Search Strategy
Date of search: 02nd May 2018
Search starting time: 31st December 2015
Full search terms
(dementia OR cognit* OR mild cognitive impairment OR Alzheimer disease OR dementia vascular OR dementia multi-infarct OR MCI OR cognitive dysfunction OR neuropsychologi* OR Health-Related Quality Of Life OR life quality OR Activities, Daily Living OR Chronic Limitation of Activity OR Limitation of Activity, Chronic OR ADL OR activities of daily living OR Drug-Related Side Effects and Adverse Reactions OR Adverse Drug Event OR Adverse Drug Reaction OR Long Term Adverse Effects OR Adverse Effects, Long Term Disease-Free Survival OR Event-Free Survival OR Adverse effects) AND (Exercise OR exercise therapy OR Acute Exercise OR Aerobic Exercise OR Exercise Training OR Exercise, Aerobic OR Exercise, Isometric OR Exercise, Physical OR Isometric Exercise OR Physical Activity OR resistance training)
Simplified search terms
(dementia OR MCI OR cognition OR Quality Of Life OR ADL OR Adverse Effects OR Drop-out) AND (exercise OR physical activity)
Searches were conducted in the following databases1:
- –
Cochrane
- –
Pubmed
- –
NICE Guidelines
- –
Embase
- –
PsycInfo
- –
Global Health Library (Including WHOLIS, PAHO, AIM, LILACS)
- –
Database of impact evaluations
- –
AFROLIB
- –
ArabPsycNet
- –
HERDIN NeON
- –
HrCak
- –
IndMED
- –
KoreaMed
- –
AJOL
Narrative descriptions of the studies that went into the analysis
GRADE tables 1–3: Physical activity interventions in adults with normal cognition
– present the evidence from three systematic reviews that assessed the efficacy of aerobic exercise22, resistance training22 and multimodal23,24 interventions in preventing/delaying dementia and/or cognitive decline in people with normal cognition. Although Barha et al. (2017)22 assessed the effects of all three types of physical activity interventions (aerobic exercise, resistance training and multimodal exercise) on cognition, a more recent systematic review of equal quality by Northey et al. (2018)24 investigating the effects of multimodal exercise on cognition was selected.
Barha et al.22 conducted a systematic review and meta-analysis to assess if different types of exercise interventions can improve domain-specific cognition in older adults with normal cognition. Two reviewers (C B and RF) conducted the extensive literature search and study evaluation independently. A total of 41 randomised controlled trials (RCTs) were included in the narrative qualitative synthesis and 39 in the meta-analysis. The interventions were classified into three categories: aerobic training, resistance training and multimodal training. The control groups varied from active balance and tone group to sedentary groups (36 studies) to multimodal control groups (1 study) and not reported (2 studies). The outcomes included domain-specific cognition (global, executive function, episodic memory, visuospatial function, word fluency and processing speed). Each cognitive domain was measured with a range of tests (e.g. global cognition with MMSE; executive function with set shifting, Stroop test; episodic memory with logical memory, immediate and delayed recall; visuospatial function with useful field of view, reaction time, clock drawing; word fluency with words by letter and category; processing speed with cancellation test, simple/choice reaction time test). The number of the participants, who were middle-aged older adults (45 years and older) without any neurodegenerative/clinical disorders, ranged from 18–666. The mean follow-up time of the 41 RCTs was 30 weeks.
The authors found that aerobic exercise had greater cognitive benefits than resistance training for all cognitive domains. Beneficial effects of aerobic exercise were observed for global cognition: standardized mean difference (SMD 0.85, 95% CI: 0.24 – 1.47, p=0.007); executive function (SMD 2.06, 1.58 – 2.55, p<0.000); visuospatial function (SMD 0.64, 0.14 – 1.15, p=0.013); word fluency (SMD 0.35, 95% CI: 0.2 – 0.5, p<0.000); processing speed (SMD 0.47, 0.24 – 0.7, p<0.000). The gains in the episodic memory (SMD 0.04, 95% CI: −0.36 – 0.45, p= 0.827) were not significant ().
