COVID-19 Cases and Deaths in Skilled Nursing Facilities in Cook County, Illinois

Abstract

Objective:

The COVID-19 pandemic has had a devastating impact on older adults residing in skilled nursing facilities. This study examined the pathways through which community and facility factors may have affected COVID-19 cases and deaths in skilled nursing facilities.

Methods:

We used structural equation modeling to examine the number of COVID-19 cases and deaths in skilled nursing facilities in Cook County, Illinois, from January 1 through September 30, 2020. We used data from the Centers for Medicare & Medicaid Services, the Illinois Department of Public Health, and the Cook County Medical Examiner’s Office to determine the number of resident COVID-19 cases and deaths, number of staff cases, facility-level characteristics, and community-level factors.

Results:

Poorer facility quality ratings and higher numbers of staff COVID-19 cases were associated with increased numbers of resident COVID-19 cases and deaths. For-profit ownership was associated with larger facilities and higher resident-to-staff ratios, which increased the number of staff COVID-19 cases. Furthermore, skilled nursing facilities with a greater percentage of White residents were in areas with lower levels of social vulnerability and were less likely to be for-profit and, thus, were associated with higher quality.

Conclusions:

For-profit ownership was associated with lower facility quality ratings and increases in the number of staff COVID-19 cases, leading to increased resident COVID-19 cases and deaths. Establishing enforceable regulations to ensure quality standards in for-profit skilled nursing facilities is critical to prevent future outbreaks and reduce health disparities in facilities serving racial and ethnic minority populations.

Adults aged 60 or older constitute 22% of the total US population^1,2 while representing 80% of all COVID-19 deaths in the United States. ³ Older adults residing in nursing homes have been particularly affected by COVID-19. ⁴ Although less than 2% of older adults live in nursing homes, ⁵ COVID-19 deaths among residents in long-term care facilities accounted for 37.7% of COVID-19 deaths in the United States by the end of 2020.^6,7 Cook County, the second largest county in the United States, with a population of more than 5.1 million, had the highest number of COVID-19 cases and deaths in Illinois, and nursing homes in Cook County were disproportionately affected. ⁸ There are roughly 15 000 skilled nursing facilities (SNFs) in the United States; Illinois has close to 700 SNFs with more than 77 900 certified beds. ⁹

Nursing home residents are at high risk of experiencing severe COVID-19 outcomes because of their age and underlying chronic health conditions. ¹⁰ People with cognitive impairment are estimated to account for 50% of nursing home residents and have difficulty complying with instructions to wear face masks and wash their hands. ¹¹ In addition, the shared rooms, public spaces, and corridors of nursing homes make it difficult for residents to implement social distancing practices.

Nursing home staff have been affected by COVID-19 as well.^12,13 Nursing homes were known to have staff shortages and high turnover rates even before the pandemic. ¹⁴ The additional burden of high numbers of staff who were unable to work because of COVID-19 infection may have exacerbated an already acute workforce shortage in SNFs. ¹⁵ Nursing home workers also move in and out of facilities on a daily basis, and nursing homes lacked adequate personal protective equipment (PPE), which may also have contributed to the transmission of COVID-19 in their facilities. ¹⁶

Multiple facility-level factors associated with an increased number of COVID-19 cases and deaths in nursing homes have been identified, including ownership type, facility quality, and facility size. However, findings on the effects of these factors have been inconsistent.^16-23 A potential reason for these inconsistencies may be due to the ways in which studies analyzed facility characteristics. Several studies examined bivariate relationships and used linear regression models that did not capture the relationships among these factors. As a result, these studies may not have identified the significant effects of some factors on nursing home COVID-19 cases and deaths. To address this issue, we used structural equation models to explore potential relationships among relevant resident, facility, and community characteristics and to estimate the total effects of independent variables on both outcomes.

