Talk:COVID-19 testing

This article was the subject of a Wiki Education Foundation-supported course assignment, between 3 April 2020 and 10 June 2020. Further details are available on the course page. Student editor(s): Elisawulfsberg (article contribs).

Will a country that only tests people admitted to hospitals have a lower "Positive / million people" than a country that tests everybody?

The column “positive/million people” will be higher in a country that only tests people admitted to hospitals, compared to a country that tests all citizens whether or not they are showing symptoms. The population of people being admitted to the hospital contains a higher percentage of people who actually have the disease, compared to the entire population of the country. Those admitted to the hospital are not a random cross section of the population but rather consist of those who have symptoms, and a group of people with symptoms will have a higher percentage of infected people than a group of people, most of whom do not have symptoms. A higher positive / hundred tests, means that there must also be a higher “positive / million people.”

Let’s take Albania. The column “Pos/mill” bears the same ratio to “Positive” as “Tests/mill” bears to “Tests,” which equals roughly 213. Let’s assume that this example represents a country that only tests people admitted to hospitals. Now let’s assume that the same country performed the same number of tests but tested everybody. They would have a lower number testing positive. Let’s assume 305 (about half). The number in the "% (Percentage of positive tests)" column would be roughly 5%, the number of “Tests/mill” would be the same, the number of “Pos/mill” would be roughly 107. So the “Pos/mill” is higher in the “only hospital testing” example. Swood100 (talk) 17:21, 22 April 2020 (UTC)

@Swood100:A country where only the hospitalized are tested will have a higher real pos/million ratio than another country with wide range of tests provided that those two countries have the same ratio on the table. And pos/million ratio is lower on the table in a country with limited tests provided that two countries have the same real ratio.

I think that's the fundamental confusion of the statement. As it state that In countries with similar spread of infection, I will assume that means countries with similar real ratio. Atzhh (talk) 19:32, 22 April 2020 (UTC)

Atzhh: What is the difference between “real ratio” and “ratio on the table”? The two examples I gave both involve Albania, which has the same spread of infection as itself. The only difference is which portion of the population is being sampled, the portion with a higher rate of infection or the portion with the lower rate of infection. Can you give me an example, using Albania for both, in which the sample with the lower rate of infection results in a higher pos/million figure? How would such a thing make sense? Swood100 (talk) 22:33, 22 April 2020 (UTC)

@Swood100: Let's speak in this way. Two countries A and B both have 1M population and 10K infections, so both have real ratio 10k per million. Assuming among 10k, 2k think they should go to hospital. A tested everyone who go to hospital and B tested everyone in the country. Then A could caught 2k positives and have the ratio on the table 2k per million and B could caught 10k positives with the ratio on the table 10k per million. The other way around, if both countries detect 2k positives (so ratio on table 2k per million), since B have tested everyone it's real ratio is again 2k. But A haven't caught anyone asymptomatic, the real infections is higher than 2k and thus real ratio is higher as well.

I read your argument again and realize that what you messed up is really the following: When a country is considering testing everyone, that means they can test everyone who need a test at hospital at first. They are not strategies of same amount of tests. A country could have a wide range of tests only if its capacity surplus its critical needs.Atzhh (talk) 23:34, 22 April 2020 (UTC)

Atzhh: Skip the following and read my response to UnladenSwallow below. Swood100 (talk) 17:53, 23 April 2020 (UTC)

Atzhh: Country B tests a sample of 1000 people at random from the community. B finds that 100 people in that sample are infected. This is 10%, since it is testing a truly random sample. Since the population is 1,000,000, B reports that the pos / million in that country is 100,000 / million.

A tests a sample of 1000 people who go to the hospital. Since A’s sample includes a higher proportion of people who are infected A finds that 300 people in that sample are infected. This is 30%. Since the population is 1,000,000 A reports that the pos / million in that country is 300,000 / million.

