The impact of facility-based transitional care programs on function and discharge destination for older adults with cognitive impairment: a systematic review

Background Older adults with cognitive impairment are frequently hospitalized and discharged to facility-based transitional care programs (TCPs). However, it is unknown whether TCPs are effective in improving their functional status and promoting discharge home rather than to long-term care. The aims of this systematic review were to examine the effectiveness of facility-based TCPs on functional status, patient and health services outcomes for older adults (≥ 65 years) with cognitive impairment and to determine what proportion post TCP are discharged home compared to long-term care. Methods The Joanna Briggs Institute Critical Appraisal Manual for Evidence Synthesis was used to guide the methodology for this review. The protocol was published in PROSPERO (registration number CRD42021257870). MEDLINE, CINAHL, PsycINFO, the Cochrane Library, and EMBASE databases, and ClinicalTrials.gov and the World Health Organization Trials Registry were searched for English publications. Studies that met the following criteria were included: community-dwelling older adults ≥ 65 years who participated in facility-based TCPs and included functional status and/or discharge destination outcomes. Studies with participants from nursing homes and involved rehabilitation programs or transitional care in the home or in acute care, were excluded. Risk of bias was assessed using the Joanna Briggs Institute Critical Appraisal Checklists. Results are in narrative form. Results Twenty-two studies (18 cohort and four cross sectional studies) involving 4,013,935 participants met inclusion criteria. The quality of the studies was mostly moderate to good. Improvement in activities of daily living (ADLs) was reported in eight of 13 studies. Between 24.4%-68% of participants were discharged home, 20–43.9% were hospitalized, and 4.1–40% transitioned to long-term care. Review limitations included the inability to perform meta-analysis due to heterogeneity of outcome measurement tools, measurement times, and patient populations. Conclusions Facility-based TCPs are associated with improvements in ADLs and generally result in a greater percentage of participants with cognitive impairment going home rather than to long-term care. However, gains in function were not as great as for those without cognitive impairment. Future research should employ consistent outcome measurement tools to facilitate meta-analyses. The level of evidence is level III-2 according to the National Health and Medical Research Council for cohort and cross-sectional studies. Supplementary Information The online version contains supplementary material available at 10.1186/s12877-022-03537-y.


Background
As a result of the growing aging population there is a greater urgency to establish and maintain effective health care systems and programs. According to the World Health Organization, the proportion of adults over the age of 60 globally will increase from 12 to 22% between 2015 and 2050 [1]. Moreover, the number of people with dementia will almost double, from 50 million people worldwide in the year 2020 to 82 million in 2030, and 152 million in 2050 [2]. Cognitive impairment (CI), which can include dementia, delirium, and unspecified CI [3,4], has a global prevalence of 5.1-37.5% among older adults aged 60-69 years, with a median of 20.1% [5]. Given the prevalence of CI in older adults and the growing number of people with dementia, there is an increasing demand for health care services that effectively meet their needs and facilitate positive health outcomes.
Systematic reviews have shown that older adults with CI have poorer health outcomes, including a higher risk for hospitalization [6], and increased risk for functional decline when hospitalized [7], and a higher risk for discharge to institutional long-term care post hospitalization [8], compared to those without CI. Moreover, recent reviews have shown that CI is associated with an increased length of hospital stay [9] and delayed discharge [10], which is problematic as these factors are associated with increased mortality, depression, and a decline in mobility and activities of daily living (ADLs) [11]. Therefore, these reviews highlight the need for specialized programs to help older adults with CI achieve positive outcomes such as improvement in functional status and discharge home.
After the acute issue is treated, some older adults remain in hospital longer due to the lack of community supports [12] or as the result of additional functional decline [13]. Thus, facility-based transitional care programs (TCPs) are one possible solution to facilitate discharge for these individuals. In this review, a facilitybased TCP is defined as a post-acute program or unit within a facility which provides short-term, restorative care [14,15] to older adults. Restorative care involves transitioning from providing full care to older adults to providing assistance to older adults, in order to maintain or improve functional abilities [16]. In terms of intensity, restorative care can involve two or more activities such as walking, mobility, and dressing for at least 15 min a day, six days a week [17]. Restorative care differs from inpatient rehabilitation programs in terms of therapy intensity, as inpatient rehabilitation programs are often high intensity, are typically 4-6 weeks in length, involve daily medical and nursing care, and 30-60 min physical and occupational therapy up to 5 times per week [18]. Throughout the literature, facility-based TCPs may be called by different names. In the United States, they may be called subacute care, post acute care, and skilled nursing facilities [14]. They are called intermediate care models in the United Kingdom, transition care programs in Australia, and transitional care programs in Canada [14]. These programs will hereafter be referred to as TCPs in this paper.
A recent scoping review found that TCPs admit older adult patients both with and without CI [14]. Moreover, functional status was the most common patient outcome, while discharge destination was a frequently used health services outcome [14]. Meta-analyses have shown that TCPs can significantly improve an older adult's ability to perform ADLs, resulting in 80% of participants being discharged home [19], and a significant reduction in hospital readmission rates [20]. However, there are no reviews to date that have determined the impact of TCPs on functional status and discharge destination outcomes for older adults with CI. Given the growing aging population and increasing number of older adults with CI who are most at risk to decline functionally, it is critical that a review be undertaken to inform the creation, modification, and maintenance of effective TCPs for this population.
The review questions were: 1) What is the effectiveness of facility-based TCPs on functional status, patient and health services outcomes for older adults (≥ 65 years) with CI? 2) What proportion of older adults with CI at the end of the TCP are discharged home compared to long-term care?

