COVID & Superspreading Part 1: Introduction

What makes COVID-19 unique from the perspective of epidemiological dynamics?   In this multi-part series, we talk about over-dispersion: notion that a small fraction of infectious individuals/events account for large portion of transmissions.   Be ready for some math.

By now, we all have heard of R0 (or Rt). Overall, it’s a really useful parameter particularly for assessing whether infections are growing in # over time.  However, over-reliance on this parameter lends to an oversimplified view of an epidemic. (e.g. R0 = 3).

The truth is R0 is merely the AVERAGE # of secondary infections occurring per case.  It doesn’t tell us how transmission is varied across individuals or changes with time.   R0=3 can present in varied ways:

So what pattern does COVID-19 fit into?    We can look at Hong Kong as an example.  According to Dr. Dillon Adams (@DrDCAdam ) publication in Nature Medicine (https://nature.com/articles/s41591-020-1092-0… ) from Jan 23-April 28, Hong-Kong identified 1038 confirmed cases (of which 18.8% were asymptomatic).  They found 3 major superspreading events:

1. a band that performed at multiple venues,

2. a wedding, and

3. a single monk who infected others in multiple temples. 

However, if you noticed most other cases led to no subsequent spread.  For every “superspreader”, there are multiple “dead-enders” – basically where infection met an end to transmission.   In their analysis, 70% cases had ZERO transmissions.

Many have heard that spread of COVID-19 is “overdispersed” and that 10-20% of infected cases are responsible for 80% of transmissions.  This puts into perspective the importance of  superspreading as a major player in epidemiological dynamics. So how can you quantify this dispersion?   There are basically 2 methods:  (1) Negative Binomial Distribution (NBD) and (2) Extreme Value Theory (EVT).

In the next part, we will focus on NBD.

End Part 1/