Insight

Today I’m going to talk a bit about why random testing is important, and why in my view, we need to have the ability to do random testing in order to relax the lockdown. I now know of three countries where they are doing testing that is closer to random than Australia. Iceland, which I talked about in my last post, and New Zealand (and Slovenia, late breaking news pointed out by Martin). New Zealand has started testing at supermarkets in a few hotspots – Christchurch, Queenstown, Auckland and the Waikato. So far all their tests are negative. Which is good but not conclusive that there aren’t cases out there that haven’t been tested.

This NZ statistics article (found via my statistician cousin) puts it very well:

One of the known unknowns about the NZ coronavirus epidemic is the number of cases we have not detected.  There will have been a mixture of people who didn’t get any symptoms, people who are going to show symptoms but haven’t yet, people who got moderately sick but didn’t get tested, and people whose deaths were attributed to some pre-existing condition without testing.

For the decision to loosen restrictions, we care mostly about people who are currently infected, who aren’t (currently) sick enough to get testing, and who aren’t known contacts of previous cases.  What can we say about this number — the ‘community prevalence’ of undetected coronavirus infection in New Zealand?

While we continue to have new cases test positive, there are likely to be other cases in the community that haven’t yet tested positive. And those cases can still spread the disease to others. While we are locked down, contacts out there are not zero – even as the non designated shopper in my community, I’m still going out for exercise every day.

I’ve linked before to several pieces of research which try to work out how many asymptomatic carriers there are. There is more information about the Italian town of Vo in this study which I have just come across (from the filename it was published in the past few days):

On the first survey, which was conducted around the time the town lockdown started, we found a prevalence of infection of 2.6% (95% confidence interval (CI) 2.1- 3.3%). On the second survey, which was conducted at the end of the lockdown, we found a prevalence of 1.2% (95% CI 0.8-1.8%). Notably, 43.2% (95% CI 32.2-54.7%) of the confirmed SARSCoV-2 infections detected across the two surveys were asymptomatic….

… The presence of a significant number of asymptomatic SARS-CoV-2 infections raises questions about their ability to transmit the virus. To address this issue, we conducted an extensive contact tracing analysis of the 8 new infections identified in the second survey (Table 3). Three of the new infections reported the presence of mild symptoms and did not require hospitalization. For Subject 1 we could not identify the source of infection. Subject 2 had contacts with four infected relatives who did not have any symptoms at the time of contact. Subject 3 reported contacts with two infected symptomatic individuals before the lockdown. Five of the 8 new infections showed no symptoms; Subjects 4 and 6 shared the same flat with symptomatic infected relatives. Subject 5 reported meeting an asymptomatic infected individual before the lockdown; Subject 7 did not report any contact with positive individuals and Subject 8 shared the same flat with two asymptomatic relatives.

From this, it seems fairly clear that transmission of the virus can occur from asymptomatic individual to asymptomatic individual. And then eventually, to someone who becomes sick with the disease. Nearly half of the infected people in the town of Vo were asymptomatic when they tested positive. So if we are only testing people with symptoms, we would expect that there are at least double the cases than are being reported right now. And those non symptomatic cases are probably a lot less careful about their social distancing. so they could keep transmission of the virus going.

As the New Zealand statistician points out, given only 1% of tests of symptomatic people there are positive, so the maximum infection in the wider community is only 1% and probably lower, you have to test a lot of people without a positive test to be comfortable you have found all the positives there are.

The question gets more extreme with smaller sample sizes: if we sample 350 people (as was done at the Queenstown PakNSave) and find no cases, what can we say about the prevalence?  The classical answer, a valuable trick for hallway statistical consulting, is that if the true rate is 3/N or higher, the chance of seeing no cases in N tests is less than 5%. So, if we see no cases in 350 people, we can be pretty sure the prevalence was less than 3/350, or about 1%.  Since we were already pretty sure the prevalence was way less than 1%, that hasn’t got us much further forward.  We’re eventually going to want thousands, or tens of thousands, of tests. The Queenstown testing was only a start.

So in my view, as in New Zealand, a key part of lifting a lockdown in any location in Australia would be to have a strong and continuous program of some form of random testing. Given the ACT has had one positive test over the past 7 days (unfortunately it was today), maybe they are the candidate for the first part of Australia to cautiously lift lockdown to some degree with a strong program of randomised and targeted testing.

Link

Today’s local(ish) lockdown news is that New Zealand is moving from Level 4 lockdown to Level 3 at midnight next Monday (27 April).

