Assessing efficacy of non-pharmacological interventions
countering cognitive decline in healthy older adults: A review
Luuk Lamens
10519300
Luuklamens94@gmail.com
Master Thesis
Submitted on: 15-03-2020
University of Amsterdam
Master Brain and Cognitive Sciences
Supervisor & first examiner: Jaap Murre
Abstract
With the aging population, the number of patients with dementia is rising with an alarming pace, with an expected threefold increase by 2040 (World Health Organization, 2006). This presents a major challenge for public health care since treating dementia patients is long-lasting and labour-intensive. Improving cognitive functions can create a certain cognitive reserve, which makes a person less vulnerable for developing dementia and might prevent dementia (Stern, 2009). Research into cost-effective ways of countering cognitive decline is, therefore, imperative at this stage. In the recent 20 years a multitude of studies have assessed potential non-pharmacological interventions to counter cognitive decline. These studies suffer, however, from heterogeneous results and methodological problems. Therefore, this review examines methodologically robust random controlled trials with an active control group and focussed on transfer effects. Intervention studies involving cognitive brain
training, video games, leisure activities or volunteering were included. 29 individual studies were critically assessed, showing mostly minimal to moderate near transfer effects. As of now, the methodological issues are too frequent and serious to demonstrate compelling evidence for any intervention to successfully counter cognitive decline in older adults.
Keywords: aging, cognitive decline, intervention, healthy older individuals, randomized controlled clinical trials, cognitive training, video games, leisure activities, volunteering
Assessing efficacy of non-pharmacological interventions countering
cognitive decline in healthy older adults: A review
Globally, the proportion of older adults is rising rapidly. In the European Union the percentage of adults of 80 years and older is expected to grow from 5% to 12% of the total population (European Commission, 2015). In the United States the number of older adults is predicted to grow from 45 million currently to 70 million by 2030 (Ortman et al., 2014). As people age, cognitive functions are more susceptible to cognitive decline (Harada et al., 2013). Disorders characterised by cognitive decline are, therefore, becoming a major health problem. Dementia, including Alzheimer’s disease, is a neurological progressive
degenerative disorder that is characterised by decline in memory, language, problem-solving and other cognitive functions that affect a person’s ability to perform everyday activities. With the demographic shift, the number of people with dementia is rising with an alarming pace, with an expected threefold increase by 2040 (World Health Organization, 2006). This presents a major challenge for public health care since treating dementia patients is long-lasting and labour-intensive. The world-wide costs of dementia were estimated at 818 billion US dollars in 2015, a 35% rise in costs compared to 2010 estimates (Wimo et al., 2017). Research into cost-effective treatment of dementia is, therefore, imperative at this stage.
Studies focussed on pharmacological therapies have, up to recently, not shown major breakthroughs in treating dementia (Casey et al., 2010). Also, research on drug-based
therapies is a costly and long-lasting process with both animal and human-based clinical trials. In the recent 20 years, a multitude of studies have assessed potential
non-pharmacological interventions to counter cognitive decline which in turn might prevent dementia. The underlying theory is that improving cognitive functions creates a certain cognitive reserve, which makes a person less vulnerable for developing dementia (Stern, 2009). Since the behavioural therapies focussed on preventing cognitive decline are easier to implement and more cost-effective than drug-based therapies, research into behavioural interventions that prevent cognitive decline in healthy older adults is necessary.
One intervention in particular has gained a substantial amount of attention by the public. Brain training applications on phones, tablets, desktops or gaming consoles are increasingly popular and already a major industry. According to SharpBrains, an
independent market research firm, people have spent 1.9 billion pounds on brain training applications in 2018 (SharpBrains, 2019). Due to the widespread usage of these applications, most studies investigating interventions to boost cognition implement cognitive brain training. Despite the claims of these applications that it will improve the user’s cognition, a scientific consensus has not been reached. Reviews in the last few years have both
demonstrated modest effects (Klimova, 2016; Mansor et al., 2019) and limited and inconsistent effects of cognitive brain training (Buitenweg et al., 2012). In addition to cognitive brain training, numerous other non-pharmacological interventions countering cognitive decline are present in the literature. Several recent reviews and meta-analyses discuss (combinations of) the following interventions to counter cognitive decline in older adults: (aerobic) exercise, voluntary work, playing (action) video games, musical experience or therapy, dancing, changing your diet, meditation and leisure activities including reading, writing, making crosswords, playing board games, engaging in art/handicrafts (Christie et al., 2017; Iizuka et al., 2019; Klimova et al., 2017). Again, the reviews present contrasting conclusions. Klimova et al. (2017) conclude that all selected interventions in their review show positive effects on cognition, Iizuka et al. (2019) conclude that 13 out of the 20 selected studies show positive effects on cognition and Christie et al. (2017) conclude the evidence for decreasing the risk of developing dementia with non-pharmacological interventions remains equivocal. Despite the mixed conclusions, both these reviews and the reviews on cognitive brain training address the difficulty in comparing results of individual studies due to their heterogeneity. Coming to a unified conclusion is difficult with the wide range of outcome measures, exposure time and study designs reported. Aside from the heterogeneity, weak methodology is also major concern addressed in these reviews.
First, the absence of an adequate control group is problematic. Without a control group an improvement on the cognitive tests after the intervention compared with baseline can be attributed to a practice (or test-retest) effect. By adding a control group and
comparing the change score (post-intervention – baseline) of the intervention group with the change score of the control group, a difference in improvement cannot be attributed to a practice effect anymore. Adding a control group is in itself, however, not sufficient to attribute an effect purely to the intervention. Many studies have a no-contact or waitlist control group. The participants in these groups have no contact with the researchers beside performing the cognitive tests at baseline and at follow-up and in a waitlist group the participants have the possibility to try the intervention after the study has finished. In terms of encouragement and social contact, this creates a contrast with the intervention group. In most of the studies, participants in the intervention group have regular contact with the researchers either by telephone or face to face. Also, depending on the intervention,
participants learn something new such as handling a gaming console. Especially with older adults, it is plausible that merely by being engaged in a new activity or having regular social contact can lead to better performance at a cognitive test. A single study with a no-contact or waitlist control group reporting an improvement on cognition is strictly no evidence in favour of the specific intervention applied but merely in favour of the act of applying an intervention. In order to tackle this problem, these social and novelty components need to be matched between control and intervention groups by implementing an active control
condition. Ideally, this active control condition is equal in exposure time, social contact and novelty to the activity implemented.
