Introduction

Assisted living (AL) represents an increasingly common residential setting for older adults with health and support needs [1, 2]. Primarily aligned with a social model of care, AL homes typically provide a secure environment, personal support services and limited health care, emphasizing a home-like setting [2]. Relative to nursing homes (NHs), AL settings have lower levels of professional staffing (no or limited on-site registered nurses, physicians, nurse practitioners), fewer health and medical services and a greater emphasis on family involvement in resident care [3–6]. These features contrast with the clinical complexity of the AL population, with residents exhibiting a high prevalence of dementia, multimorbidity, frailty and complex polypharmacy [1, 2, 6, 7]. Research on the ability of AL to provide quality care to residents with changing health needs remains sparse though studies have raised concerns about insufficient staffing levels and mix [8–10], medication use and oversight [11], and high rates of care transitions, including hospital and NH admissions [2, 8, 12, 13]. As such, AL settings may be particularly sensitive to external stresses because of their limited reserve.

Hospitalization events are common and an important indicator of the quality and availability of care in AL [8, 13]. Associated risk factors include both AL resident (e.g., health instability, polypharmacy) and home characteristics (e.g., smaller settings, lower staffing levels and the absence of on-site professional staff) [8–10]. Prior to the COVID-19 pandemic, studies within the U.S. [12, 14] and Canada [8, 13, 15] showed a significantly higher rate of hospitalization (and prolonged stays with delayed discharge) for AL residents relative to the NH population. The relative absence of professional staff, medical services and integrated interprofessional care, arguably resulting in less timely and/or appropriate care (e.g., failure to address upstream risks for falls, delirium, and infections; delayed detection or treatment of acute health changes), is felt to contribute to the higher hospitalizations observed for AL [6, 16, 17]. These hospitalization events often lead to lengthy hospital stays and worsening functional and cognitive health status for residents [18] (prompting a NH admission) [2] and increased distress for family caregivers. Yet, it is important to acknowledge that many hospitalizations are necessary and that reduced access to needed hospital care (e.g., during healthcare crises) also constitutes a relevant system quality indicator for at-risk older populations [19]. Given these considerations, it is likely that the healthcare disruptions associated with the COVID-19 pandemic (e.g., staffing shortages, reduced availability of social/medical services and family to support care) were especially challenging for AL homes [3–5, 20] potentially leading to heightened risks of adverse health outcomes for residents.

Studies of older adults in community [21, 22] and nursing home [23–28] settings showed that hospitalization and emergency department rates declined dramatically during the first pandemic wave and typically remained below historical rates during the second wave [22, 27, 28]. Other than a descriptive study of hospitalizations from residential care and nursing homes in England early in the pandemic, that showed few differences between the settings [24], research has yet to rigorously examine whether the COVID-19 pandemic and different pandemic waves (that varied in case numbers, infection virulence, vaccination rates, public health restrictions, and availability of staff and services) [28] differentially affected hospitalizations among AL compared with NH residents. This is a particularly salient research question given historical patterns of hospitalizations evident for these two settings caring for vulnerable older residents.

Our objectives were to compare rates of hospitalization events (all-cause hospitalization, hospitalization with delayed discharge, hospitalization with death) among residents of AL and NH settings in Alberta, before and following the onset of the COVID-19 pandemic, and to explore variation in rates by pandemic waves 1–4 for these two settings. We hypothesized that hospitalization rates would be higher for AL than NH residents (pre- and post-pandemic onset) and there would be significant decreases in rates for all-cause and delayed hospitalizations (but increases in hospitalization with death) that would be more pronounced for NH than AL residents and vary by pandemic wave.

Methods

Results

During the study period, there were 439,218 AL and 673,486 NH monthly records included after exclusions for deaths or hospitalizations at the time of monthly index dates (1–2% of records), representing 20,997 unique AL and 35,487 NH residents.

