Our daily updated research and data on the coronavirus pandemic.
Read more

Data on COVID-19 testing

This is the section on testing which is part of our main publication on the pandemic here: Coronavirus Disease (COVID-19).

Why data on testing is needed

Without data on COVID-19 we cannot possibly understand how the pandemic is progressing.

Without data we cannot respond appropriately to the threat; neither as individuals nor as a society. Nor can we learn where countermeasures against the pandemic are working. 

The number of confirmed cases is what informs us about the development of the pandemic.

But the confirmation of a case is based on a test. The World Health Organization defines a confirmed case as “a person with laboratory confirmation of COVID-19 infection”.1

Reliable data on testing is therefore necessary to assess the reliability of the data that informs us about the spread of the pandemic: the data on cases and deaths.

Some countries provide clear and helpful data on testing

Some countries present comprehensive, detailed and regularly updated data. Iceland (here) is one of these countries. Estonia (here) goes even further, showing breakdowns by age, gender and region.

For many countries however, available data on testing is either incomplete or else completely unavailable. This makes it impossible for their citizens and for researchers to assess the extent and significance of their testing efforts. 

Our current knowledge of COVID-19 testing – and more importantly of the pandemic itself – would be greatly improved if all countries were able to report all the testing data available to them in the way shown by the best examples. 

We need to understand what the published numbers on testing mean 

Those countries that do publish testing data often do not provide the required documentation to make it clear what the provided numbers precisely mean, and this is crucial for meaningful comparisons between countries and over time.

The key questions that any data description on testing data should answer are given in the following checklist. Clear answers to these questions are what is needed to properly interpret and compare published numbers.

For citizens to trust and understand the published data and for countries to learn from each other, it is crucial that every country provides the data on testing in a clearly documented way. We hope this checklist offers helpful guidance.

Our checklist for COVID-19 testing data

1) Is there any data?

Many countries are not yet providing official figures. Others do not do so on a regular basis. The first question to ask, then, is if there is any testing data for a given country.

Equally important is to make the available data findable. Currently, the available data is often not easy to find, because some countries are releasing figures at unpredictable intervals in ad-hoc locations (including social media or press conferences).

2) Do numbers refer to ‘performed tests’ or ‘individuals tested’?

The number of tests performed is different to the number of individuals tested. The reason for this is that it is common for COVID-19 testing that the same person is tested more than once.

Some countries report tests performed, while others report the number of individuals tested.

The source description should state clearly what is counted.

3) Are negative results included? Are pending results included?

It needs to be clear whether or not figures for the total number of tests performed, or the number of people tested, include negative test results, as well as the number of tests that are pending results.

Many sources report the number of individuals who are ‘suspected’ or have been ‘ruled out’. To be reliably included in test counts, it needs to be explicit whether such categories reflect the number of people who are awaiting test results or have tested negatively.

4) Do the figures include all tests conducted in the country, or only some? 

Figures reported by countries may only be partial if not all laboratories are reporting to the central authority.

The scope of testing data should be made explicit by the source. For instance, the US CDC make it clear that their figures do not include tests conducted in private labs.

5) Are all regions and laboratories within a country submitting data on the same basis?

Answers to the questions above may vary from region to region. In order to assess the reliability of aggregate testing data, it needs to be clear if heterogenous data is being summed together.

The US COVID Tracking Project, for instance makes it clear that their US totals combine data for tests performed and individuals tested, depending on which is reported by individual states.

6) What period do the published figures refer to?

Cumulative counts of the total number of tests should make clear the date from which the count begins. The key question that needs to be answered is whether the figures published at some date (attempt to) include all tests conducted up to that date.

Because the reporting of tests can take several days, for some countries figures for the last few days may not yet be complete. It needs to be made clear by the source if this may be the case. The US CDC, for instance, makes this clear.

7) Are there any issues that affect the comparability of the data over time?

If we want to look at how testing figures are changing over time, we need to know how any of the factors discussed above may have changed too.

The Netherlands, for instance, makes it clear that not all labs were included in national estimates from the start. As new labs get included, their past cumulative total gets added to the day they begin reporting, creating spikes in the time series.

8) What are the typical testing practices in the country?

Having a sense of how often and when individuals are tested, can help the users of these statistics understand how estimates of tests performed and individuals tested might relate to each other.

For instance, how many tests does a case investigation require? What are the eligibility criteria to be tested? Are health workers, or other specific groups, being routinely retested?

9) Might any of the information above be lost in translation?

People accessing data published in a language in which they are not fluent may misinterpret the data by mistranslating the provided text, which often includes technical terms.

Many countries report testing data in multiple languages – this helps disseminate the information to a broader audience, whilst helping prevent misinterpretations.

Our database on COVID-19 testing data

Our goal at Our World in Data is to provide testing data over time for many countries around the world.

We have started with this effort and will expand it in the coming days.

Our aim is to provide alongside the data a good understanding of the definitions used and any important limitations they might have. The checklist above is what guides our efforts.

At the end of this document you find descriptions of the data for each country. But in many cases sources do not yet provide the detailed descriptions of the data we would like. All the details we have been able to find so far are provided below.

We will be adding to the list of countries shown in the coming days.

The total number of tests performed or people tested so far

The two charts shown here show the total number of tests, or people tested, as indicated in the legend for each series.

The first chart shows the absolute number and the second shows it per thousand people of the country’s population.

