Abstract
Background:
Mild cognitive impairment (MCI) and dementia are associated with increased age. MCI is a clinical entity described as a transitional state between normal cognition and dementia. Video games (VGs) can potentially promote cognition and functional capacity since multiple cognitive domains are recruited during gameplay. However, there is still a lack of consensus regarding the efficacy of VGs as therapeutic tools, particularly in neurodegenerative diseases.
Objective:
We aimed to analyze the impact of VGs on cognition and functional capacity outcomes in MCI/dementia patients.
Methods:
We conducted a systematic review and meta-analysis study (PROSPERO [CRD42021229445]). PubMed, Web of Science, Epistemonikos, CENTRAL, and EBSCO electronic databases were searched for RCT (2000-2021) that analyzed the impact of VGs on cognitive and functional capacity outcomes in MCI/dementia patients.
Results:
Nine studies were included (n = 409 participants), and Risk of Bias (RoB2) and quality of evidence (GRADE) were assessed. Data regarding attention, memory/learning, visual working memory, executive functions, general cognition, functional capacity, quality of life were identified, and pooled analyses were conducted. An effect favoring VGs interventions was observed on Mini-Mental State Examination (MMSE) score (MD = 1.64, 95%CI 0.60 to 2.69).
Conclusion:
Although promising, the effects observed should be interpreted with caution since serious methodological shortcomings were identified in the studies included. Nonetheless, the effect observed is higher than the minimum clinically important difference (1.4 points) established to MMSE. Future studies on the current topic urge. Recommendations for the design and conduction of cognitive RCT studies are presented.
Keywords
INTRODUCTION
Neurodegenerative disorders evolve through the years and range from mild cognitive impairment (MCI) to dementia [1, 2]. MCI is a clinical entity characterized by a cognitive decline considering the patient’s age, sex, and educational level [3, 4]. MCI is associated with difficulties in more cognitively demanding instrumental activities of daily living (IADL) [5], decreased quality of life [6], and higher long-term mortality when compared to non-impaired individuals [7]. In addition, MCI patients have a higher risk of converting to dementia (RR = 13.8, 95%CI 8.44 to 22.6%) [8], which is the seventh cause of death worldwide, and a major cause of disability and dependency among older adults [9].
Although conversion to dementia happens in a considerable proportion of patients over time, epidemiological studies have shown that 17.5%of MCI patients remain stable [10], and between 16%to 24%may return to a normal or near-normal range of cognitive functioning [11, 12]. These numbers open a window of opportunity to delay cognitive decline and increase the number of years patients live with autonomy and quality of life. Furthermore, even among patients with dementia, there is evidence that cognitive interventions can positively impact global cognition [13] and potentially reduce the progression of functional decline [14]. Therefore, the development of effective and sustainable digital cognitive interventions is currently a global public health priority [15].
The development of digital cognitive interventions based on video games (VGs) has received considerable attention and investment from the scientific community during the last decade [16–18]. As a training paradigm, the use of VGs is based on the concept that their building blocks or game elements can foster learning and learning transfer [17, 19–22]. Some of these game elements are immediate feedback, small incremental increases in task difficulty, and stimulating scenarios involving tracking multiple tasks or game components simultaneously (for a brief review, see [22, 23]). Moreover, the positive and pleasant experience associated with VGs’ ludic and engaging activities has also been studied as a paradigm to promote therapeutic adherence and patient self-management in chronic diseases [16, 24].
Computerized cognitive training programs (CCTP) and brain training programs (BTP) (both used interchangeably in the literature) have previously proved to be a promising approach to promote cognition and patient engagement when compared to traditional health education programs [25] and paper-and-pencil exercises [17, 27]. However, VGs seem to have a differential impact on human cognition when compared to CCTP/BTP. For instance, in Belchior et al. [28], the authors compared participants’ cognitive performance after playing a VG (i.e., Crazy Taxy), a CCTP (i.e., Posit Science In Sight), and a no-treatment group. Results showed that participants in both intervention groups showed improvements compared to the no-treatment group; however, the pattern of cognitive gains was different between intervention groups [28]. Similarly, Perrot, Maillot, & Hartley [29] found that playing Super Mario Bros for one hour a day, three times per week for two months, was associated with improvements in spatial memory, visuospatial functions, and processing speed, whereas playing a CCTP (i.e., Kawashima Brain Training) was associated with improvements in inhibitory control capacity.