Resistance training improved executive function (SMD 0.64, 95% CI: 0.41 – 0.87, p<0.000) and visuospatial function (SMD 0.55, 95%CI: 0.14–0.95, p=0.009). No improvement was seen for episodic memory (SMD 0.07, 95% CI: 0.08 – 0.22, p=0.337) and word fluency (SMD 0.83, 95% CI: 0.87–2.53, p=0.339) and there was negative effect for global cognition (SMD −1.81, 95% CI: 2.88 −0.75, p=0.001). No study assessing impact of resistance training on processing speed was available (). The authors also conducted tests for heterogeneity among the studies (Q statistic and Higgins I2); and noted how longer interventions may have greater impact on cognitive function.
Barreto et al.23 conducted a systematic review to assess effects of multimodal exercise on dementia onset, mild cognitive impairment (MCI) onset and cognitive decline. Two reviewers conducted literature search independently. Cochrane Collaboration’s tool for evaluating risk of bias was applied to the included studies. A total of 5 randomised controlled trials (RCTs) were included in the review with total participants (n=2878) and age range (65–80 years). The mean follow-up was 12 months. Control group varied from placebo to sham exercise to social group.
For dementia onset, 3 studies with participants (n=1966) were included. These studies had high heterogeneity (I2=63.1%). The incidence of dementia was 3.7% (n=949) for exercisers and 6.1% (n=1017) for controls. Multimodal exercise was not found to reduce the risk of dementia onset (Risk ratio (RR)= 0.56 95% CI: 0.23 – 1.36, p=0.20). In a sensitivity analysis, the authors found a significant effect of exercise for reducing the risk of dementia onset by 35%. However, this must be interpreted with caution since and should not be assumed as a definitive finding. It remains undecided that exercise can decrease the risk of incident dementia (). For MCI onset, only one study was available with participants (n=1635) and a follow-up duration of 2 years. The incidence of dementia was 10.2% (n=686) for exercisers and 9.1% (n=682) for controls. Multimodal exercise was not found to reduce the risk of MCI onset (Risk ratio (RR)=1.12, 95% CI: 0.81 – 1.55, p=0.49). Drop-out rates were also reported individually for 5 studies and were not included into any meta-analysis. Mean drop-out rates for the exercise group were 21.4% and for the control group 14.2% ().23
Northey et al.24 conducted a systematic review to assess effects of exercise upon cognition in adults with normal cognition. Since this review was of similar quality as the review used for the dementia and MCI outcomes23, but more recent, it was chosen to gather the evidence on the cognitive function outcomes. After an extensive literature search, two authors (NJ and SD) independently extracted the data and assessed the risk of bias of included RCTs according to the Cochrane Collaboration Guidelines25. GRADE guidelines26 were also applied independently by two authors (NJ and SD) to evaluate the overall quality of evidence for the comparison of cognitive function between exercise and control groups. To study effects of multimodal exercise upon cognition, the review included nine RCTs with total participants (n=716) of age >50 years. Cognition was assessed across following domains: global, attention, executive function, memory and working memory. In order to assess heterogeneity, Q-statistic was applied (p <0.01) showing significant heterogeneity across included studies. Publication bias was identified through an evaluation of funnel plot asymmetry and the effect size was not large enough to downgrade the evidence.
Benefits of multimodal exercise were reported for attention (SMD 0.27, 95% CI: 0.41–0.41), executive function (SMD 0.34, 95% CI 0.2–0.47), memory (SMD 0.36, 95% CI: 0.22–0.5), working memory (SMD 0.29, 0.12–0.45) except for global cognition (SMD 0.16, 95% CI: 0.14 – 0.47) (). The authors provide positive evidence for both aerobic and resistance training (i.e. multicomponent training), in compliance with exercise recommendations for age group (>50 years) to improve cognitive functions.
Overall risk of bias in included studies varied from not-serious to serious risk of bias. Methodological inconsistences in the studies with high risk of bias were mainly due to sequence generation, random allocation, allocation concealment, incomplete outcome data, and attrition (–).
GRADE tables 4, 5, and 6: Physical activity Interventions in adults with mild cognitive impairment (MCI)
Song et al.20 conducted a systematic review assessing the effects of aerobic exercise, resistance training, and multimodal exercise on cognition in individuals with MCI. Literature search, and quality assessment of the included RCTs was carried out by two authors independently (SD and LY). A total of eleven RCTs (n=881, age range 50–94 years) were included in the systematic review. Control group varied from placebo, to stretching, to health education, to social recreational activities. The review included studies with a single component (aerobic exercise or resistance training) as well as multimodal interventions, and the results were reported on domain-specific cognitive function (global function, executive function, and memory).
