We lost faith in what the Models were telling us and now the Covid 19 surge we feared is happening
A Comparative Analysis of Mumbai & Delhi
By: Brahmani Nutakki, Harshita Magroria, Sukhada Gole
This article documents how we discovered in November 2020 the conditions for a significant risk of a surge in infections in Mumbai and Delhi existed. We could not be sure about the timing becuase that depended on future unknowable efforts to contain the spread. Our worst fear was it might happen as soon as January 2021 following the relaxation of controls for the festive season.
The timeline in Exhibit 1 shows how our work on predictive modelling was anticipating what we are seeing now.
Timeline of our models
Then when the surge didn’t happen in January and there was widespread belief that Covid 19 had been controlled we began to wonder if our analysis might be wrong. Unfortunately it has transpired the surge we had expected is now happening. In hindsight it was inevitable unless India had somehow embedded very high levels pre-existing resistance to Covid 19. The delay seems to have happened partly because of an issue we discovered in our Delhi analysis: it is critical to track infections at a local level. If a large number of very different areas are added up it can mask a major outbreak and create a ‘muddle of averages’. As a society we clearly fell victim to this and we hope the measurement systems can be reformed so we do not repeat the same mistake again.
This is the story of how we discovered these insights and we hope by publishing it now it will help to stimulate a reform for how the pandemic impact is measured and managed.
We are a group of university students from Mumbai and Hyderabad involved in developing the COVID-19 localized models for L ward, Mumbai, and Delhi respectively under the guidance of The COVID-19 Localization Modeling Group, London. This article attempts to draw similarities and differences between Mumbai and Delhi National Capital Region (NCR) in terms of their population composition and living conditions. This comparative study helps us to consider the insights from the COVID-19 localized model for L ward, Mumbai thereby focusing on the future infection trends in Delhi.
Across the world, COVID-19 has impacted each location differently. Mumbai and Delhi are examples of places that on the surface might appear to be similar but because of several factors unique to each city, the COVID-19 pandemic has developed on completely different paths. At some level, you could say: Why dwell on this now? Isn’t it too late to do anything about what has already happened? Why not just set it aside, move on and look to the future?
The reality is we are nowhere near the end of COVID-19: there is evidence that in some slum locations a very high percentage of the population have been infected, although in non-slums the percentages remain low. This means until the people in the non-slums are infected or vaccinated they are at risk and with over 4000 mutations to date there is a risk that vaccines may not offer protection against some mutations¹. Studying the very short history of the pandemic in these two cities can give us insights into what has gone well and what could have been done better. But not in the sense that we look to judge or criticise but to learn about how to chart a better path forward.
The localized model build for Delhi involves a comprehensive analysis for both slums and non-slums whereas, the COVID-19 localized models for L ward Mumbai separately study the movement of the virus in slums and non-slums areas. Through this article, we try to compare Delhi with L ward list down policy suggestions, especially for Delhi which is following Mumbai and is on the verge of a massive surge.
Further, we hope to show how systematic transparent modelling of the COVID-19 crisis can be a widely useful tool in the hands of ordinary citizens like us to understand the infection trend in a localized area. The article aims to suggest area-specific and general measures for Delhi and Mumbai to devise policies for the effective management of the pandemic.
COVID-19 has become one of the worst pandemics to hit the human race, causing an economic and political catastrophe. It forced us to evaluate our existing practices and adopt new ones. With over 10 Million cases discovered (only a small fraction of the actual since most are asymptomatic and never detected), India is the second-worst affected country, Delhi and Mumbai continue to be the worst affected cities. The discovery of the new more infectious strains of COVID-19 in the United Kingdom and South Africa is a new and more serious challenge to our understanding of the virus and our ability to overcome it. The virus continues to be a conundrum with unanswered questions of what may happen next. How can we tackle the crisis and secure our future? On that note, “The COVID-19 localisation model” co-developed by The Covid 19 Localization Modelling Group to help those unfamiliar with pandemic modelling understand the local trajectory of the virus. To better understand the movement of the virus in Mumbai and Delhi we sought to compare their infection trend concerning COVID-19. The two cosmopolitans of India closely resemble each other in terms of their population density and simultaneous habitation of slums and non-slums.
