Analysing surveillance records and hospital statistics allowed us to identify achievements and challenges in the EHF outbreak response activities in Masindi district.
With a single exception due to a HCW's refusal to be hospitalised, SEBOV did not invade the community at large (Figure ). Instead, the outbreak at community level was limited to the extended family of the index case. The outbreak response efforts have, in all likelihood, significantly contributed to this containment. The acclaim for this should be shared between the outbreak response team for its vigorous efforts to trace cases and follow up contacts, and the community for imposing a quarantine on the index family. In our experience, international experts are usually reluctant to recommend quarantine measures, being concerned by human rights issues, fearing to antagonise the quarantined community, and arguing that quarantine is difficult to enforce. However, when the community itself imposes quarantine, its effectiveness may be more straightforward to ensure. Somewhat reluctantly the response team endorsed the quarantine, knowing that it would be difficult to convince the community to give it up, and hoping that it would not only decrease social mixing and exposure of susceptible individuals to SEBOV, but also reduce conflicts between the index family and their autochthonous neighbours. Surveillance officers convinced the community to allow the family to fetch water from the nearest well outside the family's compound, and made the quarantine more acceptable to the index family by providing food and supplies, which the family members were prevented from purchasing at Kaduku Trading Centre. Far from advocating community imposed quarantine as a standard outbreak response strategy, we acknowledge that in this specific outbreak it appeared to be effective in preventing the spread of SEBOV into the community at large, and in defusing tensions between the affected family and its neighbours.
A related achievement was the avoidance of transmission chains of unknown origin. While transmission could not always be prevented, when it occurred it was at least immediately clear where it had originated: all cases came from the population of contacts under surveillance. The absence of unknown contacts makes surveillance much easier, as the many cases of acute febrile illnesses without prior contact and without more specific symptoms are then unlikely to be EHF cases. This saves resources, and avoids feelings of being overpowered by the epidemic.
African traditions are likely to have prevented transmission to young children in the index family (Figure ), the only paediatric EHF case probably originating on the Ebola ward. As described for other filoviral haemorrhagic fever (FHF) outbreaks, the tradition to keep young children away from ill family members seems to have protected the children from transmission of SEBOV [15
The absence of nosocomial transmission to patients in Masindi, with the possible exception of a single transmission on the Ebola ward from mother to infant, was another achievement. This was again an effect of early case detection and isolation, which prevented EHF patients to seek care from unsuspecting HCWs in facilities other than the Ebola ward. Infection control on the Ebola ward was then sufficiently stringent to prevent SEBOV transmission to non-cases.
The mobilisation of the local government needs was successful. The local councils stood behind the disease control team unwaveringly, even during adverse community reactions. The community, sub-county and district level proved to be particularly important structures for the support of outbreak response activities: they liaised with the community including the index family, organised the cooperation with other sectors (education, environment etc.), and facilitated the collaboration between local authorities and the various international players.
Though the difference in CFP between patients who were hospitalised in Masindi and those who were not is statistically not significant, it suggests that hospital treatment of EHF might be of some use for the patient - that isolation is useful to halt an FHF epidemic is beyond doubt [12
The presence of the field laboratory in Gulu has set a new standard for the control of major filoviral HF outbreaks. The availability of Ebola specific laboratory results within 24 to 48 h after blood sampling permits taking case management decisions based on laboratory confirmed diagnoses, while otherwise the role of the laboratory is limited to the concurrent or retrospective analysis of the epidemiological development. The result of a blood sample taken just after admission allows for the swift decision whether the patient needs to be isolated on the "confirmed" section of the Ebola ward, or whether he can be treated under normal conditions on a general ward. It is thus possible to reduce the risk of transmission to patients who have been isolated on clinical and epidemiological grounds but are de facto not EHF cases. A second test after clinical improvement allows for earlier discharge from the Ebola ward conditional on antigen clearance, thus assuring safety for contacts of convalescent patients, avoiding unnecessarily long stays in the isolation unit, decreasing stress levels for the convalescent, and reducing the work load for staff [19
]. Clearly, to have a field laboratory on site is helpful to the outbreak control efforts.
