REGULAR RESEARCH ARTICLES

Can Cognitive Exercise Prevent the Onset

of Dementia? Systematic Review of
Randomized Clinical Trials with
Longitudinal Follow-up
Michael Valenzuela, B.Sc. Hons., M.B.B.S. Hons., Ph.D.,
Perminder Sachdev, M.D., Ph.D., F.R.A.N.Z.C.P.

Objectives: Epidemiological and preclinical studies suggest that mental activity levels
may alter dementia risk. Clinical trials are now beginning to address the key issues of
persistence of effect over extended follow-up and transfer of effect to nontrained domains.
The aim of this report was to therefore systematically review results from clinical trials,
which have examined the effect of cognitive exercise on longitudinal cognitive perfor-
mance in healthy elderly individuals. Methods: MEDLINE, PubMed, and key references
were used to generate an initial list of relevant studies (N ⫽ 54). These were reviewed to
identify randomized controlled trials, which tested the effect of a discrete cognitive
exercise training regime on longitudinal (⬎3 months) posttraining neuropsychological
performance in healthy older adults. Seven RCTs met entry criteria. Prechange and
postchange scores were integrated using a random effects weighted mean difference
(WMD) meta-analytic approach (Review Manager Version 4.2). Results: A strong effect
size was observed for cognitive exercise interventions compared with wait-and-see con-
trol conditions (WMD ⫽ 1.07, CI: 0.32–1.83, z ⫽ 2.78, N ⫽ 7, p ⫽ 0.006, N ⫽ 3,194). RCTs
with follow-up greater than 2 years did not appear to produce lower effect size estimates
than those with less extended follow-up. Quality of reporting of trials was in general low.
Conclusion: Cognitive exercise training in healthy older individuals produces strong
and persistent protective effects on longitudinal neuropsychological performance. Trans-
fer of these effects to dementia-relevant domains such as general cognition and daily
functioning has also been reported in some studies. Importantly, cognitive exercise has
yet to be shown to prevent incident dementia in an appropriately designed trial and this
is now an international priority. (Am J Geriatr Psychiatry 2009; 17:179 –187)

Key Words: Dementia, prevention, cognitive exercise, mental activity, cognitive train-
ing, brain reserve, cognitive reserve

Received December 14, 2007; revised May 15, 2008; accepted June 27, 2008. From the School of Psychiatry, University of New South Wales;
and Neuropsychiatric Institute, Prince of Wales Hospital, Sydney, NSW, Australia. Send correspondence and reprint requests to Michael
Valenzuela, B.Sc. Hons., M.B.B.S. Hons., Ph.D., Neuropsychiatric Institute, Prince of Wales Hospital, Randwick, Sydney, NSW 2031, Australia.
e-mail: michaelv@unsw.edu.au
© 2009 American Association for Geriatric Psychiatry

