Drivers of COVID-19 Outcomes in Long-Term Care Facilities Using Multi-Level Analysis: A Systematic Review

This study aimed to identify the individual, organizational, and environmental factors which contributed to COVID-19-related outcomes in long-term care facilities (LTCFs). A systematic review was conducted to summarize and synthesize empirical studies using a multi-level analysis approach to address the identified influential factors. Five databases were searched on 23 May 2023. To be included in the review, studies had to be published in peer-reviewed journals or as grey literature containing relevant statistical data. The Joanna Briggs Institute critical appraisal tool was employed to assess the methodological quality of each article included in this study. Of 2137 citations identified after exclusions, 99 records met the inclusion criteria. The predominant individual, organizational, and environmental factors that were most frequently found associated with the COVID-19 outbreak comprised older age, higher dependency level; lower staffing levels and lower star and subset domain ratings for the facility; and occupancy metrics and co-occurrences of outbreaks in counties and communities where the LTCFs were located, respectively. The primary individual, organizational, and environmental factors frequently linked to COVID-19-related deaths comprised age, and male sex; higher percentages of racial and ethnic minorities in LTCFs, as well as ownership types (including private, for-profit, and chain membership); and higher occupancy metrics and LTCF’s size and bed capacity, respectively. Unfolding the risk factors collectively may mitigate the risk of outbreaks and pandemic-related mortality in LTCFs during future endemic and pandemics through developing and improving interventions that address those significant factors.


Introduction
In unravelling the complexities of the impact of the COVID-19 pandemic on older adults in long-term care facilities (LTCFs), it becomes apparent that a comprehensive understanding of the substantial individual, organizational, and environmental factors is essential.The importance of this understanding is highlighted by the disproportionate impact and the huge toll of SARS-CoV-2 coronavirus (COVID-19) infection on older adults in LTCFs, especially those with underlying health conditions who experienced a range of predominantly adverse outcomes and death [1,2].Thus, this study examined COVID-19 case(s)/outbreaks and COVID-related deaths among residents in LTCFs where the pandemic hit the hardest.
The COVID-19-related outcomes in LTCFs were highlighted in a systematic review, reporting an infection rate of 45% coupled with a 23% mortality rate [3].Despite global efforts, by October 2021, the pandemic persisted and continued to spread; it appeared to differentially affect LTCFs worldwide, albeit with varying degrees of intensity.In the USA and across Europe, the proportion of COVID-19 cases per occupied LTCF bed ranged from 2.2% (in Finland) to 50% (in the USA) [4].COVID-19-related mortality rates per total population were also widely variable, from 11% in the Czech Republic to 50% in Belgium [4].Factors influencing COVID-19-related outcomes included variation in reporting information, the reliability, accuracy, and validity of the research, virus variants, healthcare infrastructure, population density, staffing levels and training, and geographic and cultural factors [5][6][7][8].The phenomenon's complexity was further evident as COVID-19 outcomes within a specific location fluctuated over time.For instance, the Canadian Institute for Health Information (CIHI) reported that in LTCFs, the count of COVID-19-related fatalities for March, April, May, and June 2020 were 13, 804, 3692, and 948, respectively [9].
Efforts have been made to dissect this intricate and complex phenomenon by either concentrating on a single domain or a particular topic within a domain separately or by examining all three domains/levels but failing to incorporate the relevant factors and synthesize them collectively comprehensively.In the existing body of literature, various drivers and predictors have been identified that can be classified as individual (e.g., age, sex), organizational (e.g., the ratio of staff to residents), and environmental factors (e.g., community prevalence).Reviews investigating determinants of COVID-19 outcomes in LTCFs have appeared in the scientific literature since 2020; to our knowledge, no published systematic review has yet collectively addressed these factors [10][11][12].The synthesis of emerging research and a comprehensive appreciation of the individual and contextual factors which might influence COVID-19 outcomes is vital for effective management and mitigation of harm in future infectious outbreaks.
The aim of this systematic review was to identify the individual, organizational, and environmental factors which together might underly COVID-19-related outcomes in LTCF.
To achieve this goal, the following are research questions: 1.
What individual factors or characteristics, such as Frailty Index and age, impacted COVID-19 outcomes within LTCFs? 2.
What organizational characteristics and practices (e.g., staffing level, ownership status) within LTCFs affected the spread and mortality-related outcomes?3.
Did environmental factors, such as facility/room design and facility age, influence COVID-19 outbreak and death?
This approach prevents the skewing of the impact of one group of factors without considering their broader context and will suggest comprehensive areas that require further research.

