Mental Health Trends in India from 2020 to 2022: Association with Financial Stress, Food Insecurity, and COVID-19-related Illness Concerns (preprint)

This study examines the impact of pandemic-related worries on mental health in the Indian general adult population from 2020 to 2022. Using data from the Global COVID-19 Trends and Impact Survey (N = 2,576,174 respondents aged >= 18 years in India; an average weekly sample size of around 25,000), it explores the associations between worry variables (namely financial stress, food insecurity, and COVID-19-related health worries) and self-reported symptoms of depression and nervousness. The statistical analysis was conducted using complete cases only (N = 747,996). Our analysis used survey-weighted models, focusing on the three pandemic-related worries as the exposures, while also adjusting for various other covariates, including demographics and calendar time. The study finds significant associations between these worries and mental health outcomes, with financial stress being the most significant factor affecting both depression (adjusted odds ratio: 2.36, 95% confidence interval: [2.27, 2.46]) and nervousness (adjusted odds ratio: 1.91, 95% confidence interval: [1.81, 2.01]) during the first phase of the study period (June 27, 2020, to May 19, 2021). The fully adjusted models also identify additional factors related to mental health, including age, gender, residential status, geographical region, occupation, and education. Moreover, the research highlights that males and urban residents had higher odds ratios for self-reported mental health problems regarding the worry variables than females and rural residents, respectively. Furthermore, the study reveals a rise in the prevalence of self-reported depression and nervousness and their association with COVID-19-related health worries during the lethal second wave of the pandemic in May 2021 compared to the onset of the pandemic. This study shows that social media platforms like Facebook can deploy surveys to a large number of participants globally and can be useful tools in capturing mental health trends and uncovering associations during a public health crisis.

Design and Analysis Heterogeneity in Observational Studies of COVID-19 Booster Effectiveness: A Review and Case Study (preprint)

Background Observational vaccine effectiveness (VE) studies based on real-world data are a crucial supplement to initial randomized clinical trials of Coronavirus Disease 2019 (COVID-19) vaccines. However, there exists substantial heterogeneity in study designs and statistical methods for estimating VE. The impact of such heterogeneity on VE estimates is not clear. Methods We conducted a two-step literature review of booster VE: a literature search for first or second monovalent boosters on January 1, 2023, and a rapid search for bivalent boosters on March 28, 2023. For each study identified, study design, methods, and VE estimates for infection, hospitalization, and/or death were extracted and summarized via forest plots. We then applied methods identified in the literature to a single dataset from Michigan Medicine (MM), providing a comparison of the impact of different statistical methodologies on the same dataset. Results We identified 53 studies estimating VE of the first booster, 16 for the second booster. Of these studies, 2 were case-control, 17 were test-negative, and 50 were cohort studies. Together, they included nearly 130 million people worldwide. VE for all outcomes was very high (around 90%) in earlier studies (i.e., in 2021), but became attenuated and more heterogeneous over time (around 40%-50% for infection, 60%-90% for hospitalization, and 50%-90% for death). VE compared to the previous dose was lower for the second booster (10-30% for infection, 30-60% against hospitalization, and 50-90% against death). We also identified 11 bivalent booster studies including over 20 million people. Early studies of the bivalent booster showed increased effectiveness compared to the monovalent booster (VE around 50-80% for hospitalization and death). Our primary analysis with MM data using a cohort design included 186,495 individuals overall (including 153,811 boosted and 32,684 with only a primary series vaccination), and a secondary test-negative design included 65,992 individuals tested for SARS-CoV-2. When different statistical designs and methods were applied to MM data, VE estimates for hospitalization and death were robust to analytic choices, with test-negative designs leading to narrower confidence intervals. Adjusting either for the propensity of getting boosted or directly adjusting for covariates reduced the heterogeneity across VE estimates for the infection outcome. Conclusion While the advantage of the second monovalent booster is not obvious from the literature review, the first monovalent booster and the bivalent booster appear to offer strong protection against severe COVID-19. Based on both the literature view and data analysis, VE analyses with a severe disease outcome (hospitalization, ICU admission, or death) appear to be more robust to design and analytic choices than an infection endpoint. Test-negative designs can extend to severe disease outcomes and may offer advantages in statistical efficiency when used properly.

Exploring the big data paradox for various estimands using vaccination data from the global COVID-19 Trends and Impact Survey (CTIS) (preprint)

Selection bias poses a challenge to statistical inference validity in non-probability surveys. This study compared estimates of the first-dose COVID-19 vaccination rates among Indian adults in 2021 from a large non-probability survey, COVID-19 Trends and Impact Survey (CTIS), and a small probability survey, the Center for Voting Options and Trends in Election Research (CVoter), against benchmark data from the COVID Vaccine Intelligence Network (CoWIN). Notably, CTIS exhibits a larger estimation error (0.39) compared to CVoter (0.16). Additionally, we investigated the estimation accuracy of the CTIS when using a relative scale and found a significant increase in the effective sample size by altering the estimand from the overall vaccination rate. These results suggest that the big data paradox can manifest in countries beyond the US and it may not apply to every estimand of interest.