The pos / million is higher for the country whose sample has a higher rate of infection: those admitted to hospitals. Swood100 (talk) 14:52, 23 April 2020 (UTC)

@Swood100: Consider two countries with similar spread of infection: Country A with 100 million people that only tests people admitted to hospitals and Country B with 10 million people that tests all citizens whether or not they are showing symptoms. Let's assume the spread of infection is 1%. So there are 1 million infected people in Country A and 100,000 infected people in Country B. Further, let's assume that 75% are asymptomatic, 15% are mild cases, and 10% require hospitalization (the exact rates are not important, it just makes it easier to follow the argument). So there are 750,000 asymptomatic, 150,000 mild cases, and 100,000 hospitalized in Country A. Since Country A only tests those hospitalized, it performs 100,000 tests and gets 100,000 positives, which works out to 1,000 positives / million population. Country B has 75,000 asymptomatic, 15,000 mild cases, and 10,000 hospitalized. It tests all hospitalized (10,000 positives), some mild cases (let's say 1⁄3 for 5,000 positives), and some asymptomatic cases via contact tracing (let's say 1⁄15 for 5,000 positives) for a total of 20,000 positives, which works out to 2,000 positives / million population. As you can see, even though the population of Country A is 10 times that of Country B, and the countries have similar spread of infection, the positives / million population figure for Country B is actually 2 times that for Country A. That's because Country B does not restrict itself to testing only hospitalized cases and thus catches a larger share of infections. — UnladenSwallow (talk) 23:43, 22 April 2020 (UTC)

UnladenSwallow: I created the new section before I saw the existing section. I think I see what you’re saying. The column pos/million only has meaning in relation to its own tests/million and a comparison with any other country’s pos/million is meaningless because of the difference in population sizes. You’re right that for your Country A it comes to 1,000 pos /million but that is misleading since the typical reader will assume this means that the testing sample showed that 0.1% of the population is positive whereas the real number is 100%.

I don’t think that countries that only test people admitted to hospitals always have a lower pos/million. In your example if Country B created a truly random sample of 100,000 tests it would find 1,000 positives since its true rate of infection is 1%. This would give it a test/mill of 10,000 and a pos/mill of 100.

I think that the column pos/mill creates more confusion than clarity. I would remove that column entirely. In the alternative I would try to provide some explanation for what it means but I can’t think of any explanation that either could easily be understood or that justifies that column’s presence in the table. I would replace Tests/mill with “% of population tested.” Swood100 (talk) 17:53, 23 April 2020 (UTC)Swood100 (talk) 19:05, 23 April 2020 (UTC)

Is this your table? Who decides what format it will have? If you agree with my change recommendations I'd be happy to help with the changes. Swood100 (talk) 21:26, 23 April 2020 (UTC)

@Swood100:

1. …that is misleading since the typical reader will assume this means that the testing sample showed that 0.1% of the population is positive whereas the real number is 100% But it does mean that. Let me try to explain this again. Among the population of Country A there are 0.75% asymptomatic cases, 0.15% mild cases, and 0.1% severe (hospitalized) cases. Country B has similar spread of infection, so among the population of Country B there are also 0.75% asymptomatic cases, 0.15% mild cases, and 0.1% severe cases. Since Country A only tests hospitalized cases, there's no way it will register more than 0.1% positive cases (= 1,000 positive / million people). Meanwhile, Country B, depending on how extensive its testing is, can potentially register up to 0.75% + 0.15% + 0.1% = 1% cases (= 10,000 positive / million people). In practice, it won't "catch them all", but it will still register more percentage-wise than Country A.

2. …if Country B created a truly random sample of 100,000 tests it would find 1,000 positives since its true rate of infection is 1%. This would give it a test/mill of 10,000 and a pos/mill of 100. That's right, but real countries are not doing random testing. They test hospitalized cases first, symptomatic cases second, possible contacts of confirmed cases third, high-risk employees (doctors, supermarket employees, etc.) fourth, and anyone who wants tested fifth (to the best of my knowledge, this last stage is only being done by Iceland and South Korea). More importantly, there's no mention of any limits on testing in the statement in question. Countries A and B both test as much as they want, but Country A limits itself to only testing hospitalized cases (cf. 7 below).

3. I think that the column pos/mill creates more confusion than clarity. I would remove that column entirely. The column was requested by multiple editors—that's why I've added it. I think its meaning is pretty clear: it's the number of positive cases per capita (expressed in millionths of country's population).