Am J Geriatr Psychiatry 17:3, March 2009 179

Systematic Review of RCTs of Cognitive Training

T here are an estimated 27 million individuals af-

fected by dementia worldwide,1 with the cost of
care in many developed countries already outstripping
tivity group, whereas those with the highest activity
levels had their risk reduced to 33%.7
These findings can now be assessed against six of
those associated with cardiovascular disease and can- the key etiological criteria posed by Hill in 19658—
cer combined. The rate of new dementia diagnosis is set association, consistency, dose dependency, biological
to rise because of the shifting age profile of the popu- plausibility, coherency, and temporal primacy. Meta-
lation, with projections in the U.S. of over 7.7 million analysis has clearly shown a robust association be-
demented individuals by 2030. There is therefore an tween mental activity and dementia incidence, and
urgent need for development of strategies for the pre- that these are highly consistent. More recent studies
vention of dementia. Within this context, delay of have furthermore pointed to a dose-dependent pat-
symptom onset is a modest yet potentially powerful tern in this association. In terms of biological plausi-
goal: delay of dementia presentation by 5 years would bility, there is an embarrassment of riches from ani-
effectively halve the burden of disease.2 In this article, mal and human studies indicating the action of
clinical trials that have examined the impact of cogni- several potential mechanisms. These have been re-
tive training on longitudinal neuropsychological per- viewed in detail elsewhere in conjunction with the
formance will be systematically reviewed. When com- introduction of a more coherent theoretical framework
bined with preclinical and epidemiological evidence, for their interpretation.4 Potential mediatory mecha-
there are reasonable grounds to expect that mental nisms will only briefly be addressed in Discussion.
activity may be a safe and effective strategy for delay- An important conceptual question therefore remains
ing the onset of cognitive impairment in late life. as to the direction of causality: is active cognitive life
style a prospective predictor of dementia, or low
activity levels in fact an early sign of preclinical
Epidemiological Evidence for a Protective Role
disease.9 This was partially addressed in a meta-
of Mental Activity
analysis of studies focused on cognitive decline
Several international community-based cohort rather than dementia incidence,10 after adjusting
studies have now examined the link between mental for baseline level of cognition. Individuals with
activity and dementia risk. Typically, these studies high levels of mental activities were found to have
compare incidence rates in groups with either high significantly less risk for prospective cognitive de-
or low levels of educational attainment, occupational cline than those with lower activity levels, comple-
complexity, or participation in cognitive life style menting the results from the first meta-analysis.
activities. A meta-analysis of 22 such studies inte- While persuasive, such a correlational approach
grated data from over 29,000 individuals and found does not exclusively rule out reverse causality. For
consistent results3: an overall risk reduction of 46% this reason, results from intervention studies are
for high mental activity levels compared to low ac- critical to development of any future dementia
tivity (OR 0.54, CI: 0.49 – 0.59). Interestingly, the in- prevention programs and are systematically re-
dependent effects of education (OR 0.53), occupa- viewed here.
tional complexity (OR 0.56), and cognitive life style
(OR 0.50) were similar in magnitude.
More recent studies suggest that activity in the
later stage of life may also have a beneficial effect
METHODS
independent of earlier life experiences. Six cohort
studies to examine this issue have replicated a pro- Search Strategy
tective effect in the order of approximately 40%–50%,
even after simultaneous control for other risk factors We searched MEDLINE (1950-November Wk 2 2007)
including education level.4 Moreover, a number of and PubMed (www.pubmed.gov) databases for origi-
these studies also point to dose-dependent ef- nal research articles in any language which met our
fects.3,5,6 One group, for example, found that the risk criteria. Our initial search strategy included the
for dementia in a group with a moderate level of intersection of the following terms: 关“randomized
leisure activities was 50% compared to the low ac- control trial” or “randomized”兴, 关“cognitive” or

180 Am J Geriatr Psychiatry 17:3, March 2009

 Valenzuela and Sachdev

“mental”兴, 关“training” or “exercise”兴, 关“longitudinal” lar to Cohen’s d but adjusts for potential bias from
or “follow-up”兴, and 关“older adults” or “elderly”兴. studies with small sample size.11 Because results
This search produced 50 studies. This list was sup- from individual studies were represented by relative
plemented by manual searches through reference mean change scores in the training versus control
lists of published reports. Abstracts from a final total groups, negative scores indicate decline over time
of 54 studies were then reviewed to assess suitability and positive scores a longitudinal increase. Inverse
for inclusion. variance methods are then used to combine results
across studies, whereby individual effect sizes are
weighted according to the reciprocal of their vari-
Entry Criteria
ance. Finally, statistical inferences are made on the
Inclusion criteria were as follows: i) RCT design, ii) basis of the z test, which represents the overall effect
intervention through a cognitive exercise regime, size estimate relative to its SE.12 A single primary
which includes any type of training using repetitive neuropsychological outcome variable was used for
cognitive tasks over separate days for more than 1 each study to minimize colinearity bias.
week, iii) longitudinal neuropsychological follow-up
after cessation of training, defined as greater than 3
months, and iv) participation by healthy, community
dwelling older adults greater than 50 years of age.
RESULTS
The single additional exclusion criterion used was
participation by patients with clinical dementia, cog- Seven studies met inclusion criteria and represented
nitive impairment, or other major neurological or a cumulative sample of 3,194 individuals. Only three
psychiatric condition. The three most common rea- studies met more than 12 of the 22 CONSORT re-
sons for rejecting studies from our initial list were as porting criteria for RCTs. More details about these
follows: failure to investigate a cognitive training studies are available in Table 1.
intervention, not using healthy elderly, and not re- The relative effect sizes from longitudinal RCTs of
porting a primary longitudinal cognitive outcome. cognitive exercise are shown in Figure 1. All studies
have found effects in a protective direction, with four
Appraisal of Study Quality and Data Extraction out of seven showing statistically significant effects.
The cumulative WMD effect size estimate was 1.07
Seven RCTs met entry criteria and were individually with a 95% confidence interval between 0.32 and 1.83
scored on their published adherence to the CONSORT (z ⫽ 2.78. N ⫽ 7, p ⫽ 0.006, N ⫽ 3,194).
2001 reporting criteria (www.consort-statement.org; Sensitivity analysis was confined to examining the
see Table 1). Given the modest number of relevant effect of medium-term (⬍2 years) versus long-term
papers, key information was extracted by a single (ⱖ2 years) follow-up. The average effect size from
reviewer onto a standard template. Where key infor- RCTs with long-term follow-up (WMD: 1.02 CI:0.14 –
mation was missing from the published version, con- 1.89, z ⫽ 2.28, N ⫽ 5, p ⫽ 0.02, N ⫽ 3,040) was within
tact was made with the lead author and information the 95% confidence interval of those with less than
requested. 2-year follow-up (WMD: 1.16 CI: 0.37–1.96, z ⫽ 2.88,
N ⫽ 2, p ⫽ 0.004, N ⫽ 154).
Quantitative Meta-Analysis