Characterizing and Predicting Post-Acute Sequelae of SARS CoV-2 infection (PASC) in a Large Academic Medical Center in the US (preprint)

Objective: The growing number of Coronavirus Disease-2019 (COVID-19) survivors who are affected by Post-Acute Sequelae of SARS CoV-2 infection (PACS) represent a worldwide public health challenge. Yet, the novelty of this condition and the resulting limited data on underlying pathomechanisms so far hampered the advancement of effective therapies. Using electronic health records (EHR) data, we aimed to characterize PASC-associated diagnoses and to develop risk prediction models. Methods: In our cohort of 63,675 COVID-19 positive patients seen at Michigan Medicine, 1,724 (2.7 %) had a recorded PASC diagnosis. We used a case control study design comparing PASC cases with 17,205 matched controls and performed phenome-wide association studies (PheWASs) to characterize enriched phenotypes of the post-COVID-19 period and potential PASC pre-disposing phenotypes of the pre-, and acute-COVID-19 periods. We also integrated PASC-associated phenotypes into Phenotype Risk Scores (PheRSs) and evaluated their predictive performance. Results: In the post-COVID-19 period, cases were significantly enriched for known PASC symptoms (e.g., shortness of breath, malaise/fatigue, and cardiac dysrhythmias) but also many musculoskeletal, infectious, and digestive disorders. We found seven phenotypes in the pre-COVID-19 period (irritable bowel syndrome, concussion, nausea/vomiting, shortness of breath, respiratory abnormalities, allergic reaction to food, and circulatory disease) and 69 phenotypes in the acute-COVID-19 period (predominantly respiratory, circulatory, neurological, digestive, and mental health phenotypes) that were significantly associated with PASC. The derived pre-COVID-19 PheRS and acute-COVID-19 PheRS had low accuracy to differentiate cases from controls; however, they stratified risk well, e.g., a combination of the two PheRSs identified a quarter of the COVID-19 positive cohort at a 3.5-fold increased risk for PASC compared to the bottom 50% of their distributions. Conclusions: Our agnostic screen of time stamped EHR data uncovered a plethora of PASC-associated diagnoses across many categories and highlighted a complex arrangement of presenting and likely pre-disposing features -- the latter with a potential for risk stratification approaches. Yet, considerably more work will need to be done to better characterize PASC and its subtypes, especially long-term consequences, and to consider more comprehensive risk models.

A Case-Crossover Phenome-wide Association Study (PheWAS) for Understanding Post-COVID-19 Diagnosis Patterns (preprint)

Importance Post COVID-19 condition (PCC) is known to affect a large proportion of COVID-19 survivors. Robust study design and methods are needed to understand post-COVID-19 diagnosis patterns in all COVID-19 survivors, not just the ones clinically diagnosed with PCC. Objective To assess which diagnoses appear more frequently after a COVID-19 infection and how they differ by COVID-19 severity and vaccination status. Design We applied a case-crossover phenome-wide association study (PheWAS) in a retrospective cohort of COVID-19 survivors, comparing the occurrences of 1,649 diagnosis-based phenotype codes (PheCodes) pre- and post-COVID-19 infection periods in the same individual using a conditional logistic regression. Setting Patients tested for or diagnosed with COVID-19 at Michigan Medicine from March 10, 2020 through May 1, 2022. Participants 36,856 SARS-CoV-2-positive patients and 141,615 age- and sex-matched SARS-CoV-2-negative patients as a comparison group for sensitivity analysis. Exposure SARS-CoV-2 virus infection as determined by RT-PCR testing and/or clinical evaluation. Main Outcomes and Measures We compared the rate of occurrence of 1,649 disease classification codes in “pre-” and “post-COVID-19 periods”. We studied how this pattern varied by COVID-19 severity and vaccination status at the time of infection. Results Using a case-crossover PheWAS framework, we found mental, circulatory, and respiratory disorders to be strongly associated with the “post-COVID-19 period” for the overall COVID-19-positive cohort. A total of 325 PheCodes reached phenome-wide significance (p<3e-05), and top hits included cardiac dysrhythmias (OR=1.7 [95%CI: 1.6-1.9]), respiratory failure, insufficiency, arrest (OR=3.1 [95%CI: 2.7-3.5]) and anxiety disorder (OR=1.7 [95%CI: 1.6-1.8]). In the patients with severe disease, we found stronger associations with many respiratory and circulatory disorders, such as pneumonia (p=2.1e-18) and acute pulmonary heart disease (p=2.4e-8), and the “post-COVID-19 period,” compared to those with mild/moderate disease. Test negative patients exhibited a somewhat similar association pattern to those fully vaccinated, with mental health and chronic circulatory diseases rising to the top of the association list in these groups. Conclusions and Relevance Our results confirm that patients experience myriad symptoms more than 28 days after SARS-CoV-2 infection, but especially mental, circulatory, and respiratory disorders. Our case-crossover PheWAS approach controls for within-person confounders that are time-invariant. Comparison to test negatives with a similar design helped identify enrichment specific to COVID-19. As we look into the future, we must be aware of COVID-19 survivors’ healthcare needs in the period after infection. Key Points Question What patterns of clinical diagnosis tend to occur more frequently after a COVID-19 infection and how do they vary by COVID-19 severity and vaccination status? Findings In a cohort of 36,856 COVID-19-positive patients, using a case-crossover phenome-wide association analysis that controls for within-subject confounders, we found symptoms such as anxiety disorder, cardiac dysrhythmias, and respiratory failure to be significantly associated with the “post-COVID-19 period.” Patients with severe COVID-19 were more likely to receive diagnoses related to respiratory conditions in their “post-COVID-19 period” compared to those with mild/moderate COVID-19. The landscape of phenome-wide association signals for the vaccinated group featured common chronic conditions when compared to the signals in the unvaccinated group. Meaning Symptoms across multiple organ systems, especially in the mental, circulatory, and respiratory domains, were associated with the “post-COVID-19 period.” Characterization of post-COVID-19 diagnosis patterns is crucial to understand the long term and future healthcare burden of COVID-19.

