Skip to content
Briefings are running a touch slower this week while we rebuild the foundations.See roadmap
Pandemics and Biosecurity
9JUN

Ebola model puts the fork at 20% isolation

3 min read
09:58UTC

The CDC put the Bundibugyo outbreak's reproduction number at 2.51 on 5 June, and showed its fate turning on one controllable variable: how many patients get isolated in time.

ScienceDeveloping
Key takeaway

The outbreak's outcome turns on the patient isolation rate, a logistics variable, not on the virus itself.

The US Centers for Disease Control and Prevention (CDC) published modelling on Friday 5 June putting the Bundibugyo Ebola outbreak's basic reproduction number (R0, the average number of people each case infects) at 2.51, and estimating that the spillover which started it occurred around 19 February 1. The CDC is the United States' principal federal public-health agency; it released the projections through its Morbidity and Mortality Weekly Report (MMWR), its flagship epidemiological journal. Bundibugyo is one of the rarer Ebola species, named for the Ugandan district where it was first isolated in 2007, and the 2026 DRC and Uganda outbreak is the largest on record for it.

The fork in the model matters more than the R0. At the current patient isolation rate of roughly 20%, 65% of the model's simulation runs cross 20,000 cases by 22 August, approaching the scale of the 2014 to 2016 West Africa epidemic. Lift isolation to 70% and that worst case collapses to a 1% tail, with 90% of runs projecting fewer than 2,000 deaths. One controllable variable, the isolation rate, decides which of those two futures the outbreak follows.

That R0 of 2.51 sits below the 2014 Zaire-ebolavirus estimates of roughly 1.5 to 2.5 at peak, yet the projection runs worse, because isolation rather than intrinsic transmissibility is the binding control, and isolation is failing where the state apparatus is thin. The model was computed against a corrected baseline: WHO had carried 1,040 cases in late May before clearing its laboratory backlog and switching to confirmed-only figures. We lead on the projection and the correction rather than a raw count, because the counting method itself had inflated the earlier figure.

Deep Analysis

In plain English

Think of a disease's R0 number as how many people each infected person passes the illness to before they recover or die. An R0 of 2.51 means one person with this Ebola strain infects roughly two or three others, on average. For comparison, seasonal flu has an R0 around 1.3; measles is about 15. The CDC ran thousands of computer simulations of how this outbreak could play out. Right now, only about one in five patients is being isolated , kept away from others while infectious. At that rate, nearly two-thirds of the simulations ended with more than 20,000 cases by late August. But if isolation could be raised to seven in ten patients, almost none of the simulations got that bad. The hard problem is that two of the areas where the outbreak has spread are controlled by armed groups who block health workers from entering.

Deep Analysis
Root Causes

Bundibugyo's late-detected spillover around 19 February 2026 , estimated from the modelling's baseline , reflects a structural gap in DRC's zoonotic surveillance network. Ituri Province lacks the sentinel site density that would catch a haemorrhagic fever signal inside the 14-day window needed to initiate ring control before community amplification begins.

The 20% isolation rate is not primarily a capacity failure: treatment units exist in Bunia and Mongbwalu. It reflects a trust deficit built through years of contested Ebola responses in eastern DRC, where communities have torched treatment centres and attacked health workers. The CDC model treats isolation as a behavioural parameter, not an infrastructure one, making the gap a governance and community-engagement problem, not a beds problem.

What could happen next?
  • Risk

    At 20% isolation, CDC simulations give a 65% probability of 20,000+ cases by 22 August , an outcome that would place sustained pressure on global Ebola response capacity and increase the probability of exported cases beyond the current UAE-confirmed-clear incident.

    Short term · Assessed
  • Consequence

    The 70% isolation threshold required to collapse the worst-case is structurally blocked in IS-controlled Mambasa and M23-held South Kivu, meaning the model's best-case scenario requires a security intervention the health response cannot itself deliver.

    Immediate · Reported
  • Meaning

    The CDC's explicit comparison to the 2014 West Africa outbreak trajectory (also noted by Africa CDC) signals that the international community is framing this as a potential generational outbreak, not a contained regional event.

    Medium term · Assessed
First Reported In

Update #6 · Ebola outbreak gets an R0, and a fork

CDC MMWR· 9 Jun 2026
Read original
Causes and effects
This Event
Ebola model puts the fork at 20% isolation
The model converts a containment gap into a single watchable metric, telling readers which number decides whether the outbreak stays small or reaches 20,000 cases.
Different Perspectives
World Health Organization
World Health Organization
WHO's DON606 recalibration to confirmed-only reporting gives the clean baseline the CDC model rests on, but the apparent fall from 1,040 to 534 carries misinterpretation risk WHO communications have not pre-empted. The PABS deadlock ahead of IGWG7 and continuing MBP134/remdesivir assessment without authorisation make WHO the body most able to accelerate the two decisions that could change the outbreak's trajectory.
European Union (ECDC)
European Union (ECDC)
ECDC's Week 23 CDTR tracked four simultaneous non-Ebola signals: the Dermatophilus congolensis novel-transmission cluster across France, Germany and Spain; a 4.2-fold malaria surge in Mayotte; the Salmonella ST2045 multi-country cluster; and two new Saudi MERS cases. The continental early-warning layer is carrying a full multi-pathogen picture while Bundibugyo dominates headlines.
Uganda
Uganda
Uganda's 19 confirmed cases are concentrated in Kampala and Wakiso, an urban cluster that applied the 2022 Sudan-ebolavirus playbook; the Bwera border laboratory shortens cross-border confirmation to same-day. Uganda's regulatory authority must co-sign before MBP134 or remdesivir can dose any patient.
Democratic Republic of the Congo
Democratic Republic of the Congo
Kinshasa shares Bundibugyo sequence data in real time with no treaty-guaranteed access to the vaccines that data informs, and its health minister called the US entry ban discriminatory while negotiating an early lift. DRC accounts for 515 of 534 confirmed cases and faces IS-controlled access blockades in Mambasa that health authorities cannot resolve.
United States (HHS/CDC)
United States (HHS/CDC)
HHS expanded the Ebola entry ban to green-card holders on 5 June, widening a restriction expiring around 17 June against WHO advice. The CDC simultaneously published the R0=2.51 modelling, the sharpest analytical contribution to the response, from a federal bench that holds the NIH and acting CDC director roles in one person.
Imperial College London / Cori and Ferguson
Imperial College London / Cori and Ferguson
Anne Cori and Neil Ferguson place the case-fatality ratio at 30 to 40 per cent and read the 6.8-to-1 suspected-to-confirmed ratio as evidence that the laboratory figure understates true lethality. Many people die before a swab reaches them.