Methods

Conceptual Framework

This analysis examined the distribution of COVID-19 cases and deaths in SNFs in Cook County, Illinois. Previous studies document that for-profit ownership is consistently associated with lower-quality nursing home care.^24,25 Studies also indicate that nursing homes serving racial and ethnic minority residents experience a greater number of COVID-19 cases and deaths. ²⁶ Thus, we hypothesized that for-profit facilities would have lower quality ratings, greater numbers of occupied beds, and higher resident-to-staff ratios. Similarly, for-profit facilities would have a relatively greater proportion of non-White residents. We also assumed that socioeconomically disadvantaged communities would be more likely than their wealthier counterparts to have nursing home facilities with lower quality ratings and be more likely to serve non-White residents. Furthermore, we hypothesized that higher numbers of COVID-19 cases among nursing home staff would be associated with increased numbers of resident COVID-19 cases, while COVID-19 cases among nursing home staff would be associated with the rate of COVID-19 in surrounding community areas. ²¹ This conceptual framework is reflected in the structural equation models for COVID-19 cases and deaths (Figures 1 and 2).

Figure 1.

Structural equation model explaining the number of resident COVID-19 cases in 177 skilled nursing facilities in Cook County, Illinois, January 1–September 30, 2021. Values are Β (P value); only significant P values are presented. Solid lines indicate significant relationships; dotted lines indicate nonsignificant relationships. All analyses were tested at α = .05. Data source: Centers for Medicare & Medicaid Services (CMS). ²⁷

Figure 2.

Structural equation model explaining the number of resident COVID-19 deaths in 177 skilled nursing facilities in Cook County, Illinois, January 1–September 30, 2021. Values are Β (P value); only significant P values are presented. Solid lines indicate significant relationships; dotted lines indicate nonsignificant relationships. All analyses were tested at α = .05. Data source: Centers for Medicare & Medicaid Services (CMS). ²⁷

Datasets

We used 4 publicly available datasets for this analysis. We retrieved data on COVID-19 deaths that occurred from January 1 through September 30, 2020, from the Cook County Medical Examiner’s Office case archive file. ²⁸ We geocoded incident addresses and spatially matched them with the Centers for Medicare & Medicaid Services (CMS) nursing home list ⁹ to identify certified SNFs in Cook County. We then retrieved the cumulative number of COVID-19 cases and deaths for SNF residents and staff for the study period, the number of occupied beds, and ownership type (for-profit vs nonprofit) from the CMS COVID-19 dataset. ²⁷

Next, we appended the CMS Five-Star Quality Rating for SNFs in Cook County using the CMS nursing home quality rating data. ²⁹ We obtained the average numbers of residents and staff and resident racial and ethnic composition from LTCFocus, which hosts nursing home–related data compiled by the Brown University Center for Gerontology and Healthcare Research. ³⁰ To describe neighborhood characteristics, we used the Centers for Disease Control and Prevention’s (CDC’s) Social Vulnerability Index (SVI), which quantifies the potential negative effects on communities caused by external stresses on human health. ³¹ We appended the SVI to SNFs in Cook County using census tract numbers. Finally, we obtained zip code–level population COVID-19 infection rates from the Illinois Department of Public Health COVID-19 website. ³² We used existing public datasets that do not include identifiable information, and we obtained exemption (No. 2021-1991) from the University of Illinois at Chicago Institutional Review Board.

Analysis

We used 2 outcome measures for this analysis: resident COVID-19 cases and deaths. CMS defines a nursing home COVID-19 case as a case that “originated in the facility,” not a case that is admitted “from a hospital with known positive status, or an individual’s COVID-19 status that became positive within 14 days after admission.” ³³

Explanatory variables included the following: percentage of White residents, number of occupied beds, facility quality rating, resident-to-staff ratio, for-profit ownership, number of COVID-19 cases among staff, the facility zip code–level COVID-19 infection rate, and neighborhood SVI. In terms of facility characteristics, the CMS Five-Star quality ratings range from 1 to 5, with 1 being poor, 5 being excellent, and the middle 70% of facilities receive a rating of 2-4 stars. We measured resident-to-staff ratio by calculating the number of nursing home residents per staff member. Facility ownership type was a dichotomous variable, categorizing facilities into for-profit and nonprofit ownership. In addition, we included the number of COVID-19 cases among staff. CDC defines a staff COVID-19 case as “a staff member who worked at the facility for any length of time in two calendar days before the onset of symptoms for a symptomatic person; or two calendar days before the positive sample was obtained for an asymptomatic person.” ³⁴ Concerning community characteristics, we used the population COVID-19 infection rate, which was the number of COVID-19 cases per 100 000 population. The SVI is a standardized score centered at zero, with a higher score indicating greater social vulnerability.

