If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
This systematic review and meta-analysis evaluates the evidence from randomized clinical trials (RCTs) that designed brain gaming interventions to improve cognitive functions of older adults with cognitive impairments, including mild cognitive impairments and dementia.
Systematic review and meta-analysis.
Setting and Participants
Data sources—relevant randomized control trials (RCTs) were identified by a systematic search of databases including Medline, PubMed, PsycINFO, Embase, CINAHL, Web of Science, and Cochrane. RCTs were selected first based on title and abstract review and then on full-text review by independent reviewers using predefined eligibility criteria. Risk of bias (RoB) was assessed using the Cochrane RoB tool and funnel plots. The primary outcome variable was the composite score of global cognitive function.
A total of 909 participants with mild cognitive impairment or dementia from 16 RCTs were included in the systematic review. The study quality was modest, and the RoB assessment showed bias in blinding the participants and personnel. Funnel plots showed no evidence of publication bias. The meta-analysis of 14 RCTs revealed no superior effect of brain gaming compared to other interventions on global cognitive function (pooled standardized mean difference = 0.08, 95% confidence interval −0.24, 0.41, P = .61, I2 = 77%). Likewise, no superior effects were found on the cognitive domains of memory, executive function, visuospatial skills, and language.
Conclusion and Implications
The findings of this meta-analysis suggest that brain gaming compared with the control intervention does not show significant improvement in standardized tests of cognitive function. Because of considerable heterogeneity in sample size, gaming platform, cognitive status, study design, assessment tools, and training prescription, we cannot confidently refute the premise that brain gaming is an effective cognitive training approach for older adults with cognitive impairments. Recommendations for future research are included.
The socioeconomic burden of dementia on patients, families, and societies is already staggering. Alzheimer's disease (AD), a type of dementia, alone is projected to have cost Medicare and Medicaid $195 billion in 2019.
The potential benefits of developing effective CCT programs extend beyond people with dementia to include people with mild cognitive impairment (MCI). MCI is considered an at-risk state between healthy aging and dementia that is associated with subjective memory complaints in the absence of objective impairments in cognitive functions and daily-life activities.
In particular, brain gaming, a nonimmersive, user-friendly form of CCT, has gained tremendous popularity over the past decade. Although there are subjective components to define brain gaming (ie, features that enhance user engagement and motivation), the ability to adapt games based on level of difficulty and therefore provide a challenging or competitive experience to the user is one of the main criteria for inclusion as a brain gaming paradigm.
The cognitive tasks must be engineered to enhance the user's engagement and motivation with the game. This adaptability is a core feature that separates basic CCT programs from brain gaming. Electronic brain gaming software may run on desktop and laptop computers, tablets, or mobile devices (ie, iPad, tablet, phone), and gaming hardware that are accessible and frequently used by older adults.
Studies that investigate the effectiveness of brain gaming in older adults with and without cognitive impairments are vital to confirm or refute the claims that are made by the industry.
Continuing to understand and advance the utility of effective digital at-home cognitive therapies is also timely given the precautions needed to be taken during COVID-19. Telemedicine and remote rehabilitation are more common during the pandemic—for both patients who have COVID-19 and those who do not.
The potential benefits of engaging in safe, cognitively challenging, and motivating activities afforded by brain gaming may play a vital role in postcare of patients affected by COVID-19, but also in protecting against accelerated cognitive decline due to detriments on mental health.
no study has evaluated the effectiveness of nonimmersive brain gaming on cognitive functions in older adults with MCI and dementia-related AD. A scoping review conducted by our group found 13 randomized controlled trials (RCTs) investigating brain gaming in older adults with MCI and AD. The included studies demonstrate that nonimmersive electronic brain gaming is a safe, feasible, user-friendly, and potentially effective CCT intervention to maintain or improve cognitive functions among older adults with cognitive impairments.
Although some differences were found in intervention dose, type of brain gaming, and cognitive outcome measures among the included RCTs, we concluded that the studies were sufficiently homogeneous in research design to evaluate the effectiveness of brain gaming by performing a meta-analysis.
The primary objective of this systematic review and meta-analysis was to quantify the effects of brain gaming intervention on global cognitive function among older adults with MCI or dementia. Secondary objectives were to (1) assess the effect of brain gaming interventions on the cognitive domains such as memory, executive function, visuospatial skills, and language and (2) determine the effect of brain gaming interventions on secondary outcomes, such as activities of daily living (ADL), instrumental activities of daily living (IADL), depression, and quality of life (QoL). In our subgroup analysis, we hypothesized that brain gaming interventions would show larger effect sizes in adults with MCI as compared to dementia. In addition, we evaluated whether intervention dose and type of setting (home vs controlled settings) would impact the magnitude of the intervention effect.
PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) guidelines were followed for this review.
The protocol for this review was registered on PROSPERO (Central registration Depository: CRD42015023918).
The literature search was conducted using Ovid Medline, PubMed, PsycINFO, Embase, CINAHL, Web of Science, and the Cochrane Library to identify RCTs, written in English and published from inception to April 2021. The search strategy was based on four main concepts: (1) cognitive impairment or dementia; (2) outcomes (ie, cognition, ADL, QoL); (3) nonimmersive, electronic brain gaming interventions (eg, computer gaming, video gaming); and (4) study designs (controlled, randomized). Combination of multiple text words and medical subject headings (MeSH) were used to extract literature with the assistance of a medical librarian (search strategy, Supplementary Material 1). Manual search yielded additional articles from the reference list of review articles, authors’ own literature file, and Google Scholar. Authors of the studies that had insufficient information were contacted directly via e-mail. A comprehensive 2-level eligibility process was followed to identify studies for inclusion. Level 1 involved screening titles and abstracts to exclude articles that failed to meet our inclusion criteria and level 2 involved screening full texts of remaining studies. The data were independently extracted by 2 reviewers; that is, the first reviewer extracted the data and the second reviewer reviewed data for any discrepancies. Disagreements were resolved through study team discussion.
Studies included were only RCTs that examined the effect of cognitive interventions using nonimmersive, electronic brain gaming methods as defined by Sood et al
on cognition among older adults with MCI or dementia.
Brain gaming technology involves a wide range of computer technologies (hardware and software) such as desktop and laptop computers, mobile computers (ie, iPad, tablet, phone), and video game technologies.
Cognitive impairment status, that is, diagnosis of MCI or dementia, was determined by neuropsychological instruments used to define cognitive status (ie, Mini-Mental State Examination or Montreal Cognitive Assessment) or as reported by the authors (based on the criteria or cognitive evaluation) in the original study. Our primary focus was on cognition and domain-specific cognitive functions such as memory, executive functions, visuospatial functions, and language.
Studies were excluded if they were case reports, protocols, commentaries, dissertations, book chapters, letters, or conference abstracts. Decision was made to exclude brain gaming interventions that involved immersive or semi-immersive virtual reality games as these games required specialized equipment that are harder to access than nonimmersive brain games. Additionally, any non–computer-based games such as paper-and-pencil games or board games were excluded.
Risk of Bias and Quality Assessment
Two reviewers (S.L.K. and P.H.) independently completed risk of bias assessment using a standardized form and Cochrane Risk of Bias tool.
Disagreements were discussed among the reviewers and the research team until an agreement was reached. When the number of studies was at least 10, a comparison-adjusted funnel plot was drawn to assess for publication bias and small study effects. Quality assessment was performed using level-of-evidence hierarchy used in evidence-based clinical medicine as developed by the Center for Evidence-Based Medicine.
Global cognitive function was considered our primary outcome variable and was obtained from either composite scores or scores on the Mini-Mental State Examination. The composite score was calculated as the grand average mean and standard deviation (SD), derived from the postintervention mean of each cognitive domain measure.
Our secondary outcomes included domain-specific cognitive functions such as memory, executive functions, visuospatial functions, and language, reported in at least 2 studies. Other secondary outcomes were ADL, IADL, and QoL.
