Pitfalls and Solutions for Reporting on COVID-19 Mortality and Prioritization of Vaccinations in Local Communities and Neighborhoods (US)

Adelina Comas-Herrera

4 years ago

Greg Arling¹, Matthew Blaser², John R. Canar ^{3, 4}, Apostolis Sambanis⁵, Michael D. Cailas³*

¹Purdue University, School of Nursing, College of Health and Human Sciences.

² United States Environmental Protection Agency, Research Associate under an inter-agency agreement with Oak Ridge Institute for Science and Education

³ Environmental and Occupational Health Sciences, School of Public Health, University of Illinois Chicago

⁴ United States Environmental Protection Agency Region V;

⁵ Health Policy and Administration, School of Public Health, University of Illinois Chicago

*Corresponding author: mihalis@uic.edu

Introduction

Over a year has passed since the first report of a COVID-19 related death and the first major COVID-19 outbreak in a Kirkland WA nursing home.[1] Even today public reporting systems on COVID-19 mortality suffer from two major pitfalls. First, they fail to distinguish between deaths occurring in long-term care facilities (LTCF) and those in other community settings, particularly individual households. This distinction is crucial because patterns of COVID-19 mortality are distinctly different between LTCFs and households, with the latter being highly sensitivity to conditions in local communities or neighborhoods. Second, social vulnerability indices being used to prioritize vaccinations and other local mitigation efforts are poor indicators of mortality risk faced by a community. As a result, use of these vulnerability indicators may end up targeting communities with relatively low COVID-19 mortality risk while shifting resources away from communities at high risk. Reporting systems that effectively address these issues are urgently needed.

Reporting of COVID-19 Mortality

COVID-19 mortality is arguable the most reliable measure for tracking COVID-19 at the community level; confirmed cases are poor indicators because of wide variation between communities in access to testing. All of the current public reporting systems have been tracking total COVID-19 mortality from the early days of the pandemic. Reporting of COVID-19 mortality among long-term care facility residents lagged behind. When this reporting began it was done separately without regard to its implications for figures on COVID-19 mortality in the total population or among persons not residing in long-term care facilities. These implications are substantial since approximately 40% of COVID-19 deaths have been in these settings both in the US and worldwide, and the percentage remained steadily high through December 2020.[2]

The Johns Hopkins Coronavirus Resource Center [3] offers the most detailed information on both daily and weekly total COVID-19 mortality at the national, state, and county levels. The Center for Disease Control and Prevention (CDC) also reports detailed information on total COVID-19 mortality.[4] Neither source offers information on COVID-19 mortality among long-term care facility residents. The Kaiser Family Foundation [5] and the New York Times [6] [7], which were also tracking total COVID mortality, began reporting national and state-level figures for COVID-19 mortality for residents of nursing and other long-term care facilities in March 2020. Perhaps spurred on by the shockingly high rates of COVID-19 mortality being reported for LTCF residents, the Center for Medicare and Medicaid Services (CMS) came out with a nursing facility data base reporting system in late May [8]. The Atlantic COVID-19 Tracking Project (CTP) followed in August with weekly long-term care mortality reports to accompany their total COVID-19 reporting[9].

Failure to Address Different Patterns of COVID-19 in LTCFs and Households

We have shown that this approach, reporting total COVID-19 deaths without regard to deaths of residents in LTCFs, can lead to serious distortions, painting an incomplete and often inaccurate picture of mortality patterns among residents of individual households.[10] Distortions are most pronounced at the local area level, particularly at the scale of the county, city, or neighborhood within a city. In our study of Cook County (Chicago) we found that spatial and temporal patterns in mortality were distinctly different between households in local neighborhoods and LTC facilities within and between neighborhoods [11]. Combining LTCF and household mortality figures can lead to confounding and an underestimation of the effects of neighborhood characteristics on COVID-19 mortality in households. Neighborhood characteristics are only weakly related to mortality among long-term care facility residents, and within neighborhoods the COVID-19 mortality rates for residents of long care facilities are only weakly correlated with rates for household members. When we separate LTCF and household mortality rates, we find a more pronounced relationship between COVID-19 mortality in households and the proportion of racial/ethnic minorities and old people, lower SES, crowding, and employment in service sector occupations (forthcoming publication).