The Municipal Corporation of Delhi (MCD) is divided between three zones (North, Eastern, South) whereas the Municipal Corporation for Mumbai district i.e., MCGM is one integrated body making space for better policy coordination, records management and tracking supply for both medicines and medical health professionals. MCGM faces minimal state government intervention in formulating and implementing policies and the latter acting as an advisory body. On the other hand for Delhi, the central and state government are important stakeholders in policy formulation generally creating hurdles in collective action. However, during the pandemic, this scenario seemed to be reversed.
Mumbai has recorded 11 thousand deaths till February while the COVID-19 death toll in Delhi crossed 10 thousand. With due aid from the central government, the state government of Delhi appropriated funds ensuring better medical facilities and higher tests/million portraying coordination between three-tier governments. The central paramilitary units provided their managerial expertise to Delhi which helped in enhancing bed capacities and medical staff availability. The Central government released funds for all the states under the State Disaster Risk Management Fund (SDRMF). The total fund of Rupees 11092 crore was released as the first instalment and Maharashtra received the largest chunk being Rupees 1611 crores as per the reports of the Union Ministry of Home Affairs². The fund allocation statement provides no mention of the total funds allocated to Delhi under SDRMF. Now, it is imperative to ask Why the Central government did not provide a statement about the total funds allocated to Delhi? With no adequate information about the total allocations made, it would not be easy to compare the policy decisions of Maharashtra and Delhi concerning COVID-19.
Secondly, it is vital to ask, are the authorities testing enough? What is the proportion of RT-PCR tests to the Rapid antigen tests? The tests per million in Delhi are higher compared to that of Mumbai, it should be noted that 77% of total tests conducted in Delhi were Rapid Antigen tests which are less accurate than the standard RT-PCR tests³. On the other hand, since the advent of the pandemic MCGM preferred RT-PCR tests which accounted for about 69% of the total tests in September 2020. Further, the authorities in Mumbai are gradually adopting antigen tests without phasing out the more reliable RT-PCR tests⁴.
Verifiable data is a cornerstone for developing models and analysing the upcoming challenges. To provide adequate information about the virus, MCGM developed a dedicated website (https://stopcoronavirus.mcgm.gov.in/) providing details about the total cases, deaths, tests and bed facilities available. Significantly, the Ward-wise bifurcation of total cases and deaths makes it easier for our localized model to trace the infection trends and come up with area-specific policies. Although, the NCT Delhi provides citizens and academicians with data on the total cases, deaths, bed facilities and tests through a customised website (https://delhifightscorona.in/) but it lacks adequate segregated data for sub-regions hindering the process of developing a localized model.
Analysis of Mumbai
Exhibit 2 shows the results of the COVID-19 model set up by the students of the University of Mumbai for L ward, Mumbai to understand the spread of the virus in the slums and non-slums with the help of a seroprevalence survey conducted in the mid of July 2020⁵. According to the serosurvey, the presence of some kind of antibodies in 57% of the slum population and 16% in the non-slums was noticed. The slum areas will most likely leave COVID-19 behind them in the near term due to the presence of antibodies in a significant count of population⁶. This prediction is supported by the recent reports showing the decrease in cases in Dharavi⁷. Reinfection risk remains but our analysis provides a lead indicator of the likely path for Mumbai. The non-slum areas might be at risk of a surge of infections 7x the level experienced in April 2020 as eary as January 2021 due to the reopening of the economy, easy mobility and intermixing of the population.
The non-slums in L ward, Mumbai will also witness a sudden surge in deaths caused due to the virus from January 2021 while the slums will continue to experience a stable death count throughout the first half of 2021.
In Mumbai, restricted testing for the asymptomatic patients might have led to underreporting of cases⁸ also highlighted by the seroprevalence survey conducted in July and October 2020. A total of 13 Thousand more deaths were reported in Mumbai compared to 2019 indicating an underreporting of deaths that might have occurred due to the COVID-19 virus⁹. The total deaths recorded in the L ward according to MCGM were 565 until 16th January 2021. The authorities in Mumbai should avoid looking at crude numbers as an indicator to device strategies but consider the underreported deaths derived with the help of modelling to determine the future path.
