Prime Minister Narendra Modi had on March 24 announced a 21-day nationwide lockdown, the idea behind which was to restrict the movement of people across the country, in a preventive measure against the spread of COVID-19.
Starting with a ‘test lockdown’ — a 14-hour voluntary janata (public) curfew on March 22 — it has been followed by a series of regulations in the country’s novel coronavirus-affected regions. When the lockdown was imposed, the total number of confirmed COVID-19 cases in India stood at 512.
Towards the end of the first lockdown phase, many states had recommended the Prime Minister to extend the lockdown. On April 14, Modi had extended the lockdown till May 3. However, he announced an option of a conditional relaxation after April 20 for regions where the spread had been contained or was minimal.
On May 1, the Union government extended the nationwide lockdown by two more weeks, until May 17. It divided the districts into three zones based on the spread of the virus — green, red and orange — with relaxations being applied accordingly.
On May 17, the lockdown was further extended till May 31. However, this time, the announcement was made by the National Disaster Management Authority.
The government has gone on record to state that the lockdowns had slowed the growth rate of the disease. In the first phase, the rate of doubling was reduced to six days and by the second, it was eight. This had gone up to 11 and 13 days during the subsequent lockdowns.
What was the purpose of the first lockdown?
The purpose of the lockdown was to act like a speed-breaker to the spread of the disease and ‘buy time’ during this period. Globally, it was seen as an opportunity to ramp up medical infrastructure and the scale of testing. No epidemiological model says that the virus would completely vanish due to the shutdown.
The attempt during a shutdown is merely to ‘flatten the curve’ so that there is no stress on the hospital infrastructure. However, if we look at the epi-curve for India, we see that the number of cases only went up after lockdown was imposed and the curve certainly does not appear to flatten; it has rather been grown exponentially and continues to do so.
What is lockdown efficiency?
In order to view the usefulness of the lockdown, we propose an analytical framework. We examine the efficiency of lockdown both state-wise as well as phase-wise.
It is now obvious that cases are rising. However, jumping to a conclusion based on just the number of cases would be wrong, just like a state showing limited spike in cases does not imply that it’s all hunky-dory there.
On all television channels and the social media, the daily talk is all about “how many new confirmed cases are there.” While it is true that it is something to track, in reality, it is not that important since we have not tested everybody in the state and in fact, no state can. Hence, the numbers that get reported are, in reality, a function of how many tests we conduct. States like Maharashtra, Delhi and Tamil Nadu that test more tend to report more confirmed cases every day.
Several other states, like Telangana, are choosing not to conduct extensive testing, which means that they don’t show any great increase in new cases either.
If it’s not a confirmed case, it does not reflect in the total count. Therefore, to compare states, it is better if we ignore the gross testing numbers (in some cases people are tested more than once), and rather look at the positivity rate, and follow a simple algorithm.
How many people did we test?
Among the people tested, how many have tested positive for COVID-19?
If there’s a decline in the positivity rate, it may be concluded that the infection spread is slowing down. The decline would be an indication that infection rate is lowering across states. That number in India has varied between one and 15 per cent.
The positivity rate depends on the number of tests conducted, which again depends on the testing criteria. “Who is being tested?” If we are only testing those who are likely to have a high ‘prior-probability’ of the infection, such as someone arriving from abroad or from a containment zone with fever, then the positivity rate could be high.
(Prior probability, in Bayesian statistical inference, is the probability of an event before new data is collected. This is the best rational assessment of the probability of an outcome based on the current knowledge before an experiment is performed.)
But if we see the positivity rate is increasing along with the absolute number of positive cases, this would indicate a high degree of transmission of the infection. Most states in India are currently on an exponential upward trend, and the numbers are likely to go up further as more migrants are being tested, as the lockdown is being further relaxed, and as people start commuting and travelling.
The need for randomised testing
Global examples show that if both positivity rate and the number of confirmed cases are increasing, one must look at relaxing the testing criteria; open options for “test on demand” i.e. anybody walks into a testing lab and get themselves tested. One can also look at randomised antibody tests being conducted at a community level to understand the “sero-prevalence”.
