Abstract
Background.
Coronavirus disease 2019 (COVID-19) case and death rates have been variable across geography and time; an understanding of community social vulnerability and variation in COVID-19 outcomes is needed to better respond to acute public health needs and prioritize post-pandemic recovery.
Methods.
We analyzed records of confirmed COVID-19 cases (N = 106 037) and deaths (N = 2107) in Fulton County, Georgia, from 4 March 2020 to 11 November 2021. We developed the social vulnerability index-plus (SVI+) as a composite measure of 19 census tract-level social and health indicators summarizing community vulnerability. Ecological analyses included linear regression estimates of differences in weekly COVID-19 case and death rates per 100 000 population by SVI+ across 201 census tracts. Multilevel analyses included log-binomial regression estimates of relative risk of individual-level death among confirmed cases according to the SVI+. Calendar periods included Pre-Vaccine (March 2020 to 15 December 2020), Post-Vaccine/Pre-Delta variant surge (16 December 2020 to 30 June 2021), and post-vaccine/Delta variant surge (1 July 2021 to 11 November 2021).
Results.
We observed a positive association between SVI+ and both COVID-19 census tract-level case and death rates. Analyses by calendar period indicated that the strongest association between SVI+ and both case rates, and death rates was in the post-vaccine/Delta variant surge period. Analyses of individual-level death among cases showed a positive association between SVI+ and COVID-19 death after accounting for age, race, and sex. SVI+ and individual-level death were positively associated in all calendar periods.
Conclusions.
Even after vaccine availability, communities with higher social vulnerability experienced worse COVID-19 outcomes. The findings reiterate the importance of addressing social determinants by increasing public health efforts in vulnerable communities to mitigate health disparities in future pandemics.
Keywords: COVID-19, social vulnerability, determinants of health, community-level, health disparities
The United States experienced higher coronavirus disease 2019 (COVID-19) case and death rates than any other country in the world, but these rates are not evenly distributed across the population [1]. Racial disparities in COVID-19 incidence and mortality have been reported across the United States and data have shown that Black people have disproportionate COVID-19 hospitalization and death rates [2, 3]. Research has suggested that this uneven distribution of COVID-19 may be related to underlying disparities in social determinants of health [4] or social vulnerability [5-8].
The majority of studies investigating COVID-19 case and death rates were conducted at the county level [6, 8]. Although informative, county-level studies do not account for the immense variability in social vulnerability or COVID-19 outcomes within a county. Furthermore, temporal changes in the dominant severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variant and availability of vaccinations may also impact these associations. For example, socially vulnerable communities may experience under-coverage of COVID-19 vaccination, exacerbating preexisting disparities. However, the limited research on inequities in COVID-19 related outcomes at the sub-county level (eg, Census tract-level [9-11]) fail to consider temporal variation in the relationship between community vulnerability and COVID-19 outcomes and have yielded mixed findings.
The Southeastern United States was impacted heavily by COVID-19, and many southern states reported the largest numbers of COVID-19 deaths nationally [8]. In Georgia, the burden of COVID-19 deaths was highly concentrated by county; counties with higher percentages of socially vulnerable populations were impacted the most by the pandemic [12]. Describing the relationships between community-level social vulnerability, vaccination coverage, and COVID-19 outcomes over time may shed light on differential impacts of the pandemic on communities and individuals, as well as inform future rehabilitation efforts, in this high-risk setting. We examined the impact of community social vulnerability on COVID-19 outcomes across time in Fulton County, a large metropolitan county in the US south.
METHODS
Study Design and Data Sources
We conducted an ecological investigation of COVID-19 case and death rates and multilevel investigation of death among COVID-19 cases in Fulton County, Georgia. Individual-level data on all laboratory-confirmed COVID-19 cases were obtained from the Georgia Department of Public Health (GDPH) State Electronic Notifiable Disease Surveillance System (SENDSS). COVID-19 deaths were ascertained through periodic linkage with Georgia Department of Health Vital Records. Analyses were restricted to confirmed COVID-19 cases in Fulton County reported to SENDSS between 4 March 2020, and 11 November 2021, before the Omicron-predominant period. COVID-19 cases were followed for death until 25 June 2022. Vaccination data were obtained weekly from 10 June 2021, via the GDPH COVID-19 Vaccination dashboard Github [13]. Social vulnerability index (SVI) data by census tract were downloaded from the publicly available US Centers for Disease Control and Prevention (CDC)/Agency for Toxic Substances and Disease Registry 2018 Georgia database [14]. Additional data sources included Fulton County Board of Health (FCBOH) databases for age-standardized chronic disease rates and the 2018 American Community Survey for uninsured rates [15]. Individual-level COVID-19 outcomes were merged with census tract-level social vulnerability and vaccination data to create a multilevel database of 201 census tracts and 112 457 individuals.
This analysis was conducted for public health surveillance purposes in response to the COVID-19 emergency. As such, this analysis was deemed exempt from Institutional Review Board review by the GDPH Review Board.