For comparisons across the series it is important to understand the definitions of the different measures. These are provided in the country by country notes below.

Download the data: we make our full testing dataset, alongside detailed source descriptions, available on Github.

Tests per day

The two charts shown here show the daily number of tests, or people tested, in absolute terms and per thousand people respectively.

For comparisons across the series it is important to understand the definitions of the different measures. These are provided in the country by country notes below.

Download the data: we make our full testing dataset, alongside detailed source descriptions, available on Github.

Source information country by country

Below is the list of the 40 countries for which we have data. We will be adding to this list in the coming days.
You can download the full dataset along side the detailed source descriptions can be downloaded on Github.
Or else click to jump to the detailed source description and the latest data for that country:

Australia
Austria
Bahrain
Belgium
Canada
Costa Rica
Czech Republic
Denmark
Ecuador
Estonia
Finland
France
Germany
Greece
Iceland
India
Indonesia
Ireland
Italy
Japan
Lithuania
Malaysia
Netherlands
New Zealand
Norway
Pakistan
Philippines
Senegal
South Africa
South Korea
Sweden
Switzerland
Taiwan
Thailand
Tunisia
Turkey
United Kingdom
United States
Uruguay
Vietnam

Australia

Source: Australian Department for Health; State-level Governments
Short description: The source reports this data as the ‘number of tests performed’. However, based on the context, it is unclear whether this refers to the total number of tests conducted, or number of people who have been tested.

Latest estimate: 12.57 per thousand people; 319,368 in total (as of 08 April 2020).

Detailed description:

The Australian Department for Health publish weekly epidemiological reports of confirmed cases, deaths and testing which includes demographic breakdowns of cases and symptoms.

In these reports they have, since 22nd March, included a total cumulative figure of tests performed. It is not clear whether this is equivalent to the number of people tested; the fact that they also report the share of these that are positive suggests this might be the case.

Weekly epidemiological reports are also published prior to 22nd March, however, they only include figures on confirmed cases and do not include quantiative figures on the number of tests performed.

Data for 2nd April comes from the most recent Federal Ministry of Health report: these have not yet been published nationally by the Australian Department for Health since we are awaiting its latest weekly epidemiological report. The first data point is for the week ending 22nd March (143,056 cumulative tests performed). It is not clear from what date this total refers.

Austria

Source: Austria Ministry for Health
Short description: The source reports this data as the ‘number of tests carried out so far’. However, based on the context, it is unclear whether this refers to the total number of tests conducted, or number of people who have been tested.

Latest estimate: 13.75 per thousand people; 120,755 in total (as of 08 April 2020).

Detailed description:

The Austrian Ministry for Health publishes daily updates on the number of confirmed cases, deaths and tests carried out to date.

This daily update states the ‘number of tests carried out so far/to date’ i.e. the cumulative number of tests. However, what is unclear is whether this refers to the number of tests conducted, or the number of people who have been tested.

The first report of this kind on the Austrian Ministry for Health website is from the 25th of February; we do not know the first date that testing began, only that 218 tests had been carried out as of 25th February. The Austrian Ministry of Health updates its main page on Coronavirus reports daily; historic time-series trends can be compiled using previous versions of this webpage at Web Archive.

Bahrain

Source: Bahrain Ministry of Health
Short description: The source reports the ‘number of assessments’ conducted. It is unclear whether this refers to the total number of tests conducted, or the number of people tested.

Latest estimate: 31.1 per thousand people; 52,804 in total (as of 08 April 2020).

Detailed description:

The Bahrainian Ministry of Health publishes frequent updates (not daily, but with high frequency) on the number of confirmed cases, status of existing cases and number of assessments conducted.

These figures represent the cumulative tests to that given date. It is not clear whether this refers to the total number of tests conducted, or the number of people tested.

These updates are not provided daily – typically every few days, so a daily time-series is not available. However, using web archives we can construct a time-series of tests conducted over time based on these frequent updates. It is not clear when testing first began; data is only available from 5th March where it was reported that 5334 tests had been conducted.

Belgium

Source: Sciensano (Belgian institute for health)
Short description: The figures are labelled as the ‘total number of tests performed’. Note that from other countries, we see that such a label can be consistent with figures relating to the number of individuals tested.

Latest estimate: 7.25 per thousand people; 84,248 in total (as of 07 April 2020).

Detailed description:

The Belgian institute for health reports the number of tests performed daily in CSV and JSON format. The date of is the date of laboratory diagnosis, or when not available, the date of sampling is used. A codebook is available.

Canada

Source: Government of Canada
Short description: The number of people tested.

Latest estimate: 9.55 per thousand people; 359,269 in total (as of 08 April 2020).

Detailed description:

A daily time series is not available on the Government of Canada’s website. It updates the page daily to report the latest cumulative number of people tested.

To build a time series we have aggregated the data from previous version’s of the Government of Canada website from Web Archive.

Data refers to the number of patients tested: it is the sum of positive and negative results (which are also provided), hence it does not include people who have their test results pending.

By combining daily snapshots, there is a large break/jump in the series from 17th to 18th March 2020. This is likely (the Government of Canada website does not make the reasoning clear, only publishes a slightly different format of table) the result of a large backlog of tests waiting to be processed. This makes the data misleading to use in time-series format prior to 18th March. We therefore only use data from 18th March onwards for comparability.

Costa Rica

Source: Costa Rican Ministry of Health
Short description: The number of people tested.

