Thread by @MichaelSFuhrer

In a recent thread I looked at the performance of some low-covid countries against expectations from models of the expected R_eff achievable at different vaccination levels.

Today let’s examine how jurisdictions in Oceania are doing.

Thread.
1/🧵

https://twitter.com/MichaelSFuhrer/status/1441282658693124096

The effective reproductive number R_eff controls whether infections grow (R_eff > 1) or decay (R_eff < 1). We therefore need to achieve R_eff < 1 to have control over the epidemic with our public health and social measures (PHSMs).

2/🧵

The most important question then, is:

👉"Under what conditions of PHSM and vaccination can we achieve R_eff < 1?"👈

I’ll be plotting R_eff as a function of the effective vaccination expressed as a percentage of total population.

3/🧵

Details: I use generation time = 5d, and the week-on-week increase to get R_eff.

To get an “effective vaccination” I count 1-dose individuals as 2/3 of a 2-dose individual.

I plot R_eff vs vax rate two weeks before (to account for time for vax to reach effectiveness).

4/🧵

I assume vax effectiveness against onward txn 89.6%, similar to the avg in Doherty Inst report for AZ/Pf. However, I assume all ages are equally involved in txn, and vax affects the population uniformly (poor assumptions, but generally underestimates vax effectiveness).

5/🧵

Here’s the predicted R_eff from Doherty (points), along with the uniform vax approximation (thin lines), for various public health and social measures (PHSMs) and partial test/trace/isolate/quarantine (TTIQ). Effectiveness of TTIQ and PHSMs from Doherty, w/R0=8 for delta.

6/🧵

The uniform vaccination model produces straight lines for R_eff vs vax. The y-intercept is R_eff at zero vax, and the x-intercept is the inverse of the vax effectiveness.

7/🧵

You can see that the Doherty model gives slightly lower R_eff at a given vax level. This comes from the transmission matrix which weights transmission more towards working-age adults; vaccinating those ages allows vaccination to overperform the uniform-vax expectation.

8/🧵

Here’s how NSW, VIC, ACT, and NZ are tracking.

NSW tracks very close to the Doherty prediction for “high” PHSM (equivalent to Stage 4 lockdown) and “partial” TTIQ, as expected.

9/🧵

Perhaps surprisingly, ACT and NZ track very close to NSW, though their very low caseloads should allow for “optimal” TTIQ (dashed line), and NZ has until recently had a stricter lockdown protocol.

10/🧵

VIC is the underperformer, tracking worse (also with larger fluctuations), though dip around 23% vax is the near elimination of delta during lockdown #5.

No evidence that NSW is performing worse due to high caseloads impacting TTIQ, at least compared to ACT, NZ.

11/🧵

We can compare to other low-covid countries (<500/M detaths from covid over entire pandemic, suggesting infections <10% of population, in some cases much less). For each I plot only roughly the phase in the epidemic corresponding to delta.

12/🧵

Here is NSW with the Nordic low-covid countries:
NSW outperforms them, however, none of the Nordic countries are in lockdown, and are closer to “low” or “baseline” PHSMs. In that sense, they all outperform the Doherty predictions for their restriction levels.

13/🧵

Here is NSW with low-covid Asian countries (Japan, Singapore, South Korea). Japan looks a lot like NSW. South Korea seems to chug along at R=1 independent of vaccination. Singapore has recently had a big increase in cases.

14/🧵

Others have held SG out as a cautionary tale of re-opening gone wrong. Misses the point that SG is one of the most regulated technocratic countries on Earth. If they wanted to control covid, they would.

15/🧵

I conclude that SG gov't has made an intentional decision to tolerate higher cases, and this seems consistent with SG gov’t messaging. Though, they have recently re-imposed restrictions, so 🤷

16/🧵
sg.news.yahoo.com/moh-stop-provi…

We can also compare NSW to a low-covid Asian nations controlling delta with lockdowns (Thailand and Vietnam). Both outperforming NSW, achieving R=1 at much lower vax. (Note: I know little about the PHSMs or vax programs in these countries, they may be highly non-uniform.)

17/🧵

What have we learned?

➡️Most countries show R_eff within the bounds of “baseline” to “high” PHSM with “partial” TTIQ predicted by Doherty (and uniform vaccination). Some do quite a bit better than the best expectations from Doherty of “high” PHSM + partial TTIQ.

18/🧵

➡️The Doherty model therefore appears roughly realistic. It appears to be a good predictor for NSW.

➡️Many countries show a trend towards lower R_eff with high vaccination. The trends aren’t clear enough to say anything about whether the x-intercept...

19/🧵

... is correct (vaccine effectiveness), but there is evidence that vaccination has an effect on R_eff of roughly the order of magnitude predicted.

➡️Paradoxically, jurisdictions with the lowest caseloads (ACT, NZ, SK) don’t seem to have any easier time compared to...

20/🧵

... higher caseload jurisdictions. This is further evidence that elimination to zero covid may be untenable in the delta era.

➡️Achieving R<1 is the norm, not the exception. All the low-covid countries with delta outbreaks have achieved R<1...

21/🧵

...DK, FI, NO, SG, SK, JP did so without lockdown, with assistance of vaccination. SG subsequently has seen R>1. TH and VN achieved R<1 with lockdowns, at substantially lower vaccination than NSW.

22/🧵

➡️The level of PHSMs does not appear to make as dramatic a difference in R_eff as predicted by Doherty. The Nordic countries are probably closer to “baseline” PHSM but tracking between “low” and “medium”. SK and JP are performing similarly to NSW without lockdown.

23/🧵

Potential problems with this analysis:

➡️NSW, VIC, NZ, ACT have delta outbreaks in winter/spring, while the others are either in summer/autumn or are tropical. This may explain the overperformance of comparison countries with respect to Oceania.

24/🧵

➡️There are many other differences in public health responses, demographics, culture, etc. which might confound comparisons among these countries.

25/🧵

➡️R_eff from country-wide case data may poorly estimate the actual transmission potential. It also misses inhomogeneity within a county; for example, some provinces of Vietnam have eliminated or nearly eliminated delta.

26/26🧵