For Days, Samples Tested Were More Than People Tested By Exactly 890

New Delhi: Here’s a new puzzle for scientists, doctors and journalists tracking the spread of COVID-19 in India: the difference between the number of people tested by the government and the number of samples tested has been a constant and recurring figure of 890 everyday, official data show.

As of 9 p.m. on April 4, 2020, India had tested 79,950 samples and 75,000 people, data from the Indian Council of Medical Research (ICMR) show. Is this number meaningful in any way? Is it big enough? Can we take this number at face value? Experts have been debating. 

For the first few weeks of the spread of COVID-19 in India, a lot of time was spent analysing if India was testing enough. The broad consensus was that for a country of 1.3 billion people, India was testing way too little. This was due to a restricted eligibility criteria and a shortage of testing kits, with the government still scrambling to assess the quality of kits and open up calls for supply.

But here’s something else to consider: When the singer Kanika Kapoor was suspected to be positive for COVID-19, she was tested multiple times. It was confusing for many who wanted to know: Were tests being wasted on her?

In fact, repeated testing is an important step in the protocol for any accurate diagnosis. It is done to rule out the possibility of false positive or false negative results, which sometimes turn up.

So with this information also in mind, what does it mean when we say India has done 79,950 tests? Are we testing enough?  

Since March 2020, the ICMR has been releasing data--albeit erratically--for the number of samples tested and also the number of individuals tested.

The difference between these numbers remained a steady 890 as long as the numbers of individuals tested were released (the ICMR stopped including the number of people tested in its press releases from April 2). 

The fact that the difference between number of samples and people tested keeps coming up to 890, is is strange for a couple of reasons: First, given that an average patient or suspect is tested at least twice if not thrice, the difference should be much larger (more on the protocol is discussed below). Secondly, this difference should definitely vary every day given that the number of tests done, the positive results and the number of people being discharged, is also varying daily.

I tried to ask a question about this at four of the daily press briefings. On April 4, 2020, this question was taken up by Raman Gangakhedkar, head of epidemiology and communicable diseases at the ICMR. He replied, "It is difficult to say off-hand. The number of repeat tests will likely be small. Up until now, 75,000 people have been tested.”

I spoke to other epidemiologists and scientists at ICMR. A senior scientist who is coordinating the ICMR’s work on COVID-19 confirmed one basic and important detail off the record: The numbers for total samples tested include repeated tests done. 

This means that the number of people tested should be far smaller and that the difference between the two data points should be far larger. And the difference between them should probably not be a stagnant number of 890. 

Several documents published by the government list, as we explain below, how testing and repeat-testing is to be done: For anyone to be declared recovered and discharged, they need to be tested repeatedly. As of 9 a.m on April 5, 2020, 266 people had been discharged (after being declared recovered and COVID19-free), according to the government.

How does repeat-testing work?

A person who has tested positive will need to test negative twice in order to be discharged, according to the government’s ‘Containment Plan’ for COVID-19. These two negative results must come from tests conducted consecutively at a 24-hour interval. 

Sometimes, a positive test may come in between, and so the person will have to remain in isolation or in the hospital and be tested again until there are two consecutive negative results. This discharge protocol is listed in at least four documents published by the government (here, here, here and here).

This means that there have already been at least three tests conducted on the 266 people who have been discharged. So, of the total 79,500 samples tested in India, at least 798 should be repeat tests.

How to clarify why the 890 figure keeps coming up?

With all eyes on India--given its vast population and poor health system--data are valuable to understand what is happening right now, but also for researchers to study what happened in India when this crisis has finally passed.

If the Indian government releases proactive and timely data, this will be in public interest.

However, there has been discrepancy between what the Union health ministry and ICMR have been reporting. There was a break of several days when ICMR stopped putting out information on samples and individuals tested. Now they have re-started publishing these data, but they are only putting out the total number of samples tested and not the number of individuals tested.

The scenario is further complicated because, since last week, the government has been putting curbs on journalists attending the press conferences and also asking questions (read this thread on questions that have gone unanswered), restricting it to only certain media houses and reporters. Journalists can only ask questions by posting it on the government’s WhatsApp group for reporters and hope that it gets picked for answering. 