Second, an important concept of cognitive training, which is overlooked in some studies, is that of transfer. Transfer represents the degree to which a learned skill is
demonstrated in a different context, with near and far transfer referring to generalisation of the learned skill to contexts being proximal and distant to the learned skill, respectively. In an average cognitive brain training study, participants perform a cognitive test-battery at baseline and after the intervention. During the intervention, participants perform the brain training with different cognitive tests than in the test-battery. As expected, participants will perform better on the cognitive training tests they are trained on during intervention. This, however, might simply be a practice effect. Only by comparing the outcomes of the
untrained tasks of the test-battery from follow-up to baseline, improvement of cognitive functions can be measured. Near transfer entails an improvement on an untrained task of the same cognitive domain as the trained task, far transfer entails an improvement on an
untrained task of a different cognitive domain as the trained task.
Third, the absence of multiple comparison correction in this field is also a frequently encountered problem. Since most studies assess cognitive improvement by implementing a wide range of cognitive measures, the necessity of a proper multiple comparison correction is fundamental to prevent false positives.
Given the necessity for finding cost-effective ways to prevent cognitive decline and the methodological problems in the current literature, this review will investigate the effect of non-pharmacological interventions on preventing cognitive decline in healthy older adults based on methodologically robust studies. The current review will only discuss individual studies with an active control group and focus on transfer effects. Due to the extensive range of interventions present in the literature, this review will only focus on cognitive brain training, video games, leisure activities and volunteering. Furthermore, it will only use randomised controlled trials (RCTs), the most rigorous way of determining whether a cause-effect relationship exists between intervention and outcome and for assessing a cost-cause-effective treatment (Sibbald and Roland, 1998). By critically assessing these studies, this review will attempt to answer the following research question: are cognitive brain training, video gaming, leisure activities or volunteering an effective intervention to counter cognitive decline in healthy older adults?
The methodology was as follows. As a starting point, reviews were searched on Google Scholar with the following keywords: “Cognitive Decline” AND “Review” AND “Elderly” OR “Older People” OR “Older Adults” AND one of the following intervention keywords: “Brain Training” OR “Cognitive Training” OR “Cognitive Brain Training” OR “Video Games” OR “Leisure” OR “Leisure Activities” OR “Volunteering” OR “Voluntary Work”. Only peer-reviewed English written reviews published after 2010 were examined. For cognitive brain training, only reviews published after 2015 were included because of the abundance of reviews and time constraints. For any review, it was checked if it had been cited by a more recent relevant review in order to include the most recent studies. Next, RCTs were extracted from reviews if they included healthy 50+ participants, an active control group, cognitive outcome measures and one of the four interventions mentioned above. For a complete overview of the individual RCTs examined, see appendix I.
This review will explore each intervention in a separate section. Each section will examine and weigh evidence from individual RCTs in order to assess the efficacy of the intervention. Various effect size measures are reported in examined studies, table 1 presents the various effect sizes and their corresponding implication (Cohen, 1988). After discussing interventions separately, the discussion section will compare the efficacy of the interventions with each other and present relevant suggestions for future research.
Table 1: Meaning of reported effect size magnitudes.
Measure Small effect Moderate effect Large effect
Cohen’s d 0.2 0.5 0.8
Partial eta-squared (ηp2) 0.01 0.06 0.14
Cognitive Brain Training Interventions
Cognitive brain training is the most ubiquitous intervention in countering cognitive decline in older adults, both in academia and society. In our definition, cognitive brain training involves a digital application specifically targeted at training cognitive functions in healthy people. Every study including an intervention or a control group with a
commercialised video game that is not specifically created to enhance cognitive functions will be discussed in the video game intervention section.
RCTs were extracted from reviews by Klimova (2016) and Mansor et al. (2019). A total of 33 studies were examined from these reviews. After removing two duplicates, 11 studies without an active control group, five studies with a commercial video game intervention, one study with subjects with mild cognitive impairment and one study without random
participant allocation to groups, 13 studies were left to examine.
Exergame Studies
Out of these 13 studies, five involved an intervention combining cognitive training and exercise called exergaming (a concatenation of exercising and gaming). Exergames utilise video games that require bodily movement while simultaneously engaging participants in cognitive tasks.
Kayama et al. (2013) implemented a dual-task tai chi task (DTTC) in which subjects had to solve Sudoku puzzles by dragging numbers to the correct boxes with full body movements. The DTTC group (n = 26) demonstrated no significant difference in verbal fluency but did show a significant difference in shifting compared to the control group who performed regular physical training (n = 15). Shifting, assessed by the trail-making-test (TMT) significantly differed (p = 0.03, Cohen’s d = 0.18) in one of three measures and this change was mainly driven by a decrease in performance of the control group.Notably, no multiple comparison correction was applied.
A study by Schoene et al. (2015) created an exergame called STEP. With STEP, subjects must put their feet on certain locations of a platform, which are determined by playing cognitive mini games presented on a computer screen. The intervention group (n = 39) improved significantly more in four out 19 tasks compared to the control group (n = 42).
These differences were found in two measures of speed of processing (p < 0.001; Cohen’s d = 0.856 and 1.00) and two measures of working memory (p < 0.03; Cohen’s d = 0.12 and 0.31), but not in tasks related to attention and updating. A major weakness, however, of this study is the control condition in which participants were given a brochure about falling risks and were asked to continue their usual activities. With a rather passive control and no multiple comparison correction, the results should be interpreted with caution. Interestingly, a more recent study with the same STEP exergame intervention by Schattin et al. (2016)
implemented a more robust control group who performed balance exercises. Both
intervention and control activities were adapted to individual performance and exposure time (12 hours) was matched. Here, no significant difference was present in change scores for any of the tasks related to executive functions (attention, shifting, inhibition and updating).
A similar study by Eggenberger et al. (2016) also involved a platform-based exergame (DANCE) as an intervention and balancing exercises a control. Again, the intervention group did not improve more in any of the tasks related to executive functions compared to the control group. Another study by Eggenberger et al. (2015) involving the same exergame implemented a programme lasting three months with three groups: the DANCE group involving attention demanding cognitive tasks (n = 24), a group who performed a verbal memory task while walking on a treadmill (n = 22) and a group who walked on a treadmill without any cognitive task (n = 25). All conditions were matched in exposure time (12 hours). Nine cognitive tasks involved with processing speed, shifting, updating and delayed
memory were administered at baseline, after three months of training, six months follow-up and at 12 months follow-up. After applying Bonferroni correction, a significant time X intervention interaction effect (p = 0.024; small to moderate effect size r = 0.21) was found in the executive control task. Results showed greater improvement scores from three to six months follow-up in the DANCE condition compared with the memory condition. This effect disappeared, however, at 12 months follow-up. With a multitude of tests and measuring moments revealing only one small to moderate effect, the evidence is fragile.
