Unknown Chains

Contact tracing has more than one denominator. A response team may report the number of contacts it has identified, the number due for follow-up, the number reached that day, or the number of cases whose likely exposure can be linked to an earlier case. Those are different measures. Confusing them can make a response look either more complete or more broken than the evidence supports.

That distinction matters in the 2026 Bundibugyo Ebola outbreak in the Democratic Republic of the Congo. A percentage described as “outside contact lists” cannot automatically be read as the share of all transmission chains investigators failed to trace; the denominator and definition have to be stated. The WHO’s 3 July situation report gives a more precise, interpretable number: as of 1 July, response teams had identified 10,821 contacts in Ituri and North Kivu and had followed up 8,954. That is about 82.7 percent of the contacts on the identified list. WHO also reported follow-up rates of 83.2 percent in Ituri and 81 percent in North Kivu. Those figures describe listed contacts. They do not tell us what share of all infections had been found, or what share of all cases came from known transmission chains. WHO’s 3 July Disease Outbreak News reports the counts and their limits.

This is not a technical quibble. A follow-up rate can be high while new infections continue, because the contacts who are on the list may be only part of the people who were exposed. Conversely, a large number of unlinked cases does not mean that every part of the response is failing. Cases may emerge late, be detected outside health facilities, or be assigned to a zone after laboratory confirmation. The right reading depends on how each figure was defined and when it was collected.

What a contact list can tell us

When a person tests positive for Ebola, investigators seek to identify people who had relevant contact while the patient was infectious. Those contacts may be monitored for symptoms during the incubation period. If someone becomes ill, prompt testing, isolation and care can help protect others. Tracing is one part of a larger response that also requires diagnosis, safe care, infection prevention, safe and dignified burial, community engagement and reliable logistics. WHO describes these as connected outbreak-control activities.

The contact list is therefore a working instrument, not a census of every exposure. Its denominator grows as investigators find more contacts, and it changes as people complete follow-up. A “followed” contact means that the monitoring step was recorded; it does not by itself establish that every possible contact was found, that a person was reached every day, or that a future infection would necessarily be prevented.

The July report shows why precision matters. WHO recorded 10,821 identified contacts across two provinces and 8,954 followed up. Dividing those figures gives about 82.7 percent. That calculation is only a rough aggregate from the reported counts; the report’s separate province-level rates are 83.2 percent for Ituri and 81 percent for North Kivu. Neither the aggregate nor the province rates answer the question “What proportion of new cases belonged to a known chain?” Their denominator is contacts already identified, not all infections or all cases reported.

There is also a timing problem. WHO cautioned that newly reported cases and deaths could include a backlog of samples, and therefore need not represent infections acquired on the day they were reported. A chart by notification date is partly a picture of when a response system processed information. It is not a direct daily map of when transmission occurred. The same WHO report makes that limitation explicit.

What July’s figures did show

As of 1 July, WHO reported 1,460 confirmed cases and 452 deaths in the DRC across 36 health zones. Ituri accounted for 1,333 of those cases, or 91.3 percent. At least 102 confirmed cases were among health and care workers. These were serious measurements of the outbreak’s scale and distribution. They were not proof that the virus had moved uniformly through the country. The burden was concentrated, and the number of affected zones was changing.

The outbreak was unfolding in a conflict-affected humanitarian setting. WHO described displaced and mobile populations, disrupted services, insecurity affecting health facilities and constrained access for response teams as operational challenges. Such conditions can make surveillance incomplete for ordinary reasons: a team cannot reach a location, a sick person may avoid a facility, or a specimen may take time to arrive and be tested. “Unknown” is a description of what investigators currently know. It does not identify a deliberate concealment or a particular cause.

The virus also posed a diagnostic challenge. Bundibugyo disease can begin with nonspecific symptoms such as fever, fatigue, muscle pain, headache and sore throat, which overlap with malaria and other common illnesses. WHO says laboratory confirmation is needed to distinguish it reliably. That is one reason surveillance depends on functioning clinics and laboratories, not merely on the size of the contact-tracing team.