Methods
The Joanna Briggs Institute Critical Appraisal Manual for Evidence Synthesis (April 2021) [21] was used to guide the methodology for this systematic review and the results are reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 checklist [22]. The review protocol was published in PROSPERO (https:// www. crd. york. ac. uk/ prosp ero/; registration number CRD42021257870).

Search strategy
Comprehensive, systematic searches of OVID MED-LINE, CINAHL, PsycINFO, the Cochrane Library, and EMBASE databases were completed on July 15, 2021, from inception to present. The searches were updated on July 9, 2022. The search strategy was developed and refined by AC in consultation with KSM, TJFC, MTP, and a library information sciences expert (MM).
The three key search terms were: 1) transitional care programs; 2) older adults; and 3) cognitive impairment. In this review, cognitive impairment includes dementia, delirium, and non-specified cognitive impairment, as differentiating between them can be challenging [23]. Long-term care includes long-term care homes, nursing homes, and care homes [24]. Reference lists of included studies and reviews were also hand searched for relevant articles. The full search strategies and search results for each database can be found in Additional file 1.
Registries of ongoing trials from ClinicalTrials.gov and the 17 primary registries on the World Health Organization website [25], were searched independently by two reviewers (SR and AC or NZ). See Additional file 2 for registry search strategies, results and the dates the registries were last searched. Grey literature was not included in this review.

Study selection
Pilot testing of the search strategy was completed by two independent reviewers (AC, PS, SR). Titles and abstracts were screened by two independent reviewers (AC, PS, SR, CW); full texts of studies were also screened by two independent reviewers (AC, SR, PS, NZ). Disagreements were resolved by discussion and consensus with a third reviewer (the other of AC, PS, SR, SW, or NZ). Covidence systematic review software [26] was used to manage and record data decisions.
Studies were eligible for inclusion if the following criteria were met: 1) included community-dwelling older adults (mean age ≥ 65 years) with CI (dementia, delirium, and/or CI) who were hospitalized and then admitted to a facility-based TCP; 2) TCPs were delivered in skilled nursing facilities, nursing homes, subacute and post acute units in hospitals, geriatric intermediate care facilities, and convalescent care [14,15]; 3) included functional status and/or discharge destination as outcomes, with functional status defined as the ability to perform activities needed in daily life [27], measured using a validated tool, such as the Barthel Index, and discharge destinations including home, long-term care, and hospital; 4) published as a full length manuscript in a peer-reviewed journal; 5) designated as primary and secondary interventional studies (RCTs, quasi-experimental), primary and secondary observational studies (prospective cohort, retrospective cohort, cross-sectional, and case-control), and mixed-methods studies if there was quantitative data on functional status and/or discharge destination; 6) published in English.
The exclusion criteria were: 1) reviews, case studies, dissertations, conference proceedings, editorials, and qualitative studies; 2) mean age of participants < 65 years old; 3) participants living in a long-term care facility prior to hospitalization and TCP admission; 4) participants who were at the end of life (< 6 months prognosis) [28]; 5) rehabilitation programs; 6) transitional care provided in the home; 7) transitional care services provided only in acute care.