In the aftermath of the outbreak, the hostile attitude towards the index family subsided and made way for expressions of solidarity. For instance, the local group of a service club organization donated school uniforms for the orphans and contributed to the acquisition of a maize mill to compensate for the loss of labour and to help the family generating some income.
Challenges in the community
During times of crisis, the local tradition required that members of the index family stay even more closely together than they do anyway, taking food from the same plates, and sleeping close to each other in the same few huts at the centre of the compound rather than living scattered over an area of approximately 4 ha as in normal times (Figure ). This, together with giving care to sick family members, was obviously not conducive for infection control. For a long time, the eldest son of the deceased head of the family refused to comply with the response team's advice to minimise contact within the family, and to disperse in the various family houses instead of gathering around the central ones. The notion of infection and transmission was rejected, instead the son was convinced that the family had been poisoned as immigrants by its autochthonous neighbours, and advised the other family members accordingly. The concept of poisoning is widespread in sub-Saharan Africa, and gained plausibility at this occasion by the fact that autochthonous neighbours who lived among the index family remained unaffected, and by the somewhat hostile reactions the index family experienced during the outbreak because of its Kenyan roots. Its siege mentality was probably further deepened by the sudden loss of the elders, which was, particularly in the traditional African context, a significant challenge to the family's social fabric.
Members of the response team from European, American, or African countries other than Uganda acted as "trusted strangers" and go-betweens for the index family and autochthonous neighbours. They were thus able to prevent escalation, for instance when they discouraged local surveillance offers from carrying out their duties under the protection of armed guards. It is interesting to consider the circumstances how the index family's attitude eventually became more cooperative. Despite daily visits by the outbreak response team, and considerable efforts to persuade the index family to follow the team's advice, the situation only improved after the death of the eldest son, when the family members eventually accepted to give up their daily congregations and to stay in their individual houses instead (Figure ). The key intervention to change the family's attitude, however, was not undertaken by an epidemiologist or anthropologist, but by the driver of the surveillance vehicle: He held up the front page of a national newspaper ("New Vision", 6 Dec 2000), where the EHF death of a popular Ugandan doctor in Gulu made the headline "Ebola kills Dr. Lukwiya", and told the family: "Look: you are not the only ones who are affected!" It is difficult to plan for having the right intuition in the right moment - but flat hierarchies may help good ideas to emerge.
The general population of Masindi district, even at a considerable distance from the index family, was frightened. Panic reactions occurred, when a stampede was triggered by the rumour that an EHF patient wandered the hospital grounds: "100 patients flee Masindi Hospital over Ebola fear" was the newspaper headline of the day ("New Vision", 9 Dec 2000). Fear made it very difficult, at times impossible, to recruit community volunteers for activities like digging graves; international experts thus had to take on this role. Fear lead to acts of sabotage, when villagers filled in graves over night which the military had dug in advance. Fear resulted in discrimination, when HCWs found themselves banned from shops and market places. Fear gave rise to aggression, when a violent demonstration protested against the presence of EHF patients in Masindi hospital, accusing local politicians to accept them for money, or when HCWs found their house burned down after returning from work. Fear revived rivalries between Masindi and Kiryandongo inhabitants, when the former resented that patients from the index family were taken to Masindi hospital although Kiryandongo was closer by. Finally, fear alienated neighbours, who had lived peacefully with each other for decades, and locals demanded: "Those Kenyans who brought Ebola here - send them home!"
It is very important to be mentally prepared for such adverse community reactions, and to avoid anything which inflames the situation further. For instance, by the time health educators reach the villages, the community often knows already that EHF is a dangerous disease, so there is usually little point in stressing how dreadful and deadly it is. It is not helpful either to falsely state that "there is no treatment for Ebola", neglecting the availability of supportive treatment. Such statements discourage patients to accept isolation in the hospital, and can frequently be found in the media or in health education material. Instead, the community needs to be informed how its members can protect themselves, how EHF can be recognised in its early stage - particularly, that haemorrhage is not necessarily present -, and what to do when somebody falls ill with symptoms compatible with EHF. The community needs confidence instilled, not fear.