A random effects weighted mean difference

(WMD) model was used to estimate the integrated
DISCUSSION
effect size across trials using Review Manager (Ver-
sion 4.2 for Windows. Copenhagen, Denmark: The
Longitudinal RCTs of Cognitive Exercise
Nordic Cochrane Centre, The Cochrane Collabora-
tion, 2003). Review Manager uses the Hedges ad- Cognitive exercise in older adults is undoubtedly
justed g formula for estimation of the standardized effective for improving performance in the trained
mean difference of individual studies which is simi- task if one conducts follow-up assessment immedi-

Am J Geriatr Psychiatry 17:3, March 2009 181

Systematic Review of RCTs of Cognitive Training

TABLE 1. Details of Longitudinal RCTs of Mental Activity Training Included in Meta-Analysis

Initial Sample
Size (N) and Follow-up
Study Intervention Control Type Time (Months) Main Outcome
Mahncke et al. (18) Computer-based training Active contact 162 Healthy elders 3 Digit span
on several cognitive
tasks—40 sessions of
60 m over 10 weeks
Willis et al. (17) Reasoning training—10 No contact 2,832 Healthy 72 IADLs
sessions of 60 m over elders
5 weeks
Oswald et al. (19) Paper and pencil No contact 375 Healthy elders 72 Composite
training on memory, cognitive
problem solving, and score
information
processing speed
tasks—30 sessions of
90 m over 30 weeks
Derwinger et al. (20) Self-generated strategy No contact 81 Healthy elders 8 Recall of short
training—10 1 hour number
sessions twice/week sequences
for 5 weeks
Ball et al. (16) Information processing No contact 2,832 Healthy 24 Information
speed, memory, and elders processing
problem solving speed
training—10 sessions
of 60 m over 5 weeks
Stigsdotter et al. (54) Multifactorial training No contact 30 Healthy elders 42 Buschke
covering encoding selective
with imagery and reminding
method of loci, test
attentional task
exercises and
relaxation—8 weekly
sessions
Scogin et al. (21) Manual completion No contact 27 Healthy elders 36 Benton visual
focusing on learning with self-report retention test
mnemonic skills, memory
increasing encoding complaints
time, and practice
exercises

ately after the end of training.13,14 Two central ques- proximates a relative improvement of 1.2/2.6 points
tions, however, need to be addressed in order to in the MMSE, or 4.1/9.9 ADAS-Cog points, when
hypothesize whether a similar strategy may be effec- extrapolated to a community-based sample of either
tive for the prevention of dementia onset. These are older cognitively intact individuals or those with
a) transfer of effect: do improvements from particu- Mild Cognitive Impairment (MCI), respectively.15
lar training regime generalize to other nontrained Before discussing the pros and cons of this analytical
domains and functions over time, and b) persistence approach, salient individual studies will be exam-
of effect: do such effects last beyond the proximal ined in more detail.
posttraining period? The largest trial so far has been the ACTIVE
Our meta-analysis suggests that a discrete “dose” study,16 which examined the effect of 10 sessions of
of cognitive exercise in the order of 2–3 months may cognitive training on 2,832 healthy older individuals
have long-lasting and persistent protective effects on divided into three different intervention groups:
cognition over a number of years in healthy older memory training, reasoning training, and processing
individuals. The overall integrated effect size was speed training. Each intervention improved cogni-
strong in magnitude, estimated to be 1.07 (CI: 0.32– tive ability in the targeted area 2 years later. How-
1.83). In more common clinical terms, this effect ap- ever, there was neither evidence of transfer of gain to

