This blog post has received a lot of attention.

A thread.
In short, the blog post takes public misconceptions about the technical report from Doherty, makes a cursory read of the report, repeats misconceptions, accuses Doherty of intentional scientific misconduct to lead Australia to disaster.

Crackpot stuff.
Not really worth taking apart. But it’s been retweeted by a number of figures who should really know better. So let’s take a look.

Doherty Institute report is here if you want to read along.

The bulk of the blog post is based on a fundamental misunderstanding of the purpose of the section “Dynamics and consequences given timing of transition to Phase B”.
The complaint is that the models were only run for 180 days, so they give a misleading picture of the hospitalizations/deaths in each scenario.

Here's a relevant plot from the blog - looks scary, right? They cut off the bad stuff!
This section models what will happen assuming online baseline public health safety measures were implemented, and partial effectiveness of testing/tracing/isolation/quarantine (TTIQ).
This is a “worst case” scenario. It asks, what happens if we open up and let the epidemic rip, and do nothing to stop it?

Why run this model at all? The answer is in the title: to understand the *dynamics*.
This is intended to answer: In the worst case scenario, how fast does the epidemic get away from us? If it is slow (e.g. takes 180 days) then we have a chance to react with PHSMs, further vaccinations, etc.
Why are dynamics important?
The Q is, if we know we’re going to reach 80% vax, why not open at 50%/60%/70%, since we’ll get to 80% quickly enough? The answer is that things might get out of control rapidly, overwhelming TTIQ, and then health care, before we get to high vax.
Why cut it off at 180 days?

Remember that the *assumptions* are no PHSMs and no more vax. Under those assumptions, the *only* thing that stops outbreak is add'l infection-acquired immunity. # of infections is solely determined by the # vaxxed, and that’s capped at 80%.
So the long-term outcome is exactly the same in every simulation!

If you’re an epidemiologist, this is bleeding obvious, so you don’t need to state it.

But, if you’re an electrical engineer, it looks like a conspiracy.
We can check that this is the case by scaling the plots and placing them on top of one another.

If you open completely at 80% you get the same outbreak as if you had opened at 70% and proceeded to 80% 22 days later, only delayed by 40 days.
Anyway, here’s Doherty’s slide on the results. The box in upper right tells you that this is the worst case, and will be lower with PHSMs in place.
Next complaint is that Doherty is based on maintaining partially effective TTIQ. “Partial” is based on VIC’s experience at the peak of 2nd wave. To me, that's pretty ineffective. The post argues NSW is worse. Maybe, I don’t know. I’m not sure that makes Doherty misleading.
The next complaint is that Doherty starts their dynamics simulations at 30 total infections, whereas NSW has far more active cases. That’s a reasonable complaint.

But the simulations illustrate the *dynamics*, and already assume partially effect TTIQ.
So it’s simple to just read off what would happen if you started at 1,000 infections/day – it just advances where you are on the curve.
If NSW hits 80% at 1,000 cases/day, they're here.

Obviously this is bad news if you’re NSW and you plan to get to 80% vaccination, end all PHSMs, and just stop vaccinating. But that’s not the plan anyway.
NSW will have little breathing room. So they'll need to relax restrictions slowly. But the rhetoric from NSW gov't indicates they understand this. It's important to hold NSW to it, but this doesn't make Doherty misleading.
The next part comes from a basic misunderstanding of why this plot from Doherty is on a log scale on the vertical.

Answer: the various measures to reduce R are multiplicative, which becomes additive on a log scale.

Simple as that.
In other words, if TTIQ reduces Reff by half, and 60% vax reduces Reff by half, then it goes from 8 to 4 and 4 to 2, whichever order you do them in.

Boxes of the same height on this plot have the same size effect (e.g. reduce by a factor of 2).

Not a conspiracy.
Then there’s the usual cherry-picking of data…

Israel has rapid growth of cases with delta and high vax, therefore the transmission assumptions must be wrong!

