Factors associated with entry to residential care in frail older inpatients

To examine factors that may influence the risk of discharge to a residential aged care facility (RACF) in a population of frail older inpatients.


| INTRODUCTION
As the population ages, there will be a growing number of older adults at risk of entering residential aged care facilities (RACFs). 1 Risk factors for entry to an RACF for community dwellers are well-described: at present, the strongest level of evidence is for increasing age, poor self-rated health, functional impairment, cognitive impairment (including dementia diagnosis) and polypharmacy. 2Similar factors are associated with increased risk of discharge to an RACF from hospital. 3ompared with adults entering an RACF from the community, adults entering from hospital tend to be more dependent and multimorbid with a history of frequent hospitalisations. 4egardless of the setting, 'frail' individuals may have many, if not all, of these risk factors.Frailty is a state of increased vulnerability to stressors. 5Frail individuals have a reduced ability to recover from acute insults, which leads to adverse outcomes including falls, dependence, entry into residential aged care and mortality. 5he frailty index (FI) model conceptualises frailty as an accumulation of health 'deficits', whereby the more problems an individual has, the more frail they are. 6The FI model employs a well-defined method to create an index as a proportion of impairments across different domains (including mobility, cognition, chronic disease and function).As a continuous variable ranging from zero (not frail at all) to a theoretical maximum of one, the FI is a concise measure of overall health status.Approximately 30% of community dwellers assessed for Australian aged care services between 2003 and 2013 (N = 903,996) were found to be frail. 7rail community dwellers and hospital patients are at increased risk of entry to RACF. 8,9However, there may be demographic, social and health factors that not only increase the risk of entry to an RACF independently of frailty but also modulate the relationship between frailty and entry to an RACF.For example, sex may be an important risk factor for entry to an RACF from hospital, although which sex is at higher risk remains unclear. 3,4ex differences in frailty are well-described; sex modulates the relationship between frailty and death, with women facing a lower risk of death than men despite higher levels of frailty. 10It is not known whether sex also modulates the relationship between frailty and other adverse outcomes, such as discharge to an RACF.
In the FI model, it is the number of deficits, rather than the nature of the deficits, that influences the relationship between the FI and adverse outcomes. 6All deficits are weighted equally as small gains in predictive power result in significant losses to generalisability. 11Even so, some health deficits contribute to caregiver burden to a greater extent than others and, as a result, are likely to increase the risk of entry to an RACF to a greater extent than others.In community dwellers with dementia, for example, the presence of behavioural and psychological symptoms of dementia (BPSD) is a risk factor for entry to an RACF. 12aecal incontinence is another debilitating health condition that impacts frail older adults, particularly those with dementia, and places a significant burden on caregivers. 13ne might hypothesise that these health deficits would influence the relationship between frailty and discharge to an RACF from hospital.
Health assets are resources that individuals have to protect against negative health outcomes.Social support (or lack, thereof) has been associated with entry to an RACF from the community. 2In a study of hospital inpatients, living alone increased the risk of discharge to an RACF by 85%, whereas having a person supportive of discharge reduced the risk by 15%. 14Frailty, as measured by the FI, remained an independent predictor in these multivariate models.It is unclear, however, whether these factors modulated the relationship between the FI and discharge to an RACF.
Our study aimed to address this gap in the literature by examining the impact of these demographic, health and social factors on the risk of discharge to an RACF in a population of frail, older inpatients.

| Study setting and participants
This study was a secondary analysis of the Comprehensive electronic Geriatric Assessment (CeGA) data set.The Comprehensive electronic Geriatric Assessment includes 7755 assessments of inpatients referred by treating (medical or surgical) teams for geriatric medicine input regarding assessment and management of complex care needs, multimorbidity and cognitive impairment, and/or suitability for admission to subacute care or RACF.Patients were admitted to one of 27 hospitals in Queensland, Australia, between March 2007 and December 2018.
Patients were included in the current study if they were aged >60 years, were admitted from independent living and discharged either to an RACF or private residence.

Practice Impact
This study yielded novel results regarding social and health factors that may increase or decrease the risk of discharge to a residential aged care facility (RACF) in a population of frail older inpatients.Potentially modifiable factors may be targets for intervention while non-modifiable factors may inform service planning.Future work should address whether early identification and management of potentially modifiable factors decrease risk of discharge to an RACF from hospital.

| Assessment
Every patient in the CeGA data set underwent an inter-RAI™ Acute Care Comprehensive Geriatric Assessment (InterRAI AC-CGA). 15Trained nurses gathered information from in-person consultation with patients, family and allied health staff as well as from hospital medical records.The assessment period was defined as the preceding 24 h.The interRAI AC-CGA gathers information regarding multiple domains including sociodemographics, cognition, mood and behaviour, functional status, continence, health conditions and patient discharge.

