COVID-19 data: The 890 puzzle still unsolved

On April 5, 2020, IndiaSpend reported an anomaly in the Indian government’s data on COVID19--for 11 days in March, the difference between the number of people tested and the total number of samples tested, had been a repeating figure of 890.

With recovered and deceased patients, whose numbers vary every day, no longer being tested, and varying numbers of new cases being added every day, the gap between the number of samples tested and the number of individuals tested would keep changing, and not be static. (And India requires a person to test negative in two consecutive tests at 24 hours’ difference in order to be declared free of COVID19.) 

A few explanations, but none fits

Reuters followed up on IndiaSpend’s story and ICMR's Rajni Kant told them that the discrepancy was explained by the 890 people evacuated from abroad, who had been tested twice in India. A government press release on March 9, 2020, had said 890 people were evacuated through February.

However, the data published in an ICMR paper puts this explanation into question. This paper says that these 890 evacuated people had come back through February 2020, so their testing and quarantine would have begun in February with some spillover into March. But the repeating figure of 890 shows up in the data for mid-March 2020. 

This paper also said that 1,544 samples were tested from these 890 people whereas the government data shows a repetition of 890 samples. 

Reuters also reported that initial forms which were filled when a person’s sample was collected, did not ask if the person was being repeat-tested. The form currently being used does include the question, “Is it a repeated sample?” 

This means, prior to this change in the form, every repeat-test was counted as a separate individual, which inflates the number of individuals tested.

Further, there were 53 people discharged on the 11 days that this anomaly occurred. If the repeating figure of 890 only reflects tests done on the evacuated people as per ICMR’s explanation, then data for the repeat-tests of these 53 people is not in the data on those 11 days, or they were discharged without repeat-testing. 

“These repeat-tests should also be showing up in the data on these days because people discharged have to test negative twice. That's the rule. This ensures that it is safe to discharge these people," said Gautam Menon, a professor of physics and biology at Ashoka University, Sonepat, Haryana.

What do we know about repeat-tests?

The number of repeat-tests and its break-down (such as how many people have been tested twice, thrice or more times) is not a data-point that the Indian government is currently releasing. The only way to know at least the consolidated number of repeat-tests conducted is for the government to disclose the daily numbers of samples and individuals tested. 

At one of the government’s daily press briefings on April 4, 2020, Ramanan Gangakhedkar, head of epidemiology and communicable diseases at ICMR, said he would have to find out about repeat-testing data, in response to a query by IndiaSpend.

On following up, Gangakhedkar said that 853 repeat-tests had been done in India as on April 3, 2020. This includes a range of repeat tests: For example, as of April 3, 184 people were tested twice, 90 were tested thrice and one person was tested 11 times.

Bringing more integrity to India’s COVID-19 data

"I can see no straight-forward explanation for this anomaly of the 890 figure,” Menon said. “Maybe there was an error while entering the data. But I'd like to have seen some attempt by the government to explain this.”

The integrity of India's data on COVID-19 matters to people like Menon who are closely studying it. "Apart from not addressing this anomaly, the government is also releasing just the most minimal amount of data possible,” he said. “The old press releases are being deleted from the website. It is the ICMR's responsibility to collect and release as much data to the public and press so that transparency is ensured."

Since IndiaSpend’s blog was published on April 5, 2020, several journalists have posed questions on India’s repeat testing figures and about individuals tested to the Union health ministry’s press officer. The questions have so far not been picked for answering in the government’s press briefings. 

ICMR stopped releasing data for individuals tested after March 27, 2020, then re-started giving out this figure from April 9, 2020. As of April 11, 179,374 samples from 164,773 people had been tested, data show. This is a difference of 14,601. By 2.30 p.m. on April 12, 2020, India had tested 186,906 samples, of which 7,953 (4.3%) had tested positive, an ICMR representative said.

IndiaSpend has reached out to ICMR’s director general Balram Bhargava with queries on this via email and also contacted him via phone and SMS. This piece will be updated when he replies.