Although the number of scientific publications regarding VGs as therapeutic tools has been growing over the last few years [30], there is still a lack of evidence regarding the efficacy of this approach. The importance of analyzing the impact of VGs in clinical populations, especially in patients at risk of cognitive decline and decreased functional capacity (i.e., the ability to perform the physical and cognitive activities needed to execute daily-living activities and maintain an independent living with quality-of-life) [31], such as MCI and dementia patients is crucial. Therefore, this paper aims to analyze the efficacy of VGs to promote cognition and functional capacity in MCI/dementia patients, based exclusively on the evidence gathered from randomized controlled trials (RCT). To our knowledge, this is the first study focused on answering the above research question.
METHODS
This meta-analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA 2020 guidelines) [32] (see Supplementary Table 1 for PRISMA 2020 Checklist) and the GRADE Handbook for grading the quality of evidence [33]. A protocol for this review was registered with PROSPERO [CRD42021229445].
Types of studies
For this review, we included studies that: 1) were RCT studies, 2) use VGs, 3) for at least three sessions, 4) to promote cognitive functioning and functional capacity, 5) in MCI or dementia patients, and 6) analyzed intervention’s impact by comparing post-treatment to baseline scores in standardized cognitive measures. Studies that targeted cognitive differences between gamers and non-gamers or studies exclusively focused on functional capacity outcomes were excluded.
Types of participants
Inclusion criteria
Eligibility criteria for inclusion were MCI or Alzheimer’s disease (AD) dementia diagnosis determined by 1) a neurologist or treating physician according to widely accepted diagnostic criteria (e.g., DSM-IV/V, CERAD, NINCDS-ARDRA or NIAAA), or 2) a score indicative of MCI or dementia in standardized cognitive tests or dementia rating scales, such as the Clinical Dementia Rating.
Exclusion criteria
Studies focused on patients with cognitive impairments in the context of other systemic medical diseases, neuropsychiatric disorders (e.g., schizophrenia, depression), oncologic diseases (e.g., cancer survivors), acquired brain injuries (e.g., traumatic brain injury, stroke), demyelinating pathologies (e.g., multiple sclerosis), movement diseases (i.e., Parkinson’s disease), and other neurodegenerative clinical profiles not compatible with probable AD, were excluded.
Types of interventions
For this review, we included only studies that used a VG as an intervention method. Off-the-shelf VGs (i.e., commercially available), serious games (i.e., video games specifically developed to achieve a change in the player’s patterns of behavior) [34, 35], and exergames (i.e., video games where players interact with the VG through body movements) [36] were all eligible for this category. Brain training programs and CCTP were excluded from this category following the authors’ point of view that these software are better described as neuropsychological tests that underwent a superficial gamification process (i.e., implementation of game elements within the main activity) [37, 38]. This perspective is also grounded in recent findings that show that VGs and CCTP/BTP differentially impact human cognition [28, 29].
Furthermore, only interventions composed of three or more sessions were considered. This criterion was applied to exclude papers focused on the study of VG applicability to specific populations (i.e., feasibility studies) and papers that use VGs as a cognitive assessment instrument typically composed of one or two experimental sessions.
Types of comparators
Studies that used as a control condition any cognitive stimulation/training technique such as CCTP (e.g., RehaCom®, Cogweb®), BTP (e.g., Lumosity®, CogniFit®, Brain Age®), paper-and-pencil exercises or low cognitive load exercises (e.g., read, write, drawing, painting) were considered as active control groups. Participants included in waiting lists, contact as usual (with no intervention), phone contact or no contact groups were regarded as passive control groups. Studies that used a combined intervention, such as exergames, were included if participants in the control group received the complementary intervention (e.g., physical exercise).
Types of outcome measures
Critical outcomes
We considered critical/primary outcomes the scores in standardized cognitive tests of individual cognitive functions (e.g., attention, language, memory). The cognitive domain and subdomain targeted by each test were determined according to the classification provided in [39] and [40]. The domain and subdomain identified were later used as criteria to pool data for meta-analysis purposes.