| Frailty
A 52-item frailty index (FI) derived from the interRAI AC-CGA using a standardised method has been described and validated in previous studies. 16Domains included cognition, mobility, function, sensorium, continence, medical co-morbidity and medication use (Appendix S1: Figure S1).An FI was calculated for each patient by adding the number of 'deficits' and dividing by the number of variables (potential deficits).For example, a patient with 13 deficits (out of 52 potential deficits) would have an FI of 0.25.

| Risk and protective factors
Factors were determined a priori.They were female sex, presence of behavioural and psychological symptoms of dementia (BPSD), presence of faecal incontinence, being married/in a de facto relationship and living with others.Female sex, BPSD and faecal incontinence were considered risk factors, while the remaining were considered protective factors.

| Outcome
The primary outcome was the discharge destination: RACF versus private accommodation (private residence or independent living unit).

| Statistical analysis
Data analysis was performed using STATA v17. 17Pearson's chi-squared tests were used to compare differences between categorical variables while two sample t-test or Wilcoxon rank-sum test were used to compare continuous variables between patients discharged to RACF and patients discharged to a private residence/independent living unit.
Logistic regression was used to assess the relationship between factors and discharge destination, adjusted for frailty status (continuous), age and sex, and including hospital as a cluster variable.The association of frailty with each factor (sex, marital status, living arrangement, presence of BPSD and faecal incontinence) was assessed with univariate linear regression models.Logistic regression analyses with frailty × risk factor interactions were also performed.All models were checked to ensure they met the appropriate model assumptions.
The 52-item FI included faecal continence and BPSD as variables.Consequently, for analyses examining the relationships between these variables, the discharge destination and/or the FI, the FI was adjusted to remove the variables from the denominator and numerator as required.
This project was reviewed by the Office of Research Ethics (approval #: 2021/HE000811) at the University of Queensland and deemed to be exempt from ethics review under the National Statement on Ethical Conduct in Human Research.

| RESULTS
Of the 7755 patient assessments in the database, 5846 were analysed in this study.Of the 7755 patients in the database, 316 were excluded because they were under the age limit, 433 were excluded as they came from an RACF, 623 were excluded because they did not have a recorded discharge destination, 525 were transferred to other temporary accommodation, nine were missing FI data and three had died (Appendix S1: Figure S2).One thousand six hundred and seventy-eight (29%) were discharged to an RACF and 4168 (71%) were discharged to private accommodation.Participant characteristics are presented in Table 1.

| Patient characteristics
Patients had a mean (SD) age of 79.7 (8.2) years.The majority were women (54%) and born in Australia (73%).Most were not married or in a de facto relationship (62%) but were living with others (54%) at the time of admission to hospital.Patients had a mean (SD) FI of 0.44 (0.14) and the 99th percentile was 0.76 (Appendix S1: Figure S3).
Compared to those discharged to private accommodation, patients discharged to an RACF were older (81.4 vs. 79.0years, p < 0.001), were less likely to be married or in a de facto relationship (30% vs. 41%, p < 0.001) and were more likely to be living alone (51% vs. 44%, p < 0.001).There were no sex differences in discharge destination (p = 0.79).Patients discharged to an RACF were more likely to have BPSD (24% vs. 10%, p < 0.001) and be faecally incontinent (35% vs. 20%, p < 0.001) than patients discharged to private accommodation.The FI of patients discharged to RACF was significantly higher than for those discharged to private accommodation (mean FI = 0.50 vs. 0.42, respectively, p < 0.001).

| Independent risk and protective factors for discharge to an RACF
Each 0.1 increment in the FI was associated with 1.54 times the odds of being discharged to an RACF (Table 2; Model 1-adjusted for age, sex and hospital).All other factors, except for female sex, were independently associated with the primary outcome (Models 2 to 6-all adjusted for FI, age, sex [except Model 2] and hospital).
Being partnered and living with others reduced the risk of discharge to an RACF by 48% and 42%, respectively, when compared to being single or living alone.The presence of BPSD almost doubled the risk of discharge to an RACF and faecal incontinence increased the risk by 27%.

| Risk and protective factors modulating the relationship between frailty and discharge to an RACF
All factors, except sex (p = 0.34), were significantly associated with frailty (p < 0.001) in univariate linear regression models (Appendix S1: Table S1).Multivariate logistic regression analysis with interaction terms for frailty by sex, marital status, prior living arrangement, BPSD and faecal continence, on risk of discharge to an RACF, was performed.Significant interaction terms were observed for marital status and BPSD (both p < 0.001; Appendix S1: Table S2).Follow-up tests of effect sizes showed that being married or in a de facto relationship reduced the risk of discharge to an RACF beyond an FI of 0.70 (Figure 1) while BPSD increased the risk of discharge to an RACF up to an FI of 0.70 (Figure 1).