Analysis of Delhi
Exhibit 3 shows the similar COVID-19 localization model for Delhi set up by the students of the University of Hyderabad in August to understand the spread of the pandemic and gain insights on how it will spread in the near future. As per the first seroprevalence survey conducted in Delhi, about 22.86% of those surveyed developed antibodies by the mid of July 2020¹⁰. Taking this into consideration and the data collected, the localization model was built. The model developed in August 2020 shows the drastic increase in the infection rates that Delhi experienced in December. The model predicted that Delhi would see about 5 Lakh Confirmed Cases by the start of December, while the actual numbers reported were about 5,70,000. The model predicted that Delhi might hit its peak in the mid of December 2020. The model then shows a gradual decline in cases and the infections coming to a close zero in June 2021. Rerunning the model by including the data till now and considering the opening of the economy will provide accurate and insightful observations of the future of Delhi.
The localized model for Delhi predicted that the number of total deaths by January 2021 would be around 35 Thousand, although the reported deaths were around 10 Thousand. This difference in deaths is observed as the model uplifts the reported deaths to compensate for the under-reporting of deaths¹². At Delhi, significant differences have been reported in the death figures provided by the crematoriums and the Municipal Corporation. Further, the authorities refrained from testing the asymptomatic patients¹³. This raises a significant question, did we miss out on deaths that might have occurred due to COVID-19 in Delhi?
Mumbai, a preview of what will happen in Delhi?
Exhibit 4 shows that when we overlaid the results of the two models we saw that Delhi was lagging Mumbai ‘splitting the difference’ between the slum and non-slums of L ward, Mumbai. are likely to witness a surge in cases from January 2021. Thus, it seems that Delhi might be following the infection trajectories of Mumbai, but this may be hidden by the way the data is collected in Delhi: there is no separation between slum and non-slum areas. The virus is moving faster in non-slums of L ward than in Delhi, thus making it an accurate example to suggest containment strategies to the national capital region.
It is necessary to note, the model for Delhi presents a comprehensive outlook of total cases in both slums and non-slums. Due to the unavailability of serosurvey in Delhi with slum and non-slum bifurcation, it is not easy to analyse the slums and non-slums of Delhi as done for Mumbai. This presents a serious problem as this muddle of averages might mislead the authorities. There is a high possibility of a massive surge in Delhi similar to that of Mumbai and the health authorities ought to be cautious as the strategies may differ for the slums and non-slums due to limited access to sanitation, high normal contacts per day per person and relatively higher density of population.
It is peculiar to note even if the cases in Delhi surpass those in non-slums of the L ward, the total deaths in Delhi lie in between the slums and non-slums of the L ward. In Exhibit 4 the deaths per capita of Delhi are in between the slums and non-slums of Mumbai. One reason for this may be the deaths per capita in Delhi are comprising both the slums and the non-slums. Delhi also witnessed huge under-reporting of deaths and this might be another reason. The heavy testing conducted by the Delhi government might also contribute to the high cases per capita when compared to the slums and non-slums of Mumbai. Taking L ward, Mumbai as an example, the non-slums of Delhi seem to be at a higher risk of exposure to the virus in the second wave.
Insights from the Susceptible count
The localized models have attempted to trace the infections by considering the total count of susceptibles. Susceptible is the section of the population that most likely can get infected by the virus in near future. A lower percentage of population susceptible implies a higher percentage of the population infected with COVID-19. In L ward Mumbai, from the month December 2020 to February 2021, the susceptible population is below 1 lakh which implies a higher infection count. According to MCGM, L Ward reported merely 22 new cases in total on the 25th of January 2021. The MCGM authorities may be misled by the fall in the count of cases thus the localized model can be used as an insight to ramp up the efforts to tackle any unprecedented surge in cases.
The localized model for Delhi predicts that there are about 7 million Susceptibles in Delhi by January 2021. The current infection rate in Delhi might be slow, but there is a chance of a massive spike in the infections unless precautions are taken. The authorities should not be fooled by the misleading numbers and take into account the varying infection rates in slums and non-slums. Reinfections might occur and render the medical infrastructure strained unless the authorities ramp up the facilities. Are the authorities in Delhi and Mumbai ready to face the current sudden surge in the cases? Thus, the city corporations must pay attention to the total susceptible that provide a much clearer picture of the total infections (symptomatic and asymptomatic) enabling better policy formulations for concerned localized areas.
