ロジスティックとの関わり

Backround

Following the death of the Rwandese President in a plane crash in April, Rwanda has experienced three months of civil war and genocide. Over 5 million of the countries estimated 7.3 million people are believed to have been displaced from their homes. At the onset of the conflict, the RPF (Rwandan Patriotic Front) controlled a small fraction of the country in the north along the Ugandan border. RGF (Rwandan Government Forces) controlled the remaining areas of the fertile, mountainous nation.

A wave of ethnic violence began and RPF forces responded quickly with an insurgency from the north and quickly took control of much of the central and eastern portions of the country. This drove the RGF forces and much of the civilian population west, and to a lesser extent, south. By mid-July, the remaining government forces had been pushed west to the areas adjoining Zaire.

Beginning on July 14th, large numbers of RGF forces and civilians began fleeing into Goma, Zaire, in anticipation of RPF victory in the north-west of Rwanda. Over a three day period, it is believed that about 700,000 people crossed into Zaire in the Goma region. Goma, a town of 130,000 inhabitants was overwhelmed. Most of the refugees arrived initially in Goma, congesting roads, denuding foliage, and washing, bathing, and collecting water along the shores of Lake Kivu. The refugees were quickly moved to available areas north and west of the town which were acquired, and in some cases prepared, under the direction of UNHCR. The region is sheeted with lava outflows providing few water sources and is largely prohibitive to the digging of graves and latrines. The main water source in the region is Lake Kivu, which was within walking distance of only 1 of the 4 major camps.

By July 20, MSF was reporting that there were 1000 cholera related deaths occurring per day. Pictures of streets lined with uncollected bodies shocked the world. Large numbers of NGO’s moved to Goma to assist, and a cholera task force was formed by UNHCR. Part of the task force’s mandate was to document the outbreak and it’s progress. While the outbreak continues, it is likely that a vast majority of the cases have already occurred and thus, the conditions contributing to the outbreak and the magnitude of the health impact are not expected to change significantly in the coming weeks.

The Outbreak

The exact date of the first reported case of cholera is not known. As the North Kivu Region is endemic for cholera, identifying a first case may be of little value. As mentioned, MSF claimed that there were 1000 cholera related deaths by July 20th, and samples taken at this time were confirmed in Amsterdam as being vibrio cholerae 01, bio-type El tor, sero-type Ogawa. A crude surveillance system, by which clinics and NGO’s estimated the number of patients seen per day was instituted beginning on July 21st. By July 27th, this system included only actual patient tallies from rehydration centers and clinics.

Numbers of reported clinic and hospital patients in acute need of rehydration in Goma town and in the 4 major refugee camps surrounding Goma are presented below. Between July 21 and August 9, 60,848 patients were admitted for rehydration.

Unfortunately, these reports include cholera, dysentery and dehydration cases as a combined total. On July 24th, in a non-scientific examination of the cholera camp at Mugunda, WHO and UNHCR staff believed that 40% of the patients in the rehydration unit were simply exhausted and dehydrated. The MDM medical coordinator in the camp agreed with this estimation. On the same day, the MSF-B medical coordinator estimated that 30% of the patients in the MSF rehydration centers were simply dehydrated.

Assumptions:

To estimate the actual number of clinic reported cholera cases, the cases of exhaustion/dehydration, and the cases of dysentery need to be subtracted from the rehydration patients. In order to correct for the exhaustion/dehydration patients in the reports, several assumptions need to be made: a) That 35% of all reported cases on and before July 24th were simple dehydration and exhaustion. b) That when an NGO started distinguishing cholera from dysentery in their daily reports, all reported cases were truly cholera or dysentery. c) That the decrease in simple exhaustion/dehydration was uniformly incremental between the 35% on July 24th and the 0% on the first date of cholera/dysentery distinction in the daily reports.

To compensate for the inclusion of dysentery in the reported cases, the fraction of cases caused by dysentery between the first clinic daily report of total rehydration patients(July 21 or 22) and the first survey or daily registration providing dysentery as a fraction of clinic patients was considered to be constant. This means that in Mugunga from July 21 to July 27, 19% of reported cases were assumed to be dysentery. In Munigi, 24% of reported cases were assumed to be dysentery between July 21st and 27th. In Katale, this fraction was 20% from July 22nd to the 24th, and in Kibumba the fraction was taken as 13% from July 22nd to the 29th. In the 22 Goma town clinics, the fraction of dysentery cases was extrapolated from data available in 2 clinics, Mudja and Ndosho.

