Effectiveness of interventions for preventing people with dementia exiting or getting lost

Background and Objectives: 
People with dementia are at risk of exiting premises unsupervised, eloping, or getting lost, potentially leading to harmful or distressing consequences. This review aimed to estimate the effectiveness of interventions for preventing people with dementia from exiting or getting lost.

Research Design and Methods: 
A systematic review of English sources was undertaken. Health care (EMBASE, BNI, Medline, PubMed, CINAHL, PsycINFO, AMED, HTA, CENTRAL) and gray literature (OpenGrey) databases were searched using prespecified search terms. Additional studies were identified by hand-searching bibliographies of relevant reviews and included studies. Wide inclusion criteria were set to capture a range of intervention types. Data extraction and risk of bias assessment were completed independently by two reviewers. Methods were preregistered on PROSPERO.

Results: 
Individual and overall risk of bias was too high for statistical meta-analyses. A narrative synthesis was therefore performed. Twenty-five studies with 814 participants were included, investigating a range of nonpharmacological interventions aiming to prevent exiting, facilitate retrieval, educate participants, or a combination of these. Seventeen (68%) of the included studies had critical risks of internal bias to outcomes, providing no useful evidence for the effectiveness of their respective interventions. The remaining 8 (32%) studies had serious risks of bias. Narrative synthesis of results yielded no overall robust evidence for the effectiveness of any interventions.

Discussion and Implications: 
No evidence was found to justify the recommendation of any interventions included in this review. Future studies should focus on high-quality, controlled study designs.