The traditional approach to immunization policy (an appropriate concern in the U.S. right now, given the current panic over flu shots) typically involves trying to immunize as many people as possible in order to cut down a bug's chances of spreading throughout the population. This doesn't work all that well -- it turns out that, on average, you need to immunize 95% of the population if you're just getting a random sample. While it may be theoretically possible to immunize 95% of a population, the financial and logistical challenges are fairly daunting.
But human societies are not random. We have networks of interaction, easily demonstrated by checking out the various social software websites out there (Friendster and Tribe.net being two of the better-known ones). And when you start thinking about human behavior not as random individuals but as networks, you can come up with new ideas about immunization.
Human networks of acquaintances, computer networks like the Internet, and interacting protein networks in the body, all share a characteristic layout: most of the elements have only a few links to others, while a few individuals have a very large number of links. If one of these highly connected individuals in a human network becomes infected, she can become a "super-spreader," infecting all of her contacts and efficiently distributing the disease. This structure suggests a deceptively simple solution to the vaccination question: immunizing all the super-spreaders in a network slows or stops the spread of a disease as effectively as destroying a country's highway interchanges would stop traffic.
Reuven Cohen and colleagues at Israel's Bar-Ilan University have found that, rather than trying to immunize everyone in hopes of hitting the "super-spreaders," randomly selecting 20% of a population and asking each to name a single acquaintance, the immunizing that acquaintance, is an effective means of focusing in on those most at risk of spreading an infection to a large number of people. With a large enough population, you can even take a smaller sampling and only go after acquaintances mentioned by two or more people and still get great results.
Two quick thoughts:
First, it's interesting that this approach is only now being seriously addressed, since the "super sneezer" phenomenon has been known since at least the time of "Typhoid Mary" Mallon around the turn of the 20th century. Goes to show how much weight status quo thinking can carry.
Second,much like the implications for using actual lifestyle/behavioral data for health or life insurance purposes, there are serious implications from a public policy perspective. What sort of rights would a person "suspected" of being a mega-vector have? How about someone triangulated as a major spreader of AIDS? What level of statistical "proof" would be required before an entity (governmental/private/?) could take action? What sort of action?
It seems unlikely that, given the reluctance of many people to get vaccinated even when the treatment is available, we could rely on the super-spreaders to voluntarily get treated. Then what?
Okay, that's more than two thoughts, but the riffs on this, and the implications for public policy, are mind-boggling.
Good observations. It's worth noting that the "super spreaders" aren't key vectors because they're particularly infectious, but because they're particularly social. But because social behaviors change (people lose jobs, move, have falling-out with friends, etc.), a person identified as a social network node may not maintain that position over time. Simply tagging such people as "always immunize" (or quarrantine...) would likely have rapidly diminishing returns.
Just two comments:
- Curing the epidemic is in principle in contraddiction with curing the single patient, and the decision about which of the two is a priority is (should be) more a political decision than a technical issue.
-The need for an optimisation of the vaccination strategies is essentially due to the scarce resources that we, as richer countries, are willing to spend for these problems, and this clever trick should not be a way to forget this.