In summary, exergames are an interesting intervention implementing cognitive training and exercise in a stimulating way for older adults. Notably, some studies also report high levels of enjoyableness which might lead to lower levels of attrition, which is a common problem with an older sample. The evidence, however, for being an effective intervention for
countering cognitive decline is lacking. Based on the five studies mentioned above exergames show no to minimal effects on improving cognitive functions.
Cognitive Brain Training
In the literature, a distinction can be made between interventions that are
commercially available and interventions created by researchers. A popular commercially available brain training intervention is Nintendo’s Big Brain Academy. Ackerman et al. (2010) aimed to test the effect of this application on crystallised intelligence, fluid intelligence and speed of processing. As a control, participants were given newspaper and magazine articles of a scientific nature and were subsequently tested on information in these articles.
Participants completed both interventions, with each intervention lasting one month and the order counterbalanced within the sample. Neither the brain training intervention nor the reading intervention had any significant transfer effects.
Another popular brain training application, My Brain Trainer, was investigated by Walton et al. (2015). Participants were randomly assigned to train 12 cognitive tasks offered on the My Brain Trainer website (n = 16) or to train one of the 12 cognitive tasks offered by the same website, training only speed of processing (n = 12). Exposure time (nine hours) in the four-week intervention was matched between training activities. Without a multiple comparison correction, speed and accuracy on a total of four cognitive measures were assessed at baseline and post-intervention. Improved speed response on all tasks was comparable for both interventions. Improved accuracy on two working memory tasks was, however, present in the intervention group but absent in the control group. Since only time interactions are reported (comparing baseline to post-test) and no time X intervention
interactions are reported no direct statistical comparison can be made between conditions. In any case, this study only shows near transfer since the intervention group was trained on multiple tasks related to working memory.
A more specific brain training tool called Virtual Week focusses solely on prospective memory was assessed by Rose et al. (2015). Virtual Week is a game that simulates going through the course of the day and engages the user to make decisions, interact in social events and to remember to perform certain actions at appropriate times. Participants were randomly assigned to the Virtual Week group (n = 23), the music lesson group (n = 14) or the
no-contact group (n = 18). As a control activity, non-musician participants engaged in a music training programme supervised by a teacher. Exposure time (12 hours) was matched between conditions. A novel real-world prospective memory task (the call-back task), various other prospective memory tasks and a measure of daily competence, the Time instrumental activities of daily living (TIADL) (Owsley et al., 2002), were administered at baseline and after four weeks of intervention. Two out of six measures improved significantly (TIADL & the call-back task: ps < 0.05; effect sizes not reported) more in the Virtual Week condition compared to the control conditions. According to the authors, this supports the notion of far transfer. Since Virtual Week entails remembering and performing daily activities at the appropriate time, it trains similar cognitive demands which are measured with the call-back task and the TIADL. Therefore, arguing this study shows evidence for far transfer seems excessive. Furthermore, with no multiple comparison correction, no effect sizes reported and a small sample size, this study does not demonstrate solid evidence for the transfer effects.
The three studies mentioned above were all commercially available brain training applications. Many other studies, however, develop their own brain training software. Van Muijden et al. (2012) carried out a seven-week intervention with custom-made brain training games. One group performed cognitive brain training (n = 53) and the active control group watched documentaries and subsequently answered quiz questions about these
documentaries (n = 19). Cognitive measures of attention, reasoning, inhibition, shifting and updating were measured at baseline and post-intervention. After applying a Bonferroni correction, no significantly differences appear in change scores of any of the untrained tasks regarding executive functions when comparing intervention to control group.
Another intervention study by Bozoki et al. (2013) implemented a six-week intervention programme with the cognitive brain training software My Better Mind, developed by researchers. The control condition required only passive viewing of video/audio news clips and textual news stories. Exposure time (21 hours) was matched between intervention (n = 29) and control condition (n = 21). Comparing baseline - post-test change scores between groups revealed no significant differences between groups on any of the untrained tasks regarding executive functions.
All the studies above pertaining to cognitive brain training may involve sample sizes too small (n < 100) to reveal any effects. A large-scale study by Corbett et al. (2015) developed
a fully online cognitive brain training programme lasting three months with three groups: A ReaCT group, a reasoning and problem-solving cognitive training, a GCT group, general cognitive training and a control group. The control group performed internet-based tasks in which participants needed to put statements in the correct order with the help of searching the internet. Primary outcome for participants over the age of 60 (n = 2,912) was self-reported instrumental activities of daily living (IADL), primary outcome measures for participants between age 50 and 60 (n = 3,830) were tests involved with reasoning, spatial working memory, digit vigilance, verbal short-term visual memory and verbal learning. Outcome measures were measured at baseline, immediate post-intervention at three months and after six months follow-up. The ReaCT and GCT interventions both conferred significantly greater benefit on IADL than the control treatment at 6 months in those older than 60 (ReaCT: p = . 008, Cohen’s d = 0.15; GCT: p = .011, Cohen’s d = 0.16). In adults older than 50, both the ReaCT and GCT interventions conferred significant benefit to reasoning (ReaCT: p < .0001, Cohen’s
d = 0.3; GCT: p < .0001, Cohen’s d = 0.42) and verbal learning (ReaCT: p = .008, Cohen’s d =
0.18; GCT: p = .007, Cohen’s d = 0.19) at 6 months in comparison with controls. The former finding, an improvement in IADL, suggests a far transfer effect and the latter findings, improvement in reasoning and verbal learning suggest near transfer effects. Although the transfer effects of this large-scale RCT are promising, this study also has some limitations. First, due to the online nature of the study, only subjects were eligible if they had daily access to a computer. This could have led to a bias towards subject selection with a higher
education, making it difficult to generalise these results to other populations. Second, with six outcome measures and four measurement moments and no multiple comparison
correction, these results need to be interpreted with caution. In addition, effect sizes are small for far transfer effects and moderate at best for near transfer effects.
Beside the brain training software comprised of mini-games, researched have also investigated more videogame-like interventions targeted at improving cognitive functions. Stern et al. (2011) focussed on Space Fortress, a complex video game originally developed to study complex skill acquisition in young adults (Mané and Donchin, 1989). In the Space
Fortress game, players are required to protect their spaceship by shooting missiles and
destroying a space fortress with a joystick. The game was developed to engage divided attention, multi-tasking, visual scanning, working memory, long-term memory load and
motor control. One group played Space Fortress with certain instructions focussed on
enhancing executive control (n = 17) in the game, one group played Space Fortress freely as an active control (n = 18) and one group was a no-contact control (n = 19). Gameplay time was matched between the gaming conditions. Comparing baseline to immediate post-test, one of the five measures (without a multiple comparison correction) of executive functions assessed improved in the executive function condition (p < 0.05; with a large effect size of ηp2 = 0.14) but not in either control conditions. This near transfer effect, however, disappeared after three months of no gameplay.
