Human to human hantavirus

The evidence does not match the alarm

Dr Clare Craig

When severe pneumonia aboard the MV Hondius was attributed to hantavirus this month, the reporting moved quickly from a handful of sick passengers to the claim that a rodent-borne virus had begun spreading efficiently from person to person. There were comparisons with the first weeks of covid, evacuations in biocontainment units, weeks of enforced isolation, and a call from Deborah Birx for routine PCR testing of passengers who had no symptoms at all. The World Health Organization has felt the need to declare in public that this is not the start of another pandemic.

As always there are two primary false claims:

1. Human to human transmission – i.e. we are all at risk

2. 30% mortality rate – i.e. we are all going to die

Where did the alarm originate?

How good is the evidence that hantavirus spreads between people?

Hantaviruses spread between rodents, which carry it without illness and shed it for long periods. Occasionally a human inhales virus from rat faeces or urine and becomes sick. There is no evidence for human to human spread. Human beings are dead-end hosts: we become ill, sometimes severely, but we do not pass the virus on. The virus attacks the lining of small blood vessels rather than the airway itself, so even badly damaged lungs are not full of virus which can be exhaled and inhaled by others.

The single claimed exception is Andes virus in southern Argentina and Chile. The CDC states that Andes can pass between people after close, prolonged contact. However, direct person-to-person transmission of a hantavirus has never actually been observed. No one has watched one person infect another, isolated the same replicating virus from both, and, importantly, excluded a shared rodent source. The claim is an inference drawn from clusters of cases that occurred near one another in time and space. That is exactly what environmental exposure also looks like.

Human-to-human transmission of hantavirus is an unproven hypothesis that was born in 2020 and has recently been repeated until it sounds like reliable evidence.

The paper now being leaned on

The study being invoked to make sense of the cruise ship is the 2018–19 Epuyén outbreak, published in the New England Journal of Medicine in December 2020 and presented ever since as proof that Andes virus can spread between people. The story is presented as an index case of a symptomatic 68-year-old man who attended a birthday party infecting six of the guests seated near him who fell ill two to three weeks later, and cases over the course of 4 months in people who could be linked amounting to an outbreak of around thirty-four cases and eleven deaths.

It is a striking account, but it does not show what it is said to show. The index case had himself just been infected by rodents near his home, which means the entire village was living in an active rodent reservoir at the time of the party. Around a hundred people attended; five became ill. Five in a hundred is an unremarkable attack rate for environmental exposure in a place like that. Proximity to the sick man at the party is one possible explanation for those five cases. Shared exposure to the same rodents in the same valley, over the same weeks, or even at the party, is another, and the paper did not exclude it. Genetic sequencing later confirmed that the cases shared a single viral strain, but that is precisely what a common rodent source would also produce. Identical sequences tell you where a virus came from, not which direction it travelled between hosts. What they failed to ask, and which all scientist should ask, is what would the null hypothesis look like? If rat transmission caused a regional outbreak what proportion of cases would have had a social contact over the course of four months? Without that control no conclusions can be drawn.

No healthcare staff were sick in this outbreak but they claimed three instances of patient to patient hospital spread. In all three instances the patients had also been living in the infected town!

Other evidence

The only other claim was from Ferres and colleagues in 2007. They followed the household contacts of confirmed cases of an earlier outbreak and classified three of them as “definite” instances of person-to-person spread. The rest were labelled “probable” or “possible” — categories that rest on interviewer judgement rather than on anyone observing one person infect another. The definite claim was based on similar genetics without ruling out environmental exposure.

A deep problem applies to all of this work. The incubation period of Andes virus runs from 4 to 42 days, so two people exposed to the same infected rodents on the same day can fall ill more than six weeks apart. That looks exactly like one person passing the infection to the next, even when nothing of the kind has happened. Spouses share environments. Households are not independent; rural rodent exposure is constant and easy to miss.

A systematic review in 2022 looked at the whole literature and concluded that the balance of evidence does not support human-to-human transmission at all.

The numbers do not add up

The figure driving the fear is the case fatality rate, usually given as 30 to 40 per cent. Set it against the antibody surveys. In Jujuy province in northern Argentina, 6.5 per cent of the general population carries antibodies to hantavirus; a 2024 meta-analysis put seroprevalence across the Americas at around 2.4 per cent; in parts of the Gran Chaco it reaches 20 to 40 per cent. Antibody surveys are the only honest way to count how many people a virus has actually reached, because they pick up the mild and silent infections that never trouble a doctor and never enter a case register. They are not flawless — false positive results will exist — but they are the right instrument for the job, and no plausible margin of error in them comes close to reconciling these numbers with a death rate of one in three. If a third of all infections were fatal, the cumulative toll in northern Argentina would be impossible to miss. It is not there. The 30 to 40 per cent figure is the death rate among the sickest patients who reach hospital and are recognised, not the death rate across everyone the virus infects. The same Jujuy survey found many confirmed infections that never progressed to respiratory distress and a fatality rate of 13.3 per cent, well below the headline. Even mainstream coverage of the present outbreak now concedes that testing only the symptomatic inflates the apparent mortality. The disease is commoner, and milder, than the number that frightens people.

The cruise ship

A rat aboard a vessel working South American ports is hardly an exotic suggestion, and a single infected rodent in the food stores or ventilation would account for several cases over the following weeks without requiring the virus to have changed at all. The sequence recovered from a patient is an ordinary Andes virus from the regional rodent reservoir, with no significant mutations reported.

The timing fits this reading rather than the alarming one. The first patient fell ill five days after the ship sailed. With an incubation that can be as short as four days, that is most easily explained by an infection picked up at or just before boarding; it sits at the short end of the normal range, not outside it, and gives no reason to suppose the virus has gained some new ability to spread person to person.



The failure was not letting people off the boat earlier.

No change in total number of cases or in the virus

The WHO’s own 2019 report states that between 2013 and 2018 Argentina recorded an average of 100 confirmed cases annually. The Epuyén cluster of 34 fell entirely within that envelope — it didn’t push the national total above the normal range. So even if you accept every claimed transmission, the supposed “super-spreader” event produced no detectable signal in the country’s annual count. A virus newly capable of efficient human spread would do exactly the opposite.

The NEJM paper found the Epuyén outbreak was caused by a single spillover event, and the strain was closely related to ordinary endemic cases including the 1996 El Bolsón strain — i.e. nothing genetically new. The same is reported for the Hondius: ordinary Andes virus, no significant mutations.

Premature certainty, again

The ingredients here are familiar: very small numbers, a frightening name, an incubation period long and variable enough to disguise coincidence as a chain, and a public conversation that has raced ahead of the evidence. The remedy now being proposed — PCR testing of passengers who have no symptoms — is the one least suited to the situation. When the prior probability of infection is low, most positive results from a highly sensitive test will be misleading, which is the lesson the last few years should have taught and apparently did not.

Andes virus may, on rare occasions and after prolonged close contact, pass from one person to another — though even that has never been properly proven. What is certain is that this is a long way from a virus capable of seeding an outbreak through a ship’s air handling. The fear is moving a good deal faster than the virus, and on the evidence so far the virus is doing nothing it has not always done.

Source: HART's Substack