Reported Ebola cases in the Democratic Republic of the Congo have fallen by a third since early September. In Nord-Kivu the laboratory data support that fall. In Ituri, where three in four cases have occurred, the number of samples tested fell by about as much as the cases did.
UKOilWatch · 2 October 2026 · OilWatch Network Analysis
On 22 August we published the result of a twelve-week test of the Compound Cascade Modelling Framework against the Bundibugyo Ebola outbreak in eastern Congo. One pre-committed trigger had fired and the reading moved upward. The article said it would be updated again only when a further trigger resolved.
None has. That rule is why this is a new article and not a third update to From Hormuz to Bundibugyo.
What has changed is the shape of the question. The count is no longer rising at its August pace, and the World Health Organization says it is starting to see "encouraging signs that we are gaining ground". In the same fortnight a displacement camp holding about 1,000 known Ebola contacts emptied, and its transit centre was burned.
Both are true at once, so this piece asks something narrower than whether the outbreak is getting worse. Is the fall in reported cases real? To answer it we read 57 daily situation reports from Congo's Institut National de Santé Publique (INSP), covering 4 August to 29 September, and extracted the figures that bear on the reach of surveillance and not only the size of the outbreak.
Where the outbreak stands
As of 29 September the DRC had reported 8,224 confirmed cases and 3,982 deaths, a crude case fatality ratio of 48.4%. The crude ratio understates mortality among resolved cases, because deaths lag cases; community deaths that are never tested push it the other way. At the close of the August test window the figures were 5,290 and 2,516.
The outbreak has reached 63 of 167 health zones in seven provinces. The two newest are Bulu in Sud Ubangi, in the north-west on the river corridor, and Dungu in Haut-Uélé, bordering South Sudan. WHO says this expansion raises the risk of cross-border transmission.
It is the largest and deadliest Ebola outbreak in Congo's history and the second largest recorded anywhere, behind West Africa in 2014–2016 (28,616 cases, 11,310 deaths). Confirmed deaths stand at 35% of the West African total, four and a half months after the outbreak was declared.
The pace has slowed. Confirmed cases averaged 92 a day in the week to 18 August, 85 a day in the week to 8 September and 58 a day in the week to 29 September. Deaths fell from 50 a day to 29 over the same span.
The institutions agree on the direction and disagree on the weight. WHO reports that incidence in Ituri has declined gradually since a mid-August peak and that Nord-Kivu peaked in mid-September. Africa CDC says cases have levelled off but that "we cannot talk about control".
The test: cases, tests and positivity
A fall in reported cases can mean fewer infections, or fewer infections found. The daily reports allow a rough test of which, because they give the number of new samples tested in each province as well as the number positive.
If cases fall while testing holds steady, the share of samples that are positive must fall. That is the signature of a real decline. If cases fall because fewer samples are tested, positivity holds or rises. That is the signature of reduced reach.
The test is not conclusive, because a real decline also produces fewer suspected cases to sample. What it shows is where the reported fall is corroborated and where it is not.
Vaccination does not explain the September fall. Ervebo is being given to frontline workers within a study that began on 19 September. Ituri had vaccinated 731 people by 29 September, and WHO says the vaccine's protection against Bundibugyo is unknown.
Nord-Kivu passes. In the week to 15 September the province recorded 33 cases a day with 20% of samples positive. In the week to 29 September it recorded 17 cases a day with 11% positive. Testing held between 150 and 166 samples a day throughout. The September surge in Nord-Kivu was real, and so is the fall since.
Ituri does not. In the week to 8 September Ituri recorded 51 cases a day from 295 samples a day, 17% positive. In the week to 29 September it recorded 36 cases a day from 214 samples a day, again 17% positive. Cases fell 30% and tests fell 27%.
In between, testing dropped to about 170 samples a day and positivity rose to 24%. That is what happens when sampling narrows to the likeliest cases. The reports do not say why Ituri's testing fell. It began to fall in the second week of September.
Ituri's earlier decline looked different. Between mid-August and late August its cases fell from 68 a day to 51, while testing held near 290 samples a day and positivity dropped from 25% to 17%. That fall was corroborated. September's is not.