Important outcomes
Scores obtained in screening (e.g., MMSE, MoCA), global cognition (e.g., ADAS-Cog, CDR), functional capacity (e.g., Lawton IADL), psychological well-being and quality of life assessment instruments were considered important/secondary outcomes.
Search methods for identification of studies
Electronic searches
PubMed, Cochrane Central Register of Controlled Trials (CENTRAL), Web of Science (WoS), Epistemonikos, and EBSCO –Psychology & Behavioral Sciences Collection electronic databases were searched on the 21 January 2021. A set of keywords related to VGs, cognitive impairment, cognitive interventions, and IADL were identified and searched (for a detailed description of the search string, please see PROSPERO [CRD42021229445]). Electronic databases were restricted to studies published: 1) between 01.01.2000 to 31.12.2020, 2) with full-text availability, 3) in English, Portuguese, or Spanish language, and 4) with humans as study’s subject.
Hand searching
We manually searched the reference list of the included studies, reviews (narrative and systematic), and meta-analysis identified during the screening process to identify additional studies. Furthermore, protocol trials identified during this phase were also searched and separately analyzed to determine if the results were published and could be included in our review.
Data collection and analysis
Selection of the studies
One reviewer (FFB) conducted the electronic database search, imported the results to Zotero reference manager software [41] and removed the duplicates. Next, two reviewers (FFB and FR) independently screened the remaining titles based on title and abstract eligibility, followed by a full-text analysis. This phase was conducted using the Rayyan QCRI web application [42] to facilitate the screening process. Finally, disagreements concerning article eligibility were solved by a third element (AV or LC).
Data extraction and management
Data extraction was performed by one reviewer (FFB) and confirmed by a second reviewer (FR). The following elements were searched and gathered: Publication details: authors, title, year of publication, and journal Game: name, main activity, game elements, type (i.e., serious, commercial, exergame), and platform (e.g., game console, computer, tablet) Sample characteristics: MCI or dementia participants, MCI subtype, age, educational level, percentage of males, and confidence level using technology Intervention: single versus multiple cognitive domains, individual versus group format, purpose (training versus rehabilitation), frequency and duration, setting (hospital versus patient home), and user-experience (i.e., the subjective impression of a user towards a product before, during and after the use that includes user’s impression regarding fun of use, aesthetics, emotions, stimulation or attractiveness; UX) [43, 44] Study design: type of control, sample size, dropout rate, and type of statistical analysis Outcomes: baseline and post-intervention tests scores and criteria used to determine improvement.
Risk of bias
The risk of bias and quality of evidence assessment was performed independently by FFB and FR and confirmed by a third element (AV, DAS, and JC). Cochrane Collaboration’s Risk of Bias tool 2 (RoB2) [45, 46] was used to assess the risk of bias in individual studies. Five domains were analyzed: randomization process, deviations from the intended interventions, missing outcome data, measurement of the outcome, and selection of the reported results. Signaling questions were answered with Yes, Probably yes, No information, Probably no, and No risk of bias. Quality of evidence assessment was conducted using GRADEpro application [33].The data gathered for each outcome were assessed for very serious, serious or not serious risk of bias in the following domains: risk of bias, inconsistency, indirectness, imprecision, and other considerations.
Data analysis
Different scales can be used to assess the same cognitive outcome and, so, standardized mean difference (SMD) with 95%confidence intervals were calculated to compare treatment effects. Mean difference (MD) with 95%confidence intervals were calculated when the same test was identified across studies to assess the same outcome. Estimates of treatment effect were calculated using a random-effect model and the inverse variance method. Heterogeneity was assessed using the Chi-square test and the I2 statistics using as reference I2 > 40%as indicative of high statistical heterogeneity [33].
RESULTS
Results of database search
As shown in Fig. 1, from the initial set of references screened (n = 966), thirty-four articles were screened for inclusion based on full-text analysis, of which twenty-seven were excluded (see Supplementary Table 2 for exclusion reasons by study). One of the articles included [47] was made available online in 2020, and for this reason, it was identified during the electronic database search. However, its official date of publication is 2021. Two articles [48, 49] were added after analyzing the reference lists. In the end, nine studies were included in the systematic review and seven studies in the meta-analysis (studies included are marked in bold font in the Reference section).
