Role of Multi-resolution Vulnerability Indices in COVID-19 spread: A Case Study in India (preprint)

Introduction: The outbreak of COVID-19 has differentially affected countries in the world, with health infrastructure and other related vulnerability indicators playing a role in determining the extent of the COVID-19 spread. Vulnerability of a geographical region/country to COVID-19 has been a topic of interest, particularly in low- and middle-income countries like India to assess the multi-factorial impact of COVID-19 on the incidence, prevalence or mortality data. Datasets and Methods Based on publicly reported socio-economic, demographic, health-based and epidemiological data from national surveys in India, we compute contextual, COVID-19 Vulnerability Indices (cVIs) across multiple thematic resolutions for different geographical and spatial administrative regions. These multi-resolution cVIs were used in regression models to assess their impact on indicators of the spread of COVID-19 such as the average time-varying instantaneous reproduction number. Results Our observational study was focused on 30 districts of the eastern Indian state of Odisha. It is an agrarian state, prone to natural disasters and one of the largest contributors of an unprotected migrant workforce. Our analyses identified housing and hygiene conditions, availability of health care and COVID preparedness as important spatial indicators. Conclusion Odisha has demonstrated success in containing the COVID-19 infection to a reasonable level with proactive measures to contain the spread of the virus during the first wave. However, with the onset of the second wave of COVID, the virus has been making inroads into the hinterlands and peripheral districts of the state, burdening the already deficient public health system in these areas. The vulnerability index presented in this paper identified vulnerable districts in Odisha. While some of them may not have a large number of COVID-19 cases at a given point of time, they could experience repercussions of the pandemic. Improved understanding of the factors driving COVID-19 vulnerability will help policy makers prioritise resources and regions leading to more effective mitigation strategies for the COVID-19 pandemic and beyond.

Resurgence of SARS-CoV-2 in India: Potential role of the B.1.617.2 (Delta) variant and delayed interventions (preprint)

India has seen a surge of SARS-CoV-2 infections and deaths in early part of 2021, despite having controlled the epidemic during 2020. Building on a two-strain, semi-mechanistic model that synthesizes mortality and genomic data, we find evidence that altered epidemiological properties of B.1.617.2 (Delta) variant play an important role in this resurgence in India. Under all scenarios of immune evasion, we find an increased transmissibility advantage for B.1617.2 against all previously circulating strains. Using an extended SIR model accounting for reinfections and wanning immunity, we produce evidence in support of how early public interventions in March 2021 would have helped to control transmission in the country. We argue that enhanced genomic surveillance along with constant assessment of risk associated with increased transmission is critical for pandemic responsiveness.

Network-based Modeling of COVID-19 Dynamics: Early Pandemic Spread in India (preprint)

Modeling the dynamics of COVID-19 pandemic spread is a challenging and relevant problem. Established models for the epidemic spread such as compartmental epidemiological models e.g. Susceptible-Infected-Recovered (SIR) models and its variants, have been discussed extensively in the literature and utilized to forecast the growth of the pandemic across different hot-spots in the world. The standard formulations of SIR models rely upon summary-level data, which may not be able to fully capture the complete dynamics of the pandemic growth. Since the disease spreads from carriers to susceptible individuals via some form of contact, it inherently relies upon a network of individuals for its growth, with edges established via direct interaction, such as shared physical proximity. Using individual-level COVID-19 data from the early days (January 30 to April 15, 2020) of the pandemic in India, and under a network-based SIR model framework, we performed state-specific forecasting under multiple scenarios characterized by the basic reproduction number of COVID-19 across 34 Indian states and union territories. We validated our short-term projections using observed case counts and the long-term projections using national sero-survey findings. Based on healthcare availability data, we also performed projections to assess the burdens on the infrastructure along the spectrum of the pandemic growth. We have developed an \href{https://bayesrx.shinyapps.io/COV-N/}{interactive dashboard} summarizing our results. Our predictions successfully identified the initial hot-spots of India such as Maharashtra and Delhi, and those that emerged later, such as Madhya Pradesh and Kerala. These models have the potential to inform appropriate policies for isolation and mitigation strategies to contain the pandemic, through a phased approach by appropriate resource prioritization and allocation.

