Unlocking the 40-day national lockdown in India: There is no magic key

This article completes a three part series on the novel corona virus outbreak in India. You can read the first part (pre-lockdown) here and the second part (studying lockdown) here. You can also access a related pre-print here. All data source, code and interactive plots are available at covind19.org.

India, a democracy of 1.34 billion people has been aggressive in adopting strong public health intervention measures including sealing its borders, introducing mandatory social distancing including a lockdown early in the course of the COVID-19 epidemic. India crossed 100 cases on March 14 (2 deaths) and crossed 10,000 cases on April 14 (358 deaths). The rates at which the number of cases and deaths have doubled in India (Figures 1a and 1b respectively) are slower than other COVID-19 affected countries at the same stage of the epidemic. There have been several speculative hypotheses around this observation of slower spread, including genetics, immunity, temperature, humidity, BCG vaccinations, use of antimalarials, younger population and possibly their multifactorial interactions. Or, could it really be the effect of early prevention and lockdown?

Figure 1a. Doubling graph for case-counts in India and other countries using data till April 21. Interactive plots are available at covind19.org.

Since COVID-19 related case counts are a mystery in themselves [FiveThirtyEight, The Guardian] there are many unknowns here for the government and the people to feel confident that the above explanations are indeed true and widespread community transmission has really not happened yet [Quartz, Hindustan Times]. Regardless of the variation in the projected counts from any epidemiologic model (e.g., Ray et al. 2020, Institute for Health Metrics and Evaluation, Imperial College London, Nature, Singh and Adhikari 2020, Center for Disease Dynamics, Economics & Policy, Tang, Li & Li 2020, Islam et al. 2020, Zhou et al. 2020, Zhan et al. 2020), further amplified by their imperfect data inputs, there is absolutely no doubt that there is a huge cost for waiting to act in a pandemic [New York Times]. On April 14, Prime Minister Narendra Modi extended the 21-day lockdown to May 3rd [BBC].

Figure 1b. Doubling graph for death counts in India and other countries using data till April 21. Interactive plots are available at covind19.org.

It is clear that such draconian national lockdowns cannot go on for months for a country like India, or for any nation. We need to think about pragmatic strategies to “unlock” the lockdown and be prepared for reopening the country. Unfortunately, unlocking the lockdown is not just turning a magic key. A well planned multi-phase exit strategy will constitute of measures of suppression, a robust infrastructure for testing, disease surveillance, contact tracing and isolation strategies, healthcare support and capacity for treatment of COVID-19 patients. The three critical T’s from a public health point of view are: Test, Trace and Treat. Accompanied with the three T’s is the careful fourth T of titrating resources to jointly achieve the public health goals of mitigation and the economic goals of reactivation of the workforce and recovery.

1. Was the extension from 21 to 40 days justified from a public health perspective?

Yes! We carry out a natural experiment via simulation. Let us pretend we had data and information that the government had in hand from March 1-April 14 in terms of the number of reported case-counts and death counts. We build models [Ray et al. 2020, Wang et al. 2020] for predicted case-counts and death counts under five different lengths of the lockdown 21-, 28-, 40-, 50- and 60-days. We clearly see from Figure 2a that a lockdown of 40, 50 and 60 days have substantially stronger efficacy than a 21 or 28-day lockdown. For example, by May 1, May 15 and June 15 a 40-day versus 21-day lockdown would prevent approximately 4500, 113 thousand, and 4.2 million cases and 90, 2300 and 84 thousand deaths, respectively (assuming a 2% case-fatality rate). In terms of daily case counts, seen in Figure 2b, we estimate that there would be 23 thousand new cases under a 21-day lockdown compared to 2100 new cases under a 40-day lockdown on May 15, a difference of 21,000 new cases. The upper credible interval (CI) for these numbers are quite high [covind19.org].

