Which Specimen, Which Test, Which Day?

A 28-year-old laboratory assistant at the Irkutsk Anti-Plague Research Institute broke a tube of live bacteria while preparing samples in late September. That date comes from regional media. On September 29 she was admitted to hospital with severe pneumonia and put on a ventilator. She died early on October 2 (RFE/RL; Moscow Times). By then more than 60 staff were confined inside the institute, about 200 contacts were under observation, and five hospitals were under quarantine. On October 5 Rospotrebnadzor said tests of her biological material “found no microorganisms linked to her professional activity” and gave the cause of death as “pneumonia of unknown aetiology” (Euronews). The WHO could only pass on what it had been told (UN News). I don’t know whether she had plague, and nobody outside Russia does either. That gap is the thing worth fixing.

“No microorganisms found” describes a test, not a patient. A trained plague worker who reports an exposure and then develops pneumonia will almost certainly get antibiotics on admission, and a few days of treatment can sterilise blood cultures. Antibodies to the F1 capsule take one to two weeks to appear, and she died about a week after the reported exposure. The WHO’s plague case definition accepts species-specific Y. pestis PCR on a clinical sample as confirmation. So whether the negative means anything depends on three facts that haven’t been published: which specimen, which test, and which day relative to the first antibiotic dose. A checkable result has been produced before. When Malcolm Casadaban of the University of Chicago died in 2009, four separate labs worked on his blood isolate. CDC matched the strain to the attenuated KIM D27 he had been handling. The autopsy then found hereditary hemochromatosis: iron overload had probably given a strain that is crippled specifically in iron acquisition the iron it lacked. All of it was published in MMWR 60(7).

The proposal is to make that outcome automatic, so it doesn’t depend on a state’s willingness to share. Anti-doping already works this way: the B-sample is sealed at collection because tester and tested don’t trust each other. Institutes that handle pathogens on the IHR Annex 2 “always assess” list, pneumonic plague included, would commit in advance, the way states already declare high-containment labs under the Biological Weapons Convention’s confidence-building measures. The trigger would be a staff admission for febrile respiratory illness within a set window after documented work with the agent. That would automatically send a sealed split specimen (blood, and lung tissue after a death) to a neutral receiver such as the WHO BioHub at Spiez. Results would come out in a fixed, no-blame format: assay, specimen type, collection date, antibiotic start, outcome. The base rate says this would cost little. A scoping review counted 309 lab-acquired infections worldwide from 2000 to 2021, 8 of them fatal, and 2 of those 8 deaths were Y. pestis. Events like this are rare and consequential, which is exactly when a cheap standing protocol is worth having.

There is also a long precedent for the alternative. In 1939 a Soviet vaccine researcher, Abram Berlin, caught pneumonic plague during aerosol experiments, travelled to Moscow and stayed at the Hotel National. He died, and so did his barber and the doctor who isolated himself with him. Every other contact was quarantined by the NKVD, and the hotel was disinfected at night so word wouldn’t get out (JAMA; RBTH). The containment was excellent and secret at once. Irkutsk so far fits that pattern: about 200 contacts traced within days, no secondary cases, no data. My bet is that by December 31, 2026, no specimen-level result for this case will be published or passed on by the WHO, and the quarantines will end without a stated cause of death. What would prove the proposal unnecessary is post-mortem lung PCR published together with a named alternative cause. If that doesn’t happen, the people who were watched, and the neighbouring governments of Mongolia, China and Kazakhstan, are left with a negative they can’t check. That costs more than mailing a frozen aliquot to Switzerland would have.