Dynamic Structural Impact of the COVID-19 Outbreak on the Stock Market and the Exchange Rate: A Cross-country Analysis Among BRICS Nations (preprint)

COVID-19 has impacted the economy of almost every country in the world. Of particular interest are the responses of the economic indicators of developing nations (such as BRICS) to the COVID-19 shock. As an extension to our earlier work on the dynamic associations of pandemic growth, exchange rate, and stock market indices in the context of India, we look at the same question with respect to the BRICS nations. We use structural variable autoregression (SVAR) to identify the dynamic underlying associations across the normalized growth measurements of the COVID-19 cumulative case, recovery, and death counts, and those of the exchange rate, and stock market indices, using data over 203 days (March 12 - September 30, 2020). Using impulse response analyses, the COVID-19 shock to the growth of exchange rate was seen to persist for around 10+ days, and that for stock exchange was seen to be around 15 days. The models capture the contemporaneous nature of these shocks and the subsequent responses, potentially guiding to inform policy decisions at a national level. Further, causal inference-based analyses would allow us to infer relationships that are stronger than mere associations.

A Comparison of Five Epidemiological Models for Transmission of SARS-CoV-2 in India (preprint)

BackgroundMany popular disease transmission models have helped nations respond to the COVID-19 pandemic by informing decisions about pandemic planning, resource allocation, implementation of social distancing measures and other non-pharmaceutical interventions. We study how five epidemiological models forecast and assess the course of the pandemic in India: a baseline model, an extended SIR (eSIR) model, two extended SEIR (SAPHIRE and SEIR-fansy) models, and a semi-mechanistic Bayesian hierarchical model (ICM). MethodsUsing COVID-19 data for India from March 15 to June 18 to train the models, we generate predictions from each of the five models from June 19 to July 18. To compare prediction accuracy with respect to reported cumulative and active case counts and cumulative death counts, we compute the symmetric mean absolute prediction error (SMAPE) for each of the five models. ResultsFor active case counts, SMAPE values are 0.72 (SEIR-fansy) and 33.83 (eSIR). For cumulative case counts, SMAPE values are 1.76 (baseline) 23. (eSIR), 2.07 (SAPHIRE) and 3.20 (SEIR-fansy). For cumulative death counts, the SMAPE values are 7.13 (SEIR-fansy) and 26.30 (eSIR). For cumulative cases and deaths, we compute Pearson’s and Lin’s correlation coefficients to investigate how well the projected and observed reported COVID-counts agree. Three models (SAPHIRE, SEIR-fansy and ICM) return total (sum of reported and unreported) counts as well. We compute underreporting factors as of June 30 and note that the SEIR-fansy model reports the highest underreporting factor for active cases (6.10) and cumulative deaths (3.62), while the SAPHIRE model reports the highest underreporting factor for cumulative cases (27.79).ConclusionsIn this comparative paper we describe five different models used to study full disease transmission of the SARS-Cov-2 disease transmission in India. While simulation studies are the only gold standard way to compare the accuracy of the models, here we were uniquely poised to compare the projected case-counts against observed data on a test period. Prediction of daily active number of cases does show appreciable variation across models. The largest variability across models is observed in predicting the “total” number of infections including reported and unreported cases. The degree of under-reporting has been a major concern in India. 

A comparison of five epidemiological models for transmission of SARS-CoV-2 in India (preprint)

Many popular disease transmission models have helped nations respond to the COVID-19 pandemic by informing decisions about pandemic planning, resource allocation, implementation of social distancing measures and other non-pharmaceutical interventions. We study how five epidemiological models forecast and assess the course of the pandemic in India: a baseline model, an extended SIR (eSIR) model, two extended SEIR (SAPHIRE and SEIR-fansy) models, and a semi-mechanistic Bayesian hierarchical model (ICM). Using COVID-19 data for India from March 15 to June 18 to train the models, we generate predictions from each of the five models from June 19 to July 18. To compare prediction accuracy with respect to reported cumulative and active case counts and cumulative death counts, we compute the symmetric mean absolute prediction error (SMAPE) for each of the five models. For active case counts, SMAPE values are 0.72 (SEIR-fansy) and 33.83 (eSIR). For cumulative case counts, SMAPE values are 1.76 (baseline) 23.10 (eSIR), 2.07 (SAPHIRE) and 3.20 (SEIR-fansy). For cumulative death counts, the SMAPE values are 7.13 (SEIR-fansy) and 26.30 (eSIR). For cumulative cases and deaths, we compute Pearsons and Lins correlation coefficients to investigate how well the projected and observed reported COVID-counts agree. Three models (SAPHIRE, SEIR-fansy and ICM) return total (sum of reported and unreported) counts as well. We compute underreporting factors as of June 30 and note that the SEIR-fansy model reports the highest underreporting factor for active cases (6.10) and cumulative deaths (3.62), while the SAPHIRE model reports the highest underreporting factor for cumulative cases (27.79).

Reconciling epidemiological models with misclassified case-counts for SARS-CoV-2 with seroprevalence surveys: A case study in Delhi, India (preprint)

Underreporting of COVID-19 cases and deaths is a hindrance to correctly modeling and monitoring the pandemic. This is primarily due to limited testing, lack of reporting infrastructure and a large number of asymptomatic infections. In addition, diagnostic tests (RT-PCR tests for detecting current infection) and serological antibody tests for IgG (to assess past infections) are imperfect. In particular, the diagnostic tests have a high false negative rate. Epidemiologic models with a latent compartment for unascertained infections like the Susceptible-Exposed-Infected-Removed (SEIR) models can provide predictions for unreported cases and deaths under certain assumptions. Typically, the number of unascertained cases is unobserved and thus we cannot validate these estimates for a real study except for simulation studies. Population-based seroprevalence studies can provide a rough estimate of the total number of infections and help us check epidemiologic model projections. In this paper, we develop a method to account for high false negative rates in RT-PCR in an extension to the classic SEIR model. We apply this method to Delhi, the national capital region of India, with a population of 19.8 million and a COVID-19 hotspot of the country, obtaining estimates of underreporting factor for cases at 34-53 times and that for deaths at 8-13 times. Based on a recently released serological survey for Delhi with an estimated 22.86% seroprevalence, we compute adjusted estimates of the true number of infections reported by the survey (after accounting for misclassification of the antibody test results) which is largely consistent with the model outputs, yielding an underreporting factor for cases from 30-42. Together with the model and the serosurvey, this implies approximately 96-98% cases in Delhi remained unreported and whereas only 109,140 cases were reported on July 10, the true number of infections varied somewhere between 4.4-4.6 million across different estimates. While repeated serological monitoring is resource intensive, model-based adjustments, run with the most up to date data, can provide a viable option to keep track of the unreported cases and deaths and gauge the true extent of transmission of this insidious virus.

Examining the Effect of COVID-19 on Foreign Exchange Rate and Stock Market -- An Applied Insight into the Variable Effects of Lockdown on Indian Economy (preprint)

Since March 25, 2020, India had been under a nation-wide lockdown announced as a response to the spread of SARS-CoV-2 and COVID-19 and has resorted to a process of 'unlocking' the lockdown over the past couple of months. This work attempts to examine the effect of novel coronavirus 2019 (COVID-19) and its resulting disease, the COVID-19, on the foreign exchange rates and stock market performances of India using secondary data over a span of 112 days spanning between March 11 and June 30, 2020. The study explores whether the causal relationships and directions among the growth rate of confirmed cases (GROWTHC), exchange rate (GEX) and SENSEX value (GSENSEX) are remaining the same across different pre and post-lockdown phases, attempting to capture any potential changes over time via the vector autoregressive (VAR) models. A positive correlation is found between the growth rate of confirmed cases and the growth rate of exchange rate, and a negative correlation between the growth rate of confirmed cases and the growth rate of SENSEX value. However, on applying a vector autoregressive (VAR) model, it is observed that an increase in the confirmed COVID-19 cases causes no significant change in the values of the exchange rate and SENSEX index. The result varies if the analysis is split across different time periods - before lockdown, the four phases of lockdown, and the first phase of unlock. Nuanced and sensible interpretations of the numeric results indicate significant variability across time in terms of the relation between the variables of interest. The detailed knowledge about the varying patterns of dependence could potentially help the policy makers and investors of India in order to develop their strategies to cope up with the situation.

Lockdown Effect on COVID-19 Spread in India: National Data Masking State-Level Trends (preprint)

Introduction India has been under four phases of a national lockdown from March 25 to May 31 in response to the COVID-19 pandemic. Unmasking the state-wise variation in the effect of the nationwide lockdown on the progression of the pandemic could inform dynamic policy interventions towards containment and mitigation. Methods Using data on confirmed COVID-19 cases across 20 states that accounted for more than 99% of the cumulative case counts in India till May 31, 2020, we illustrate the masking of state-level trends and highlight the variations across states by presenting evaluative evidence on some aspects of the COVID-19 outbreak: case-fatality rates, doubling times of cases, effective reproduction numbers, and the scale of testing. Results The estimated effective reproduction number R for India was 3.36 (95% confidence interval (CI): [3.03, 3.71]) on March 24, whereas the average of estimates from May 25 - May 31 stands at 1.27 (95% CI: [1.26, 1.28]). Similarly, the estimated doubling time across India was at 3.56 days on March 24, and the past 7-day average for the same on May 31 is 14.37 days. The average daily number of tests have increased from 1,717 (March 19-25) to 131,772 (May 25-31) with an estimated testing shortfall of 4.58 million tests nationally by May 31. However, various states exhibit substantial departures from these national patterns. Conclusions Patterns of change over lockdown periods indicate the lockdown has been effective in slowing the spread of the virus nationally. The COVID-19 outbreak in India displays large state-level variations and identifying these variations can help in both understanding the dynamics of the pandemic and formulating effective public health interventions. Our framework offers a holistic assessment of the pandemic across Indian states and union territories along with a set of interactive visualization tools that are daily updated at covind19.org.

Predictions, role of interventions and effects of a historic national lockdown in India's response to the COVID-19 pandemic: data science call to arms (preprint)

Importance: India has taken strong and early public health measures for arresting the spread of the COVID-19 epidemic. With only 536 COVID-19 cases and 11 fatalities, India - a democracy of 1.34 billion people - took the historic decision of a 21-day national lockdown on March 25. The lockdown was further extended to May 3, soon after the analysis of this paper was completed. Objective: To study the short- and long-term impact of an initial 21-day lockdown on the total number of COVID-19 cases in India compared to other less severe non-pharmaceutical interventions using epidemiological forecasting models and Bayesian estimation algorithms; to compare effects of hypothetical durations of lockdown from an epidemiological perspective; to study alternative explanations for slower growth rate of the virus outbreak in India, including exploring the association of the number of cases and average monthly temperature; and finally, to outline the pivotal role of reliable and transparent data, reproducible data science methods, tools and products as we reopen the country and prepare for a post lock-down phase of the pandemic. Design, Setting, and Participants: We use the daily data on the number of COVID-19 cases, of recovered and of deaths from March 1 until April 7, 2020 from the 2019 Novel Coronavirus Visual Dashboard operated by the Johns Hopkins University Center for Systems Science and Engineering (JHU CSSE). Additionally, we use COVID-19 incidence counts data from Kaggle and the monthly average temperature of major cities across the world from Wikipedia. Main Outcome and Measures: The current time-series data on daily proportions of cases and removed (recovered and death combined) from India are analyzed using an extended version of the standard SIR (susceptible, infected, and removed) model. The eSIR model incorporates time-varying transmission rates that help us predict the effect of lockdown compared to other hypothetical interventions on the number of cases at future time points. A Markov Chain Monte Carlo implementation of this model provided predicted proportions of the cases at future time points along with credible intervals (CI). Results: Our predicted cumulative number of COVID-19 cases in India on April 30 assuming a 1-week delay in people's adherence to a 21-day lockdown (March 25 - April 14) and a gradual, moderate resumption of daily activities after April 14 is 9,181 with upper 95% CI of 72,245. In comparison, the predicted cumulative number of cases under "no intervention" and "social distancing and travel bans without lockdown" are 358 thousand and 46 thousand (upper 95% CI of nearly 2.3 million and 0.3 million) respectively. An effective lockdown can prevent roughly 343 thousand (upper 95% CI 1.8 million) and 2.4 million (upper 95% CI 38.4 million) COVID-19 cases nationwide compared to social distancing alone by May 15 and June 15, respectively. When comparing a 21-day lockdown with a hypothetical lockdown of longer duration, we find that 28-, 42-, and 56-day lockdowns can approximately prevent 238 thousand (upper 95% CI 2.3 million), 622 thousand (upper 95% CI 4.3 million), 781 thousand (upper 95% CI 4.6 million) cases by June 15, respectively. We find some suggestive evidence that the COVID-19 incidence rates worldwide are negatively associated with temperature in a crude unadjusted analysis with Pearson correlation estimates [95% confidence interval] between average monthly temperature and total monthly incidence around the world being -0.185 [-0.548, 0.236] for January, -0.110 [-0.362, 0.157] for February, and -0.173 [-0.314, -0.026] for March. Conclusions and Relevance: The lockdown, if implemented correctly in the end, has a high chance of reducing the total number of COVID-19 cases in the short term, and buy India invaluable time to prepare its healthcare and disease monitoring system. Our analysis shows we need to have some measures of suppression in place after the lockdown for the best outcome. We cannot heavily rely on the hypothetical prevention governed by meteorological factors such as temperature based on current evidence. From an epidemiological perspective, a longer lockdown between 42-56 days is preferable. However, the lockdown comes at a tremendous price to social and economic health through a contagion process not dissimilar to that of the coronavirus itself. Data can play a defining role as we design post-lockdown testing, reopening and resource allocation strategies. Software: Our contribution to data science includes an interactive and dynamic app (covind19.org) with short- and long-term projections updated daily that can help inform policy and practice related to COVID-19 in India. Anyone can visualize the observed data for India and create predictions under hypothetical scenarios with quantification of uncertainties. We make our prediction codes freely available (https://github.com/umich-cphds/cov-ind-19) for reproducible science and for other COVID-19 affected countries to use them for their prediction and data visualization work.