Figure 2a. Cumulative graph for forecasting models assuming a quick adherence under 21-, 28-, 40-, 50- and 60-day lockdown scenarios using observed data through April 14

Balancing between collateral damage from a shutdown and public health considerations, an extension to 40 days appear to be a rational choice. Coincidentally, the word quarantine came from the Italian word quarantina which in the 17-th century meant 40 days.

Figure 2b. Incidence graph for forecasting models assuming a quick adherence under 21-, 28-, 40-, 50- and 60-day lockdown scenarios using observed data through April 14

2. Where would case-counts stand after a 40-day lockdown?

Depends on how we behave after the lockdown. After 40 days of isolation if we are overzealous in re-connecting, or if there are super-spreader events, the lockdown will not have its maximal chance of helping us. We again used our computer as a field of natural experiments and generated predictions based on what we know as of today (April 21), assuming three different types of human behavior after a 40-day lockdown on May 3. These different behavior types are reflected through the change in the contact transmission probabilities in our models and the changes to the basic reproduction number R0. We estimated the posterior mean of R0 in our Bayesian eSIR model to be 2.26 [95% Credible Interval (CI): (1.06, 3.94)].

We see that under a quick compliance to the lockdown and a cautious return we will stand at roughly 60 thousand [upper 95% CI: 395 thousand] and 307 thousand [upper 95% CI: 411 thousand] cases on May 15 and June 1, respectively. On the other hand, an overzealous return (i.e., the “Normal (pre-intervention)” curve in Figure 3) predicts 62 thousand [upper 95% CI: 1.8 million], and 2.3 million [upper 95% CI: 13 million] cases on those dates, respectively.

Figure 3. Predicted cumulative case counts in India per 100,000 people from April 30 to June 15 from the eSIR model are shown, based on observed data until April 21

3. How many COVID-19 cases would the 40-day lockdown prevent? How many lives saved?

21-day lockdown vs only social distancing: Comparing to the world where there was no lockdown and only perpetual social distancing, based on data up to March 25, we would have predicted 2 million cases and 40 thousand deaths on April 15. Instead we have observed 12,370 total cases and 422 deaths on April 15 under lockdown since March 25 (according to covid19india.org through April 15).

40-day lockdown vs only social distancing: By May 3, the projections with and without a 40-day lockdown are 37 thousand and 23 million cases showing a difference of 23 million cases. By June 15, the projections with and without lockdown are 2.5 million and 116 million cases, representing a difference of 113 million cases. The lockdown has possibly saved millions of cases and thousands of lives. It will be hard to know definitively whether the lockdown was an overreaction as we will never get to observe the counterfactual world.

Table 1. Predicted case counts from Bayes eSIR model with [upper 95% Credible Interval]

4. Is India testing enough?

Not yet! We used data from covid19india.org for India and Our World in Data for other countries to compare testing strategies across countries in the world having 1000 or more cases as of April 19. The scatter plot (Figure 4) shows the cumulative cases versus tests per million, and India is low on both axes with 297 tests done per million and 13 positives per million. Compare that to two other countries that have been exceptional and unique with their testing strategies: Iceland has 122,199 tests/million and 4,990 positives/million, and South Korea has 10,834 tests/million and 207 positives/million [Asia Pacific Foundation of Canada, World Economic Forum].

Figure 4. Cumulative number of cases per million versus tests per million until April 19. Linear regression model: Cumulative confirmed cases per million ~ cumulative tests per million. Fitting the model on data from all 53 countries, the adjusted R-squared is 0.46, with the fitted regression line in blue and 95% confidence interval in the yellow area. Excluding Iceland, the adjusted R-squared is 0.39, with the fitted regression line in red and 95% confidence interval in the grey area. Interactive plots are available at covind19.org.