4. I would replace Tests/mill with “% of population tested.” That would result in many zeros after the decimal point for most countries. When the numbers grow sufficiently, it will make sense to change these columns to "/ 100,000 people", then "/ 10,000 people", etc.

5. Is this your table? Only pages starting with User:EditorsName may be considered "owned" by an editor (with certain limitations). Everything else belongs to everyone. Who decides what format it will have? Such decisions are supposed to be made collectively as a result of discussions on the tables' talk pages Template talk:COVID-19 testing by country and Template talk:COVID-19 testing by country subdivision. Since there were almost no discussions, I had to make those decisions myself based on suggestions of other editors.

6. When I first saw 77.59.125.228's comment in § Testing policy influence, I had a reaction similar to yours: wait, what? It took me some time to think it through. The statement is certainly not obvious.

7. If the amount of tests per capita is fixed and very limited (say, 100,000 for Country A and 10,000 for Country B in my example above), then your interpretation is actually the right one: Country A, by spending its test "budget" only on hospitals, will get a higher "Positive / million people" figure than Country B. But the statement in question doesn't say that the amount of tests is fixed.

8. Perhaps we can re-write the statement like this:

The figures below are influenced by a country's testing availability and policy. Among countries that have the same spread of infection: Countries that experience test shortages and therefore only test hospitalized cases, will have higher % (Percentage of positive tests) and lower Positive / million people figures, because they won't be registering mild and asymptomatic cases. Countries that have unlimited tests, but only test people with symptoms and those who were in contact with confirmed cases, will have mid-range % (Percentage of positive tests) and Positive / million people figures. Countries that have unlimited tests and test anyone who wants tested, will have lower % (Percentage of positive tests) and higher Positive / million people figures, because they will be registering the most asymptomatic cases.

— UnladenSwallow (talk) 14:47, 24 April 2020 (UTC)

UnladenSwallow: Take two countries that are like Country A in every respect except for whom they test. A1 only tests hospital admissions, A2 conducts its tests without regard to the symptoms of the person tested (and therefore conducts more random testing). At the beginning of testing, and as long as they perform the same number of tests A1 will have a greater Positive/million people. At some point A1 will run into a testing wall since it is limited to testing those who are symptomatic and who show up at the hospital. A2 will not have that wall and will go on testing and it will have a greater number of positives to find since it is not limited to those who are symptomatic. When A2 starts conducting a greater number of tests we have introduced a second difference between the countries, one known to result in a higher pos/mill. Eventually the number of positives A2 finds will exceed the number found by A1 and at that point its pos/mill will become larger.

The original intent of the explanatory text was simply to deal with the Percentage of positive tests issue, which will always favor A1. To try to incorporate the pos/mill issue into this would require that we discuss not just who is tested but how many tests are performed. We would have to explain the above crossover point or at least we couldn’t say that A1 always has a lower pos/mill since until it falls behind in number of tests performed that won’t be true. How about the following, which splits it up into two sentences, each of which only deals with one concept:

If two countries are alike in every respect, including having the same spread of infection, the country that only tests people admitted to hospitals will have a higher figure for “% (Percentage of positive tests)” than a country that tests all citizens, whether or not they are showing symptoms.

If two countries are alike in every respect, including which people they test, the one that tests more people will have a higher “Positive / million people.” Swood100 (talk) 14:02, 27 April 2020 (UTC)

@UnladenSwallow: Did you see my suggestion above? Swood100 (talk) 20:54, 28 April 2020 (UTC)

Semi-protected edit request on 23 April 2020

In the section "Approaches to testing" paragraph 5 please add the sentence: An open source software "Laboratory Optimizer for Mass Testing" provides the planning, reporting, and data analytics capabilities for samples pooling, implementing binary search and other group testing algorithms.

