India to see 2.87 lakh COVID-19 infections per day by winter, says study
India will see 2.87 lakh COVID-19 infections per day by the winter of 2020-21, a study by a team of the Sloan School of Management of the Massachusetts Institute of Technology (MIT) has said.
The study, authored by Hazhir Rahmandad, Tse Yang Lim, and John Sterman said the true magnitude of the pandemic was difficult to be ascertained from official data due to the following reasons: dramatic variation in testing rates across countries, differences in attribution of deaths to COVID-19, and significant false-negative rates.
“Effective response to the COVID-19 pandemic requires an understanding of its global magnitude and risks. Yet, more than 15 weeks after WHO declared a global pandemic, we do not know the true number of cases nor the infection fatality rate. The data we have are highly variable; as of late June 2020, countries have reported cumulative cases ranging between 0.18 and 2,850 per 1,00,000 and deaths to cases (approximate case fatality rates: CFR) ranging between 0 and 24 per cent. We do not fully know what explains such large variance,” the authors said.
Related News: India’s COVID-19 caseload increases to 7,42,417
The study also lists the top ten countries by projected daily infection rates at the end of winter 2020-21: India (287000 infections per day), the US (95.4), South Africa (20.6), Iran (17.0), Indonesia (13.2), UK (4.2), Nigeria (4.0), Turkey (4.0), France (3.3), and Germany (3.0).
The team used a dynamic epidemiological model for the study. The model integrated data on cases, deaths, and excess mortality to estimate asymptomatic transmission, disease acuity, hospitalization, and behavioural and policy responses across nations over time.
“Establishing such basic aspects has partly been slowed by a large asymptomatic fraction and similarity of COVID-19 symptoms with other common illnesses. Moreover, official case and fatality data are generated through testing,” the authors said.
What is CFR and IFR
The study team took into account the case fatality rate (CFR) and infection fatality rate (IFR) to arrive at their conclusions.
In epidemiology, CFR is the ratio of deaths from a certain disease compared to the total number of people diagnosed for a certain time prevalence. A CFR can be expressed as a percentage and represents a measure of disease severity. CFRs are most often used for diseases with discrete, limited-time courses, such as outbreaks of acute infections like the COVID-19. A CFR can only be considered final when all the cases have been resolved (either died or recovered). The preliminary CFR, for example, during the course of an outbreak with a high daily increase and long disease-resolution time, would be substantially lower than the final CFR.
The consensus right now is that the IFR or the ratio of deaths among all the infected individuals for this disease is in the range of 0.5 per cent (and below). This means, based on the data available now, one can conclude that this disease is probably not as dangerously infectious as it was thought out to be in March.
Related News: WHO takes note of ’emerging evidence’ on airborne spread of COVID-19
The IFR is closely related to the CFR, but unlike CFR, the IFR attempts to additionally account for all asymptomatic and undiagnosed infections.
The IFR differs from the CFR in that it aims to estimate the fatality rate in all those with infection: the detected disease (cases) and those with an undetected disease (asymptomatic and not tested group).
Individuals who are infected but always remain asymptomatic, are said to have subclinical infections. As a rule, the IFR will always be lower than the CFR as long as all deaths are accurately attributed to either the infected or the non-infected class.
As there are more tests and the hypothetical assumption is that the entire population is tested, then the CFR would tend towards IFR. But if that is not the case, there is still going to be these super-spreaders who goes about generously spreading the virus.
Prevalence multiple
The CFR-IFR ratio gives a ‘prevalence multiple’ and this is an estimate of the likelihood of infection in a given population. So, if we take the case of India, the CFR currently is about 3 per cent and the IFR, according to estimates, is about 0.26 per cent [median]. This is much lower than what it was originally thought out to be, then this prevalence multiple is 3/0.26 = 11.
This means, the number of infected in the population out there is: 11 (x) total reported cases (-) reported cases, that is 10x the number of reported cases. This is likely to be the true infection rate (TIR).
‘Wrong’ trend
The issue is that one could get fooled by the ‘national’ trend and end up wrongly prioritizing resources because each state might have different CFRs and hence different TIRs.
Assuming that our mortality rate reports are correct, this means that a state like Gujarat could have 20 times more infection than what the reported cases are; Maharashtra, Delhi about 15 times more; and Tamil Nadu, about four times more.
Simpson’s Paradox
Simpson’s Paradox occurs when a trend that appears in a dataset when separated (stratified) reverses when the data is aggregated. This is what we are seeing in India as we look at national-level data and the relatively small chunks of data the government has been communicating about a decrease in the overall trend.
Paradox in action
There also might be some truth in the rationalizations expressed by our health department, like, young people and many in search of livelihood are the ones who are more exposed now; old people are protected more; we have better treatment protocols; tests are high etc.
The aggregate data might look peaceful. But a zoom into the respective major cities or even districts is likely to throw up altogether different results, especially around the ‘hotspots.’
The large decline in COVID morbidity in states like Maharashtra, Tamil Nadu, Gujarat, etc is now influencing the overall national trend in India. But one must not be misled by this and assume that such a singular kind of ‘national trend’ can be applied locally as well. The indication is clear: one should not view this as one pandemic wave across the Indian subcontinent as a whole but, in fact, as different pandemic waves hitting different geographies at different times.
Stratified data
For a true understanding of how deadly the virus is in the Indian context, the ‘stratified data’ should be analyzed. This could say how readily the virus attack and kills ‘different groups’ of people. The risk of dying from COVID-19 can vary considerably depending on age, state of public health in states, socioeconomic status, and underlying health conditions. Also relevant is better data-driven policy decisions and high-quality surveys of stratified groups and histopathology of cases.