Social inequality and the syndemic of chronic disease and COVID-19: county-level analysis in the USA

Abstract

Background

Given the effect of chronic diseases on risk of severe COVID-19 infection, the present pandemic may have a particularly profound impact on socially disadvantaged counties.

Methods

Counties in the USA were categorised into five groups by level of social vulnerability, using the Social Vulnerability Index (a widely used measure of social disadvantage) developed by the US Centers for Disease Control and Prevention. The incidence and mortality from COVID-19, and the prevalence of major chronic conditions were calculated relative to the least vulnerable quintile using Poisson regression models.

Results

Among 3141 counties, there were 5 010 496 cases and 161 058 deaths from COVID-19 by 10 August 2020. Relative to the least vulnerable quintile, counties in the most vulnerable quintile had twice the rates of COVID-19 cases and deaths (rate ratios 2.11 (95% CI 1.97 to 2.26) and 2.42 (95% CI 2.22 to 2.64), respectively). Similarly, the prevalence of major chronic conditions was 24%–41% higher in the most vulnerable counties. Geographical clustering of counties with high COVID-19 mortality, high chronic disease prevalence and high social vulnerability was found, especially in southern USA.

Conclusion

Some counties are experiencing a confluence of epidemics from COVID-19 and chronic diseases in the context of social disadvantage. Such counties are likely to require enhanced public health and social support.

NI and BL are joint first authors.

IK and MM are joint senior authors.

Introduction

In the past, pandemics have disproportionately affected poorer populations, widening existing social inequalities.^1–3 The COVID-19 has now spread throughout the USA, and there is growing evidence of disparities between different socio-economic groups in mortality from COVID-19.⁴ Yet the extent of these disparities, and how they might be addressed, requires further characterisation.

Chronic conditions, including chronic obstructive pulmonary disease, heart disease, diabetes and chronic kidney disease, have emerged as important risk factors for severe illness from COVID-19 infection,⁵ and there is a particular concern that socially disadvantaged populations might be affected by a confluence of epidemics from chronic diseases and COVID-19, which may exacerbate each other: a concept known as ‘syndemics’, defined as ‘the presence of two or more disease states that adversely interact with each other, negatively affecting the mutual course of each disease trajectory, enhancing vulnerability and which are made more deleterious by experienced inequities.’⁶

Social disadvantage has been variously measured. The US Centers for Disease Control and Prevention (CDC) has previously employed a Social Vulnerability Index (SVI; constructed using census data) to identify counties that are especially vulnerable to the economic and social consequences of a major environmental disaster.^{7 8} However, empirical data on its value in identifying areas at particular risk during the current COVID-19 pandemic are scarce.

This study explores the social disparities in the effect of the COVID-19 pandemic between US counties. It aims to quantify the relation between county-level social disadvantage (as measured by SVI) and COVID-19 incidence and mortality. Such findings will inform our understanding of the social determinants of COVID-19, and the utility of measures such as SVI in the public health response to the pandemic. The study also aims to describe the geographical distribution of socially disadvantaged counties that have both high prevalence of chronic diseases and high mortality from COVID-19 (ie, those counties that are experiencing a syndemic of diseases), which are likely to need particular public health and social support over the course of the pandemic.

Methods

County-level social vulnerability was assessed using the SVI developed by the US CDC, with data from the 2018 American Community Survey. Details of the parameters and methods used to construct the index have been described in detail elsewhere.^{7 8} In brief, it is a composite measure of 15 socioeconomic and demographic factors reported in the US Census (including the level of poverty, unemployment, crowded housing and health insurance coverage). The index was developed to identify counties that were most vulnerable to the social and economic impacts of major environmental disasters, and as such most likely to require public assistance after such an event; higher SVI score indicates greater social vulnerability.