At the 30-minute daily briefings, the government is only taking up about four or five questions. Getting the issue of repeat-testing clarified was itself difficult, and with the curbs on the press, it is likely to continue to be difficult to get clarity on this matter. But the question remains open and relevant for anyone crunching the data to bear in mind and try to explain.

(Anoo Bhuyan is a special correspondent at IndiaSpend.)

New Delhi: Here’s a new puzzle for scientists, doctors and journalists tracking the spread of COVID-19 in India: the difference between the number of people tested by the government and the number of samples tested has been a constant and recurring figure of 890 everyday, official data show.

As of 9 p.m. on April 4, 2020, India had tested 79,950 samples and 75,000 people, data from the Indian Council of Medical Research (ICMR) show. Is this number meaningful in any way? Is it big enough? Can we take this number at face value? Experts have been debating. 

For the first few weeks of the spread of COVID-19 in India, a lot of time was spent analysing if India was testing enough. The broad consensus was that for a country of 1.3 billion people, India was testing way too little. This was due to a restricted eligibility criteria and a shortage of testing kits, with the government still scrambling to assess the quality of kits and open up calls for supply.

But here’s something else to consider: When the singer Kanika Kapoor was suspected to be positive for COVID-19, she was tested multiple times. It was confusing for many who wanted to know: Were tests being wasted on her?

In fact, repeated testing is an important step in the protocol for any accurate diagnosis. It is done to rule out the possibility of false positive or false negative results, which sometimes turn up.

So with this information also in mind, what does it mean when we say India has done 79,950 tests? Are we testing enough?  

Since March 2020, the ICMR has been releasing data--albeit erratically--for the number of samples tested and also the number of individuals tested.

The difference between these numbers remained a steady 890 as long as the numbers of individuals tested were released (the ICMR stopped including the number of people tested in its press releases from April 2). 

The fact that the difference between number of samples and people tested keeps coming up to 890, is is strange for a couple of reasons: First, given that an average patient or suspect is tested at least twice if not thrice, the difference should be much larger (more on the protocol is discussed below). Secondly, this difference should definitely vary every day given that the number of tests done, the positive results and the number of people being discharged, is also varying daily.

I tried to ask a question about this at four of the daily press briefings. On April 4, 2020, this question was taken up by Raman Gangakhedkar, head of epidemiology and communicable diseases at the ICMR. He replied, "It is difficult to say off-hand. The number of repeat tests will likely be small. Up until now, 75,000 people have been tested.”

I spoke to other epidemiologists and scientists at ICMR. A senior scientist who is coordinating the ICMR’s work on COVID-19 confirmed one basic and important detail off the record: The numbers for total samples tested include repeated tests done. 

This means that the number of people tested should be far smaller and that the difference between the two data points should be far larger. And the difference between them should probably not be a stagnant number of 890. 

Several documents published by the government list, as we explain below, how testing and repeat-testing is to be done: For anyone to be declared recovered and discharged, they need to be tested repeatedly. As of 9 a.m on April 5, 2020, 266 people had been discharged (after being declared recovered and COVID19-free), according to the government.

How does repeat-testing work?

A person who has tested positive will need to test negative twice in order to be discharged, according to the government’s ‘Containment Plan’ for COVID-19. These two negative results must come from tests conducted consecutively at a 24-hour interval. 

Sometimes, a positive test may come in between, and so the person will have to remain in isolation or in the hospital and be tested again until there are two consecutive negative results. This discharge protocol is listed in at least four documents published by the government (here, here, here and here).

This means that there have already been at least three tests conducted on the 266 people who have been discharged. So, of the total 79,500 samples tested in India, at least 798 should be repeat tests.

How to clarify why the 890 figure keeps coming up?

With all eyes on India--given its vast population and poor health system--data are valuable to understand what is happening right now, but also for researchers to study what happened in India when this crisis has finally passed.

If the Indian government releases proactive and timely data, this will be in public interest.

However, there has been discrepancy between what the Union health ministry and ICMR have been reporting. There was a break of several days when ICMR stopped putting out information on samples and individuals tested. Now they have re-started publishing these data, but they are only putting out the total number of samples tested and not the number of individuals tested.