The major risks highlighted from our comparative study of Delhi and L ward, Mumbai is:
- Delhi was 12 weeks behind Mumbai last year and may still be behind Mumbai in the trajectory leading to a surge of cases in Delhi.
- The muddle of averages of slums and non-slums in Delhi is misleading the authorities.
- The decline in cases has misled the medical authorities thus it is vital to ensure strict adherence to COVID-19 guidelines.
- There are fewer insights about the advent of reinfections, making the study of total susceptibles important in assessing the impact of the virus.
The important insights from Delhi and Mumbai are:
- The non-slum areas of Mumbai might experience a surge of infections 7x the level that of April in the month from January 2021.
- The models of Delhi and Mumbai predict a considerably lower susceptible count in the months of January-February 2021 indicating a higher infection count of the COVID-19 virus.
- The total deaths registered might be a deflated count of the actual deaths that have occurred in Delhi and Mumbai.
What should be the possible way forward for the authorities in Delhi and Mumbai? The medical authorities at Delhi should continue with extensive testing facilities as they did in the early months of the pandemic. Further, with the aid of the Central government, the makeshift beds made available for quarantine facilities (Jawaharlal Nehru stadium) should continue to limit the contacts per person per day of the infected person. They should keep in mind the varying spread of the virus in slums and non-slum areas and track the spread properly. A proper serosurvey has to be conducted bifurcating the slums and non-slums. While in Mumbai, the seroprevalence survey conducted in October recorded antibodies in the 18% population of non-slums¹⁵; an increase in the earlier results of July. Thus, making it vital for MCGM to gear up the medical facilities to accommodate the current boom in the cases. Further, the authorities can develop distinct strategies for slums and non-slums as they have different characteristics. It is necessary to promote synergy between the Elected Representatives of the urban local bodies and ward — level medical officials.
Following are the suggestions for the medical authorities both at Delhi and Mumbai:
- Conduct antigen tests in localized areas to estimate the total infected population.
- Conduct health protocol awareness drives with help of community leaders.
- Incentivize adherence to hygiene measures by positive nudge or penalties.
- Rigorously record reinfections in the locality.
- Use the model to make better estimates of total susceptibles, total infections and deaths.
- Large-scale execution of vaccination drive while working to eliminate vaccine hesitancy.
The viruses often undergo mutations and new variants may be more infectious than the original strain of the virus. The B.1.1.7 variant of COVID-19 first appeared in the United Kingdom in September 2020¹⁶. According to reports, the new variant could be up to 70 per cent more transmissible than the old variant. What’s more worrying is that this strain is affecting people in the age group of 30–60 years which is fairly young. Preliminary data from the UK shows the virus is spreading quickly in parts of southern England, displacing other variants that have been circulating for months¹⁷. Until the 16th of January 2021, a total of 116 cases of the UK strain were reported in India¹⁸. Currently, the situation in the UK is alarming wherein the authorities had to announce a national lockdown. The Ministry of Health in India should take cognizance of the situation. Moreover, this more infectious strain of the COVID-19 virus is likely to increase the reproduction rate of the virus causing an additional burden on the health infrastructure. So the authorities should gear up to face the upcoming challenges.
Analysis shows that test positivity rates are doubling every day in several states of India. It is faster than the growth rates seen in the UK when the B.1.1.7 variant took off at the end of 2020, suggesting a rapid community spread. At the height of the winter surge in London, by comparison, doubling time 9 days. The resurgence is strongest in Maharashtra which is seeing daily infections of more than 30 thousand. 20% of its tests are coming back positive.
The COVID-19 pandemic is a lesson for us to focus on developing our health infrastructure both at public and private levels. The Central government should spend a higher percentage of GDP to build health capital ensuring the long-term sustainability of India in tackling future health crises following the method of decentralized governance.
“The findings, discoveries and opinons in this article are our own; as are any errors or omissions — We all have agreed to voluntarily assist you (TC-19LMG) further in the process of building the localised COVID-19 models. It shall be a one of a kind learning experience for us.”
