Post-lockdown strategy

Lockdown and after

Print edition : May 08, 2020

A civic worker sprays disinfectant on beds at a temporary hospital facility in Mumbai on April 10. Photo: Rajanish KaKade/AP

In India, decisions on strategies to contain COVID-19 are being made on the basis of unreliable data, but two things are clear: the number of infections is still on the rise and once the lockdown is lifted there will be a resurgence of cases. The government needs to come up with a calibrated approach to manage both in the post-lockdown period.

AT the end of the first week into the second phase of the country’s extended lockdown (April 21), the number of confirmed COVID-19 cases, the number of those recovered and the number of deaths stood at 18,601 (14,759 active), 3,252 and 590 respectively.

The rates of increase both in the number of confirmed cases and in the number of deaths (in percentage terms) have come down to single digits (Figures 1 and 2). The plots in Figure 2, which is on a logarithmic scale (with each unit on the y-axis increasing by a factor of 10), are linearly increasing, which is what they would look like if the number of cases (and correspondingly the number of deaths) are still on an exponentially increasing curve.

The two phases of lockdown have certainly achieved their objective of slowing down the transmission of the infection in the population. This, as mentioned above, is also evident from the fact that the rate of increase in the number of infections has been coming down since the lockdown began. In the second phase, in fact, it has come down from a 10-15 per cent rate of increase to under 10 per cent, a little flattening of the curve. That this should happen is intuitively obvious when physical distancing between people and person-to-person contacts are minimised; the chance of one individual passing the infection to another has been greatly reduced.

But the growth factor (the ratio of the changes in the number of confirmed cases between two consecutive days), that is, the slope of the curve, is still consistently above 1, which means that the curve of infections is still exponentially increasing, though a little slower than before the lockdown, and the country is still far from peaking and then declining. Only when the slope stays around 1 for some days can one say that the curve has hit the inflection point from where the slope will become less than 1; the peak number can be expected to be around twice the number of infections at the inflection point.

Doubling time

Another way of looking at the progression of the epidemic in India is the so-called doubling time: the number of days it takes for the number of infections on a given day to double itself. It is this metric that the government is using to show the efficacy of the lockdown. In its press update of April 20, the Health Ministry said: “The doubling rate of COVID-19 cases calculated using growth over the past seven days indicates that India’s doubling rate for the week before lockdown [March 18-24] was 3.4 and has improved to 7.5 as on 19th April, 2020 (for the last seven days).”

This is not surprising because, as pointed out above, if you cut off nearly all possibilities of transmission of infection by totally switching off social/societal interactions, this will naturally happen.

At the State level, however, there is a wide dispersion in doubling times as is evident from the Ministry’s April 19 data (see table). The Chennai-based National Institute of Epidemiology has on its website presented a nice way of looking at this with data for eight representative States: Andhra Pradesh, Delhi, Kerala, Maharashtra, Rajasthan, Tamil Nadu and Telangana (Figure 3; unfortunately, this graph has not been updated with latest data, and the plots are only up to a few days before April 19).

The growth has been plotted beginning with the (normalised) day “zero” for these States, which is when the number of confirmed cases crossed 50. For comparison, plots of growth with doubling times of three and five days have also been given in the same graph. As can be seen, except for Kerala (which is clearly an outlier in this set), although most of them were curving towards the five-day doubling curve, some slower than the others, there are problematic States such as Maharashtra and Delhi. When this plot is combined with the Ministry data, it is clear that if States such as Delhi, Rajasthan and Tamil Nadu have managed to improve their doubling times—that is, their plots have gone below the five-day doubling curve—this has happened only in the last few days.

The important question at this juncture, however, is, What is the planned strategy of exit on May 3 when the second phase of lockdown ends? But before one addresses that question, one has to keep in mind an important caveat while interpreting all the data above and in the accompanying graphs. The caveat is that all the official data about doubling times, and so on, should be taken with a huge dose of salt because it is well known that the surveillance and testing strategy in India has been grossly inadequate. According to model estimates (see separate story), the infection detection rate in India is only about 2 per cent, though it is improving slowly (Figure 4) with the adoption of the newly broadened criteria and the ongoing efforts at widespread deployment of rapid sero-testing kits (notwithstanding reported proneness to error).

So, what is the sanctity or even remote accuracy of the baseline caseload data from which doubling times are being calculated? For example, many of the cases that get picked up on a given date could actually not be new cases at all; they could be old (probably a few days into the incubation period or subclinical/mild and asymptomatic) cases that were not picked up earlier—because of the inefficient testing strategy and very low detection rate—and are now being detected either by chance or because of the gradually improving detection rate. Unfortunately, all policy decisions, including strategies such as lockdowns to contain the spread of infection, are based on such extremely unreliable and inaccurate data; thousands of undetected and mild/asymptomatic cases are probably spreading the infection without the virus carriers actually aware that they are infectious.

When Dr T. Jacob John, the well-known virologist formerly with the Christian Medical College (CMC), Vellore, was asked what he thought the impact of the lockdown was, he said in an email response: “Lockdown has two parameters. Lockdown will flatten the curve; that is for sure, knowing epidemiology. [But] first, what you call lockdown, was it truly lockdown or lockdown in name, too many leaks and not truly lockdown but partial lockdown. Second, how exactly can we measure flattening of the curve? I haven’t heard of an a priori method defined/designed to monitor flattening of the slope. So, we don’t have good evidence for either, not because they are ineffective, but we don’t have criteria to measure…. Our daily counts are not any sign of “flat curve” except in Kerala (well managed) and in Goa (well managed and some luck too, I suspect). Whatever happened till two weeks under lockdown, I will assign to infections occurring before lockdown; after that, infections [occurred] under lockdown. So, yes, some speed acceleration should be expected.”