Another video game approach to training cognitive functions was undertaken by Anguera et al. (2013). In NeuroRacer, a custom-designed video game, players need to maintain a car in the middle of a winding road (a visuomotor tracking task) while simultaneously responding or inhibiting a response to signs appearing on the side of the road (a perceptual discrimination task). Participants were randomly assigned to the multi-tasking training (n = 16), in which both the drive and sign tasks were performed
simultaneously, or the single-tasking training (n = 15), in which in separate sessions drive and sign tasks were performed, or to a no-contact control group (n = 15). Equivalent difficulty, engagement and exposure time (12 hours) was ensured in both training
conditions. Cognitive outcome tests on attention, updating, dual tasking, working memory and delayed recognition were administered at baseline, four weeks after training and after six months of additional follow-up. Of the 11 outcome measures, multi-tasking training improved significantly more (p< 0.05; with a medium to large effect sizes Cohen’s d = 0.54-0.67) in three (two working memory and one sustained attention task) compared to single-task training. According to the authors, working memory and sustained attention loads are equal between training conditions, implying far transfer effects. Pre-test data is missing, however, making it unclear whether groups were matched at baseline. Moreover, without multiple comparison correction and a small sample size the evidence is weakened. Although the study implements a solid active control condition, it provides little compelling evidence of transfer effects.
It is difficult to identify the cause behind the contrasting results due to the wide variety of outcome measures applied. A major improvement would, therefore, be to use identical outcome measures with proper multiple comparison corrections.
Based on these studies, commercial brain training software lacks empirical evidence for improving overall cognitive performance, as it often claims. It appears that more specific brain training software such as Virtual Week, aimed to improve one specific cognitive
function, has more potential. At best, these specific brain training applications might generate small to moderate near transfer effects.
The brain training software developed by researchers both led to null-findings and moderate near and far transfer effects. Similar to the commercially developed games, it appears hard to effectuate transfer effects with an intervention aimed at a range of cognitive functions. Even with thousands of participants, the effect size is still small to moderate for near and far transfer effects. On the other hand, the two studies implementing brain training games which require more specific cognitive demands show opposite effects. An explanation for these opposite effects might lie in the active control condition. In the Space Fortress study, participants in the intervention condition were given instructions to pay attention to different sub-goals of the game separately, whilst the control condition could play the game freely. The boundless control condition might have led to unintended similarities with the intervention condition, obscuring effects of the intervention. In the NeuroRacer study, however, the difference between the intervention and active control condition was more controlled, with a more restrained active control condition. Even with a well-matched control condition though, the transfer effects are little compelling due to methodological weaknesses of the study.
Taken together, evidence of brain training programmes developed by researchers shows more potential than commercialised programmes. Improvement on only a minority of cognitive measures tested and methodological impairments, however, lead to little
compelling evidence for cognitive brain training as a successful intervention to counter cognitive decline in older adults.
Video Game Interventions
Video games have gained increased attention as an intervention for countering cognitive decline in older adults. Video games seem a promising cost-effective treatment: they are widely available, usually have low costs and are intended to be entertaining. In contrast to the studies discussed in the previous section, the studies here involve video games developed by commercial parties and do not serve an educational purpose.
RCTs were extracted from reviews by Mansor et al. (2019) and Toril et al. (2014). A total of 46 studies from these reviews were examined. After removing 13 duplicates, 13 studies without an active control condition, 12 studies solely focussed on cognitive brain training games and one study without random participant allocation to groups, seven studies were left to examine. Of these seven studies, six involve a comparison between cognitive brain training and video game training.
Individual Studies
Dustman et al. (1992) was one of the first studies assessing the effect of video game play on cognitive functions in older adults. The study comprised three groups: a video game group (n = 20), a movie-watching group (n = 20) and a no-contact control group (n = 20). The video game group were free to choose among dozen available games, most played were
Breakout, Ms. Pac-Man, Galaxian and Frogger. During the 11-week intervention, exposure time
(33 hours) in the video game and movie watching group was matched. Participants were cognitively assessed at baseline and post-intervention with an assessment battery focussed on sustained attention, visuomotor coordination and tracking, shifting, finger agility, visual memory, immediate verbal memory and complex RT. Results showed a significant time X group interaction effect (p < 0.012; effect size could not be extracted from the article) with an improvement in one of the nine cognitive tasks. Performance in the complex RT task was improved in the video game group but not in the movie-watching group. Considering the video games involve training in processing speed, this finding constitutes a near transfer effect. With a total of nine statistical tests and no multiple comparison correction, however, this finding should be interpreted with caution.
Nouchi et al. (2012) also investigated a classical computer game, Tetris, but now in comparison with cognitive brain training. Subjects either played the popular cognitive brain
training game Brain Age (n = 14) or Tetris (n = 14), both developed by Nintendo. Exposure time between Brain Age and Tetris group was matched (6 hours). At baseline and
post-intervention, global cognitive status, executive function, attention and speed of processing was assessed. Results showed a significant time X intervention effect for four out of eight tests measured. A multiple comparison correction was not applied. For all measures of executive function and two measures of speed of processing, improvement in only the Brain
Age group (ps < 0.05; with medium to large effect size ηp2 = 0.12-0.19) was present. Since both
speed of processing and executive function was trained with the Brain Age games, this constitutes a near transfer effect.
Peretz et al. (2011) compared cognitive brain training (CogniFit Personal Coach) with an array of classical computer games. Games included Tetris, Snake, playing tennis, memory games and arithmetic games. Exposure time (16 hours) was matched between the video game (n = 55) and the personalised cognitive training (n = 66). At baseline and after 12 weeks of intervention, the NexAde cognitive test battery of six tasks was administered. There was a significant (p <0.0019) improvement in the cognitive training group as compared to the computer games group in 3 out of 9 cognitive domains: visual-spatial working memory, visual-spatial learning and focused attention (Cohen’s d = 0.43, 0.51 and 0.63 respectively). Notably, no multiple comparison correction was performed, increasing the chance of a type I error. All three cognitive domains were trained with the CogniFit training, making these findings near transfer effects.