Materials and Methods
The Preferred Reporting Items for Systematic reviews and Meta-Analyses statements (PRISMA-S) checklist guided the conduct and reporting of the current systematic review [13].

Search Strategy and Data Sources
The systematic search was conducted with the collaboration of a librarian scientist using OVID MEDLINE, EMBASE, EBSCOhost CINAHL, and the Wiley Cochrane Database of Systematic Reviews.The search results included those published from inception to 23 May 2023.The details of search terms and strategies are shown in Table S1.

Eligibility Criteria
The PICO model was employed to design this review [14]; P (population): older adults residing in LTCFs or nursing homes regardless of sex and their health status, I (Intervention/exposure): LTCFs or nursing homes that reported case(s) with confirmed COVID-19-related outcomes, C (comparison): characteristics between LTCFs or residents with and without COVID-19-related outcomes, O (outcomes): the determinants that were associated with at least one confirmed COVID-19 case or death-related COVID-19 among residents in LTCFs.
To be included in the review, studies had to be original observational studies, published in peer-reviewed journals or as grey literature since 2020, and contain original statistical data on factors associated with COVID-19 outbreaks or COVID-19-related deaths among LTC residents.Studies were excluded if members of the study population were transferred to acute hospitals due to variations in care levels, staffing, and in the environment [15].Inclusion of hospital transfers might introduce a potential sampling bias, for example, a study of hospitalized residents diagnosed with COVID-19 exhibited a significantly elevated mortality rate (ranging from 51.6% to 59.3%), contrasting with control samples (8.1% to 9.7%) within LTCFs [16].Additional grounds for exclusion were met when the study's participants resided in alternative types of facilities, such as retirement homes, where residents had limited medical and daily assistance needs, or when the studies were oriented towards assessing the effects of interventions on outcomes related to COVID-19.A search for publications in languages other than English confirmed that no substantial (>5%) body of literature was excluded from the review.

Data Extraction
After removing the duplicates and screening the titles and abstracts, the full text of all potentially eligible articles was retrieved and assessed against the inclusion criteria.The Joanna Briggs Institute (JBI) critical appraisal tools for cross-sectional studies (eightitem scale), case-control (ten-item scale), and cohort studies (eleven-item scale) were used to appraise the methodological quality of the included studies [17].The studies were categorized into three levels based on quality assessment scores with scores ranging from zero to 8, 10, and 11 for each study type, respectively.Those in level one predominantly fulfilled the criteria and exhibited minimal bias risk, whereas level three studies failed to meet multiple criteria or demonstrated a significant risk of bias (Table S2).We also used a modified version of the JBI data extraction tool to include specific data about participants, context, methods, and outcomes (including individual, organizational, and environmental factors that may influence COVID-19 outbreaks and deaths).Prior to the study, the extraction tool was pilot tested on five studies by MKD and HMH using various methodological designs to make sure all related data were obtained.The tool was then refined and finalized through author consensus on conceptualizing and adding richness to the concepts of organizational, environmental, and individual factors using ecological models, focused on unravelling the complex interplay between factors which might influencing behaviour or care [18].Each stage of the data collection was accompanied by team review and discussion until a consensus was reached.

Data Analysis
The multi-level analysis was employed to identify three levels of determinants by incorporating both deductive and inductive analysis [19].The first author initially extracted tentative codes and themes.Then, the last author reviewed the results from each round and updated these.The analytical process was reviewed by the whole research team, and disagreements were resolved by discussion.We began by deductively coding the quantitative data against the three a priori codes: individual, organizational, and environmental factors.The data under each level were then categorized under two outcome groups: COVID-19 outbreaks and COVID-19-related deaths.Data under each category were then coded again, which led to emerging subcodes (the second level of the coding scheme) which revealed corresponding determinants for each level.Subcodes were then inductively clustered into two major categories for individual determinants and environmental factors and three major categories for organizational factors.The three a priori levels comprised seven themes and 98 subcodes.We synthesized data narratively and presented these under the three levels.

