Here's the graph for Anzac Day, Saturday 25th April from Chris McDowall | The Spinoff

An extension of the plot showing our progress under L4 restrictions, projecting forward for the next month.

A few comments/notes

• NZ showing clearly exponential decrease in local case numbers over the last 2.5 weeks of L4 lockdown
• Halving time for new local cases is ~5 days (somewhat longer than the doubling time was initially on the route up!)
• Forward projection is based on the last 2.5 weeks data (black part of the NZ curve) and is applicable only for the current L4 restrictions
• From NZ, HK, Taiwan & other case details it seems some infections are being confirmed up to 3 weeks (and occasionally longer?) after likely infection date.
• Hence have adopted no new local case for 3 weeks as 'effective elimination/eradication(?)'; ie. once drop to a rate of 1/3 weeks.
• New local cases can still occur after that due to infection from imported cases (= new outbreak?).  Possibly also if there are undetected local cases.  However, if infectious the probably of those not showing up as new local cases in three weeks seems very low.  Even if asymptomatic, others infected by the undetected case would start showing up within a three week period?
• On the trend apparent so far, under L4 restrictions it would take us another - 

   

  • ~1 week to drop to 1 new local case per day.
  • ~4 weeks to reach 'effective elimination'.  
• It is going to be very interesting to see what impact moving to L3 has on the downwards trend.
• It will likely be 1 - 2 weeks after L3 starts before the impact starts to become apparent
• The daily new cases show a distinct weekly cycle and sensitively to lab testing or (it appears less so), numbers of tests samples taken.

   

  • An uptick of new local case numbers on each of the last three Thursdays. These will typically be from samples taken 2 days previously, and lab testing performed the day before.
  • A small down tick on 13 April (Easter Monday).  
  • Another (larger) down tick on Monday 20 April.  There were fewer than normal test samples taken on both 19 & 20 Apr.  Seems possible this down tick was caused by less lab testing on the Sunday (assuming that was the case)?
• The weekly cycle is not obvious when plot 'new' cases announced in the press releases vs press release date (see earlier post #2471087).  But there was an eerie parallel between the NZ & AU case numbers (up ticks & down ticks in sync) so I suspect the underlying cause is the weekly testing / lab testing cycle.
• Assuming L3 does not have a big impact on the downwards trend, it seems to me that L3 should be retained for at least three weeks to minimise the risk of undetected local cases causing additional new local cases. 
• It could take 2 weeks before the impact of dropping to L3 becomes fully apparent - even more reason not to end L3 after only 2 weeks. 
• It will though require at least four weeks of L3 (assuming L3 impact small) for 'effective elimination'.
• Main threats probably come from some of our larger bubbles - rest homes, hostels, large family bubbles (which potentially may get bigger / more of under L3)  .

Note: plot shows average daily rate of new local cases per million population.  Daily rate = #new cases/time since last local case.  For NZ, we have been getting new cases every day so the 'time since last case' is 1 day.  But HK & Taiwan have dropped below that level so their 'time since last case' is now more than 1 day; eg.  Yesterday HK had their first new local case in 6 days so the value plotted is (1/6)/population in millions (7.48).  This value has been plotted at yesterdays date (25/4).  From an epidemiological(?) point of view it would possibly be better plotted at mid-point of the time interval since the penultimate local case.  The trend does look a bit more rational plotted that way (more linear downwards slope - steeper than NZ).  However, in the plot I have left it plotted at the end date of the interval to better show when the last local case occurred.  In the case of Taiwan, that was 12 Apr; ie. no local case for 13 days so far (to yesterday).

NZ local cases include confirmed and probable. Same is true for HK though they have had only one probable case.  Not so clear for Taiwan as they do not provide the data in a table.  Have to wade through multiple scores of daily reports.  The data in those is as at date of report when some cases are still under investigation.  Note also that in HK and Taiwan all confirmed cases are hospitalised during isolation.

Data sources

• NZ:  https://www.health.govt.nz/our-work/diseases-and-conditions/covid-19-novel-coronavirus/covid-19-current-situation/covid-19-current-cases/covid-19-current-cases-details (last accessed 25 April 2020)
• HK:  https://www.chp.gov.hk/files/pdf/local_situation_covid19_en.pdf (last accessed 26 April 2020)
• Taiwan: https://www.cdc.gov.tw/En/Bulletin/List/7tUXjTBf6paRvrhEl-mrPg?page=1 (last accessed 25 April 2020)

Incidentally, the new local case yesterday in HK was a 'close contact of an imported case', so perhaps a 'new outbreak' ...?

But this does highlight that going forward, tight quarantining of arrivals from overseas will be important.  I do have concerns that 2 weeks may not be sufficient in all cases, at least without two negative tests at the end of the two weeks. 

The HK data include a significant number of 'possibly local' cases that appear often to be people who returned to HK more than two weeks before they tested positive (many up to only three weeks though).  These cases presumably cannot be classified as imported because they come out of quarantine before testing positive.  Hence cannot rule out local infection post-quarantine?

A couple of (very) recent articles re asymptomatic/presymptomatic 

Presymptomatic SARS-CoV-2 Infections and Transmission in a Skilled Nursing Facility
https://www.nejm.org/doi/full/10.1056/NEJMoa2008457

and related editorial

Asymptomatic Transmission, the Achilles’ Heel of Current Strategies to Control Covid-19
https://www.nejm.org/doi/full/10.1056/NEJMe2009758

Here's the graph for Sunday 26th April from Chris McDowall | The Spinoff

Here's the graph for Monday 27th April from Chris McDowall | The Spinoff

Could someone please clarify a point for me regarding probable cases. Suppose we have an existing known cluster event, and Person A gets infected maybe indirectly. They develop only mild symptoms and do not present themselves for testing. Person A then infects Person B. While the the virus is incubating in Person B, Person A gets better. Person B then gets sick and gets tested. During contact-tracing, Person A is interviewd and discloses that they had had mild symptoms. So Person A counts as a probable case? But obviously they are not a new case.

So yesterday when the headlines say '5 new cases, 1 confirmed, 4 probable' , it could be that the 4 are not new, they are just old cases completely recovered, and discovered by contact-tracing interviewing?

A probable case is one without a positive laboratory result, but which is treated like a confirmed case based on its exposure history and clinical symptoms.

JessieB:

 

So yesterday when the headlines say '5 new cases, 1 confirmed, 4 probable' , it could be that the 4 are not new, they are just old cases completely recovered, and discovered by contact-tracing interviewing?

 

The definition of a 'new' case has been discussed a fair bit here. My understanding is that the 'new' count for the day is cases that have been reported to the MoH in the 24 hours to 9am that day (i.e. mostly the previous day). These are obviously not people who have caught the disease that day, nor who have got symptoms, nor who have been tested, nor who have tested positive.

So, in your scenario the 4 are 'new' in the sense that they have not been reported before, but they may have already recovered.

Here's the graph for Tuesday 28th April from Chris McDowall | The Spinoff