Data extraction
Data were independently extracted by two reviewers (AC and SR, PS, CI, NZ, or TC) using a pre-piloted extraction form created with Microsoft Excel 2019. Information about the study design and methodology, TCP characteristics (staff complement, description of TCP services, inclusion and exclusion criteria), participant characteristics, and all outcome measures were extracted from the articles. The outcomes were reported according to the classification of outcomes as outlined in McGilton et al. [14]. The primary outcomes were functional status and discharge destination post TCP. The secondary outcomes were divided into patient outcomes, such as mortality, and health services outcomes, such as rehospitalization [14]. Disagreements between individual judgments were resolved by discussion and consensus. Authors [29][30][31][32][33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48] were contacted to ascertain any required information that was missing or unclear and data provided directly by the authors [29,30,47] was included in this review (See PRISMA diagram, Fig. 1). Information that was not in the study was reported as 'NR' (not reported). Extracted data for this review can be found in Additional file 4.

Risk of bias assessment
The Joanna Briggs Institute Critical Appraisal Checklists [49] were used to assess the risk of bias in the included studies. The checklists were completed by two independent reviewers (AC, KSM, PS, TJFC) and disagreements were resolved by discussion and consensus. No studies were excluded on the basis of quality. Fair quality was assigned to studies if less than or equal to 50% of the checklist items were given a rating of yes, moderate quality if 51-80% of items were given a rating of yes, and good quality if greater than 80% of items were given a rating of yes, based on the rating system used by Benenson et al. [50].

Synthesis of results
Results were synthesized in narrative form, using tables and figures according to outcome measure. A meta-analysis was not performed due to heterogeneity in the outcome measurement tools and data measurement times. There was also heterogeneity in the patient populations; although all participants had some form of CI, some studies focused on specific populations, such as veterans or older adults with heart failure or traumatic brain injury. Figure 1 shows the PRISMA flow diagram which outlines the study selection process. The search of the databases yielded 14,556 articles, the search of the trial registers yielded 12 articles, and hand searches of reference lists of included studies yielded 222 articles, with 2,714 duplicates in total. After completing title and abstract and full text screening, 30 studies met the inclusion criteria. Among the 30 studies that included older adults with CI, only 22 performed a subgroup analysis or had separate data for older adults with CI. Therefore, data was extracted from the 22 studies which reported information on a total of 22 TCPs and 4,013,935 study participants.

Performance of ADLs
Thirteen studies assessed the impact of TCPs on functional status [29, 30, 38, 42, 44-46, 51, 53, 55, 57, 58, 62], see Table 2. Functional status was primarily measured as performance of ADLs, with the Minimum Data Set ADL score (n = 8) being the most commonly used tool [42, 44-46, 55, 57, 58, 62]. Performance of ADLs was measured at multiple time points, with assessment most often at admission, discharge, and at 1-month. For functional status outcomes, only those reported from admission to discharge, or first time point are reported below, but follow up time points are found in Table 2. Improvement in functional status was reported in eight studies for older adults with CI [29,30,42,44,51,53,55,62]; however, overall, a greater percentage of participants without CI experienced functional improvement compared to those with CI. Improvement in performance of ADLs was reported in 28.4% [51] to 53.3% [55] with dementia, 46.2% with dementia and delirium [55], 51.9% with delirium [55], and 57.4% of participants with CI [62], compared with 30.6% [51] to 68.8% [62] of participants without CI. Moreover, gains in functional status scores were smaller for older adults with CI [30], dementia [29,53], and delirium [42], compared to those without CI. Furthermore, poor functional status post TCP was reported in four studies [38,45,57,58] and having CI was associated with significantly less improvement in one study [46].

Patient outcomes
Other patient outcomes were assessed in six studies [29,45,54,55,62,63], with mortality (n = 5) [30,54,55,62,63] being the most common (Additional file 3: Table S4). Three-month mortality ranged from 8.2% [54] to 33.7% [55] for participants with CI, compared to a range from 5.7% [27] to 12.8% [55] for those with no CI. Six-month mortality rate for older adults with delirium was 25.0% and 18.3% for those with subsyndromal delirium, compared to only 5.7% for those without delirium [63]. Furthermore, 1-year mortality for older adults with CI ranged from 38.8% [62] to 49.1% [55], compared to a range from 24.4% [55] to 26.2% for those without CI [62]. There were improvements between admission and at four weeks in the Mini-Mental State Examination, Geriatric Depression Scale, and Mini Nutritional Assessment scores in older adults with dementia, however, those without CI had greater improvements in the Geriatric Depression Scale than those with CI [29].

Health services outcomes
Health services outcomes were measured in five studies [29,30,54,55,62] (Additional file 3. Table S5), with mean length of TCP stay being most commonly evaluated [29,30,62]. Mean TCP length of stay for older adults with CI ranged from 28.6 days [29] to 37.2 days [30], compared to a range from 27.5 days [62] to 31.7 days for older adults without CI [30]. Between 13.4 and 16.4% of participants with dementia were re-hospitalized within 30 days [54], while 17.2% of older adults with delirium and dementia, 26.4% of older adults with delirium but no dementia [55], and between 13.8% and 16.8% of patients without dementia [54] were re-hospitalized. Between 24.6% [54] and 38.7% [30] of participants with dementia and 34.3% of older adults with CI [30] were re-hospitalized within 90 days [54], compared to between 22.3% [30] and 27.2% [54] of older adults with no CI.

Measurements
Loomer 2019 [44] Improvement in self-care and mobility (difference between admission and discharge scores for self-care and mobility came from Section GG of the MDS 3.0 v1.14.1) Improvement in composite self-care and composite mobility scores was calculated by subtracting participants' admission score from their discharge score. The percentage of participants whose composite scores stayed the same or improved were calculated Expected self-care and mobility (a dichotomous variable to determine residents whose scores were the same or higher than expected At discharge: Percentage of residents whose observed self-care and mobility performance improved or stayed the same between admission and discharge: MDS ADL-Change in score between admission and discharge The ADL score ranges from 0 (no impairment) to 28 (total dependence). It is considered an improvement in function if their discharge ADL score was less than their admission score     MDS 3.0 scale ranges from 0 to 28, with higher scores indicating greater impairment. ADL change was calculated as the admission score minus the discharge score, and so positive scores indicate improvement and negative scores indicate decline discharged to long-term care. Moreover, participants with dementia in facility-based TCPs were less likely to be discharged to home (adjusted odds ratio (aOR) 0.53 [28] and aOR 0.4 [52]) compared to participants without dementia. Finally, participants with CI were less likely to be discharged home (odds ratio (OR) 0.46), more likely to be discharged to the nursing home or be deceased after two months (OR 2.95), and more likely to transfer to another TCP after two months (OR 1.96), compared to those without CI [60].
Beyond the percentage of participants discharged home, four studies [47,48,51,54] specified the percentage of participants who had a successful community discharge, that is, they were discharged from TCP to the community within 90-100 days of TCP admission [47,51,54] and, within 30 days of discharge from TCP, they were not hospitalized [47,48,51,54], were not admitted to a nursing home [47,48,51], and did not die [47,48,51]. Between 24.6% [47] and 68.0% [48] of older adults with any form of CI, compared to 58.1% [47] and 62.9% [54] of older adults with no CI is Successful Discharge (defined as being discharged alive from a skilled nursing facility (SNF) to the community within 90 days of SNF admission without subsequent inpatient healthcare utilization for 30 continuous days; ** = Outcome is community discharge rate (metric used on Nursing Home Compare is the rate of beneficiaries who are able to leave the SNF by 100 days after hospital discharge and remain in the community (i.e., alive and outside the hospital and nursing home) for at least 30 days after SNF discharge; *** = Outcome is Successful Discharge (discharge to community within 100 days of a nursing home admission, defined as: Discharge to the community within 100 days (allowing for interim discharges from Community Living Center to hospital if the Minimum Data Set noted that return was anticipated, observation stays, and emergency room use), and no unplanned admissions to a hospital, a nursing home or observation stay, and not dying within 30 days following discharge; **** = Outcome is Successful Discharge to the community (During the 30 subsequent days the veteran did not die, was not readmitted to a hospital for an unplanned inpatient stay, and was not admitted to a nursing home): No * indicates that it is the percentage of older adults with CI discharged home, and does not specify that it needs to have been a "successful" discharge as defined in the 4 studies with a * had a successful community discharge. Furthermore, only one study [51] looked at both successful discharge (57.4% of older adults with dementia) and functional decline. Improvement in functional status was found in 28.4% of participants with dementia, while 45.5% had no improvement, and 26.1% had missing data [51].

Quality of studies
Although the majority of the studies were rated moderate to good quality, the heterogeneity of the outcome measures, measurement times, and patient populations as well as study designs in the included studies, in addition to the lack of RCTs in this review precluded meaningful meta-analysis. Furthermore, as all the studies included in this review were observational, there is a risk of bias due to lack of randomization. Therefore, only determination of associations was possible.

Discussion
The results of this systematic review reveal that TCPs help improve outcomes for older adults with and without CI [29,30,42,44,51,[53][54][55]62]. However, a greater percentage of participants without CI had improvements in ADLs and better patient and health services outcomes compared to those with CI. In terms of discharge destination, older adults with CI were more often discharged home than to long-term care, however, a greater percentage of participants without CI were discharged home [38,45,47,54,55,60]. There was also a wide range in the percentage of older adults with CI who had a successful discharge home [47,48,51,54].
A meta-analysis by Hang et al. [19] on communitydwelling older adults in TCPs found a significant improvement in modified Barthel Index functional score between admission and discharge (pooled mean difference of 17.65 points (95% confidence interval [5.68, 29.62], p = 0.004). However, Hang et al. 's meta-analysis did not focus on community-dwelling older adults with CI; instead, they focused on community-dwelling older adults in general. In this review, community-dwelling older adults with CI in TCPs also had an improvement in ADLs which was reported in eight of 12 studies. However, the study by Miu, Chan, & Kok [30] used the modified Barthel Index and found a smaller increase in functional score for those with dementia than that reported in Hang et al. [19]. Similarly, overall, functional improvement found in this present review was smaller for older adults with CI than for those without CI.
Although participants with CI had less functional improvement in TCPs than those without CI, it is likely that having older adults with CI who remain in hospitals once their acute medical condition is treated is not ideal. A previous review by Hartley et al. and an article by Pedone and colleagues demonstrated that having CI on hospital admission is a risk factor for functional decline [7,65]. Therefore, the improvements in functional status in TCPs indicate that these settings may be a better option for older adults with CI, rather than remaining in acute care where there is the risk of functional decline.
The meta-analysis by Hang and colleagues found that 80% of older adult participants in TCPs were discharged home [19]; however, this is a stark difference from the 25.9-68% of older adults with CI discharged home in the current review. Prior research on hospitalized older adults who have CI found that living alone and having responsive behaviours (e.g., verbal or physical behaviours related to care provision) at admission were negatively associated with discharge home [66]. Therefore, behavioural and psychological symptoms may influence discharge outcomes [66]. Thus, the lower percentage of participants with dementia being discharged home from TCPs may be due a variety of factors; future research to determine the facilitators and barriers to being discharged home is needed. In terms of discharge to longterm care, a review by Fogg and colleagues found that between 8.3-22.4% of hospitalized patients with CI (mild CI, CI, dementia) compared to 3.5-19.4% with no CI (p = 0.001), transitioned to nursing homes post TCP [9], slightly less than what was found in the present review (4.1-40%). Moreover, these reviews highlight the need for specialized interventions to increase the percentage of older adults with CI who can be discharged to their home.
Furthermore, given the role of TCPs in improving safety of transitions, there is a need to consider the difference between promoting increased discharge home and promoting successful discharge home. Discharged home means that the older adults are not discharged to a different facility such as long-term care. Successful discharge was defined slightly differently by each of the four studies; it means that, within 30 days of discharge to home, the older adult avoids re-hospitalization [47,48,51,54], admission to nursing home [47,48,51], and death [47,48,51]. Moreover, adverse events such as falls [67], functional decline [68], and medication-related adverse events [69] can all contribute to re-hospitalization risk. Given the percentage of older adults with CI who were re-hospitalized post TCP [48,51,54,55] as well as the wide range for the percentage of older adults with CI who had a successful discharge home [47,48,51,54], there is a need for interventions to promote safe, successful transitions to the home that reduces the risk of adverse events. Indeed, Toles and colleagues' study involving persons with dementia, their care partners, TCP staff, and home health nurses found that transitions from TCPs to home involve several important and unique care needs [70]. These included care planning specific to the needs of persons with dementia; the need to prepare care partners to manage dementia symptoms at home; difficulty connecting care partners and older adults with dementia to community supports; and the need for support for care partners to address their own needs [70]. Other considerations to reduce adverse events that can result in rehospitalization include medication management [69], addressing information needs of care partners, such as providing instructions on how to transfer the older adults in and out of a wheelchair, and scheduled post-TCP medical follow-up appointments [71].
This present review also demonstrates that various health care professionals are involved in the different TCP models of care. One model which has resulted in positive functional status and patient outcomes included an interprofessional team that focused on a reablement approach [29]. A reablement approach in older adults with dementia involves maintaining function for as long as possible, regaining lost function when it is possible to do so, and adapting when lost function cannot be regained [72]. In Lee et al. 's prospective cohort study, a TCP with a physical reablement program consisting of a comprehensive geriatric assessment, ADL training, exercises, and care plans with functional goals resulted in improvements in all patient outcome measures, including functional status, instrumental ADLs, and cognitive function for older adults with dementia [29]. However, discharge destination was not an outcome assessed in this study. The reablement approach could be adopted by TCPs and tested for the impact on both functional status and discharge destination in future studies. This model could also be compared and evaluated with other models in order to determine best practices for this population.