The outcomes in the included studies reported continuous data (mean and SD) and used different outcome measures. Therefore, standardized mean differences (SMDs) with 95% confidence intervals (CIs) were used to estimate the treatment effect to facilitate comparisons across all outcomes. SMDs were pooled and the inverse-variance random effects model was used considering the variability in methodology, participants, and intervention characteristics across studies. SMD between 0.20 and 0.49 represented a small effect, SMD between 0.50 and 0.79 a moderate effect, and SMD of 0.80 and higher a large effect.
The Z test was used to determine the treatment effect with a statistical significance threshold of P < .05. Heterogeneity was assessed using the chi-square statistic (2-tailed P < .10) using the Higgins I2 criteria in accordance with the Cochrane Collaboration thresholds, where 25%, 50%, and 75% imply small, moderate, and large heterogeneity, respectively.
Subgroup analysis was conducted to compare the treatment effects in studies with different diagnoses (MCI vs dementia), intervention dosage (intense vs mild, where intense is categorized as more than 3 formal sessions per week whereas less intense interventions is categorized as up to 3 formal sessions per week),
and intervention setting (home vs laboratory or clinic).
Figure 1 depicts the PRISMA flowchart of the systematic review and meta-analysis. After duplicate studies were removed, a total of 1291 original studies were initially screened for eligibility. Following title and abstract screening, 207 were full-text articles were independently reviewed by 2 authors. Sixteen studies were included in the systematic review. Two studies were excluded from the meta-analysis. Authors of these 2 studies were contacted but we could not retrieve the necessary data required to conduct the analysis.
The 16 studies included in this systematic review encompassed 909 participants with mean age ranging from 67 to 82 years. Of those, 461 (51%) participants were males as detailed in Table 1. Twelve studies (75%) included participants with MCI, whereas 3 studies (19%) included participants with dementia. Only 1 study focused on both MCI and dementia.
Table 1Study Sample Characteristics Table According to Cognitive Status
The type of control group varied across studies, with sample sizes ranging from 11 to 195 participants. Eleven studies used an active comparison group such as other nongaming computer-based activities,
Seven of 16 studies were conducted in the United States, 2 each were conducted in Italy and China, and 1 each in Australia, Eastern Slovakia, Greece, Republic of Korea, and United Kingdom (Supplementary Table 1). Although the type of brain gaming varied considerably across studies, most studies (n = 14, 88%) used a computer platform. Intervention periods ranged between 4 and 16 weeks. The training frequency varied between 2 and 15 sessions per week and the duration per session varied between 20 and 100 minutes.
Risk Bias Assessment and Quality of Studies
Based on OCEBM level of evidence, all RCTs were rated as level 1B, except 1 study that was 1C.
Overall, the quality of the studies was modest (Figure 2). The results of risk of bias assessment revealed that description of blinding of the participant and personnel was mostly unclear or low. In addition, the blinding of outcome assessment (detection bias) was largely unclear or low. Attrition bias was low. The funnel plot suggested no evidence of publication bias for overall cognitive functions from the composite scores (Supplementary Figure 1).
Global cognitive function from composite scores
Fourteen studies were included for calculation of global cognition function (Figure 3). The overall effect size was small (SMD –0.08, 95% CI –0.24, 0.41) and nonsignificant (P = .61). The heterogeneity across the studies was high (I2 = 77%).
Global cognitive function from Mini-Mental State Examination
Six studies used Mini-Mental State Examination as outcome tool to determine global cognition (Supplementary Figure 2). The meta-analysis of global cognition revealed a small (SMD –0.07, 95% CI –0.46, 0.59), nonsignificant effect size (P = .49) and moderate heterogeneity across the studies (I2 = 70%).
Secondary Outcomes: Cognitive Domains
Seven studies reported memory outcomes and were pooled to determine the effect of brain gaming on memory (Supplementary Figure 3). The overall effect size was small (SMD –0.17, CI –0.40, 0.06) and nonsignificant (P = .16). There was small heterogeneity across the studies (I2 = 21%).
