By 2050, the older adult population is expected to double on a global scale [1] followed by a growing demand for high-quality support among this population. There is a variety of disabilities related to aging. However, a considerable percentage of people aged >65 years are affected by cognitive impairments. According to the Alzheimer’s Association, approximately 15% to 20% of people aged >65 years are affected by mild cognitive impairment (MCI) [2]. It is a progressive, terminal brain disorder for which the cure and cause are unknown [3]. Worldwide, the number of Alzheimer disease (AD) cases is rising; estimates indicate that 44 million people currently have AD or dementia. As the Baby Boomer generation passes the age of 65 years, the number of people experiencing AD is expected to reach 76 million by 2030 [4,5]. The consequences of these cognitive disorders include poor judgment, decreased activity, and memory loss. They are difficult to manage; thus, it would be highly beneficial to develop proper and effective interventions for the older adult population.

These disorders imply a decline in various cognitive functions such as memory [6,7], executive function [8,9], attention [10], and processing speed [11]. Processing speed refers to the time that an individual takes from receiving a stimulus to reacting to it. Slowed processing speed is a typical characteristic of cognitive aging. It implies a slowing rate at which individuals perform decision-making as well as motor and perceptual tasks. Moreover, processing speed represents one of the critical predictors of performance related to cognitive tasks in older adults; it is the basis of a primary hypothesis for age-related cognitive decline [11,12]. Accordingly, aging research primarily aims to design methods for improving and maintaining cognitive functions in older adults.

Serious games are one of the interventions that have been used to improve and maintain processing speed in older adults. Serious games represent a type of video game used for training and educational purposes in different contexts [13,14]. Serious games have shown promising results in improving cognitive functions such as global cognition [15-18], processing speed [16-20], memory [16-23], executive function [16-22,24], and attention [16,17,19,20,22]. The most common types of serious games used for improving cognitive abilities are exergames (video games that require physical exercise as part of playing) and cognitive training games (video games that include cognitively stimulating activities designed to maintain or promote the users’ cognitive abilities).

Several studies have explored the therapeutic impact of serious games on processing speed. However, the evidence from these studies is still fragmented. Aggregating the evidence through systematic reviews is important to draw conclusions about the effectiveness of serious games in improving processing speed. A total of 5 systematic reviews have pooled findings from these studies, but they are undermined by some limitations [16-20]. Specifically, these reviews (1) focused on older adults without cognitive impairment [16,18-20], (2) included quasi-experiments or pilot randomized controlled trials (RCTs) [17-19], (3) did not assess the quality of the meta-analyzed evidence [17,18,20], (4) only focused on a specific type of serious game such as cognitive training games [16,17,20] and exergames [18], or (5) did not compare the effect of serious games with a specific comparator (eg, no intervention, conventional exercises, or conventional cognitive activities) [16-18,20]. Accordingly, the aim of this review was to examine the effectiveness of serious games on the cognitive processing speed of an older adult population living with cognitive impairment. To address the aforementioned limitations, we (1) focused on older adults with cognitive impairment; (2) included only RCTs; (3) appraised the quality of the meta-analyzed evidence using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach; (4) included all types of serious games; and (5) compared the effect of serious games with that of a specific comparator. In this review, effectiveness refers to the degree to which serious games are successful in improving cognitive processing speed.

Only RCTs that evaluated the effectiveness of serious games in improving processing speed among older adults with cognitive impairment were included in this study. To be more precise, we considered studies that included serious games available on any digital platform, such as PCs, video game consoles (eg, Xbox and PlayStation), mobile phones, tablets, handheld devices, Nintendo, or any other type of digital device. The game had to be the major component of the intervention and used solely for therapeutic purposes. Studies involving serious games in combination with other interventions were included if the control group underwent the same adjacent intervention. Games that were not based on digital technology (eg, paper-and-pencil games and board games) or that were used for monitoring, screening, diagnosis, or research were excluded.

The study population was older adults (aged ≥60 years) with any type of cognitive impairment or disorder as confirmed by checking the inclusion criteria or baseline scores against defined diagnostic criteria (eg, Mini-Mental State Examination). Older adults without cognitive impairment, health care providers, and caregivers were beyond the scope of this review. No restrictions were applied regarding gender or ethnicity.

The outcome of interest in this review was cognitive processing speed. No restrictions were applied regarding the outcome measures. This review did not consider studies that focused only on cost-effectiveness, acceptance, feasibility, satisfaction, or cognitive abilities other than processing speed. The focus of this review was on postintervention data (ie, outcome data collected just after the intervention) rather than follow-up data (ie, outcome data collected a period after the intervention).

