A systematic review and network meta-analysis comparing various non-pharmacological treatments for older people with mild cognitive impairment

Non-pharmacological therapy appeared to alleviate Mild Cognitive Impairment (MCI) symptoms and signs, according to systematic studies. This network meta-analysis aimed to assess the impact of non-pharmacological therapies on improving cognition in individuals with MCI and identified the most effective intervention. We reviewed six databases in search of potentially relevant studies of non-pharmacological therapies such as Physical exercise (PE), Multidisciplinary intervention (MI), Musical therapy (MT), Cognitive training (CT), Cognitive stimulation (CS), Cognitive rehabilitation (CR)，Art therapy (AT), general psychotherapy or interpersonal therapy (IPT), and Traditional Chinese Medicine (TCM) (such as acupuncture therapy, massage, auricular-plaster and other related systems) and others. Excluded the literature such as missing full text, missing search results, or no reporting specific values and combined with the inclusion criteria and exclusion criteria in this article, the literature ultimately included in the analysis addressed the following seven non-drug therapies PE, MI, MT, CT, CS, CR, AT. Mini-mental state evaluation paired meta-analyses were undertaken by taking weighted average mean differences with confidence intervals (CI) of 95%. The network meta-analysis was conducted to compare various therapies. A total of 39 randomized controlled trials, including two three-arm studies, with 3157 participants were included. PE was most likely to be the most effective intervention to slow down the cognitive ability of patients (SMD = 1.34, 95%CI: 0.80, 1.89). CS and CR had no significant effect on cognitive ability. The non-pharmacological therapy had the potential to greatly promote the cognitive ability of the adult population with MCI. PE had the best chance of being the best non-pharmacological therapy. Due to the limited sample size, substantial variability among different study designs, and the potential for bias, the results should be regarded with caution. Our findings should be confirmed by future multi-center randomized controlled, high-quality large-scale studies.