Beside these rather simplistic mini-games, more complex games were also
investigated such as the well-known Super Mario. Perrot et al. (2018) compared cognitive improvement in participants who played Kawashima Brain Training n = 12) with participants who played Super Mario Bros freely (n = 12) and with a no-contact control group (n = 11).
Kawashima Brain Training is specifically designed to improve cognitive functions, especially
in older adults, with focussed mini games demanding inhibition, working memory and spatial orientation. Super Mario Bros is a platform game which involves a player controlling a character who runs, ducks and jumps to attain coins or boosts, avoid danger and in this way move to the next level. Exposure time was matched between conditions (24 hours). Cognitive measures of speed of processing, executive functions and visuospatial abilities were
the brain training software, the mean brain age of the cohort improved from 58.6 years at the beginning to 34.2 years at the end of the training. In comparing brain training directly with video gaming only a significant difference on one of seven cognitive measures was revealed. Brain training led to greater improvement on the Stroop test, reflecting improved response inhibition (p < 0.05; effect size could not be extracted from the article). Since response inhibition was trained in the brain training program, this constitutes a near transfer effect. However, with seven tests and no multiple comparison correction, this finding should be interpreted with caution. A similar study by Boot et al. (2013) compared MarioKart DS, the racing game from the same enterprise, with another popular cognitive brain training game
Brain Fitness. Brain Fitness also consists of several mini games focussed on training memory,
reaction time, language skills and mathematical ability. Training lasted 12 weeks and
exposure time was not matched between racing game group (n = 14; 56 hours) and cognitive brain training group (n = 21; 22 hours). Results showed no positive effect on any of the cognitive measures assessed (for either gaming intervention compared to a no-contact control group (n = 20). Notably, enjoyableness and compliance ratings for the racing game intervention was much lower than for the cognitive brain training, which might have affected the performance in the racing game.
Another racing game, Crazy Taxi, was also compared to cognitive brain training software InSight (Belchior et al., 2019). Crazy Taxi, is a driving game with key features that include rapid navigation through an urban environment, attending to speed and roadway features. It is thought to elevate perceptual, cognitive and motor loads. InSight consists of five tasks aimed at improving visual and cognitive performance, which increased in difficulty depending on the participants progress. Exposure time (60 hours) was matched between the
Crazy Taxi group (n = 17) and the InSight group (n = 19) during the 12-week intervention.
Several versions of The Useful Field of View test, measures of speed of processing, divided and selective attention, were administered at baseline, immediate post-intervention and at three months follow-up. Results revealed significant differences on two out of nine cognitive measures tested, with greater improvement in visual attention and speed of processing in the
Insight group compared to the Crazy Taxi group (ps < 0.05; Cohen’s d = 0.5 and 0.8), both at
immediate post-intervention and three months follow-up. Visual attention and speed of processing were trained with InSight, so a near transfer effect is present. Importantly, as
noted by the authors, these results need to be interpreted with caution since the sample size was small and no multiple comparison correction was applied.
A final study involving video games investigated two types of video games: the classic strategic game Tetris and the fast-paced action game Medal of Honor (Belchior et al., 2013). Medal of Honor is a first-person shooter game involving missions with multiple objectives such as reaching a target behind enemy lines or freeing hostages from enemy hands. Two additional conditions consisted of performing a Useful Field of View (UFOV) training program or a no-contact control condition. UFOV training is a customized attention training programme using a computer touch screen. Exposure time (nine hours) was
matched for the Medal of Honor (n = 14), Tetris (n = 15) and UFOV group (n = 16). At baseline and after three weeks of intervention the UFOV test, measuring speed of processing, selective and divided attention, was administered. Comparing the three interventions revealed no differences in transfer effects on any of the assessed tasks.
Conclusion
Similar to the studies on cognitive brain training, the studies implementing a video game intervention are heterogeneous. Video games range from simple arcade games to fast-paced first-person shooter games. Moreover, many outcome measures are tested without a proper multiple comparison correction. As with studies on cognitive brain training, using identical cognitive measures with proper multiple comparison corrections would aid cohesion of results considerably.
Most of the studies examined compared cognitive brain training directly to
commercial video games. Two studies show no transfer effects to cognitive tasks for either intervention. The remaining four studies show greater improvements in untrained cognitive tasks after cognitive brain training compared to video game playing. Further examination of the evidence in favour of cognitive brain training studies bears little persuasion. In most studies only a fraction of the cognitive tasks administered reveal significant improvements. Moreover, small sample sizes and absence of multiple comparison corrections further weaken the evidence. Far transfer effects are unequivocally absent for either cognitive brain training or video game playing and near transfer effects are minimal to moderate at best for cognitive brain training.
In conclusion, evidence for video games as an effective way in countering cognitive decline in older adults is lacking. An explanation for this might be that video games are not developed for older adults. Most video games, especially racing and action games, are fast-paced, complex and sometimes without clear instructions (e.g. one can run freely in the
Super Mario game). As a complete opposite, cognitive brain training software is especially
developed for older adults. Most brain training software consists of mini games with explicit instructions and training for each game. Since the complex and boundless character of action video games leads to lower compliance (Boot et al., 2013), it is doubtful whether this
Leisure Activity Interventions
Leisure activities are defined as activities that individuals engage in for enjoyment or well-being beside their routine activities of daily living. Leisure activities seem a promising intervention for countering cognitive decline because it includes elements of physical and intellectual activities, social exchanges and are intended to bring enjoyment. The range of leisure activities in the literature is broad. This review will leave out leisure activities involving solely physical exercise, dietary changes or cognitive training programmes.
RCTs were extracted from reviews by Iizuka et al. (2019), Klimova et al. (2017) and Wanchai and Phrompayak (2019). A total of 39 studies were examined. After removing one duplicate, nine studies with participants with cognitive decline, seven studies involving only physical exercise, two studies involving a dietary intervention, one study involving training on cognitive tasks, eight studies without an active control group, four studies without random participant allocation to groups, one study without cognitive outcome measures and one study of which the full text was not available, five studies were left to examine in detail.
Individual Studies
Crossword puzzles are a leisure activity often performed by older adults. Murphy et al. (2014) carried out an RCT with two participants groups: one group did a crossword puzzle on a daily basis (n = 19) and one group kept a gratitude diary in which they daily recorded three things they felt grateful for (n = 18). A significant interaction between
condition and time of test was revealed (p = 0.04; moderate effect size reported without exact data). Performing daily crossword puzzles led to greater improvement in verbal fluency compared to keeping a gratitude diary. In a verbal fluency task, participants need to name as much words as possible starting with a certain letter. It is, therefore, sensible to assume this finding resembles a near transfer effect. Although the intervention was externally valid, no measures were taken to ensure compliance in either intervention groups. To tackle this, many other studies implement an intervention programme with supervised meetings.