On March 1, 2020, there were 9,485 AL and 14,319 NH residents, representing 251 AL and 181 NH sites. The two settings were comparable in terms of resident age (mean 81 years), sex (63–67% female), number of chronic conditions (34% with 5+ conditions) and the presence of select conditions including dementia (58–62%) (Table 1). Relative to NH residents, those in AL showed significantly lower levels of ADL and cognitive impairment, depressive symptoms, health instability, frailty, and responsive behaviours. There were few differences in AL resident characteristics over time except for a significant decrease in the proportion of residents with meaningful depressive symptoms. Several NH resident characteristics also remained stable over time, however there were declines during the study period in the prevalence of frailty, multiple chronic conditions (5+) and specific conditions (dementia, arthritis) among residents in this setting.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Yearly characteristics of Alberta Assisted Living (AL) and Nursing Home (NH) residents,^a comparing select time periods.

https://doi.org/10.1371/journal.pone.0306569.t001

AL residents were more likely than NH residents to reside in the South zone, smaller homes and those owned/operated by voluntary non-profit and private for-profit organizations (rather than by AHS or contracted non-profit operators) (Table 2). The urban/rural distribution of residents was comparable for both settings. COVID-19 vaccination of both AL and NH residents started January 1, 2021 and by May 28, 2021, 85% had received their second dose. By study end about 70–80% had received a third dose.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Yearly home characteristics of Alberta Assisted Living (AL) and Nursing Home (NH) residents,^a comparing select time periods.

https://doi.org/10.1371/journal.pone.0306569.t002

There was a statistically significant decrease in the rate of all-cause hospitalization during the peak of pandemic wave 1 (April 2020) for residents of both AL (adjRR 0.60, 95% CI 0.51–0.71) and NHs (adjRR 0.74, 95% CI 0.64–0.85) relative to historical comparisons (Table 3, Fig 1 and S1 Fig). Over subsequent months up to and including pandemic wave 3, the hospitalization rate among NH residents increased slightly but remained significantly lower relative to historical comparisons (adjRR 0.82 to 0.85). For AL residents, the hospitalization rate increased after the initial wave 1 decline and showed no statistically significant difference relative to historical periods during the peak of wave 2 (December 2020) and the subsequent month (January 2021). During January 2021, the hospitalization rate among AL residents increased by 27% relative to that among NH residents (period*setting interaction p = 0.013). Both settings showed significantly lower hospitalization rates during pandemic waves 3 (April 2021) and 4 (September 2021), with a more pronounced decrease during wave 4 (AL: adjRR 0.76, 95% CI 0.66–0.88; NH: adjRR 0.65, 95% CI 0.56–0.75).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Monthly all-cause hospitalization rate (per 100 person-days) for historical (2018/19 combined) and 2020–2021 pandemic periods, among Assisted Living (AL) and Nursing Home (NH) residents.

https://doi.org/10.1371/journal.pone.0306569.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Adjusted rate ratios^a for all-cause hospitalization, hospitalization with delayed discharge, and hospitalization with death, comparing COVID-19 pandemic vs historical (2018/19) monthly periods, among Alberta Assisted Living (AL) and Nursing Home (NH) residents.

https://doi.org/10.1371/journal.pone.0306569.t003

Among AL residents, there were no statistically significant associations between the selected pandemic periods and rate of hospitalization with a delayed discharge (Table 3, Fig 2 and S2 Fig), whereas among NH residents, there was a statistically significant increase during both the peak of wave 1 (adjRR 1.94, 95% CI 1.31–2.85, with a statistically significant period*setting interaction, p = 0.008) and August 2020 (adjRR 1.84, 95% CI 1.19–2.86). Of note, both settings showed a statistically significant increase in rate of hospitalization with a delayed discharge during October 2020 (not an a priori exposure period of interest). This increase was more pronounced for NH (adjRR 2.57 (1.69–3.91) than AL (adjRR 1.55 (1.17–2.04) residents (period*setting interaction, p = 0.048).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Monthly hospitalization with delayed discharge rate (per 100 person-days) for historical (2018/19 combined) and 2020–2021 pandemic periods, among Assisted Living (AL) and Nursing Home (NH) residents.

https://doi.org/10.1371/journal.pone.0306569.g002

Among AL residents, there was a statistically significant increase in rate of hospitalization with death during the peak of wave 2 and in January 2021, followed by a significant decrease in this outcome during the peak of wave 3 (Table 3, S3 Fig). For NH residents, the only statistically significant association was for an increase in rate of hospitalization with death during wave 2 relative to historical period.

