ObjectiveTo systematically review the efficacy and safety of non-pharmacological interventions for sleep disturbance in dementia, and to provide evidence for clinical practice.MethodsDatabases including CNKI, WanFang Data, VIP, PubMed, EMbase and The Cochrane Library were searched to collect randomized controlled trials (RCTs) on non-pharmacological interventions for sleep disturbance in dementia from inception to May 2020. Two reviewers independently screened literature, extracted data, and assessed risk of bias of included studies. Meta-analysis was then performed using RevMan 5.3 software.ResultsA total of 9 RCTs were included, involving 720 patients. Light therapy was the most commonly used treatment, followed by special activity and sleep education program. The results of meta-analysis showed that compared with the control intervention, light therapy could improve sleep efficiency (MD=2.21, 95%CI 1.09 to 3.33, P=0.0001) and the night-time sleep (MD=14.27, 95%CI 5.01 to 23.53, P=0.003) of patients with dementia in the community and nursing institutions, special activity could increase the night-time sleep (MD=29.74, 95%CI 20.44 to 39.04, P<0.00001), and sleep education program could also improve sleep efficiency (MD=6.19, 95%CI 5.22 to 7.16, P<0.00001) and night-time sleep (MD=33.95, 95%CI 25.40 to 42.50, P<0.00001). In addition, it was superior to obtain 120 or 60 minutes of light exposure than 30 minutes to improve the quality of sleep (RR=−2.62, 95%CI −3.56 to −1.68, P<0.001) and reduce daytime sleep (RR=−4.75, 95%CI −5.71 to −3.42, P<0.001). However, there was significant difference in incidence of adverse reactions between groups of 120 minutes and 30 minutes of light exposure (RR=2.57, 95%CI 1.44 to 4.58, P=0.001).ConclusionsThe current evidence shows that non-pharmacological intervention can improve sleep efficiency and night-time sleep in patients with dementia. Due to limited quantity and quality of the included studies, more high quality studies are required to verify above conclusions.