“In the coming weeks we will continue to focus on testing people with symptoms suggestive of Covid-19, and hunting out any undetected cases that might exist. We will continue to undertake community and other sentinel testing as part of our ongoing surveillance.”

Testing on asymptomatic workers at places where recent cases had been recorded would be scaled up, he said.

Level 3 in New Zealand is similar (but not the same) as the rules we have in Australia (they still can’t get their hair cut, but the schools will be open in a similar way to ours – only for people who need to go). New Zealand only had 9 new confirmed or probable cases today, which is promising, so hopefully their numbers will continue to go down.

Life glimpses

Yesterday we had a birthday party for a friend on zoom – the vanilla cake looked delicious but unfortunately was not in our zoom location so we stuck with the packet of chips we had opened in celebration. The singing worked better than I expected (not well) given that zoom effectively only allows one person to speak at a time. The children in the various households found our attempts to use all of zoom’s features (choose your own background… raise your hand to speak etc) quite hilarious. But we had a good time.

Bit of Beauty

Today’s bit of beauty is musical for a change. Many of my readers know I’m a member of a community choir, and I’m really missing the weekly chance to sing together. But we have a whatsapp group sharing musical beauty from around the world, and today’s bit of beauty is too good not to share. The London City Voices singing You’ve got a Friend, campaigning for Women’s Aid (a UK helpline) during the time of lockdown. As the London City Voices say, sadly coercive and controlling behaviour has escalated during the COVID-19 pandemic and women and children are becoming increasingly isolated.

17 Comments

  1. Thanks Jennifer
    New York Governor Cuomo has just announced random antibody sample testing in NY to gain a better understanding of the level of community infection. He correctly sees it as a critical part of the re-opening process.

  2. There is no particular benifit knowing whoo is a asymptomatic carrier or how many there are. Firstly, test onky gives a result to the point of the test. The very lack of social distancing in getting a test merely increases the risk of infection. We cant know if the person is positive after the test, only that they were negative before hand
    Secondly, for reason given above test dont negate the need for social distancing and personal hygiene practices. We must assume anyone is a carrier. Testing might agenda complacency too.

    Finally, testing is a scarce medical resource particularly the health workers carrying it out. There are more important things they can be doing than risk their own health (depleting scarce human resources)
    Australia current practive of screening often remotetly of ill people: a doctor can readily tell if an illness is not CONVID19, then testing only those who may be ill to infirm best treatment. We also test health workers so they can continue at work if experining a minor health concern.

    Not all data is useful. Dont give importance to what is measured; measure what is important.

    Meten is Weten!

    1. There is no doubt that when tests were scarce they needed to be prioritised to those with symptoms and/or high risk. I am not sure that tests are scarce in Australia now. I also agree that hygiene and physical distancing must be maintained, irrespective of the results of random testing. However, there is material value in a confident estimate of the extent of currently unknown infection. There is plenty of speculation and debate about the risk of children infecting other children (and their families), as well as their teachers, if schools are open. An understanding of the likelihood would help us assess this risk, as well as the risk of other proposed easing of rules.

      2. Hi Martin, I think the issue with this study is that the blood tests were not random – all the high school ones were from one high school, and all the primary school ones from one primary. Not saying the results are under- or over-stated, just not a “representative sample” as quoted in the SMH. Cheers, Karen

      3. Thanks, Karen, I agree that would obviously represent a significant flaw in the research. I have re-read the report and it is not obvious to me how you reached that conclusion. Would you mind pointing out the reference to blood tests being confined to only two schools? Thank you

      4. On page 2, under the heading “High Schools” – “In one high school, of the 75 close contacts who underwent blood testing at approximately 1 month after contact with the initial cases while infectious, only 1 student had antibodies detected, indicating infection had occurred.” And on the same page under the heading “Primary Schools” – “Only one secondary case (nose/throat swab positive) was identified in the 168 close contacts. In the same primary school that had this secondary case, 21 close contacts underwent blood testing. The same student whose nose/throat swab tested positive also had antibodies detected through serology testing, consistent with their known recent infection.” This section on primary schools also indicates that at least some of the serology testing of individuals overlapped with swab testing, thus further casting doubt on the level of transmission discussed in the report.

      5. Author

        Thanks Martin and Karen, I’ve revisited this topic today, I’ll be interested to see what you both think.

  3. You have got a friend was like a singing mosaic. Thank you.. love Marta

    1. An Italian friend noted to me they are rolling out a program of random testing to what looked to me (google translate permitting) 150,000 people for this very reason

    2. Carole King’s song was in part a response to her friend James Taylor who was struggling with mental health issues at the time (‘I’ve seen lonely times when I could not find a friend’). Good song to think of in these Covid-19 times.
      Am so impressed with the thinking in this blog and all the great responses!

