Fake blood is seen in test tubes labelled COVID-19 in this illustration. Photo: Reuters/Dado Ruvic.
A new model of the COVID-19 pandemic by researchers at the Armed Forces Medical College (AFMC), Pune, and INHS Asvini, India’s oldest naval hospital, in Mumbai, has concluded that if India is able to isolate and successfully quarantine at least 50% of all people infected by the new coronavirus today, the case growth rate would peak sometime in April to 7,000-9,000 new cases per day and fall off rapidly by up to 90% after.
This study is the latest in a series that have estimated India’s total case load over time due to the spread of the new coronavirus, and almost all of which one thing in common: the outcome that if India does nothing, it could have millions of active cases by May. However, India has done something already – in the form of the national lockdown instituted from March 25 to April 14, and there is some hope that restrictions will ease soon if only for pressing socio-economic reasons. This said, there is little clarity on what could come after in terms of the epidemic itself, and this is where disease modelling experts can fill the gaps.
The current study in particular, a pre-proof copy of which was published online on April 2, is one of the more optimistic ones to have been published because, if nothing else, of its favourable view of the national lockdown. It finds that the sooner India improves from finding and quarantining 1% of all those infected to 50% of all those infected, the drastically better the outcomes will be for the whole population. For example, if this improvement happens over a period of 7 days, the study estimates India will a maximum of 70,000 total active cases; if over 14 days, then a little over 82,000 total active cases; and if over 21 days, then around 95,000 total active cases.
These predictions are akin to those made by the Indian Council of Medical Research (ICMR). Gautam Menon, a professor of biology and physics at Ashoka University, Haryana, agreed as he called the study “very standard, very like the ICMR model”. Menon also raised questions about the authors claiming their study is based on a stochastic model – which implies at least one of the model’s inputs will vary randomly – “but the equations are deterministic”. Finally, he noted that unlike many other more detailed models, the study didn’t provide an age-wise breakdown of the case and mortality loads, and didn’t address the problem of asymptomatic infections.
Indeed, and by extension, the study also does not seem to question the government’s lack of aggressive testing for COVID-19 patients. Without such tests, India won’t be able to identify everyone, and certainly not isolate half or more as the study recommends. As a result, a substantial number of people could be left behind in the population who still harbour the infection and could potentially lead to new case clusters after the lockdown has ended.
Second, and as an extension, if the quarantining strategy is not completely effective vis-à-vis its goal of eliminating all potential sources of new infections from the population, the virus could bounce back once the lockdown is lifted. Multiple studies published in March 2020, based on studies of patients in China, Japan, Germany and those onboard the Diamond Princess cruise ship, already support one hypothesis that around 30% of people who contract the new coronavirus may never develop any outward signs of infection yet still be able to spread it to others. So tests that overlook people with mild symptoms could in fact be overlooking many of the people still at risk of infecting others.
A previous modelling study published as a preprint paper 1, by researchers affiliated with Cambridge University and the Institute of Mathematical Sciences, Chennai, also found that a 21-day lockdown would only delay, but not eliminate, an exponential growth of the case load. Instead, the authors recommend India implement three lockdowns of 21, 28 and 18 days, with a few days in between each, or one long lockdown of 49 days – to completely smother the growth curve.
However, Suvrat Raju, a physicist at the International Centre for Theoretical Sciences, Bangalore, sharply criticised this paper in a (public) Facebook post on April 2, in which he called the model’s conclusions “absurd” and said their inputs were compromised by the fact that they were based on slightly outdated Indian government data, and therefore quite unreliable considering the government’s staunch reluctance to test more. That is, in the absence of more tests, there is presumed to be a big difference between the number of people we know are infected and the number of people who are actually infected.
As some reporters wrote over at FiveThirtyEight with reference to the wide variation in models’ predictions: “To determine the fatality rate, you have to divide the number of people who have died from the disease by the number of people infected with the disease. In this case, we don’t really have a reliable count for the number of people infected – so, to put it mathematically, we don’t know the denominator.”
Ultimately, and irrespective of the legitimacy of these models’ predictions, public health experts – together with experts in many allied fields of study – have been unrelenting with their demand that the government test more people. Even the AFMC and INHS Asvini study, without directly referring to this lacuna in India’s strategy to contain the local coronavirus epidemic, invokes this shortcoming.
In a slight departure from the standard SEIR (‘susceptible’, ‘exposed’, ‘infected’ and ‘recovered’) model that epidemiologists use to predict the spread of infectious diseases, the authors of the new study insert a ‘quarantined’ category between ‘I’ and ‘R’, and note that it refers to an individual who has been isolated but who, in reality, may not always be detectable.
* There’s a catch.
Therefore whose conclusions could potentially be more contested than if it was a peer-reviewed paper↩