A second series locates where Ituri's fall happened. Each report counts the alerts received in the province and the alerts validated as suspected cases.
If fewer people with symptoms were coming forward, suspected cases would have fallen with the tests. They did not. Ituri validated about 182 suspects a day in the week to 8 September and about 183 in the week to 29 September, and alerts rose from about 810 a day to about 1,100. Over the same span samples tested fell from 1.6 per suspected case to between 0.8 and 1.2.
The fall in Ituri's testing came between identifying a suspected case and getting a sample to a laboratory. That is consistent with the UN's report that insecurity has delayed the transport of samples from Fataki to Bunia. It does not show that more infections went unconfirmed, because the reports do not say how many suspects were sampled.
Ituri accounts for a little over half of the national fall since early September, and Nord-Kivu for the rest. So a little under half of the bend in the reported curve is supported by the laboratory data. The remainder is unconfirmed, and it sits in the province with 76% of all cases.
What has held or improved
The framework commits to reading evidence symmetrically, so the measures that have held or improved come before the ones that have not.
- Cases found after death. About one confirmed case in three is identified only by swabbing a body in the community: 33% in the latest week, and between 27% and 39% in each of the past eight. That has held, at a poor level: those people were never isolated while infectious, and post-mortem detection at this rate is itself a driver of transmission.
- Contact listing. Contacts under follow-up per case confirmed in the previous 21 days rose from 13 on 25 August to 21 in the latest week. Africa CDC's working figure is about 60 contacts per case, so listing has gone from roughly a fifth of that benchmark to roughly a third.
- Goma. The INSP health-zone tables show one cumulative confirmed case in Goma in every report that carries the table. Médecins Sans Frontières has suspended admissions at its Goma treatment centre because of the small number of active cases.
- Uganda. The outbreak there was declared over on 27 August, with 20 confirmed cases and two deaths.
- Capacity. WHO reports that beds rose from 929 to 1,510 between late July and 21 September and laboratories from 15 to 26. Africa CDC says nearly $2.9 billion was pledged on 24 September. Pledges are not disbursements, and WHO says another 1,400 health professionals are needed to staff the new beds.
What has deteriorated
The number of listed contacts who are not seen each day has roughly doubled. In Ituri the weekly median rose from about 1,200 in the week to 8 September to about 2,700 in the week to 29 September. In Nord-Kivu it rose from about 1,450 to about 3,050.
Nationally, 78.7% of listed contacts were seen on 29 September, against a required 95%. Nord-Kivu was at 71.4%.
This is the known network fraying. The unknown one may be larger. Africa CDC's director-general, Jean Kaseya, said on 25 September that more than 80% of newly confirmed cases were people who were not on any contact list. In August the agency had put that share at more than 70%.
No INSP report publishes that share. It cannot be tracked from the daily data, and it rests on those two statements.
The reports do record the violence behind the numbers: a screener shot dead at the Vulemba checkpoint in Nord-Kivu, a burial team attacked in Beni, and response activities suspended in Nizi, in Ituri, for insecurity. The UN adds that fighting has restricted access to the Fataki, Drodro and Lita health zones.
Kigonze
Kigonze camp, outside Bunia, sheltered about 19,000 displaced people. The UN's senior Ebola coordinator in Congo, Julien Harneis, says it now lies abandoned. Its Ebola transit centre burned down on the evening of 29 September; the UN does not say how it started, and the Associated Press reports that soldiers burned it.
The UN dates the security incidents at the camp from 20 September. Reuters reports that security forces entered looking for weapons and that clashes followed. Citing two health officials, it reports that about 1,000 Ebola contacts lived in the camp and that only about a fifth have since been traced.
Kigonze is not the only site. More than 3,000 people fled Rhoe camp, also in Ituri, after fighting intensified on 27 September. UNHCR counts 88 confirmed cases and 56 deaths across Congo's displacement sites and considers that an underestimate.
This is the mechanism the May article described: conflict acting on the system around the pathogen and leaving the pathogen unchanged. Three things follow from it, and one does not.