PRISMA 2020 flow diagram for new systematic reviews which included searches of databases, registers, and other sources. CENTRAL, Cochrane Central Register of Controlled Trials; EBSCO, Psychology & Behavioral Sciences Collection electronic databases; WoS, Web of Science; RCT, randomized controlled trial; CG, control group.
Results of ongoing trials, literature reviews, and conference proceedings search
Forty-five clinical trials protocols were identified, four of which regarded protocols of articles identified during the database search. The remain 41 protocols were excluded because they were: on-going trials (n = 4); focused on different populations (n = 24); did not analyze cognitive outcomes (n = 2); control group did not receive the complementary intervention (n = 2); participants were not randomized (n = 1); interventions did not use VGs (n = 2) or used CCTP/BTP (n = 3); or it was not possible to access to the protocol after authors been contacted twice (n = 3). Forty-five literature reviews (i.e., narrative or systematic) were identified and hand-searched for relevant papers. One abstract of a conference proceeding was excluded due to a lack of information after two attempts to contact the corresponding author.
Included studies
Participants included
Nine studies were included in the systematic review, encompassing 409 participants (26.9%men). Two hundred and twelve participants were included in the intervention group (75.6±5.0 years; 25.5%men), and 197 in the control group (76.9±4.8 years; 28.4%men).
Mild cognitive impairment: Five studies focused on MCI patients (n = 262; mean sample size: n = 52). A total of 129 MCI patients were included in the intervention group (72.2±4.3 years; 24.3%men) and 133 in the control group (73.2±4.6 years; 30.1%men). MCI diagnosis was determined by: 1) clinical criteria only [50]; 2) standardized tests [48]; 3) clinical criteria together with a standardized tests [51]. Two studies reported that MCI diagnosis was performed before [47] or right before study enrollment [52]. However, no further information regarding the instruments used to determine the MCI diagnosis was presented. In addition, only one study specified the subtype of MCI (i.e., amnestic MCI) [50].
Dementia: Four studies focused on dementia were identified. In two of the studies included, patients with different etiologies of dementia were included (e.g., vascular dementia, mixed dementia) [53, 54]. Since we only considered AD type dementia in our review, the authors of these studies were contacted and asked for the data regarding only the participants diagnosed with AD dementia. Additionally, in a three-arm study [53], we only considered the data from AD patients included in the VG intervention group and the group that received the complementary intervention (i.e., aerobic exercise). A total of 147 patients with AD dementia were identified (mean group size: n = 37) and included in the intervention group (n = 83; 81.4±6.2 years; 27.7%men) or in the control group (n = 64; 83.0±5.2 years; 25.0 %men). Dementia diagnosis was determined by: 1) clinical criteria only [54, 55]; or 2) clinical criteria together with a standardized test [49, 53] (see Supplementary Table 3 for a detailed description of the methodological characteristics of the included studies).
Interventions included
Regarding the type of VGs used, seven were applied games [48–51, 53–55] and two were commercially available VGs [47, 52] used for training (n = 8) [48–55] or rehabilitation purposes (n = 1) [47]. Two studies used commercially available exergames [47, 52], and four studies used applied exergames [51, 53–55]. Regarding the use of VR technology, three studies used non-immersive VR [47–49], and one study used a fully immersive VR software program [51].
The most used game platform was computer (n = 4) [47–49, 55], followed by fitness machines [53, 54], game console [52], tablet [50], and head-mounted display [51]. Five VGs replicate daily-life activities such as shopping [48], juice making and house tiding [51], cycling [53, 54], and walking [49]. In the remaining studies, cognitive challenges/tasks with [47, 55] or without a physical activity component [50] were used. All VGs aimed to target multi cognitive domains such as episodic memory [48, 53], visual [47, 50] and motor abilities [47, 52], memory [47–49, 52], executive functions [47, 53], global cognition [50, 55], among others. Sessions were delivered in individual (n = 5) [48–50, 53], in group [55] or in mixed format [54]. Session format was not possible to determine in two studies [47, 51]. Two studies aimed to determine interventions’ impact on functional capacity [48, 55], and one study on quality of life [54]. Another study assessed the impact of VGs on neuropsychiatric symptoms, such as depression, anxiety, and apathy [50]. Participants included in the active control group performed paper-and-pencil exercises [47–49] or received the complementary intervention (i.e., physical activity [52, 53]; health education program [51]; arts, crafts and walking activities [54]. Two studies [50, 55] offered contact as usual to participants in the control group (i.e., inactive control group).