Coverage of testing: We further investigated the coverage of testing for each of the above 53 countries until April 19, 2020 (Figure 5). Since units of tests differ across countries and some are not reported, the percentage of tested population is approximated by dividing the number of tests by the size of the population. In Iceland, 12.2% of total population have been tested, which is the highest among 53 countries, and no other country has tested more than 6% of total population. Though the United States has done about 3.8 million tests and far more in absolute numbers than any other country, only 1.2% of the population have been tested. The percentages of tested population in the United Kingdom and South Korea are 0.6% and 1.1%, respectively. Less than 0.1% of the total population have been tested in India (0.04% to be precise). In order to get to 1% of the population tested, India needs to be at 13.5 million tests of which approximately 550,000 has been done till date.

Figure 5. Total tests and percentage of tested population across countries until April 19, 2020. The percentage of tested population is approximated by the number of tests divided by the total population and is presented on top of the bar. Except India, the starting date for the above countries in our analysis is when there were over 100 tests. India starts on March 13, 2020 when the number of tests began to be reported consistently.

Variation of rate of positive tests over time: We examined the number of daily tests and positives for some countries over the course of the pandemic, namely, United States, United Kingdom, Iceland, South Korea, Turkey and India, to demonstrate the variation of test positive rate over time (Figure 6). For the United States and United Kingdom, more positive cases are detected as the number of tests increases and infection rates increase. The average of the daily positive rates (daily positive rate defined as the number of positives divided by the number of tests each day) in the United States is 0.166 (range: 0.054–0.28) and in the United Kingdom is 0.141 (range: 0–0.682). One can see in Figure 6 that there is a strong increasing trend in the plots for the fraction of positive tests during the growth of the pandemic as the testing strategies allowed mostly testing symptomatic patients. Thus, the fraction positives increase over time and then decrease. However, for India we do not see such an obviously increasing trend of detected positive cases yet. Even though the number of tests has increased in the last 10 days, the number of daily discovered positive cases still remain stable. During the past week (April 13–19, 2020), the average daily positive cases is 1329 (range: 966–2,267) out of an average of 29,405 (range: 21,806–37,000) daily tests. Overall, from March 13, 2020 to April 19, 2020, India has been seeing a low average positive rate (0.041) and a narrow range (0.011–0.071) which is encouraging.

Figure 6. Number of daily confirmed positive tests and total tests reported for selected countries

5. What are sensible strategies for future testing?

There are three primary goals of a testing strategy (1) estimating the prevalence in the asymptomatic and in the general population (using both RT-PCR and serology tests) (2) for identifying hotspot areas that are at risk of crossing the threshold of critical number of cases/infection rate so that the explosive growth of infection can be contained (using RT-PCR tests) (3) Use of serology tests looking for the presence of antibodies to know how many people were indeed infected. Currently, RT-PCR tests are only being offered to the patients who are clearly symptomatic or have come into direct contact with COVID19 positive patients in India [Indian Council of Medical Research COVID-19 Testing Strategy].

Some guiding principles for future RT-PCR Diagnostic testing strategies

  • Regularly offer a low number of tests to a random sample of subjects across the country, on some days increase the number of tests
  • Offer more tests in high contact, high density areas
  • Trace and test contacts: for rapid case-identification and contact tracing, managing the regional contact network and mobility data obtained through cell phone may be helpful to prioritize testing
  • Cluster testing (neighborhood, community dwellings, places of religious gathering) should be considered for tracking super spreading events.
  • Testing those that are at high risk of exposure (delivery personnel, store personnel, frontline health workers, essential workers)
  • All states need to ramp up testing and release credible data, there is tremendous variation even across populated states in this domain
  • Set up proxy alerts: Syndromic surveillance in health systems, survey of symptoms (different modalities: by phone, door to door, facebook), conduct temperature scans, analyze social media data for cues regarding covid-19 ontology.
  • Risk stratification based on thorough analysis of current data in terms of COVID outcomes (asymptomatic, mild symptoms, hospitalization, critical care, intubation) and prioritize and ration tests for the high-risk group [Richardson et al, JAMA].
  • Create risk scores of exposures for different labor sectors based on occupational exposure and distribute tests in high exposure groups including migrant workers in a chartered manner

Use of Antibody Tests: The antibody test is not useful in real-time identification of COVID-19 positive cases [NPR]. Note that there is also no consensus that an anti-body test proves immunity [Abbasi 2020, ABC News]. It is well documented that a big portion of the COVID-19 infected population is asymptomatic (Science Alert, Washington Post, Centre for Evidence-Based Medicine). The antibody test helps identify the past silent infections. This will help us to identify the true proportion of the population actually infected by the virus, symptomatic or asymptomatic. The antibody test can give us a good understanding of what proportion of the pediatric population was infected by the virus as children may be largely asymptomatic.