Challenges in the health system
Recruiting public HCWs for the follow-up of contacts was difficult for fear of contamination; working in mobilisation teams in communities where transmission had not yet occurred was more popular. Even more difficult was the recruitment of clinical staff for work on the isolation unit for fear of occupational transmission of Ebola virus. The few HCWs who volunteered were quickly overworked. Staff meetings were called in, appealing to the solidarity of HCWs not to let down their volunteering colleagues. An agreement was reached resulting in more hospital staff working on the Ebola ward, but only for a few days each. While this strategy improved the situation, the resulting high turnover had two downsides: it put a considerable burden on expatriate staff for training a high number of hospital staff, and it prevented hospital staff from accumulating experience that would have had a positive effect on confidence and work safety.
Fear remained a major factor among HCWs. While more staff now agreed to work on the Ebola ward, not all of them trusted the protective gear and dared to get close enough to the patients for providing nursing care and supportive treatment. Oral rehydration is seen as an important component of supportive treatment [18
]. However, when weak or confused patients lie in basic hospital beds without back support, they depend on others for helping them drink. Insufficient assistance for oral rehydration may have contributed to the initially high CFP, which with 76% was higher than what was observed in Gulu (51%, computed from own data and [3
]) and other SEBOV outbreaks (Sudan 1976, 53% [1
]; Sudan 1979, 65% [2
]; Sudan 2004, 41% [6
]). The much lower CFP after reinforcement arrived (20%) may be explained by improved quality of care, particularly improved rehydration, due to better staffing, reduced fear in and enhanced self-confidence in HCWs. This explanation is in line with observations of HCWs' practice on the ward. Alternative explanations for the difference in CFP include a lower viral load in occupational cases and virus attenuation.
HCWs' fears were sustained by the continuing occurrence of occupational transmission after the introduction of barrier-nursing. For most cases, breaches of barrier nursing by absent-minded or careless HCWs could be identified retrospectively. A recipe against absent-mindedness may be the 'buddy' system, which was introduced in later outbreaks, whereby each HCW touching a patient and becoming contaminated is accompanied by a second HCW who monitors the compliance with safety procedures and warns his buddy if a breach is imminent. Special attention should be given to the training and supervision of nursing aides, cleaners, drivers etc. When HCWs are careless and refuse to heed advice, they have to be prevented from working on the isolation ward for the safety of themselves and their colleagues.
Health care for EHF patients collapsed towards the end November, when the influx of patients increased drastically (9 patients within 3 days) and deaths from the virus increased including six HCWs and patients in the Ebola ward. These events revived fears and even panic among many staff members. Clinical staff stayed away from the Ebola ward, drivers absconded, and community volunteers could no longer be recruited, for digging graves (a no-risk activity) or as cleaners. When the crisis reached its maximum, there were four dead bodies on the Ebola ward for more than 48 h. Because body fluids tend to leak out of corpses in abundance, dead bodies are a significant source of contamination and must be disinfected, put into a body bag and buried swiftly. Furthermore, the prolonged presence of a dead body on the isolation ward is frightening and appalling and undermines the willingness of hospital staff to work on the Ebola ward and of probable cases, namely among HCWs, to accept hospitalisation. The nosocomial transmission of SEBOV from mother to child likely happened in this phase. Only the prevailing confusion can explain why the infant was not separated from its mother suffering from EHF.