Suzuki et al. (2014), for instance, implemented a 12-week training programme for picture book reading. Picture book reading to children is expected to enhance memory and cognitive functions, since it involves multi-tasking where one needs to memorise the tale
while showing pictures. The active control group participated in monthly lectures about elderly health maintenance. The picture book reading group (n = 29) participated in weekly sessions of two hours whilst the lecture group (n = 29) participated in monthly lectures of which the duration is not reported, indicating unmatched exposure time. Only two out of 13 cognitive tasks showed a significant interaction effect in favour of the picture book reading group, both being measures of logical memory (LM II: p = 0.011; Cohen’s d = 0.103, ∆LM: p = 0.034; Cohen’s d = 0.220). With no multiple comparison correction performed and small effect sizes, this study demonstrates weak near transfer effects.
Another leisure activity programme was implemented by Medeiros et al. (2011). Participants were randomly assigned to an autobiographical writing workshop (n = 18), an active control group involved with oral reminiscing (n = 18) or to a no-contact control group (n = 15). Unlike oral reminiscing in which recall happens ‘in the moment’, autobiographical writing requires more demanding reconstruction of past experiences. Exposure time (12 hours) was matched between conditions. Autobiographical memory, new verbal and visuospatial learning and memory aspects were assessed at baseline, immediate post
intervention (eight weeks after baseline) and at long-term follow up (34 weeks after baseline). No significant time X group interactions were found for any of the cognitive measures assessed.
Perhaps an exposure time of 12 hours spread out over eight weeks is too little to bear any cognitive enhancements. Klusmann et al. (2010) carried out a more extensive (exposure time = ~110 hours) and longer lasting (25 weeks) intervention programme. Participants were assigned to a computer course (n = 81), an aerobic exercise programme (n = 80) or a no-contact control group (n = 69). The computer course covered creative matters as well as coordinative and memory assignments. At baseline and after six months of intervention, cognitive tasks focussed on episodic memory, working memory and executive functions were administered. After a Bonferroni correction had been applied, none out of seven cognitive tests showed a differential improvement in comparing the computer course with the exercise condition.
A final study involving leisure activities by Park et al. (2013) involved a sophisticated and robust study design. The study consisted of six conditions including three intervention, productive engagement, conditions and three control, receptive engagement, conditions. The
productive engagement conditions comprised a digital photography class (n = 29), a quilting class (n = 35) or a dual-condition (n = 42) involving both quilting and digital photography. The three receptive engagement conditions comprised a social condition (n = 36), a placebo condition (n = 39) and a no treatment condition (n = 40). In the social condition subjects participated in social activities without active skill acquisition (e.g. cooking, watching movies/sports, going on field trips together). In the placebo condition, subjects engaged in solitary activities relying on existing knowledge or activities that have not been reliable linked by empirical evidence to cognitive improvement but are commonly thought of as cognitively engaging (e.g. documentaries, informative magazines such as National Geographic, puzzles and classical music CDs). In the no treatment condition participants were required to keep a checklist of activities. All participant groups (except no treatment) spend similar time on their activities, with about 15-18 hours per week for 14 weeks, totalling to approximately 210 hours of exposure time. To ensure participants performed an activity of some interest to them, participant allocation was not completely random since participants could exclude one of the productive engagement conditions. At baseline and at immediate post-intervention cognitive tasks were administered which focussed on speed of processing, mental control, episodic memory and visuospatial processing. Bonferroni corrections were applied
adequately. A significant condition X time interaction effect was observed for one of the four cognitive tests: episodic memory. The productive engagement conditions improved
significantly more over time (p = 0.002) compared to the receptive engagement conditions. There was no significant difference within the productive engagement conditions nor within the receptive engagement conditions. Productive engagement was also compared with the social condition, revealing greater improvement in episodic memory in productive
engagement groups compared to social activities (p = 0.04). Productive engagement conditions were also separately compared with placebo condition, revealing greater
improvement in episodic memory in the photo condition compared to receptive cognitively engaging activities (p = 0.01; net effect size = 0.54, for details on effect size calculations see Park et al. (2013)). Although improvements in episodic memory are rarely reported in cognitive brain training studies, it was expected here due to the nature of the digital photography class. The class was, according to the authors, particularly demanding of
episodic memory given that participants had to remember many complex instructions to use the software and the camera.
Conclusion
Taken together, the studies discussed present a wide variety of leisure activity interventions. Perhaps the most externally valid of all, doing a daily crossword puzzle, showed near transfer effects of a moderate effect size. Since neither compliance of
performing the puzzles nor compliance of not performing puzzles in the control condition was ensured, the study design is not robust. More controlled study designs implemented group activities such as picture book reading, autobiographical writing or following a computer course. As the first demonstrated small near transfer effects with the lack of a multiple comparison correction and the second and third demonstrated no differences between interventions applied, the evidence is not convincing.
The most sophisticated and insightful study design has been implemented by Park et al. (2013). Based on their results, it can be concluded that productive engagement leads to more cognitive improvement than receptive engagement. More importantly, productive engagement activities might be more facilitative of cognitive enhancement than socialising alone. To be explicit, social activities might enhance cognitive functions, but productive engaging activities appear to facilitate greater improvements. Notably, only the digital photography intervention led to cognitive improvements. It is, therefore, excessive to state productive engagement activities are superior to social activities alone, since this only pertains to a digital photography class.
Based on the examined studies it can be concluded that leisure activities show little evidence for bearing cognitive enhancements. For a leisure activity to bear cognitive enhancements, a long-lasting and time-intensive programme with productive engaging activities shows most potential.
Volunteering Interventions
Volunteering is defined as unpaid, non-compulsory work done through an organisation and for the benefit of people outside the person’s household (Guiney and Machado, 2017). Volunteering seems a viable option for countering cognitive decline since it can be implemented on a large scale and is in its essence cost-effective since there are no to minimal costs involved. Furthermore, volunteering can entail a wide spectrum of activities that will interest a large fraction of older population.
Individual studies have been extracted from several reviews (Anderson et al., 2014; Cattan et al., 2011; Guiney and Machado, 2017; Jenkinson et al., 2013; von Bonsdorff and Rantanen, 2011). A total of 141 studies have been examined and up to recently, to our knowledge, there is only one RCT that focusses on the effect of volunteering on improving cognitive functions in healthy older adults. The RCT called Experience Corps does not, however, implement an active control condition. Also, participants in the Experience Corps programme were reimbursed with $150-$200 per month to compensate for travel and food expenses, conflicting with the definition of volunteer work. Despite these issues, the
Experience Corps is the only RCT and gives valuable insights in volunteering as an
intervention to counter cognitive decline. The intervention trial design from which four individual studies emerged will therefore be examined in detail, followed by the individual results from the studies.
