Bengaluru: As public health experts continue to wonder why India has such low rates of diagnostic tests for COVID-19 per capita, a recently published study by ICMR scientists indicates the council may have ignored its own analysis on the need for more aggressive testing.
A mathematical model prepared by ICMR scientists almost two months ago suggested that simply isolating symptomatic international air passengers could not have helped delay a COVID-19 epidemic in India.
In the model, published in a paper this month, scientists from the Indian Council of Medical Research (ICMR) compared a scenario of ‘no airport screening’ with three other scenarios in which airport screening detected all symptomatic cases. The researchers found that even flawless screening couldn’t have delayed a COVID-19 epidemic in India by more than 2.9 days because such screening wouldn’t capture infectious people who weren’t yet showing any symptoms (a.k.a. pre-symptomatic cases).
The model also suggested that the only strategy that could appreciably ‘flatten the curve’ for India would be extensive testing of symptomatic people with no travel history, allowing up to 50% of all COVID-19 infections to be identified.
These findings raise serious questions about ICMR’s reluctance to widen testing beyond people with a travel history until only last weekend.
Jacob John, a professor of community medicine at Vellore’s Christian Medical College, said, “I believe the testing of only travel-related cases was a serious error.” As of March 17, ICMR had tested only 500 community cases of severe acute respiratory illness (SARI). The agency changed its strategy only on March 20, to include all SARI and pneumonia cases.
Further, after WHO director general Tedros Adhanom Ghebreyesus asked all countries to “test, test, test”, ICMR officials claimed to have written to WHO’s India office that Ghebreyesus’s statement was unwarranted because it didn’t apply to countries like India with no evidence of community transmission.
When asked why ICMR waited so long to widen community testing, despite the agency’s own model suggesting otherwise, the paper’s coauthor and epidemiologist Tarun Bhatnagar said logistical constraints had played a part. “This is not Singapore or South Korea, which are like a single Indian state. We have to take a balanced approach to see that we are using our resources in an optimal manner,” he told The Wire Science.
However, by ICMR’s own admission, India had ample capacity to conduct tests for COVID-19, and so a shortage wasn’t the cause for the slow ramp-up. Raman Gangakhedkar, another coauthor of the paper, has said previously that ICMR was only using 10% of its testing capacity. The reason it wasn’t expanding further, he explained, was because there was no evidence of community transmission.
This conclusion was in turn seemingly based on negative test results from 500 SARI samples – a sample size that other experts have called inadequate for a country of over 1.3 billion people.
The futility of airport screening
ICMR’s paper, published in the Indian Journal of Medical Research, estimates how quickly the entry of infected people from China would result in 1,000 COVID-19-positive cases in India. The authors describe the scenario of 1,000 cases as an “epidemic” because it would imply extensive transmission within the country.
To calculate the time to an epidemic, the authors used air travel data from China to India between October 2019 and March 2020. Based on this and the rate at which infected people have been known to transmit the virus, they estimated how long India would take to witness an epidemic in three scenarios.
In the first scenario, airport-screening caught all symptomatic cases. In the second, it caught an additional 50% of all pre-symptomatic cases. In the third scenario, the additional number of pre-symptomatic cases rose to 90%.
A growing body of research suggests that pre-symptomatic people could be passing the virus to other people. For example, data from Singapore and China’s Tianjin indicates that COVID-19 patients were transmitting the microbe around two or three days before the onset of symptoms.
The ICMR scientists found that in the first scenario – the one closest to reality – the epidemic hardly changed its course. Identifying 100% of symptomatic cases put off the epidemic by a maximum of 2.9 days compared to no airport screening at all. Testing travellers with no symptoms helped a lot more – creating a maximum delay of 20 days – but India did not do that in reality.
The importance of community testing
In another simulation in the study, the authors considered an alternative intervention, where community cases (with no travel history) were tested widely for COVID-19. They estimated that if such testing succeeded in identifying half of all COVID-19 cases within three days of them showing symptoms, the epidemic’s peak could be flattened greatly in each of India’s four metropolises: Delhi, Mumbai, Chennai and Kolkata.
This was true for both optimistic and pessimistic assumptions. The optimistic scenario captures a situation in which each individual in an outbreak infects an average of 1.5 other people, a number known as R0 [Footnote]R0, pronounced R-zero, is the total number of new infections seeded by every infected person at the start of an outbreak[/footnote]. Also, pre-symptomatic people don’t infect anyone at all.