We checked potential multicollinearity among the variables and found that variance inflation factor scores were <2, indicating no evidence of multicollinearity among the independent variables. To explore the spatial distribution of SNFs in Cook County, we mapped their geographic locations and the number of deaths using ArcGIS version 10.8.1 (Esri). We performed descriptive statistics to compare characteristics of for-profit and nonprofit facilities. We also conducted multivariate linear regression analysis to examine resident COVID-19 cases and deaths. We then used structural equation models to explore potential connections among the variables for both resident COVID-19 cases and deaths. All analyses were tested at α = .05. We used Stata version 16 (StataCorp, LLC) for statistical analysis.

Results

Overall, our analytic dataset included 177 SNFs. In these SNFs, there were 6270 resident cases, 1845 resident deaths, and 3666 staff cases during the study period. SNFs in Cook County were highly clustered on the north side of Chicago (Figure 3), and facilities in these areas had more deaths than facilities located in less clustered areas. In contrast, facilities with no COVID-19 deaths were more likely to be located outside city boundaries, on the north and west sides of Cook County.

Figure 3.

Skilled nursing facilities in Cook County, Illinois, and number of COVID-19 deaths by facility, January 1–September 30, 2021. Map created in ArcGIS version 10.8.1 (Esri). Data sources: Centers for Medicare & Medicaid Services (CMS).^9,27

Of the 177 SNFs in Cook County, 81.4% were for-profit and 18.6% were nonprofit facilities. The mean number of resident COVID-19 cases was 53.2 in for-profit facilities and 35.4 in nonprofit facilities. Similarly, the mean number of resident COVID-19 deaths was 14.6 in for-profit facilities and 11.2 in nonprofit facilities. For-profit facilities were larger on average than nonprofit facilities (mean number of beds, 187 vs 138). However, the percentage of occupied beds did not differ significantly between for-profit and nonprofit facilities (69.1% vs 67.7%, respectively; P = .61). The overall quality rating was significantly higher for nonprofit facilities than for-profit facilities (4.2 vs 2.5; P < .001).

The multivariate linear regression models explained the number of COVID-19 cases and the number of deaths in relation to facility and community characteristics (Table 1).Overall, lower quality ratings, for-profit ownership, and a greater number of staff COVID-19 cases were associated with a greater number of resident COVID-19 cases. Lower quality ratings and for-profit ownership were also associated with a larger number of resident COVID-19 deaths.

Table 1.

Results of multivariate linear regression models explaining the number of resident COVID-19 cases (n = 6270) and deaths (n = 1845) in 177 skilled nursing facilities in Cook County, Illinois, January 1–September 30, 2021

Variable	Β (SE) [P value] a
	No. of cases b	No. of deaths b
% White residents c	0.07 (0.09) [.44]	0.03 (0.04) [.25]
No. of occupied beds d	0.03 (0.04) [.42]	−0.01 (0.01) [.22]
CMS quality rating e	−3.86 (1.78) [.03]	−1.53 (0.51) [<.001]
Resident-to-staff ratio c	−3.35 (3.12) [.29]	0.80 (0.90) [.37]
For-profit ownership d	11.93 (5.87) [.04]	1.54 (1.69) [.36]
No. of staff COVID-19 cases d	1.52 (0.13) [<.001]	0.37 (0.04) [<.001]
Zip code–level population COVID-19 infection rate f	−0.06 (0.09) [.51]	−0.01 (0.03) [.82]
Social Vulnerability Index g	6.19 (9.24) [.50]	−0.05 (2.66) [.99]

Abbreviation: CMS, Centers for Medicare & Medicaid Services.

^aSignificant at P < .05.