Returning to the question about the post-May 3 strategy, extension of the lockdown is not a viable solution, according to Dr Jayaprakash Muliyil, an epidemiologist also from the CMC: “The virus is out there; you cannot stop it. Studies have shown that subclinical infections are as high as 50-80 per cent of the total infections, many of which go undetected. So, the question of containing the virus does not arise. You cannot keep the country under lockdown for a year or more till a vaccine is developed. The social and economic costs of that would be a disaster for the country.”

“The only solution to get rid of the virus,” Dr Muliyil said, “is by achieving a certain level of herd immunity. Roughly 60 per cent of the population needs to get infected to achieve that. This is what happened with the other pandemic, H1N1. When you lift the lockdown, there will be resurgence [of infections] because the virus is there and because of the lockdown a large number are still susceptible [to the virus]. The population will get infected.” 

When he was asked about the consequent mortality in the population that this would lead to, Dr Muliyil said: “Fortunately, the virus seems to affect only the elderly, above the age of 60, severely. So, you have to ensure that the elderly are well protected. You don’t have to cut them off entirely, and the burden of securing herd immunity should be taken by the young. It may be a tough ask, but there is no other alternative. We have a demographic advantage there. Only about 10-15 per cent are old requiring to be protected from the virus unlike in Europe. You let the young go about their normal work—but, of course, following all the safe health practices like social distancing, washing hands, etc.—they will get infected but most of them will come out of it unscathed. A small fraction of the young may require hospitalisation, and some requiring ICU [intensive care unit] facilities. Hopefully, the country is now prepared after the lockdown to extend proper health care to this younger lot who may get affected severely. With this strategy, herd immunity can be achieved.” Dr John echoed this view as well. “If life gets back to pre-lockdown situation,” he said, “I will give 4 weeks [for resurgence]; if ‘calibrated’, [it will be] slower. If all our seniors and ‘vulnerables’ are cocooned (reverse quarantined) for calibrated lift of lockdown, and if all wear mask in all situations of human interactions, even if physically distanced, then we can be a bit bolder to speed up lockdown-lifting. If we can be smart and innovative, I will begin serologic surveys for IgG antibody (not difficult to make and scale). Had we started in January, we would have that [antibody kits] in hand by end March. Every COVID case points to 4 infections without symptoms. So, we must have a huge number of immune persons. If they are so certified [by IgG tests], they can kick-start economic activities.”

In the context of achieving herd immunity in India, there is an eye-opening piece of new research work from the United States’ Centres for Disease Control and Prevention (CDC), which is yet to be published but is available on its website. By systematically and carefully re-examining the dynamics of the spread of infection in Wuhan in the early phase of the epidemic (January 15-30) and based on data in 140 confirmed case reports obtained from the China CDC, Steven Sanche of the Los Alamos Laboratory and others have re-evaluated the basic epidemic parameters, in particular the basic reproduction number (R, or “R-nought”), which gives the number of infections that a single index case can potentially lead to. It is thus a measure of the contagiousness of the disease. Chinese researchers had earlier estimated this parameter to have a value of 2.2 to 2.7, somewhat higher than the figure for influenza, which is about 1.3.

This new work, however, revises this number drastically. Assuming a serial interval (the time between onset of symptoms in the index case to the onset of symptoms in the infectee) of six to nine days, the new value for R that Sanche’s team obtains is 5.7, which means that the virus is highly contagious. “With a high growth rate of the outbreak, R is, in general, high, and the longer the latent and infectious periods, the higher the estimated R,” write the CDC authors. This indeed seems to be the case for the new virus causing COVID-19, which one can note empirically from the speed with which the infection spread around the world, within just four months. Its known longer incubation period also ties in with the higher value of R, as the CDC team noted.

Now a higher value of R has an immediate implication for both pharmaceutical and non-pharmaceutical intervention strategies. For example, as the paper points out, the threshold for vaccine efficacy and herd immunity for disease extinction is given as [1− 1/R]. For an R of 2.2-2.7, the earlier value, this threshold is only 55-63 per cent. But at 5.7 the threshold rises to 82 per cent. That is, for herd immunity to be achieved to stop transmission, over 82 per cent of the population has to be immunised, either by getting exposed to the virus or by vaccination. The question, therefore, is, what is the implication of this new R for countries, India in particular?

“This data on R cannot be immediately extrapolated to the Indian situation,” said Dr Muliyil. “We have a very complex situation in India. Firstly, R depends on population density as well. We have nearly 70 per cent rural and 30 per cent urban population and at the same time 80 per cent work in urban areas. There will be a wide variation in R. I guess, it could be more than 2.4-2.5 but would be less than this new number. In any case, going from 60 per cent getting infected to 80 per cent will happen very quickly since we have no control over the virus. It will happen in about 3 months [after the lockdown ends],” he added. “Higher the R, higher the herd immunity needed for good herd effect. [But] without a reliable IgG antibody survey in the community, we do not know the current ‘infection prevalence’,” Dr John pointed out. “Since we started late with infection, we would have been slow to infect people widely; however, we do not know what proportion is already infected. In any case, I allow two months for graph catching up with natural path kept ‘frozen’ by lockdown,” he added.

This article is closed for comments.
Please Email the Editor