Experience Corps
In Baltimore, Maryland (USA), a programme places older adults in public elementary schools in a project to have high impact on the academic outcomes of the children in
kindergarten to third grade. The programme called Experience Corps has been ongoing since the 90s and involves over 2,000 volunteers training 30,000 students in more than 20 cities (American Association of Retired Persons, n.d.). The programme is designed to increase the volunteer’s physical, social and cognitive activity simultaneously, ultimately leading to health benefits. All volunteers are trained on three modules. First, supporting children in literacy development by reading to and with children. A tool book helps volunteers with assessing children’s reading levels, building children’s vocabulary and asking questions about the book. Second, volunteers are trained to support in library functions. This includes
shelving or cataloguing books, helping at school libraries and helping children pick books. Third, volunteers are trained to teach children in conflict resolution. In a supervised recess programme, volunteers played a variety of both quiet group activity games and board games with the children. The hypothesis was that a programme designed to improve children’s success in school would attract a diverse population of older adults including adults who might otherwise not engage in programmes aimed at their own health promotion. For a full description of the Experience Corps programme, see Fried et al. (2004). In three out of four studies, volunteers were randomly assigned to the Experience Corps programme to help for a full academic year for 15 hours per week or to a wait-list control group which was eligible to participate in Experience Corps after the study ended. The most recent study by Carlson et al. (2015) had a slightly different control group. Those randomised to the control group were, in addition to being placed on the waitlist, approached for volunteer work of short duration and of low time demands, including helping at health fairs, city festivals and senior health centres. Sample size, time between baseline and follow-up measurements and outcome measures differs between studies.
The first study reporting results from the Experience Corps trial was by Fried et al. (2004). Intervention group (n = 69) and no-contact control group (n = 56) were tested at baseline and at four to eight-month follow-up. The difference in follow-up length is because volunteers started in either November, January or March and were all tested at the end of the academic year in June. With a 15 hour per week intensity, this amounts to approximately 240-480 hours of exposure time. Outcome measures entailed amount of engagement in physical, social and cognitive activities outside the program, data were gathered with self- and interviewer-administered questionnaires. Cognitive activities were divided in high (e.g. crossword puzzles), moderate (e.g. cooking) and low (e.g. television viewing) cognitive intensity activities. Results revealed no cognitive measure to increase more in the intervention than in the control condition.
Although, no cognitive enhancements are found, the unconventional and
self-reported outcome measures of cognitive performance might be unable to detect any transfer effects. Therefore, the second study on the Experience Corps by Carlson et al. (2008) involved well-validated cognitive tests. Again, the study involved an intervention group (n = 62) and a no-contact control group (n = 48) that were tested at baseline and four to eight-month
follow-up with an exposure time of (240-480 hours). Cognitive tests involved measures of executive functions, verbal and visuospatial memory and psychomotor speed. The time X group interaction effect was non-significant for all tests, indicating no transfer effects.
The third study emerging from the Experience Corps trial also investigated the effects on executive functions, now with additional neuroimaging data that will not be examined in this review. Carlson et al. (2009) involved an intervention group (n = 8) and a no-contact control group (n = 9) with tests administered at baseline and six months follow-up,
indicating an exposure time of 360 hours. Executive function was measured with the Flanker task. A significant time X group interaction effect was found, with a greater improvement in the intervention group compared to the control group (p < 0.04; effect size not reported). With no multiple comparison correction applied and a small sample size, these results need to be interpreted with caution.
Lastly, Carlson et al. (2015) focussed on potential verbal memory improvements, again with additional neuroimaging data that will not be examined in this review. As
mentioned before, compared to the previous three studies, this study involved a more active control condition. This time, participants assigned to the control condition were referred to the Baltimore City Commission for other low-activity volunteer opportunities. If interested, participants could volunteer at a selection of low-time demanding events. There is, however, no data reported on compliance or hours spent on these low-activity volunteering events. Verbal memory was assessed by the Rey Auditory Verbal Learning Test (RAVLT) at baseline and at annual follow-ups during the 2-year trial exposure (exposure time = 960 hours). Unfortunately, raw data or tests on the verbal memory scores alone are not reported. Memory changes are only reported in relation to hippocampal volume changes from
baseline to two-year follow-up. Here, a significant positive correlation between hippocampal volume changes and verbal memory is present in the intervention group (r = 0.39; p = 0.02) and absent in the control group. Without data on the verbal memory test alone, it is not possible to assess whether cognitive improvement was significantly greater in the intervention group compared to the control group.
Conclusion
Evidence for cognitive enhancement in older adults as a result of volunteering is minimal to modest at best. Of the four studies examined, two present no differential effects on cognitive improvement between control and intervention group, one presents a
significantly greater improvement in executive functions in a small sample and one presents a positive correlation between hippocampal volume changes and verbal memory
improvement but lacks data on cognitive improvement alone.
Moreover, one must consider that all these studies implemented a completely to mostly passive control condition. In the most recent study by Carlson et al. (2015), the control group could partake in volunteering activities of low time demands. It is, however,
extremely unlikely the exposure time of 15 hours per week for two academic years was matched to a control condition with activities of low time demands.
For now, the evidence in favour of volunteering as an effective intervention to counter cognitive decline is lacking. RCTs with active control conditions with matched exposure time are necessary to further determine the potential of volunteering in countering cognitive decline.
Discussion
The current review aimed to assess the efficacy of non-pharmacological interventions on countering cognitive decline in healthy older adults by carefully examining
methodologically robust studies. Studies concerning cognitive brain training, video game interventions, leisure activity interventions and volunteering interventions were examined in detail. Apart from volunteering interventions, only studies involving RCTs with an active control condition were assessed. Efficacy of interventions were based on near and far transfer effects to untrained tasks involving cognitive performance. Overall, evidence for convincing far transfer effects is absent in all interventions. Near transfer effects are present in cognitive brain training interventions and leisure activity interventions. Video game interventions show least potential and for volunteering interventions it is too soon to accurately estimate its potential as an intervention countering cognitive decline.
