Our search was restricted to published intervention studies. We selected a study for inclusion if it: (a) was published in English; (b) included a control or comparison group; (c) published outcome data from the intervention under study; (d) reported hospitalization as one of the outcomes; (e) included community-dwelling older adults; and (f) enrolled participants with dementia. We excluded studies that involved people under the age of 50 (to avoid a priori omission of any studies that enrolled persons with early-onset dementia), older adults living in nursing homes, interventions solely targeting caregiver outcomes, purely lifestyle interventions, and pharmaceutical trials.

A professional research librarian obtained articles from database searches (from 1990 through December 2013) of PubMed, CINAHL, Cochrane, Embase, PsycINFO, Wholis, Proquest, EBSCO, and WOS (see Fig. 1 for PRISMA-formatted chart with search strategy and details). Additional sources included reference lists from published studies and representatives of organizations who have published work in the area of dementia (eg, Institute of Medicine; University of California at Los Angeles). Four hundred thirty-four references were identified in this initial search. Then, through an iterative process, all authors systematically identified articles that met the inclusion criteria. See Figure 1 for details of the steps included in the systematic review, including MESH terms. This process resulted in 10 articles for analysis.17–26

A list of data to be extracted was compiled by the group, and 2 authors (K.J.D. and E.A.P.) extracted data independently and thereafter resolved any discrepancies via discussion. Data were extracted on the study design, participant characteristics, features of the interventions, and outcomes of the interventions. Intervention descriptions were categorized by the format in which they were delivered, the content and intensity (number of contacts with intervention participants and intervention period duration) of the intervention, and the target (ie, patient, physician, or caregiver) of the intervention content. When provided in the article, quantitative outcome data were also extracted, including statistical test results. All data extracted are provided in Tables 1 to 3.

As the focus of this review was to identify promising interventions for the prevention of hospitalizations among community-dwelling older adults, data analysis concentrated on the features of the interventions. Specifically, the authors examined characteristics of the interventions that might influence findings related to hospitalization. In the Results section, tables provide intervention details including the mechanisms that might underlie any observed intervention effects.

As shown in Table 1, of the 10 included studies, 6 were conducted in the United States. Study participants were recruited from a range of settings, including health care organizations, assisted living facilities, and community-based agencies; 1 study24 used newspaper announcements and a health insurance registry of Alzheimer disease drug users. Half of the studies randomized the participant with dementia along with their caregiver. Sample sizes ranged from 43 to 408. The mean age of dementia participants across studies was 70.4 years. Dementia severity ranged from mild to severe; 4 studies focused on moderate-severity dementia, and in 1 study, dementia severity was not provided. The dementia type where specified was most often Alzheimer; in 5 studies, the dementia type was not characterized. Table 2 summarizes features of interventions in the included studies.

Although all included studies examined hospitalization as an outcome, hospitalization was a primary outcome in just half (Table 3). Most studies did not describe if health care utilization was tracked in real-time, such that that information would be known to the study interventionist and thereby permit him/her to take preemptive actions, either in the time period immediately leading up to admission (to prevent the admission) or at/shortly after the time of discharge (to prevent readmission). A significant reduction in hospital admissions was not found in any of the included studies, although 1 study did observe a reduction in hospital days.24

In interpreting this main finding of our review, methodology as well as intervention content and delivery merit consideration. Of the 2 studies with nonsignificant effects on hospital admission, one did not use a randomized design (which tends to overestimate intervention effects),19 and the other was likely underpowered to detect a significant between-group difference in hospitalization.18 The study that measured a reduction in hospital days did not report hospitalization frequency by study group.24 In the studies that used the most rigorous (randomized, controlled trial) design, an effect on risk of hospital admission was either not observed or another measure of hospitalizations (hospital days) was reported.18,20,23,24,26,27

Apart from methodologic limitations of included studies, the majority of interventions did not explicitly target either control of common chronic conditions (eg, congestive heart failure) or prevention of unintentional injuries (eg, falls, burns), both of which are common reasons for hospitalization in this age group and in persons with dementia.3,4,28 Other common reasons for hospitalization include acute infections (eg, urinary tract infection, pneumonia) and dehydration4,29; none of the interventions appeared to address these issues via caregiver education/training or other methods. Overall, a medical focus was lacking in most interventions tested.

Recommendations for future work on this topic are several. First, in light of the risks of hospitalization for persons with dementia, greater emphasis on this outcome in intervention studies is warranted. The advent of a Healthy People 2020 objective13 specific to this issue—namely to “reduce the proportion of preventable hospitalizations in persons with diagnosed Alzheimer disease and other dementias”—may increase the focus on this as an outcome in future intervention research with this population.

Second, future studies should design and test interventions aimed at preventing avoidable acute health care utilization. In other words, interventions should have an explicit focus on keeping persons with dementia out of the hospital, whenever possible. In our effort to be comprehensive with this review, we chose not to exclude studies where the conceptual link between the focus of the intervention (eg, self-care or coping strategies for caregivers) and our outcome of interest was not obvious/apparent. In the future, studies could test directed approaches that might include: counseling family caregivers about the risks of hospitalization, the most common reasons for hospitalization, and clinical signs of acute deterioration; educating caregivers about chronic condition control and simplifying medication regimens to optimize adherence; providing 24-hour access to nurse telephone care consultation for advice on concerns related to acute changes in the person with dementia; completion of advance directives with explicit statement of preferences regarding hospitalization; and notification of study interventionists if a patient with dementia presents to an emergency department for care. Although attention has been paid to potentially preventable hospitalizations among persons with dementia,2–4,13 research is needed on all-cause hospitalizations. A wide range of outcomes related to hospitalization for persons with dementia, including complications and adverse events, is also relevant. And, for the person admitted to hospital who has dementia, hospitalization due to any cause is just as relevant an outcome, as the risk of an adverse consequence of hospitalization is presumably no different regardless of the reason for hospitalization.

Third, based on observational studies examining reasons for hospitalization of persons with dementia,3,4,28 future intervention studies should test approaches to managing chronic conditions and preventing injuries in persons living in the community with dementia. Current work in the area of dementia and multiple chronic conditions may help shape this understanding.30,31 Effective interventions for the prevention of falls in community-dwelling elders with dementia are lacking; this is another worthy—albeit challenging—avenue of investigation.

Limitations of our review should be noted. The review includes only those articles written in English and published studies. However, the review did benefit from using multiple sources to search for eligible studies, including a professional research librarian search of online databases and interviews with professionals with expertise in dementia and healthcare utilization. Data for this review were abstracted from descriptions available in the article. Thus, the characterizations of the intervention and outcomes were not verified through potential secondary published sources that may have described the study design or interventions and no input was sought from the authors. This review intentionally included studies that examined any measure of hospitalization as an outcome. As a consequence, the heterogeneity of outcomes and assessment methods precluded performing a formal meta-analysis. Despite these limitations, this review provides an important summary of the state of the evidence regarding interventions and their effect on hospitalizations of persons living with dementia.

We support the prevailing belief among many in the field that dementia is not “just another diagnosis” on a patient’s problem list. Dementia impairs an individual’s ability to manage his/her other chronic conditions, to recognize and articulate the onset of new physical or emotional symptoms, and to seek assistance/care in the face of alterations in one’s overall condition. Dementia implies that self-management support must be available, and due to its progressive nature, changing care needs over time.32,33 Thus, a diagnosis of dementia should shape the focus and the approach to management of all other health issues in the context of routine outpatient care. The next generation of intervention research must identify the elements of high-quality outpatient care for persons with established dementia.