Days on which no data were reported (Mugunga Camp only) were assumed to have seen a number of rehydration patients equal the average of the day preceding and following. If one NGO provided combined cholera and dysentery figures while the other NGO’s reported them separately, the fraction of cases which were from dysentery in the combined figure was assumed to be equal to the fraction seen in the other NGO’s clinics.

For purposes of mortality estimation, it is assumed that there were 700,000 refugees in the Goma region over the course of the outbreak. This is a crude figure based on the present estimate of about 550,000 in the camps, 50,000 near the roads and surrounding areas, and 100,000 which have died or been repatriated.

Estimated Cholera Cases & Deaths in Reporting Clinics:

Thus, a best estimate of the number of cholera cases seen in the reporting clinics is shown above. Note that the peak of the curve, July 26th, is the same in both graphs but that the peak of the estimated cholera cases curve is somewhat more pronounced. Between July 21st and August 9th, there was an estimated 36,464 cases seen by the reporting clinics. Half of these cases were seen by July 27th.

During this same period, clinics reported 4090 in-clinic deaths associated with cholera, dysentery, and dehydration. While an estimated 60% of the rehydration patients over this period are believed to have had cholera, estimating the fraction of deaths due to cholera is particularly problematic because the cases fatality rates for the three conditions were probably very different. Thus, the total number of in-clinic deaths from cholera is certainly less than 4000, and the in-clinic case fatality rate was probably less than 10 percent.

Overall Mortality:

Data on non-clinic mortality over the course of the epidemic is limited. Data are available from agencies collection corpses, but some of these values are suspect because the collectors may have inflated figures thinking that they would be payed a fixed amount per body collected. At some, but not all grave sites, cadavers were counted as they were buried. Finally, 2 surveys examining mortality have been completed in Katale and Kibumba.

The number of corpses reportedly collected by NGO’s is presented here. Especially during the first days of corpse collection, transportation limitations meant that bodies were often not collected until several days after the deaths. Thus, this curve probably lags behind a true mortality curve. The peak of the corpse collection curve is on July 28th, 2 days after the peak in reported cholera cases.

Over the period July 21st to August 9th, 46,505 corpses were collected. As the refugees arrived primarily between July 14th and the 17th, these collection figures correspond to a period beginning around July 15th and ending August 8th. Over this 25 day period, this accounts for a crude mortality rate (CMR) of 26.6 per 10,000 per day if the at risk population was 700,000. While some of these reports may have been inflated, most are believed to be accurate and these figures do not account for private burials.

A survey done by MSF-H in Katale (pop. 80,000) found that over the period July 15th to August 4th, the population had experienced a CMR of 41.3/10,000/day. A similar survey conducted in Kibumba (pop. 180,000) found a CMR of 28.1/10,000/day over the period July 14th to August 8th. As mortality had decreased dramatically by August, and as the 2 lower CMR figures include this period of low mortality, all three values are roughly comparable. In the Katale survey, 90% of deaths were associated with diarrhea while the figure was 85% in Kibumba. Estimating the specific fraction of deaths attributable to cholera is not possible. A conservative estimate that 60% of all diarrhea deaths were assumed to be from cholera given that: 60% of rehydration patients during this period are estimated to have had cholera, the case-fatality rate for cholera probably exceeds that of simple exhaustion, and less than a third of diarrheal illnesses were recorded as dysentery until August when overall mortality had fallen. Thus, using the two camp surveys, the population weighed average fraction of deaths attributable to diarrhea is 86.5 percent. If 86.5% of the 46,505 corpses collected were from diarrheal deaths, and 60% of those were from cholera, this implies that at least 24,136 of the corpses can be attributed to cholera. To get a total estimate number of cholera cases, the cholera deaths and the reported cases should be added together without double counting those who may have visited a clinic and then died. If only 10% of the 36,464 clinic reported cases died (the highest daily in-clinic death/case ratio seen), this implies that about 56,950 non-fatal cases or deaths from cholera were recorded over this period, or an attack rate of 8 percent. As the estimates are conservative, and reporting of cases and deaths incomplete, the actual attack rate was probably somewhat higher.

