Thread by @maiamajumder

New pre-print by myself & @mandl:

Early basic reproduction number estimates for #nCoV2019 range from 2.0 to 3.3 (based off of publicly reported confirmed cases through 1/22/20 & subject to change) [ssrn.com/abstract=35246…].

Short explainer & several caveats in the thread below.

@mandl The basic reproduction number (R_0) is a measure of transmissibility that aims to describe the average number of people a new case *in a fully susceptible population* will infect. (Most of the time, this number isn’t actualized thanks to interventions as simple as hand-washing.)

@mandl Today, the @WHO reported their own estimates (though I haven't seen the methods yet), which were R_0 = 1.4 to 2.5. Their estimates are within the bounds of those we obtained on 1/18/20 (as shown in Panel B of the figure above); our estimates skew slightly higher by 1/22/20.

@mandl @WHO (There are many possible reasons for this slight difference, but I can't say for sure until I've seen the methods used by WHO researchers. Needless to say, the overlap should be encouraging given that we used public data and [I assume] they had access to more info than we did!)

@mandl @WHO For the related #SARS-Coronavirus, estimates for R_0 range from 2-5, so our early estimates for #nCoV2019 (though very preliminary) shouldn’t come as too much of a surprise. That said, they’re likely to fluctuate (perhaps considerably so) as more data (and info) become available.

@mandl @WHO The model we’ve used for our R_0 estimate is phenomenological, which means that it doesn’t aim to *explain* what’s happening on the ground but rather to *describe* it. This is a good option in information-scarce situations, like at the start of a novel viral outbreak (AKA now).

@mandl @WHO However, *because* we’re at the start of a novel viral outbreak, there are a number of critical assumptions we had to make to run the model. The first is that human-to-human transmission is happening in a meaningful way, which the WHO has already suggested (and the data agree).

@mandl @WHO The second is that the average time between two consecutive #novelcoronavirus cases in a chain of transmission – the serial interval – is similar to the related #MERS-CoV & #SARS-CoV. (We made this assumption because we don't know what the serial interval for #nCoV2019 is yet.)

@mandl @WHO The third is that the cumulative case count data that have been made publicly available so far are reasonably accurate. Thankfully if this assumption is wrong, we can easily recalibrate the model to reassess the situation... Which is exactly what we intend to do moving ahead.

@mandl @WHO As more information becomes available, we fully intend to report changes to our estimates as publicly and transparently as possible. This early in an outbreak, preliminary estimates like ours may be useful for decision-making but should be considered fluid and ever-shifting.

@mandl @WHO I bet I’ll have more thoughts later, which I’ll add to this thread. However, for now: please be advised that this work is a *pre-print*. This means that it has not yet undergone peer review and as a result, our findings should be treated as provisional. Thank y'all! [/fin]

As promised, I’m back! Overnight, @C_Althaus and his team shared their own R_0 estimates using stochastic simulations (which is a method very different from ours). With this in mind especially, I’m very encouraged that our results are so similar. This is open science at work.

https://twitter.com/C_Althaus/status/1220646280449462274

Another day, another update to this thread! Yet another R_0 estimate is out (with yet another set of methods), yielding mean results (~2.6) similar to those reported by both our team and Christian’s team. Useful summary thread too. Work by @MRC_Outbreak, including @neil_ferguson.

https://twitter.com/mrc_outbreak/status/1221032729971773440