Estimating the effectiveness of shielding during pregnancy against SARS-CoV-2 in New York City during the first year of the COVID-19 pandemic

Pregnant patients have increased morbidity and mortality in the setting of SARS-CoV-2 infection. The exposure of pregnant patients in New York City to SARS-CoV-2 is not well understood due to early lack of access to testing and the presence of asymptomatic COVID-19 infections. Before the availability of vaccinations, preventative (shielding) measures, including but not limited to wearing a mask and quarantining at home to limit contact, were recommended for pregnant patients. Using universal testing data from 2196 patients who gave birth from April through December 2020 from one institution in New York City, and in comparison, with infection data of the general population in New York City, we estimated the exposure and real-world effectiveness of shielding in pregnant patients. Our Bayesian model shows that patients already pregnant at the onset of the pandemic had a 50% decrease in exposure compared to those who became pregnant after the onset of the pandemic and to the general population.

Characterizing features of outbreak duration for novel SARS-CoV-2 variants of concern

Characterizing the dynamics of epidemic trajectories is critical to understanding the potential impacts of emerging outbreaks and to designing appropriate mitigation strategies. As the COVID-19 pandemic evolves, however, the emergence of SARS-CoV-2 variants of concern has complicated our ability to assess in real-time the potential effects of imminent outbreaks, such as those presently caused by the Omicron variant. Here, we report that SARS-CoV-2 outbreaks across regions exhibit strain-specific times from onset to peak, specifically for Delta and Omicron variants. Our findings may facilitate real-time identification of peak medical demand and may help fine-tune ongoing and future outbreak mitigation deployment efforts.

Association between City-wide Lockdown and COVID-19 Hospitalization Rates in Multigenerational Households in New York City

BackgroundCity-wide lockdowns and school closures have demonstrably impacted COVID-19 transmission. However, simulation studies have suggested an increased risk of COVID-19 related morbidity for older individuals inoculated by house-bound children. This study examines whether the March 2020 lockdown in New York City (NYC) was associated with higher COVID-19 hospitalization rates in neighborhoods with larger proportions of multigenerational households. MethodsWe obtained daily age-segmented COVID-19 hospitalization counts in each of 166 ZIP code tabulation areas (ZCTAs) in NYC. Using Bayesian Poisson regression models that account for spatiotemporal dependencies between ZCTAs, as well as socioeconomic risk factors, we conducted a difference-in-differences study amongst ZCTA-level hospitalization rates from February 23 to May 2, 2020. We compared ZCTAs in the lowest quartile of multigenerational housing to other quartiles before and after the lockdown. FindingsAmong individuals over 55 years, the lockdown was associated with higher COVID-19 hospitalization rates in ZCTAs with more multigenerational households. The greatest difference occurred three weeks after lockdown: Q2 vs. Q1: 54% increase (95% Bayesian credible intervals: 22 - 96%); Q3 vs. Q1: 48%, (17 - 89%); Q4 vs. Q1: 66%, (30 - 211%). After accounting for pandemic-related population shifts, a significant difference was observed only in Q4 ZCTAs: 37% (7 -76%). InterpretationBy increasing house-bound mixing across older and younger age groups, city-wide lockdown mandates imposed during the growth of COVID-19 cases may have inadvertently, but transiently, contributed to increased transmission in multigenerational households. FundingNational Center for Advancing Translational Sciences; Clinical and Translational Science Center at Weill Cornell Medical College.

Association between overcrowded households, multigenerational households, and COVID-19: a cohort study