COVID-19 outcomes by cancer status, type, treatment, and vaccination (preprint)

Background: Observational studies have identified patients with cancer as a potential subgroup of individuals at elevated risk of severe SARS-CoV-2 (COVID-19) disease and mortality. Early studies showed an increased risk of COVID-19 mortality for cancer patients, but it is not well understood how this association varies by cancer site, cancer treatment, and vaccination status. Methods: Using electronic health record data from an academic medical center, we identified 259,893 individuals who were tested for or diagnosed with COVID-19 from March 10, 2020, to February 2, 2022. Of these, 41,218 tested positive for COVID-19 of whom 10,266 had a past or current cancer diagnosis. We conducted Firth-corrected, covariate-adjusted logistic regression to assess the association of cancer status, cancer type, and cancer treatment with four COVID-19 outcomes: hospitalization, intensive care unit (ICU) admission, mortality, and a composite "severe COVID-19" outcome which is the union of the first three outcomes. We examine the effect of the timing of cancer diagnosis and treatment relative to COVID diagnosis, and the effect of vaccination. Results: Cancer status was associated with higher rates of severe COVID-19 infection [OR (95% CI): 1.18 (1.08, 1.29)], hospitalization [OR (95% CI): 1.18 (1.06, 1.28)], and mortality [OR (95% CI): 1.22 (1.00, 1.48)]. These associations were driven by patients whose most recent initial cancer diagnosis was within the past three years. Chemotherapy receipt was positively associated with all four COVID-19 outcomes (e.g., severe COVID [OR (95% CI): 1.96 (1.73, 2.22)], while receipt of either radiation or surgery alone were not associated with worse COVID-19 outcomes. Among cancer types, hematologic malignancies [OR (95% CI): 1.62 (1.39, 1.88)] and lung cancer [OR (95% CI): 1.81 (1.34, 2.43)] were significantly associated with higher odds of hospitalization. Hematologic malignancies were associated with ICU admission [OR (95% CI): 1.49 (1.11, 1.97)] and mortality [OR (95% CI): 1.57 (1.15, 2.11)], while melanoma and breast cancer were not associated with worse COVID-19 outcomes. Vaccinations were found to reduce the frequency of occurrence for the four COVID-19 outcomes across cancer status but those with cancer continued to have elevated risk of severe COVID [cancer OR (95% CI) among those fully vaccinated: 1.69 (1.10, 2.62)] relative to those without cancer even among vaccinated. Conclusion: Our study provides insight to the relationship between cancer diagnosis, treatment, cancer type, vaccination, and COVID-19 outcomes. Our results indicate that it is plausible that specific diagnoses (e.g., hematologic malignancies, lung cancer) and treatments (e.g., chemotherapy) are associated with worse COVID-19 outcomes. Vaccines significantly reduce the risk of severe COVID-19 outcomes in individuals with cancer and those without, but cancer patients are still at higher risk of breakthrough infections and more severe COVID outcomes even after vaccination. These findings provide actionable insights for risk identification and targeted treatment and prevention strategies.

Estimating COVID-19 Vaccination Effectiveness Using Electronic Health Records of an Academic Medical Center in Michigan (preprint)

