survey team was forced to evacuate. The cluster was not replaced. In Kabizo, eight households (1.5%) refused to participate in the survey and 48 (8.0%) were absent; in Masisi nine (1.0%) refused and 29 (3.0%) were absent; and in Kitchanga 18 (1.8%) refused and 55 (5.1%) were absent.

In Kabizo, 92 episodes of violence against individuals were reported, including two women who reported being raped. In Masisi, 914 episodes were reported, including 103 rapes (11.3%, 95% CI: 8.2 - 16.1). In Kitchanga, 922 episodes were reported including 44 rapes (4.8%, 95% CI: 3.3 - 6.3). Forced labour was the most commonly reported type of violence against individuals in Masisi 41.9% (383/914, 95% CI: 35.3 - 48.7) and Kitchanga 51.5% (475/922, 95% CI: 44.1 - 58.9). Beatings were the most commonly reported act of individual violence in Kabizo: 66.3% (61/92, 95% CI: 54.7 - 76.2). At all sites, over 95% of perpetrators were reported to be a combatant (rebel or military) or an "incivique". In Kabizo, 11.6% (63/542, 95% CI: 8.8 - 15.3) of households reported at least one household member being subjected to violence during the recall period; in Masisi 99.1% (897/905, 95% CI: 98.2 - 100.0); and in Kitchanga 50.0% (509/1018, 95% CI: 46.8 - 54.1).

Theft or destruction of property was also common during the recall period. In Kabizo 87.7% (451/514, 95% CI: 84.4 - 91.2) of households reported theft or destruction of major items; in Masisi 96.0% (869/905, 95% CI: 92.2 - 98.0); and in Kitchanga 57.4% (585/1018, 95% CI: 48.8 - 65.6).

Access to fields was either not possible or limited for 55.4% (300/542, 95% CI: 51.5 - 59.2) of households in Kabizo, 41.3% (186/451, 95% CI: 32.9 - 50.3) in Masisi, and 79.7% (811/1018, 95% CI: 73.5 - 85.9) in Kitchanga.

At all sites, 70% of households reported having at least one member fall ill and require care outside the home in the 2 weeks prior to the survey. In Kabizo, 94.8% (364/384, 95% CI: 92.0 - 96.6) received care, of whom 84.4% (/359, 95% CI: 80.4 - 87.7) did not pay for their consultation or treatment. In Masisi, 81.7% (515/630, 95% CI: 75.9 - 87.5) of the sick received care, of whom 87.6% (451/515, 95% CI: 83.4 - 91.6) did not pay. In Kitchanga, 89.8% (651/725, 95% CI: 86.3 - 93.3) of the sick received care, of whom 88.9% (579/651, 95% CI: 86.5 - 92.8) did not pay.

Our results, obtained in insecure field conditions, show the ongoing impact of the conflict on civilian populations, in particular the violence inflicted upon them. Although mortality rates were under emergency thresholds in our surveys, the proportion of violent death was high, reaching over 30% in the global population in two sites and 58% and 71% in the population 5 years and over in the same sites. While the CMRs in our surveys were lower than that of other surveys undertaken in camps during acute crises (4.1/10 000/day in Ituri, DRC 2005), they are similar to results reported from eastern DRC as part of a nationwide survey carried out in 2006 - 2007 [4, 10], The proportion of deaths reported as violent in two of our survey sites is similar to the 67% reported in the Ituri survey, but much higher than the 0.6% violent deaths, reported in the cumulative results of the eastern region in the 2006 - 2007 survey[4, 10].

Our results may under-represent the true extent of deaths due to violence (and violence against individuals), since some villages in Kabizo and Masisi were excluded from the survey for security reasons. These sites, and that from which the team was evacuated, were likely to have been more strongly affected by violence than those in which the survey could be conducted. In addition, the MSF programmes cover a wider more inaccessible area than could be covered in the surveys. The programmes include areas only reachable by mobile teams supporting government health centres that cannot be accessed on a regular basis because of insecurity. These are populations that continue to be affected by conflict, are constantly moving, and likely have high levels of trauma and mortality, as yet unrecorded.

