Computerised cognitive training for preventing dementia in people with mild cognitive impairment


The terms 'cognition' and 'cognitive function' describe all of the mental activities related to thinking, learning, remembering, and communicating. There are normal changes in cognition with age, There are also diseases that affect cognition, principally dementia, in which cognition is impaired to the point of affecting a person's ability to manage daily activities. More common than dementia is a condition often described as mild cognitive impairment (MCI), in which mild impairment of cognition, more than expected from age alone, can be detected on testing, but by which daily functioning is largely unaffected. For some people, MCI is a stage on the way to developing dementia. There is a lot of interest in anything that might prevent further decline in cognition in people with MCI. One thing that has been suggested as a means of doing this is computerised cognitive training (CCT). Cognitive training consists of a set of standardised tasks intended to 'exercise the brain' in various ways. These days, cognitive training exercises are often delivered via computers or mobile technology, so that people can do them on their own at home. We wanted to know whether CCT is an effective way for people with MCI to maintain their cognitive function and reduce their risk of going on to develop dementia.

What we did

We searched the medical literature up to 15 March 2018 for trials in which a group of people with MCI had participated in CCT for at least 12 weeks and had been compared with another group that had not received any CCT. This 'control' group could have taken part in an alternative activity instead, or group members could have received no intervention at all. For the comparison to be as fair as possible, it should have been decided at random whether people were in the CCT or control group. We were primarily interested in whether study participants developed dementia and in their overall cognitive function, but we also looked for evidence on particular cognitive skills, daily activities, quality of life, mood, or mental well-being, and any harmful effects.

What we found

We found eight trials with 660 participants to include in the review. Seven of the trials (623 participants) compared CCT to an alternative activity. None of the included trials examined development of dementia, so this review presents no evidence on whether taking part in computerised cognitive training will help to prevent dementia. Our main finding in relation to all of the other outcomes in which we were interested was that the overall quality of the evidence was very low. This very low quality was mainly due to small sample sizes, problems with study methods, and differences between trials. Therefore, although we found some evidence for a few benefits of CCT for cognition, we were highly uncertain about study results and consider it likely that future research might lead to different results.

Our conclusions

Unfortunately, it is not yet possible to answer our review question with any certainty. We think it remains an important area for further study. We would like to see larger studies, which would be more able to detect effects of CCT, and longer studies, which are needed to show whether there are any benefits, whether benefits are long-lasting, and whether there is a chance of preventing or delaying the development of dementia.

Authors' conclusions: 

Currently available evidence does not allow us to determine whether or not computerised cognitive training will prevent clinical dementia or improve or maintain cognitive function in those who already have evidence of cognitive impairment. Small numbers of trials, small samples, risk of bias, inconsistency between trials, and highly imprecise results mean that it is not possible to derive any implications for clinical practice, despite some observed large effect sizes from individual studies. Direct adverse events are unlikely to occur, although the time and sometimes the money involved in computerised cognitive training programmes may represent significant burdens. Further research is necessary and should concentrate on improving methodological rigour, selecting suitable outcomes measures, and assessing generalisability and persistence of any effects. Trials with long-term follow-up are needed to determine the potential of this intervention to reduce the risk of dementia.

Read the full abstract...

The number of people living with dementia is increasing rapidly. Clinical dementia does not develop suddenly, but rather is preceded by a period of cognitive decline beyond normal age-related change. People at this intermediate stage between normal cognitive function and clinical dementia are often described as having mild cognitive impairment (MCI). Considerable research and clinical efforts have been directed toward finding disease-modifying interventions that may prevent or delay progression from MCI to clinical dementia.


To evaluate the effects of at least 12 weeks of computerised cognitive training (CCT) on maintaining or improving cognitive function and preventing dementia in people with mild cognitive impairment.

Search strategy: 

We searched to 31 May 2018 in ALOIS ( and ran additional searches in MEDLINE, Embase, PsycINFO, CINAHL,, and the WHO portal/ICTRP ( to identify published, unpublished, and ongoing trials.

Selection criteria: 

We included randomised controlled trials (RCTs) and quasi-RCTs in which cognitive training via interactive computerised technology was compared with an active or inactive control intervention. Experimental computerised cognitive training (CCT) interventions had to adhere to the following criteria: minimum intervention duration of 12 weeks; any form of interactive computerised cognitive training, including computer exercises, computer games, mobile devices, gaming console, and virtual reality. Participants were adults with a diagnosis of mild cognitive impairment (MCI) or mild neurocognitive disorder (MND), or otherwise at high risk of cognitive decline.

Data collection and analysis: 

Two review authors independently extracted data and assessed risk of bias of the included RCTs. We expressed treatment effects as mean differences (MDs) or standardised mean differences (SMDs) for continuous outcomes and as risk ratios (RRs) for dichotomous outcomes. We used the GRADE approach to describe the overall quality of evidence for each outcome.

Main results: 

Eight RCTs with a total of 660 participants met review inclusion criteria. Duration of the included trials varied from 12 weeks to 18 months. Only one trial used an inactive control. Most studies were at unclear or high risk of bias in several domains. Overall, our ability to draw conclusions was hampered by very low-quality evidence. Almost all results were very imprecise; there were also problems related to risk of bias, inconsistency between trials, and indirectness of the evidence.

No trial provided data on incident dementia. For comparisons of CCT with both active and inactive controls, the quality of evidence on our other primary outcome of global cognitive function immediately after the intervention period was very low. Therefore, we were unable to draw any conclusions about this outcome.

Due to very low quality of evidence, we were also unable to determine whether there was any effect of CCT compared to active control on our secondary outcomes of episodic memory, working memory, executive function, depression, functional performance, and mortality. We found low-quality evidence suggesting that there is probably no effect on speed of processing (SMD 0.20, 95% confidence interval (CI) -0.16 to 0.56; 2 studies; 119 participants), verbal fluency (SMD -0.16, 95% CI -0.76 to 0.44; 3 studies; 150 participants), or quality of life (mean difference (MD) 0.40, 95% CI -1.85 to 2.65; 1 study; 19 participants).

When CCT was compared with inactive control, we obtained data on five secondary outcomes, including episodic memory, executive function, verbal fluency, depression, and functional performance. We found very low-quality evidence; therefore, we were unable to draw any conclusions about these outcomes.