IntroductionThe role of overcrowded and multigenerational households as a risk factor for COVID-19 remains unmeasured. The objective of this study is to examine and quantify the association between overcrowded and multigenerational households, and COVID-19 in New York City (NYC). MethodsWe conducted a Bayesian ecological time series analysis at the ZIP Code Tabulation Area (ZCTA) level in NYC to assess whether ZCTAs with higher proportions of overcrowded (defined as proportion of estimated number of housing units with more than one occupant per room) and multigenerational households (defined as the estimated percentage of residences occupied by a grandparent and a grandchild less than 18 years of age) were independently associated with higher suspected COVID-19 case rates (from NYC Department of Health Syndromic Surveillance data for March 1 to 30, 2020). Our main measure was adjusted incidence rate ratio (IRR) of suspected COVID-19 cases per 10,000 population. Our final model controlled for ZCTA-level sociodemographic factors (median income, poverty status, White race, essential workers), prevalence of clinical conditions related to COVID-19 severity (obesity, hypertension, coronary heart disease, diabetes, asthma, smoking status, and chronic obstructive pulmonary disease), and spatial clustering. Results39,923 suspected COVID-19 cases presented to emergency departments across 173 ZCTAs in NYC. Adjusted COVID-19 case rates increased by 67% (IRR 1.67, 95% CI = 1.12, 2.52) in ZCTAs in quartile four (versus one) for percent overcrowdedness and increased by 77% (IRR 1.77, 95% CI = 1.11, 2.79) in quartile four (versus one) for percent living in multigenerational housing. Interaction between both exposures was not significant ({beta}interaction = 0.99, 95% CI: 0.99-1.00). ConclusionsOver-crowdedness and multigenerational housing are independent risk factors for suspected COVID-19. In the early phase of surge in COVID cases, social distancing measures that increase house-bound populations may inadvertently but temporarily increase SARS-CoV-2 transmission risk and COVID-19 disease in these populations.

Analysis and visualization of epidemics on the timescale of burden: derivation and application of Epidemic Resistance Lines (ERLs) to COVID-19 outbreaks in the US

The 2020 COVID-19 pandemic produced thousands of well-quantified epidemics in counties, states, and countries around the world. Comparing the dynamics and outcomes of these nested epidemics could improve our understanding of the efficacy of non-pharmaceutical interventions (NPIs) and help managers with risk assessment across multiple geographic levels. However, cross-outbreak comparisons are challenging due to their variable dates of introduction of the SARS-CoV-2 virus, rates of transmission, case detection rates, and asynchronous and diverse management interventions. Here, we present a graphical method for comparing ongoing COVID-19 epidemics by using disease burden as a natural timescale for comparison. Trajectories of growth rates of cases over the timescale of lagged deaths per-capita produces coherent visual comparisons of epidemics that are otherwise incoherent and asynchronous in the timescale of calendar dates or incomparable using non-stationary measures of burden such as cases. Applied to US COVID-19 outbreaks at the county and state level, this approach reveals lockdowns reducing transmission at fewer deaths per-capita early in the epidemic, reopenings causing resurgent summer epidemics, and peaks that while separated in time and place actually occur at points of similar per-capita deaths. Our method uses early and minimally mitigated epidemics, like that in NYC in March-April 2020 and Sweden in later 2020, to define what we call "epidemic resistance lines" (ERLs) or hypothesized upper bounds of epidemic speed and burden. ERLs from less-mitigated epidemics allow benchmarking of resurgent summer epidemics in the US. In particular, the unmitigated NYC epidemic resistance line appears to bound the growth rates of 3,000 US counties and funnel growth rates across counties to their peaks where growth rates equal zero in the fall and winter of 2020. Corroboration of upper-bounds on epidemic trajectories allowed early predictions of mortality burden for unmitigated COVID-19 epidemics in these populations, predictions that were more accurate for counties in states without mask-wearing mandates. We discuss how this method could be used for future epidemics, including seasonal epidemics caused by influenza or ongoing epidemics caused by new SARS-CoV-2 variants. Press SummaryWhy, despite no statewide mask-wearing mandates or other restrictions like restaurant closures, did South Dakotas COVID-19 epidemic peak not in January, when seasonal forcing wanes, but in early November? Why are we not seeing a resurgent epidemic in Florida or Texas, where non-pharmaceutical interventions have been relaxed for months? How can we compare the current outbreak in India with other countries epidemics to contextualize the speed of the Indian outbreak and estimate the potential loss of life? We have developed a new method of visualizing epidemics in progress that can help to compare distinct COVID-19 outbreaks to understand, in specific cases like South Dakota, why they peaked when they did. The "when" in this case does not refer to prediction of a calendar date, but rather a point in the accumulation of deaths in a given locale due to the disease in question. The method presented in this paper therefore essentially uses population-based burden of disease as a timescale for measuring epidemics. Just as the age of a car can be measured in years or miles, the age of a COVID-19 epidemic can be measured in days or deaths per-capita. Plotting growth rates of cases as a function of per-capita deaths 11 days later produces a real-time visual comparison of epidemics that are otherwise asynchronous in time. This approach permits both direct comparison across local outbreaks that may be disparate in time and/or place, as well as benchmarking of any outbreak against known exemplars of archetypal response strategies, such as New York Citys unmitigated urban outbreak in Spring 2020 and Swedens uncontained summer 2020 epidemic. Whether comparing the speed of resurgent outbreaks following relaxation in US states like Florida or the peak mortality burden in fall outbreaks across thousands of US counties with and without statewide mask-wearing mandates, this method offers a simple, intuitive tool for real-time monitoring and prediction capability connecting epidemic speed, burden, and management interventions. While our findings point to compelling epidemiological hypotheses for peaks in less-regulated states, future work is needed to confirm and extend our results predicting mortality burden at the peak of confirmed cases in the ongoing and evolving COVID-19 pandemic.