County-level data on COVID-19 cases and deaths up to 10 August 2020 were obtained from USAFacts, a source of COVID-19 data used by the CDC (https://www.cdc.gov/coronavirus/2019-ncov/cases-updates/previouscases.html). County-level prevalence of five major chronic conditions associated with COVID-19 severity among adults aged ≥18 years were obtained from US CDC, using data from the 2018 Behavioural Risk Factor Surveillance System (BRFSS) and the US Census population.⁵ Information was obtained on the prevalence of obesity (body mass index ≥30 kg/m²), diabetes mellitus (types 1 and 2), chronic obstructive pulmonary disease, heart disease (angina or coronary heart disease and myocardial infarction) and chronic kidney disease.⁵

Statistical analysis

We used Poisson regression, with the log of the population as an offset, to model the association between SVI and COVID-19 cases and deaths. We added a scale parameter to address the issue of potential overdispersion,⁹ and used US States as a fixed-effect to adjust for state-level confounding factors.¹⁰ We checked the robustness of these analyses by using negative binomial regression models. The same approach was used to examine the association between SVI and prevalence of the chronic conditions. We also mapped the joint distribution of COVID-19 mortality and each of the five chronic conditions in the most-deprived counties (fourth and fifth quintiles combined). Statistical analysis was performed using Stata statistical software V.14.2 (StataCorp) or Python (V.3.7.7).

Results

After excluding one county due to missing information on SVI, 3141 counties from 50 states of the USA and the District of Columbia were analysed, representing more than 327 million people. Relative to the lowest SVI quintile, counties in the highest quintile had a higher proportion of people in poverty (9.2% vs 23.6%), in unemployment (3.3% vs 8.7%), without health insurance (6.6% vs 14.3%), and from ethnic minority populations (9.7% vs 46.6%), but lower per capita income (US$ 32 000 vs US$ 21 000) (online supplemental table S1).

By 10 August 2020, a total of 5 010 496 cases and 161 058 deaths from COVID-19 had been reported. The rate of COVID-19 cases and deaths increased in a dose-response manner with increasing levels of SVI (table 1). Compared with the lowest SVI quintile, the adjusted rates of COVID-19 cases in the second, third, fourth and fifth quintile were 24% (95% CI 16% to 33%), 39% (95% CI 30% to 49%), 76% (95% CI 65% to 87%) and 111% (95% CI 97% to 126%) higher, respectively, while the rate of COVID-19 deaths was 19% (95% CI 9% to 30%), 22% (95% CI 12% to 33%), 77% (95% CI 63% to 92%) and 142% (95% CI 122% to 164%) higher, respectively. In sensitivity analyses, additionally adjusting for county-level proportions of people ≥65 years did not materially alter the findings. Neither was there any effect of restricting the analyses to counties with at least one COVID-19 case or death, or when using negative binomial regression models as opposed to Poisson regression models (online supplemental table S2).

Table 1

COVID-19 cases and deaths in 3141 US counties as of 10 August 2020, by quintiles of Social Vulnerability Index

	Social Vulnerability Index
	First quintile	Second quintile	Third quintile	Fourth quintile	Fifth quintile
	(Least vulnerable)				(Most vulnerable)
COVID-19 cases
Total cases	368 312	808 447	1 088 614	1 614 575	1 130 548
Crude rate per 100 000; median (IQR)	474 (279–786)	594 (331–1008)	670 (372–1272)	1009 (485–1849)	1816 (1020–2827)
Adjusted rate ratio (95% CI)	1.0 (Ref.)	1.24 (1.16 to 1.33)	1.39 (1.30 to 1.49)	1.76 (1.65 to 1.87)	2.11 (1.97 to 2.26)
COVID-19 deaths
Total deaths	13 611	28 043	34 007	47 093	38 304
Crude rate per 100,000; median (IQR)	4 (0–15)	8 (0–23)	10 (0–27)	15 (4–41)	31 (12–68)
Adjusted rate ratio (95% CI)	1.0 (Ref.)	1.19 (1.09 to 1.30)	1.22 (1.12 to 1.33)	1.77 (1.63 to 1.92)	2.42 (2.22 to 2.64)

Rate ratios adjusted for US states as a fixed effect (to adjust for state-level confounding factors) in the Poisson regression models using the log of the population as an offset term, and a scale parameter, set as the Pearson χ2 statistic divided by the residual df, to address the issue of potential overdispersion. Data: US Centers for Disease Control and Prevention and USAFacts.

CI, confidence interval; df, degrees of freedom; IQR, interquartile range.