As illustrated in Supplementary Figure 4, the pooled data for 8 studies demonstrated no superior effect of brain gaming on executive function (SMD –0.03, CI –0.30, 0.24; P = .82). The heterogeneity was small across the studies (I2 = 34%).
Data from 3 studies were pooled to determine the effect of brain gaming on visuospatial functions (Supplementary Figure 5). The overall effect size was small (SMD –0.09, CI –0.37, 0.18) and nonsignificant (P = .51). Heterogeneity across the studies was small (I2 = 0%).
Supplementary Figure 6 shows the pooled data from 3 studies, demonstrating a large effect (SMD 1.28, CI –0.23, 2.78) that was nonsignificant (P = .10) in favor of the brain gaming intervention on language. However, there was considerable heterogeneity across the studies (I2 = 96%).
Secondary Outcomes: Other
As shown in Supplementary Figure 7, the pooled data from 2 studies demonstrating a small (SMD 0.04, CI –0.78, 0.86), nonsignificant (P = .93) effect size on ADL, with no heterogeneity across studies (I2 = 0%).
Supplementary Figure 8 pools the data from 6 studies, showing a small (SMD 0.14, CI –0.13, 0.42) and nonsignificant (P = .31) effect size in IADL. There was no heterogeneity across the studies (I2 = 0%).
Supplementary Figure 9 shows a small (SMD –0.09, CI –0.56, 0.39) and nonsignificant (P = .72) effect size on depression based on data from 3 studies. There was small heterogeneity across the studies (I2 = 13%).
Quality of life
Data from 2 studies were pooled to demonstrate large heterogeneity across the studies (I2 = 98%) on QoL. The overall effect size was small but nonsignificant (P = .75), equally favoring the brain gaming and the control interventions (Supplementary Figure 10).
Subgroup Analysis on Effects of Brain Gaming
MCI vs dementia
Figure 4 displays the results of brain gaming on overall cognitive function in MCI and dementia, separately. Subgroup analysis based on diagnosis suggest that participants with dementia did not benefit more than participants with MCI from brain gaming on overall cognitive functions (SMD –0.19, CI –0.54, 0.16, vs SMD 0.16, CI –0.23, 0.54, respectively). Studies that focused on MCI demonstrated higher heterogeneity (I2 = 82%) vs studies that focused on dementia (I2 = 0%).
Weekly intervention dosage (sessions per week)
Studies were categorized based on dosage sessions per week similar to the one used by Bahar-Fuchs et al,
as more intense (ie, more than 3 formal sessions per week) vs less intense interventions (ie, up to 3 formal sessions per week). We found no significant differences between intervention dosage of brain gaming (SMD 0.00, CI –0.30, 0.30, vs SMD –0.18, CI –0.37, 0.74, P = .57) on overall cognitive function (Supplementary Figure 11).
Subgroup analysis revealed that categorized by setting did not change the benefit of brain gaming (Supplementary Figure 12). Studies focused on clinical and laboratory settings demonstrated higher heterogeneity (I2 = 87%) vs studies focused on home (I2 = 42%) and others (I2 = 1%).
The aim of our systematic review and meta-analysis was to evaluate the effectiveness of brain gaming—a subdomain of computerized cognitive training (CCT)—for adults with cognitive impairments. The evidence base for brain gaming in older adults with cognitive impairments has grown rapidly, partly driven by unsubstantiated claims from commercial application developers that brain gaming can maintain or improve cognitive functions. Based on our systematic review of 16 studies and our meta-analysis of 14 studies, we conclude that brain gaming is not more effective than control interventions in improving cognitive functions among adults with MCI or dementia. However, because of considerable heterogeneity of the included studies in terms of study design (eg, training prescription, gaming platform, and setting), we cannot confidently refute the premise that brain gaming is an effective cognitive training approach in this population.
Our conclusions largely resonate with a recent review on the effectiveness of 12 or more weeks of CCT (including immersive and nonimmersive brain gaming) on maintaining or improving cognitive function in MCI.
CCT interventions did not prove to be more efficacious than other interventions on speed of processing, verbal fluency, and quality of life. The low quality of evidence of the included studies hampered the authors’ ability to make firm conclusions about the effectiveness of CCT in MCI.