COVID-19 reopening strategies at the county level in the face of uncertainty: Multiple Models for Outbreak Decision Support

Policymakers make decisions about COVID-19 management in the face of considerable uncertainty. We convened multiple modeling teams to evaluate reopening strategies for a mid-sized county in the United States, in a novel process designed to fully express scientific uncertainty while reducing linguistic uncertainty and cognitive biases. For the scenarios considered, the consensus from 17 distinct models was that a second outbreak will occur within 6 months of reopening, unless schools and non-essential workplaces remain closed. Up to half the population could be infected with full workplace reopening; non-essential business closures reduced median cumulative infections by 82%. Intermediate reopening interventions identified no win-win situations; there was a trade-off between public health outcomes and duration of workplace closures. Aggregate results captured twice the uncertainty of individual models, providing a more complete expression of risk for decision-making purposes.

High-Quality Masks Can Reduce Infections and Deaths in the US

ObjectivesTo evaluate the effectiveness of widespread adoption of masks or face coverings to reduce community transmission of the SARS-CoV-2 virus that causes Covid-19. MethodsWe employed an agent-based stochastic network simulation model, where Covid-19 progresses across census tracts according to a variant of SEIR. We considered a mask order that was initiated 3.5 months after the first confirmed Covid-19 case. We evaluated scenarios where wearing a mask reduces transmission and susceptibility by 50% or 80%; an individual wears a mask with a probability of 0%, 20%, 40%, 60%, 80%, or 100%. ResultsIf 60% of the population wears masks that are 50% effective, this decreases the cumulative infection attack rate (CAR) by 25%, the peak prevalence by 51%, and the population mortality by 25%. If 100% of people wear masks (or 60% wear masks that are 80% effective), this decreases the CAR by 38%, the peak prevalence by 67%, and the population mortality by 40%. ConclusionsAfter community transmission is present, masks can significantly reduce infections.

Using ILI surveillance to estimate state-specific case detection rates and forecast SARS-CoV-2 spread in the United States

Detection of SARS-CoV-2 infections to date has relied on RT-PCR testing. However, a failure to identify early cases imported to a country, bottlenecks in RT-PCR testing, and the existence of infections which are asymptomatic, sub-clinical, or with an alternative presentation than the standard cough and fever have resulted in an under-counting of the true prevalence of SARS-CoV-2. Here, we show how publicly available CDC influenza-like illness (ILI) outpatient surveillance data can be repurposed to estimate the detection rate of symptomatic SARS-CoV-2 infections. We find a surge of non-influenza ILI above the seasonal average and show that this surge is correlated with COVID case counts across states. By quantifying the number of excess ILI patients in March relative to previous years and comparing excess ILI to confirmed COVID case counts, we estimate the syndromic case detection rate of SARS-CoV-2 in the US to be less than 13%. If only 1/3 of patients infected with SARS-CoV-2 sought care, the ILI surge would correspond to more than 8.7 million new SARS-CoV-2 infections across the US during the three week period from March 8 to March 28. Combining excess ILI counts with the date of onset of community transmission in the US, we also show that the early epidemic in the US was unlikely to be doubling slower than every 4 days. Together these results suggest a conceptual model for the COVID epidemic in the US in which rapid spread across the US are combined with a large population of infected patients with presumably mild-to-moderate clinical symptoms. We emphasize the importance of testing these findings with seroprevalence data, and discuss the broader potential to use syndromic time series for early detection and understanding of emerging infectious diseases.