Results
Of the 2474 studies' records identified via the systematic search strategy, 99 records met the inclusion criteria (Figure 1).levels comprised seven themes and 98 subcodes.We synthesized data narratively and presented these under the three levels.
A significant portion (40%) of the articles explored the impact of individual, organizational, and environmental factors concurrently.
The included articles examined these factors using 3 to 150 subcategorized variables, with a median of 23.Seventy-one percent of the articles were of moderate quality.Variability in findings was observed among studies classified as either low quality (6%) or high quality (22%).Variability in results was observed explicitly across articles that performed adjustment analyses or effect analyses (83%) and those that did not undertake them (16%).
The findings are organized into three primary sections, focusing on individual, organizational, and environmental factors.Within these sections, various subgroups are linked to both COVID-19 outbreak and COVID-related deaths.The presentation of data in Tables S3-S8 illustrates this structure, wherein each section is further divided into subsections emerged from an inductive analysis of the data.

Individual Factors
Individual factors were conceptualized as the attributes of LTCF residents that were related to COVID-19 infections and deaths.

Individual Factors Related to Outbreaks
Identified individual factors were linked to 13 areas and fell into two groups: (1) sociodemographic background (age, sex, and socioeconomic status) and (2) condition-specific factors (comorbid conditions, scores of health status instruments, smokers, co-existent medications, seroprevalence, body mass index (BMI), resident dependency level, frailty index, duration of stay in LTCFs, and hospitalization experience).
Table S3 details the individual factors associated with outbreaks in LTCFs.

Individual Factors Related to COVID-19-Related Deaths
Fifteen areas of individual factors associated with COVID-19-related death emerged and were categorized as follows: (1) sociodemographic background (age, sex, and social engagement level) and (2) condition-specific factors (cognitive/mental status, comorbidity, symptoms, lab test results, nutritional status, BMI, degree of dependence and level of needed care, frailty, co-existent medications, duration of stay in LTCFs, and hospitalization experience).
Table S4 details the individual factors associated with death in LTCFs.

Organizational Factors
Organizational factors were conceptualized as the internal attributes and organizational characteristics of an LTCF.

Organizational Factors Related to Infection Outbreaks
Eight areas were identified across three groups: (1) LTCFs quality indicators (quality rating star, quality performance), (2) staffing (staffing levels, infected staff, nursing staff assignment, and employment status, (3) ownership and membership affiliation (types of ownership, chain membership), (4) Medicare and Medicaid coverage, and (5) LTCF's racial and ethnic composition.
Table S5 contains details about the organizational factors associated with COVID-19 outbreaks in LTCFs.

Organizational Factors Related to COVID-Related Deaths
COVID-related deaths were associated with six organizational areas that we further categorized into five groups: (1) LTCFs quality indicators (star rating, quality performance), (2) staffing (staffing levels, infected staff), (3) ownership and chain affiliation (ownership types, chain membership status), (4) Medicaid and Medicare coverage, and (5) LTCF's racial and ethnic composition.
Table S6 depicts detailed information about the impact of organizational factors on COVID-19-related death.

Environmental Factors
Environmental factors were conceptualized as the social and physical environments in which residents reside.

Environmental Factors Related to COVID Outbreaks
Two groups of environmental factors covering 13 areas were found to affect COVID-19 outbreaks: (1) community factors (outbreaks in counties/communities, outbreak in the community where staff live, location of LTCF, community sociodemographic status (e.g., racial/ethnic composition of the community and socioeconomic status), and (2) physical characteristics (number of beds, crowding index, occupancy rate, new admissions, structural design of the rooms, using the Green House model, accessing ventilator-dependent units, mechanical recirculation of air, and type of care provided in the ward.
Detailed information on these environmental factors is provided in Table S7.

Environmental Factors Related to COVID-Related Deaths
Ten environmental factors associated with COVID-19-related death were identified and further categorized into two groups: (1) community factors (outbreaks in counties/communities, high-density communities, public transportation use by workers, disability support program use, location of LTCF, community sociodemographic status (proportions of ethnic minorities and socially deprived communities), and (2) LTCFs physical characteristics (bed numbers, levels of crowding, structural design of the rooms, using the Green House model, accessing ventilator-dependent unit).
Table S8 contains detailed information about the impact of organizational factors on COVID-19-related death.