Regarding exposure to the intervention, a total of 6,740 min (≈112 h) were delivered to participants included in the experimental group, ranging between 200 and 2,400 min. The parameters of exposure varied between studies with the total number of sessions raging between 8 and 36 sessions (20.29±11.04), session duration between 20 and 100 min (47.9±25.6), weekly frequency between 2 and 5 sessions (mode = 3), and intervention length between 4 and 24 weeks (mode = 4).
Adherence, user-experience (UX), and confidence level using technology
Drop-out rate was within the expected range for cognitive interventions (mean = 5.7%[0%to 13.60%]). The main reasons for drop-out were: lost contact [51, 52]; worsening of the symptoms [52]; discontentment with treatment [52, 54]; admission to home care [54]; non-adherence to the intervention’s frequency [47]; and death [54]. Only two studies [50, 54] included measures that aimed to assess participants’ subjective and emotional experience (i.e., UX) while playing VGs by analyzing participants’ level of satisfaction, enjoyment, desire to continue to play, and adherence to the intervention. Also, only two studies assessed participant’s confidence level using technology [50, 54].
Risk of bias in the included studies
The eligibility criteria required studies to be randomized; however, five studies were classified as having a high overall risk of bias. The remaining four studies were assessed as having some concerns regarding the overall risk of bias (see Fig. 2). The main reasons for this classification were lack of information regarding the method used to generate the random sequence, lack of allocation concealment, participants and personnel were not blind to intervention assignment, and unavailability of the trial protocol before study conduction. Furthermore, concerns were raised regarding the measurement of critical outcomes due to baseline group differences [48], modifications in the method of administration of the cognitive tests [52], and timeline differences in post-treatment assessment sessions [50].
Risk of bias of the included studies (n = 9), +Low risk of bias; ?Some concerns; High risk of bias.
Effects of interventions
Two studies were excluded from the meta-analysis due to a lack of information after the authors had been repeatedly contacted [47, 55]. A total of 293 participants were included in our analysis (n = 154 in the experimental and 139 in the control group). As shown in Fig. 3, based on the scores gathered from four studies [50–52, 54], a large effect was identified on MMSE score favoring VGs group when compared to control group (MD = 1.64, 95%CI 0.60 to 2.69 I2 = 0%, p = 0.002).
Forest plot of the estimated effect with 95%confidence interval of VG interventions impact on general cognition (MMSE) scores when compared to the control group (n = 4).
Furthermore, VGs were associated with a non-significant improvement in attention outcomes (see Fig. 4a). The effects on task-switching, memory/learning and visual working memory outcomes were not statistically significant (see Fig. 4b-d). Additionally, we found moderate to high statistical heterogeneity for some of these outcomes (i.e., task-switching and memory/learning).
Forest plot of the estimated effect with 95%confidence interval of VG interventions impact on (a) attention, (b) task-switching, (c) memory/learning, and (d) visual working memory scores when compared to the control group.
No subgroup analysis was conducted for any of the outcomes due to the small number of studies included. Data regarding the impact of VGs on functional capacity measured through the Lawton IADL scale was reported in one study [48], and we found no significant differences between groups. Another study reported data on quality-of-life [54] measured through EQ-5D-5L, and no significant differences between VGs and control group were found. Finally, data regarding apathy symptoms measured through the Apathy Evaluation Scale was identified in one study [50], and we found no significant differences between groups (see Supplementary Table 4. Cognitive tests and data reported in the original studies grouped by cognitive outcome).
DISCUSSION
This meta-analysis aimed to analyze the impact of VGs on cognition and functional capacity in MCI/dementia patients by analyzing the data gathered from RCT studies. From the nine studies identified, seven were included in the quantitative analysis, and data was pooled for five cognitive outcomes: general cognitive function, attention, task-switching, memory/learning, and visual working memory.