Latest estimate: 1.18 per thousand people; 5,963 in total (as of 08 April 2020).

Detailed description:

The Costa Rican Ministry of Health produce daily update reports of confirmed cases, deaths and test results.

These daily reports state the daily number of people who have been tested – given as the number of confirmed cases (‘Confirma contagio’) and number of people with negative results (‘Descarta contagio’) (thus the figures do not include pending tests). We can use these daily updates to construct a full time-series. This has been made downloadable.

Daily reports with figures on testing are only available dating back to 11th March. We therefore do not know the first date of testing, or daily figures prior to this date.

Czech Republic

Source: Ministry of Health
Short description: The number of tests performed.

Latest estimate: 9.28 per thousand people; 98,681 in total (as of 07 April 2020).

Detailed description:

The Ministry of Health publishes a data set containing incremental and cumulative daily numbers of COVID-19 tests performed according to laboratory reports. It is updated daily and accessible in CSV and JSON format.

Denmark

Source: Danish, Statens Serum Institut
Short description: The number of people tested.

Latest estimate: 9.91 per thousand people; 57,434 in total (as of 07 April 2020).

Detailed description:

The Danish Statum Serum Institute provides epidemiological update reports on confirmed cases, deaths, testing and demographic details of each.

It provides daily statistics on the number of people tested for COVID-19, and the number and share of people that tested positive.

The most recent epidemiological status reports provides the daily data for the previous two weeks (dating back to the 25th March), with a cumulative figure from 7 Jan – 25 March given.

The daily data for earlier dates can be seen in a chart that the Insitute show in their monitoring reports page.

Ecuador

Source: Government of Ecuador
Short description: The source reports this as the number of samples tested (‘muestras’). However this figure equals the sum of confirmed, negative (‘descartados’) and suspected ‘cases’, which might suggest that individuals are being counted. It may be that in the case of Ecuador the distinction is less important, if few people are tested more than once.

Latest estimate: 0.9 per thousand people; 15,526 in total (as of 08 April 2020).

Detailed description:

The Government of Ecuador publish daily updates in the form of situation reports and summary infographics. These report the number and status of confirmed cases, deaths and number of samples tested. This data is available daily from 18th March; reports and infographics prior to this date do not include the number of samples tested. But all figures are dated cumulative since 29th February. The first published figure is 762 samples tested since 29th February, as of 18th March.

It’s important to note that the dates here do not refer to the number of tests on a given day (or even when results come out), but instead reflect the date of the published report. The report notes that: “The difference between samples taken minus confirmed and discarded cases is due to suspected cases for which test results are pending”. This makes clear that total samples includes pending tests.

Estonia

Source: Estonian Central Health Information System and Patient Portal
Short description: The source reports this as the ‘number of tests administered’. However, based on the context, it is unclear whether this refers to the total number of tests conducted, or number of people who have been tested.

Latest estimate: 19.08 per thousand people; 24,813 in total (as of 07 April 2020).

Detailed description:

The Estonian Central Health Information System and Patient Portal provides a daily update of the number of tests administered on a given day. However, it is currently unclear whether this refers to the number of tests administered or number of people tested. It also reports a figure on the share of these tests which were positive; this would suggest it reflects the number of people of tested, however we cannot say with certainty without further information.

A complete time-series from 25th February is available, and is updated daily.

Finland

Source: Finnish Department of Health and Welfare COVID-19 data dashboard
Short description: The number of samples tested.

Latest estimate: 6.54 per thousand people; 36,501 in total (as of 07 April 2020).

Detailed description:

The Finnish Department of Health and Welfare publishes a dashboard of confirmed cases, deaths and tests sampled.

The dashboard provides daily figures of samples tested per day (and cumulative test numbers on a separate tab). We have copied this time-series as presented in the original source. It’s important to note that when figures exceed 1000, the Finnish Department of Health and Welfare report samples to the nearest hundred (e.g. 1.6k).

From the available documention we cannot be entirely certain whether ‘samples tested’ refers to the number of tests or number of people tested, given the general ambiguity in the usage of these terms internationally. However, a literal interpretation of the figures as the number of samples tested is consistent with further descriptions of the testing data provided here.

The data description notes that tests are being rationed and allocated according to the following strategy: “Samples are taken primarily from patients with severe symptoms of respiratory infection and from health and social care staff to ensure staff health and labor adequacy….Attention has also been paid to vulnerable groups of patients with some underlying disease and to those over 70 years of age.”

This data series extends to the 27th February, when test figures were very low (9 tests per day). It is likely this is was the first day, or close to the first day of testing.

France

Source: Agence nationale de santé publique
Short description: The testing figures are labelled as ‘number of tests’ in the source, but as of 2 April we are unsure of the right interpretation of the units.

Latest estimate: 3.41 per thousand people; 224,254 in total (as of 31 March 2020).

Detailed description:

The source publishes epidemiological updates at irregular intervals. The latest update, as of 2 April, was published on the 24 March. The updates include a figure for the number of tests (‘Nombre de tests’). However the context in which the figures are published suggests that these numbers may actually refer to the number of individuals tested. A breakdown into positive and negative results are given, along with a positive test rate.

Such a breakdown seems counter-intuitive if the figures relate to the number of tests performed rather than the number of individuals tested. In the former case, it would mean that the tests from people who had two two tests – e.g. both positive – would be added to tests of people who had five tests – e.g. three negative and two positive.