Discussion
The COVID-19 pandemic led to excess deaths of older adults in LTCFs, particularly early on, when there were few effective treatments and no vaccination programs [117].This, coupled with staffing shortages, and the relative shortages of adequate personal protective equipment [118] exacerbated the already parlous state within LTCFs.
There is an abundance of information on the contributors to outbreaks and deaths.In looking across the included citations, the most influential factors begin to emerge.Figure 2 depicts factors with the greatest support.
Among the individual factors identified in contributing to outbreaks were a larger representation of older age, increased resident dependency, sex (female), and higher Frailty Index, as well as presence of comorbidities and cognitive decline/dementia.
Arranging individual factors based on the strongest research support reveals the following sequence of their influence on mortality: comorbidity, in particular older age, male sex, increased dependency, cognitive deterioration/dementia, and frailty levels, and co-morbidities.
and further categorized into two groups: (1) community factors (outbreaks in counties/communities, high-density communities, public transportation use by workers, disability support program use, location of LTCF, community sociodemographic status (proportions of ethnic minorities and socially deprived communities), and (2) LTCFs physical characteristics (bed numbers, levels of crowding, structural design of the rooms, using the Green House model, accessing ventilator-dependent unit).
Table S8 contains detailed information about the impact of organizational factors on COVID-19-related death.