182 Am J Geriatr Psychiatry 17:3, March 2009

 Valenzuela and Sachdev

FIGURE 1. Summary of Key Findings From RCTs of Mental Activity Training on Longitudinal Cognitive and Functional
Change

Notes: Treatment and control means refer to posttest–pretest differences and SD to SD. Studies are listed in chronological order from earliest
to most recent. Method of integration is weighted mean difference (WMD), which represents the overall standardized difference between
treatment and control post-pre effects. See Table 1 for study details.

other domains nor any effect on instrumental activ- Although the cumulative effect size and concor-
ities of daily living (IADLs). dance across longitudinal trials of cognitive exercise
A 5-year follow-up to the ACTIVE study has, how- is promising, some caution against over interpreta-
ever, been recently reported with change in IADLs tion is recommended. Primary outcome measures,
used as a main outcome measure.17 Reasoning train- for example, differed widely across the trials, as did
ing specifically protected against functional decline the duration, precise nature, and frequency of the
over this extended follow-up period compared with interventions (Table 1). Second, only the single most
any of the other interventions or the control wait- clinically relevant primary outcome variable was en-
and-see condition. This is therefore the first major tered into this meta-analysis per trial, generally at the
clinical trial to show a significant transfer of effect: exclusion of secondary outcomes which tended to be
directed cognitive exercise producing robust and en- less robust. Third, rigorous adherence to CONSORT
during benefits on a general functional outcome that clinical trials guidelines has been an exception rather
is highly relevant to dementia onset. than the rule. In general, the quality of trials in this
Mahncke et al.18 conducted a RCT that was notable area has been disappointing.
for the use of computer-based cognitive exercises, On the other hand, it is encouraging that those stud-
allowing individuals to face tasks of increasing dif- ies with longer term follow-up showed no evidence of
ficulty as their skill levels progress. Neuropsycholog- less potent effects than those with more modest follow-
ical tests immediately after the end of the training up. Claims for persistence of effect would therefore
period found verbal memory performance improved appear justified. Finally and perhaps most signifi-
by up to 25% of a SD, and testing 3 months later cantly, two of the more recent clinical trials have shown
showed that short-term memory performance re- that their training protocols generalize to domains be-
mained enhanced. yond the narrow focus of the exercise regime.17,19 Well-
The Sim-A clinical trial compared the effects of designed cognitive exercise interventions may thereby
cognitive, physical, and combined training in healthy have potential to transfer to those domains critical for
older individuals over a 5-year period.19 Despite the the development, and thus prevention, of dementia
randomization procedure being incomplete, 30 pa- such as general cognitive function and instrumental
per-and-pencil cognitive training sessions produced activities of daily activity.
a significant effect over both the 12 month and 5-year
follow-up periods. Moreover, this effect seemed to Could Mental and Physical Activity
transfer to a measure of general cognition. Other Be Synergistic?
smaller studies with samples of less than 100 indi-
viduals have found positive trends but have lacked A systematic review by Colcombe and Kramer22 of
power20,21 (see Fig. 1). RCTs examined the effects of physical exercise on the

