Has the World Health Organisation (WHO) or any government anywhere conducted random tests for Covid-19?

It is announced today that in Britain they are going to step up the testing from 1,500 people per day to 10,000. Will any of these be from a random cohort?

It would seem to me that one random batch of say 5,000 people would tell us a lot about the extent of infection in the community, and hence the relative dangers of getting the disease.

If the results of an election can be accurately predicted from a random opinion poll survey of 1000 people, then it would seem that we could learn a lot about the virus from random testing - in different parts of the country - and something well worth doing!


The small Italian town of Vò has tested and retested its entire population.

Through testing and retesting of all 3,300 inhabitants of the town of Vò, near Venice, regardless of whether they were exhibiting symptoms, and rigorous quarantining of their contacts once infection was confirmed, health authorities have been able to completely stop the spread of the illness there.


The first testing round, carried out on the town’s entire population in late February, found 3 per cent of the population infected, though half of the carriers had no symptoms. After isolating all those infected, the second testing round about 10 days later showed the infection rate had dropped to 0.3 per cent.

"Aggressive testing helps Italian town cut new coronavirus cases to zero", Financial Times, 2020-03-17


The Dutch have begun a project to test donated blood for the antibodies to SARS-COV-2. (sources: Reuters, NLtimes)

The motivation for this project seems to be similar to that in your discussion. The blood antibody test is different than the swab tests, however. The swab PCR tests seem to be designed for patients who have a significant probability of having the COVID-19 infection.

Regarding your point about the utility of testing in different areas, Tom Frieden wrote recently that

"testing is absolutely, crucially important ...in areas with few or no cases, to inform containment and isolation strategies and facilitate contact tracing. Seattle would likely have had a much smaller outbreak if testing there had been widely available sooner."

  • The term you're looking for is probably (public health) surveillance. ;) – JJ for Transparency and Monica Mar 18 at 20:48
  • I don't see how this is answering the question. This is not a debate forum to post "answers" like "you're right, we need testing" to a question that asks "is there testing going on?". – Fizz Mar 19 at 8:08
  • @Fizz - Got it. Edited with an actual answer now. – Matt Giambusso Mar 19 at 13:51
  • 1
    This is the golden test that everyone was waiting for. The results will finally show what the true infection rate is, including patients who were asymptomatic. – JonathanReez Mar 20 at 6:24

I am unable to find any instances of random samples for Covid-19 tests being drawn from the general population (at the time of writing, see other answers for more recent developments). Current random test samples seem to have been drawn from populations most likely to be infected with the virus, such as healthcare professionals or those who have recently returned from communties most at risk, or those already presenting with influenza-like symptoms.

The ramping-up of tests in the UK that the question refers to relates to the overall testing capacity of the NHS. According to the NHS website:

Tests for coronavirus are only done if there's a high chance you could have the illness.

This may be because:

  • you have been in close contact with someone with confirmed coronavirus
  • in the last 14 days you have travelled to a country or area with a high coronavirus risk – see our coronavirus advice for travellers

However, on 26th February, Public Health England announced that it would conduct random tests in a selection of GP practices and NHS hosptials on patients who are suffering from "severe respiratory infections but who do not display Covid-19 symptoms".

The Department of Health in Hawaii yesterday announced that they too would be conducted random tests for the virus, however these tests are also to be conducted on "Samples collected for influenza testing from patients with respiratory symptoms", and not on samples collected from the general population.

In the Netherlands, random tests were conducted on hospital staff in Brabant, resulting in 4% being identified as having contracted the virus. Again, though, I cannot find any evidence on tests having been conducted on a general sample of the population as a whole.

In New York, in a briefing delivered on Monday, Governor Andrew M. Cuomo gave an insight into the current testing protocol:

The testing - we'll go through the latest numbers, but let's also remember the context for testing if we can: The more you test, the more positives you will find, and you are testing primarily a suspect group because we're testing people who we believe came in contact with a positive person. We want that data because we want to find out who's positive so we can isolate them and reduce the spread. But it is not a random sample, it is not statistically representative of anything. It's testing a particular universe that we believe may very well have been exposed to a positive person. So it's not statistically, I don't know what it means, I take it as good news because I want to be finding the positives so we can isolate them and we can reduce the spread. And that's what the testing is all about.

It seems, therefore, that given the rapid spread of the disease and ongoing battle for health services to simultaneously cope with the current situation whilst also preparing for the future spread of the virus, testing capabilities are being prioritised for patients most at risk of having contracted the illness, in order to provide treatment and quarantine those who test positive. While statistical analysis of a random sample of the population might be interesting, it is not particularly helpful for health services, most of which acknowledge that community spread is inevitable.

  • This is a creditable answer which confirms what I largely thought was the case. But it does seem to me that the small effort of doing some random surveys would tell us a lot that we do not know at the moment. Especially since the President of the US, has alleged that Covid-19 is probably no more dangerous than seasonal flu. I think we desperately need to know if that is the case, and more urgently that it should be comprehensively refuted if it is not. – WS2 Mar 11 at 14:49
  • 3
    @WS2 I think the crux of it comes down to urgency. We know that testing is resource-intensive if the whole of the NHS can only perform 10,000 tests a day across the UK. While a completely random sample might give heath services some idea of the current level of community spread, it would be at the expense of current patients that are at risk of or suspected of having the virus who may have to wait longer to be tested. As the situation is evolving day by day, the results of the test would also become outdated very quickly. – CDJB Mar 11 at 14:55
  • @WS2 - right now our tests are only beginning to become reasonably accurate and these tests are in limited by supply and processing capacity. If these limitations weren't there: meaning that we had an ample supply of tests and had spare processing capacity then 100% we ought to be doing random tests of the population. – Mayo Mar 11 at 16:28
  • 1
    @CDJB The people who have so far died, may well have similarly died of seasonal flu, had they got it. How can we know if we remain ignorant of how many are infected? Italy has virtually closed down - no shops open anywhere - other than food shops. Britain could follow within days. It seems imperative to me that we better understand what is going on. – WS2 Mar 12 at 8:04

The Dutch National Institute for Public Health and the Environment held a small random sample test among hospital staff in the Dutch province of Noord-Brabant. Outcome indicated that about 4% of staff was infected with COVID-19.

From 6 to 8 March 1097 hospital workers were tested of which 3.9 per cent was indeed infected with COVID-19. The percentages vary per hospital and are between 0 and 10 per cent. Two hundred patients were also tested. Of these patients, an average of 9 per cent was infected with the virus. Again, there are differences between hospitals. The people who were tested already had symptoms.


This is an important question to discuss. From what I understand, the shortage of tests and the need to devote them to diagnosis of persons who have symptoms, or to test for those have who have been exposed but who may be asymptomatic, may make it hard to do.

But given the advent of exact test logistic regression software that can produce odds ratios generalizable to the larger population, with rather low survey sizes (widely used in epidemiology and medicine studies), it seems to me that random tests are valuable even with small sample sizes. I would imagine participation would be very high with any kind of cash incentive or perhaps a novel incentive: a test of your spouse or other family member as a bonus.

The widest use seems to be in India: [ICMR: 500 random tests for Covid-19 come up negative (yes, an expenditure of resources, but reassuring to see )][1] Lots of reports on this work in India if you google it.

  • 4
    Welcome, Please take a look at our tour if you haven't done so already. Note that we are looking for "answers" and not "discussion". You could expand the section on the testing in India (don't just say "google it") and trim some of the more opinion based paragraphs on the benefits of testing, since they don't answer the question. – James K Mar 21 at 17:31

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .