Effectiveness of Acute Geriatric Unit Care Using Acute Care for Elders Components: A Systematic Review and Meta-Analysis

Objectives To compare the effectiveness of acute geriatric unit care, based on all or part of the Acute Care for Elders (ACE) model and introduced in the acute phase of illness or injury, with that of usual care. Design Systematic review and meta-analysis of 13 randomized controlled and quasi-experimental trials with parallel comparison groups retrieved from multiple sources. Setting Acute care geriatric and nongeriatric hospital units. Participants Acutely ill or injured adults (N = 6,839) with an average age of 81. Interventions Acute geriatric unit care characterized by one or more ACE components: patient-centered care, frequent medical review, early rehabilitation, early discharge planning, prepared environment. Measurements Falls, pressure ulcers, delirium, functional decline at discharge from baseline 2-week prehospital and hospital admission statuses, length of hospital stay, discharge destination (home or nursing home), mortality, costs, and hospital readmissions. Results Acute geriatric unit care was associated with fewer falls (risk ratio (RR) = 0.51, 95% confidence interval (CI) = 0.29–0.88), less delirium (RR = 0.73, 95% CI = 0.61–0.88), less functional decline at discharge from baseline 2-week prehospital admission status (RR = 0.87, 95% CI = 0.78–0.97), shorter length of hospital stay (weighted mean difference (WMD) = −0.61, 95% CI = −1.16 to −0.05), fewer discharges to a nursing home (RR = 0.82, 95% CI = 0.68–0.99), lower costs (WMD = −$245.80, 95% CI = −$446.23 to −$45.38), and more discharges to home (RR = 1.05, 95% CI = 1.01–1.10). A nonsignificant trend toward fewer pressure ulcers was observed. No differences were found in functional decline between baseline hospital admission status and discharge, mortality, or hospital readmissions. Conclusion Acute geriatric unit care, based on all or part of the ACE model and introduced during the acute phase of older adults' illness or injury, improves patient- and system-level outcomes.

A dults aged 65 and older constitute the "core business" of hospitals. 1 Although they represent 13% of the population in the United States 2 and 14% of the population in Canada, 3 older adults account for 43% of inpatient hospital days in the United States 4 and 40% in Canada. 5 This trend is likely to continue given population aging. 3 During hospitalization for an acute event such as illness or injury, older adults are at risk of experiencing functional decline and iatrogenic complications, including falls, pressure ulcers, and delirium, which further contribute to functional decline. 6 Hospital-acquired functional decline is associated with greater hospital expenditures, institutionalization, and mortality in older adults 7 even after controlling for comorbidity and illness severity. 8 Therefore, early intervention (before an acute episode is resolved) is critical because of the short length of time during which older persons can recover functional losses, resume their former lives, and avoid institutionalization. 9 Dedicated geriatric units, based on a prehabilitation 10 and function-focused 11 model of care called Acute Care for Elders (ACE), have been designed specifically to prevent functional decline and related complications in older adults admitted to the hospital for an acute event. 12,13 In response to an increasingly older and complex hospital population, some service providers have adopted the ACE model on hospital units where older adults are admitted. 13 However, the overall effect of acute geriatric unit care, based on all or part of the ACE model and introduced during the acute phase of illness or injury, is unclear and unquantified.
Two systematic reviews of acute geriatric unit care based on the ACE model have been conducted, 14,15 but the authors did not present results of meta-analyses, supporting the need for this current review. Three prior reviews combined data from studies conducted with individuals in the acute and subacute illness phases; 14,16,17 the results have limited validity for individuals in the acute phase of an illness or injury. One meta-analysis 18 imputed means for missing standard deviations for cost and lengthof-stay outcomes in almost 30% of included studies, 18 which may have resulted in an underestimation of the overall effect. Last, no meta-analysis of acute geriatric unit care included iatrogenic complications, which are critical indicators of quality hospital care. 19 The purpose of this study was to determine the effectiveness of acute geriatric unit care, based on all or part of the ACE model components and introduced in the acute phase of illness or injury, in reducing iatrogenic complications, functional decline, length of hospital stay, poor discharge destination outcomes, mortality, costs, and hospital readmissions in older adults.

METHODS
A systematic review was performed that compared acute geriatric unit care, in which all or part of the ACE model components were introduced in the acute phase of illness or injury, with usual care using the Cochrane Collaboration Protocol. 20

Eligibility Criteria
Eligible studies included published and unpublished randomized controlled and quasi-experimental trials with parallel controls that compared acute geriatric unit care with usual care for adults aged 65 and older in the acute illness or injury phase. 17 Acute geriatric unit care included at least one of the five ACE model components or principles: 13,21,22 patient-centered care, defined as care activities (assessments and protocols) to prevent declines in activities of daily living (ADLs), mobility, continence, nutrition, skin integrity, mood, sleep, and cognition; frequent medical review, defined as activities to minimize the adverse effects of treatments on older adults' functioning; early rehabilitation, defined as the participation of physical or occupational therapists in daily team meetings for the purposes of initiating rehabilitation or standard provision of physical or occupational therapy; early discharge planning, defined as activities to facilitate return to the community; and prepared environment, defined as environmental modifications to facilitate physical and cognitive functioning. Usual care was defined as any care not provided on an acute geriatric unit.
Eligible studies included at least one primary (iatrogenic complications or functional decline) or secondary (length of hospital stay, discharge destination, mortality, costs, or hospital readmissions) outcome. Iatrogenic complications included falls (defined as the number of individuals who experienced ! 1 falls), pressure ulcers (defined as the number of individuals who experienced skin breakdown), or delirium (defined as the number of individuals diagnosed with ! 1 delirium episodes) during hospitalization. Functional decline was defined as loss of independence at discharge in at least one of five basic ADLs: transfers, toileting, dressing, eating, or bathing, 10 as measured as the Barthel Index or the Katz ADL scale, 2 weeks before hospital admission or upon hospital admission. Length of hospital stay was defined as the total number of days in the hospital or as the time between study admission and discharge if total number of days in the hospital was not provided. Discharge destination included discharge to home (defined as own home or with family) or nursing home (defined as nursing home, sheltered living, or hostel). Mortality refers to number of deaths during hospitalization. Costs were defined as total hospital costs associated with care for the duration of hospital stay. Cost data were standardized to U.S. dollars for a common price year of 2000, the last study year with published cost data. Hospital readmissions refer to the number of individuals readmitted one or more times to an acute care hospital within 1 or 3 months after discharge from the study hospital.
Studies unavailable in English or French, involving individuals undergoing elective surgical procedures or receiving palliative care, including social admissions, or with historical control groups were ineligible.

Search Strategy and Study Selection
An information specialist conducted the literature search with input from team members with expertise in the clinical area to identify keywords reflective of the ACE model (Appendix S1 of the electronic supplementary material). Electronic databases searched were as follows: Evidence-Based Medicine Reviews consisting of the Cochrane Library, DARE, HTA, NHSEED and ACP; MEDLINE; EMBASE; CINAHL; Proquest Dissertations and Theses; PubMed; Web of Science; SciSearch; PEDro; Sigma Theta Tau International's registry of nursing research; Joanna Briggs Institute; CRISP; and OT Seeker. Internet search engines included Google, Yahoo, Scirus, Healia, and HON. Handsearching was conducted in the Gerontologist, Age and Ageing, JAMA, and bibliographies of all included articles and previous systematic reviews.
Two reviewers independently screened abstracts of the retrieved citations for potential inclusion. Disagreements about eligibility were resolved by consensus between two reviewers. Where consensus could not be reached, a third team member independently reviewed the abstract and determined final inclusion. When necessary, the complete article was reviewed to determine eligibility.

Data Extraction and Risk of Bias Assessment
Two reviewers independently extracted relevant data from each included article and entered the data into a standardized data extraction form. Information categories included study design, participants, ACE components, healthcare providers, occasions of measurement, and outcomes. Two reviewers independently assessed each study's risk of bias using six defined domains: sequence generation; allocation concealment; blinding of participants, personnel, and outcome assessors; completeness of outcome data; selective reporting; and other sources of bias. 20 Study authors were contacted if additional data were required. Disagreements on data extraction and risk of bias assessments were resolved by consensus with assistance of a third team member.

Data Analysis
When sufficient data were available and studies were comparable in terms of outcomes, meta-analyses were performed using REVIEW MANAGER software. 20 When data were neither retrievable from study authors nor derivable from available data, values were not assumed for the purposes of meta-analyses. In studies in which per-protocol and intention-to-treat data were reported, the latter were analyzed. Continuous and dichotomous outcomes were analyzed using a random-effects model to calculate weighted mean differences (WMDs) and risk ratios (RRs), respectively, with 95% confidence intervals (CIs). P < .05 was considered statistically significant for an overall effect. P < .10 was considered statistically significant for heterogeneity. 23 Degree of heterogeneity is reported according to the I 2 statistic, which refers to the degree of variation between studies. 20 Because of the potential for clinical heterogeneity of study populations and ACE components and the associated risk of a false-negative I 2 statistic, 24 the CIs of individual studies contained in the forest plots were also examined. 20 In situations in which heterogeneity was statistically significant or was not statistically significant but there was minimal overlap of the CIs, sensitivity analyses were performed whereby studies were systematically removed from meta-analyses to determine robustness of findings. Decisions for removing studies were based on their potential sources of variability; studies conducted on surgical units were removed first, followed by studies conducted on medical-surgical units, and then studies that did not implement all five ACE components, beginning with studies that implemented the fewest components.

Risk of Bias
Selection bias resulting from inadequate sequence generation was low in seven of the 13 studies. 21,22,25,26,30,31,41 Three studies that used randomization provided insufficient information to draw conclusions in this domain 28,35 or were considered not to have been properly randomized. 29 Three studies were determined to have high risk of selection bias because randomization was not performed. [38][39][40] Risk of selection bias resulting from inadequate allocation concealment was low in six of the 13 studies. 21,22,25,26,31,41 Allocation was not concealed in one study, resulting in a high risk assessment. 30 In all other studies, risk of bias was unclear because allocation information was not provided. 28,29,35,[38][39][40] Risk of performance bias related to double blinding (participants and personnel) was unclear because seven studies did not provide this information. 22  low risk of performance bias, 21,26,31 whereas three studies were not double blinded and were considered to have high risk of performance bias. 25,30,41 Risk of detection bias related to blinding of outcome assessors was unclear because 11 studies did not provide this information. 21,22,25,26,[28][29][30]35,[38][39][40] One study was considered high risk because outcomes assessors were blinded to only one of several outcomes. 41 Only one study reported that outcome assessors were blinded and was considered to have low risk of bias. 31 Risk of attrition bias related to completeness of outcome data was low in six studies 22,[28][29][30][31]41 and unclear in one. 38 Six studies were considered to have high risk of bias because of postrandomization exclusions 25,39 or attrition. 21,26,35,40 Risk of reporting bias due to selective reporting was low 22,28,31,35,41 or unclear. 25,29,30,[38][39][40] One study 26 was considered to have high risk of reporting bias because its length-of-stay and cost data were missing. None of the 13 studies appeared to be at risk of other sources of bias that were not addressed in prior domains.

Effectiveness of Acute Geriatric Unit Care
Eleven meta-analyses were performed. Unpublished data were obtained from study authors to perform metaanalyses on functional decline between baseline 2-week prehospital admission status and discharge, 39 length of hospital stay, 27,29,39,41 mortality, 35 and costs. 21,41 Four sensitivity analyses were conducted for functional decline between baseline hospital admission status and discharge, length of hospital stay, discharge to nursing home, and costs.
Delirium was reported in three studies. 25,31,39 Metaanalysis of these three studies showed that acute geriatric unit care was associated with significantly less occurrence of delirium than usual care in acutely ill or injured older adults (RR = 0.73, 95% CI = 0.61-0.88; P = .001).

Functional Decline
Functional decline between baseline 2-week prehospital admission status and discharge was reported in six studies. 21,22,31,[39][40][41] Meta-analysis of these six studies indicated that individuals receiving acute geriatric unit care were 13% significantly less likely to experience functional decline between their baseline 2-week prehospital admission status and discharge than those receiving usual care (RR = 0.87, 95% CI = 0.78-0.97; P = .01).
Functional decline between baseline hospital admission status and discharge was reported in four studies. 21,22,40,41 Meta-analysis of these four studies showed that, compared to usual care, individuals receiving acute geriatric unit care experienced no significant difference in risk of functional decline between baseline hospital admission status and discharge (RR = 0.83, 95% CI = 0.64-1.08; P = .16). Significant statistical heterogeneity was observed for this comparison. With removal of one outlier study 40 during sensitivity analysis, statistical heterogeneity was resolved, although the effect remained nonsignificant ( Table 1).
Six studies reported whether participants were discharged to a nursing home. 21,25,26,[28][29][30] Meta-analysis of these six studies identified no significant effect, although significant statistical heterogeneity was observed between studies for this comparison (Table 1). With the removal of three outlier studies 26,28,29 that resolved the heterogeneity, a meta-analysis identified that individuals receiving acute geriatric unit care were significantly less likely than those receiving usual care to be discharged to a nursing home (RR = 0.82, 95% CI = 0.68-0.99; P = .04).

Hospital Readmissions
Two studies reported on hospital readmissions within 1 month of discharge, 21,31 and three reported on hospital readmissions within 3 months of discharge. 22,25,41 Metaanalysis of these five studies identified no significant difference in hospital readmissions within 1 or 3 months of discharge between individuals receiving acute geriatric unit care and those of individuals receiving usual care (Table 1).

Post Hoc Analyses
Post hoc subgroup meta-analyses were performed in the three studies that examined the effect of the full ACE model on the study outcomes. Results remained significant (length of hospital stay) 21,22,41 or nonsignificant (functional decline between baseline hospital admission status and discharge, mortality, and hospital readmissions) 21,22,41 or were inconclusive because of heterogeneity (discharge home) 22,41 or no longer significant (functional decline between baseline 2-week prehospital admission status and discharge 21,22,41 and costs 21,27,41 ). The last may have been because of low power resulting in a Type I error.

DISCUSSION
This is the first combined systematic review and metaanalysis of acute geriatric unit care based on all or part of the ACE model components and the first to examine iatrogenic complications and functional decline between baseline hospital admission status and discharge. Results from meta-analyses demonstrate that acute geriatric unit care including one or more ACE components and introduced during the acute illness or injury phase has significant beneficial effects over usual care in reducing falls, delirium, functional decline between baseline 2-week prehospital admission status and discharge, length of hospital stay, discharge to a nursing home, and costs and in increasing discharges to home. In addition, a nonsignificant trend of finding fewer pressure ulcers was observed. Given the demographic and health characteristics of the average  25 Barnes et al. 41 Collard et al. 26 Counsell et al. 21 Fretwell et al. 28 González-Montalvo et al. 29 Harris et al. 30 Landefeld et al. 22 Olofsson 31 Somme et al. 35 Vidan et al. 39 6,612 Asplund et al. 25 Barnes et al. 41 Counsell et al. 21 Covinsky et al. 27 Outlier removed Asplund et al. 25 Barnes et al. 41 Counsell et al. 21 Covinsky  25 Barnes et al. 41 Counsell et al. 21 Landefeld et al. 22  study participant, these findings are mainly applicable to octogenarians admitted through the emergency department with acute illnesses or injuries and other morbidities.

Implications for Practice and Policy
The findings have relevance for clinicians, hospital administrators, policy-makers, and funders. By implementing all or part of the ACE model components during older adults' acute illness or injury phase, clinicians may anticipate small to moderate beneficial effects on the outcomes found to be significant in the meta-analyses. Although further research is needed, clinicians may also anticipate fewer pressure ulcers. Patient-centered care, frequent medical review, early rehabilitation, and early discharge planning were provided in more than half the studies and may represent the optimal ACE components for positive outcome achievement. Interdisciplinary team work was also a unique characteristic of acute geriatric unit care and may be important for clinicians to consider in their practice. The findings are applicable to the care of acutely ill and injured older adults on medical, surgical, and medicalsurgical units, and they address concerns about limited applicability and benefit of ACE to nonmedical patient groups and units. 42 Older adults with acute injuries typically have comorbidities that precipitated the injury and complicate its management and therefore benefit from a functionfocused prehabilitation approach.
Hospital administrators may anticipate cost savings of approximately $246 per hospital stay in U.S. dollars standardized to 2000 and more than a half-day shorter hospital stay than with usual care. Older adults account for 50% of Canadian 43 and 45% of U.S. 44 hospital expenditures. With projected increases in age demographics in both countries, 3 this cost difference may represent a significant future source of financial saving to both healthcare systems. This finding addresses cost-ineffectiveness 12 and cost-prohibitiveness 45 barriers to adopting the ACE model.
By establishing ACE as the preferred model of care, policy-makers can play an influential role in its adoption and in the improvement of patient-and system-level outcomes. By changing reimbursement or charge rates and by establishing targets for cost and resource efficiency for older people's care, funders can create the external and substantive structural incentives needed to move ACE into the "mainstream of hospital care." 46

Comparison with Previous Research
The findings of the current study are similar to those of an earlier meta-analysis of older adults with medical disorders 18 that found that acute geriatric unit care, which may or may not have included the ACE components, had significant effects on preventing functional decline between baseline 2-week prehospital admission status and discharge, increasing discharges home, and reducing costs and nonsignificant effects on mortality and hospital readmissions. However, in contrast to the earlier meta-analysis, which found nonsignificant or inconclusive effects on length of hospital stay or discharges to a nursing home, 18 the current meta-analysis identified significant reductions in length of hospital stay and discharges to a nursing home after acute geriatric unit care in which ACE components were provided in varying degrees. These findings concur with a prior narrative analysis comparing ACE with usual care units. 15 This review included six new randomized 31,35,41 and quasi-experimental 29,39,40 trials, which resulted in larger sample sizes of many of the meta-analyses than in prior research. CIs were also more precise for most outcomes than were those of prior meta-analyses. 18

Strengths and Limitations of the Review
This review had little missing data because six study authors 21,27,29,35,39,41 provided unpublished data, minimizing publication bias. The review included a small number of studies with limited information regarding study methods, which restricted the ability to draw conclusions regarding level of bias in several domains. Although randomization was used in most studies, six 21,25,26,35,39,40 had postrandomization exclusions or did not report related information, which may have contributed to an overestimation of effect sizes. Sample sizes in the meta-analyses on iatrogenic complications were modest, which may have influenced the imprecision of the estimates.
Although this review included a diverse group of individuals admitted to medical, medical-surgical, or surgical units, heterogeneity was low in the majority of meta-analyses, supporting validity of the results. It was not possible to perform subgroup meta-analyses (medical vs surgical) because three studies did not report results separately for medical and surgical patients 26,28,38 and because of the potential for bias with small and uneven distribution of groups. 20

Implications for Future Research
This review highlights the limited number of studies examining the effectiveness of acute geriatric unit care based on all or part of the ACE model components on outcomes of importance to older adults and service providers, specifically iatrogenic complications, costs, and hospital readmissions. With increasing concerns about safe and fiscally responsible care that does not result in hospital readmissions, 47 future research should examine the effectiveness of acute geriatric unit care on these outcomes.
Future research should explore the effectiveness of the ACE components with surgical patients. As ACE continues to be adopted and tested in the care of older surgical patients, future researchers may conduct subgroup analyses to compare its effectiveness in medical patients with its effectiveness in surgical patients.
Most studies restricted entry to individuals admitted through the emergency department. Given the importance of community services, 48 future trials should include older adults admitted to the hospital through avenues other than the emergency department. Future trials should also provide more-detailed descriptions of the methods used to facilitate assessment of the risk of bias and interpretation of results.
A prior meta-analysis 18 examining the effectiveness of admission to acute geriatric units excluded studies that limited admission to acutely injured individuals but included