| DISCUSSION
We investigated factors associated with entry to residential aged care in a group of older inpatients undergoing comprehensive geriatric assessment.Our study yielded novel results regarding social and health factors that may modify the risk of discharge to an RACF in this population.First, using a validated FI derived from comprehensive geriatric assessment, we demonstrated that older inpatients were severely frail (mean FI = 0.44).While the FI ranges theoretically from zero to one, only 1% (approximately) of older adults in community and residential aged care settings have been found to have an FI >0.7. 18n this study, however, 4% of patients exceeded this value.This finding is consistent, however, with another study conducted in older hospital patients receiving geriatric medicine input in the postacute setting 19 indicating that the older inpatient population includes the 'frailest of the frail'.
Second, we found that increasing frailty, BPSD and faecal incontinence were independent risk factors, and married/de facto relationships and living with others were independent protective factors for discharge to an RACF from hospital.While the association between frailty and entry to residential aged care has previously been recognised, 9 we ascertained that the presence of BPSD and being married or in a de facto relationship influenced this association.
In this study, being married or in a de facto relationship had a protective effect regardless of FI.The Australian Longitudinal Study of Ageing found that social networks and confidants are protective factors 20 and while a spousal relationship may offer social support, it is the physical and functional support that may moderate the relationship between frailty and entry into residential aged care.While marital status is not a modifiable risk factor, it may prompt clinicians to identify opportunities to assist single older adults at risk of entering residential care.For example, it may trigger a referral for aged care assessment with view to commencing an in-home care package or a referral to local services to provide social and emergency support. 21ehavioural and psychological symptoms of dementia increased the risk of discharge to RACF up to an FI of 0.70 in our study.The relationship between BPSD and frailty identified in this study is consistent with previous research linking frailty with dementia 22 and dementia with entry to residential aged care. 23While severely disruptive and distressing behaviours and symptoms may indicate more advanced dementia (with associated functional and physical limitations), they may also indicate severely strained carer-caregiver relationships (with or without advanced dementia).Consequently, patients with BPSD may require additional care regardless of their frailty status.As a potentially modifiable risk factor, management of BPSD, and of dementia more broadly, may decrease the risk of discharge to an RACF among frail older adults.
While faecal was independently associated with FI and with discharge to an RACF in this study population, it did not influence the relationship between frailty and discharge destination.This may be due to the fact that in more robust patients, faecal incontinence is likely to result from other medical comorbidities and may be easily managed alone or with some assistance, whereas in frail patients, faecal incontinence is likely to arise in the setting of severe functional, physical or cognitive impairment. 13e hypothesised that female patients would be more likely to be discharged to an RACF and that sex would modulate the relationship between frailty and discharge destination.Approximately two-thirds of RACF residents in Australia are women 24 and women (in general) are impacted by key risk factors including higher frailty, higher rates of dementia and increased likelihood of widowhood and living alone to a greater extent than men. 25,26However, we did not find evidence for sex differences in the risk of discharge to an RACF or for a 'sex x frailty' interaction in this study.The lack of sex differences in frailty in older inpatients has been described before, and one explanation for this finding is that men are likely to experience a more severe acute illness, which may eliminate sex differences in frailty on admission to hospital. 27,28he strengths of this study include the comprehensive data collection using a standardised instrument from a large cohort of inpatients across several medical disciplines and sites.This study also has important limitations.For example, we used a cross-sectional data set that cannot tell us about the causality of the relationships identified in our results.Furthermore, generalisability of the study findings is limited by the study sampling: CeGA is not available in all Queensland Hospitals, not all frail older patients undergo CeGA and not all patients entering RACF from hospital undergo CeGA.Consequently, the sample may not be representative of frail older inpatient populations across Queensland, or more widely.Finally, we selected five variables of interest for this exploratory analysis.It is likely that there are other important factors that influence the risk of discharge to an RACF in frail inpatients.This is an important future line of enquiry.

| CONCLUSIONS
It is well-documented that older adults prefer to remain living in the community 29,30 ; it is important to identify risk and protective factors for entry to residential aged care, particularly potentially modifiable factors that may be targets for intervention.In the hospital setting, understanding risk and protective factors may inform effective service planning and support individuals and their caregivers through this significant transition.This study yielded novel results regarding social and health factors that may increase or decrease the risk of discharge to an RACF in a population of frail older patients.Future work should address whether early identification and management of mutable factors can decrease risk of discharge to RACF from hospital.

F I G U R E 1
Probability of discharge to an RACF by FI for those (A) who are or are not married/in a de facto relationship and (B) with or without BPSD.CI, confidence interval; FI, frailty index; BPSD, behavioural and psychological symptoms of dementia; RACF, residential aged care facility.
T A B L E 1Abbreviations: BPSD, behavioural and psychological symptoms of dementia; IQR, interquartile rangeRACF, residential aged care facility; SD, standard deviation.
Multivariate logistic regression of risk and protective factors for discharge to an RACF.Note: Models were adjusted for FI, age, sex (exc Model 2) and hospital as a cluster variable.Reference categories: male sex, not married/de facto, living alone, no BPSD, not faecally incontinent.
T A B L E 2