Both sensitivity analyses produced comparable findings to those presented in Table 3.

Rates of all-cause hospitalization and hospitalization with a delayed discharge were consistently higher among AL than NH residents throughout the study period (adjRR of 1.4–1.6 and 4.5–6.5, respectively; Figs 1 and 2, S2 Table). AL residents consistently showed higher median delayed discharge bed days (e.g., during January 2021, median 30 [IQR 6–52] bed days for AL vs median 9.5 [IQR 4–28] for NH residents, S3 Table).

Among the top 10 ICD-10-CA codes for the most responsible diagnosis (for hospitalizations during April 1, 2020 and January 1, 2021 vs historical counterparts) (S3 Table), dementia/mental health/delirium and injuries/fractures were more common and ranked higher among AL vs NH residents (especially during the pandemic months), whereas sepsis and renal failure/UTIs were more common and ranked higher during pandemic months among NH residents. Respiratory diseases were among the top two diagnoses for both settings but accounted for smaller proportion of diagnoses during pandemic periods (replaced by COVID-19 as the top condition in January 2021).

The monthly rates of COVID-19 hospitalizations were highest during wave 2, decreased following the implementation of vaccinations, and increased again during wave 4 (S4 Fig). During waves 2 and 4 the rates were higher among AL than NH residents. The proportion of all hospitalizations that were for COVID-19 was comparable for AL and NHs during December 2020 (27.0% vs 26.1%) and slightly higher for AL in January 2021 (19.5% vs 14.9%) and September 2021 (7.7% vs 5.6%). Monthly rates of positive COVID-19 tests were similar in AL and NHs (except for wave 1) and aligned with peak surges observed for COVID-19 hospitalizations (S5 Fig).

Discussion

In this population-based study of residents of AL and NHs in Alberta Canada, both settings showed pronounced decreases in all-cause hospitalizations during the first (40% and 26%, respectively) and fourth (24% and 35%, respectively) COVID-19 pandemic peaks but they varied significantly in this outcome during wave 2, and in the secondary hospital outcomes examined during selected pandemic periods. Whereas hospitalization rates among NH residents remained significantly below historical levels throughout, they increased for AL residents (by 27% relative to NHs) and were not significantly different from pre-pandemic rates during this period. For hospitalization with a delayed discharge, rates increased significantly during wave 1 (and into August 2020) for NH residents only, and though both settings showed a significant increase in rates of hospitalization with death during wave 2, this increase was larger and persisted longer (beyond peak of wave 2) for AL residents. The two settings also showed differences in the distribution of most responsible hospital diagnoses during pandemic periods and in COVID-19 associated hospitalizations (higher in AL than NH for waves 2 and 4), the latter despite generally comparable rates of positive COVID-19 tests.

The substantial decline in hospitalizations noted for both settings during the peak of wave 1 (consistent for all-cause and emergent admissions) aligns with other studies of chronically ill older adults in community [21] and long-term care settings [23–25, 28]. Few studies of congregate care have examined changes in healthcare use for AL residents [24] or for periods beyond 2020 (following the introduction of COVID-19 vaccinations). Expanding on earlier work, our findings reveal that with one exception (wave 2), pandemic periods examined up to the end of 2021 were associated with significantly lower hospitalizations from both settings. Complex and inter-related health system, home and resident/family factors may explain these lower rates, including the protection of acute care beds for the influx of expected COVID-19 cases early on in the pandemic [21], a reluctance by hospitals to accept AL or NH residents and/or by care homes or family to transfer residents to hospital because of the perceived risk of infection [25], changes in resident conditions typically associated with hospitalization [23, 24, 27], and a failure to recognize emergent health conditions requiring acute care given staffing shortages and family absence [27]. From a resident quality of care perspective, these findings may be viewed positively (e.g., there were fewer unnecessary hospital transfers from care home settings) [18, 43, 44] or negatively (e.g., there were more residents with serious unmet clinical care needs at heightened risk for death or functional and cognitive decline) [19, 24, 45]. Both possibilities warrant further investigation.

