Effectiveness and analysis of factors predictive of discharge to home in a 4‐year cohort in a residential transitional care unit

Abstract Objective The aim of this study was to evaluate the effectiveness and identify factors predictive of home discharge in a cohort of patients admitted to the residential Transitional Aged Care Program (r‐TACP) after a stay in an acute hospital. Methods A retrospective observational cohort study of patients admitted to a single r‐TACP unit between 1 January 2014 and 31 December 2017 was carried out. Baseline patient characteristics and discharge outcomes were analyzed. Results Three hundred sixty‐nine patients were admitted during the study period. The discharge outcomes were as follows: 68% returned home, 17% went onto residential care, 14% were readmitted to hospital, and 1% died. Factors associated with not returning home were increased age, increased comorbidities, and lower Barthel Index on admission to the r‐TACP. Conclusion Our r‐TACP is an effective program that successfully returns the majority (67.8%) of older patients home after an acute hospital admission. Older patients with greater comorbidities and poorer baseline functional status in our program were less likely to return home.

the samples had been collected from three different services. With heterogeneity and a small sample size, the authors could not draw any definitive conclusion. 2 Subsequently, a national evaluation summarized 23 reports (2443 patients) on the TACP operation completed between September 2006 and May 2007 1 and concluded that older patients who received TACP were deemed to be frail, with 37% returning to hospital at least once within 3 months, and 47% within 6 months.
However, the report did demonstrate reduction of residential aged care facility (RACF) placement and hospital re-presentation rates at 6 months for those who used the TACP compared with historical controls. Of note is that the report grouped both c-TACP and r-TACP together in its analysis and there was heterogeneity in the services model (49% received c-TACP, 42% received r-TACP, and 9% received both), making it difficult to conclude definitively. 1 While theoretically, this is a service that could reduce the length of hospital stay, reduce cost, and help patients return home, few reports in the Australian setting have rigorously evaluated these salient outcomes. 3,4 Although there are a few other studies assessing outcomes of similar programs, these outcomes varied. Different patient characteristics and service models might have contributed to the differences in results and, importantly, none have reported on an r-TACP. [5][6][7] Summing up, despite the small number of studies published formally and informally to date, the results produced have been conflicting regarding the utility of TACPs. Importantly, there have been no published consensus data about patient selection into TACPs that would lead to desirable outcomes; this is especially the case for the r-TACP, a place where frail older patients are preferentially discharged instead of to the c-TACP. As Parsons et al 6 there is a paucity of information about what type of patients would benefit most with desirable outcomes, such as "returning home" or "not being readmitted to hospital." Allen et al 8 in their systematic review of the c-TACP also noted the variability of outcomes and service models, and we note that there is no r-TACP study included in their review.
With this background, we set out to study if patients' baseline characteristics would be good predictors for desirable outcomes (returning home, not readmitted to hospital) in a single residential TACP center.

| Primary objective
Our primary objective was to evaluate the effectiveness and identify the baseline factors associated with successful discharge home from the r-TACP unit.

| Secondary objective
Our secondary objective was to evaluate the length of stay in the r-TACP unit and the improvement in function of the r-TACP cohort.

| Study design and service model
This was a retrospective study assessing 4 years' data set up since the inception of the r-TACP in metropolitan South Western Sydney.
The r-TACP unit consists of 13 beds, based within a nongovernmentrun aged-care facility (at hostel level) that also provides respite and permanent care for other residents. The r-TACP unit is located separately from the other services and is staffed with designated on-site nursing and part-time allied health as well as being serviced by threetimes-weekly geriatrician visits. In our local health district, there exists another TACP model, namely, the c-TACP, whereby patients receive rehabilitation services from a multidisciplinary team at their own homes (following acute illnesses) instead of a residential facility. Patients are usually less frail but they also require prior approval from the Aged Care Assessment Team. Patients from the two models are serviced by separate teams and not mixed together.
There are 13 beds in our r-TACP unit with nursing staffed at the hostel level and salaries paid by the nongovernment organization that owns the facility. In addition, the rehabilitation is serviced by a multidisciplinary team led by a part-time geriatrician and paid for by our Health Department (Table 1).

