Objectives: We aimed to identify the best form of cognitive therapy among 3 main cognitive interventions of Alzheimer’s disease (AD) including cognitive training (CT), cognitive stimulation (CS), and cognitive rehabilitation (CR). Design: Systematic review and Bayesian network meta-analysis. Setting and Participants: An exhaustive literature search was conducted based on PubMed, Embase, the Cochrane Central Register of Controlled Trials, PsycINFO, the China National Knowledge Infrastructure database, the Chinese Biomedical Literature database, the Wan Fang database, and Web of Science and other database and randomized controlled trials were identified from their inception to May 1, 2018. Older adult participants diagnosed with AD were recruited. Measures: We conducted a Bayesian network meta-analysis (NMA) to rank the included treatments. Cognitive functions were measured based on the Mini-Mental State Examination (MMSE). A series of analyses and assessments, such as the Pairwise meta-analysis and the risk of bias, were performed concurrently. Results: Only 22 studies were included in our analysis based on a series of rigorous screenings, which comprised 1368 participants. No obvious heterogeneities were found in NMA (I 2 ¼ 32.7%, P ¼ .07) after the data were pooled. The mean difference (MD) of CT [MD ¼ 2.1, confidence interval [CI]: 1.0, 3.2), CS (MD ¼ 0.92, CI: 0.20, 2.0), and CR (MD ¼ 2.0, CI: 0.73, 3.4) showed that CT and CR could significantly improve cognitive function as measured by MMSE in the treatment group whereas the CS was less effective. CT had the highest probability among the 3 cognitive interventions [the surface under the cumulative ranking curve (SUCRA) ¼ 84.7%], followed by CR (SUCRA ¼ 50.0%) and CS (SUCRA ¼ 47.4%). Conclusions/Relevance: Our study indicated that the CT might be the best method for improving the cognitive function of AD patients. The findings from our study may be useful for policy makers and service commissioners when they make choices among different alternatives. Published by Elsevier Inc. on behalf of AMDA e The Society for Post-Acute and Long-Term Care Medicine. Alzheimer’s disease (AD) is a chronic, progressive, fatal, and prevalent neurodegenerative condition, and it is frequently preceded by a prodromal mild cognitive impairment stage. Clinically, AD is predominately characterized by memory failure, as well as word- finding difficulties and attention, visuospatial functioning, and exec￾utive functioning deficits. Impairments in self-care ability and social activity appear as the disease progresses.1e3 Currently, the number of individuals having dementia worldwide is approximately 47 million, and approximately 70% of these in￾dividuals are affected by AD.4 Deaths resulting from AD increased by 71% from 2000 to 2013 in the United States.5 The aging population, which continues to exhibit rapid growth, is expected to reach 74.7 million by the year 2030 and double in 2050 compared with 46.8 The authors declare no conflicts of interest. * Address correspondence to Yong Xu, MD or Chen-Wei Pan, MD, PhD, School of Public Health, Medical College of Soochow University, No. 199 Ren Ai Road, Suzhou, China 215123. E-mail addresses: childhealth@suda.edu.cn (Y. Xu), pcwonly@gmail.com (C.-W. Pan). https://doi.org/10.1016/j.jamda.2018.09.017 1525-8610/Published by Elsevier Inc. on behalf of AMDA e The Society for Post-Acute and Long-Term Care Medicine. JAMDA journal homepage: www.jamda.com JAMDA xxx (2018) 1e9 million in 2012.6e8 In addition, the global economic cost for dementia patients to date has been estimated to be $315 billion.9 These findings emphasize the need for practical approaches that could retard the progression of patients with AD and other dementias and reduce the families’ dementia-related burden or adverse effects that jeopardize their health and financial security. Regarding the lethal magnitude of AD, it is listed in fourth place after cancer, heart disease, and cerebrovascular disease, and it remains incurable. Any approach that could potentially slow down the progression of AD is worth disseminating. Clare and Woods10 classified cognitive interventions into 3 categories, including cognitive training (CT), cognitive stimulation (CS), and cognitive rehabilitation (CR). The specific intervention that is most effective remains debatable. CT, as a standardized task involving guided practice, should be implemented repeatedly, targeting a series of specific cognitive functions (such as memory, attention, language, or executive functions) to maintain or restore the impaired functions; however, in most cases, the mediated domains of functioning, such as basic and instrumental activities of daily living,11e13 social skills, and behavioral disturbances, may also be targeted.14 Moreover, in some cases, relevant training has been facilitated by medications or caregivers as a component of supportive intervention.2,15 CS (such as reality orientation therapy, learning therapy, and memory training) refers to a more specific method and is offered through group sessions, with an emphasis on social interaction.16 It focuses on engaging in a series of relevant activities to help individuals produce improvements in cognition and, in some cases, behavior.17,18 However, in some cases, attention, language, problem-solving, and other functions are included in the global method of CS therapy for its operation requirements.10 CR should be applied with an individualized approach focused on improving functioning in the daily context.19 Rehabilitation is implemented following a natural trajectory over time that changes according to the background of the patients (such as the extent of cognitive impairment and their social context).20 It is often divided into 2 basic major strategies for solving the memory diffi- culties; one strategy is to take full advantage of the remaining memory ability (such as identifying the important information in their life or performing real-life practical skills to help them enhance their memory),21,22 whereas the other strategy is to establish a method of compensation (such as developing memory aids and adjusting or controlling the environment around patients so that the demands on their memory are reduced).19,23 As an extension of traditional meta-analyses, network metaanalysis is instrumental when investigators are interested in summarizing 2 or more treatment results and the hierarchy of these results by estimating the comparative efficacy and summarizing and interpreting a broader picture of the evidence base.24e26 Therefore, we aimed to provide a relatively reliable conclusion via a comprehensive literature search and Bayesian network meta-analysis to identify the optimal cognitive intervention for patients with AD.