Impact of excluding nursing home COVID-19 cases when assessing the relationship between county-level social distancing behavior and COVID-19 cases across the US during the early phase of the pandemic, February 2020-May 2020
Open Access
- 30 November 2021
- journal article
- research article
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 16 (11), e0260151
- https://doi.org/10.1371/journal.pone.0260151
Abstract
To conduct a cross-sectional nationwide study examining how exclusion of nursing home COVID-19 cases influences the association between county level social distancing behavior and COVID-19 cases throughout the US during the early phase of the pandemic (February 2020-May 2020). Using county-level COVID-19 data and social distancing metrics from tracked mobile devices, we investigated the impact social distancing had on a county’s total COVID-19 cases (cases/100,000 people) between when the first COVID-19 case was confirmed in a county and May 31st, 2020 when most statewide social distancing measures were lifted, representing the pandemic’s exponential growth phase. We created a mixed-effects negative binomial model to assess how implementation of social distancing measures when they were most stringent (March 2020-May 2020) influenced total COVID-19 cases while controlling for social distancing and COVID-19 related covariates in two scenarios: (1) when COVID-19 nursing home cases are not excluded from total COVID-19 cases and (2) when these cases are excluded. Model findings were compared to those from February 2020, a baseline when social distancing measures were not in place. Marginal effects at the means were generated to further isolate the influence of social distancing on COVID-19 from other factors and determine total COVID-19 cases during March 2020-May 2020 for the two scenarios. Regardless of whether nursing home COVID-19 cases were excluded from total COVID-19 cases, a 1% increase in average % of mobile devices leaving home was significantly associated with a 5% increase in a county’s total COVID-19 cases between March 2020-May 2020 and about a 2.5% decrease in February 2020. When the influence of social distancing was separated from other factors, the estimated total COVID-19 cases/100,000 people was comparable throughout the range of social distancing values (25%-45% of mobile phone devices leaving home between March 2020-May 2020) when nursing home COVID-19 cases were not excluded (25% of mobile phones leaving home: 163.84 cases/100,000 people (95% CI: 121.81, 205.86), 45% of mobile phones leaving home: 432.79 cases/100,000 people (95% CI: 256.91, 608.66)) and when they were excluded (25% of mobile phones leaving home: 149.58 cases/100,000 people (95% CI: 111.90, 187.26), 45% of mobile phones leaving home: 405.38 cases/100,000 people (95% CI: 243.14, 567.62)). Exclusion of nursing home COVID-19 cases from total COVID-19 case counts has little impact when estimating the relationship between county-level social distancing and preventing COVID-19 cases with additional research needed to see whether this finding is also observed for COVID-19 growth rates and mortality.This publication has 17 references indexed in Scilit:
- Characteristics of U.S. Nursing Homes with COVID‐19 CasesJournal of the American Geriatrics Society, 2020
- Strong Social Distancing Measures In The United States Reduced The COVID-19 Growth RateHealth Affairs, 2020
- Public Health Response to the Initiation and Spread of Pandemic COVID-19 in the United States, February 24–April 21, 2020Morbidity and Mortality Weekly Report (MMWR), 2020
- Mathematical assessment of the impact of non-pharmaceutical interventions on curtailing the 2019 novel CoronavirusMathematical Biosciences, 2020
- Examining the effect of social distancing on the compound growth rate of COVID-19 at the county level (United States) using statistical analyses and a random forest machine learning modelPublic Health, 2020
- Impact of Social Vulnerability on COVID-19 Incidence and Outcomes in the United StatesPublished by Cold Spring Harbor Laboratory ,2020
- New Approaches for Calculating Moran’s Index of Spatial AutocorrelationPLOS ONE, 2013
- Using the Margins Command to Estimate and Interpret Adjusted Predictions and Marginal EffectsThe Stata Journal: Promoting communications on statistics and Stata, 2012
- Estimates of Crowding in Long-Term Care: Comparing Two ApproachesHerd: Health Environments Research & Design Journal, 2011
- An Investigation of the Effects of Social Desirability on the Validity of Self-Reports of Cancer Screening BehaviorsMedical Care, 2005