COVID-19 Outcomes
We examined confirmed COVID-19 cases and deaths per 100 000 population at the census tract level. Numerator data (cases and deaths) were obtained from GDPH SENDSS, whereas denominator data were drawn from the Census 2019 population estimates. COVID-19 cases were defined as confirmed (polymerase chain reaction [PCR] positive) and antigen-positive cases reported through electronic lab reporting, SENDSS, and provider reports. COVID-19 deaths included those where COVID-19 was determined to be the primary cause or contributed to death [16].
Individual-level COVID-19 death were also analyzed.
Community Measures
The primary study exposure was the SVI+, a composite measure of 19 social vulnerability indicators at the census tract level. The SVI+ built upon the CDC’s SVI, which originally included 15 census tract indicators covering 4 themes: socioeconomic status, household composition and disability, minority status and language, and housing and transportation. The SVI+ additionally incorporated a fifth health-related theme with 4 indicators: age-adjusted hospital discharge rates for diabetes, major cardiovascular disease, and cancers, as well as the percentage of uninsured persons. The inclusion of the health theme was motivated by data showing that these specific morbidities and lack of healthcare access can lead to severe COVID-19 outcomes [17, 18]. The 5 themes were standardized and averaged into the SVI+ measure, where a higher score indicated greater vulnerability. The primary analysis focused on the SVI+ summary and its 5 themes, with additional analyses examining the 19 individual indicators separately. SVI+ indicators were treated as time-fixed in the analysis.
Apart from SVI+ indicators, we included cumulative vaccination coverage by census tract, defined as the proportion of residents with 2 doses of the Pfizer/Moderna vaccines or 1 dose of the Janssen vaccine. Vaccination data were time-varying and available starting July 2021.
Individual Covariates
In individual-level death analyses, we included age, sex, race, and ethnicity. Approximately 90% of the Fulton County population identifies as White alone or Black alone, 8% as Asian alone, and <2% as Native American or another racial group [19]. We classified Asian, Native American, and multiracial individuals as a composite “other” racial and ethnic group category because we did not have sufficient sample size to analyze them separately in our data set. We examined race as a social rather than biological construct, interpreting any association between Black race and mortality as a reflection of social factors. These factors include lower average socioeconomic position, worse neighborhood conditions, experiences of racism, and other systemic risks that contribute to poorer health outcomes [20].
Statistical Analysis
We described the distribution of vulnerability indicators and COVID-19 outcomes across census tracts, categorizing them as least, medium, or most vulnerable based on SVI+ tertiles. Choropleth maps and scatter plots illustrated the spatial distribution of SVI themes, overall SVI+, COVID-19 outcomes (case and death rates), and vaccination rates.
Ecological analyses included linear regression models to evaluate associations between social vulnerability measures and case/death rates at the census tract level. Model coefficients represent rate differences across census tracts. We first examined bivariate associations of vulnerability measures and vaccination coverage with COVID-19 case and death rates. Adjusted models included all 5 SVI themes and vaccination coverage. To assess how social vulnerability related to COVID-19 outcomes over time, we defined 3 calendar periods: pre-vaccine (March 2020 to 15 December 2020), post-vaccine/pre-Delta variant surge (16 December 2020 to 30 June 2021), and post-vaccine/Delta variant surge (1 July 2021 to November 2021).
Multilevel analyses of community social vulnerability and individual death were conducted using Poisson models with generalized estimating equations to account for clustering within census tracts. Model coefficients were exponentiated to report risk ratios for death according to social vulnerability metrics.
All analyses were conducted using SAS v9.4 (Cary, North Carolina, USA).
RESULTS
Table 1 shows the distribution of 19 indicators comprising the SVI+ by level of vulnerability across 201 census tracts in Fulton County, Georgia. The SVI+ mean was 0.17 (range: 0.05–0.33) for least vulnerable tracts, 0.50 (range: 0.34–0.66) for medium vulnerable tracts, and 0.83 (range: 0.67–1.00) for most vulnerable tracts. Indicators of social disadvantage—particularly indicators of socioeconomic composition (eg, percentage of persons below poverty)—and poor population-level health status (eg, hospital discharge rates for diabetes) tended to be highest in the most vulnerable census tracts.
Table 1.