It cannot yet be read in the case data. Ebola's incubation period runs to 21 days, so cases among people who dispersed in late September would surface through mid-October, and their contacts later. Reported cases in the Bunia health zone fell from 16 a day in early September to 7 a day in the latest week. That proves nothing either way, because people who have left a zone are not tested in it.
The official daily reports do not register it. No INSP report for 20 to 29 September mentions the camp emptying or the loss of its contacts. Earlier reports name Kigonze only as a transit centre for suspected cases, saturated at the end of August.
Ituri's contact list shrank in the same week. Contacts under follow-up in the province fell from 17,438 on 21 September to 12,476 on 28 September, with no explanation in the reports. Contacts leave a list when they complete 21 days, when a health zone fails to report, or when they are lost. The reports do not say which.
What does not follow is attribution. Ituri's testing began to fall in the second week of September, before the incidents at Kigonze. Unseen contacts rose in Nord-Kivu as well as in Ituri. Kigonze is one observed failure inside a wider loss of access, and the data cannot separate its effect.
Nor is the direction of its effect on transmission known. The camp was itself a site of transmission: Reuters reported in June that more than 30 people had died there in just over a month, many from Ebola. Dispersal reduces crowding and destroys visibility. The second effect is certain. The first is not yet measured.
A measurement trap
One widely quoted comparison overstates the gap in contact tracing. Dr Kaseya set about 30,000 listed contacts against the roughly 420,000 that 7,000 cases should generate at 60 contacts each.
The two numbers are not the same kind of thing. The 30,000 is the number under follow-up now, and a contact leaves the list after 21 days. The 420,000 counts every case since May.
Set like against like and the gap narrows. In the three weeks to 29 September, 1,381 cases were confirmed. At 60 contacts each that is about 83,000 expected, against 30,751 listed. That is roughly a third, where the quoted comparison implies a fourteenth.
A third is still a large shortfall. But measured the same way against Africa CDC's benchmark, the three weeks to 25 August came to about a fifth, so listing has improved.
The May triggers
No trigger from the May article has fired since the August update. One has become readable.
| Trigger set in May | Direction | Status | Basis |
|---|---|---|---|
| 5,000 cases by end of August | Up | Fired 16 August | Recorded in the August update |
| Case in Mombasa, Dar es Salaam, Nairobi or a West African capital | Up | Not fired | No confirmed case in another country since Uganda. A fourth case was imported into Europe on 1 October; all four were medical or humanitarian staff |
| Health-worker infection rate above 5% for four weeks | Up | Unresolved | The denominator was published once, on 4 August, then dropped (see the corrections below) |
| PHEIC escalated | Up | Not fired | The Emergency Committee met again on 18 August; no pandemic-emergency determination |
| Emergency authorisation of a Bundibugyo vaccine | Down | Not fired | WHO recommends Ervebo only within research protocols; a 20,000-person study began on 19 September |
| PHEIC downgraded or withdrawn | Down | Not fired | In force since 17 May |
| Declines in both Goma and Kampala | Down | Kampala met; Goma readable | One cumulative case in Goma and no growth |
| A responding agency says containment is achievable on the current trajectory | Down | Not fired | WHO, 16 September: "We have many, many months ahead of us" |
Goma needs a paragraph. The August update said the Goma half of that trigger could not be assessed from province-level data. The INSP health-zone tables do assess it: one cumulative confirmed case, in every report that carries the table.
The trigger as written asks for declining counts, and one case cannot decline, so it has not formally fired. Read plainly, the transport hub the May article worried about most did not take off. That belongs on the downward side of the ledger.
Two cautions apply. Goma is under M23 control, and we cannot assess how complete surveillance is there. And Médecins Sans Frontières refers to a "small number of active cases" in Goma, which sits oddly beside a single cumulative confirmed case.
Corrections to the record
The earlier article and its updates stay as published. Five things in them need correcting or qualifying here.
Uganda's dates. The August update said Uganda declared its outbreak over on 28 July. That date was the interruption of local transmission. WHO and Africa CDC announced the end of the outbreak on 27 August.
The health-worker denominator. The August update said the number of front-line responders was not published. It was, once. The INSP report for 4 August gives "134 sur 2 884 exposés (4,6 %)" for front-line staff in Ituri. The next day's report tabulates 141 infected in Ituri, 14 in Nord-Kivu and 4 in Haut-Uélé. The table was dropped from the reports on 6 August.
