Temporal Analysis of Social Determinants Associated with COVID-19 Mortality

Shayom Debopadhaya, John Erickson, Kristin P. BennettmedRxiv 2021 · DOI

Abstract This study examines how social determinants associated with COVID-19 mortality change over time. Using US county-level data from July 5 and December 28, 2020, the effect of 19 high-risk factors on COVID-19 mortality rate was quantified at each time point with negative binomial mixed models. Then, these high-risk factors were used as controls in two association studies between 40 social determinants and COVID-19 mortality rates using data from the same time points. The results indicate that counties with certain ethnic minorities and age groups, immigrants, prevalence of diseases like pediatric asthma and diabetes and cardiovascular disease, socioeconomic inequalities, and higher social association are associated with increased COVID-19 mortality rates. Meanwhile, more mental health providers, access to exercise, higher income, chronic lung disease in adults, suicide, and excessive drinking are associated with decreased mortality. Our temporal analysis also reveals a possible decreasing impact of socioeconomic disadvantage and air quality, and an increasing effect of factors like age, which suggests that public health policies may have been effective in protecting disadvantaged populations over time or that analysis utilizing earlier data may have exaggerated certain effects. Overall, we continue to recognize that social inequality still places disadvantaged groups at risk, and we identify possible relationships between lung disease, mental health, and COVID-19 that need to be explored on a clinical level.

Introduction

Early COVID-19 research often provided static snapshots of risk factors, identifying vulnerable populations without considering how these risks evolved over time. As the pandemic progressed, shifting from urban centers to rural areas and coinciding with various public health interventions, the profile of high-risk communities changed. This study addresses the need for a temporal analysis to distinguish between persistent social threats and those mitigated by systemic protections.

Methodology

The researchers conducted an ecological study across 3,093 US counties using negative binomial mixed models (NBMM) at two distinct time points: July 5, 2020, and December 28, 2020. The analysis utilized a two-stage pipeline to ensure robustness and avoid the "false discovery" of associations.

Figure 1: Study methodology flowchart illustrating the two-stage pipeline for identifying high-risk factors and screening social determinants.

In the first stage, 19 established high-risk factors—including racial distribution, age (65+), comorbidities (COPD, diabetes), and state-level mask mandates—were quantified. In the second stage, these factors served as controls while screening 40 additional social determinants of health (SDoH). Results were reported as Mortality Rate Ratios (MRR), indicating the change in mortality per one standard deviation increase in the explanatory variable.

The study revealed significant shifts in how socioeconomic factors influenced mortality. Early in the pandemic, higher median income was associated with increased mortality (1.23 MRR in July), likely reflecting the initial impact on wealthy, densely populated international travel hubs. By December, this trend reversed (0.95 MRR), with higher income becoming a protective factor as the virus spread into lower-income rural regions.

Figure 2: Mortality Rate Ratios (MRR) for socioeconomic and quality of life social determinants, comparing data from July 5 and December 28, 2020.

Persistent risks were also identified. A lack of high school education remained a consistent predictor of higher mortality across both time points (1.15 MRR in July; 1.13 MRR in December). Conversely, the impact of air quality (Average Daily PM2.5) diminished significantly from a 1.37 MRR in July to 1.11 MRR in December, suggesting that while environmental exposure is a risk, other factors became more dominant as the pandemic matured.

Demographic and Clinical Insights

Racial disparities were stark in July, particularly for Hispanic (1.25 MRR) and Native American (1.18 MRR) populations. While these ratios decreased by year-end, they remained statistically significant. Interestingly, the percentage of the population under 18 emerged as a significant predictor by December (1.11 MRR), potentially highlighting the role of younger demographics in community transmission.

The study also noted a "Lung Disease Paradox." While adult chronic lung disease is a known individual risk factor, counties with higher prevalence actually saw decreased mortality in December. In contrast, pediatric asthma was linked to increased mortality, suggesting a need for clinical investigation into the biological interactions between different respiratory conditions and COVID-19.

Conclusion and Limitations

This temporal analysis underscores that social inequality continues to place disadvantaged groups at risk, though the specific drivers of mortality evolved throughout 2020. While the study is limited by the ecological fallacy—describing community-level rather than individual-level risks—it identifies critical areas for future clinical research, particularly regarding mental health, pediatric asthma, and the long-term efficacy of public health policies in protecting vulnerable populations.

Summary updated

Abstract

ABSTRACT This study examines how social determinants associated with COVID-19 mortality change over time. Using US county-level data from July 5 and December 28, 2020, the effect of 19 high-risk factors on COVID-19 mortality rate was quantified at each time point with negative binomial mixed models. Then, these high-risk factors were used as controls in two association studies between 40 social determinants and COVID-19 mortality rates using data from the same time points. The results indicate that counties with certain ethnic minorities and age groups, immigrants, prevalence of diseases like pediatric asthma and diabetes and cardiovascular disease, socioeconomic inequalities, and higher social association are associated with increased COVID-19 mortality rates. Meanwhile, more mental health providers, access to exercise, higher income, chronic lung disease in adults, suicide, and excessive drinking are associated with decreased mortality. Our temporal analysis also reveals a possible decreasing impact of socioeconomic disadvantage and air quality, and an increasing effect of factors like age, which suggests that public health policies may have been effective in protecting disadvantaged populations over time or that analysis utilizing earlier data may have exaggerated certain effects. Overall, we continue to recognize that social inequality still places disadvantaged groups at risk, and we identify possible relationships between lung disease, mental health, and COVID-19 that need to be explored on a clinical level. CCS CONCEPTS Applied computing → Health informatics .
BibTeX
@misc{debopadhaya2021temporal,
  author = {Shayom Debopadhaya and John Erickson and Kristin P. Bennett},
  title = {{Temporal Analysis of Social Determinants Associated with COVID-19 Mortality}},
  year = {2021},
  doi = {10.1101/2021.06.22.21258971},
  howpublished = {medRxiv},
}
Figure