The geographic situation of Kabizo may be linked to the few reported deaths from violence. During the recall period of the survey the relatively large town of Kabizo was overrun and residents fled when the first skirmishes occurred, most escaping the violence; no further attacks occurred in the survey area. In Masisi and Kitchanga, numerous offensives were conducted on multiple villages, which might have led to less warning (as demonstrated by the unanticipated attack on the village in which one survey team was working), resulting in more fatalities.

We did not carry out verbal autopsies in the field. For that reason we present the reported causes of death as either violent or non violent, since we consider that respondents could reliably report this information.

Our results, unlike others obtained in eastern DRC, show mortality linked to disease well below emergency thresholds[2–4, 10]. However, our survey sites were not representative of all of North Kivu and reflect the situation of only selected populations for whom humanitarian aid had been in place for at least a year. The medical care provided to these populations was comprehensive and free of charge, and our results suggest that access to care was good. Although not quantified in our survey, water and hygiene activities were also in place in all sites. Together, these activities could have contributed to the disease-related mortality being lower than that reported in other surveys, and demonstrate that high disease-related mortality rates are not inevitable.

The exposure of the population to violence is also revealed in the number of violent events perpetrated against individuals. Few are exempted, with nearly half the households included in our survey affected during the 8-month recall period.

The episodes of violence against individuals, particularly rape, are likely to have been under-represented in our survey results. Rape is still a taboo subject in many communities of DRC, and reporting can have severe negative consequences for the victim. In our survey nearly all perpetrators were identified as combatants, and there may be under-reporting of violence perpetrated within the community. Another limitation in the data on violence is that we only recorded one type of violence per violent episode. Although this limitation does not affect the number of episodes, it might under-represent what are considered less severe forms of violence--eg, an episode of violence might be recorded as rape when the person was subjected to both rape and beating.

Our results also reveal the prevalence of forced labour in two of the survey sites. Civilians, usually men, are used to carry material for combatants, frequently for long distances and usually under threat of violence if they do not comply. The frequency of forced labour in our survey areas is higher than reported previously[2].

The high proportion of households who had basic non-food items, livestock, or food-stocks stolen, and who had limited or no access to their fields exposes more of the precarious existence of these populations. While in these sites NGOs have generally replaced essential items, the same might not be true for other populations of North Kivu. The poor access to fields and loss of livestock also suggests that households may struggle to meet even basic needs in the short and medium terms. Although we did not collect specific information about why fields could not be accessed, it might be linked to the distance the population have been displaced. In insecure conditions, individuals are unlikely to stay overnight at their fields, which would be necessary if they are more than a few hours walk from their current residence.

Our results show the continual instability of the population, many of whom had fled at least once during the 8-month recall period, many more than once, and most as a result of direct attacks on their homes at the time of the displacement. This is in contrast to rumours within the international community that villagers had been fleeing pre-emptively, before combatants arrived. In general, the displaced did not flee far from their homes, most walking for less than 2 days before settling. An area of direct conflict may be located very close to an area where the population settles. The challenge faced by humanitarian organisations is to provide aid for the dispersed populations and to ensure the security of their teams.

The prevalence of violence as seen in our survey results is high and reflects what is reported by field teams. The MSF programmes see many survivors of sexual violence (in one site, an average of 68 per month from a population estimated at around 140 000; L Berryman, personal communication), although it is thought that the team manages to reach only a small proportion of survivors due to lack of access to information and issues of stigma and confidentiality. MSF mental-health programmes in the region treat people who have commonly experienced a combination of different traumatic events: having to flee and hide from enemies, houses being destroyed, family members, neighbours, and friends being killed or disappearing, witnessing someone being killed or raped, being raped, and properties, livestock, and fields being confiscated. Their main complaints are anxiety related, with sleeping disorders and intense psychological distress the most frequent symptoms (L Berryman, personal communication).

The results of these surveys are both positive and negative. Positive, in that access to health care was good, and disease-related mortality was lower than shown in previous surveys. Negative, in that the direct consequences of violence were experienced by an extremely large proportion of the population. The civilian population suffers high levels of violence, is regularly displaced, and has property stolen. These results show that, although high disease-related mortality rates are not inevitable, the population of North Kivu continues to suffer the appalling effects of this devastating conflict, despite the peace agreements and elections.