Higher levels of SVI was also associated with higher prevalence of each of the five chronic conditions (online supplemental table S3). Compared with the lowest quintile, the adjusted prevalence of obesity, diabetes, chronic obstructive pulmonary disease, heart disease and chronic kidney disease in the highest quintile was 24% (95% CI 22% to 26%), 41% (95% CI 39% to 43%), 32% (95% CI 28% to 35%) and 25% (95% CI 22% to 28%) and 32% (95% CI 30% to 35%) higher, respectively.

Maps of the joint distribution of COVID-19 mortality with prevalence of each of the chronic conditions among the most deprived counties (fourth and fifth SVI quintiles together) shows marked geographical clustering of areas with both high COVID-19 mortality and high prevalence of chronic diseases (ie, those counties experiencing a syndemic of diseases), particularly in the southern US states of Mississippi, Georgia, Louisiana and Texas (figure 1). However, there was substantial variation in both COVID-19 mortality and chronic disease prevalence among counties, with many reporting low COVID-19 mortality despite high prevalence of chronic disease and social vulnerability. In analyses that compare the characteristics of counties with high levels of both COVID-19 mortality and chronic diseases (ie, counties experiencing a syndemic of diseases) to those counties with low levels of both COVID-19 mortality and chronic diseases, the proportions of people ≥65 years did not differ between groups, whereas the proportion of people in poverty, unemployed, without health insurance or from an ethnic minority population was substantially higher among counties that experienced the syndemic (online supplemental table S4).

Figure 1. Geographical distribution of county-level prevalence of major chronic conditions and COVID-19 mortality rates as of 10 August 2020, among the most socially vulnerable counties in the USA. Analyses restricted to counties in the 4th and 5th quintiles of the Social Vulnerability Index of the US Centers for Disease Control and Prevention. Maps give overlapping quintiles of county-level prevalence of major conditions ( online supplemental table S3) with COVID-19 mortality rates (table 1) up to 10 August 2020. Data: 2018 Behavioral Risk Factor Surveillance System; COVID-19 deaths from USAFacts.

Discussion

In this study of 3141 US counties, county-level SVI was strongly associated with both cases and deaths from COVID-19, with those in the highest SVI quintile having about double the rate of cases and deaths of those in the lowest quintile. SVI was also associated with the prevalence of each of the five major chronic conditions, and the mapping of counties indicated some areas (especially in the southern USA) are experiencing a confluence of epidemics from chronic diseases and COVID-19 in the context of social vulnerability.

Social disadvantage is strongly related to overcrowded living conditions, reduced likelihood of working conditions that permit physical distancing, and reduced likelihood of seeking and using healthcare services, among other factors that might exacerbate the impact of spread and clinical course of infectious disease epidemics.^11–14 Such populations are also more likely to suffer from chronic diseases, as found in the present report. Given chronic conditions, including heart disease, obesity, and diabetes, are themselves major risk factors for the severity of COVID-19 infections, and it is becoming clear that the combined effects of social disadvantage itself on the potential for COVID-19 to propagate, together with the cocurrent chronic disease epidemic among these counties, is driving the disproportionate impact of COVID-19 on these populations.^15–17

The social disruption accompanying the response to the COVID-19 pandemic has in turn adversely affected the management of chronic diseases, as well as the socioeconomic circumstances of the poorest counties, highlighting the complex syndemic nature of COVID-19. The mapping of counties in the present report illustrates that some areas are particularly affected by chronic diseases, social disadvantage, and the COVID-19 pandemic. However, there also some counties with high social vulnerability and high prevalence of chronic disease that have not experienced high rates of COVID-19 deaths, suggesting there may be important lessons from the response to the pandemic in these counties that could be applicable elsewhere.

It is also worth noting that some counties had lower prevalence of chronic conditions despite higher social vulnerability, including parts of Texas and California, which might be attributable to health and social policies in these states.^{18 19} For example, California has been at the forefront of implementing a range of policy interventions to address smoking, such as an increase in cigarette tax.^18–20

Our study has some limitations. First, the reporting of COVID-19 infections and deaths is dependant on testing, and the associations between SVI and rates of COVID-19 cases and deaths may be even stronger if testing is less prevalent in socially vulnerable areas. Second, the prevalence of chronic diseases examined in this study were obtained from BRFSS in 2018, which may be affected by non-response, and differential responses in the marginalised populations. Third, these data do not distinguish the duration or severity of the chronic diseases examined. Lastly, although there were strong associations between social vulnerability and disease occurrence, it is not possible to exclude some residual confounding by other social or demographic factors (including age). Also, given the ecological fallacy, the findings should not be interpreted at an individual level.