Discussion
The COVID-19 pandemic led to excess deaths of older adults in LTCFs, particularly early on, when there were few effective treatments and no vaccination programs [117].This, coupled with staffing shortages, and the relative shortages of adequate personal protective equipment [118] exacerbated the already parlous state within LTCFs.
There is an abundance of information on the contributors to outbreaks and deaths.In looking across the included citations, the most influential factors begin to emerge.Among the individual factors identified in contributing to outbreaks were a larger representation of older age, increased resident dependency, sex (female), and higher Frailty Index, as well as presence of comorbidities and cognitive decline/dementia.
Arranging individual factors based on the strongest research support reveals the following sequence of their influence on mortality: comorbidity, in particular older age, male sex, increased dependency, cognitive deterioration/dementia, and frailty levels, and comorbidities.The individual factors are unsurprising.The population residing in LTCFs represents the most vulnerable older adults, with complex comorbidities and a high prevalence of Alzheimer's disease and related dementias.The greater likelihood of outbreaks in those exhibiting responsive behaviours is predictable, given that adherence to the strict infection control required is more difficult for these individuals and their dependency upon staff much greater.Cognitive impairment was the most extensively researched comorbidity and a significant predictor of COVID-related outcomes in LTCFs.This result ties well with a systematic review and meta-analysis [119,120].Underlying mechanisms identified as being implicated in the increased risk were neuroinflammation, nonadherence to COVID-19 prevention measures, and the impact of coexisting comorbidities [119,120].
When looking at organizational factors, parallels emerged for outbreaks and mortality.For outbreaks, staffing levels (an inverse association), star and subset domain ratings for the facility, higher percentages of racial and ethnic minorities in LTCFs, ownership types (including private, for-profit, and chain membership), and presence of infected staff were corroborated across studies.The organizational factors with the highest level of research support for mortality were similar but with a slightly different rank order: LTCF's racial and ethnic composition, private/for-profit ownership and chain membership, staffing levels, star and subset domain ratings, and the presence of infected staff.A lower quality of LTCF care has previously been linked to for-profit status ownership [121], and perhaps this finding illustrates the same paradigm.
Unsurprisingly, staffing levels, which also may be lower in private and for-profit homes, and which were further stressed during the pandemic, were identified as a significant risk in outbreak and death analyses.A higher risk of adverse COVID-19 outcomes was identified within LTCFs having larger proportions of racial/ethnic minority residents; it was noted that the disparities in nursing home outcomes attributed to race were not solely a result of race.Instead, these disparities were rooted in underlying inequalities inherent in healthcare and nonhealthcare sectors, ultimately leading to poorer health outcomes for racial/ethnic minority residents [89,95].
Not all studies attempted to control for socioeconomic factors underlying ethnic differences in susceptibility, either that of the residents, their care staff, or the socioeconomic status of the facility or of the community in which it resided, exposing the inter-relatedness of the three strata analyzed.
The variability regarding the influence of staffing levels can be attributed to three main factors.Firstly, discrepancies or challenges in methods and study quality may play a role, as different studies utilized diverse analysis techniques and might have overlooked adjusted analyses.This aligns with the findings of Harrington and colleagues [82], who found that low total nurse staffing hours were associated with outbreaks when the model was not adjusted for factors, including health deficiencies, bed size, ownership, and total nurse staffing hours.Adjusting for multi-level factors might have yielded different results.Studies also used various databases to gain information about the characteristics of the LTCFs in varying time periods before the pandemic; these might not reflect the characteristics of the LTCF staff attrition that occurred due to COVID-19 [122].Furthermore, data related to the COVID-19 outcomes were often collected in different time frames, with little attention to longitudinal as the pandemic progressed.Methodological challenges may also arise in connection with interpretive analysis, leading to divergent results.For example, when interpreting staff-resident ratios, discrepancies may occur based on whether the numerator considers solely full-time staff or encompasses casual workers, as noted in a distinct systematic review [41].Secondly, concerning differences in staffing policies, exemplified by variations in pre-COVID-19 staffing regulations governing casual and part-time employment or work across multiple LTCFs, could potentially contribute to the transmission of diseases [123].Thirdly, contextual factors may contribute to the variation in the effects of staffing levels on outcomes; results could be influenced by whether the staff members come from communities experiencing active outbreaks, thereby affecting the potential for disease transmission within LTCFs [73].The availability of personal protective equipment (PPE) and consistent infection control and prevention training throughout the pandemic could also shape outcomes [91].
Given the conflicting data regarding the effects of staffing.a more comprehensive investigation is warranted.This could involve retrospective longitudinal studies that span the entirety of the pandemic, utilizing repeated measures designs.Additionally, utilizing time-sensitive data related to COVID-19 outcomes and the specific attributes of the facilities under study would provide a more comprehensive understanding of the impact.To capture the whole picture of a phenomenon, performing an adjusted analysis by including the individual, organizational, and environmental covariates would be required.
The environmental factors that garnered the most substantial support regarding outbreaks and mortality alike were the number of beds/crowding index/occupancy rate, outbreaks in counties/communities, community sociodemographic status (racial/ethnic composition of the community/socioeconomic status), high-density communities and structural design of the rooms.Similarly, the strongest environmental factors contributing to COVID-19-related mortality included the number of beds/crowding index/occupancy rate, outbreaks in counties/communities, location of the LTCF, and structural design of the rooms.
A scoping and systematic review demonstrated that certain studies show a relationship between the for-profit status of care facilities and related attributes, like sufficient staffing, access to personal protective equipment (PPE), and testing provisions.These factors are then tied to increased adverse COVID-19 outcomes.The reviews also indicated that the influence of ownership is intricate and holds significance [124,125].
The higher occupancy metrics (including number of beds, crowding index, occupancy rate) and occurrences of outbreaks in counties and communities where LTCFs were located were predominant environmental factors, which is aligned with a previous systematic review wherein the congregate physical environment in LTCFs was described as a factor that exacerbates the outbreaks and mortality risk [10].Larger LTCF size, location, and interaction with the community with high COVID-19 rates were described as the strongest and most consistent predictors of COVID-19 outcomes [10].
These findings illustrate the inter-relatedness of the classification of strata, homes in poorer neighbourhoods, drawing their staff from less privileged communities (all factors associated with outbreaks) may also have residents who similarly themselves have a high proportion of "at risk" factors.A lack of single rooms, making isolation in the face of infection more difficult, is perhaps a predictable finding, as is the existence of high-density living and older home design (perhaps more "institutional").Findings such as those in the "Green House" model (smaller, more home-like communities) are perhaps less expected but again may illustrate resident-based risk, rather than an institutional factor.
Discrepancies within articles regarding both COVID-19 outbreaks and mortality may be partially attributed to outcome reporting bias.Notably, clarity is absent regarding whether the prevalence of COVID-19 encompasses asymptomatic residents as well or only those confirmed with tests.Similarly, for mortality rates it remains uncertain whether deaths among residents with positive COVID-19 status died because of, or with, COVID-19.This was evident in a study where 22.7% of COVID-19 cases resulted in death, of which only 24.8% were classified as COVID-19-related deaths [66].
Our study does have limitations.Observational studies are prone to biases, such as reverse causation and residual confounding [126].The appraisal tool used for this study may not explicitly focus on issues like reverse causality or other forms of endogeneity in observational studies.
We did not include studies focusing mainly on residents transferred to the hospital to treat COVID-19 infection or those that implemented interventions that could be considered organizational.Included study designs were observational, describing strength of association rather than allowing inference of causality.Analyses also failed to consider the complexity of the interrelationship between factors; of the studies included here, only 40% took into account individual, organizational, and environmental (both internal and external to the facility) factors collectively, all including varying covariates/cofounders.Conversely, around 30% of the articles focused solely on one of these three strata, potentially introducing confounding effects due to unmeasured variables.
Many studies did not provide a fully adjusted analysis, which could have reduced the bias in the parameter estimates, so the results might have been overestimated.There was no consistent pattern of adjustment in analytical models across the papers, which may have affected the results.The sample size in seven studies was small and included only one to six LTCFs.More than half of the studies took place in the USA, with data from five studies obtained from one state.This may introduce a systematic bias, being based upon a common administrative structure.The remaining locations form an unrepresentative sample of international LTCF.
Furthermore, non-English papers were not included in this review; thus, a small percentage of the overall number of relevant articles have been excluded.We were also unable to take an intersectional lens in analyzing social determinants of the health of residents, their care providers, and the communities in which the LTCFs were situated.Some of the factors identified in this review are supported by very little evidence, requiring further studies.Furthermore, few LTCF care delivery models were covered, and the Green House model was the focus of only one study.
The results of this study offer comprehensive insights into the complexities of the phenomena and may support the development of modeling which incorporates the multilevel nature of factors influencing outcomes.The findings may also inform decisionmakers about those that need to be taken into account to mitigate the impact of this and future pandemics in LTCFs.We have highlighted areas that have not been rigorously researched and those factors that can be considered covariates to control for in future analyses.We also encourage using concurrent data as the validity of the study's results may be compromised if using non-contemporary data (e.g., two-year-old data regarding characteristics of the organizations).

