The agenda today has three items; the first decides the other two. First: what the evidence actually allows us to claim about the licensed Ebola vaccine during a Bundibugyo outbreak. Second: what the responsible institutions have chosen to do, given that evidence. Third: what that choice says about the profession, and about how the next outbreak will be handled. The first item decides the other two — and on August 31, 2026, the World Health Organization answered it with a phrase that most communications offices would spend a week burying. The phrase was: not enough.
The setting matters. This is a filovirus outbreak in a country that has lived through repeated Ebola emergencies, where public trust in vaccination is itself a fragile resource. A document that overclaims protection would spend that trust for a short-term appearance of resolve. A document that underclaims protects it. Which is why the sentence WHO chose to publish deserves to be read slowly, twice.
Read the memo in order, because the numbers and the reasoning follow the same line. The line runs through a virus called Bundibugyo, a member of the ebolavirus family, and a vaccine called Ervebo that was designed against a different member of that family — the Zaire strain. An outbreak is underway in the Democratic Republic of the Congo, and when a licensed vaccine exists and a related virus is spreading, the natural pressure is to deploy the tool and call the matter handled. WHO has published a full emergency guidance document for exactly this situation, and the document is striking not for what it authorizes but for what it refuses to assert.
Item one: the gap is on the table
Let me lay out the evidence in order, because the shape of it is the story. There is a body of animal data, a body of immunology, and a body of observational evidence, and all of it leans in the same direction: the vaccine may protect against Bundibugyo too, not just against the Zaire strain it was built for. That is the technical meaning of cross-protection — a tool built against one strain doing useful work against a close cousin. The guidance gives that hypothesis its due; it says so plainly, with the sources lined up behind it. And then it says the hard part, in the flat tone of an audit note: the existing evidence is not enough to determine whether Ervebo confers clinically meaningful protection against Bundibugyo infection in humans.
I want to slow down on that sentence, because it is the most important sentence in the document and the easiest to skim. It is not saying the vaccine is useless. It is not saying the vaccine is dangerous. It is saying that the evidence standard has a threshold, and the evidence has not reached it — not for a clinical claim, and not for a deployment claim. Consider what “clinically meaningful” is doing in that sentence. It is a threshold, not a slur. A vaccine could reduce severe disease, shorten illness, or do neither, and an observational dataset can nudge us in a direction without ever proving it. The guidance refuses to let a nudge become a claim. It also refuses the opposite error — it does not declare the vaccine useless, because the animal and immunology evidence is real and deserves weight. The discipline is in holding both. In a field where every outbreak produces pressure to declare victory early, that is a precise, defensible, and genuinely rare thing to say out loud. I was an analyst long enough to know how tempting the alternative is. The alternative is the sentence that says the vaccine is likely to help and we are deploying it, full stop. The alternative is a hope wearing a recommendation’s clothes.
Item two: what the guidance actually authorizes
So what does the guidance do, once it has named the gap? The honest summary is that it converts uncertainty into procedure. Ervebo is to be used, yes — but only inside a research framework, and the guidance explicitly recommends the design: a ring randomized controlled trial. That is the design where you vaccinate the contacts of confirmed cases and compare them against a control group, and it is chosen here not because it is convenient but because it is the structure that can actually produce the evidence the first item says is missing. The outbreak becomes, in effect, a controlled experiment in the middle of an emergency — which sounds cold until you remember that the experiment is also the fastest honest path to knowing whether the tool works.
The rest of the guidance reads like a checklist written by people who have watched promises fail before. Maintain the existing outbreak control measures — the vaccine is an addition, not a replacement. Inform people honestly that the vaccine’s efficacy against this virus is unknown — the consent form tells the truth even when the truth is awkward. And accelerate the development of a Bundibugyo-specific vaccine — because the gap, once named, becomes a line item on someone’s work plan rather than an uncomfortable topic nobody raises. Each of these items is modest on its own. Together they form a coherent position: we do not know, we will say so, we will structure our action so that not-knowing becomes learning, and we will not stop building toward a tool that is made for this virus specifically.