Distribution of Community Social Vulnerability Indicators and COVID-19 Outcomes by Level of Vulnerability, Fulton County, Georgia
| Social Vulnerability Measures, 2018 |
Fulton County Mean (Min, Max) |
Least Vulnerable Tracts(n = 67) Mean (Min-Max) |
Medium Vulnerable Tracts (n = 66) Mean (Min-Max) |
Most Vulnerabl Tracts (n = 68) Mean (Min-Max) |
|---|---|---|---|---|
| SVI+ summary theme | 0.5 (0.0, 1.0) | 0.2 (0.0, 0.3) | 0.5 (0.3, 0.7) | 0.8 (0.7, 1.0) |
| Socioeconomic status themea | 0.4 (0.0, 1.0) | 0.1 (0.0, 0.2) | 0.4 (0.0, 1.0) | 0.8 (0.4, 1.0) |
| Percentage of persons below poverty estimate | 18.9 (1.1, 79.8) | 5.8 (1.1, 20.4) | 16.6 (1.9, 55.0) | 34.4 (15.5, 79.8) |
| Percentage of civilian (age 16+) unemployed estimate | 7.9 (0.0, 35.5) | 3.1 (0.0, 8.8) | 6.0 (0.0, 22.3) | 14.6 (3.5, 35.5) |
| Per capita income estimate, $1000 | 42.5 (44.7, 145.9) | 70.9 (40.2, 146.0) | 37.8 (44.7, 102.3) | 18.5 (7.30, 31.6) |
| Percentage of persons with no high school diploma (age 25+) estimate | 9.6 (0.0, 44.6) | 2.4 (0.0, 7.2) | 9.8 (0.9, 44.6) | 16.6 (6.5, 34.5) |
| Household composition and disability themea | 0.4 (0.0, 1.0) | 0.1 (0.0, 0.5) | 0.3 (0.0, 0.9) | 0.7 (0.1, 1.0) |
| Percentage of persons aged 65+ estimate | 12.0 (0.0, 90.5) | 12.3 (1.9, 30.6) | 11.4 (0.0, 90.5) | 12.3 (1.0, 31.9) |
| Percentage of persons aged 17 and younger estimate | 21.2 (0.0, 59.3) | 19.8 (2.7, 32.4) | 18.7 (0.0, 32.2) | 25.0 (7.2, 59.3) |
| Percentage of persons of civilian noninstitutionalized population with a disability estimate | 11.4 (1.0, 36.5) | 6.2 (2.5, 13.2) | 10.8 (1.0, 36.5) | 17.3 (4.9, 31.8) |
| Percentage of single parent households with children under 18 estimate | 10.3 (0.0, 56.4) | 4.3 (0.0, 12.6) | 9.1 (0.0, 30.7) | 17.6 (3.0, 56.4) |
| Minority status and language themea | 0.6 (0.0, 1.0) | 0.4 (0.0, 0.9) | 0.6 (0.1, 1.0) | 0.7 (0.4, 1.0) |
| Percentage minority (all persons except White, non-Hispanic) estimate | 61.6 (3.3, 100.0) | 27.5 (3.3, 64.6) | 64.9 (22.9, 100.0) | 92.8 (62.9, 100.0) |
| Percentage of persons (age 5+) who speak English less than well estimate | 1.8 (0.0, 22.4) | 1.3 (0.0, 6.6) | 2.6 (0.0, 16.4) | 1.5 (0.0, 22.4) |
| Housing type and transportation themea | 0.5 (0.0, 1.0) | 0.3 (0.0, 0.8) | 0.5 (0.1, 1.0) | 0.7 (0.2, 1.0) |
| Percentage of housing in structures with 10 or more units estimate | 31.4 (0.0, 95.6) | 30.4 (0.0, 94.9) | 35.8 (0.5, 95.6) | 28.4 (0.0, 87.6) |
| Percentage of mobile homes estimate | 0.6 (0.0, 15.0) | 0.4 (0.0, 2.9) | 0.6 (0.0, 10.4) | 0.9 (0.0, 15.0) |
| Percentage of households with more people than rooms estimate | 1.9 (0.0, 11.7) | 0.8 (0.0, 4.9) | 2.0 (0.0, 11.7) | 2.9 (0.0, 11.5) |
| Percentage of households with no vehicle available estimate | 14.2 (0.0, 52.6) | 4.1 (0.0, 16.4) | 11.7 (0.7, 33.8) | 26.8 (6.2, 52.6) |
| Percentage of persons in group quarters estimate | 3.9 (0.0, 91.3) | 1.3 (0.0, 40.5) | 7.6 (0.0, 91.3) | 2.8 (0.0, 55.2) |
| Health theme | 0.5 (0.0, 1.0) | 0.2 (0.0, 0.5) | 0.5 (0.1, 0.9) | 0.8 (0.5, 1.0) |
| Hospital discharge rates for cancers | 225.0 (67.9, 790.0) | 172.2 (83.5, 309.0) | 212.2 (67.9, 790.0) | 290.4 (136.9, 467.8) |
| Hospital discharge rates for diabetes | 62.0 (3.6, 195.2) | 18.3 (3.6, 56.3) | 52.6 (8.7, 161.4) | 114.1 (42.8 195.2) |
| Hospital discharge rates for cardiovascular conditions | 1350.1 (99.7, 7867.6) | 628.5 (99.7, 1517.9) | 1337.7 (364.8, 7867.6) | 2086.3 (875.0, 3784.7) |
| Percentile percentage of persons uninsured | 11.2 (0.0, 40.4) | 4.5 (0.0, 12.1) | 12.2 (2.5, 37.2) | 16.8 (9.1, 40.4) |
| COVID-19 Outcomes, March 2020- November 2021 |
Median (IQR) |
Median (IQR) |
Median (IQR) |
Median (IQR) |
| Cases per 100 000 population | 9841 (3298) | 8697 (3229) | 9884 (3091) | 10 740 (2674) |
| Deaths per 100 000 population | 124 (179.5) | 60 (73.1) | 110 (108.4) | 239 (156.0) |
Each theme was derived from the percentile of indicators listed below it.
Abbreviations: COVID-19, coronavirus disease 2019; SVI+, social vulnerability index-plus.
Indicates the theme and component indicators were directly taken from the US Centers for Disease Control and Prevention (CDC) SVI.