At this point, the closure of the Ebola ward in Masindi, and the transfer of all patients to Gulu was considered by the District Ebola response team as last option. Fortunately, this measure was averted by the arrival of a high-level delegation from the MoH, and of a team from Gulu district consisting of senior international experts and experienced local staff. The direct interaction between high ranking MoH officials and the Masindi hospital staff boosted morale, while the increase in experts allowed reviving certain activities (e.g. public mobilisation). Gulu local staff turned the tide by fearlessly clearing the Ebola ward of dead bodies, thus acting as role models. With the approval of MoH the District Government paid out special allowances to attract and retain HCWs and volunteers involved in the Ebola response. Local hospital staff volunteered again to work on the Ebola ward and agreed to work there long enough to gather sufficient experience, and community members approached the response team asking whether they could join in the effort.
In retrospect it became apparent that the follow-up of contacts was not as comprehensive as it should have been. Patients who had been discharged from the 'probable' section of the Ebola ward as non-cases as well as staff working on the Ebola ward were not followed up systematically. In future outbreaks, the possibility of nosocomial and occupational transmission on the Ebola ward should be taken into account when establishing lists of contacts for follow-up.
For the VHF field laboratory to be fully supportive to the control efforts there must be an appropriate strategy for sampling, testing and communicating results. Isolation must not be delayed until the diagnosis has been laboratory confirmed for any length of time, as this would put contacts of probable cases at risk of transmission. Instead, probable cases must be isolated on clinical and epidemiological grounds alone, possibly in a holding area outside the isolation ward if that is more acceptable to patients. Given that the isolation ward is a very frightening place for most, probable cases who know that laboratory results will be available in a few days, notably HCWs who are probable cases themselves, may refuse isolation without laboratory confirmation. HCWs who decide on isolation may be less inclined to insist if the probable case is one of their colleagues. This occurred repeatedly during the Masindi outbreak, and at least one case of occupational Ebola infection occurred at Kiryandongo Hospital due to the delayed transfer of a probable case to the Ebola ward.
Antigen capture ELISA and PCR were the main tests to confirm current infection. At the time, the latter was believed to become positive one or two days earlier than the ELISA, which would be an obvious advantage to reduce in-hospital transmission between patients. On the other hand, positive PCR results repeatedly turned out to be false. This created considerable and unnecessary anxiety in the patients and put them at avoidable risk for in-hospital transmission. Specificity of PCR needed further improvement if the method was to become an essential tool for a BSL 4 field laboratory. Five years later, PCR was the dominant test during the Marburg HF outbreak in Uige/Angola [20
The communication of laboratory results was problematic in two ways. Firstly, it was initially unclear who was responsible and entitled to receive laboratory results, so that conflicting information circulated, which resulted in an unnecessary burden on laboratory staff to respond to repeat queries. In the future, a medically qualified individual should be identified from the beginning to act as single contact person for the communication between laboratory and hospital. Secondly, oral communication by telephone in a setting where many family names are similar or identical and where many team members do not share the same mother tongue lead to several misunderstandings. This problem can be solved by using written communication by email, transmitted via telephone landlines, mobile phone networks, satellite connection or high frequency radio
Clinical records were not available for analysis. It was therefore almost impossible to investigate whether variations in the quality of supportive care may explain the striking differences in CFP between the two phases of the outbreak response. Two explanations have been offered for the records' absence. Firstly, it was reported that they were sent to Gulu to be analysed jointly with the clinical records from there. However, colleagues in Gulu denied having ever received them. In any case, clinical records from Masindi Ebola ward were property of Masindi hospitals, and should not have been removed; instead, copies should have been sent to Gulu. Secondly, it has been suggested that the possibly contaminated records may have been destroyed at the end of the epidemic to avoid that they could become the origin of renewed transmission, or that they are incriminated should such transmission occur from other origins. The risk that dry paper acts as fomite for the transmission of filovirus appears to be small given their susceptibility to drought and sunlight [21
], but experiments could provide valuable data on viral survival rates on paper. Ultimately, guidelines on handling clinical records from Ebola wards should be agreed by the major players and approved by WHO to avoid unnecessary loss of data urgently needed to assess the effectiveness of treatment regimes for filoviral infections.