This study highlights the importance of policy interventions to tackle the pandemic that more explicitly focus on health equity and social justice. A greater attention to, and proportionate resource mobilisation for, disadvantaged counties is needed (including improved opportunities for COVID-19 testing and policies to support appropriate physical distancing), to ensure this pandemic does not widen existing social inequalities.

What is already known on this subject

• In the past, pandemics have disproportionately affected poorer populations, widening existing social inequalities. Given the effect of chronic diseases on risk of severe COVID-19 infection, the present pandemic may have a particularly profound impact on socially disadvantaged counties. As the COVID-19 pandemic has now spread throughout the USA, there is growing evidence of disparities between different socio-economic groups in incidence of, and mortality from, COVID-19. Yet the extent of these disparities, and how they might be addressed, requires further characterisation.

What this study adds

• This national study of 3141 US counties used the Social Vulnerability Index (a widely-used measure of social disadvantage) to stratify counties, and found that as of 10 August 2020, the most socially disadvantaged counties had, on average, twice the rate of COVID-19 cases and deaths relative to the least disadvantaged counties. Similarly, the prevalence of major chronic conditions was also substantially higher in the most disadvantaged counties. Some counties are experiencing a confluence of epidemics from COVID-19 and chronic disease in the context of social disadvantage, which may exacerbate each other and further widen social inequalities. There was evidence of particular geographic clustering of such counties in the southern USA.

NI and BL receive salary support from the Clinical Trial Service Unit and Epidemiological Studies Unit (CTSU), Nuffield Department of Population Health (NDPH), University of Oxford. CTSU receives core support from the UK MRC, British Heart Foundation (BHF), and Cancer Research UK (CRUK). Support was also received from the US Centres for Disease Control and Prevention (CDC) Foundation (with support from Amgen). BL acknowledges support from UK Biobank, the National Institute for Health Research Biomedical Research Centre (Oxford, UK), and the BHF Centre of Research Excellence (Oxford, UK).

Data availability statement

All the data used in this study are publicly available and properly cited. Please contact [email protected] for more information.

Ethics statements

Patient consent for publication

Not required.

Ethics approval

Since all the data were anonymous, aggregated without any personal information, and publicly available, ethics approval was deemed waived.

Footnote

Twitter @HDambhaMiller

Contributors NI and BL conceptualised the study with the input from IK, MM, SS, AME, HD-M and GC. IK and MM were the cosenior authors. NI did the statistical analysis. NI and BL wrote the first draft of the manuscript. All authors contributed to data interpretations, and critical revisions of the manuscript. NI and BL are the guarantors.

Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

Disclaimer Sponsors had no role in the design, analysis, or dissemination of the study. The views expressed in this article are those of the authors and not necessarily those of the entities the authors are affiliated with and/or supported by.

Map disclaimer The depiction of boundaries on this map does not imply the expression of any opinion whatsoever on the part of BMJ (or any member of its group) concerning the legal status of any country, territory, jurisdiction or area or of its authorities. This map is provided without any warranty of any kind, either express or implied.

Competing interests None declared.

Patient and public involvement statement Patient and public involvement was not applicable since this study did not involve patients and public directly. However, our findings will be appropriately disseminated to the public through personal and social communication tools.

Provenance and peer review Not commissioned; externally peer reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

References

¹ Fanning PJ. Influenza and inequality: one town’s tragic response to the great epidemic of 1918. Amherst: University of Massachusetts Press, 2010.

² Fry CV. ‘Never Let a Good Crisis Go To Waste’: The impact of the 2014 ebola epidemic on West African science. MA: USA: Massachusetts Institute of Technology, 2019.