The later test: Kenya’s imported case

The July numbers were not the end of the story. In its 23 October update, WHO reported 9,704 confirmed cases and 4,687 deaths across 66 health zones in eight DRC provinces, based on information through 20 October. WHO also reported imported cases in Kenya and Nigeria. In Kenya, nine cases, including one death, had been reported by 20 October; the eight additional cases were all known contacts of the first imported case, and WHO said there was no evidence of wider community transmission at that time. The 23 October WHO update provides the date, counts and qualification.

That Kenyan cluster illustrates what contact tracing can do when cases are identified and their contacts followed: it can help authorities find linked cases early enough to isolate and care for people, while monitoring for signs of wider spread. It does not prove that every infection is known, that no onward transmission can occur, or that the same performance will hold in a different district. A small, well-defined cluster and a large, geographically dispersed outbreak test different parts of the system.

The contrast also corrects a common mistake in reading outbreak statistics. “Eight cases were known contacts” is a statement about the relationship of those cases to a known index case, under the investigation available on that date. “Eighty percent of all transmission chains are untraceable” would be a much broader claim. The first does not establish the second. It is better to report the actual denominator and the limit of what it means than to turn a dramatic shorthand into a measure of the whole epidemic.

A simple model, carefully labelled

A small numerical example can show why unidentified exposure matters without pretending to estimate Ebola’s reproduction number. Imagine 100 confirmed cases, of which investigators can link 60 to a known chain and cannot yet link 40. If each group generates the same average number of further infections, the unlinked group represents 40 percent of those cases for which the tracing system has not yet established a chain. This is just a classification example; the 60/40 split and any transmission rate are invented, not data from the DRC outbreak.

The example cannot tell us whether transmission is growing. To estimate that, epidemiologists need information about timing, infectious periods, contact patterns, isolation, immunity, care and the probability that cases are detected. Nor does “unlinked” mean “unobserved forever.” A chain can become clear after interviews, laboratory results or a later case investigation. The category records uncertainty at a particular moment.

This is why surveillance work should be judged with several measures at once: how quickly suspected cases are tested, how many contacts are identified, whether listed contacts are reached during the relevant period, where new cases occur, and whether people can access safe care. A strong follow-up rate is encouraging, but it cannot compensate for infections never detected. A high count of unlinked cases is concerning, but it does not tell us which response action is failing without more context.

Trust and access are infrastructure

Contact tracing is not only a spreadsheet problem. It requires people to share names, movements and symptoms with teams they trust. In a setting where clinics are difficult to reach, treatment may be feared or misunderstood, and insecurity can interrupt daily life, a technically correct form cannot create cooperation by itself. WHO’s report calls for response efforts led by local leaders and anchored in communities. It also notes that deaths in the community and attacks affecting health facilities complicate early detection and access to care.

The countercase to the image of a broken net is that response systems can learn and adapt. In Uganda, WHO reported no new cases after 21 June in its early July report, and all but ten of 831 listed contacts had completed their 21-day follow-up by 2 July. That does not mean the same result was guaranteed in the DRC; WHO’s October updates show continued transmission there. It does show that focused tracing and follow-up can close a defined cluster when cases, contacts and resources are sufficiently bounded.

Border restrictions are not a substitute for that work. In July, WHO advised against restrictions on travel or trade with the DRC and Uganda. Its October advice likewise opposed restrictions on affected countries. Screening, information-sharing, rapid testing and readiness at points of entry can support response; broad restrictions may shift movement to less visible routes and disrupt trade without finding a contact. The useful question is not how a border looks on a map, but whether health teams can identify and reach people who may have been exposed.

“Unknown chains” is a serious warning, but it is not a self-explanatory statistic. The July contact-follow-up count described people already identified; the October Kenyan cluster showed that known contacts could be found and monitored in one setting; the DRC’s continuing spread showed the scale of the wider challenge. Each measure answers a different question. Good outbreak reporting keeps them separate. Good forecasting does too: it changes its view when a new case, a completed follow-up period or a better denominator changes what is known.

CDC Global photographed Ebola contact-tracing teams awaiting their daily assignments in West Africa on 18 August 2014. Historical file image from an earlier outbreak; it is not a 2026 response team or a current contact list. Credit: CDC Global, via Wikimedia Commons. Image source · CC BY 2.0 Generic