The figures in the time series relate to tests performed since 24 February. In the 24 March update, in addition to the previously reported hospital tests, laboratory tests are also reported. Only the total number of laboratory tests performed between 24 February and 24 March is presented, without a time series. Our figure for 24 March adds the cumulative number of laboratory tests since 24 February (6,500) to the hospital tests figure (101,046). As such 24 March represents a break in the series.

Germany

Source: Robert Koch Institut
Short description: The number of tests performed. Note that the number of laboratories reporting in the data seems to vary from week to week.

Latest estimate: 15.97 per thousand people; 1,317,887 in total (as of 05 April 2020).

Detailed description:

To determine how many laboratory tests regarding SARS-CoV-2 are carried out per calendar week in Germany and how many tests are positive or negative, the RKI has started a Germany-wide laboratory query. However, the number of laboratories reporting data seems to vary from week to week.

The report published on 8 April states that “from the beginning of the collection up to and including calendar week 14/2020”:
– The cumulative total of samples tested was 1,317,887;
– For calendar week 14 (which ends on 5 April), 143 labs reported 392,984 samples tested;
– For calendar week 13 (which ends on 29 March), 149 labs reported 360,964 samples tested;
– For calendar week 12 (which ends on 22 March), 152 labs reported 348619 samples tested;
– For calendar week 11 (which ends on 15 March), 114 labs reported 127457 samples tested.
– For calendar week 10 (which ends on 8 March), 48 labs reported 87,863 samples tested.

By subtracting each weekly change from the cumulative total, we retrospectively work out the cumulative totals by the end of each week.

Since laboratories can post-check the tests of past calendar weeks in the RKI test number query, previous figures may be revised upwards slightly in subsequent reports. The source is explicit that these figures refer to tests performed and that this will not equal the number of people tested, because of multiple tests per person.

Greece

Source: The National Organization of Public Health
Short description: The number of people tested.

Latest estimate: 2.57 per thousand people; 28,584 in total (as of 07 April 2020).

Detailed description:

The Greek National Organization of Public Health publishes daily COVID-19 reports on confirmed cases, deaths and samples tested.

It refers to its testing data as the number of people who have tested positive and negative to date: this means figures represent the number of people tested, excluding those pending results.

The official website provides a list of daily reports. Unfortunately many reports are missing meaning a full daily time-series cannot be constructed. It is unclear how far back testing extends; the earliest available report is on 20th March when 7172 people had been tested.

Iceland

Source: Office of the Director of Public Health
Short description: The source reports this as the ‘number of samples tested’. However, based on the context, it is unclear whether this refers to the total number of tests conducted, or number of people who have been tested.

Latest estimate: 90.16 per thousand people; 30,947 in total (as of 07 April 2020).

Detailed description:

The Icelandic Office of Public Health publishes a complete time-series of daily samples tested. This time-series is broken down by the two main labs where these tests are sampled.

It is not clear from the context whether this reflects the number of tests, or the number of people who are tested.

The daily time-series data only extends back to 27th February. However, the source shows clearly via annotation on the testing chart that between 1st and 26th February, 41 samples had been tested. We do therefore not know the days across which these samples were taken, only that as of 26th February, the total cumulative number of samples tested was 41.

India

Source #1: Indian Council of Medical Research
Short description: The number of people tested.

Latest estimate: 0.02 per thousand people; 26,798 in total (as of 27 March 2020).

Detailed description:

The ICMR reports separate figures for both “samples tested” and “people tested” at press conferences and in press releases (shown separately in the charts above). No other details are provided.

The press releases from ICMR do not always stay online for very long. The reason for this is unknown, but the releases are being backed up at this Github repository. Some figures announced during press conferences are also available in articles published by Bloomberg Quint.

On some occasions there appear to have been more than one update released per day. Where we are aware of multiple observations for the day, we show the number for the earlier release.

Source #2: Indian Council of Medical Research
Short description: The number of samples tested.

Latest estimate: 0.07 per thousand people; 101,968 in total (as of 06 April 2020).

Detailed description:

The ICMR reports separate figures for both “samples tested” and “people tested” at press conferences and in press releases (shown separately in the charts above). No other details are provided.

The press releases from ICMR do not always stay online for very long. The reason for this is unknown, but the releases are being backed up at this Github repository. Some figures announced during press conferences are also available in articles published by Bloomberg Quint.

On some occasions there appear to have been more than one update released per day. Where we are aware of multiple observations for the day, we show the number for the earlier release.

Indonesia

Source: Emerging infections, Indonesian Ministry of Health
Short description: Figures are labelled as “specimens accepted”, but the context suggests the figures may relate to the number of people tested.

Latest estimate: 0.05 per thousand people; 14,571 in total (as of 08 April 2020).

Detailed description:

The Indonesian Ministry of Health updates its COVID-19 dashboard twice a day but doesn’t keep time series of past data. The two URLs (1, 2) seem to lead to the exact same dashboard.

The dashboard shows the latest number of “Spesimen Diterima”, which translates to “specimens accepted”. However the figures equal to total confirmed and negative cases (‘Kasus Konfirmasi’, ‘Kasus Negatif’), strongly suggesting that these numbers actually refer to the number of people tested.

We found past values using Internet Archive’s Wayback Machine.

Ireland

Source: Department of Health
Short description: The units are unclear. Some press releases mention “tests carried out”, while others give figures for “people tested”.

