Virulence and transmission

Why are diseases we catch by touch usually milder than those carried to us by mosquitoes?

There is a comfortable old story that says pathogens always evolve towards mildness, because a parasite that kills its host destroys its own home. It is tidy, it sounds like common sense, and it is wrong. Malaria and cholera have been with us for a very long time and neither has become gentle. Meanwhile the viruses we pass by handshake and doorknob are, on the whole, a nuisance rather than a threat. What is doing the sorting, if not simple prudence?

Start by asking what selection actually acts on. Not the pathogen's comfort, and not ours — only on how many new hosts a lineage infects before it stops. So ask what damage buys. Damage is largely a by-product of replication: more virus or bacteria in the tissue means more shed into the world, and more chance of reaching someone else. If harm were free of any benefit, it would simply be selected away. It is not free.

But push replication further and something turns. A host who is dead, hospitalised, or too ill to leave the room stops meeting people. So the same variable — how hard the pathogen multiplies — raises the rate of transmission while shortening the duration of it. Two opposing effects on one dial. Where do you expect the total to peak?

Not at zero, and not at the extreme. Somewhere in between. That is the trade-off hypothesis, formalised by Anderson and May in the early 1980s: selection favours whatever level of virulence maximises lifetime transmission, and there is no reason that level should be harmless. Mildness is one possible answer to the optimisation, not its inevitable destination.

Now the part that answers your question. The trade-off's second term — the shortening of the infectious period — depends entirely on how the pathogen gets to the next host. Suppose transmission requires the host to walk about and breathe near people. Then immobilising them is catastrophic for the pathogen, and the optimum sits low. Now suppose a mosquito comes to the host, or the host's diarrhoea reaches a water supply, or a nurse's hands carry the organism to the next bed. A prostrate patient transmits perfectly well — arguably better, since a feverish, still, uncovered body is easier for a mosquito to find, and severe cholera produces vastly more infectious fluid than mild cholera. The penalty for immobilising the host has been paid by somebody else.

This is Paul Ewald's argument, and it makes a testable prediction: vector-borne, waterborne, and hospital-acquired pathogens should tolerate higher virulence than those needing a mobile, sociable host. Malaria, yellow fever, cholera and typhoid on one side; rhinoviruses on the other.

Does the evidence hold? Partly. The clearest natural experiment is myxoma virus, released into Australian rabbits in 1950 at a strain killing over 99% of those infected. It did not become benign. It settled at intermediate grades killing most but not all rabbits — roughly 70 to 95% — exactly where a trade-off model would put it, while the rabbits independently evolved resistance. Beyond that case the record is mixed: the trade-off has support in some systems, including the relationship between HIV set-point viral load and both infectiousness and time to AIDS, and weak or absent support in many others. Ewald's comparative data have been criticised for treating related pathogens as independent observations. Treat the framework as a good way to ask the question rather than a law that settles it.

Three refinements matter.

First, virulence is not a property of the pathogen alone. It is what happens when this organism meets this host, and much of the damage in severe disease is done by our own immune response rather than by the microbe directly. The same strain can be trivial in one host and lethal in another.

Second, transmission mode is a tendency, not a rule, and counterexamples are easy: smallpox spread by ordinary close contact and killed around 30% of those it infected. Other pathways to high virulence exist — a pathogen that mainly circulates in another species has little reason to spare us at all, which is why rabies and Ebola behave as they do in humans.

Third, and most importantly, the trade-off model predicts intermediate virulence, not declining virulence. A newly emerged pathogen may evolve up as well as down, depending on where it started relative to its optimum. The practical payoff is the interesting bit: if transmission mode shapes the optimum, then clean water, mosquito nets, and hospital hygiene are not only barriers to spread. By making it costly again for a pathogen to flatten its host, they change what selection rewards.

• Why a pathogen circulating mainly in another species has no optimum with respect to us at all.
• How vaccines that block disease but not transmission may shift the optimum upward.

Every term the collection defines is gathered in the glossary.