There is a governance principle buried in the research-protocol requirement, and it deserves to be named: in an emergency, evidence is not a luxury that can be postponed. It is the one asset that, if produced, pays for every later decision — including the decision about whether the tool should be used in the next outbreak, and the decision about how the affected population should be counseled. The ring design is slower than mass vaccination would be, arguably, and in the middle of an outbreak the word slower carries real weight. But it is the design that yields an answer rather than an impression. The guidance treats the outbreak as both a crisis to manage and a question to answer, and it refuses to let the first role silence the second.
Item three: the difference between a decision and a hope
Now the third item, which is the one I actually care about. The board’s question in a crisis is not whether to act; it is how to act without pretending the numbers are better than they are. A hope is what you reach for when you want the outcome; a decision is what you make when you have weighed the outcome against the evidence. WHO’s guidance is a decision. It does the unglamorous work of binding its own hands: the vaccine may only be used where it will produce data, the claim of protection will not be made, and the patient will be told what is not known. Those constraints are the entire point. The most reliable predictor of whether an emergency response will be honest is whether it has already written down, in advance, what it will refuse to claim.
Let me say plainly what the default institutional move looks like, because it is important to recognize the shape of what WHO did not do. The default is to deploy the tool, emphasize its promise, and let the evidence question drift to the back of the file. That path is always tempting because it is always defensible in the short term — no one is blamed for trying. But it is precisely the path that erodes public trust when the promise fails to materialize. The more decisive move, the one WHO chose, is the harder one to defend in a briefing room and the easier one to defend in history: say what you know, say what you do not, and bind the deployment to the search for the answer. Decisive is often mistaken for loud; here it takes the quieter form of a binding constraint that the institution imposes on itself.
Let me be concrete about what this looks like on the ground, because governance documents have a way of floating above the places they govern. Picture a clinic in the affected region: a confirmed case is reported, the ring of contacts is defined, and a vaccination team arrives with a protocol, a consent form, and a script that does not include the phrase “this will protect you.” A parent asks directly whether the shot will keep the child safe. The honest answer — we do not know yet, which is exactly why we are asking you to help us find out — is harder to say and easier to respect. That is the concrete moment the guidance is really about: not the meeting in Geneva, but the conversation in the clinic where the truth has to survive contact with fear.
I have watched enough outbreak coverage to notice the default move: turn every tool into a promise, then let the promise do the talking. The cost of that move is never paid at the press conference. It is paid later, in eroded trust, when the promised protection turns out to have been thinner than advertised, and the next vaccination campaign starts from behind. The guidance’s decision to publish the unknown is therefore not a weakness; it is the strongest asset a response can hold, because it cannot be embarrassed by events. When the data improves — when the trial produces an answer, whichever way it points — the document will still stand: we told you what we did not know, and then we found out.
Where the memo lands
Let me close by returning to the agenda. The first item — what the evidence allows — was answered with an honest limit. The second item — what to do — was answered with a protocol designed to close that limit. The third item — what it means — is answered by the simple observation that the institutions responsible for the response chose discipline over convenience. That is the difference between a decision and a hope. A hope says: this should work. A decision says: here is what we do not know, here is how we will find out, and here is how we will behave while we are finding out.
The forecast I would put in the memo is modest but firm. This guidance will be cited by the next outbreak, whatever virus it involves, because it establishes a template: name the evidence gap, restrict the unproven tool to research use, keep the baseline measures running, and tell people the truth about what remains unknown. Not every agency will follow it, and not every field where it applies will welcome it. But the template now exists, published, signed, and dated. The numbers and the reasoning follow the same line, and the line is a good one to have drawn.
One last note, in the margin. I began this memo expecting to write about vaccines, and I find myself writing about honesty — or rather, about the point where the two become the same thing. Decision-making in a crisis is usually praised for speed. This document is evidence that the more valuable virtue is precision — knowing exactly what you do not know, and organizing the whole response around that knowledge instead of around the wish. If I am right, the August 31 guidance will matter less for the Bundibugyo outbreak it was written for than for the ones it will shape afterward. The board’s question was not whether to use the tool. It was how to use it without lying about what it can do. That question now has an answer on the record, and the answer will keep.