COVID-19 Case Rates
Between March 2020 and November 2021, there were a total of 106 037 confirmed COVID-19 cases recorded by the Georgia Department of Public Health. There was substantial variation in cumulative COVID-19 cases per 100 000 between the most and least vulnerable census tracts (Table 1). Median case rates were highest in the most vulnerable tracts (10 740, interquartile range [IQR]: 2674) and medium vulnerable tracts (9884, IQR: 3091) and were lowest in the least vulnerable tracts (8697, IQR: 3229). COVID-19 case rates were generally higher in southern and western census tracts (Figure 1).
Figure 1.
Spatial distribution of COVID-19 outcomes, vaccination coverage, and social vulnerability in Fulton County, Georgia. This figure displays census tract-level choropleth maps for key COVID-19 outcomes and social vulnerability indicators in Fulton County, Georgia. Top Left: Cumulative COVID-19 case rates per 100 000 population from 4 March 2020 to 11 November 2021. Top Right: Cumulative COVID-19 death rates per 100 000 population for the same period. Bottom Left: Percentage of the population fully vaccinated against COVID-19 from 10 June 2021 to 11 November 2021. Bottom Right: The CDC Social Vulnerability Index (SVI+) composite score for 2018, grouped into terciles representing least, medium, and most vulnerable tracts. Higher case and death rates, lower vaccination coverage, and greater social vulnerability appear geographically clustered in southern census tracts, highlighting potential inequities in the local pandemic response. Abbreviations: CDC, Centers for Disease Control and Prevention; COVID-19, coronavirus disease 2019; SVI+, social vulnerability index-plus.
COVID-19 Death Rates
There were total of 2107 deaths recorded by the Georgia Department of Public Health during the study period. Like case rates, we observed variation in cumulative COVID-19 death rates: median deaths rates per 100 000 were highest in most vulnerable tracts (239, IQR: 156.0) and medium vulnerable tracts (110, IQR: 108.4), and lowest in least vulnerable tracts (60, IQR: 73.1). Death rates were also generally higher in the southern and western census tracts.
The southern and western tracts also had relatively lower coverage of the vaccine and higher levels of social vulnerability (Figure 1). The distribution of individual SVI+ themes and overall SVI indicators similarly varied geographically by census tract (Supplementary Figure 1).
Census Tract-level Association Between Social Vulnerability and COVID-19 Case Rates
Across the full study period, there was a positive association between vulnerability and COVID-19 case rates (Table 2). There were on average 45 more COVID-19 cases per 100 000 population for every unit higher of the SVI+ summary (SVI+ summary crude rate difference [RD]: 45.5; 95% confidence interval [CI]: 29.0, 62.0). Similarly, there were strong positive associations between each of the SVI+ subthemes and COVID-19 case rates across census tracts, with the strongest association between the health subtheme and case rates (case RD: 57.7; 95% CI: 36.7, 78.6).
Table 2.
Census Tract-level Associations of Community Social Vulnerability With COVID-19 Case and Death Rates, 4 March 2020 to 11 November 2021
| Vulnerability Measures |
Case Rate Differences (95% CI) |
P Value |
Death Rate Differences (95% CI) |
P Value |
|---|---|---|---|---|
| Overall | ||||
| SVI+ summary | 45.5 (28.9, 62.0) | <.01 | 5.3 (3.7, 6.9) | <.01 |
| Socioeconomic theme | 35.4 (20.4, 50.3) | <.01 | 4.1 (2.4, 5.9) | <.01 |
| Household composition and disability theme | 35.9 (21.3, 50.5) | <.01 | 4.0 (2.6, 5.5) | <.01 |
| Minority status and language theme | 2.8 (−16.1, 21.7) | 0.77 | 1.2 (−0.4, 2.7) | .14 |
| Housing type and transportation | 29.2 (9.8, 48.6) | <.01 | 3.2 (1.1, 5.3) | <.01 |
| Health theme | 57.7 (36.7, 78.6) | <.01 | 6.0 (3.5, 8.6) | <.01 |
| Percentage of residents with 2 vaccines | N/A | N/A | N/A | N/A |
| Pre-vaccinea | ||||
| SVI+ summary | −73.2 (−101.0, −45.9) | <.01 | 4.6 (2.3, 6.9) | <.01 |
| Socioeconomic theme | −70.7 (−95.2, −46.1) | <.01 | 2.9 (0.8, 4.9) | <.01 |
| Household composition and disability theme | −45.6 (−71.4, −19.9) | <.01 | 3.9 (1.7, 6.2) | <.01 |
| Minority status and language theme | −65.7 (−101.0, −30.8) | <.01 | 0.6 (−1.8, 2.9) | .64 |
| Housing type and transportation theme | −40.5 (−74.1, −6.94) | 0.02 | 4.3 (1.6, 7.0) | <.01 |
| Health theme | −55.3 (−85.8, −24.9) | <.01 | 4.1 (1.7, 6.6) | <.01 |
| Percentage of residents with 2 vaccines | n/a | n/a | n/a | n/a |
| Post-vaccine/Pre-Delta variant surgeb | ||||
| SVI+ summary | 36.0 (23.4, 48.7) | <.01 | 3.6 (1.9, 5.3) | <.01 |
| Socioeconomic theme | 33.0 (22.0, 44.1) | <.01 | 3.2 (1.0, 5.3) | <.01 |
| Household composition and disability theme | 16.3 (6.5, 26.1) | <.01 | 2.7 (1.4, 4.0) | <.01 |