Latest estimate: 6.18 per thousand people; 30,213 in total (as of 31 March 2020).

Detailed description:

The Irish Department of Health publishes daily press releases, which sometimes (about once a week) include an updated cumulative total for national tests.

Early documents (e.g. 9 March) gave a number of “people tested”, but more recent ones mention “tests carried out”. As of 4 April we are not exactly sure what these numbers describe (tests or people).

No other details about what the numbers refer to are known.

Italy

Source: Ministero della Salute
Short description: We currently have very little information about this data. More details will be provided as we find out more.

Latest estimate: 13.65 per thousand people; 807,125 in total (as of 08 April 2020).

Detailed description:

In particular the unit being counted is not currently known. The figures are provided in a table under a heading of ‘Tamponi’ (‘Swabs’). However, the context of the table suggests the figures may relate to the number of people tested.

Japan

Source: Ministry of Health, Labor and Welfare Press Release
Short description: The number of people tested.

Latest estimate: 0.37 per thousand people; 46,172 in total (as of 06 April 2020).

Detailed description:

A time series has not yet been released. The figures are the daily reports provided for the total number of people tested. Daily changes in the number of people tested are included in parentheses. From the 3rd April, the daily changes do not reconcile with the cumulative totals. Moreover, it is clear that revisions to past data are simply added to the new cumulative total, making the time series of ‘daily tests’ erratic.

This includes two cases where the cumulative number of people tested falls: (1) 19/03 and (2) 25/03. For case (1) see footnote 4 which indicates a past mistake has been noticed and the cumulative figure revised on the date to adjust for this. For case (2), we could not find the associated footnote.

It isn’t clear what exact date these cumulative tests date back to, but it is earlier than 10 Feb when the source reports 938 people had been tested. Prior to the 10 Feb, the press releases provide reports of coronavirus infections for the Diamond Princess cruise ship quarantined in Yokohama port. We do not report these numbers in the time series as it is unclear how they relate to the cumulative totals.

Google translate was used while compiling this data so this may affect our interpretation of the data.

Lithuania

Source: Ministry of Health
Short description: The number of analysed samples. Note that from other countries, we see that such a label can be consistent with figures relating to the number of individuals tested.

Latest estimate: 8.97 per thousand people; 25,599 in total (as of 06 April 2020).

Detailed description:

A time series has not yet been released. The Ministry of Health provides fairly regular updates of the number of samples that have been analyzed for suspected coronavirus, to date. It is not clear the exact date these cumulative figures date back to.

The source provides the number of confirmed cases, deaths, survivors, and the number investigated the previous day but it is not entirely clear whether pending samples are included in the total to date or not.

Google translate was used while compiling this data so this may affect our interpretation of the data.

Malaysia

Source: Ministry of Health Malaysia
Short description: The number of cases tested, including pending.

Latest estimate: 1.77 per thousand people; 58,240 in total (as of 08 April 2020).

Detailed description:

A time series has not yet been released. The figures provided are daily updates to the total number of cases tested. It is not clear the exact date these cumulative figures date back to. The earliest release we have been able to find begins from the 14th February.

The source provides a breakdown of the number of positive, negative, and still pending tests. We report total tests as the sum of positive and negative tests, excluding pending tests. The title of the table indicates that these figures relate to cases. As far as we are aware, cases are equivalent to individuals tested.

In a previous version of this page, we reported total tests as the sum of positive, negative, and pending tests. However, since April 7th 2020, the source has not reported the number of pending tests each day. For this reason, we have updated the time series so that total tests is equal to the sum of positive and negative tests (excluding pending tests) for all days on which we report data.

Netherlands

Source: Dutch National Institute for Public Health and the Environment
Short description: The number of people tested.

Latest estimate: 5.91 per thousand people; 101,534 in total (as of 07 April 2020).

Detailed description:

The Dutch National Institute for Public Health and the Environment releases ‘virological daily reports’. Rather than providing a full time series, each release provides daily test figures for a handful of the most recent days and then shows the cumulative total prior to that. It appears that figures are retrospectively updated, because the chain of daily figures from previous releases does not perfectly agree with the cumulative total reported in subsequent releases (though the discrepancies are relatively small).

Available to us are those reports retrievable through Web Archive. The earliest is from 26 March. A report for everyday since then could not be retreived. We construct our series taking the cumulative total as reported in the most recent report, and filling in any gaps from the daily figures available in the most recent report. The daily tesing figures we provide are calculated based on the change in the this cumulative figure we contruct. As such they may disagree slightly with the figures published in earlier releases.

The source reports the number of labs reporting in a given daily or cumulative figure. These have varied over the course of the reports, and this will affect the consistency of the time series we present.

The earliest figure we have is a cumulative total (reported on the 26 March) for the 14 March. The release states that ‘all laboratories in the Netherlands that perform diagnostics for SARS-CoV-2 have been asked from 9 March report these data daily.’ The figure given notes that 24 labs were counted in this cumulative figure. More recent releases count around 35 labs in the cumulative totals provided. But there appear to be reporting lags – the number of labs reporting for the latest one or two days are a good deal lower than for previous days.

New Zealand

Source: New Zealand Ministry of Health
Short description: The number of tests performed.

Latest estimate: 9.7 per thousand people; 46,875 in total (as of 07 April 2020).

Detailed description:

The New Zealand Ministry of Health produce daily update reports of confirmed cases, deaths and tests. This information includes demographic information of confirmed cases.