| Patients' data collection
All patients admitted to the unit between 1 January 2014 and 31 December 2017 were identified from the preexisting r-TACP electronic database. Two reviewers (S. Z. and Y. L.) reviewed all patients' data and retrieved relevant information from the database. S. Z. did the preliminary data collection, which was counterchecked by Y. L., and any discrepancy was reconciled under the supervision of C. U. and D. K. Y. C. Study data were extracted from the database.
Where there were missing data, hospital records were reviewed by

| Discharge destination and other patient outcomes
The main outcome measures of this study were factors associated with home as the discharge outcome at the end of the r-TACP unit stay. There were four discharge outcomes: returning home, readmission to hospital, discharge to RACF, or death. Other study outcomes were length of stay at the r-TACP unit and change in BI on discharge from the r-TACP unit compared to on admission to the r-TACP unit.

| Reasons for hospital admission
The reasons for hospital admission prior to transfer to the r-TACP unit between the two groups (those who returned home and those who did not) are shown in Table 3. The most common diagnosis was fractures (38.5%), followed by medical illness other than falls (26.6%), falls (13.8%), and stroke (7%).

| Discharge destination and other patient outcomes
A total of 250 patients (67.8%) returned home after r-TACP. Fortyfive patients (12.1%) were transferred to low-level residential care, 18 (4.9%) were transferred to high-level residential care, 48 (13.0%) were readmitted to hospital, three (1.0%) died at the r-TACP unit, and five (1.4%) patients were discharged to another destination.
The average length of stay at the r-TACP unit was 46.8 ± 22.5 days.
Patients who returned home stayed on average 5 days longer than patients who did not return home but this did not reach statistical significance (48.3 vs 43.6 days, P = .083).
Functional status was measured by the baseline BI on admission to the r-TACP unit and the final BI on discharge from the r-TACP unit.
Patients who were readmitted to hospital or died during r-TACP were pragmatically scored a final BI score of zero in the database and were therefore excluded in the statistical analysis. Overall, the mean BI on admission was 65.8 ± 11.5 and increased to 86.5 ± 10.7 on discharge, achieving a mean gain of 20.3. Patients who did not return home had lower baseline BI and achieved smaller gains in BI on discharge compared to patients who returned home (17.3 vs 21.1, P = .020).

| Baseline factors associated with home as discharge destination
Baseline factors predictive of returning home were identified from a backward stepwise logistic regression model. The base model included demographic factors of age group, CCI, presenting principal diagnoses for hospitalization, baseline BI, and dementia (

| D ISCUSS I ON
Out of the 369 patients in our r-TACP over 4 years, the majority (68%) returned home. We also found that CCI was a reasonable predictor associated with homeward outcome. A mean CCI of 3 was associated with the less desirable outcome of RACF placement, while a lower mean CCI of 2.2 was associated with a homeward discharge destination. The mean gain in BI was 20.3 and this was achieved at the expense of 46.8 days' stay in the program.
A 2008 report 1 found that by the 3-month follow-up, 35% of patients had been readmitted to hospital at least once, 45% were in permanent residential care, and 14% had died. By 6 months, these numbers had increased to 43%, 58%, and 20%, respectively.
Unfortunately, we do not have the follow-up data of 3 and 6 months to compare, and even if we did, the previous data were from a decade ago, and many variables that may affect outcomes are not readily available for consideration.
The CCI was initially validated in an acute hospitalized setting to predict 1-year mortality, with higher scores being associated with higher mortality risk. 9 While overreliance on this simple tool to select patients for our r-TACP may not be ideal, it does give us confidence that the current patient selection process appears to yield reasonable results (68% of patients were able to return home). If too many RACF-bound patients are admitted, it creates patient flow challenges for hospitals as patients frequently wait in acute hospital beds until an RACF bed is available.

| CON CLUS ION
Our r-TACP is an effective transitional program that successfully returns the majority of frail older patients home after an acute hospital admission. Older patients with greater comorbidities and poorer baseline functional status in our program were less likely to return home although our current selection process appears to be adequate in screening out many patients who may not benefit. The r-TACP model of care can be considered for use in similar hospitals with similar patient demographics.

ACK N OWLED G EM ENTS
We acknowledge with gratitude Friedbert Kohler, Elayne Armer, and John Loy for their assistance in the preparation of the paper.

CO N FLI C T S O F I NTE R E S T
Nothing to disclose. Abbreviations: CI, confidence interval; OR, odds ratio.