Am J Geriatr Psychiatry 17:3, March 2009 183

Systematic Review of RCTs of Cognitive Training

cognitive abilities of older adults. The overall effect Mental stimulation is a robust trigger for the in-
size from this review was 0.48 for exercisers and 0.16 duction of brain-derived neurotrophic factor and
across the control groups, indicative of a relative nerve growth factor.24,28 These molecules are vital for
effect size of approximately 0.32, less than half the neural cell survival and proliferation, and knockout
effect seen in the current meta-analysis of longitudi- of these genes leads to severe impairments in learn-
nal cognitive training studies. It is important to note ing and synaptic plasticity.29,30 Electrophysiological
that RCTs integrated by these authors did not feature measures of synaptic plasticity such as long-term
longitudinal follow-up, and thus the applicability of potentiation are also augmented.31 Similarly, enrich-
these findings for the prevention of dementia re- ment induces profound increases in synaptogenesis,
mains unclear. by as much as 150%–200% in quantitative studies.32
The issue of potential interactive effects has been This effect is especially salient to clinical dementia, as
assessed in two RCTs who studied cognitive train- synaptic density is arguably the most accurate bio-
ing, physical training, and their combination in physical correlate of cognitive impairment.33,34 Fur-
comparison to a control condition. A preliminary thermore, dozens of studies have now shown that
study without longitudinal follow-up found that enrichment can increase neurogenesis in the adult
the average difference between pretraining and hippocampus.35 There remain, however, a number of
posttraining memory scores was significantly unresolved issues,36 not least whether generation of
higher in the combined group than in either the new neurons in the adult brain is of any functional
aerobic or mental training alone.23 As mentioned, significance.37,38
the Sim-A trial also examined the effect of 30 ses- Early studies in animals found that the mass and
sions of combined training after 5 years of follow- girth of the brain increased after a period of en-
up. Cognitive training alone produced a prepost richment.39 Remarkably, similar regional effects
effect size of d ⫽ 0.13 (p ⬍0.001), whereas the have been found in humans using volumetric MRI
control and physical training conditions alone had techniques after a period of behavioral training40
no significant longitudinal effects. The combined and physical exercise.41 Whether mental or physi-
physical and cognitive training condition, how- cal training can retard the rate of hippocampal
ever, resulted in an effect size more than thrice the atrophy seen in A.D. is unknown, yet we have
cognitive training alone of d ⫽ 0.75 (p ⬍0.001). shown that lifetime levels of mental activity are
Preliminary clinical trials evidence suggests that inversely related to rate of hippocampal atrophy in
combining physical and mental exercise may pro- healthy older individuals42 and that cognitive ex-
duce greater cognitive benefits over time than ei- ercise by healthy older individuals can increase
ther intervention alone. Capacity for such syner- levels of putatively neuroprotective metabolites in
gistic potentiation clearly needs further evaluation. the hippocampus specifically.43 Cognitive training
has also been shown to increase the efficiency of
resting state metabolism in the frontal lobe.44
Biological Mechanisms Underlying
Perhaps the most intriguing data has come from
These Benefits
reports that mental or physical activity may di-
For over four decades, the “environmental en- rectly interfere with A.D. pathophysiology. Enrich-
richment” paradigm has been the dominant exper- ment and voluntary running in transgenic A.D.
imental technique for analyzing the combined ef- mice have been associated with decreased levels of
fects of mental and physical activity on the amyloid pathology in several studies, by as much
mammalian brain.24 Enrichment involves changing as 50%.45– 47 Conflicting data,48 however, means
the home conditions of the animal to include that replication of such an effect in humans using
greater opportunities for exploration of novel toys molecular imaging is a high priority.
and mazes, more contact with other animals and ad-
ditional opportunities for voluntary exercise.25,26 Com- Challenges Facing RCTs of Cognitive Exercise
prehensive accounts of the biological changes induced
by mental4 and physical exercise27 are available else- The balance of available data suggests that preven-
where. tion of dementia is a realistic ambition for interven-

184 Am J Geriatr Psychiatry 17:3, March 2009

 Valenzuela and Sachdev

tions based on cognitive exercise. Importantly, this Details of the intervention also need close attention
claim has yet to be tested in a prospective, double- for there remains no consensus on the nature of the
blind randomized control trial. This is therefore an optimal cognitive exercise. In general, multidomain
international priority. Several key challenges will cognitive exercises seem to produce more robust re-
need to be addressed for these to proceed efficiently sults than single-domain training. Computerized de-
and effectively. livery of such training so that individuals can be
Careful selection of the initial cohort has emerged continually challenged across the training period at a
as a primary issue. One reason that the ACTIVE personalized level also makes intuitive sense and will
study did not find any effects on functional outcomes assist in standardized and replicable administration.
at 2-year follow-up was because the control group Attrition of participants is an important issue as
exhibited almost no decline over this time. Given for any longitudinal clinical trial. In the ACTIVE
that interventions in this context are unlikely to in- study, 294 of the original 2,832 participants were not
crease functional performance, but rather slow or able to complete 2-year follow-up assessment, an
halt the rate of impairment, a level of naturalistic average attrition rate of 5% per annum. Interestingly,
decline in the comparison group proves to be vital. this compares favorably with attrition rates from
Moreover, the type of “supernormal” participants MCI drug studies that vary from 12% to 27% per
that often volunteer for these types of trials can se- annum.53 Similarly, the potential for adverse events
verely limit generalization of results. The average from trials of cognitive exercise is probably low, with
baseline MMSE score in the ACTIVE study was, for none reported thus far, yet this may reflect general
example 27.3, significantly greater than the reference poor reporting standards rather than a true absence
population. This begs the question as to what one of adverse events.
expects to achieve through intervention on a 75-year Finally, improved design of the control condition
old person with a MMSE of 28/30 and no functional is required. The experiences of those on no contact
limitations? waiting lists are vastly different from the interven-
For these reasons, an “at-risk” group is prefera- tion experience beyond the cognitive factors, which
ble, yet how one defines this has its own challenges. investigators assume are etiologically salient. Com-
MCI, for example, has a number of practical and ing into to the research institute to see investigators
ontological difficulties, including competing defi- and study staff, talking and interacting with other
nitions and highly onerous screening require- participants and the general sense of “doing some-
ments.49 Other options include raising the mini- thing positive” all need to be better accounted for
mum age of entry, using “borderline” MMSE entry through use of active control arms in future RCTs.
cut-offs, or selecting on APOE4 or presence of
other risk factors. Counterbalancing the at-risk se- How Should We Advise Our Older Patients?
lection strategy is a perceived fear that these indi-
viduals may already be “too far down” a patho- Given the minimal risk for harm from taking on
logical process for behavioral interventions to additional mental activities—and the high likelihood
work. Yet as a number of studies have found,50 –52 that this may in fact prove beneficial from a longitu-
it is possible to slow the rate of cognitive decline dinal cognitive perspective—it is suggested that
through mental activity even in early dementia. older adults maintain a robust level and range of
This concern may therefore be overemphasized mental activities, particularly after retirement. It is
and borderline cognitive function does not appear also important to emphasize that no amount of men-
to exclude individuals from completing cognitive tal activity is sufficient to guarantee against develop-
training. Indeed, to maximize generalization of re- ing dementia or age-related cognitive decline.
sults to community and clinical settings, exclusion
on the basis of comorbidities should be minimized.
It is, however, acknowledged that those with vi-
sual impairment will find it difficult to learn and CONCLUSIONS
practice cognitive exercises that rely on visual, Growing epidemiological and clinical trials evi-
written, and spatial stimuli. dence suggests that cognitive exercise may be an

Am J Geriatr Psychiatry 17:3, March 2009 185

Systematic Review of RCTs of Cognitive Training

effective strategy for delaying the onset of cogni- robust mental activity, particularly after retire-
tive impairment in older adults. A number of plau- ment for optimal cognitive health.
sible neurobiological mechanisms may account for
these benefits. Clinical trials that address the re-
maining issues are therefore required. Trials will The authors thank Dr. Sharon Tennstedt for pro-
need careful attention to patient selection, to de- viding additional data from the ACTIVE study. This
sign of cognitive exercise and control conditions, work is supported by the National Health and Medical
and importantly, to improve quality of reporting. Research Council of Australia Program grant 350833.
While awaiting such corroborating results, it may MV is a Vice-Chancellor’s research fellow at the Uni-
be prudent to advise older individuals to maintain versity of New South Wales.