^bData source for number of COVID-19 cases and deaths for skilled nursing facility residents and staff members: CMS COVID-19 dataset. ²⁷

^cData source: LTCFocus, Brown University Center for Gerontology and Healthcare Research. ³⁰

^dData source: CMS COVID-19 dataset. ²⁷

^eData source: CMS Five-Star Quality Rating System. ²⁹

^fData source: Illinois Department of Public Health. ³²

^gData source: Centers for Disease Control and Prevention. ³¹

Neither the percentage of White residents nor the 2 community-level variables (the SVI score and the population COVID-19 infection rate) were associated with resident COVID-19 cases or deaths. The number of occupied beds was associated with the number of resident COVID-19 cases but was no longer significant when for-profit ownership was introduced.

In the structural equation analysis explaining the number of resident COVID-19 cases (Table 2), the SVI had significant direct effects on the percentage of White residents and population COVID-19 infection rates (Figure 1). SNFs with a lower percentage of White residents were more likely to be in communities with a higher SVI, and communities with higher SVI scores were associated with higher population COVID-19 infection rates. For-profit ownership was directly associated with facilities with a lower percentage of White residents, poorer facility quality ratings, higher numbers of occupied beds, and greater resident-to-staff ratios. The percentage of White residents also had a direct effect on the quality rating, which in turn affected resident COVID-19 cases. In contrast, the number of occupied beds and resident-to-staff ratio were associated with staff COVID-19 cases, which then directly affected the number of resident COVID-19 cases.

Table 2.

Standardized coefficients for direct, indirect, and total effects of the structural equation model explaining the number of resident COVID-19 cases (n = 6270) and deaths (n = 1845) in 177 skilled nursing facilities in Cook County, Illinois, January 1–September 30, 2021

Variable	B (SE) [P value] a
	Direct	Indirect	Total
No. of cases b
Population COVID-19 infection rate c	—	0.05 (0.08) [.53]	0.04 (0.08) [.53]
No. of staff COVID-19 cases d	1.53 (0.13) [<.001]	—	1.53 (0.13) [<.001]
No. of occupied beds d	0.03 (0.04) [.45]	0.10 (0.03) [<.001]	0.13 (0.05) [.01]
CMS quality rating e	−4.42 (1.45) [<.001]	—	−4.42 (1.45) [<.001]
Resident-to-staff ratio f	−1.89 (2.96) [.52]	−7.81 (2.79) [.01]	−9.70 (3.97) [.02]
For-profit ownership d	—	6.85 (3.62) [.06]	6.85 (3.62) [.06]
% White residents f	—	−0.11 (0.04) [.01]	−0.11 (0.04) [.01]
Social Vulnerability Index g	—	5.59 (2.65) [.04]	5.59 (2.65) [.04]
No. of deaths b
No. of resident COVID-19 cases b	0.15 (0.02) [<.001]	—	0.15 (0.02) [<.001]
Population COVID-19 infection rate c	—	0.01 (0.02) [.53]	0.01 (0.02) [.53]
No. of staff COVID-19 cases d	0.13 (0.04) [<.001]	0.23 (0.04) [<.001]	0.15 (0.02) [<.001]
No. of occupied beds d	—	0.03 (0.01) [<.001]	0.03 (0.01) [<.001]
CMS quality rating e	−0.42 (0.34) [.22]	−0.63 (0.23) [.01]	−1.05 (0.39) [0.01]
Resident-to-staff ratio f	0.89 (0.70) [.21]	−1.83 (0.66) [.01]	−0.94 (0.95) [.01]
For-profit ownership d	—	2.18 (0.86) [.01]	2.18 (0.86) [.01]
% White residents f	—	−0.03 (0.01) [.01]	−0.03 (0.01) [.01]
Social Vulnerability Index g	—	1.39 (0.68) [.04]	1.39 (0.68) [.04]

Abbreviations: —, does not apply; CMS, Centers for Medicare & Medicaid Services.

^aSignificant at P < .05.