 Not done: Please read our external links guidelines. You also did not link to any reliable sources. Sorry, I cannot fulfill this request. Aasim 17:23, 24 April 2020 (UTC)

@Awesome Aasim: Thanks, I've fixed the link format and changed it to an external reliable source Kirill.vechera (talk) 11:31, 28 April 2020 (UTC)

Not Done. This is advertising. Graham Beards (talk) 12:03, 28 April 2020 (UTC)

@Graham Beards: It's no more advertising that links to researches on pool testing, but instead of abstract ideas it's implemented idea, a ready to use non-commercial software. The same kind of solution as Origami Assays published a paragraph below. If you consider the word 'marketplace' in the domain name, it's a strange passion of all EU bureaucrats name every kind of a systematic list of every sort as a marketplace. Kirill.vechera (talk) 16:51, 28 April 2020 (UTC)

Procedural close this does not look like a reliable source. Please start a new discussion and discuss the edit first before using the {{editprotected}} template. Aasim 17:05, 28 April 2020 (UTC)

Semi-protected edit request on 23 April 2020

Change reference 220, " "US Historical Data". The COVID Tracking Project. 22 April 2020." " to " "US Historical Data". The COVID Tracking Project. 22 April 2020." " in order to link to the currently-orphaned wiki page for the COVID Track project at this address: https://en.wikipedia.org/wiki/COVID_Tracking_Project BlabberBobsBigBootleggers (talk) 20:25, 23 April 2020 (UTC)

• Is this WP:RELIABLE? I was expecting a .gov domain (like the vast majority of the countries) and not a .com. *edit*: also 2020 coronavirus pandemic in the United States uses CDC and JHU numbers, so maybe this article should follow the same sources? Feelthhis (talk) 19:23, 25 April 2020 (UTC)

 Done In future, please post edit requests to embedded templates on the template page itself. Whether or not the source is reliable can also be discussed on the template talk page. Darylgolden^(talk) Ping when replying 08:26, 27 April 2020 (UTC)

Russia first testing date

Please, add first Russia testing date (its missing from article). Source from other article: On 24 January, the first testing systems were developed and deployed to laboratories around the country. --109.169.202.201 (talk) 10:12, 24 April 2020 (UTC)

References

• Done ASDFS (talk) 19:38, 1 May 2020 (UTC)

Be accurate about Accuracy

Accuracy is an important topic which is worth serious discussion. However, it is currently devoted to rambling about the inaccuracy of the tests from China. This is not only odd but also harmful for people who care about accuracy (I suppose all do).

So we should probably rename the section to "Accuracy of tests from China". But even then, it neglects to mention that most tests are supplied from China, and they can be quite accurate too. For example, in citation [109], it has the following paragraph, which is conveniently left out

"... Their use was immediately suspended and new tests sourced from a different Chinese supplier. They arrived last week and had an accuracy rate of about 90%, according to the Turkish official."

Also, according to the following source, the anti-body tests used in the broadly known Stanford study are from China too.

https://www.buzzfeednews.com/article/stephaniemlee/stanford-coronavirus-study-bhattacharya-email

Since I am not an "established" author, someone please revise it. — Preceding unsigned comment added by 2601:647:5600:33c2:bc:c2db:a083:f4f (talk) 23:53, 24 April 2020 (UTC)

The problem with first CDC test and other tests

I’ll summarize what I’ve read, the CDC developed a test that look for three genetic markers. The original WHO test was only two. Third marker produced unreliable results during validation and a positive interpretation required all three so the CDC asked state laboratories not to use them. Eventually some of these test kits where are used with third markers results Ignored.

Also where’s the information on antibody testing? LabCorp is now today offering the IGG Elisa test to the general public. I believe it has 99.5% specificity meaning 1 in 200 chance of being tripped up by antibodies to another coronavirus and good sensitivity if used 2-3 weeks after exposure. Technophant (talk) 09:16, 27 April 2020 (UTC)

Please read and use this guy’s articles. Consider him to be a reliable secondary source. Read the linked articles about the problems with false negatives, trust your symptoms. I’m recovering from this myself so I’m trying to get the antibody testing after having two false negative swabs, quick tests and rapid pcr. He says that this is a feature of the disease progression, not test methods. Mine presented more like strep throat initially and never had a cough. Claims China cumulatively had a staggering 30% false negative rate which challenges many assumptions about the data. Technophant (talk) 09:49, 27 April 2020 (UTC)

Cyprus

Details for Cyprus' testing numbers can be found in this University of Cyprus' website: https://covid19.ucy.ac.cy/ RegGeo (talk) 17:19, 27 April 2020 (UTC)