The scenario is further complicated because, since last week, the government has been putting curbs on journalists attending the press conferences and also asking questions (read this thread on questions that have gone unanswered), restricting it to only certain media houses and reporters. Journalists can only ask questions by posting it on the government’s WhatsApp group for reporters and hope that it gets picked for answering. 

At the 30-minute daily briefings, the government is only taking up about four or five questions. Getting the issue of repeat-testing clarified was itself difficult, and with the curbs on the press, it is likely to continue to be difficult to get clarity on this matter. But the question remains open and relevant for anyone crunching the data to bear in mind and try to explain.

(Anoo Bhuyan is a special correspondent at IndiaSpend.)


13 responses to “For Days, Samples Tested Were More Than People Tested By Exactly 890”

  1. Many congratulations for the deep and meaningful analysis. Outstanding job. It’s very rare that one finds journalists doing such excellent analysis. As a professional data scientist, I completely agree with you on this issue. Sincere congratulations on insightful work. Please keep this up. (Interestingly, ICMR has removed many press releases–you have posted a link to those in answer to one of the replies though.)

  2. IndiaSpend team, I checked for press releases on the ICMR website for all the dates where data have been mentioned in the table used in the article. There aren’t any press releases which support or speak about the numbers mentioned in the table used above. Even if we assume that the ICMR has maliciously removed all those press releases after looking at your story, then I believe all those press releases would have been in PDF format for sure, and most probably, your team would have downloaded one copy for your reference to cite the numbers. I request you to share those PDF copies so that the numbers can be easily referred. I also went and heard the press briefings for most of the dates mentioned in the data table above. There is actually no mention of the separate numbers quoted for samples tested and individuals tested. The number which was quoted time and again was for samples tested and no data for individuals tested were reported. The only day when samples and individuals separately were quoted when figures of 79,000 for samples and 75,000 for individuals was mentioned, and even on that, the ICMR representative said it is not readily available with him. He just said approximately. Now, please share more details. Else, I would have to file a defamation notice with the help of my lawyer friends against this website for publishing misleading and false content.

    • Thank you for your comment. You’re right, the ICMR has not maintained a record of its past press releases. These numbers are very much a part of the ICMR’s press releases for the dates mentioned in the story, though (you can find the archive here: https://www.indiaspend.com/wp-content/uploads/2020/04/ICMR-releases.zip). They have also been widely reported in the media. We are surprised by the tone of your comment, which quickly went from requesting information to threatening legal action.

  3. Do you folks have a PDF copy of the press releases for all the dates mentioned in the table above? I checked on the ICMR website and there aren’t any press releases that speak about the numbers mentioned–about samples and individual people tested. Have those guys removed those press releases from the website?

  4. But, that precisely is against the protocol, given that most of testing is only 30% accurate overall. As on April 7, I have heard of at least two cases who died of COVID-19 who earlier tested negative! If there is some negligence, then it is very unfortunate.

  5. There is nothing surprising. The numbers are cumulative, so any difference that was there initially (for whatever reason) stays. The number is static at 890 because from March 19 onwards, one sample was tested per person. However, this does not mean that one person was tested only once–as the same person can be tested on March 19, 23 and 28, and would be counted in the records for those days. So there is no conspiracy going on here.

  6. This table has cumulative frequency. Once you convert it back to absolute numbers, the picture will be clear. Till March 19, more than one sample was collected per individual. After March 19, one sample per individual was collected–so no change in the counts.

  7. Is there a single place where the government publishes all corona-related data? I see good data on covid19india.org, but it would be better if the government can publish all data–and more importantly, accurate data.

  8. It is possible that they are adding the number of persons against number of samples. For example, if a person (say, Kanika Kapoor) has been tested several times, they may be counting it as one (unique case).

  9. The question of state-wise tests is equally important. Karnataka seems to have a lower incidence of the virus. Is this due to fewer tests or a genuine case of a lower incidence of the virus? The answers are not clearly visible.

Leave a Reply to Rohan Prasad Cancel reply

Your email address will not be published.

*

code