Importance: Systematic characterization of the protective effect of vaccinations across time and at-risk populations is needed to inform public health guidelines and personalized interventions. Objective: To evaluate the vaccine effectiveness (VE) over time and determine differences across demographic and clinical risk factors of COVID-19. Design, Setting, and Participants: This test negative design consisted of adult patients who were tested or diagnosed for COVID-19 at Michigan Medicine in 2021. Variables extracted from Electronic Health Records included vaccination status, age, gender, race/ethnicity, comorbidities, body mass index, residential-level socioeconomic characteristics, past COVID-19 infection, being immunosuppressed, and health care worker status. Exposure: The primary exposure was vaccination status and was categorized into fully vaccinated with and without booster, partially vaccinated, or unvaccinated. Main Outcomes and Measures: The main outcomes were infection with COVID-19 (positive test or diagnosis) and having severe COVID-19, i.e., either being hospitalized or deceased. Based on these, VE was calculated by quarter, vaccine, and patient characteristics. Results: Of 170,487 COVID-19 positive adult patients, 78,002 (45.8%) were unvaccinated, and 92,485 (54.2%) were vaccinated, among which 74,060 (80.1%) were fully vaccinated. COVID-19 positivity and severity rates were substantially higher among unvaccinated (12.1% and 1.4%, respectively) compared to fully vaccinated individuals (4.7% and 0.4%, respectively). Among 7,187 individuals with a booster, only 18 (0.3%) had a severe outcome. The covariate-adjusted VE against an infection was 62.1% (95%CI 60.3-63.8%), being highest in the Q2 of 2021 (90.9% [89.5-92.1%]), lowest in Q3 (60.1% [55.9-64.0%]), and rebounding in Q4 to 68.8% [66.3-71.1%]). Similarly, VE against severe COVID-19 overall was 73.7% (69.6-77.3%) and remained high throughout 2021: 87.4% (58.1-96.3%), 92.2% (88.3-94.8%), 74.4% (64.8-81.5%) and 83.0% (78.8-86.4%), respectively. Data on fully vaccinated individuals from Q4 indicated additional protection against infection with an additional booster dose (VE-Susceptibility: 64.0% [61.1-66.7%] vs. 87.3% [85.0-89.2%]) and severe outcomes (VE-Severity: 78.8% [73.5-83.0%] vs. 94.0% [89.5-96.6%]). Comparisons between Pfizer-BioNTech and Moderna vaccines indicated similar protection against susceptibility (82.9% [80.7-84.9%] versus 88.1% [85.5-90.2%]) and severity (87.1% [80.3-91.6%]) vs. (84.9% [76.2-90.5%]) after controlling for vaccination timing and other factors. There was no significant effect modification by all the factors we examined. Conclusions and Relevance: Our findings suggest that COVID-19 vaccines offered high protection against infection and severe COVID-19, and showed decreasing effectiveness over time and improved protection with a booster.

Global Prevalence of Post-Acute Sequelae of COVID-19 (PASC) or Long COVID: A Meta-Analysis and Systematic Review (preprint)

Importance As SARS-CoV-2 pervades worldwide, considerable focus has been placed on the longer lasting health effects of the virus on the human host and on the anticipated healthcare needs. Objective The primary aim of this study is to examine the prevalence of post-acute sequelae of COVID-19 (PASC), commonly known as long COVID, across the world and to assess geographic heterogeneities through a systematic review and meta-analysis. A second aim is to provide prevalence estimates for individual symptoms that have been commonly reported as PASC, based on the existing literature. Data Sources PubMed, Embase, and iSearch for preprints from medRxiv, bioRxiv, SSRN, and others, were searched on July 5, 2021 with verification extending to August 12, 2021. Study Selection Studies written in English that consider PASC (indexed as ailments persisting at least 28 days after diagnosis or recovery for SARS-CoV-2 infection) and that examine corresponding prevalence, risk factors, duration, or associated symptoms were included. A total of 40 studies were included with 9 from North America, 1 from South America, 17 from Europe, 11 from Asia, and 2 from other regions. Data Extraction and Synthesis Data extraction was performed and separately cross-validated on the following data elements: title, journal, authors, date of publication, outcomes, and characteristics related to the study sample and study design. Using a random effects framework for meta-analysis with DerSimonian-Laird pooled inverse-variance weighted estimator, we provide an interval estimate of PASC prevalence, globally, and across regions. This meta-analysis considers variation in PASC prevalence by hospitalization status during the acute phase of infection, duration of symptoms, and specific symptom categories. Main Outcomes and Measures Prevalence of PASC worldwide and stratified by regions. Results Global estimated pooled PASC prevalence derived from the estimates presented in 29 studies was 0.43 (95% confidence interval [CI]: 0.35, 0.63), with a higher pooled PASC prevalence estimate of 0.57 (95% CI: 0.45, 0.68), among those hospitalized during the acute phase of infection. Females were estimated to have higher pooled PASC prevalence than males (0.49 [95% CI: 0.35, 0.63] versus 0.37 [95% CI: 0.24, 0.51], respectively). Regional pooled PASC prevalence estimates in descending order were 0.49 (95% CI: 0.21, 0.42) for Asia, 0.44 (95% CI: 0.30, 0.59) for Europe, and 0.30 (95% CI: 0.32, 0.66) for North America. Global pooled PASC prevalence for 30, 60, 90, and 120 days after index test positive date were estimated to be 0.36 (95% CI: 0.25, 0.48), 0.24 (95% CI: 0.13, 0.39), 0.29 (95% CI: 0.12, 0.57) and 0.51 (95% CI: 0.42, 0.59), respectively. Among commonly reported PASC symptoms, fatigue and dyspnea were reported most frequently, with a prevalence of 0.23 (95% CI: 0.13, 0.38) and 0.13 (95% CI: 0.09, 0.19), respectively. Conclusions and Relevance The findings of this meta-analysis suggest that, worldwide, PASC comprises a significant fraction (0.43 [95% CI: 0.35, 0.63]) of COVID-19 tested positive cases and more than half of hospitalized COVID-19 cases, based on available literature as of August 12, 2021. Geographic differences appear to exist, as lowest to highest PASC prevalence is observed for North America (0.30 [95% CI: 0.32, 0.66]) to Asia (0.49 [95% CI: 0.21, 0.42]). The case-mix across studies, in terms of COVID-19 severity during the acute phase of infection and variation in the clinical definition of PASC, may explain some of these differences. Nonetheless, the health effects of COVID-19 appear to be prolonged and can exert marked stress on the healthcare system, with 237M reported COVID-19 cases worldwide as of October 12, 2021.