| Minority status and language theme | 3.9 (−8.1, 15.9) | 0.52 | 0.3 (−1.0, 1.5) | .66 |
| Housing type and transportation | 27.4 (13.5, 41.3) | <.01 | 1.4 (−0.6, 3.4) | .18 |
| Health teme | 45.4 (29.8, 60.9) | <.01 | 4.9 (1.5, 8.3) | <.01 |
| Percent of residents w/2 vaccines | n/a | n/a | n/a | n/a |
| Post-vaccine/Delta variant surgec | ||||
| SVI+ summary | 143.9 (120.9, 166.9) | <.01 | 8.9 (5.8, 11.8) | <.01 |
| Socioeconomic theme | 117.7 (96.55, 138.9) | <.01 | 7.3 (4.2, 10.4) | <.01 |
| Household composition and disability theme | 114.2 (93.2, 135.2) | <.01 | 6.0 (3.6, 8.4) | <.01 |
| Minority status and language theme | 54.61 (26.0, 83.20) | <.01 | 3.5 (0.7, 6.3) | .02 |
| Housing type and transportation | 84.0 (54.8, 113.2) | <.01 | 4.0 (0.2, 7.7) | .04 |
| Health theme | 153.7 (123.5, 183.8) | <.01 | 10.7 (5.8, 15.7) | <.01 |
| Percentage of residents with 2 vaccines | −2.1 (−2.7, −1.5) | <.01 | −0.1 (−0.2, 0.1) | .30 |
Abbreviations: CI, confidence interval; COVID-19, coronavirus disease 2019; SVI+, social vulnerability index-plus.
Pre-vaccine included cases from March 2020 to 15 December 2020.
Post-Vaccine/Pre-Delta variant surge included cases between December 16 and 30 June 2021.
Post-vaccine/Delta variant surge included cases between 1 July 2021November 2021.
Associations between vulnerability measures and case rates varied by calendar period, however. In the pre-vaccine period, we observed statistically significant inverse associations between all 5 SVI+ subthemes and the SVI+ summary and case rates. In the post-vaccine/pre-Delta variant surge period, the SVI+ Summary and 4 of 5 subthemes were positively associated COVID-19 case rates (SVI+ Summary case RD: 36.0; 95% CI: 23.4, 48.7). In this time period, the strongest association was observed between the health subtheme and case rates (case RD: 45.4; 95% CI: 29.8, 60.9). The positive associations of the SVI+ summary and its subthemes with case rates persisted in the post-vaccine/Delta variant surge (SVI+ Summary case RD: 143.9; 120.9, 166.9). During post-vaccine/Delta variant surge, the health subtheme again showed the strongest association with case rates (case RD: 153.7; 95% CI: 123.5, 183.8). In contrast, we observed an inverse association between census tract-level vaccination coverage and case rates in this calendar period. There were on average 2 fewer cases of COVID-19 per population for every percentage point of the population that had received 2 doses of vaccine (case RD: −2.07; 95% CI: −2.66, −1.48).
Census Tract-level Association Between Social Vulnerability and COVID-19 Death Rates
Across the full study period, there were on average 5 more COVID-19 deaths per 100 000 population for every unit higher of the SVI+ summary (death RD: 5.3; 95% CI: 3.7, 6.9) (Table 2). In sum. 4 of 5 vulnerability subthemes were significantly and positively associated with higher mortality rates across the full study period, with the strongest association observed between the health subtheme and COVID-19 mortality rates (death RD: 6.0; 95% CI: 3.5, 8.6), whereas the socioeconomic, household composition, and health subthemes were consistently associated with higher COVID-19 mortality rates across all three time periods. Associations of SVI+ summary and its subthemes were consistently positive in all calendar periods. The magnitude of associations for each theme and death rates were largely similar in the pre-vaccine and in the post-vaccine/pre-Delta variant surge periods. Although the direction of association remained positive in the post-vaccine/Delta variant surge period for all vulnerability measures, the absolute rate difference nearly doubled in many instances (eg, socioeconomic subtheme death RD = 3.2; 95% CI: 1.0, 5.3 in the post-vaccine/pre-Delta and RD = 7.3; 95% CI: 4.2, 10.4 in the post-vaccine/Delta period).
Analysis of Individual COVID-19 Death by Community Social Vulnerability
In multilevel analyses, there was a positive association between the census tract-level SVI+ Summary and individual-level COVID-19 death among cases after accounting for age, race, and sex (risk ratio [RR] = 1.91; 95% CI: 1.38–2.64) (Table 3). The association between census tract-level SVI+ Summary and COVID-19 death was positive and of comparable magnitude in all calendar periods. Similarly, Black race, older age, and male sex were significantly associated with higher risk of death among COVID-19 cases in each calendar period (Table 3). In the post-vaccine/Delta variant surge when deaths were at their peak, Black race was the strongest predictor of death (RR = 1.60; 95% CI: 1.22, 2.10). In supplemental analyses of vulnerability subthemes and death, none was significantly associated with death in all three periods, yet each subtheme was significantly associated with death in at least 1 period [Supplementary Figure 2].