Conversely, 2 meta-analyses recently reported positive effects of CCT in older adults with cognitive impairments. Hu and colleagues found that CCT significantly improves cognitive functions especially related to various constructs of memory in participants with subjective cognitive decline and MCI.
Methodologic differences between the included RCTs, such as inconsistencies in study design, training prescription (duration and intensity), type of training program, outcome measures, and severity of cognitive dysfunction, may have led to the ambiguity of conclusions among the systematic reviews and meta-analyses. In particular, the scope of studies included in our systematic review and meta-analysis may be an explanation for the discrepancy between our findings and those of others.
Previous reviews did not particularly focus on the brain gaming literature, rather including several CCT paradigms, such as immersive virtual reality technology or CCT without adaptability in difficulty of training. Perhaps the ease of use, adaptability, and engaging elements that typically define brain gaming come at the expense of effectiveness of targeted, immersive CCT interventions or nonadaptive training.
Participants with subjective cognitive complaints showed twice as much benefit from CCT compared to participants with MCI. Based on our analyses, we found no difference in effectiveness of brain gaming interventions across these subgroups.
There is no consensus among reviews whether benefits of CCT generalize to ADLs or QoL measures. Coyle et al
on the other hand, reported most improvements on several psychosocial functions, including depression, QoL, and neuropsychiatric functions, among individuals with MCI. Our review did not reveal any benefits of brain gaming on ADL and QoL outcomes; this was expected because we included studies that only assessed brain gaming as the intervention. As highlighted by Harvey and colleagues, CCT by itself is not the typical strategy aimed at improving functional outcomes in clinical populations.
Strengths of our review include a wide search of the available literature on brain gaming interventions. There was considerable heterogeneity across studies in study design; thus, we advocate that our findings should be interpreted with caution. Although our review included only RCT designs, many of the included RCTs had small sample size, had short-term interventions, and were pilot studies. Larger RCTs, with consistent outcome reporting, would improve the ability to generate grounded conclusions regarding the effects of brain gaming. For example, for domain analysis, each domain was measured by a variety of outcome measures, which created heterogeneity in the meta-analyses. Another limitation to our work is that we assessed interventions that incorporated only brain gaming in the experimental group. Several studies, not included in our analysis, assessed multimodal cognitive training, which includes, for example, brain gaming combined with a pharmacologic intervention,
reviewed many of these multimodal games and concluded that more studies are needed to understand the advantages of a multimodal cognitive intervention approach for older adults with MCI.
Conclusions and Implications
The currently available data on brain gaming, designed to improve cognitive function in older adults with MCI and dementia, suggests that this approach does not improve cognitive function compared with the control group. Although individual studies continue to suggest a promising effect, collectively, the data do not bear this out. The considerable heterogeneity among the studies in terms of overall study design reflects a need for the research community to focus on salient design features and outcomes measurements so that the field can move forward in determining the best CCT to improve cognitive functions of older individuals with cognitive impairments.
We would like to acknowledge the role of the American Congress of Rehabilitation Medicine ACRM and contribution of members of the Measurement Networking Group and Applied Cognition Geriatric Team—Lilian Hoffecker, Carrie Ciro, Michael Cary, Michelle Thai, and David Berbrayer—for their contribution during early stages of data collection and analysis.
This study was conducted by the members of the Cognition in Aging Taskforce (previously called Applied Cognition Geriatric Taskforce) from the American Congress of Rehabilitation Medicine . This article was supported by a VA RR&D Career Development Award ( RX001850-01 to S. L. Kletzel ), the Swedish ALF Foundation of Umeå University and Västerbotten County Council , the Swedish Stroke Foundation (to X. Hu), and National Institute on Aging of the National Institutes of Health under Award Number ( K01AG058785 to H. Devos ). This work was also supported by National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR) ( 90SFGE0002 to S. Krishna n) and ( H133G130200 and 90IF0055-01 P.C. Heyn ). Any opinions contained in this document are those of the authors and do not necessarily represent the views of any of the above stated funding agencies.