Poorly Performing Vulnerability Indices

Another reporting weakness is in the application of social vulnerability indices being used widely by public health departments for strategizing about vaccinations and other responses to COVID-19. The CDC’s SVI [12] and the CCVI [13], two widely reported indices, have inherent weaknesses that limit their effectiveness. The CDC’s SVIs was developed primarily to address losses from a single event, such as a natural or environmental disaster, and not for a disease involving a series of events spreading over an extended time period and subject to multiple waves. The CCVI expands on the CDC’s SVI by giving greater attention to population old age (65 or older), healthcare resources, epidemiological factors, and nursing population. Yet, these factors are among many demographic, social and economic characteristics in the overall index. We found that these SVIs performed poorly in predicting the loss of life in counties or cities and in neighborhoods.[14] They tend to classify high risk areas (in terms of deaths) as areas of low vulnerability which implies low vaccination priority.

Addressing the Shortcomings of Reporting on COVID-19 Mortality

To overcome the limitations of current reporting systems, we developed prototype COVID-19 dashboards for a metropolitan center [15] (Chicago) and for counties across Midwestern states [16]. These dashboards report COVID-19 mortality in long-term care and the wider community, classify communities according to their vulnerability, and provide periodically updated figures on vaccinations (upcoming). The separation of LTCF and household COVID-19 mortality is crucial because vaccination roll-outs have been far more rapid and comprehensive in LTCFs [17], which should lead to a steeper drop in mortality rates compared to household settings. Rankings are used to categorize the geographic units: neighborhoods within the metropolitan center or counties within the state. The geographic units are ranked independently along two dimensions: COVID-19 mortality rate (number of deaths per population in either the LTCF or household setting), a multifactorial vulnerability index (currently the CDC SVI. Based on these rankings, we classify geographic units into tertiles or quartiles that visualize their classifications or combination of classifications. For example, one pair of dashboard maps shows Chicago neighborhoods classified by COVID-19 death rates (LTCF and household) and vulnerability index.[18] Neighborhoods with both a history of COVID-19 deaths and high vulnerability can be targeted for vaccinations or other mitigation efforts. Another forthcoming map will show counties in Illinois classified by COVID-19 mortality (LTCF and household) and percentage of the population with COVID-19 vaccinations. With bi-weekly updates the maps can be used for both prioritizing rollout and monitoring the success of vaccination programs. Enhancements to the dashboard will include time series graphs for the three dimensions in LTCF and household populations. These graphs can be used to visualize joint relationships in COVID-19 mortality trends between LTCF and households, and patterns of vaccination uptake and mortality.

SUGGESTED CITATION

Arling G, Blaser M, Canar JR, Sambanis A, Cailas MD (2021) Pitfalls and Solutions for Reporting on COVID-19 Mortality and Prioritization of Vaccinations in Local Communities and Neighborhoods. Article in LTCcovid.org, International Long-Term Care Policy Network, CPEC-LSE.

Footnotes:

[1] https://www.cdc.gov/media/releases/2020/s0229-COVID-19-first-death.html

[2] https://ltccovid.org/wp-content/uploads/2021/02/LTC_COVID_19_international_report_January-1-February-1-1.pdf

[3] https://coronavirus.jhu.edu

[4] https://www.cdc.gov/nchs/nvss/covid-19.htm

[5] https://www.kff.org/coronavirus-covid-19/issue-brief/state-covid-19-data-and-policy-actions/

[6] https://www.nytimes.com/interactive/2020/us/coronavirus-us-cases.html

[7] https://www.nytimes.com/interactive/2020/us/coronavirus-nursing-homes.html

[8] https://data.cms.gov/stories/s/COVID-19-Nursing-Home-Data/bkwz-xpvg/

[9] https://covidtracking.com/nursing-homes-long-term-care-facilities

[10] https://p3rc.uic.edu/wp-content/uploads/sites/561/2020/08/Analyzing_COVID-19_Methods_508-1.pdf