But Denmark has delta, and zero growth of cases with high vax, modest PHSMs.
Last, the conclusion suggests that there is an alternative, which is, I guess, suppress delta to zero with PHSMs.

First, this is entirely in line with Doherty; the plan is to use vaccination and mild to moderate PHSMs to suppress outbreaks, lockdown only if abs necessary.
But the counterexample given is that Taiwan suppressed their delta outbreak with PHSMs alone, so maybe we should just do that.

But, this is incorrect, Taiwan's outbreak was alpha. No country has suppressed a delta outbreak the size of NSW's with PHSMs alone.
Long thread. Bottom line: it's a technical report. And it's technically correct. The data aren't presented in a way that is *intentionally* misleading. They might be presented in a way that they might mislead a non-epidemiologist, but that is not the intended audience.

• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with Michael S Fuhrer

Michael S Fuhrer Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @MichaelSFuhrer

15 Sep
The UK is often held up as a cautionary tale regarding covid and re-opening.

Let's have a look at what happened in the UK and see if there are parallels to what is happening in Australia.

1/🧵 Image
At the beginning of 2021 the UK was fighting a crushing wave of alpha with months of lockdown. As that wave receded, the UK began to release restrictions.

Restrictions were released at a very early stage of the vaccination program:

The UK "picnic day" and end of local-area restrictions to movement occurred on 29 March at 5.6% of total population vaxxed.

Outdoor dining/pubs opened 12 April at 11.5% vax.

3/ 🧵
Read 12 tweets
13 Sep
Some first impressions from the OzSAGE model (

1)Not enough info is given to understand how the model works. They don’t explain how R_eff is calculated, for example, from the restriction settings.

@RichardfromSyd1 @OrinCordus @RageSheen @Globalbiosec
2)They don’t explain their assumptions about ongoing vaccination beyond 80%. This is key to understanding the model and it’s left unstated.
3)They say they adapt a peer-reviewed model, but that model was not used to model time-varying restrictions, and in fact did not include masks or lockdowns at all.
Read 12 tweets
1 Sep
Here is a long/technical thread on my attempt to reverse-engineer the assumptions from the Doherty Institute report to the National Cabinet (linked) regarding delta severity, PHSM effectiveness, and vaccine effectiveness to match observations.
Here are the figures from the Doherty Institute report. Note that there are a couple versions of these figures in the report; these assume the “all adults” strategy for vaccination.
The figures attempt to show graphically the effect of various interventions on the transmission potential (TP). TP is in essence R_eff, but calculated on the population level, as I understand it. When TP>1, cases grow, and TP<1, cases decline.
Read 34 tweets
23 Aug
It is mind-boggling how fast one’s worldview needs to adapt to changing reality during this pandemic.

Taking stock of some new realities:
1) Delta is tougher than we thought. It is now safe to say that, even at current vax, delta is marginal to escape optimal TTI + lockdown. Some outbreaks fizzle out. But when they don’t, short/sharp lockdowns have become long at best, uncontrolled outbreaks at worst...
...This is ultimately inevitable, and your risk analysis for vax should account for the fact that everywhere is either in outbreak now or only months/weeks from outbreak. Get whatever vax is available to you, as soon as possible, and encourage everyone you know to do the same.
Read 11 tweets
21 Aug
10 days ago it was conceivable that NSW could control delta and reach low covid (by int'l standards) well before 80% vax, safely open and reduce covid simultaneously.

Now appears that NSW will barely control covid, at high caseloads (~500/Mpop/day) when reaching 80% vax.
1/ Image
Lines on the graph are extrapolations from 9, 14, and 20 August. In 11 days the extrapolated date of effective* 80% vax of 16+ has moved forward 15 days, and the date of achieving R=1 through vax has moved backwards by 22 days. Separation has gone from 6 weeks to 4 days.
This is a result of the higher observed R, which seems robust.

Likely scenario: NSW will achieve 80% vax of 16+ in October after 3.5 months lockdown, with cases in several 1,000s/day. Pressure to remove lockdown will be intense.
Read 6 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!