Recent releases refer to ‘total tested to date’. This is suggestive that the figures relate to the number of people tested. However earlier releases suggest the figures refer the number of tests (variously labelled as ’lab tests to date’, the ‘number of tests processed to date’, ‘completed tests’).

We assume that the figures provided refer to the same units over time, but we are uncertain whether the units are the number of tests performed or the number of people tested.

Unfortunately, despite publishing daily reports and media releases, not all reports include the latest testing figures. This means we cannot construct a complete daily time-series with available data.

These reports are clear that test figures are cumulative from 9th March.

Norway

Source: Norwegian Institute of Public Health
Short description: The number of people tested, not including testing in private health services.

Latest estimate: 20.42 per thousand people; 111,299 in total (as of 06 April 2020).

Detailed description:

The Norwegian Institute of Public Health produce daily update reports of confirmed cases, deaths and test results. This information includes demographic information of confirmed cases.

These daily reports state the cumulative number of people who have been tested – including the number and share who tested positive. We can use these daily updates to construct a full time-series. Note that the source states that “Changes in test criteria may… have resulted in slightly fewer tests tested per day. Data for the last few days is incomplete and will be updated.” They also note that “The figure contains only numbers from laboratories that have reports that include the sampling date. “[via Google Translate].

In addition to the aggregate number of people tested, the source provides a regional breakdown, in which it is noted that tests in ‘private labs that analyze tests across the country’ are excluded. The sum of this breakdown is lower than the aggregate figure provided. We take this as an indication that the main aggregate figure we report includes all private testing.

Daily reports with figures on testing are only available dating back to 16th March. We therefore do not know the first date of testing, only that as of 16th March, 18062 people had been tested.

Pakistan

Source: Government of Pakistan
Short description: The figures are labeled ‘tests conducted’, but from the context of the tables in which they are given it seems they may refer to the number of people tested.

Latest estimate: 0.17 per thousand people; 35,875 in total (as of 05 April 2020).

Detailed description:

The government of Pakistan publishes a dashboard showing up-to-date national data.

The specific graph showing testing data is named “COVID 19 | Overview”. It stacks the number of tests conducted on top of the number of confirmed cases, indicating that the figures may actually refer to the number of people tested.

The figures sometimes shows important 1-day increases due to the inclusion of more labs.

Philippines

Source: Department of Health (DOH) Philippines
Short description: The number of cases tested.

Latest estimate: 0.04 per thousand people; 4,367 in total (as of 03 April 2020).

Detailed description:

A time series has not yet been released. The DOH provide somewhat regular updates of the total number of cases in the Philippines. It is not clear the exact date these cumulative figures date back to; however, the first situation report and recorded number of cases dates back to the 28th January.

The source provides a breakdown of the ‘Laboratory status of patients’ split by the number of confirmed cases, negative cases, and cases pending test results. We report the sum of confirmed and negative cases, excluding those still pending. As far as we are aware, cases are equivalent to individuals tested.

The COVID-19 dashboard and the Philippines Department of Health (DOH) website figures differ for the 03/04 and 29/03 despite the dates and times of the updates being identical in each case.

Senegal

Source: Ministry for Health and Social Action
Short description: The figures are labelled as the number of ‘tests performed’. Note that from other countries, we see that such a label can be consistent with figures relating to the number of individuals tested.

Latest estimate: 0.12 per thousand people; 2,072 in total (as of 06 April 2020).

Detailed description:

The Senegalese Ministry for Health and Social Action publish daily press releases which detail the number of tests performed, and the number of positive confirmed cases. It is not totally clear whether the number of tests performed is equivalent to the number of people tested.

These daily press releases report the number of tests performed each day, dating back to 17th March. Although daily press releases date further back than this, they do not detail the number of tests performed, only the number of confirmed cases. Our time-series data therefore only exists from 17th March onwards.

South Africa

Source: National Institute for Communicable Diseases (NICD)
Short description: The figures are labelled as the number of ‘tests conducted’. Note that from other countries, we see that such a label can be consistent with figures relating to the number of individuals tested. It may be that in the case of South Africa the distinction is less important, if few people are tested more than once.

Latest estimate: 0.97 per thousand people; 56,873 in total (as of 05 April 2020).

Detailed description:

The South African National Institute for Communicable Diseases (NICD) publishes daily updates and graphics on the number of confirmed cases, deaths and tests conducted nationally and by province. This is published consistently on its offical Twitter account.

It reports the number of ‘tests conducted’. It also reports the number of positive cases, and sometimes the number of ‘negative cases’. This is suggestive that ‘tests conducted’ refers to the number of people tested, but this is unclear from the aviable documentation.

The NICD began publishing these daily updates on 11th March, allowing us to develop a time-series from this date forward. We do not know the first date of testing, only that as of 11th March, 645 tests had been conducted.

South Korea

Source: South Korea CDC
Short description: The number of cases tested.

Latest estimate: 9.06 per thousand people; 466,804 in total (as of 06 April 2020).

Detailed description:

KCDC have provided daily updates in English since 21 January. The figures they provide relate to ‘cases’, where this signifies an individual considered eligible for testing due to their symptoms, travel history or contact history.

Currently the figures shown below include those cases pending test results (currently roughly 15,000 cases). When we next update we will switch to figures that exclude this category.