[11] https://journals.uic.edu/ojs/index.php/ojphi/article/view/11506

[12] https://www.atsdr.cdc.gov/placeandhealth/svi/index.html

[13] https://precisionforcovid.org/ccvi

[14] https://indigo.uic.edu/articles/preprint/A_Data_Driven_Approach_for_Prioritizing_COVID-19_Vaccinations_in_the_Midwestern_United_States_Untitled_Item/14136056

[15] https://indigo.uic.edu/articles/report/MCVD_Prioritizing_Vaccinations_in_Cook_County_Illinois/14233601

[16] https://indigo.uic.edu/articles/preprint/A_Data_Driven_Approach_for_Prioritizing_COVID-19_Vaccinations_in_the_Midwestern_United_States_Untitled_Item/14136056

[17] https://www.ahcancal.org/Data-and-Research/Center-for-HPE/Documents/CHPE-Report-Vaccine-Effectiveness-Feb2021.pdf

[18]https://indigo.uic.edu/articles/report/MCVD_Prioritizing_Vaccinations_in_Cook_County_Illinois/14233601

Sources for Footnotes

First confirmed deaths from COVID-19 – CDC: https://www.cdc.gov/media/releases/2020/s0229-COVID-19-first-death.html

Johns Hopkins Coronavirus Resource Center: https://coronavirus.jhu.edu

NY Times Report: https://www.nytimes.com/interactive/2020/us/coronavirus-us-cases.html and https://www.nytimes.com/interactive/2020/us/coronavirus-nursing-homes.html

CDC COVID Mortality – Provisional death certificates – Place of death along with other characteristics: https://www.cdc.gov/nchs/nvss/covid-19.htm

CMS reporting system and data: https://data.cms.gov/stories/s/COVID-19-Nursing-Home-Data/bkwz-xpvg/

KFF Reporting: https://www.kff.org/coronavirus-covid-19/issue-brief/state-covid-19-data-and-policy-actions/

Total COVID-19 deaths nationally and by state, and LTC

CTP

https://covidtracking.com/data

https://covidtracking.com/nursing-homes-long-term-care-facilities

CDC SVI

https://www.atsdr.cdc.gov/placeandhealth/svi/index.html

Social vulnerability refers to the potential negative effects on communities caused by external stresses on human health. Such stresses include natural or human-caused disasters, or disease outbreaks. Reducing social vulnerability can decrease both human suffering and economic loss. CDC Social Vulnerability Index (CDC SVI) uses 15 U.S. census variables to help local officials identify communities that may need support before, during, or after disasters.

CCVI

https://precisionforcovid.org/ccvi

PVI – NIH

https://www.niehs.nih.gov/research/programs/coronavirus/covid19pvi/details/index.cfm

UIC-PHGIS Articles and Story Maps

https://journals.uic.edu/ojs/index.php/ojphi/article/view/11506

A Data Driven Approach for Prioritizing COVID-19 Vaccinations in the Midwestern United States: https://indigo.uic.edu/articles/preprint/A_Data_Driven_Approach_for_Prioritizing_COVID-19_Vaccinations_in_the_Midwestern_United_States_Untitled_Item/14136056

https://p3rc.uic.edu/wp-content/uploads/sites/561/2020/08/Analyzing_COVID-19_Methods_508-1.pdf

https://indigo.uic.edu/articles/report/Midwest_Comprehensive_Visualization_Dashboards_COVID-19_MCVD/13650440/1

https://ltccovid.org/2020/12/15/article-summary-a-second-wave-of-covid-19-in-chicago-usa-a-case-for-urgent-action/

https://ltccovid.org/project/a-comprehensive-analytic-framework-for-covid-19-mortality-in-long-term-care-facilities-and-neighborhoods-applicable-to-major-metropolitan-centers/

https://pubhealthgis.uic.edu/covid-19-dashboard-maps/