That figure is a cumulative share of exposed staff, not the four-week rate the trigger specifies, so the trigger stays unresolved. But it sat just under the 5% threshold in early August, and nothing comparable has been published since. Africa CDC put health-worker deaths at 50 in late September.
"Tens of thousands". The May article criticised a commentator for treating the West African toll as a projection. The objection was to the sourcing, and it stands: an envelope is not a forecast. But on 12 August WHO's Director-General said the outbreak was, at its then pace, "on track to eclipse the West African Ebola outbreak of 2014 to 2016". An institution has now said it. The pace has slowed by more than a third since that remark.
The mortality range. In May we gave a range of mid-thousands to low tens of thousands of deaths over 12 to 24 months. We do not revise it here, because no trigger licenses a revision. But confirmed deaths stood at 3,982 after four and a half months. At the last fortnight's 29 a day they would pass 5,000 in early November; that is arithmetic and not a forecast. The lower end of the range is being approached far earlier than its horizon implied, and confirmed deaths are a floor. Africa CDC estimated in late August that the outbreak may be three times larger than official figures indicate.
The geography. The May trigger named Mombasa, Dar es Salaam and Nairobi. The outbreak has instead spread north and west, into Haut-Uélé toward South Sudan and into Sud Ubangi on the river corridor. The trigger watched the wrong direction.
New triggers, to be read on 30 October
These are committed on 2 October, before the data they will be read against exists. Three can be computed by anyone from the INSP daily reports. Their thresholds are set from the range each measure has held over the past eight weeks.
| Measure | Latest week | Moves the reading up | Moves the reading down |
|---|---|---|---|
| Ituri positivity, once testing is back to 290 samples a day (seven-day mean) | 17% on 214 samples a day | 17% or higher | 13% or lower |
| Share of confirmed cases found after death | 33% | Above 39% for two consecutive weeks | Below 27% for two consecutive weeks |
| Contacts under follow-up per case confirmed in the previous 21 days | 21 | Below 15 | Above 30 |
The Ituri thresholds are not arbitrary. At 290 samples a day, 13% positivity yields about 38 cases a day, which is the level now reported. That would confirm September's decline. At 17% it yields about 49, the early-September level, and the decline was lost reach. Between the two the answer is partial and will be reported as partial.
The thresholds can be told apart. At about 1,500 samples a week, a positivity of 17% carries a margin of roughly two percentage points either way. Daily positivity swings far more, which is why every trigger is read on seven-day figures.
One caution applies to the reading. The rainy season degrades road access in the east from October, so a further fall in Ituri's testing could be seasonal as well as a loss of reach.
Two further triggers rest on events and statements.
- Up: a confirmed case, other than a managed medical evacuation, in South Sudan, Rwanda, Burundi, Kinshasa or Mbandaka. Rwanda is included because Gisenyi faces Goma across the border.
- Down: Africa CDC states that the three conditions Dr Kaseya set on 25 September are met. They are that new cases arise from contact lists, that no deaths occur in the community, and that no new health workers are infected.
One outcome is defined in advance as unresolved. If Ituri's testing stays below 220 samples a day through to 30 October, the question of whether its decline is real remains open, and that will be reported as the finding.
The rule from May carries over unchanged. If only upward triggers fire and the reading does not move up, or only downward ones fire and it does not move down, the model is being held hostage by priors and should be corrected.
What we are not claiming
- That cases in Ituri are not falling. The data do not confirm the fall. They do not refute it either.
- That Kigonze has caused new transmission. No data yet show that.
- That the response is failing. Contact listing improved, capacity expanded, and Uganda and Goma held.
- That the outbreak will reach the West African toll.
- That the framework predicted Kigonze. It did not. The May triggers contained no measure of containment being physically removed, and that gap in their design is recorded here.
What we are claiming is narrower. Reported cases have fallen by a third. A little under half of that fall is corroborated by laboratory data. The rest is unverified, in the province where most cases are. The number of known contacts going unseen has doubled. And one observed loss of containment, at Kigonze, is not yet visible in the data and is not mentioned in the official daily reports.