Conclusions
This review has identified potentially modifiable individual, organizational, and environmental risk factors for COVID-19-related outbreaks and deaths in LTCFs for older adults.Action to address these factors is a matter of urgency.To address the risk factors identified in this systematic review, several important actions are recommended.Initially, focus should be devoted to enhancing staffing levels by employing recruitment of full-time staff and training programs to ensure that residents receive sufficient support and mitigating the risk of cross-contamination and transmission within the facility by proactively developing proper strategies.Furthermore, improving quality rankings and performance standards may enhance the overall quality of care.Addressing disparities and racial and ethnic barriers to effective healthcare services in LTCFs and communities requires strategic interventions to mitigate underlying healthcare inequalities.To address specific challenges for residents with dementia in LTCFs, specific plans need to be developed.Initiatives for community engagement and support can enhance resources and tackle social determinants of health.Lastly, prioritizing further research to gain a deeper understanding of these complex interactions will inform evidence-based interventions and policies for managing future pandemics.
In light of the factors identified, including age, sex, dependency level, dementia prevalence, quality performance metrics, staffing levels, racial compositions within LTCFs, ownership structure, bed count, occupancy rate, and community/county characteristics, we recommend a meta-analysis that includes more comparisons to estimate the effectiveness of these factors on COVID-19 outcomes in LTCFs.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/healthcare12070807/s1.File S1: PRISMA Checklist.Table S1: Search strategies.Table S2: The methodological quality assessment of the included studies.Table S3.Individual factors related to outbreaks.Table S4.Individual factors related to COVID-19-related deaths.Table S5.Organizational factors related to infection outbreaks.Table S6.Organizational factors related to COVID-related deaths.Table S7.Environmental factors related to COVID outbreaks.Table S8.Environmental factors related to COVID-related deaths.References [127,128] are cited in the supplementary materials.

Figure 1 .
Figure 1.PRISMA flow-chart for the systematic review process.

Figure 1 .
Figure 1.PRISMA flow-chart for the systematic review process.
Figure 2 depicts factors with the greatest support.

Table 1 .
Description of studies included in the systematic review.