Available tests / PCR based

Available tests PCR based When scientists from China first released information on the COVID‑19 viral genome on 11 January 2020, the Malaysian Institute for Medical Research (IMR) successfully produced the “primers and probes” specific to SARS-CoV-2 on the very same day. The IMR's laboratory in Kuala Lumpur had initiated early preparedness by setting up reagents to detect coronavirus using the rt-PCR method.[94] The WHO reagent sequence (primers and probes) released several days later was very similar to that produced in the IMR's laboratory, which was used to diagnose Malaysia's first COVID‑19 patient on 24 January 2020.[95]

Public Health England developed a test by 10 January,[96] using real-time RT-PCR (RdRp gene) assay based on oral swabs.[97] The test detected the presence of any type of coronavirus including specifically identifying SARS-CoV-2. It was rolled out to twelve laboratories across the United Kingdom on 10 February.[98] Another early PCR test was developed by Charité in Berlin, working with academic collaborators in Europe and Hong Kong, and published on 23 January. It used rtRT-PCR, and formed the basis of 250,000 kits for distribution by the World Health Organization (WHO).[99] The South Korean company Kogenebiotech developed a clinical grade, PCR-based SARS-CoV-2 detection kit (PowerChek Coronavirus) on 28 January 2020.[100][101] It looks for the "E" gene shared by all beta coronaviruses, and the RdRp gene specific to SARS-CoV-2.[102]

In China, BGI Group was one of the first companies to receive emergency use approval from China's National Medical Products Administration for a PCR-based SARS-CoV-2 detection kit.[103]

In the United States, the CDC distributed its SARS-CoV-2 Real Time PCR Diagnostic Panel to public health labs through the International Reagent Resource.[104] One of three genetic tests in older versions of the test kits caused inconclusive results due to faulty reagents, and a bottleneck of testing at the CDC in Atlanta; this resulted in an average of fewer than 100 samples a day being successfully processed throughout the whole of February 2020. Tests using two components were not determined to be reliable until 28 February 2020, and it was not until then that state and local laboratories were permitted to begin testing.[105] The test was approved by the FDA under an EUA.[citation needed]

US commercial labs began testing in early March 2020. As of 5 March 2020 LabCorp announced nationwide availability of COVID‑19 testing based on RT-PCR.[106] Quest Diagnostics similarly made nationwide COVID‑19 testing available as of 9 March 2020.[107]

In Russia, the COVID‑19 test was developed and produced by the State Research Center of Virology and Biotechnology VECTOR. On 11 February 2020 the test was registered by the Federal Service for Surveillance in Healthcare.[108]

On 12 March 2020, Mayo Clinic was reported to have developed a test to detect COVID‑19 infection.[109]

On 19 March 2020, the FDA issued EUA to Abbott Laboratories[110] for a test on Abbott's m2000 system; the FDA had previously issued similar authorization to Hologic,[111] LabCorp,[112] and Thermo Fisher Scientific.[113][114] On 21 March 2020, Cepheid similarly received an EUA from the FDA for a test that takes about 45 minutes on its GeneXpert system; the same system that runs the GeneXpert MTB/RIF.[115][116]

On 13 April, Health Canada approved a test from Spartan Bioscience. Institutions may "test patients" with a handheld DNA analyzer "and receive results without having to send samples away to a [central] lab."[117][118]

On April 16, 2020 | The Australian Therapeutic Goods Administration(TGA) entered Kogene Biotech in the Australian Register of Therapeutic Goods(ARTG), as issued to World Systems Corporation the sponsor, for the approval to supply Severe Acute Respiratory Syndrome(SARS) associated coronavirus IVDs, under certificate number DV-2020-MC-05067-1 for the PowerChek™ 2019-nCoV Real-time PCR Kit which provides fast and accurate detection testing for the COVID-19 2019-nCOV coronavirus, specifically targeting the E gene for beta Coronavirus and the RdRp gene for 2019-nCoV in bronchoalveolar lavage fluid, sputum,nasopharyngeal swab and oropharyngeal swab. Patricklewiki (talk) 01:24, 1 May 2020 (UTC)