Implications for practice, policy, and future research
This review provided supportive evidence regarding the impact of TCPs on improvements in ADLs, patient and health services outcomes, and the greater percentage of discharges home than to long-term care for older adults with CI. However, practitioners and policymakers should take into consideration the level of evidence from this review, given the lack of RCTs and quasi-experimental studies.

Practice
In practice, health care teams can consider TCPs as possible discharge destinations for older adults with CI who are not yet ready to be discharged home. Given that participants with CI gained smaller improvements in ADLs, it is critical to identify patients with any form of CI, so that additional or specialized resources, such as recreational therapy, behavioural supports, or Geriatric Psychiatry, can be allocated to help improve their outcomes. Moreover, in order to improve the safety of transitions, TCPs should consider including informational support to care partners on dementia care, connecting care partners and older adults with CI with community resources, and providing support for the needs of care partners.

Policy
Given the findings of improved ADLs in older adults, TCPs may be better settings than acute care for this population and as such should be transferred to these settings as soon as they are medically stable. Thus, policymakers involved in the creation or modification of future TCPs should ensure timely access to TCPs for persons with CI. Policymakers should also consider the rate of successful discharges for older adults with CI as a quality measure for TCPs.

Research
Although this review showed that there were improvements in ADLs for older adults with CI associated with TCPs, causality cannot be implied due to the lack of RCT evidence.
Thus, there is a need for RCTs to be conducted to compare TCPs for older adults with CI with usual care, and to assess whether improvements in functional status translate into an increase in the percentage of older adults with CI who are discharged home. Second, there is a need to develop and test reablement interventions in TCPs that focus on maintaining and improving functional status in older adults with CI; a reablement program may be one solution [29]. Third, further studies are required to assess and measure other health outcomes such as complex functioning required to perform IADLs, in addition to the performance of ADLs, since living in the community requires more than just physical capabilities [73]. Fourth, future studies should utilize standardized functional status measurement tools among older adults with CI in TCPs in order to facilitate meta-analyses. Fifth, studies should include both discharge destination and rate of successful discharge to community as outcome measures, to demonstrate effectiveness of TCPs. Finally, there is a need for quantitative and qualitative studies to determine the factors, such as social supports and resources, barriers, and facilitators, that can have an impact on discharge destination for this population, and for intervention studies to address the barriers.