A Preemptive Strike on Future Pandemics

A BU-led research team publishes new findings that show how detecting and correctly assigning deaths can help mitigate the impact of pandemics

By Maureen Stanton, Hariri Institute

COVID-19 caused catastrophe and chaos across the globe. According to some disease forecasters, the world could face another COVID-like pandemic within 10 years. To improve our response to future pandemics, it’s imperative to strengthen essential systems.

Tracking death counts attributed to a fast-moving lethal virus like COVID-19 is fundamental to helping policy makers understand a pandemic’s impact and guide the allocation of scarce resources. To help improve pandemic preparedness, a BU-led research team has been working to uncover uncounted COVID-19 fatalities. Their new study reveals that the total number of COVID-19 deaths was likely 16% higher than the official U.S. COVID-19 mortality statistics, and that the undercounting of these deaths persisted throughout the first 30 months of the pandemic. One theory for this is the wide disparities in how causes of deaths are determined across the country. This highlights the critical need for reforms to the death investigation system to ensure data accuracy and equitable allocation of resources. 

The study appears online in PNAS, the official journal of the National Academy of Sciences.  

“Our research shows that the death investigation system likely missed many COVID-19 deaths even beyond the initial phase of the pandemic,” says Andrew Stokes, PhD, an associate professor of Global Health, affiliate faculty of the Hariri Institute, and corresponding author of the study. “This suggests that limited preparedness to a new disease was only part of the story and that the underlying issues with the system persisted beyond the initial emergency. Leveraging the availability of county-level data and spatial modeling techniques, we were also able to identify areas across the U.S. where the undercounts were most pronounced.” 

To determine the total number of uncounted COVID-19 deaths, the researchers developed statistical models that assessed the difference between deaths reported as COVID-19, and natural deaths not attributed to COVID-19. To avoid skewing results, the statistical model excluded causes of death due to external factors such as accidents, suicides, homicides, and others causes where COVID-19 would not likely be directly implicated. 

The researchers used death certificate data to compare cause of death reports in over 3,000 counties across the United States from March 2020 to August 2022. They estimated that 1.2 million excess natural-cause deaths occurred in US counties during the first 30 months of the pandemic. Excess deaths refer to the difference between the number of observed deaths and expected deaths during pre-pandemic times. 163,000 of these excess natural-cause deaths were not reported to COVID-19. Regionally, the gap between excess natural-cause mortality and reported COVID-19 mortality was largest in nonmetropolitan counties, the West, and the South. 

“Rapid detection of non-typical mortality patterns could pinpoint the emergence of local novel disease clusters and become an important tool for more effective pandemic mitigation,” says Yannis Paschalidis, distinguished professor of engineering, director of the Hariri Institute, and principal investigator of a National Science Foundation project at Boston University on Predicting and Preventing Epidemic to Pandemic Transitions. 

Associating excess natural-cause mortality with reported COVID-19 Deaths 

While previous studies have attempted to untangle excess deaths unrelated to COVID-19 infection from those related to the infection, this is the first paper to tackle the complex problem with county-level data, and the first to employ “time correlation analysis” whereby the researchers tracked and correlated time periods and peak reporting counts of natural-cause deaths and COVID deaths. 

“The goal of the time correlation analysis was to understand how likely it is that mechanisms other than a COVID-19 infection are responsible for excess mortality beyond deaths assigned to COVID-19,” says lead author Eugenio Paglino, PhD student of demography and sociology at the University of Pennsylvania. “If we know that on average there are 100 deaths assigned to diabetes in a year but observe 130 deaths during the first year of the pandemic, the most likely explanation is that 30 of these deaths are related to the pandemic either directly or indirectly.” 

Using time correlation analysis, the researchers determined that many excess deaths reported to non-COVID-19 natural causes were actually unrecognized COVID-19 deaths, and that contrary to current literature, the gaps were not limited to the start of the pandemic but persisted throughout. 

“This work is important because our ability to detect and correctly assign deaths during an epidemic goes to the heart of our understanding of the disease and how we organize our response,” says Dr. Nahid Bhadelia, MD, founding director of BU Center on Emerging Infectious Diseases at Boston University. 

Pinpointing Communities Most Impacted 

The investigators sought to understand the causes for such gaps by analyzing the location, time periods, and causes of deaths reported. The data suggests that large gaps may be attributed to divergent death reporting processes. 

It is broadly recognized that death investigation practices vary greatly from county to county. While in-hospital deaths are typically certified by medically-trained examiners, many death investigations are conducted by coroners who lack medical training, such as justices of the peace and elected officials who may have political motivations to not count COVID-19 as a cause of death. Death investigation systems also vary in budgets, equipment, and training. These inconsistencies can negatively affect the accuracy of mortality data on deaths and death rates. 

By observing the mortality patterns at the county level, the study reveals the pandemic’s impact on specific communities. This information is vital to help policymakers identify opportunities to target resources to the communities most impacted by a pandemic. Moreover, it demonstrates that to help ensure effective and equitable response in future emergencies, it’s critical to consider geographic variation in the quality of mortality data and the potential for delayed or incomplete data. 

“Geographic variation in the quality of cause of death reporting not only adversely affected pandemic response in areas where COVID-19 deaths were underreported, but it also reduced the accuracy of our national surveillance data and modeling,” says Katherine Hempstead, a senior policy adviser at the Robert Wood Johnson Foundation. “Standardizing and improving death investigation and certification should be a public health priority.”  

Adds study co-author Maria Glymour, chair and professor of epidemiology at the BU School of Public Health, “Resources and commitment to ensure accurate death investigations are essential. These findings of uncounted COVID-19 deaths indicate those resources are lacking in many communities.”

For more information: access a permanent repository with project replication code, further details about the data sources used, and estimates generated through the study available at https://osf.io/sqdbk/.