The evident rise in hospital admissions for AL residents in wave 2 (back to historical levels), may reflect an increased likelihood for AL homes to transfer residents with COVID-19 and other conditions to hospital and/or greater illness severity in AL vs NH residents (perhaps facilitated by less effective isolation measures or greater interactions among AL residents) during this period [20]. In both settings, positive COVID-19 cases among residents were especially high during wave 2, dropped dramatically following widespread resident vaccinations, and increased again during wave 4. It may also be that AL residents experienced more adverse health outcomes (e.g., inappropriate medication use [46], falls/injuries [47], worse mental health and/or disease progression) and potentially avoidable hospitalizations during this surge because of reductions in already low levels of professional staffing and services [8–10, 12–14] and absence of family caregivers [3]. Our data regarding setting-specific changes in the distribution of most responsible diagnoses for hospitalizations during waves 1 and 2 and consistently elevated rates of hospitalization (including admissions with delayed discharge and higher bed stays) for AL relative to NHs throughout pre- and post-pandemic periods partially support this explanation.

Though there is no evidence of changes in AL or NH admission and discharge criteria during the pandemic [48], there may have been differences in the enactment of these criteria across care settings and/or adoption of strategies to prevent avoidable hospitalizations (e.g., increased communication between NH sites and emergency department physicians or involvement of community paramedics, [49]) that contributed to the setting-specific hospitalization findings noted above.

The increased rate ratios for hospital admission with a delayed discharge shown for NH residents early in the pandemic, and for both settings during Oct 2020, may be related to the reduced availability of NH and AL beds as homes addressed concerns of overcrowding and the need for isolation spaces during periods of peak COVID-19 cases [21, 50]. These findings may also reflect hesitation among physicians, staff and/or families to have residents discharged back to AL or NHs during peak surges because of a limited availability of staff to care for residents who likely had increased health needs.

Both settings experienced a significant increase in hospitalization with death during the peak of wave 2 relative to historical periods (coinciding with peak COVID-19 cases and hospitalizations), but the magnitude of this increase was larger and persisted into early 2021 for AL residents. As suggested by our data on most responsible diagnoses, AL residents were more likely to be admitted with health conditions associated with increased hospital bed days and mortality (e.g., dementia, mental health issues, delirium) [51]. They may also have presented with more severe health conditions related to less timely detection of emerging health issues and limited ability to augment care in the AL home [8, 13]. These hospital deaths are disconcerting given residents’ and families’ preferred place of death (at home) and that many residents likely died alone because of public health restrictions.

Conclusions

Hospitalizations represent an important quality of care indicator for AL and NH populations, reflecting the importance of potentially preventable hospital transfers as well as equitable access to appropriate and timely medical care. This study showed that the associations between various pandemic waves and hospitalization events varied in important ways for AL compared with NH residents, further highlighting the unique challenges faced by different congregate care settings during COVID-19 [4, 5] and their varying responses to this external stress. Our findings support the presence of setting-specific drivers of residents’ healthcare use and outcomes both prior to and during COVID-19 that require careful consideration by continuing care operators, health care providers and policy decision makers. These setting-specific differences should also inform future interventions [44, 52, 53] aimed at strengthening the integration, communication and support between congregate care and acute care settings. Considerations of the clinical complexity of AL residents and hospitalization trends observed (both historically and during COVID-19) strongly support previous calls for a blended social-medical model of care in AL homes [1, 6, 16]. Strategies to be investigated include more routine access to primary care physicians and/or nurse practitioners (both on-site and via telemedicine) [54], enhanced levels of well-trained professional staff, clearly documented advance directives, and the development of policies and procedures specific to the appropriate detection, treatment, and management of the chronic physical and mental health conditions common in AL residents.

Supporting information

Acknowledgments

We would like to thank the Alberta SPOR SUPPORT Unit for their assistance in accessing the data used in this study.