Table 3.
Multilevel Analysis of the Association Between Community Social Vulnerability and Individual Risk of Death
| SVI Theme |
Overall RR (95% CI) |
Pre- Vaccine RR (95% CI) |
Post- Vaccine/Pre- Delta Variant Surge RR (95% CI) |
Post- Vaccine/Delta Variant Surge RR (95% CI) |
|---|---|---|---|---|
| SVI+ summary | 1.91 (1.38, 2.64) | 1.90 (1.23, 2.94) | 2.17 (1.43, 3.30) | 1.91 (1.09, 3.33) |
| Age | 1.08 (1.08, 1.09) | 1.09 (1.08, 1.09) | 1.09 (1.08, 1.10) | 1.06 (1.05, 1.07) |
| Men (Ref: Women) | 1.55 (1.38, 1.74) | 1.65 (1.47, 1.85) | 1.56 (1.33, 1.83) | 1.50 (1.19, 1.89) |
| Race (Ref: White race) | ||||
| Black race | 1.37 (1.18, 1.59) | 1.29 (1.06, 1.57) | 1.39 (1.10, 1.76) | 1.60 (1.22, 2.10) |
| Other race | 0.86 (0.69, 1.08) | 0.88 (0.64, 1.22) | 0.65 (0.38, 1.09) | 1.38 (0.94, 2.03) |
Abbreviations: CI, confidence interval; COVID-19, coronavirus disease 2019; SVI+, social vulnerability index-plus.
DISCUSSION
We investigated the relationship between social vulnerability at the census tract level and COVID-19 outcomes in Fulton County, Georgia. Using data collected for surveillance purposes, we found that census tract-level COVID-19 case and death rates increased from the beginning of the pandemic through 11 November 2021. Across the study period, census tracts with higher social vulnerability persistently experienced greater COVID-19 deaths per capita, and individuals residing in these tracts had a higher risk of death following COVID-19 infection, a pattern that continued during the post-vaccine/Delta variant surge period, after vaccines were widely available. Furthermore, in multilevel models, we found that individual residents living in communities with higher social vulnerability, even after accounting for race and age, had a higher risk of mortality among COVID-19 cases across all calendar periods, with the highest relative risk of death during the post-vaccine/Delta variant surge period. Notably, this reflects a higher relative risk of death rather than an absolute increase in mortality rates, underscoring that this pattern is not merely a statistical phenomenon.
To our knowledge, this is the first study to investigate spatial-temporal trends in case and death rates at the census tract-level. Our findings that social vulnerability is an important consideration for understanding the context of the burden of COVID-19 in a community is in line with previous research on COVID-19 vulnerability and social determinants of health which has shown higher burden of COVID-19 among counties with more social vulnerability [4-6, 8, 9, 21]. This census tract-level analysis suggests that analyzing inequities in COVID-19 outcomes at finer levels of geographic disaggregation can provide valuable information about the unequal transmission and impact within a county.
Strong associations were observed for socioeconomic and health subthemes with both COVID-19 case and death rates across all time periods, with the largest rate differences emerging during the post-vaccine/Delta variant surge when the reported case and mortality burdens were highest. These results reinforce prior studies linking adverse socioeconomic determinants to poor health outcomes [22, 23]. Research has also shown that poorer counties in the United States experience higher COVID-19 incidence and mortality [24, 25], with Georgia data further supporting the association between lower income and increased COVID-19 deaths [26, 27]. This analysis expands existing research by demonstrating the impact of socioeconomic and health determinants within a diverse county, offering insights for targeted public health strategies.
We describe an inverse association between social vulnerability and COVID-19 case rates in the early pandemic. This phenomenon may reflect under-testing in census tracts with high levels of social vulnerability in that period rather than a lower true burden of infection. Socially vulnerable communities faced barriers to testing, such as limited sites, lack of transportation, and socioeconomic constraints, which likely contributed to under-detection [28, 29]. Higher test positivity rates in these areas further support this interpretation [30]. By later periods, expanded public health initiatives improved testing accessibility, making case rates more reflective of actual transmission [31, 32]. In contrast, mortality rates were higher in more vulnerable tracts throughout. It is plausible that once infected, people in more vulnerable census tracts have fewer resources to manage their health and survive than do people living in less vulnerable census tracts. Taken together, we note the importance of considering the broader literature and triangulating across multiple lines of evidence when interpreting surveillance findings to inform public health action.
Another key contribution is the simultaneous examination of community vulnerability and individual-level demographic factors in relation to mortality risk. Individuals in high-vulnerability communities had a higher risk of COVID-19 death, independent of race, age, and sex. Additionally, Black individuals had an increased risk of COVID-19 mortality, independent of community social vulnerability. Although prior research has examined aggregate racial composition and COVID-19 outcomes at the zip code [4] or county level [33], few studies have assessed the independent effects of social vulnerability and individual race on COVID-19 mortality.