Meanwhile, the pessimistic scenario works with an R0 of 4 and assumes that pre-symptomatic people infect half as many as symptomatic people do.
The model suggests that this strategy is much more effective at flattening the curve than airport-screening. For example, in the optimistic scenario, with no intervention, an outbreak in Delhi peaked at over 1.5 million cases within 200 days. However, if community testing identified 50% of COVID-19 cases, this peak fell to less than 250,000 cases, and occurred after 600 days.
Many assumptions used in ICMR’s model, which was ready by February 27 (the date on which the paper was submitted to the journal for consideration), are likely not valid any more. Scientists have used more recent data from multiple countries to revise their estimates of R0’s value, to around 2-2.5, according to Srinivasan Venkatramanan, an infectious-disease modeller at the University of Virginia.
Also, while it seems like pre-symptomatic people do infect others, how often they do so compared to symptomatic people remains unknown.
The model also had several limitations. For example, it only used data from passengers coming from China, neglecting other countries with a high COVID-19 case load. Second, it calculated the impact of community-based interventions only in four metropolises. Third, it assumed that each infectious person had an equal likelihood of mixing with everyone else in the city they were in. More realistic models factor in the fact that an individual is more likely to meet some people, say, their family members, over others.
The first two factors could have led the authors to underestimate the size of the peak and lengthened the time to epidemics, Srinivasan said, while the third could have done the opposite.
But both John and Srinivasan said the study’s conclusion still holds: that early and widespread community testing is critical. This agrees with multiple other mathematical models of COVID-19 spread that have appeared since, they said.
John added that since India delayed such widespread testing, it might now be too late to prevent an exponential rise in cases. “We may have missed the opportunity of a South Korean-type response, to test and isolate,” he said.
Among the many things South Korea did to control its COVID-19 epidemic was to ask its private sector to begin developing diagnostic kits within a week of its first case. By March 20, thus, it was able to test 316,664 people for COVID-19. Other experts have credited this strategy with bringing the total number of new cases in South Korea down to a low of 64 new cases on March 21, down from a peak of 909 in February. In comparison, private sector labs in India will only begin participating in the testing exercise this week, while the total number of tests so far — including travel-related and community cases – stands at a little over 27,000.
There have already been several indications that the COVID-19 outbreak in India may be shifting gears. While it took nearly one-and-half months for India to cross 100 cases after the first patient was identified on January 31, it took only four days for the case count to double from 288 to 599 last week. As of March 27, India had confirmed a total of 745 cases.
On March 24, Prime Minister Narendra Modi announced a three-week countrywide lockdown, in which all interstate airplane, train and bus transport would be halted, and government offices shut. This lockdown may slow the addition of new confirmed cases, but is unlikely to stall the epidemic altogether, experts said. The epidemic, according to John, “is like a massive train: the momentum is difficult to slow down once community transmission sets in.”
Few estimates exist of how many Indians could eventually get COVID-19. One of them, from a collaboration of epidemiologists called the COV-IND-19 Study Group, used Indian data until March 16 to project that the country could have at least 58,643 people with COVID-19 by May 15. Yet another estimate from the Washington-based Centre for Disease Dynamics Economics and Policy calculated that without any interventions, India would have 100 million active cases at the peak of the epidemic in April-May 2020, of which 2-4 million would need to be hospitalised. (However, its methods are not yet known.)
The lockdown could make a dent in such scenarios but only if it is accompanied by numerous other measures: more community testing, with health officials and researchers rapidly sharing their results, doctors accessing sufficient quantities of personal protective equipment, and interventions that ensure food and essential supplies aren’t impacted by the lockdown, Bhramar Mukherjee, an epidemiologist at the University of Michigan and a member of the COV-IND-19 Study Group, said.
The widespread agreement on the need for aggressive community testing raises the following question: how many more tests has ICMR conducted among community cases since the last time it announced these numbers (826 on March 19)? The Economic Times reported earlier this week that the Union health ministry’s joint secretary said 2,000 tests had been conducted, although The Wire Science couldn’t independently confirm this. Prabhdeep Kaur, an ICMR scientist, told The Wire Science it has been difficult to get this data since individual states have been coordinating the efforts.
Note: This article was edited at 5 pm on March 28, 2020, to add a note about CDDEP’s methods.
Priyanka Pulla is a science writer.
The reporting for this story was funded by a public health journalism grant to Priyanka Pulla from The Thakur Family Foundation.