Sero-prevalence is the number of persons in a population who test positive for a specific disease, based on serology (blood serum) specimens; it is often presented as a percentage of the total specimens tested or as a proportion per 1,00,000 persons tested. Although positively identifying the occurrence of diseases is usually based upon the presence of antibodies for that disease (especially with viral infections such as COVID-19, Herpes Simplex, HIV), care, however, must be taken to assure that this number is not significant if the specificity of the antibody is low.
Considering that a large number of people could be asymptomatic carriers, a ‘sero-surveillance’ will help us in understanding the trend of the infection. It is likely that as more and more asymptomatic transmission happens, the virulence of the infection will also decrease over time.
Sero-surveillance provides estimates of antibody levels against vaccine preventable diseases (VPDs), and is considered the gold standard for measuring population immunity due to past infection or vaccination. It can predict the herd immunity in a population for an infectious disease. Sero-surveillance is an important component of disease surveillance and complements notification, hospitalisation, mortality and immunisation coverage data.
National sero-surveillance programmes are well established in many countries across the globe, and has now been advocated by the Indian Council of Medical Research (ICMR) as well.
— ANI (@ANI) May 30, 2020
We define lockdown efficiency as the ability of a state to reduce its daily positivity rate (in percentage) over a lockdown period. We examine each lockdown and rank the states on the basis as mentioned above. We also look at case fatality rate, defined as ratio of number of deaths to the number of confirmed cases.
In this analysis, we cover the impact of Non-Pharmacological Interventions (NPI) and what we can learn from it. It is now clear that the pandemic is not going away anytime soon, and we need to be more careful as India’s epi-curve peaks.
The Centre has announced the fifth phase of the lockdown, or what they are calling the ‘Unlock 1.0’, i.e. to unlock all the economic activities in a phased manner. This includes the opening of restaurants and malls on June 8. Even though lockdown 5.0 will last till June 30, it is meant to be only in the containment zones. Also the night curfew hours for areas outside these zones have been revised to 9 pm to 5 am, from the earlier 7 pm to 7 am.
When we look into the details of the MHA circular of ‘Unlock 1.0’, it is clear that this will play out differently in different states as the individual state governments have been given the freedom to prohibit certain activities which they deem fit. The R0 for different states ahead of the lockdown has been illustrated in the Choropleth map.
Reproduction number or R0
R0, pronounced as ‘R nought’, is a mathematical term indicating how contagious and infectious a disease spread can be. It is also referred to as the reproduction number. As an infection is transmitted to new people, it reproduces itself. R0 tells us the mean number of people who can contract a contagious disease from one infected person. It specifically applies to a previously infection-free population.
An R0 below 1 suggests that the number of cases is shrinking, possibly allowing the states to open up. An R0 above 1 indicates that the number of cases is growing, perhaps necessitating renewed lockdowns or other containment measures.
But R0 is a complex metric and not that easy to model either. The term is borrowed from the study of demographics, where it is used to describe the reproductive birth rates. R0 for COVID-19 has not been easy to arrive at as these figures are changing constantly.
But in general, studies now estimate that the pathogen that causes COVID-19 has an R0 of 2 to 2.5. That’s significantly higher than the annual flu and is in the lower-end ranges for SARS, another kind of coronavirus. One can, therefore, conclude that states having R0 lower than this range, are in a slightly better condition than the ones having R0 above the consensus estimate range.
It’s important to note for Unlock 1 that if R0 becomes less than 1, the pandemic extinguishes. Therefore, the policy goal for states should be to reduce R0. One can reduce R0 by isolating people who are infected or by quarantining contacts of infected. Another way to reduce R0 is by social distancing, i.e. maintaining at least one-metre distance from others.
Pre-print research papers indicate that coronavirus can travel only about a meter in the air. So, social distancing reduces the prior-probability of infection. This R0 value, however, is only a modeled estimate and can be affected by other externalities.
For this analysis of R0, we have taken data from the IIT Delhi’s PRACRITI dashboard.