India is adopting a new testing strategy with the rapid antibody test for the hotspot areas for symptomatic cases [Times of India]. If the antibody test turns positive on the symptomatic cases, the patient will be admitted to the isolation ward of the hospital. If the test turns negative on the symptomatic cases, then the patient will be quarantined, and the RT-PCR test will be done on these cases. Such a testing strategy is cost-effective, as not enough RT-PCR tests are available. Once the patient is in the isolation ward, they will get discharged only if their RT-PCR test becomes negative. Serology tests and daily temperature reporting may need to be mandatory requirements for return to work in some sectors or for attending schools and colleges.

As one is thinking about these tests, sensitivity and specificity of the tests should be taken into account for both RT-PCR and antibody tests (RT-PCR sensitivity: ~70%, ~70%, 60%; antibody test sensitivity: ~98%; ICMR asks states not to use rapid antibody tests). Multiple tests per individual may be needed to confirm infection status and infection-free status.

6. What are some strategies as and when we reopen?

The coronavirus will not disappear with the lockdown. Cases will keep appearing and will need to be contained. The spread of the coronavirus can surge again if there are no community mitigation strategies in place after the lockdown is lifted. The question we should focus on is not what case-counts are needed for re-opening but do we feel we have the infrastructure to manage the epidemic and contain the new and prevalent cases going forward? What is a rational exit strategy? To answer this question, we first take a look at possible strategies for sub-optimal mitigation, and then review the phased re-opening plan by the United States. We emphasize that India needs to focus at a state level when considering modulated re-opening strategies, and review Kerala’s success story.

Some authors have argued that herd immunity may work for youthful India. As per WHO, safely lifting restrictions requires ensuring that the transmission of the coronavirus is under control, public health systems are in place to “detect, test, isolate and treat every case and trace every contact”, new cases from outside can be managed, hotspot risks are minimized, and people in the community are aware and engaged in this public health response [NPR]. Strategies for sub-optimal mitigation include limited travels, no large gathering, getting the workforce back in phases, teleworking wherever possible, and strictly following recommended public health/hygiene guidelines like hand washing, social distancing, and nose/mouth covering when out in public.

In the US, a phased re-opening has been suggested in a plan by the Centers for Disease Control and Prevention and the Federal Emergency Management Agency. The plan provides guidance on how the stringent coronavirus measures in US may be lifted in three phases beginning in May. The type of mitigation strategy will vary by state and will depend on community readiness — as indicated by community transmission of the coronavirus and healthcare capacity. For all communities, the disease and public health indicators will be monitored closely with a focus on protecting the most vulnerable until a vaccine is available or the spread ends, and adjustments to mitigation measures are to be made as needed.

In India, it is important to recognize that readiness to re-opening will vary at the local and the state levels. The cumulative case- and death-counts across the states are substantially different. While Maharashtra is seeing an exponential growth in the death-counts with 200 deaths as of April 19, Kerala has seen only 2 deaths (Figure 7a and 7b) and has the highest percentage of recovery of COVID-19 positive patients although both states had similarities in the early phase of the outbreak [Economic Times]. Kerala’s success is attributed to its aggressive testing and contact tracing, longer quarantine periods, building of shelters for migrant workers stranded by the national lockdown that was implemented with only a few hours’ notice, and distribution of cooked meals to the needy [Washington Post]. Each state will possibly resume normal activities at their own pace after ensuring the six criteria listed by WHO for safe mitigation. Communities that have not seen significant outbreak may be re-opened soon. Communities that were hotspots formerly must be re-opened in a controlled manner. On the other hand, communities that are currently seeing hotspots of infection may need to continue stringent lockdown measures for a longer period while actively monitoring disease dynamics and healthcare capacity. Consequently, state-to-state mobility may still be restricted.