^bData source for number of COVID-19 cases and deaths for skilled nursing facility residents: CMS. ²⁷

^cData source: Illinois Department of Public Health. ³²

^dData source: CMS. ²⁷

^eData source: CMS Five-Star Quality Rating System. ²⁹

^fData source: LTCFocus, Brown University Center for Gerontology and Healthcare Research. ³⁰

^gData source: Centers for Disease Control and Prevention. ³¹

In the structural equation analysis explaining the number of resident COVID-19 deaths, we first examined resident COVID-19 deaths without the number of resident COVID-19 cases in the model (data not shown). The result was nearly identical to the model explaining resident COVID-19 cases and showed that the quality ratings and the number of resident COVID-19 cases were associated with resident COVID-19 deaths. We then included the number of resident COVID-19 cases in the model. The number of resident COVID-19 deaths was positively associated with both the number of resident and staff COVID-19 cases. The quality rating no longer had a direct effect on the number of resident COVID-19 deaths, but the number of staff COVID-19 cases continued to be significantly associated with the number of resident COVID-19 deaths.

When we examined the total effects of variables in the structural equation models for the number of resident COVID-19 cases and deaths, we found that facilities with higher percentages of White residents and higher quality ratings had fewer resident COVID-19 cases and deaths. In contrast, larger facilities with more occupied beds, for-profit facilities, and facilities with more staff COVID-19 cases had greater numbers of resident COVID-19 cases and deaths. Facilities located in areas with higher levels of social vulnerability also had more resident COVID-19 cases and deaths. The population COVID-19 infection rate was not associated with resident COVID-19 cases or deaths at SNFs. Interestingly, a higher resident-to-staff ratio was associated with fewer resident COVID-19 cases but not with the number of resident COVID-19 deaths.

Discussion

This study explored COVID-19 cases and deaths among older adults residing in SNFs in Cook County. Cook County, the second largest county in the United States, had the highest number of COVID-19 cases and deaths in the state, and SNFs in Cook County were particularly affected by the pandemic. ⁸ The body of literature on factors associated with COVID-19 cases and deaths in SNFs is relatively small and often shows conflicting findings. Our analysis took a comprehensive look at the factors that have been shown to be associated with COVID-19 cases and deaths in SNFs by exploring the structural relationships among relevant factors.

Our findings showed that SNFs with lower quality ratings and greater numbers of staff COVID-19 cases had higher numbers of resident COVID-19 cases. Although some studies have found that larger facilities with a greater number of occupied beds had higher numbers of resident COVID-19 cases,^17,20 our results show that the number of occupied beds affected the number of resident COVID-19 cases indirectly through the number of staff COVID-19 cases.

More importantly, factors associated with resident COVID-19 cases were strongly associated with ownership type. For-profit facilities were more likely than nonprofit facilities to have lower quality ratings, larger size, and a higher resident-to-staff ratio, all of which ultimately were associated with more resident COVID-19 cases. The average quality rating of for-profit facilities in Cook County was 2.5, compared with a quality rating of 4.2 for nonprofit facilities. More than 97% of facilities that had a rating of <2 stars were for-profit, whereas a little more than 50% of facilities with a 5-star quality rating were for-profit. This finding suggests that for-profit facilities with a lower quality rating may have lacked access to critical resources such as PPE.^15,17,20,22 We also showed that the number of staff COVID-19 cases was significantly associated with resident COVID-19 cases. Similar findings have been reported by others.^21,23 Conversely, a recent study documented that nursing home staffing levels did not necessarily change during the pandemic, and staff hours per resident day actually increased in nonprofit nursing homes. ³⁵ Although we do not have evidence on the nature of the transmission of the virus between staff and residents, ²⁰ many facilities had challenges accessing PPE that would protect both staff and residents ¹⁵ and continued to have staff shortages. ³⁶ Interestingly, our data showed that a higher resident-to-staff ratio was related to fewer staff COVID-19 cases. This result may reflect the fact that the resident-to-staff ratio was measured when staff COVID-19 cases were measured, which means that facilities with fewer staff COVID-19 cases would have had more staff still working in the facility, consequently, a greater resident-to-staff ratio.

