There is one question underneath this and I want to hold it, because the answer is several answers and the first one is misleading. How does an Ebola outbreak run for three or four months without anybody noticing?
The Democratic Republic of the Congo confirmed an outbreak of Ebola disease caused by Bundibugyo virus on 15 May. The World Health Organization declared a public health emergency of international concern two days later. This week the confirmed case count stands at 344 across three provinces.
The epidemiological reconstruction places the earliest infections in January or February, in Mongbwalu, in Ituri.
What everybody believes about detection
That it got much better after 2014, and the belief is not baseless.
The West Africa outbreak of 2014 was the event that rebuilt this field. It ran through Guinea, Liberia and Sierra Leone, killed more than eleven thousand people, and demonstrated that the international system for spotting and stopping a filovirus outbreak did not work.
What followed was genuine reform: revisions to how the International Health Regulations are implemented, a standing emergencies program at the WHO, the response network that can deploy specialists within days, licensed vaccines and therapeutics that did not exist in 2014, and considerable investment in laboratory capacity across the region.
All of that is real. Several outbreaks since have been detected within weeks and stopped inside a hundred cases, and that is what the machinery was built to do.
What the record shows this time
Here is the number that should be the headline and is not.
At the point of international confirmation, this outbreak was assessed at around 246 suspected cases and 65 suspected deaths.
The West Africa outbreak, when it was announced in March 2014, was roughly a fifth of that size.
So the reform did not fail at the margin. The outbreak that the improved system caught was, on the day it caught it, about five times larger than the one that overwhelmed the unimproved system was on the day that one was caught. Everything built since 2014 was in place, and the starting position this time is worse.
The vaccine that does not apply
Before the answers, the fact that changes the shape of this outbreak and is routinely omitted.
There is a licensed Ebola vaccine and it has worked well. It has been deployed in ring vaccination campaigns, it has demonstrably shortened outbreaks, and its existence is the single largest difference between the current era and 2014.
It is licensed against Zaire ebolavirus. This outbreak is Bundibugyo virus.
Those are different species within the same genus, and immunity to one is not assumed to confer useful protection against the other. The vaccine that everybody thinks of when they hear the word Ebola is, on the current evidence, not a tool available for this.
That is why the announcement earlier this month of a partnership to develop a Bundibugyo vaccine is significant and also why it is not relief. A vaccine entering development during an outbreak arrives, at the earliest, for the outbreak after this one.
So the response here is the pre 2014 toolkit: find cases, trace contacts, isolate, treat supportively, and bury safely. All of which works, and all of which depends entirely on detection, which is where this began.
What a public health emergency declaration does
Also worth being precise about, because the phrase sounds like an intervention and is closer to a signal.
A declaration under the International Health Regulations is a determination by the WHO director general that an event is a public health risk to other states through international spread and potentially requires a coordinated response. It triggers temporary recommendations: on screening, on travel and trade, on what member states should do.
What it does not do is deploy anybody or release a fund automatically. It is a statement of seriousness that unlocks attention, and attention is usually what unlocks money, but the sequence is a request rather than a mechanism.
The four answers, in ascending order of discomfort
The strain. This is Bundibugyo virus, not Zaire ebolavirus. It is a distinct species, first identified in 2007, and it is the one clinicians have seen least. Case fatality is generally lower than Zaire, which sounds like good news and is not, because a lower fatality rate means more survivors moving around and a presentation that looks less alarming to somebody deciding whether to report it.
The presentation. Early filovirus disease is fever, headache, muscle pain, fatigue and gastrointestinal symptoms. In a region with endemic malaria and typhoid, that is an ordinary week. Nothing about the first several cases would have prompted a differential diagnosis that included a hemorrhagic fever, and the tests that would answer the question are not sitting in a clinic in Ituri.
The geography. Ituri is not well served by anything. Laboratory confirmation requires a sample reaching a facility that can run it, and the distance and the roads and the security situation determine how long that takes and whether it happens at all.
The money. And this is the one that is being discussed least. Surveillance is not an event; it is a payroll. It is community health workers who are paid to notice and report, district officers who are paid to investigate, and sample transport that somebody funds. Substantial reductions in international health assistance over the past two years have fallen disproportionately on exactly those lines, because they are the least visible thing in a health budget and the easiest to cut without anybody noticing for a while.
You notice when an outbreak runs for four months.
What I am not saying
That the funding cuts caused this. I cannot demonstrate that and neither can anybody else yet, and the honest position is that four contributing factors are present and their relative weights are unknown.
A retrospective analysis will eventually attempt to separate them, and it will be hard, because the counterfactual requires knowing what a fully funded surveillance network would have caught in a region where the other three factors still applied.
What can be said now is narrower and still uncomfortable. Detection is the part of the system that produces no visible output when it is working. A surveillance network that is functioning perfectly generates nothing to report, which makes it the single easiest budget line to reduce and the one whose reduction is discovered last.
The question
So: how does this happen for four months?
Not through one failure. Through an unfamiliar strain producing an unremarkable presentation in a place with thin laboratory access, watched by a network that has been getting quietly cheaper.
Each of those on its own is survivable. The system was designed on the assumption that they would not all be true at once, and the design assumption was not written down anywhere, because nobody makes an explicit decision that a region will be watched less. It happens by subtraction, one line at a time, and the result is legible only in retrospect and only as a date: January or February, in Mongbwalu, and then nothing until May.