  4. Hi Jennifer, one of your conclusions states “So if we are only testing people with symptoms, we would expect that there are at least double the cases than are being reported right now.”. This doesnt allow for two factors in the Australian experience 1) our cases are predominantly the result of imported overseas travellers, they are not the result of organic growth within the community 2) quarantine of overseas travellers since 29 March. Both of these factors should mean that asymptomatic cases should be less than the number being reported right now. These two factors are forgotten in a lot of the comparisons currently being undertaken between Australia and overseas. Cheers, Karen

    1. Author

      Hi Karen, in one sense, I agree that our cases are not part of organic growth, but it seems to me that if 50% of cases are asymptomatic, any measure of positive cases must be missing half of them – whether they came from overseas, or whether they came into the community. Many of the asymptomatic cases (as with the symptomatic ones) would have “recovered”by now – so wouldn’t be active spreaders, but I think the active “cases” must be at least double what is being reported. And that assumes that all of the mild symptomatic people were tested, which is more likely here than in most countries, but not certain.

      1. Thanks Jennifer. I guess the question of organic vs non-organic growth comes down to the question “If we import one symptomatic case from overseas, did we also import their corresponding asymptomatic case?” I’m not sure of the answer to this. Regarding quarantine, in NSW last week there were about 100 diagnosed cases, of which around 75 were from local transmission and 25 from overseas travellers (in quarantine). So doesn’t that mean the estimate of asymptomatic cases in the community that week should only be around 75 rather than 100?

  5. If “you have to test a lot of people without a positive test” I’m not sure that supporting “a strong and continuous program of some form of random testing” necessarily follows.

    It’s hard to decide though, without knowing how many tests we’re actually talking about. FWIW My back of the envelope calculations went: there are 2,500 ICU beds in Au. That’s .001% of the population. Assuming it takes — what a month? — to flatten the curve through lock-downs and the virus doubles it’s spread in normalish times weekly you’d want to know when the infected population got to 625 people, or .00025% of the population to give you time to react. To find out if .00025% of the population had it through random testing and using your rule of thumb of N/3, you’d have to randomly test about 12,000 people a week? That’s more than half a million tests a year. And you thought telemarketers were annoying. Is that kind of number viable? I mean, genuine question, is it?

    I’m sure there’s lots of things I’ve neglected — like half the cases being symptomatic, and a minority of the infections needing an ICU, and some of my assumptions are pretty ropey. You can probably come up with a better number than me for how many random tests you’d have to actually do to make it worthwhile. I’d be really interested.

    But I do feel that there are four plausible ways out of lock-down – herd immunity through vaccination (a year away at best, may not be possible), herd immunity through infection (may not even be a thing, and how many deaths are you happy to wear?), a reliable treatment (hydroxychloroquine anybody?), …and I must admit it intrigued me when the penny finally dropped; the non-medical route, which may involve random testing, but definitely does involve general societal lockdown until there are a manageable number of cases, then afterwards aggressively isolating everyone who comes in contact with a new infection. To make it an effective approach, however, you’d have to (a) keep the borders closed (or limit movements to similarly virus-free places) (b) put a ton of resources into finding those contacts of the infected. I’m guessing people suffering from Covid-19 don’t have super reliable memories about where they went and who they saw a week and a half ago? — but you know you know, maybe you could bolster the process with some technological help. Maybe… an app…

    1. Author

      Very good point Peter – even if we think theoretically that random testing is a good idea, it is important to understand what level of testing we think is actually feasible. My own view is that the way out of this for us and NZ is to keep the cases to zero, or such a low level that we can aggressively jump on the ones that appear, and contract trace and isolate to stop them getting out again. But unless the disease is completely eliminated, some level of social distancing has to keep the R0 below 1 to make that work.

      1. That seems to be the most viable short-term exit strategy. Exports and imports can presumably largely go on as before (if we wish) but inbound and outbound travel, largely gone. That clearly won’t be great for the economy, but the least bad option I suppose. It’ll be interesting to see how bad exactly.

        It’ll also be interesting to see who else we let in to, (or wants to join) our virus free bubble. A bunch of other islands perhaps?

        (As for whether 12,000 tests a week is viable. ABSOLUTELY apparently. You get a test, you get a test. EVERBODY gets a test according to the paper.)

Comments are closed.