Stopping the SuperSpreader Epidemic: the lessons from SARS (with, perhaps, applications to MERS)

Abstract

I discuss the so-called SuperSpreader epidemic, for which SARS is the canonical examples (and, perhaps, MERS will be another). I use simulation by an agent-based model as well as the mathematics of multi-type branching-processes to illustrate how the SS epidemic differs from the more familiar uniform epidemic (e.g., caused by influenza). The conclusions may surprise the reader: (a) The SS epidemic must be described by at least two numbers, such as the mean reproductive number (of "secondary" cases caused by a "primary case"), R0, and the variance of same, call it V0; (b) Even if R0 > 1, if V0 >> R0 the probability that an infection-chain caused by one primary case goes extinct without intervention may be close to one (e.g., 0.97); (c) The SS epidemic may have a long "kindling period" in which sporadic cases appear (transmitted from some unknown host) and generate a cluster of cases, but the chains peter out, perhaps generating a false sense of security that a pandemic will not occur; (d) Interventions such as isolation (or contact-tracing and secondary case isolation) may prove efficacious even without driving R0 below one; (e) The efficacy of such interventions diminishes, but slowly, with increasing V0 at fixed R0. From these considerations, I argue that the SS epidemic has dynamics sufficiently distinct from the uniform case that efficacious public-health interventions can be designed even in the absence of a vaccine or other form of treatment.