Across Canada, the COVID-19 epidemic has been marked by a conflation of microepidemics across settings, including congregate living facilities (eg, long-term care homes [LTCHs], retirement homes, and shelters), essential workplaces, or households [1-6]. Many congregate facilities experienced outbreaks, with residents experiencing a 3- to 5-fold higher test positivity rate than the community-dwelling population [1,3,4,7,8]. Meanwhile, data on the community-dwelling population suggested a consistent pattern of geographic clustering of cases with 50% of cases occurring in approximately 25% of the population and disproportionately affecting those residing in neighborhoods that are less affluent and with a higher proportion of essential workers [9,10].

What is less understood is how congregate facilities may be connected with neighborhood disparities. Emerging data on congregate facility outbreaks suggested that facility-level spread among staff, among residents, and between staff and residents may have been partially triggered by undiagnosed infections and lack of support for effective isolation among staff [3,7]. Reducing transmission in the wider community may reduce outbreaks in congregate settings. However, data on facility staff are limited with few studies that characterized the socioeconomic, living, and working conditions of staff [11] and, thus, the extent to which social and structural determinants of community-level transmission might influence outbreaks in congregate facilities. Surveillance data on COVID-19 cases offer an opportunity to examine the pattern of cases among facility staff and other health care workers (HCWs) against community cases and neighborhood disparities. Lorenz curves, Gini coefficients, and the Hoover index, traditionally used as measurements of economic inequality [12,13], have been used in a range of health care research to measure health inequality [14,15]; for example, these measurements of health inequality have been recently applied in the context of the COVID-19 pandemic [10,16]. We sought to adapt Lorenz curves, Gini coefficients, and the Hoover index to compare the concentration of cases using neighborhood-level rates and neighborhood-level social and structural determinants across 3 mutually exclusive subgroups: community, facility staff (LTCHs, retirement homes, shelters), and other HCWs not working in congregate settings (eg, only working in hospitals). Our overarching objective was to determine if and how the pattern and magnitude of inequality and concentration in COVID-19 cases among HCWs differed between facility staff and other HCWs.

We found that the distribution of COVID-19 cases among facility workers mirrored neighborhood heterogeneity and social and structural disparities, a pattern that was less evident with other HCWs. Facility staff cases reflected greater inequalities by social and structural determinants than cases among other HCWs, and with some determinants (income and other essential workers), greater inequalities were seen with facility staff cases compared to community cases.

Cases among facility staff more closely reflected the geographic distribution of community cases than the distribution of cases among other HCWs and in the community. The similar distributions of facility staff and community cases could occur if there was an equal distribution of facility staff living across neighborhoods and the infection rate ratio between facility staff and the community was the same across neighborhoods. The implication of this potential mechanism is that most infections among facility staff would have been acquired outside the facility or workplace. The Lorenz curve patterns may also occur if facility staff were more likely than other HCWs to live in harder-hit neighborhoods with greater social and structural disparities, irrespective of workplace exposures in congregate facilities.

Although our study was centered on the Greater Toronto Area, the findings are likely to be generalizable to other large, urban cities and metropolitan areas with similar patterns of social and structural inequalities. In Canada, most congregate facilities are concentrated in large cities [34], and previous research comparing 16 cities demonstrated a similar neighborhood clustering of COVID-19 cases by social and structural determinants [10]. Data on the neighborhood characteristics of LTCH staff in the United States [11] suggest workers tend to live in lower-income neighborhoods, and individual-level data in Canada suggest that a high proportion of LTCH staff identify as racialized women with low household income [35]. In current LTCH staffing models across Canada and the United States, approximately 60% to 90% of workers who provide direct care to residents are providing services as unregulated staff (personal support workers, care aides, orderlies, and nurse assistants) [36-40] and receive the lowest wages in the health care sector at or just above minimum wage [36], often in the context of contract or casual work without benefits [36,37,41-45].

Our study suggests that cases among facility staff may disproportionately intersect with household exposures that are connected with other essential workplaces or amplified in the context of household density [21]. Our study was limited by a lack of confirmed denominators for setting-specific workers in the region, but data from England and Wales suggest a 2-fold increased rate of COVID-19 among LTCH workers versus other HCWs [46]. Based on government reports, just over 100,000 employees serve 78,000 LTCH residents in the province [36], such that the ratio of staff to LTCH residents in Ontario is approximately 1 staff for every 0.78 LTCH resident. If we extrapolate the provincial ratio to the Greater Toronto Area, where 28,316 LTCH residents reside, the city would have approximately 36,303 LTCH staff. With 4241 cases among LTCH staff in the Greater Toronto Area during our study period and a total population in the Greater Toronto Area of 7.1 million [17], the cumulative rate of COVID-19 cases among LTCH staff (at 11,682 per 100,000) would have been 10 folds higher than that of the community (1174 cases per 100,000).

This study has several limitations. First, we derived the dissemination area–level social determinants from the 2016 Census data, which may not be representative of the population during the COVID-19 pandemic. Second, the occupation status obtained from Ontario’s centralized surveillance system could have been misclassified due to possible misinterpretation of the question in self-reporting; further, facility staff or HCWs may work in multiple settings. Third, we did not have data on the residence of all HCWs across the various congregate settings to compare neighborhood-level per-HCW rates of cases. Finally, data were not available to link cases among HCWs to specific facilities and to directly examine how cases in communities influenced outbreaks in congregate settings.

The findings have implications for COVID-19 modeling and interventions. The magnitude of inequalities can be used as calibration or validation targets for epidemic and prediction models to reproduce the observed pattern of cases in relation to the distribution of overall cases and by social and structural determinants. In doing so, detection systems (eg, neighborhood wastewater surveillance) designed to predict the potential for exposures in congregate facilities could leverage data on underlying vulnerabilities in the neighborhood of facility staff residence. A study in the United States found that neighborhood characteristics of LTCH staff’s residences were the most important predictor of LTCH outbreaks [11]. These data could then be used to implement strategies to mitigate risks. For example, proximal strategies to reduce community-level transmission risks conferred through social and structural inequalities have the potential to reduce workplace exposure risks. Examples include systematically addressing the lived realties of workers that make physical distancing challenging (eg, household density) or that remain barriers to effective isolation and quarantine (eg, precarious job security and absence of benefits such as paid sick leave), with interventions such as wraparound care including access to food, medications, and child and senior care (especially in the context of multigenerational households) to facilitate staff quarantine and isolation. Prioritizing vaccination coverage in the hardest-hit neighborhoods is another example of indirectly reducing workplace exposures in LTCHs, retirement homes, and shelters. Finally, the findings highlight an urgent need for a long-term commitment and resources to comprehensively address social and structural barriers at a systems level (integration of health, education, social services, public health, and labor) given the long-standing history of infectious disease outbreaks in facilities and disparities experienced by its staff even before the COVID-19 pandemic [37].