Exposure and risk factors for COVID-19 and the impact of staying home on Michigan residents

COVID-19 has had a substantial impact on clinical care and lifestyles globally. The State of Michigan reports over 80,000 positive COVID-19 tests between March 1, 2020 and July 29, 2020. We surveyed 8,041 Michigan Medicine biorepository participants in late June 2020. We found that 55% of COVID-19 cases reported no known exposure to family members or to someone outside the house diagnosed with COVID-19. A significantly higher rate of COVID-19 cases were employed as essential workers (45% vs 19%, p = 9x10-12). COVID-19 cases reporting a fever were more likely to require hospitalization (categorized as severe;OR = 4.4 [95% CI: 1.6-12.5, p = 0.005]) whereas respondents reporting rhinorrhea was less likely to require hospitalization (categorized as mild-to-moderate;OR = 0.16 [95% CI: 0.04-0.73, p = 0.018]). African-Americans reported higher rates of being diagnosed with COVID-19 (OR = 4.0 [95% CI: 2.2-7.2, p = 5x10-6]), as well as higher rates of exposure to family or someone outside the household diagnosed with COVID-19, an annual household income < $40,000, living in rental housing, and chronic diseases. During the Executive Order in Michigan, African Americans, women, and the lowest income group reported worsening health behaviors and higher overall concern for the potential detrimental effects of the pandemic. The higher risk of contracting COVID-19 observed among African Americans may be due to the increased rates of working as essential employees, lower socioeconomic status, and exposure to known positive cases. Continued efforts should focus on COVID-19 prevention and mitigation strategies, as well as address the inequality gaps that result in higher risks for both short-term and long-term health outcomes.

SARS-CoV-2 infection fatality rates in India: systematic review, meta-analysis and model-based estimation (preprint)

IntroductionFervorous investigation and dialogue surrounding the true number of SARS-CoV-2 related deaths and implied infection fatality rates in India have been ongoing throughout the pandemic, and especially pronounced during the nations devastating second wave. We aim to synthesize the existing literature on the true SARS-CoV-2 excess deaths and infection fatality rates (IFR) in India, through a systematic search followed by viable meta-analysis. We then provide updated epidemiological model-based estimates of the wave 1, wave 2 and combined IFRs using an extension of the Susceptible-Exposed-Infected-Removed (SEIR) model, using data from April 1, 2020 to June 30, 2021. MethodsFollowing PRISMA guidelines, the databases PubMed, Embase, Global Index Medicus, as well as BioRxiv, MedRxiv, and SSRN for preprints (accessed through iSearch), were searched on July 3, 2021 (with results verified through August 15, 2021). Altogether using a two-step approach, 4,765 initial citations were screened resulting in 37 citations included in the narrative review and 19 studies with 41 datapoints included in the quantitative synthesis. Using a random effects model with DerSimonian-Laird estimation, we meta-analyze IFR1 which is defined as the ratio of the total number of observed reported deaths divided by the total number of estimated infections and IFR2 (which accounts for death underreporting in the numerator of IFR1). For the latter, we provide lower and upper bounds based on the available range of estimates of death undercounting, often arising from an excess death calculation. The primary focus is to estimate pooled nationwide estimates of IFRs with the secondary goal of estimating pooled regional and state-specific estimates for SARS-CoV-2 related IFRs in India. We also try to stratify our empirical results across the first and the second wave. In tandem, we present updated SEIR model estimates of IFRs for waves 1, 2, and combined across the waves with observed case and death count data from April 1, 2020 to June 30, 2021. ResultsFor India countrywide, underreporting factors (URF) for cases (sourced from serosurveys) range from 14.3-29.1 in the four nationwide serosurveys; URFs for deaths (sourced from excess deaths reports) range from 4.4-11.9 with cumulative excess deaths ranging from 1.79-4.9 million (as of June 2021). Nationwide pooled IFR1 and IFR2 estimates for India are 0.097% (95% confidence interval [CI]: 0.067 - 0.140) and 0.365% (95% CI: 0.264 - 0.504) to 0.485% (95% CI: 0.344 - 0.685), respectively, again noting that IFR2 changes as excess deaths estimates vary. Among the included studies in this meta-analysis, the IFR1 generally appear to decrease over time from the earliest study end date to the latest study end date (from 4 June 2020 to 6 July 2021, IFR1 changed from 0.199 to 0.055%), whereas a similar trend is not as readily evident for IFR2 due to the wide variation in excess death estimates (from 4 June 2020 to 6 July 2021, IFR2 ranged from (0.290-1.316) to (0.241-0.651) %). Nationwide SEIR model-based combined estimates for IFR1 and IFR2 are 0.101% (95% CI: 0.097 - 0.116) and 0.367% (95% CI: 0.358 - 0.383), respectively, which largely reconcile with the empirical findings and concur with the lower end of the excess death estimates. An advantage of such epidemiological models is the ability to produce daily estimates with updated data with the disadvantages being that these estimates are subject to numerous assumptions, arduousness of validation and not directly using the available excess death data. Whether one uses empirical data or model-based estimation, it is evident that IFR2 is at least 3.6 times more than IFR1. ConclusionWhen incorporating case and death underreporting, the meta-analyzed cumulative infection fatality rate in India varies from 0.36%-0.48%, with a case underreporting factor ranging from 25-30 and a death underreporting factor ranging from 4-12. This implies, by June 30, 2021, India may have seen nearly 900 million infections and 1.7-4.9 million deaths when the reported numbers stood at 30.4 million cases and 412 thousand deaths (covid19india.org) with an observed case fatality rate (CFR) of 1.35%. We reiterate the need for timely and disaggregated infection and fatality data to examine the burden of the virus by age and other demographics. Large degrees of nationwide and state-specific death undercounting reinforce the call to improve death reporting within India.