Thanks to Maurice Glucksman and Dr Kim Warren for their guidance and contributions to this article; to our student coaches Andre Nemec and Hasna Virk at the Covid 19 Localisation Modelling Group who provided advice on setting up the models. Thanks to our Mentors, Fellows, Interns at the Praja Foundation who supported us and worked with us on the original L Ward Model and to Taraak Ropolu who also worked on the Delhi Model and helped with this analysis.
Notes and References
- “The coronavirus is mutating — does it matter?”, https://www.nature.com/articles/d41586-020-02544-6
- “Centre Releases Rs 17,287 Crore As States Seek Funds To Fight Covid-19”, https://www.bloombergquint.com/coronavirus-outbreak/centre-releases-rs-17287-crore-as-states-seeks-funds-to-fight-covid-19
- “Testing must for those negative on antigen test in Delhi: Harsh Vardhan”, https://www.hindustantimes.com/delhi-news/testing-must-for-those-negative-on-antigen-test-in-delhi-harsh-vardhan/story-8GAZa2c0uKoWCzICtlnXyH.html
- “Covid-19 in Mumbai: BMC relies more on antigen tests since September”, https://www.hindustantimes.com/mumbai-news/covid-19-in-mumbai-bmc-conducting-more-antigen-tests-than-rt-pcr/story-sm8uKhLT6XaqZ5kdFnq0II.html
- “57% of Mumbai slum dwellers have Covid antibodies: Sero survey”, https://timesofindia.indiatimes.com/city/mumbai/57-of-mumbai-slumdwellers-have-covid-antibodies-sero-survey/articleshow/77231640.cms
- “A Second Wave imminent in Mumbai? Implications for Covid 19 from L Ward”, https://medium.com/@info_33771/second-wave-of-covid-19-mumbai-implications-from-l-ward-dba865c5203f
- “Mumbai: Zero new Covid-19 cases in Dharavi second time”, https://timesofindia.indiatimes.com/city/mumbai/mumbai-zero-new-covid-19-cases-in-dharavi-second-time/articleshow/80410255.cms
- “Mumbai: No test, only quarantine for asymptomatic contacts”, https://timesofindia.indiatimes.com/city/mumbai/mumbai-no-test-only-quarantine-for-asymptomatic-contacts/articleshow/75150122.cms
- “13,000 more deaths in Mumbai this year between March & September”, https://timesofindia.indiatimes.com/city/mumbai/13k-more-deaths-in-city-this-year-between-march-sept/articleshow/78920631.cms
- “29% of Delhi has antibodies for Covid-19: Second serosurvey”, https://www.hindustantimes.com/india-news/29-of-delhi-has-antibodies-for-covid-19-second-sero-survey/story-JE4rKi7lMDYaOgAacWoCVK.html
- “Delhi govt continues to under-report COVID-19 death, LN Hospital has 230 deaths, AIIMS 65”, https://www.nationalheraldindia.com/india/delhi-govt-continues-to-under-report-covid-19-death-ln-hospital-has-230-deaths-aiims-65
- “No tests for asymptomatic patients in Delhi now”, https://economictimes.indiatimes.com/news/politics-and-nation/no-tests-for-asymptomatic-persons-in-delhi-now/articleshow/76183062.cms
- “COVID-19: What You Need To Know To Interpret the Sero-Survey Results Right”, https://science.thewire.in/the-sciences/covid-19-icmr-delhi-mumbai-seroprevalence-surveys-results-confidence-interval-statistical-significance/
- “2nd serosurvey in Mumbai shows fall in the level of antibodies among slum dwellers”, https://indianexpress.com/article/cities/mumbai/2nd-sero-survey-in-mumbai-shows-fall-in-level-of-antibodies-among-slum-dwellers-6664156/
- “About Variants of the Virus that Causes COVID-19”, https://www.cdc.gov/coronavirus/2019-ncov/transmission/variant.html
- Wondering what the new coronavirus strain is all about? A doctor explains | Lifestyle News, The Indian Express
- “116 people infected with UK variant of Covid-19 in India: Government”, https://www.thehindubusinessline.com/news/116-people-infected-with-uk-variant-of-covid-19-in-india-government/article33587916.ece