Our results point towards a positive effect of VGs on general cognition in MCI/dementia patients. An estimated effect of MD = 1.64 points (95%, CI 0.60 to 2.69, I2 = 0%, p = 0.002) was found on MMSE score, which is higher than the minimal clinically important difference (MCID) of 1.4 points previously established for MMSE score [56]. MCID is defined as the smallest difference in an outcome as a result of an intervention which is perceived as a significant improvement by patients (i.e., with practical daily-life implications) as well as other stakeholders (i.e., careers, researchers and investors) [57]. Integration of patient-centered outcomes, such as MCID, has been encouraged by renowned organizations such as the International Consortium of Health Outcomes Measures (ICHOM) [58]. Such outcomes help provide meaningful interpretations and are essential to bringing forward useful suggestions into the design of effective interventions.
Previously, two meta-analysis [27, 59] have found a significant but small size effect on global cognition in MCI/dementia patients associated with cognitive interventions (g = 0.35, 95%CI = 0.20 to 0.51, p < 0.001 and g = 0.23, 95%CI 0.03 to 0.44, p = 0.03, respectively). A broad range of CCTP/BTP was considered in these studies (i.e., any CCTP/BTP intervention administered on a personal computer or gaming console, with a clear cognitive rationale). Since VGs seem to have a differential impact on human cognition compared to CCTP/BTP, we could speculate if the effect observed in our study is associated with the exclusive inclusion of VGs in our analysis.
Nonetheless, the results found in our study should be interpreted with caution. First, a small number of studies, with small sample sizes were included in the meta-analysis. Therefore, it was not possible to conduct subgroup analyses concerning the patient’s cognitive status (i.e., MCI versus dementia), type of MCI (e.g., amnestic versus non-amnestic MCI), or severity of dementia. Nonetheless, a random-effects model was applied to account for the studies’ heterogeneity. Second, the studies included in the meta-analysis were classified as having a high or unclear risk of bias, and consequently, it is not possible to conclude with a high degree of confidence that the estimated effect found on MMSE score is an accurate estimation of VG’s impact on human cognition. The inclusion of studies classified as having a high or unclear risk of bias is not supported by GRADE guidelines [33]. Nonetheless, we opted to conduct a more conservative assessment of the studies included. This contributed to identifying design and methodological limitations in the cognitive trials analyzed (see Implications for research section). Finally, using cognitive screening tests (e.g., MMSE) for clinical diagnosis and assessment of the impact of VGs also hampers the conclusions drawn.
Functional capacity was assessed in two [48, 55] out of the nine studies included. In both studies, participants showed a similar performance at baseline; however, the authors did not further analyze the data and the impact of VGs on this outcome. In a five-year longitudinal study [60], authors concluded that MCI patients showed impairments in grooming, telephone use, housework, and money management compared to non-cognitive impaired older adults. These authors also showed that functional and cognitive decline seems to co-occur [60] and, therefore, the capacity to live independently with quality-of-life is compromised from a very early stage of the disease. The absence of functional capacity outcomes in the studies here included may be explained by the fact that despite the growing number of instruments aiming to assess functional capacity, there is still an absence of validated instruments sensitive to the specific improvements that MCI/dementia patients might experience associated to the additional cognitive stimulation provided by VGs. On the other hand, there was an unexpected absence of progression/conversion to dementia measures in the studies included. Monitoring of neurodegenerative diseases is crucial for therapeutic adjustments (cognitive and pharmacological) and gives patients and their families time to plan future steps and accommodate present necessities [61].
The confidence level using technology and UX were assessed only in two studies [50, 54]. However, user’s confidence level using technology highly impacts participants’ ability to use and adapt to new technologies and must be acknowledged when interpreting the interventions’ results [62]. Moreover, the set of usability aspects such as efficiency, effectiveness, learnability, perspicuity, controllability, and the hedonic elements, which are part of the concept of UX [43], should also be taken into account when analyzing the intervention impact [1, 63]. Neglecting users’ capacity to adapt to technology, the meaning attributed to VGs, and how they experienced it may compromise short- and long-term adherence to VG-based interventions [43].