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.

Estimating the wave 1 and wave 2 infection fatality rates from SARS-CoV-2 in India (preprint)

Objective There has been much discussion and debate around the underreporting of COVID-19 infections and deaths in India. In this short report we first estimate the underreporting factor for infections from publicly available data released by the Indian Council of Medical Research on reported number of cases and national seroprevalence surveys. We then use a compartmental epidemiologic model to estimate the undetected number of infections and deaths, yielding estimates of the corresponding underreporting factors. We compare the serosurvey based ad hoc estimate of the infection fatality rate (IFR) with the model-based estimate. Since the first and second waves in India are intrinsically different in nature, we carry out this exercise in two periods: the first wave (April 1, 2020 - January 31, 2021) and part of the second wave (February 1, 2021 – May 15, 2021). The latest national seroprevalence estimate is from January 2021, and thus only relevant to our wave 1 calculations. Results Both wave 1 and wave 2 estimates qualitatively show that there is a large degree of “covert infections” in India, with model-based estimated underreporting factor for infections as 11 . 11 (95% credible interval (CrI) 10 . 71-11 . 47) and for deaths as 3 . 56 (95% CrI 3 . 48 - 3 . 64) for wave 1. For wave 2, underreporting factor for infections escalate to 26 . 77 (95% CrI 24 . 26 – 28 . 81) and to 5 . 77 (95% CrI 5 . 34 – 6 . 15) for deaths. If we rely on only reported deaths, the IFR estimate is 0 . 13% for wave 1 and 0 . 03% for part of wave 2. Taking underreporting of deaths into account, the IFR estimate is 0 . 46% for wave 1 and 0 . 18% for wave 2 (till May 15). Combining waves 1 and 2, as of May 15, while India reported a total of nearly 25 million cases and 270 thousand deaths, the estimated number of infections and deaths stand at 491 million (36% of the population) and 1.21 million respectively, yielding an estimated (combined) infection fatality rate of 0 . 25% . There is considerable variation in these estimates across Indian states. Up to date seroprevalence studies and mortality data are needed to validate these model-based estimates.

Estimating COVID-19 Related Mortality in India: An Epidemiological Challenge with Insufficient Data (preprint)

The harrowing second wave of COVID-19 in India has led to much discussion over the quality and timeliness of reporting of deaths attributed to the pandemic. In this brief report, we aim to present the existing evidence, as well as the broader complexities surrounding the mortality burden of COVID-19 in India. This article sheds light on the following epidemiological issues: (1) general and India-specific challenges to COVID-19 death reporting, (2) latest COVID-19 mortality estimates in India as of May 16, 2021, (3) the apparent scale of uncaptured COVID-19 deaths, and (4) the role of disaggregated historic mortality trends in quantification of excess deaths attributed to COVID-19. We conclude with a set of high-level policy recommendations for improving the vital surveillance system and tracking of causes of death in India. We encourage direct efforts to integrate health data and indirect strategies for cross-validation of registered deaths. Such system-wide advances would drastically aid epidemiological research efforts and strengthen India’s position to overcome future public health crises.

Estimating the Infection Fatality Rate from SARS-CoV-2 in India (preprint)

There has been much discussion and debate around the underreporting of COVID-19 infections and deaths in India. In this short report we first estimate this underreporting factor for infections from publicly available data released by the Indian Council of Medical Research from national seroprevalence surveys. We then use a rigorous compartmental epidemiologic model to estimate the undetected number of infections and deaths. We compare the serosurvey-based ad hoc estimates with the model-based estimates. Both estimates qualitatively show that there is a large degree of “covert infections” in India, with model-based estimated underreporting factor for infections as 11.11 (95% CI 10.71-11.47) and for deaths as 3.56 (95% CI 3.48- 3.64). This implies approximately 91% of infections and 72% of deaths related to COVID-19 remain unreported in India. While seroprevalence data are not available across all states and union territories in India in a uniform way, epidemiologic models can prove to be useful tools to estimate the extent of underreporting. These estimates enable us to calculate the infection fatality rate (IFR) for India. If we rely on only reported deaths the IFR estimate is 0.13% while taking underreporting of deaths into account, the IFR estimate is 0.46%. There is considerable variation in these estimates across states.Funding Statement: This research is supported by the Michigan Institute of Data Science (MIDAS), Precision Health Initiative and Rogel Scholar Fund at the University of Michigan. Bhramar Mukherjee's research is supported by NSF DMS 1712933 and CA 046592.Declaration of Interests: The authors have declared no competing interest

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. 