³ Grantz KH, Rane MS, Salje H, et al. Disparities in influenza mortality and transmission related to sociodemographic factors within Chicago in the pandemic of 1918. Proc Natl Acad Sci U S A 2016; 113: 13839–44. doi:10.1073/pnas.1612838113 http://www.ncbi.nlm.nih.gov/pubmed/27872284

⁴ 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:10.15585/mmwr.mm6915e3 http://www.ncbi.nlm.nih.gov/pubmed/32298251

⁵ Razzaghi H, Wang Y, Lu H, et al. Estimated County-Level Prevalence of Selected Underlying Medical Conditions Associated with Increased Risk for Severe COVID-19 Illness - United States, 2018. MMWR Morb Mortal Wkly Rep 2020; 69: 945–50. doi:10.15585/mmwr.mm6929a1 http://www.ncbi.nlm.nih.gov/pubmed/32701937

⁶ The Lancet. Syndemics: health in context. Lancet 2017; 389: 881. doi:10.1016/S0140-6736(17)30640-2 28271823

⁷ Flanagan BE, Gregory EW, Hallisey EJ, et al. A social vulnerability index for disaster management. J Homel Secur Emerg Manag 2011; 8: 3. doi:10.2202/1547-7355.1792

⁸ Centers for Disease Control and Prevention. CDC SVI 2018 documentation, 2020.

⁹ Islam N, Sharp SJ, Chowell G, et al. Physical distancing interventions and incidence of coronavirus disease 2019: natural experiment in 149 countries. BMJ 2020; 370: m2743. doi:10.1136/bmj.m2743 http://www.ncbi.nlm.nih.gov/pubmed/32669358

¹⁰ Gunasekara FI, Richardson K, Carter K, et al. Fixed effects analysis of repeated measures data. Int J Epidemiol 2014; 43: 264–9. doi:10.1093/ije/dyt221 http://www.ncbi.nlm.nih.gov/pubmed/24366487

¹¹ U.S. Bureau of Labor Statistics. Job flexibilities and work schedules — 2017-2018: data from the American time use survey. U.S. Bureau of Labor Statistics 2019.

¹² Khullar D, Chokshi DA. Health, income, and poverty: where we are and what could help. Health Affairs Health Policy Brief 2018. doi:10.1377/hpb20180817.901935

¹³ Kwarteng JL, Schulz AJ, Mentz GB, et al. Associations between observed neighborhood characteristics and physical activity: findings from a multiethnic urban community. J Public Health 2014; 36: 358–67. doi:10.1093/pubmed/fdt099 http://www.ncbi.nlm.nih.gov/pubmed/24159053

¹⁴ Andersen RM, Yu H, Wyn R, et al. Access to medical care for low-income persons: how do communities make a difference? Med Care Res Rev 2002; 59: 384–411. doi:10.1177/107755802237808 http://www.ncbi.nlm.nih.gov/pubmed/12508702

¹⁵ Guan W-J, Ni Z-Y, Hu Y, et al. Clinical characteristics of coronavirus disease 2019 in China. N Engl J Med 2020; 382: 1708–20. doi:10.1056/NEJMoa2002032 http://www.ncbi.nlm.nih.gov/pubmed/32109013

¹⁶ Wang D, Hu B, Hu C, et al. Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in Wuhan, China. JAMA 2020; 323: 1061–9. doi:10.1001/jama.2020.1585 http://www.ncbi.nlm.nih.gov/pubmed/32031570

¹⁷ Islam N, Khunti K, Dambha-Miller H, et al. COVID-19 mortality: a complex interplay of sex, gender and ethnicity. Eur J Public Health 2020; 30: 847–8. doi:10.1093/eurpub/ckaa150 http://www.ncbi.nlm.nih.gov/pubmed/32745211

¹⁸ Texas Cancer Registry. Tobacco-Associated cancers in Texas, 2006-2015. Texas Department of State Health Services, 2018.

¹⁹ Pierce JP, Shi Y, McMenamin SB, et al. Trends in lung cancer and cigarette smoking: California compared to the rest of the United States. Cancer Prev Res 2019; 12: 3–12. doi:10.1158/1940-6207.CAPR-18-0341 http://www.ncbi.nlm.nih.gov/pubmed/30305281

²⁰ Wolstein J, Babey S, Diamant AL. Obesity in California. Los Angeles, CA: UCLA Center for Health Policy Research, 2015.