Hypochondriacs might be good for everyone’s health, to judge from a study that’s just been published. The paper describes a computer model of the uptake of vaccinations by a population, and shows that just a few individuals whose fear of the disease prompts them to take a vaccine can trigger others to do the same. As a result the population as a whole is more likely to reach a critical mass of vaccinated people, in other words reach “herd immunity”.
With the appearance of potentially fatal infectious diseases such as swine flu and bird flu (H5N1), not to mention the threat of AIDS and other sexually transmitted diseases, forecasting epidemics has become a central concern in modern medicine. It’s vital both for developing strategies for mitigation and for assessing the emergency measures that will be needed if a major outbreak occurs.
Time was when modelling epidemics tended to involve rather simple models in which virtual people – or other animals – were situated on a checkerboard-like grid and infected their neighbours. While this basic idea – that you risk catching a disease from others with whom you come into close proximity – remains at the core of all epidemic computer models, these days programmes such as the US National Institutes of Health’s Models of Infectious Disease Agent Study (MIDAS) make use of extremely sophisticated schemes that try to capture the way real people move around (international travel makes it easy for diseases to hop continents), how they are connected in social networks, and how they make decisions about prevention or avoidance.
The trick with these methods is to not lose sight of the wood for the trees: to distinguish the kinds of things one might expect in the real world from outcomes that just depend on a specific model’s details. That’s why relatively simple models (albeit complex compared to those a decade or so ago) like the one used by Xiao-Tao Liu of Lanzhou University in China and colleagues are still valuable.
Liu and coworkers say that most epidemiological models tend to assume that people all behave in much the same way, namely that they do what others do. But that may not actually be the case. The researchers propose that a few will be “committed” to vaccination, perhaps because they have a low threshold for feeling at risk (or strongly held convictions), and they will want to be immunised as soon as they hear that someone is infected. If the model of Liu and colleagues stands up, it implies that these committed individuals can have a disproportionate effect, seeding vaccine uptake in others while also disrupting clusters of people forming who reinforce each other’s refusal to receive a vaccine.
There are two important messages to take away from a study like this, regardless of whether one thinks the model captures any real-life behaviour. First, epidemic outbreaks may be very sensitive to the details of how we make our choices about health, and in particular how those choices come about through the influence of other people. Second, the outcomes are often non-intuitive: you’d never guess, by looking at the decision-making rules alone, what the overall outcome might be.
The basic problem for vaccination programmes is that of the free-rider. If there’s an individual cost to vaccination, whether it is inconvenience or a perceived risk of the vaccine itself, people might opt not to get vaccinated themselves but to count on the protection that comes from everyone else doing so. If too many people do this, the herd immunity crumbles and the whole population stands at risk – as seemed to be happening during the scare over the (spurious) link of the MMR vaccine to autism.
An earlier study of vaccination behaviour, in which people were assumed to imitate others who adopted “successful” strategies, predicted that the proportion of a population that takes a vaccine could oscillate over time – first high, then low, and so on. Such oscillations are often a sign of a precarious balance of influences, which push the system first one way and then the other. You find them, for example, in the interdependence of predators and their prey in ecology, or in traffic flow, where they might lead to a jam. In such cases, small influences can have disproportionately big effects. That’s one reason why Liu and colleagues wondered whether individuals committed to vaccination might affect the overall behaviour markedly even if they were few in number.
In the researchers’ model, each individual (called an agent) decides whether to vaccinate or not in successive “rounds” of decision, which can correspond for example to successive “flu seasons”. This choice is made after seeing the outcomes of decisions made both by oneself and one’s immediate neighbours. If, for example, you didn’t vaccinate and did get infected, then in the next round you might copy someone who did get vaccinated.
But there is also a certain fraction of agents dispersed through the population that is always committed to vaccination. The researchers find that even a relatively small proportion of these agents (say, 5-10%) can significantly reduce the scale of an epidemic. What’s more, the effect is much stronger when all the agents are assigned specific locations on a checkerboard array – that is, in each round they “meet” the same the people they interact with regularly, like we do in reality – rather than being randomly reshuffled into a homogeneous mixture. Above a certain threshold proportion of committed agents, an epidemic can be confined to patches, rather than spreading through the whole population. And the committed agents can disrupt the appearance of clusters of free-riders, which might otherwise seed the emergence of a global epidemic.
Do such committed individuals really exist, though? It seems plausible, but Liu’s colleague Zhi-Xi Wu at Lanzhou admits it is still a supposition. However, if they do then the results imply that they can act as influential “role models” for others, and are most effective if they are dispersed quite uniformly in the population. The same should apply to other behaviours: a few committed individuals who always practice safe sex might limit the spread of sexually transmitted diseases, and perhaps even non-medically-related behaviours such as responsible energy use and recycling could be propagated this way – not just among individuals but, say, businesses too. That could be an empowering message: one all too often hears the lament that individuals can’t make much difference to the world, but the power of social feedback implies that you might have more influence than you think.