Figure 7a. Cumulative COVID-19 case count by state/union territory as of April 22

While we are focusing on reducing the impact of an infectious disease and strengthening our healthcare capacity, we need to keep in mind that it is equally important to reinforce social-safety nets on a massive scale [Ideas for India]. With 22% of the population living below the poverty line, and 90+% of India’s workforce employed in the informal sector, a sizable portion of the Indian population is experiencing loss of income and savings, and will soon experience dire poverty and starvation due to the extended nationwide lockdown. This may result in defiance of public health measures and can negate the benefit of lockdown. Several economists, including Nobel laureates Amartya Sen and Abhijit Banerjee, have outlined economic policies and measures that the government can take to provide social security to the needy [Ideas for India, The Indian Express].

Figure 7b. Cumulative COVID-19 death count by state/union territory as of April 22

Final Message

There has been constant tension and competing interest between economics and public health in the debates and politics around this crisis. However, filtering out the partisan divide and political agenda, the ultimate global objective function that we are trying to minimize is human suffering and loss. This may imply focusing more on mitigation at one time, and focusing more on restarting the workforce at another. We may have to wear masks, maintain social distancing, avoid large gatherings, limit travel and mobility, work of from whenever possible for a foreseeable future. We know a lot more about the virus now than two months ago, scientists are working round the clock to learn more. We have to wait till we see steady decline in the number of cases and get closer to the tail of the incidence curve and have enough tests to fully re-open. We need to be prepared to deal with the number of new cases and are able to curb any new surges. Otherwise we will undo all the work the people and government of India have done in beating the initial dire model projections through their committed compliance to social distancing.

The entire world is getting used to the notion of the new normal where we need to be prepared for future partial/complete lockdowns and a pause-restart-drive mode of living, learning and leading until there is a vaccine or FDA approved treatment. The COVID-19 problem has underscored the structural inequities and disparities that exist in our society. Be it economics or health, the distribution of loss and suffering has not been uniform across all strata of society. From Detroit to Dharavi, the narrative is the same. The virus curve will peak and flatten but humanity will be at its apex if we could hold on to the lessons we have learned during the pandemic. We remain hopeful that with strong partnership between people, government, scientists and policymakers, India’s containment of the COVID-19 epidemic in a low-resource setting will remain exemplar for the rest of the word.

The COV-IND-19 Study Group is comprised of an interdisciplinary group of scholars and data scientists. The following members contributed to this piece:

  • Maxwell Salvatore — Departments of Biostatistics and Epidemiology and Center for Precision Health Data Science, University of Michigan
  • Debashree Ray — Departments of Epidemiology and Biostatistics, Johns Hopkins University
  • Jiacong Du — Departments of Biostatistics and Epidemiology and Center for Precision Health Data Science, University of Michigan
  • Lili Wang — Department of Biostatistics, University of Michigan
  • Sourish Das — Chennai Mathematical Institute
  • Michael Kleinsasser — Departments of Biostatistics and Health Behavior and Health Education, University of Michigan
  • Alexander Rix — Department of Biostatistics and Center for Precision Health Data Science, University of Michigan
  • Daniel Barker — Department of Biostatistics, University of Michigan
  • Bhramar Mukherjee — Departments of Biostatistics and Epidemiology and Center for Precision Health Data Science, University of Michigan

Contact Bhramar Mukherjee (bhramar@umich.edu) with questions and inquiries.

An interdisciplinary group of scholars and data scientists who use of data and modeling to generate timely reports and recommendations about COVID-19 in India

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store