Influence of response measures:

The Goma cholera epidemic was met with a massive response by the relief community. Without community based studies to examine the case fatality rate for untreated cholera in this population, no estimate of the impact of treatment measures can be made. Disorder and insufficient supplies of ORS in the clinics was common in the earliest days of the response. This was in part due to the vast concentrations of people in the camps and the chaos and congestion such human concentrations produced. Often, most of the water, and in some cases, most of the ORS was being consumed by healthy residents who simply were thirsty. Data from within the camp clinics imply that case fatality rates were high (>10%) but rapidly decreased with increased organization and increased resources.

A priority response by the relief community was the rapid provision of water. On July 26th, it was estimated that water collected in Goma at the lake by refugees was 10 times greater in volume than the water transported by trucks. Thus, in terms of water quality, it is likely that most refugees were receiving untreated water before moving to the camps. The figure to the left shows the amount of water, in terms of liters per person per day being consumed in the three main camps in the Goma region. While the effort and organization involved in this increased supply is remarkable, this figure shows that the increased provision came too late to affect the course of the cholera epidemic. The lag time associated with water provision was largely unavoidable given: the lack of available tankers in Eastern Zaire in late July, the rock sheeting and lack of water sources near the initial northern camps of Munigi and Kibumba, and the vast numbers of people who needed to be served. Yet, this supply probably ended the significant occurrence of death due to simple dehydration and may have a considerable benefit with regard to the ongoing dysentery epidemic as well as for preventing a host of diarrheal illnesses.

Likewise, the provision of latrines was insufficient and too late to influence the course of the cholera epidemic. As of August 12th, UNHCR estimates that there is only 1 latrine per 1029 inhabitants in Mugunga, 1 per 500 in Kibumba, and 1 per 184 persons in Katale. Soap, the main tool used to promote personal hygiene, was not distributed in the camps until well into August.

Conclusions:

  1. A cholera epidemic struck the predominantly refugee population surrounding Goma, Zaire in mid-July, 1994. By August 9th, at least 8% of the population had visited a hospital or died as a result of cholera infection.
  2. Given that the vast majority of cholera infections (about 90%) are usually asymptomatic, the course of the epidemic was probably controlled exclusively by population immunity.
  3. The considerable measures regarding water supply provision made by the relief community probably did not influence the course of the epidemic. Geological conditions, logistic constraints, and a lack of available local resources made the provision of the requisite 15-20 liters per person per day a virtual impossibility. Measures to improve sanitation would have had to be in place before the camps were occupied or in the initial couple of days in order to have a significant impact on cholera transmission. But, the continued latrine construction may affect ongoing or other potential epidemics.
  4. More than 80% of the reported cholera cases and deaths to date occurred before August 1st. This implies that few of the refugees at present pose a significant health risk should they repatriate to cholera free areas.
  5. Given the local geography, geology, the health conditions of the refugees, the suddenness of the refugees arrival, and the endemicity of cholera in the region, the cholera epidemic in Goma was probably unavoidable. Although, the extent of the outbreak and number of deaths may have been diminished through a cholera preparedness plan and its implementation.

Recommendations:

  1. Future potential responses to cholera outbreaks should focus on prevention in the earliest phases. In future mass arrivals to cholera endemic areas, immediate responses (which require few resources) should include: bucket to bucket chlorination brigades to chlorinate surface water collected by individuals, preparation and protection or reservation of defecation fields in the selected camp sites.
  2. Latrine construction program should be seen as a priority, ideally using refugee labor. The need for this service remains in the Goma region, especially in Mugunga and Kibumba Camps.
  3. Clinics should organize their service and secure their areas before starting to treat patients. Preparation of a site and planning of patient triage and flow should be done by experienced personnel. As is done by many NGO’s, staff with experience in rehydration of cholera patients should be identified on rosters so that NGO’s and agencies can optimize treatment in the early days of an epidemic, before the majority of patients present to the clinics.
  4. Community outreach networks should be a first line response and should be arranged as soon as possible. The workers are invaluable because they can: give ORS in the community diminishing the workload in the clinics, they can increase the fraction of seriously ill people who get to the clinic, and they can give NGO’s a mechanism for implementing various programs such as latrine construction and distribution.
  5. Sites of potential influx (such as Bukavu at present) should develop cholera preparedness plans and request appropriate emergency response materials (8kg chlorine powder per 100,000 expected per day, WHO cholera treatment kits or equivalent, money for paying 1 chlorinator per 1000 people, resources to pay 1 health care worker per 1000 people, water storage and transportation materials, and ideally, acquisition of sites with water sources and reserved defecation zones).