Changes in COVID-19-related outcomes and the impacts of the potential risk factors over time: a follow-up analysis (preprint)

ImportanceCharacteristics of COVID-19 patients changed over the course of the pandemic. Understanding how risk factors changed over time can enhance the coordination of healthcare resources and protect the vulnerable. ObjectiveTo investigate the overall trend of severe COVID-19-related outcomes over time since the start of the pandemic, and to evaluate whether the impacts of potential risk factors, such as race/ethnic groups, changed over time. DesignThis retrospective cohort study included patients tested or treated for COVID-19 at Michigan Medicine (MM) from March 10, 2020, to September 2, 2020. According to the quarter in which they first tested positive, the COVID-19-positive cohort were stratified into three groups: Q1, March 1, 2020 - March 31, 2020; Q2, April 1, 2020 - June 30, 2020; Q3, July 1, 2020 - September 2, 2020. SettingsLarge, academic medical center. ParticipantsIndividuals tested or treated for COVID-19. ExposureExamined potential risk factors included age, race/ethnicity, smoking status, alcohol consumption, comorbidities, body mass index (BMI), and residential-level socioeconomic characteristics. Main Outcomes and MeasuresThe main outcomes included COVID-19-related hospitalization, intensive care unit (ICU) admission, and mortality, which were identified from the electronic health records from MM. ResultsThe study cohort consisted of 53,853 patients tested or treated for COVID-19 at MM, with mean (SD) age of 44.8 (23.1), mean (SD) BMI of 29.1 (7.6), and 23,814 (44.2%) males. Among the 2,582 patients who tested positive, 719 (27.8%) were hospitalized, 377 (14.6%) were admitted to ICU, and 129 (5.0%) died. The overall COVID-positive hospitalization rate decreased from 41.5% in Q1 to 12.6% in Q3, and the overall ICU admission rate decreased from 24.5% to 5.3%. Black patients had significantly higher (unadjusted) overall hospitalization rate (265 [41.1%] vs 326 [23.2%]), ICU admission rate (139 [21.6%] vs 172 [12.2%]), and mortality rate (42 [6.5%] vs 56 [4.0%]) than White patients. Each quarter, the hospitalization rate remained higher for Black patients compared to White patients, but this difference was attenuated over time for the (unadjusted) odds ratios (Q1: OR=1.9, 95% CI [1.25, 2.90]; Q2: OR=1.42, 95% CI [1.02, 1.98]; Q3: OR=1.36, 95% CI [0.67, 2.65]). Similar decreasing patterns were observed for ICU admission and mortality. Adjusting for age, sex, socioeconomic status, and comorbidity score, the racial disparities in hospitalization between White and Black patients were not significant in each quarter of the year (Q1: OR=1.43, 95% CI [0.75, 2.71]; Q2: OR=1.25, 95% CI [0.79, 1.98]; Q3: OR=1.76 95% CI [0.81, 3.85]), in contrast to what was observed in the full cohort (OR=1.85, 95% CI [1.39, 2.47]). Additionally, significant association of hospitalization with living in densely populated area was identified in the first quarter (OR= 664, 95% CI [20.4, 21600]), but such association disappeared in the second and third quarters (Q2: OR= 1.72 95% CI [0.22, 13.5]; Q3: OR=3.69, 95% CI [0.103, 132]). Underlying liver diseases were positively associated with hospitalization in White patients (OR=1.60, 95% CI [1.01, 2.55], P=.046), but not in Black patients (OR=0.49, 95% CI [0.23, 1.06], P=.072, Pint=.013). Similar results were obtained for the effect of liver diseases on ICU admission in White and Black patients (White: OR=1.75, 95% CI [1.01, 3.05], P=.047; Black: OR=0.46, 95% CI [0.17, 1.26], P=.130, Pint=.030). Conclusions and RelevanceThese findings suggest that the COVID-19-related hospitalization, ICU admission, and mortality rates were decreasing over the course of the pandemic. Although racial disparities persisted, the magnitude of the differences in hospitalization and ICU admission rates diminished over time. Key PointsO_ST_ABSQuestionsC_ST_ABSHow did the overall hospitalization and intensive care unit (ICU) admission rates change over the course of the pandemic and how did they vary by race? FindingsIn this cohort study of 2,582 patients testing positive for COVID-19, the unadjusted hospitalization rate decreased from 50.5% in Q1 (March 10, 2020, to March 31, 2020) to 17.9% in Q3 (July 1, 2020, to September 2, 2020) for Black patients, and from 23.2% in Q1 to 13.8% in Q3 for White patients. After adjusting for age, sex, sociodemographic factors, and comorbidity conditions, the odds ratios of hospitalization between White and Black patients were not significant in each quarter of the year 2020. No significant associations between ICU admission and race/ethnic groups were identified in each quarter or the entire three quarters. MeaningThese findings suggests an appreciable decline in hospitalization and ICU admission rates among COVID-19 positive patients. The hospitalization and ICU admission rates across race/ethnic groups became closer over time.