For a complete overview of the individual RCTs examined, see appendix I. In total 17 out of 29 studies show any transfer effect, with two showing both far and near transfer effects and the other 15 studies showing only near transfer effects. Looking at interventions separately, it shows transfer effects in 12 out of 19 studies in favour of cognitive brain training, one out of seven in favour of video games, three out of five in favour of leisure activities and one out of four in favour of volunteering. Note the total number of studies here does not amount to 29 since six studies both involved video game and cognitive brain training interventions. The proportion of studies reporting transfer effects might appear promising but are subject to serious methodological issues. Despite systematically selecting the most methodologically robust studies from recent reviews, methodological weaknesses were encountered often. In the 17 studies showing transfer effects, a total of 128 cognitive test measures have been assessed, of which only 33 showed any transfer effects. With these large number of tests, it is problematic that only one study applied a multiple comparison
correction. Moreover, seven studies failed to report effect sizes or important data needed to assess effect size, further impairing the strength of evidence in favour of these interventions countering cognitive decline. Although statistical significance is of great importance to determine whether there is an effect of the intervention, the magnitude of an effect is even more crucial to assess clinical relevancy for any of the interventions. Of the 11 studies reporting effect sizes for any transfer effects, five studies report small effect sizes, two report
moderate effect sizes and three report moderate to large effect sizes. Notably, of the studies reporting moderate or large effect sizes, one study lacked baseline data, one study reported fading of the effect in follow-up data and all studies lacked a multiple comparison correction.
Despite these methodological issues, some interesting insights can be gained by examining these studies. As expected with the heterogeneity in interventions applied, the exposure time and duration of the intervention does not seem to be a clear predictor of leading to transfer effects. Exposure times as little as six hours or intervention duration as short as four weeks can lead to transfer effects. Regarding the outcome measures assessed, there is also no apparent cognitive domain or test that is more susceptible to transfer than others. Cognitive transfer effects of varying effect sizes have been reported in cognitive tasks measuring executive functions (response inhibition, shifting, updating, working memory, attention), speed of processing, logical memory, episodic memory, prospective memory, daily competence and verbal fluency.
Since aging comes with cognitive decline across a variety of cognitive domains (Harada et al. 2013), interventions countering cognitive decline are also targeted at multiple cognitive domains, this pertains especially to commercial brain training programmes. Although these brain training programmes usually consist of several mini games intended to target a subset of cognitive domains, the approach of training many cognitive domains within one intervention appears ineffective. Cognitive brain training programmes focussed on training one cognitive domain show more potential in countering cognitive decline. Perhaps the inability to train a wide range of cognitive domains simultaneously also impairs the efficacy of video game interventions. These video games, especially action or racing games, demand a wide variety of simultaneous cognitive processes which might not be as beneficial as training one cognitive domain at the time. It is, therefore, advisable to advance research focussed on cognitive brain training applications on training and testing specific cognitive domains. The somewhat explorative nature of training and subsequently testing a wide a variety of cognitive outcome measures bears unreliable transfer effects and hinders the methodological strength of the evidence found. As argued by Simons et al. (2016),
improvement on speed of processing appears most likely to transfer to other domains since it is a rate-limiting factor in memory and problem-solving processes.
As for the other types of interventions involving leisure activities and volunteering, these entail a more indirect way of improving cognitive domains. Due to the indirect nature of these interventions, rigid training of one specific cognitive domain is hard to accomplish. Advantages of these interventions, however, also lie in the differences with the cognitive brain training or video game interventions. In contrast to volunteering and leisure activities, cognitive brain training and video game intervention studies sometimes suffer from low compliance and high attrition rates (Ackerman et al., 2010; Boot et al., 2013) in elderly samples due to the technical nature of these interventions. In addition, leisure activities and volunteering offer a wider spectrum of activities that generally include a larger social component. Perhaps the most intricate and insightful study design implemented by Park et al. (2013), addressed this social component specifically. The design consisted of a separate control condition with social activities without a cognitively engaging component and a control condition with solitary activities with cognitively engaging activities. Park et al. (2013) showed the combination of social activity and cognitive engagement in a digital photography class leads to more cognitive improvement than either only social or only cognitively engaging activities. This is an important finding since most active control conditions are implemented to control for the social or motivational component of the intervention applied, whether it is cognitive brain training, playing video games or a leisure activity. Unfortunately, the weight of either social or cognitive components to cognitive improvements could not be extracted from this study design. Future research could implement an identical leisure activity in both groups but with the difference of doing this activity solo or in groups to determine the separate influence of social stimulation on cognitive improvement.
Although the study by Park et al. (2013) involves an intricate study design, it does involve a common methodological struggle in studies regarding leisure activities and interventions. In their study, participants could exclude one of the offered intervention activities. The struggle for researchers here is choosing between a robust randomised controlled trial or an externally more valid approach by letting participants choose an intervention they are interested in. Although Park et al. (2013) is the only study examined in this review with this issue, the prevalence of the issue is underestimated here since this review mostly selected RCTs. Although it might be unwanted to have subjects participate in
an activity they would not choose themselves, there are adverse consequences to letting participants influence their intervention activity. A possible consequence is that participants show greater cognitive improvements due to inherent motivation or enthusiasm for the followed intervention and not due to the intervention itself. Since the amount of RCTs is sparse in interventions focussed on leisure activities and especially volunteer work,
researchers need to establish the potential of these interventions with a fully random study design.
In conclusion, considerable methodological advances need to be taken in all interventions countering cognitive decline. As of now, the methodological issues are too frequent and serious to demonstrate compelling evidence for any intervention to successfully counter cognitive decline in older adults. Still, pursuing research in the interventions
presented in this review is worthwhile, albeit solely for the potentially superior cost-effectiveness of the interventions compared to drug-based interventions. Cognitive brain training applications can be distributed to large populations with low costs. Leisure activity and volunteering programmes are relatively small-scale but have to potential to interest a much larger population with a wide spectrum of more socially engaging activities. With the alarming rise in dementia patients, the need for a cost-effective intervention is crucial. For these interventions to be implemented on a large-scale though, methodologically robust research is absolutely necessary to reliably demonstrate countering of cognitive decline.
(Ackerman et al., 2010; Chuang et al., 2015; Corbett et al., 2015; Eggenberger et al., 2015, 2016; Kayama et al., 2014; Schättin et al., 2016; Schoene et al., 2015; Van Muijden et al., 2012; Walton et al., 2015) (Bozoki et al., 2013; Rose et al., 2015) (Anguera et al., 2013; Stern et al., 2011) (Belchior et al., 2013, 2019; Boot et al., 2013) (Dustman et al.,
1992; Nouchi et al., 2012; Peretz et al., 2011; Perrot et al., 2019) (Murphy et al., 2014) (De Medeiros et al., 2011; Klusmann et al., 2010; Park et al., 2014; Suzuki et al., 2014) (Carlson et al., 2008, 2009, 2015; Fried et al., 2004)