The next four weeks will resolve some of that. We will read the triggers on 30 October and move the reading in whichever direction they dictate.
Why this is on an oil site
Not because of fuel. We looked for evidence that fuel cost or supply is constraining the Ebola response. We found passing references to fuel prices in logistics reporting and nothing that measures an effect. We do not claim a link between the diesel market and this outbreak.
The connection is method. Readers of these sites know the argument that price is lying and runway is the story. A quoted price can sit still while the physical buffer behind it drains.
A case count is the same kind of number. It is produced by an instrument made of alerts, samples, laboratories and tracers. It can fall because the thing measured fell, or because the instrument reached less of it.
The discipline is the same in both fields. Find the series that measures the instrument, and read the headline number against it. For crude that is inventory and transit. Here it is samples tested.
Method, limits and sources
The series comes from 57 INSP daily situation reports, numbers 82 to 138, covering 4 August to 29 September 2026. Figures were extracted from the published PDFs with AI assistance. Each figure was read at least twice. One report was read in full and three were spot-checked against the source text.
Daily counts are by date of notification: each report adds the day's confirmations to the previous total, so past days are not revised as later reports arrive. In eight weeks there were three one-case adjustments. Cumulative totals reconcile from each report to the next. They match figures published by WHO, ECDC, MSF and Africa CDC on seven dates.
The alerts table for Ituri and Nord-Kivu was read in a second pass; report 126 is left out of the suspected-case averages because it validates 1,058 Ituri alerts against about 200 on other days.
The limits matter.
- The share found after death and the contacts-per-case ratio are calculated here. INSP does not publish them.
- Province-level test counts begin on 15 August, so the mid-August positivity figure for Ituri rests on four days.
- Contact totals swing when health zones fail to report, which is why contact figures are weekly medians. Reports 116, 121 and 136 contradict themselves on contacts.
- Positivity is a rough discriminator. It cannot prove a decline is unreal.
- The figure of about 1,000 contacts at Kigonze rests on two unnamed officials quoted by Reuters.
Sources
- Institut National de Santé Publique, daily situation reports 82 to 138, including report 138 of 29 September and report 82 of 4 August
- WHO, Disease Outbreak News DON618, 25 September 2026
- WHO, second meeting of the IHR Emergency Committee, temporary recommendations, 24 August 2026
- WHO, emergency guidance on the use of licensed Ebola vaccine during Bundibugyo outbreaks, 1 September 2026
- WHO Regional Office for Africa, Uganda ends Ebola outbreak, 27 August 2026, and Scaling up to stay ahead of the Ebola outbreak, 30 September 2026
- ECDC, Ebola disease outbreak in the DRC, updated 1 October 2026, and fourth imported case in Europe, 1 October 2026
- US CDC, Ebola outbreak: current situation, 1 October 2026
- UN News, Ebola response faces setback as transit centre burns down in DR Congo camp, 1 October 2026, and Calls grow for urgent humanitarian access, 28 September 2026
- Associated Press, An Ebola treatment center was burned down as the death toll in Congo passes 4,000, 2 October 2026
- Reuters, Thousands flee Congo Ebola-hit camp, 1 October 2026
- Reuters, Congo Ebola contact tracing falls far short, Africa CDC chief says, 25 September 2026
- Reuters, Ebola response making progress in Congo, but outbreak far from over, WHO says, 16 September 2026
- Euronews, Congo's Ebola outbreak on track to surpass deadliest in history, WHO chief says, 13 August 2026
- Bloomberg via Business Standard, Congo Ebola outbreak may be 3 times larger than official count: Africa CDC, 28 August 2026
- Associated Press, Congo Ebola outbreak tops 8,000 confirmed cases, 28 September 2026
- Al Jazeera, Ebola cases surpass 8,000 as DRC struggles to control outbreak, 28 September 2026
- Médecins Sans Frontières, Ebola disease outbreak 2026: how MSF is responding
This is a follow-up to From Hormuz to Bundibugyo (29 May 2026) and its updates of 18 July and 22 August. Cross-published on ukoilwatch.com, eurooilwatch.com and americasoilwatch.com.