We showed that for-profit facilities were more likely than nonprofit facilities to serve non-White residents. Several studies have documented that facilities with a larger percentage of White residents are associated with fewer COVID-19 cases.^18,20,26 Our finding indicates that resident COVID-19 cases may be associated with resident racial and ethnic composition via facility quality, such that White residents are more likely to reside in facilities with a higher quality rating and, therefore, have fewer COVID-19 cases. We showed that non-White residents are more likely to reside in for-profit nursing homes and, thus, have an increased risk of exposure to COVID-19 infection compared with residents in nonprofit nursing homes, contributing to racial and ethnic disparities in COVID-19 outcomes. The disproportionate impact of COVID-19 on racial and ethnic minority populations has been well documented among the general population, and our finding documents the differential impact of COVID-19 among racial and ethnic minority populations in nursing homes.

Relatedly, our results indicate that the social vulnerability of the neighborhood in which long-term care facilities are located was indirectly associated with an increase in the number of resident COVID-19 cases and that facilities in more socially vulnerable neighborhoods were more likely to have a higher proportion of non-White nursing home residents. Previous literature has documented that communities with a high degree of social vulnerability are at a higher risk of COVID-19 cases.^37,38 To our knowledge, no studies have yet examined the association between the social vulnerability of the community in which a long-term care facility is located and COVID-19 cases among SNF residents.

The number of staff COVID-19 cases was associated with the number of resident COVID-19 deaths. It is important to note that when we examined the number of deaths without the number of resident COVID-19 cases in the model, the same factors associated with resident COVID-19 cases were also related to resident deaths. However, when we included the number of resident COVID-19 cases in the model, the quality rating and resident-to-staff ratio were not directly associated with the number of resident deaths; rather, the pathway was through resident COVID-19 cases. This finding may indicate that facility characteristics associated with an increased number of COVID-19 cases are the same factors affecting resident COVID-19 deaths; in other words, facilities that had higher numbers of COVID-19 cases also had a higher number of deaths.

Limitations

Several limitations are worth noting. First, our analyses were limited to variables that were available in public datasets during the study period. Our analysis included data through September 30, 2020. Thus, we were not able to expand our findings to later months of the pandemic, although we have no reason to believe contributing factors described herein would have been different in later months. Second, we could not control for resident case-mix and risk characteristics, including age and presence of multiple preexisting chronic conditions. ² Third, we could not determine the direction of transmission between staff and residents using cross-sectional data.

Conclusion

Despite these limitations, our findings have substantial policy implications concerning the level of infection among staff as a strong and consistent factor associated with resident COVID-19 cases and deaths. Prior studies of COVID-19 cases have shown a strong relationship between COVID-19 infections among staff and residents.^18,21 Recent CDC findings indicate that challenges exist in uptake of COVID-19 vaccines among nursing home staff, which appears to lag behind vaccine uptake among residents. ³⁹ These findings have important implications concerning a modifiable risk factor. Because a rise in the number of COVID-19 cases among staff increases the number of resident cases and deaths, efforts should be made to increase the pay and paid medical leave for these frontline workers. Dean et al ⁴⁰ argued that the presence of unions in facilities was associated with a 30% decrease in COVID-19 deaths in New York State, which may suggest that better pay and benefits could reduce COVID-19 mortality.

Nationally, COVID-19 cases per 1000 residents in SNFs peaked in June and July 2020 at 11.5 cases, declined to 6.3 cases in mid-September, and increased again to 23.2 cases in November. ²¹ CMS established the Coronavirus Commission for Safety and Quality in Nursing Homes in September 2020 to respond to this crisis. ⁴¹ CMS recommended measures to reduce COVID-19 transmission, including weekly testing, an increased supply of PPE, infection control training, and improved employee benefits. Despite these recommendations, nursing homes continued to be hot spots for COVID-19 deaths through November 2020. Although the national vaccine rollout in facilities for both residents and staff has substantially reduced the incidence of COVID-19, staff uptake of vaccines appears to lag behind resident uptake. COVID-19 uncovered weaknesses in our long-term care system that have been worsening for decades. It is imperative for public health analysts, policy makers, and health care providers to learn from the current pandemic. This is not the first and will not be the last time that nursing home residents are exposed to natural disasters.