Optimal test allocation strategy for COVID-19 (preprint)

Testing for active SARS-CoV-2 infections is key to controlling the spread of the virus and preventing severe disease. A central public health challenge is defining test allocation strategies in the presence of limited resource and/or multiple competing test options. In this paper, we provide a mathematical frame-work for defining an optimal strategy for allocating viral tests. The framework accounts for imperfect test results, testing in certain high-risk patient populations, practical constraints in terms of budget and/or total number of available tests, and the goal of testing. In our proposed approach, tests can be allocated across population strata defined by symptom severity and other patient characteristics, allowing the test allocation plan to prioritize higher risk patient populations. We extend our proposed method to address the challenge of allocating two different types of tests with different cost and accuracy (for example the expensive but more accurate RT-PCR test versus the cheap but less accurate rapid antigen test), administered under budget constraints. We provide a R Shiny web application allowing users to explore test allocation strategies across a variety of pandemic scenarios. This work can serve as a useful tool for guiding public health decision-making at a community level, adapted to different stages of the pandemic. We use distribution of tests in New York City during the initial wave of the COVID-19 outbreak as a sample example. We also show how this framework can be useful to reopening of college campuses.

COVID-19 prediction in South Africa: Understanding the unascertained cases -- the hidden part of the epidemiological iceberg (preprint)

BackgroundUnderstanding the impact of non-pharmaceutical interventions remains a critical epidemiological problem in South Africa that reported the largest number of confirmed COVID-19 cases and deaths from the African continent. MethodsIn this study, we applied two existing epidemiological models, an extension of the Susceptible-Infected-Removed model (eSIR) and SAPHIRE, to fit the daily ascertained infected (and removed) cases from March 15 to July 31 in South Africa. To combine the desirable features from the two models, we further extended the eSIR model to an eSEIRD model. ResultsUsing the eSEIRD model, the COVID-19 transmission dynamics in South Africa was characterized by the estimated basic reproduction number (R0) at 2.10 (95%CI: [2.09,2.10]). The decrease of effective reproduction number with time implied the effectiveness of interventions. The low estimated ascertained rate was found to be 2.17% (95%CI: [2.15%, 2.19%]) in the eSEIRD model. The overall infection fatality ratio (IFR) was estimated as 0.04% (95%CI: [0.02%, 0.06%]) while the reported case fatality ratio was 4.40% (95% CI: [<0.01%, 11.81%]). As of December 31, 2020, the cumulative number of ascertained cases and total infected would reach roughly 801 thousand and 36.9 million according to the long-term forecasting. ConclusionsThe dynamics based on our models suggested a decline of COVID-19 infection and that the severeness of the epidemic might be largely mitigated through strict interventions. Besides providing insights on the COVID-19 dynamics in South Africa, we develop powerful forecasting tools that allow incorporating ascertained rate and IFR estimation and inquiring into the effect of intervention measures on COVID-19 spread. Key MessagesO_LIThis study delineated the COVID-19 dynamics in South Africa from March 15 to July 31 and confirmed the effectiveness of the main non-pharmaceutical intervention-- lockdown, and mandatory wearing of face-mask in public places using epidemiological models; C_LIO_LICOVID-19 spread in South Africa was found to be associated with both low ascertained rate and low infection fatality ratio; C_LIO_LIAccording to the long-term forecast, by December 31, 2020, the cumulative number of ascertained cases and total infected would reach roughly 801 thousand and 36.9 million respectively. C_LI

EXTENDING THE SUSCEPTIBLE-EXPOSED-INFECTED-REMOVED(SEIR) MODEL TO HANDLE THE HIGH FALSE NEGATIVE RATE AND SYMPTOM-BASED ADMINISTRATION OF COVID-19 DIAGNOSTIC TESTS: SEIR-fansy (preprint)

The false negative rate of the diagnostic RT-PCR test for SARS-CoV-2 has been reported to be substantially high. Due to limited availability of testing, only a non-random subset of the population can get tested. Hence, the reported test counts are subject to a large degree of selection bias. We consider an extension of the Susceptible-Exposed-Infected-Removed (SEIR) model under both selection bias and misclassification. We derive closed form expression for the basic reproduction number under such data anomalies using the next generation matrix method. We conduct extensive simulation studies to quantify the effect of misclassification and selection on the resultant estimation and prediction of future case counts. Finally we apply the methods to reported case-death-recovery count data from India, a nation with more than 5 million cases reported over the last seven months. We show that correcting for misclassification and selection can lead to more accurate prediction of case-counts (and death counts) using the observed data as a beta tester. The model also provides an estimate of undetected infections and thus an under-reporting factor. For India, the estimated under-reporting factor for cases is around 21 and for deaths is around 6. We develop an R-package (SEIRfansy) for broader dissemination of the methods.

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).