Our multilevel findings support previous research showing that counties with lower percentages of socially vulnerable populations have better health outcomes [2-11]. By demonstrating that higher social vulnerability is linked to greater COVID-19 incidence and mortality—and that disparities persist at the individual level—this study underscores the need for equitable public health interventions. Proactively addressing social vulnerabilities in pandemic planning is critical. Using vulnerability data to identify high-risk communities before public health emergencies can guide targeted outreach and resource allocation. Strengthening partnerships between health, social service, housing, and transportation agencies can improve pandemic preparedness.
During crises, real-time analysis of social vulnerability and surveillance data can enhance resource distribution. Although states like Georgia developed COVID-19 data dashboards, social vulnerability was not fully integrated. Our findings demonstrate the feasibility of combining surveillance data with social determinants to optimize resource allocation in future crises. Embedding social vulnerability metrics into pandemic response planning can help mitigate inequities and improve health outcomes in future public health emergencies.
LIMITATIONS
Case and death rate analyses depend on accurate reporting and likely underestimate actual cases and deaths. Our analysis was limited to surveillance data up to 11 November 2021, excluding the Omicron-predominant period. The rise of at-home COVID-19 testing likely led to underreporting after December 2021 [34]. Additionally, we used static 2018 measures of social vulnerability, which may not reflect pandemic-induced changes. We were also unable to account for individual-level socioeconomic factors, such as employment as an essential worker, due to data limitations.
CONCLUSION
Even after vaccine availability, communities in Fulton County, Georgia, with higher social vulnerability experienced significantly more COVID-19 deaths per capita, and cases in these areas faced a higher risk of mortality. These findings highlight the value of surveillance data in identifying social disparities in pandemic outcomes and the persistence of these disparities despite improved prevention and treatment interventions. Addressing social vulnerability is crucial for achieving health equity and mitigating disparities in future public health crises.
Supplementary Material
Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Acknowledgments.
This work was supported in part by the Robert W. Woodruff Foundation through funding to the Emory COVID-19 Response Collaborative and the COVID-19 Health Equity Dashboard.
Footnotes
Potential conflicts of interest. The authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
References
- 1.Stokes EK, Zambrano LD, Anderson KN, et al. Coronavirus disease 2019 case surveillance--United States, January 22–May 30, 2020. MMWR Morb Mortal Wkly Rep 2020; 69:759–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Garg S, Kim L, Whitaker M, et al. Hospitalization rates and characteristics of patients hospitalized with laboratory-confirmed coronavirus disease 2019—COVID-NET, 14 states, March 1–30, 2020. MMWR Morb Mortal Wkly Rep 2020; 69:458–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Wadhera RK, Wadhera P, Gaba P, et al. Variation in COVID-19 hospitalizations and deaths across New York City Boroughs. JAMA 2020; 323:2192–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Akanbi MO, Rivera AS, Akanbi FO, Shoyinka A. An ecologic study of disparities in COVID-19 incidence and case fatality in Oakland County, MI, USA, during a state mandated shutdown. J Racial Ethn Health Disparities 2020; 8:1467–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Dasgupta S, Bowen VB, Leidner A, et al. Association between social vulnerability and a County’s risk for becoming a COVID-19 hotspot—United States, June 1–July 25, 2020. MMWR Morb Mortal Wkly Rep 2020; 69:1535–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Karaye IM, Horney JA. The impact of social vulnerability on COVID-19 in the U.S.: an analysis of spatially varying relationships. Am J Prev Med 2020; 59:317–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Neelon B, Mutiso F, Mueller NT, Pearce JL, Benjamin-Neelon SE. Spatial and temporal trends in social vulnerability and COVID-19 incidence and death rates in the United States. PLoS One 2021; 16:e0248702. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Nayak A, Islam SJ, Mehta A, et al. Impact of social vulnerability on COVID-19 incidence and outcomes in the United States. medRxiv 20060962 [Preprint]. 17 April, 2020. Available from: 10.1101/2020.04.10.20060962. [DOI] [Google Scholar]
- 9.Biggs EN, Maloney PM, Rung AL, Peters ES, Robinson WT. The relationship between social vulnerability and COVID-19 incidence among Louisiana census tracts. Front Public Health 2021; 8:617976. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Oates GR, Juarez LD, Horswell R, et al. The association between neighborhood social vulnerability and COVID-19 testing, positivity, and incidence in Alabama and Louisiana. J Community Health 2021; 46:1115–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kim SJ, Bostwick W. Social vulnerability and racial inequality in COVID-19 deaths in Chicago. Health Educ Behav 2020; 47:509–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Georgia Department of Public Health. Daily status report. Available at: https://dph.georgia.gov/covid-19-daily-status-report. Accessed 12 April 2022.
- 13.Georgia Department of Public Health. COVID-19 vaccine dashboard. GA DPH vaccine distribution dashboard. Available at: https://experience.arcgis.com/experience/3d8eea39f5c1443db1743a4cb8948a9c. Accessed 11 November 2021.
- 14.Centers for Disease Control and Prevention/ Agency for Toxic Substances and Disease Registry/ Geospatial Research, Analysis, and Services Program. CDC/ATSDR social vulnerability index [2018] database [GA]. Available at: https://www.atsdr.cdc.gov/placeandhealth/svi/data_documentation_download.html. Accessed 12 April 2022.