Implications for practice
Our study presented encouraging, although preliminary results regarding a positive effect of VGs on general cognition; however, the scarcity of high-quality studies do not allow us to draw firm conclusions regarding the benefits of using VGs in patients with MCI/dementia. However, it is important to highlight that no adverse side effects were reported (neither physical nor psychological).
Video games have been gaining popularity among older adults. In 2019, 65%of American adults played VGs, and 22%to 25%of players aged 55–64 years have been VG consumers for more than 25 years [64]. This consuming trend is already present in research and therapeutic contexts, with an increasing amount of digital cognitive platforms being developed every year [65]. From this point of view, health professionals should be aware of the difficulties that less technological savvy older adults face when they first contact with digital platforms and devices. For instance, in our review, only two studies [50, 54] gathered information regarding participants previous experience with technology and video games. In addition, clarifying patients’ doubts and stereotypes regarding the use VGs and technology as therapeutic instruments is an essential part of promoting patient’s long-term adherence to digital healthcare services, such as digital cognitive interventions.
Implications for research
The methodological limitations identified in the studies included in our systematic review are the same that have been reported in the scientific literature for more than 15 years [66]. While the nature of cognitive interventions makes it challenging to follow some of the RCT guidelines [67], the quality of the evidence produced could be improved if simple measures were adopted. For instance, study design can be improved by 1) including information about the method used to generate the random sequence and the allocation concealment; 2) integrating a third element responsible for the cognitive assessment, blind to participant allocation; 3) making available the trial protocol before the research is published, using free online platforms, such as Open Science Framework [68]; and 4) using an extension of CONSORT guidelines –CONSORT-SPI [69] to guide the design and conduction of cognitive RCT studies. Also, a detailed description of how (i.e., clinical criteria) and when the clinical diagnosis was performed highly improves both the external validity of the individual studies and the conclusions of meta-analytic studies. Information regarding the type of MCI and the severity of dementia would also improve the analysis. We tried to control some of these aspects by including specific MCI/dementia diagnosis eligibility criteria. Nonetheless, the requirements established were insufficient since, in the sample of studies included, only one study mentioned MCI subtype [50], and two studies presented vague descriptions of how clinical diagnostics were performed [47, 52]. Future studies overcoming the limitations mentioned above and including psychological (e.g., depression, anxiety), quality-of-life [58], and functional capacity outcomes, will significantly improve the quality of the evidence produced by cognitive RCT studies.
CONCLUSION
To our knowledge, this is the first meta-analysis based on RCT studies that aimed to analyze the impact of VGs on cognitive and functional capacity outcomes in MCI/dementia patients. A promising preliminary effect size of MD = 1.64 points (95%, CI 0.60 to 2.69, I2 = 0%, p = 0.002) was identified on MMSE score favoring VGs interventions. Previous studies that focused on a wide range of digital cognitive interventions found smaller effect sizes, suggesting that specific VGs characteristics are better learning catalysts when compared to other cognitive approaches. However, these results should be interpreted with caution since a small number of studies contributed to the estimated effect, and serious methodological shortcomings were also identified in the studies included. Further studies with a more robust methodological design are needed to determine the impact of VGs on human cognition and behavior with a higher degree of confidence.
Footnotes
ACKNOWLEDGMENTS
We would like to thank all authors whose papers were identified during the screening phase and kindly provided valuable information that supported our decision process. Moreover, the authors want to express a special acknowledgement to Maria Ana Mello e Castro for her thorough revision of the English language of the manuscript.
This research is carried out as part of the doctoral studies of the first author [Ref: PDE/BDE/127784/2016] and for which she received scholarships from the following entities: Fundação para a Ciência e a Tecnologia and Nippon Gases Portugal through European Social Found and Human Capital Operational Programme, co-financed by Portugal 2020 and European Union. This project also received financial support from FCT through LASIGE Research Unit funding, ref. UIDB/00408/2020 and ref. UIDP/00408/2020, and Clinical Research in Non-communicable Neurological Diseases Laboratory, Instituto de Medicina Molecular João Lobo Antunes, Faculdade de Medicina, Universidade de Lisboa, Portugal.