The daily updates show the change each day and the current totals. These form a consistent chain all the way back to 21 January. We are not aware of any significant issues affecting comparisons over time.

Sweden

Source: Public Health Agency
Short description: The number of people tested.

Latest estimate: 3.65 per thousand people; 36,900 in total (as of 29 March 2020).

Detailed description:

The weekly report gives the cumulative total of individuals analyzed since the diagnostics were set up in Sweden.

The report published on 31 March states up to week 13 (ending 29 March)
– A total of 36,900 individuals have been analyzed;
– During week 13, 12,300 individuals were analyzed.
– During week 12, 10,300 individuals were analyzed.
– During week 11, 9,000 individuals’ samples were.
– During week 10, 4,300 individuals were analyzed.

By subtracting each weekly change from the cumulative total, we can retrospectively work out the cumulative totals by the end of each week:
– By 29 March: 36900 total people tested;
– By 22 March: 36900 – 12300 = 24600 people tested;
– By 15 March: 24600 – 10300 = 14300 people tested;
– By 8 March: 14300 – 9000 = 5300 people tested;
– By 1 March: 5300 – 4300 = 1000 people tested.

Figures are often rounded off to the nearest hundred, indicating that they might not be extremely precise. The report does not indicate whether pending tests are included in the counts.

Switzerland

Source: Federal Office of Public Health
Short description: The figures are labelled as the ‘number of tests carried out’ (‘durchgeführten Tests’). Note that from other countries, we see that such a label can be consistent with figures relating to the number of individuals tested.

Latest estimate: 19.83 per thousand people; 171,938 in total (as of 08 April 2020).

Detailed description:

In a daily report on the epidemiological situation in Switzerland and the Principality of Liechtenstein, the Federal Office of Public Health publishes an up-to-date cumulative total of tests carried out across the country, since the first laboratory-confirmed case recorded on February 24.

The report available on the website only shows the latest figure. For previous values:
– We used archive.org for days where the PDF file was archived;
– Gaps were filled for 3 days based on data sent to us on Twitter.

Figures are often rounded off to the nearest thousand, indicating that they might not be extremely precise. The report does not indicate whether pending tests are included in the counts.

Taiwan

Source: Taiwan Centers for Disease Control (CDC)
Short description: The figures are labelled in the source only as ‘tested’. It is unclear whether this relates to the number of individuals tested, or the number of samples tested.

Latest estimate: 1.56 per thousand people; 37,219 in total (as of 06 April 2020).

Detailed description:

The Taiwanese Centers for Disease Control (CDC) host a dashboard in which they publish the total ‘tested’ to date, and ‘new from yesterday’. It is not clear whether this represents the total number of tests performed, or the number of people tested. The number ‘tested’ is greater than the sum of confirmed cases and ‘excluded’ (i.e. negative) test results. This would suggest it does not equal the total number of people tested, or if it does, pending results without a reported outcome are also included.

Although the CDC only show the last day’s figures on this dashboard, we can construct a time-series by looking at previous versions on web archive. Where possible we have tried to take the total test counts at the same time every day (those published at 00:30h). Unfortunately this time of publishing from CDC was not always completely consistent. This may lead to small discrepancies between ‘total tested’ and ‘new from ‘yesterday’ figures. Wherever possible we took the total test count as of 00:30 as the prefered figure; where a day of data was missing we used the ‘new from yesterday’ figure to calculate the total for the previous day. If there are small discrepancies with other sources, this timing issue is likely to be the cause. In any case it is likely to be minor.

The date from which the total test figures date back to is not known; the CDC dashboard is unavailable in web archives prior to 21st March.

Thailand

Source: Thailand, Department of Disease Control
Short description: The number of people tested.

Latest estimate: 0.35 per thousand people; 24,474 in total (as of 04 April 2020).

Detailed description:

The Thailand Department of Disease Control issues daily situation reports which detail the status of confirmed cases, deaths and people being assessed.

We provide the figures reported as ‘persons under investigation’ (PUI). PUI is the sum of confirmed cases, negatively tested people the number of people pending test results. All three figures are reported individually from 3rd March onwards.

Tunisia

Source: Tunisian Ministry of Health
Short description: Figures are provided both in terms of the number of cases tested and the number of tests. It may be that in the case of Tunisia the distinction between the number of individuals tested and the number of tests performed is less significant, if few people are tested more than once.

Latest estimate: 0.62 per thousand people; 7,351 in total (as of 04 April 2020).

Detailed description:

The Tunisian Ministry of Health dashboard provides daily figures of the total number of cases tested (‘Cas testés’). They also provide a time series for the number of tests per day (‘Nombre de tests journaliers’). The figures we provide relate to the latter.

The time series figures sum up to a number slightly higher than the cumulative figure provided for cases tested. We understand this to be equivalent to the number of individuals tested. It is unclear whether the difference between the numbers relates to a genuine distinction in the two figures between the number of indivudals tested and the number of tests performed, or if the differences are due instead to reporting delays or other unintended discrepancies.

The first date in the time series of daily tests is 10th March, for which 28 tests were reported.

Turkey

Source: Turkish Ministry of Health
Short description: The source reports this as the ‘total number of tests’. However, based on the context, it is unclear whether this refers to the total number of tests conducted, or number of people who have been tested.

Latest estimate: 2.96 per thousand people; 247,768 in total (as of 08 April 2020).

