Comparative efficacy of multiple non-pharmacological interventions for behavioural and psychological symptoms of dementia: A network meta-analysis of randomised controlled trials.

Non-pharmacological interventions can improve the behavioural and psychological symptoms of dementia (BPSD). However, the optimal non-pharmacological treatments remain controversial. This study aimed to compare the efficacy of multiple non-pharmacological methods and identify the optimal therapy for BPSD. Potential randomised controlled trials (RCTs) were searched and selected from 15 databases and sources from the inception of the databases until 1 October 2022. Two independent authors implemented study screening, data extraction, and methodological quality assessment. Primary outcome was reduction of Neuropsychiatric Inventory (NPI). The secondary outcome were changes of Cornell Scale for Depression in Dementia (CSDD), the Cohen-Mansfield Agitation Inventory (CMAI), the Quality of Life in Alzheimer's Disease (QoL-AD), the Abilities of Daily Living scale, and the Apathy Evaluation Scale. Meta-analyses were performed using STATA v15.0 and ADDIS v1.16.8. The GRADE approaches were utilised to evaluate evidence quality. The present study included 43 RCTs with 4978 participants. The global methodological quality of the RCTs was moderate. Regarding NPI reduction, with moderate-certainty evidence, exercise plus treatment as usual (TAU) outperformed TAU (mean difference [MD]: -7.13; 95% confidence interval [CI]: -13.22, -0.76) and ranked as the optimal treatment. For reduction in CSDD, with low- to moderate-certainty evidence, massage plus TAU (MD: -15.26; 95% CI: -20.13, -10.52) and music plus TAU (MD: -2.40; 95% CI: -4.62, -0.12) were associated with greater reduction compared with TAU. For reduction in CMAI, with moderate-certainty evidence, aromatherapy plus massage (MD: -15.84; 95% CI: -29.76, -2.42) and massage plus music (MD: -13.12; 95% CI: -25.43, -0.76) were significantly more effective than TAU. For improvement in QoL-AD, with critically low- to low-certainty evidence, there were no statistical differences between any of non-pharmacological treatments and TAU. Due to the limited number of included studies, network meta-analysis was not performed for other outcomes. In conclusion, non-pharmacological treatments are effective for overall symptoms, depression, and agitation. Exercise plus treatment as usual may be an optimal non-pharmacological intervention for improving the overall BPSD. This may help to guide patients, doctors, and policymakers.