- 15.U.S. Census Bureau. Health Insurance in the United States: 2018-Tables. American Community Survey 5-Year Estimates Subject Tables, Table HI-05, Table HI-10-Table HI-14, 2018. Available at: https://www.census.gov/data/tables/time-series/demo/health-insurance/acs-hi.2018.html#list-tab-776654388. Accessed 11 November 2021. [Google Scholar]
- 16.Georgia Department of Public Health. Georgia DPH COVID-19 status report data definitions and faqs for COVID-19. COVID-19 status report. Available at: chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://ga-covid19.ondemand.sas.com/docs/GA_COVID19_Dashboard_Guide.pdf. Accessed 11 November 2021. [Google Scholar]
- 17.Centers for Disease Control and Prevention. Underlying conditions and the higher risk for severe COVID-19. U.S. Department of Health & Human Services. Available at: https://www.cdc.gov/covid/hcp/clinical-care/underlying-conditions.html. Accessed 15 November 2021. [Google Scholar]
- 18.Davies J, Krishnathasan D, Khairani CD, et al. Association between insurance status and cardiovascular outcomes in patients with COVID-19. J Am Coll Cardiol 2024; 83:2300. [Google Scholar]
- 19.U.S. Census Bureau. n.d. QuickFacts: Fulton County, Georgia. U.S. Department of Commerce. Available at: https://www.census.gov/quickfacts/fact/table/fultoncountygeorgia/PST045224. Accessed 5 March 2025. [Google Scholar]
- 20.Khan SS, McGowan C, Seegmiller LE, Kershaw KN. Associations between neighborhood-level racial residential segregation, socioeconomic factors, and life expectancy in the US. JAMA Health Forum 2023; 4:e231805. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Lee J, Ramírez IJ. Geography of disparity: connecting COVID-19 vulnerability and social determinants of health in Colorado. Behav Med 2022; 48:72–84. [DOI] [PubMed] [Google Scholar]
- 22.Shaw KM, Theis KA, Self-Brown S, Roblin DW, Barker L. Chronic disease disparities by county economic status and metropolitan classification, behavioral risk factor surveillance system, 2013. Prev Chronic Dis 2016; 13:E119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Berman AE, Miller DD, Rahn DW, et al. A county-level analysis of socioeconomic and clinical predictors of COVID-19 incidence and case-fatality rates in Georgia, March–September 2020. Public Health Rep 2021; 136:626–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Geranios K, Kagabo R, Kim J. Impact of COVID-19 and socioeconomic status on delayed care and unemployment. Health Equity 2022; 6:91–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Little C, Alsen M, Barlow J, et al. The impact of socioeconomic status on the clinical outcomes of COVID-19: a retrospective cohort study. J Community Health 2021; 46:794–802. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Moore JX, Langston ME, George V, Coughlin SS. Epidemiology of the 2020 pandemic of COVID-19 in the state of Georgia: inadequate critical care resources and impact after 7 weeks of community spread. J Am Coll Emerg Physicians Open 2020; 1:527–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Richmond HL, Tome J, Rochani H, Fung IC, Shah GH, Schwind JS. The use of penalized regression analysis to identify county-level demographic and socioeconomic variables predictive of increased COVID-19 cumulative case rates in the state of Georgia. Int J Environ Res Public Health 2020; 17:8036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Lee RM, Handunge VL, Augenbraun SL, et al. Addressing COVID-19 testing inequities among underserved populations in Massachusetts: a rapid qualitative exploration of health center staff, partner, and resident perceptions. Front Public Health 2022; 10:838544. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Society for Healthcare Epidemiology of America. Social determinants of health are painfully clear in COVID-19 testing, resources, and outcomes. SHEA. Available at: https://shea-online.org/social-determinants-of-health-are-painfully-clear-in-covid-19-testing-resources-and-outcomes/. Accessed 5 March 2025. [Google Scholar]
- 30.Agency for Toxic Substances and Disease Registry. Innovative uses of SVI during COVID-19. Centers for Disease Control and Prevention. Available at: https://www.atsdr.cdc.gov/place-health/php/covid19response/innovative-uses-of-svi-during-covid-19.html. Accessed 5 March 2025. [Google Scholar]
- 31.Sheikhattari P, Barsha RAA, Shaffer E, Bhuyan I, Elliott B. Lessons learned to improve COVID-19 response in communities with greatest socio-economic vulnerabilities. BMC Public Health 2023; 23:659. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Zhong Z, Sengupta R, Paynabar K, Waller LA. Multi-objective allocation of COVID-19 testing centers: improving coverage and equity in access. arXiv 2110.0927 [Preprint]. 21 September, 2021. Available from: 10.48550/arXiv.2110.09272. [DOI] [Google Scholar]
- 33.Millett GA, Honermann B, Jones A, et al. White counties stand apart: the primacy of residential segregation in COVID-19 and HIV diagnoses. AIDS Patient Care STDS 2020; 34:417–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Rader B, Gertz A, Iuliano AD, et al. Use of at-home COVID-19 tests—United States, August 23, 2021–March 12, 2022. MMWR Morb Mortal Wkly Rep 2022; 71:489–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.