All types of RCTs (ie, parallel, cluster, crossover, and factorial) were considered in this review, whereas pilot RCTs, quasi-experiments, observational studies, and reviews were excluded. We included journal articles, conference proceedings, and dissertations, whereas abstracts, conference posters, commentaries, proposals, and editorials were excluded. This review was restricted to only studies written in the English language and published since 2010. We did not apply restrictions on the country of publication, comparator, or study settings.

We identified relevant studies by using the following process. First, all the retrieved studies were imported into EndNote (Clarivate Analytics) to find duplicate publications and remove them. Second, the titles and abstracts of all the retrieved studies were checked by 2 reviewers working independently (the first and second authors). Finally, both reviewers independently read the complete texts of the studies included in the previous step. Any disagreements were resolved via discussion. Steps 2 and 3 had an interrater agreement (Cohen κ) of 0.88 and 0.96, respectively.

Before extracting the data, we pilot-tested the data extraction form with 2 of the included studies. Microsoft Excel was used by 2 reviewers (the first and second authors) to independently extract data from the included studies. Any disputes in the extracted data between the reviewers were resolved through discussion. The first and corresponding authors of the included studies were contacted to retrieve outcome data (eg, mean, SD, and sample size) if they were missing from the published articles. The data extraction form is provided in Multimedia Appendix 3.

Two reviewers (the first and second authors) independently examined the risk of bias in the included studies using the Risk of Bias 2 tool [27]. This tool evaluates the risk of bias in 5 areas of RCTs: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the results [27]. The reviewers held discussions to resolve any disagreements, and the interrater agreement was 0.85.

Narrative and statistical methods were used to summarize the collected data. In our narrative synthesis, we used text and tables to describe the characteristics of the included studies (study metadata, population, interventions, comparisons, and outcome measures). The results of the experiments were compiled and classified according to the comparator: no or passive interventions, conventional exercises, and other serious games. When 2 or more studies from the same comparator submitted sufficient data (ie, mean, SD, and number of participants in each intervention group), meta-analyses were performed using Review Manager (RevMan 5.4; The Cochrane Collaboration). As the type of data for the outcome of interest (processing speed) was continuous and the instruments used to evaluate the outcome varied across the included trials, the standardized mean difference (SMD; Cohen d) was used to estimate the overall effect of each study. We also chose the random effects model for the analysis because of the high clinical heterogeneity among the meta-analyzed trials in terms of serious game characteristics (eg, type, duration, frequency, and period), population characteristics (eg, sample size, mean age, and health condition), and outcome measures (ie, tools and follow-up period).

To assess the degree and statistical significance of heterogeneity in the meta-analyzed studies, we calculated I² and a chi-square P value, respectively. A chi-square P value of ≤.05 indicated heterogeneous meta-analyzed studies [28]. When I² ranged from 0% to 40%, 30% to 60%, 50% to 90%, and 75% to 100%, the degree of heterogeneity was considered as insignificant, moderate, substantial, or considerable, respectively [28].

We used the GRADE approach [29] to appraise the overall quality of the evidence resulting from the meta-analyses. The GRADE approach examines the quality of evidence based on 5 domains: risk of bias, inconsistency (ie, heterogeneity), indirectness, imprecision, and publication bias [29]. The quality of the meta-analyzed evidence was independently assessed by 2 reviewers (the first and second authors). Any differences between the reviewers were resolved via discussion, and the interrater agreement of the reviewers was 0.94 [30].

As shown in Figure 1, a total of 548 records were found after searching 8 electronic databases. Using the EndNote software, of the 548 records, 98 (17.9%) duplicates were removed. Checking the titles and abstracts of the remaining records resulted in the exclusion of 63.1% (346/548) for several reasons shown in Figure 1. Reading the full text of the remaining 104 publications resulted in the exclusion of 88 (84.6%) studies (Figure 1). This review included 16 RCTs in total [31-46]. Of these 16 RCTs, 13 (81%) were included in the meta-analyses [31-43].

The included studies were published between 2012 and 2021 and came from 11 different countries (Table 1). Except for a book chapter, all the included papers (15/16, 94%) were peer-reviewed articles. Parallel RCTs were the most common type of trial used in the included studies (14/16, 88%). The sample sizes of the included studies ranged from 20 to 195, with an average of 79.9. The average age of the participants was reported in 94% (15/16) of the studies and varied between 66 and 82.9 years, with an average of 75 years. Male participants in 94% (15/16) of the studies ranged from 23.5% to 71%, with an average of 47.5%. The mean Mini-Mental State Examination score of the participants was reported in 88% (14/16) of the studies and ranged from 18.6 to 28.1, with an average of 24.4. The most common disorder among the participants in the included studies was MCI (11/16, 69%). Participants were drawn from clinical (10/16, 62%), community (4/16, 25%), and clinical and community (2/16, 12%) settings.

We identified 16 distinct serious games used in the studies (Table 2). More than one game was used in 6% (1/16) of the studies. The serious games in the trials were divided into 2 categories based on the treatment modality they provided: cognitive training games (12/16, 75%) and exergames (4/16, 25%). In 88% (14/16) of the studies, the games were designed with a “serious” objective from the start (designed serious games). By contrast, games in the remaining 12% (2/16) of the studies were not planned as serious games from the start but were instead used for a serious purpose (purpose-shifted games). Serious games were played under the supervision of health care providers or caregivers in most studies (11/16, 69%). The game duration in the included studies ranged from 25 to 100 minutes. The frequency of playing the games ranged from 2 to 7 times per week, but it was <4 times per week in approximately two-thirds of the studies (11/16, 69%). The duration of the interventions ranged from 4 to 25 weeks, but it was <13 weeks in three-fourths of the studies (13/16, 81%).

Half of the studies (8/16, 50%) delivered no or passive interventions to the comparison groups (eg, reading newspaper articles, surfing the internet, or watching a documentary program), whereas these groups received active interventions (eg, conventional exercises and other serious games) in 62% (10/16) of the studies (Table 3). A total of 12% (2/16) of the studies delivered both active and passive interventions as comparators. The active interventions had a duration ranging from 25 to 100 minutes. The active interventions were performed once to 7 times per week. The active comparator duration ranged from 5 to 25 weeks. The outcome of interest (ie, processing speed) was assessed using 11 different tools, with the Trail Making Test A being a frequently used tool in the included studies (7/16, 44%). The outcome of interest was measured after the intervention in all the included studies (16/16, 100%). In total, 44% (7/16) of the studies followed the participants after the interventions, and the follow-up period varied between 4 and 261 weeks. The number of participants who dropped out of the included studies ranged from 0 to 28.

As shown in Figure 2, a total of 31% (5/16) of the studies were judged to have a low risk of bias in the “randomization process” domain. With regard to the “deviations from the intended interventions” domain, there was a low risk of bias in 75% (12/16) of the studies. The risk of bias because of missing outcome data was low in 81% (13/16) of the studies. All the included studies (16/16, 100%) were judged to have a low risk of bias in the “measuring the outcome” domain. In half of the included studies (8/16, 50%), the risk of bias was rated as low in the “selection of the reported results” domain. According to these judgments, only 19% (3/16) of the studies were judged to have a low risk of bias in the last domain (ie, overall bias). Reviewers’ judgments about each “risk of bias” domain for each included study are presented in Multimedia Appendix 4 [31-46].

This study compiled evidence regarding the effectiveness of serious games in improving processing speed among older adults with cognitive impairment. Our review revealed that serious games are as effective as no or passive interventions and conventional exercises in improving processing speed and that there is no difference between cognitive training games and exergames when compared with conventional exercises. The nonsuperior effect of serious games over no or passive interventions and conventional exercises may have been due to the following reasons: (1) serious games in the included studies targeted cognitive abilities in general rather than processing speed specifically and (2) the sample size in most of the included studies (10/16, 62%) was small (<100).

Our findings are consistent with findings of a previous review that showed no statistically significant difference in processing speed between the cognitive training game group and the passive or active intervention group among older adults with MCI or dementia [17]. However, our findings are not consistent with findings of previous reviews [16,18-20]. Specifically, 4 reviews showed that cognitive training games [16,19,20] and exergames [18] are more effective than other interventions (passive or active) in improving processing speed among healthy older adults. This inconsistency may be attributed to the fact that (1) all these reviews focused on older adults without cognitive impairment only, whereas our review focused on older adults with cognitive impairment [16,18-20]; (2) they included quasi-experiments and pilot RCTs [18,19]; and (3) they did not compare the effect of serious games with that of a specific comparator (eg, no intervention, conventional exercises, or conventional cognitive activities) [16,18-20].

Serious games did not have a superior effect on processing speed among older adults with cognitive impairment in comparison with no or passive interventions and conventional exercises. However, this finding should be cautiously interpreted for the following reasons: (1) the quality of the evidence from all meta-analyses was very low mainly because of high risk of bias, high heterogeneity, and imprecision of the estimated total effect sizes; (2) the number of studies included in some meta-analyses was small; and (3) the sample sizes in many meta-analyzed studies were small. Therefore, until more robust evidence is available, serious games should be offered or used as a supplement rather than as an alternative intervention targeting processing speed. Future reviews should investigate the long-term impact of serious games on other cognitive abilities and disorders among people from different age groups with or without cognitive impairment. Further trials should be undertaken to investigate the effect of serious games that specifically target processing speed rather than cognitive abilities in general.