References
1. Wimo A, Jonsson L, Winblad B: An estimate of the worldwide in healthy older adults using a brain plasticity-based training
prevalence and direct costs of dementia in 2003. Dement Geriatr program: a randomised, controlled study. Proc Natl Acad Sci USA
Cogn Disord 2006; 21:175–181 2006; 103:12523–12528
2. Katzman R: Education and the prevalence of dementia and Alz- 19. Oswald W, Gunzelmann T, Rupprecht R, et al: Differential effects
heimer’s disease. Neurology 1993; 43:13–20 of single versus combined cognitive and physical training with
3. Valenzuela MJ, Sachdev P: Brain reserve and dementia: a system- older adults: the SimA study in a 5-year perspective. Eur J Ageing
atic review. Psychol Med 2006; 36:441– 454 2006; 3:179 –192
4. Valenzuela M, Breakspear M, Sachdev P: Complex mental activity 20. Derwinger A, Stigsdotter Neely A, Backman L: Design your own
and the ageing brain: molecular, cellular and cortical network memory strategies! Self-generated strategy training versus mne-
mechanisms. Brain Res Rev 2007; 56:198 –213 monic training in old age: an 8-month follow-up. Neuropsychol
5. Fratiglioni L, Wang HX, Ericsson K, et al: Influence of social Rehabil 2005; 15:37–54
network on occurrence of dementia: a community -based longi- 21. Scogin F, Bienias J: A three-year follow-up of older adult partici-
tudinal study. Lancet 2000; 355:1315–1319 pants in a memory-skills training program. Psychol Aging 1988;
6. Fabrigoule C, Letenneur L, Dartigues JF, et al: Social and leisure 3:334 –337
activities and risk of dementia: a prospective longitudinal study. 22. Colcombe S, Kramer S: Fitness effects on the cognitive function
J Am Geriatr Soc 1995; 43:485– 490 of older adults: a meta-analytic study. Psychol Sci 2003; 14:125–
7. Verghese J, Lipton R, Katz M, et al: Leisure activities and the risk 130
of dementia in the elderly. N Engl J Med 2003; 348:2508 –2516 23. Fabre C, Chamari K, Mucci P, et al: Improvement of cognitive
8. Hill A: The environment and disease: association or causation? function via mental and/or individualised aerobic training in
Proc R Soc Lond 1965; 58:295–300 healthy elderly subjects. J Sports Med 2002; 23:415– 421
9. Gallacher J, Bayer A, Ben-Shlomo Y: Commentary: activity each 24. Mohammed AH, Zhu S, Darmopil S, et al: Environmental enrich-
day keeps dementia away— does social interaction really pre- ment and the brain, in Progress in Brain Research. Edited by
serve cognitive function? Int J Epidemiol 2005; 34:872– 873 Hofman M, Boer G, Hotmaat A, et al. New York, Elsevier, 2002,
10. Valenzuela M, Sachdev P: Brain reserve and cognitive decline: a pp 109 –133
nonparametric systematic review. Psychol Med 2006; 36:1065– 25. Nithianantharajah J, Hannan A: Enriched environments, experi-
1073 ence-dependent plasticity and disorders of the nervous system.
11. Hedges L, Olkin I: Statistical Methods for Meta-analysis. San Di- Nat Rev Neuroscience 2006; 7:697–709
ego, Academic Press, 1985 26. van Praag H, Kempermann G, Gage FH: Neural consequences of
12. Deeks J, Higgins J: Statistical algorithms in Review Manager 5. environmental enrichment. Nat Rev Neurosci 2000; 1:191–198
The Cochrane Collaboration, 2007 27. Hillman C, Erickson K, Kramer A: Be smart, exercise your heart:
13. Rebok G, Carlson M, Langbaum J: Training and maintaining mem- exercise effects on brain and cognition. Nat Rev Neurosci 2008;
ory abilities in healthy older adults: traditional and novel ap- 9:58 – 65
proaches. J Gerontol B 2007; 62:53– 61 28. Ickes B, Pham T, Sanders L, et al: Long-term environmental
14. Verhaeghen P, Marcoen A, Goossens L: Improving memory per- enrichment leads to regional increases in neurotrophin levels in
formance in the aged through mnemonic training: a meta-analys- rat brain. Exp Neurol 2000; 164:45–52
tic study. Psychol Aging 1992; 7:242–251 29. Patterson S, Abel T, Deuel T, et al: Recombinant BDNF rescues
15. Pyo G, Elble R, Ala T, et al: The characteristics of patients with deficits in basal synaptic transmission and hippocampal LTP in
uncertain/mild cognitive impairment on the Alzheimer Disease BDNF knockout mice. Neuron 1996; 16:1137–1145
Assessment Scale-cognitive subscale. Alzheimer Dis Assoc Disord 30. Genoud C, Knott G, Sakata K, et al: Altered synapse formation in
2006; 20:16 –22 the adult somatosensory cortex of brain-derived neurotrophic
16. Ball K, Bersch D, Helmers K, et al: Effect of cognitive training factor heterozygote mice. J Neurosci 2004; 24:2394 –2400
interventions with older adults—a randomised control trial 2400. 31. Artola A, von Frijtag J, Fermont P, et al: Long-lasting modulation
JAMA 2002; 288:2271–2281 of the induction of LTD & LTP in rat hippocampal CA1 by
17. Willis S, Tennstedt S, Marsiske M, et al: Long term effects of behavioural stress and environmental enrichment. Eur J Neurosci
cognitive training on everyday functional outcomes in older 2006; 23:261–272
adults. JAMA 2006; 296:2805–2814 32. Levi O, Jongen-Relo A, Feldon J, et al: ApoE4 impairs hippocam-
18. Mahncke H, Connor B, Appelman J, et al: Memory enhancement pal plasticity isoform-specifically and blocks the environmental

186 Am J Geriatr Psychiatry 17:3, March 2009

 Valenzuela and Sachdev

stimulation of synaptogensis and memory. Neurobiol Dis 2003; 44. Small G, Silverman D, Siddharth P, et al: Effects of a 14-day
13:273–282 healthy longevity lifestyle program on cognition and brain func-
33. Terry RD, Masliah E, Salmon DP, et al: Physical basis of cognitive tion. Am J Geriatr Psychiatry 2006; 14:538 –545
alterations in Alzheimer’s disease: synapse loss is the major cor- 45. Lazarov O, Robinson J, Tang Y, et al: Environmental enrichment
relate of cognitive impairment. Ann Neurol 1991; 30:572–580 reduces Aß levels and amyloid deposition in transgenic mice. Cell
34. Scheff S, Price DA: Synaptic pathology in Alzheimer’s disease: a 2005; 120:701–713
review of ultrastructural studies. Neurobiol Aging 2003; 24:1029 – 46. Costa D, Cracchiolo J, Bachstetter A, et al: Enrichment improves
1046 cognition in AD mice by amyloid-related and unrelated mecha-
35. Kempermann G: Adult Neurogenesis. New York, Oxford Univer- nisms. Neurobiol Aging 2006; 28:831– 844
sity Press, 2006 47. Adlard P, Perreau V, Pop V, et al: Voluntary exercise decreases
36. Olson A, Eadie B, Ernst C, et al: Environmental enrichment and amyloid load in a transgenic model of Alzheimer’s Disease. J Neu-
voluntary exercise massively increase neurogenesis in the adult hip- rosci 2005; 25:4217– 4221
pocampus via dissociable pathways. Hippocampus 2006; 16:250–260 48. Janowsky J, Melnikova T, Fadale DJ, et al: Environmental enrich-
37. van Praag H, Schinder A, Christie B, et al: Functional neurogenesis
ment mitigates cognitive deficits in a mouse model of Alzheimer’s
in the adult hippocampus. Nature 2002; 415:1030 –1034
Disease. J Neurosci 2005; 25:5217–5224
38. Meshi S, Drew M, Saxe M, et al: Hippocampal neurogenesis is not
49. Visser PJ, Brodaty H: MCI is not a useful concept. Int Psychoge-
required for behavioural effects of environmental enrichment.
riatr 2006; 18:402– 409
Nat Neurosci 2006; 9:729 –731
50. Olazaran J, Muniz R, Reisberg B, et al: Benefits of cognitive-motor
39. Rosenzweig M, Bennett E: Effects of differential environments on
intervention in MCI and mild to moderate Alzheimer disease.
brain weights and enzyme activities in gerbils, rats, and mice. Dev
Psychobiol 1969; 2:87–95 Neurology 2004; 63:2348 –2353
40. May A, Hajak G, Steffens T, et al: Structural brain alterations 51. Loewenstein D, Acevedo A, Czaja S, et al: Cognitive rehabilitation
following 5 days of intervention: dynamic aspects of neuroplas- of mildly impaired Alzheimer disease patients on cholinesterase
ticity. Cereb Cortex 2007; 17:205–210 inhibitors. Am J Geriatr Psychiatry 2004; 12:395– 402
41. Colcombe S, Eriksson E, Scalf P, et al: Aerobic exercise training 52. Sitzer D, Twamley E, Jeste D: Cognitive training in Alzheimer’s
increases brain volume in aging humans. J Gerontol: Med Sci disease: a meta-analysis of the literature. Acta Psychiatr Scand
2006; 61:1166 –1170 2006; 114:75–90
42. Valenzuela M, Sachdev P, Wen W, et al: Lifespan mental activity 53. Jelic V, Kivipelto M, Winblad B: Clinical trials in mild cognitive
predicts diminished rate of hippocampal atrophy. PLoS One impairment: lesson for the future. J Neurol Neurosurg Psychiatry
2008; 3:e2598 2005; 77429 – 438
43. Valenzuela MJ, Jones M, Wen W, et al: Memory training alters 54. Stigsdotter Neely A, Bäckman L: Long-term maintenance of gains
hippocampal neurochemistry in healthy elderly. Neuroreport from memory training in older adults: two 3 1/2-year follow-up
2003; 14:1333–1337 studies. J Gerontol: Psychol Sci 1993; 48:P233–P237

Am J Geriatr Psychiatry 17:3, March 2009 187