Detailed description:

The Turkish Ministry of Health now publishes a daily chart of Coronavirus confirmed cases, deaths, recoveries, total tests, and daily tests. What is unclear is whether ‘total tests’ here refers to the number of tests completed, or the number of people who have been tested.

These daily charts are published on the Turkish Ministry of Health’s website – previous versions can be found on Web Archive. This only extends back to 27th March. Prior to this date, we rely on figures reported by the Turkish Minister for Health at daily press conferences. In these reports they detail the latest update of confirmed cases, deaths and tests conducted.

We are not aware of the date that testing began: only that as of 18th March – the first data point in our series – 10,018 tests had been conducted.

United Kingdom

Source: Public Health England/Department of Health and Social Care
Short description: The total number of people tested.

Latest estimate: 3.46 per thousand people; 232,708 in total (as of 08 April 2020).

Detailed description:

A time series is not yet released. The figures we provide relate to the daily updates provided for the total number of people tested. It is not clear the exact date that these cumulative figures date back to.

The source provides a breakdown of this total into positive and negative results – as such pending results are evidently not included in the total.

No further details are currently known.

United States

Source #1: COVID Tracking Project
Short description: The data are an aggregation of figures released by individual states. The Project aim to report on the number of people tested, including private labs, but not all states report their figures in this way.

Latest estimate: 6.61 per thousand people; 2,189,766 in total (as of 08 April 2020).

Detailed description:

This is a collaborative project launched in order to fill some of the important gaps in the testing figures being collated by the CDC.

Testing data is gathered from individual states, as reported in state health department websites, data dashboards and press releases from officials.

States currently report testing figures in a range of different ways: some report the number of tests performed, others the number of people tested; some include private labs, others not; some report negative test results, others only positive test results; some include pending tests, others do not (below we show figures that exclude explicitly pending results).

Moreover, many states do not explicitly provide details about these important factors needed to interpret the data they provide.

There are issues in comparing the figures over time. The totals given for early on in the outbreak do not include all states. One significant uncertainty is the extent to which the rapid rise in tests seen from the mid-March in part reflects states beginning to report private lab tests.

Overall the figures should be seen as providing a general indication of testing output, rather than a specific count of a given indicator. Given the very incomplete coverage and reporting delays of the CDC data, it provides a very important additional perspective.

The Project documents their work in lots of detail. See the link provided above for full details.

Source #2: US CDC
Short description: The number of specimens tested in public labs only.

Latest estimate: 0.58 per thousand people; 192,166 in total (as of 31 March 2020).

Detailed description:

In addition to CDCs labs, the figures include specimens tested at public health labs in 49 states, New York City, Puerto Rico, USAF, and 15 California Counties.

As such the figures do not include private lab tests, which are occurring in significant numbers.2

Daily figures are provided since 18 January. CDC suggest that figures within 7 days of the update “are incomplete because of the lag in time between when specimens are accessioned, testing is performed, and results are reported”. As such we only provide figures after this 7 day cut off.

The CDC documents very well what is and is not included in its published testing figures. See the source linked to above for more details.

Uruguay

Source: Ministry of Public Health
Short description: The testing figures are labelled as tests performed (‘tests realizados’) in the source, but we are currnetly unsure of the right interpretation of the units.

Latest estimate: 1.61 per thousand people; 5,614 in total (as of 06 April 2020).

Detailed description:

The Ministry of Public Health in Uruguay publishes a dashboard providing daily figures of tests performed (‘tests realizados’). However the context in which the figures are published suggests that these numbers may actually refer to the number of individuals tested. A breakdown into positive and negative results are given, along with a positive test rate. The number of ‘positive tests’ is equal to the number of confirmed cases.

Such a breakdown seems counter-intuitive if the figures relate to the number of tests performed rather than the number of individuals tested. In the former case, it would mean that the tests from people who had two two tests – e.g. both positive – would be added to tests of people who had five tests – e.g. three negative and two positive.

As well as a cumulative figure for negative and positive tests, the source provides a time series from the 22nd March onwards. However these two sets of numbers appear somewhat inconsistent. The reasons for this inconsistency are not known. The inconsistency is reflected in the data we provide: the cumulative figures we show are as published on the dashboard and recoverable through web archives; the daily figures are as given in time series provided in the most recent update.

For more detials on the construction of the series, see the observation-by-observation source notes in the data download linked to at the top of this section.

Vietnam

Source: Vietnamese Ministry of Health
Short description: The source reports this as the ‘number of samples tested’. However, based on the context, it is unclear whether this refers to the total number of tests conducted, or number of people who have been tested.

Latest estimate: 1.12 per thousand people; 110,088 in total (as of 08 April 2020).

Detailed description:

The Vietnamese Ministry of Health update its disease situation statistics page daily with the latest data on confirmed cases, deaths and samples tested.

It refers to its testing data as ‘samples tested’, and reports confirmed cases and negative samples separately; ‘sample tested’ generally matches the sum of confirmed cases and negative samples (though there are occasionally some small discrepancies – we report the sum of positive and negative in this case). This would suggest that ‘samples tested’ is equivalent to the number of people tested. However, it’s unclear whether ‘negative samples’ includes some instances of resampling for the same person. We therefore cannot say with certainty that these are equivalent.

Using web archives we can reconstruct a daily time-series of total samples by looking at previous versions of the updated website. These daily updates extend back to 2nd March. We cannot say with certainty when testing began, only that as of 2nd March 1753 samples had been tested.