In the analysis of aerobic exercise intervention (), only one of the two studies that reported on global cognition showed a significant improvement of global cognition following aerobic exercise intervention (SMD 0.58 95% CI:0.18–0.98)27. For the effect on executive functioning Three aerobic exercise studies were pooled, with no significant effect being detected (SMD 0.03; 95% CI −0.26 to 0.32). Similar results were reported for memory, both as immediate (SMD 0.01; 95% CI −0.22 to 0.24) and delayed recall (SMD 0.01; 95% CI −0.21 to 0.23).
When interventions specifically focused on training exercise were analysed (), a small to moderate improvement on global cognition was reported (SMD 0.41; 95% Cl 0.01 to 0.80). However, executive function improved only in one of the three studies included in the pooled analysis, the one characterised by the longest and intensive intervention28. Furthermore, resistance training did not seem to improve either immediate (SMD 0.12; 95% Cl −0.24 to 0.48) nor delayed (SMD 0.19; 95% Cl −0.17 to 0.55) recall as measures of memory.
The main analysis of the systematic review focused on multidomain exercise interventions (). For global cognition, beneficial effects were reported after pooling data from seven studies (SMD 0.30 95%CI: 0.10–0.49, p=0.02). For executive functioning, no significant beneficial effects were reported (SMD 0.12 95%CI: 0.04–0.29, p=0.14), after combining data from nine studies. For memory, no significant beneficial effects were reported after pooling data from eight studies (SMD 0.04 95%CI: −0.06–0.15, p=0.43). The authors also reported effects of multimodal exercise intervention on health-related quality of life assessed using questionnaires in two RCTs29,30. No meta-analysis was conducted but no significant beneficial effect on health-related quality of life was reported. Drop-out rates of 20.8% and 25% were also reported by same RCTs.
The included RCTs mostly had serious risk of bias due to concerns in allocation concealment, and selection in several studies included in the analysis. Although publication bias was not formally investigated the search was limited only to a small number of databases and no other source was included. Heterogeneity was investigated across studies for each analysis, using I2xref> statistics, which revealed no significant results.
Since these types of intervention studies in this population are limited and are subject to various methodological flaws, more RCTs with rigorous study designs are needed. Future studies are recommended specifying (1) recruitment methods; (2) systematic recruitment of participants to ensure sample representativeness; (3) appropriate controls to enhance internal validity of the findings; (4) clear description of numbers and reasons for withdrawals and drop-outs; (5) approaches to improve adherence; (6) outcomes evaluating the transferability of cognitive gains to psychological well-being and quality of life; (7) comparison group- and individual-based exercises on cognitive and psychological outcomes; (8) long-term follow-ups. These would allow researchers to detect the effects of physical exercise on delaying the progression to dementia among such cohort.
Additional Evidence
The evidence (low to high quality), obtained from the systematic search presented here above, points towards a beneficial effect of physical activity in reducing the risk of dementia and cognitive decline in particular, especially for aerobic exercise and in people with normal cognition. Observational evidence also confirms this conclusion.
In 2016, Lafortune et al.31 conducted a quick systematic review on mid-life behavioral risk factors (including physical activity, diet, smoking, alcohol, weight change, as well as leisure, cognitive activity, and social networks) associated with healthy ageing, dementia, disability and frailty in later life. Concerning the association between mid-life physical activity and cognition/dementia, ten longitudinal cohort studies were included in the qualitative synthesis. The review reported consistent evidence of the association between mid-life physical activity and lower risk of dementia as well as better cognitive functioning, in older life.
The following year after, a systematic review and meta-analysis32 of longitudinal studies investigating the potential protective role of physical activity against cognitive decline, all-cause dementia, Alzheimer’s disease (AD) and vascular dementia was published by Guure and colleagues. 25 studies for all cause dementia with follow-up of at least 12 months and up to 28, were included. The overall sample size was n=117410. The meta-analysis reported a protective effect for high level of physical activity on all-cause dementia (OR 0.79, CI 0.69–0.88), cognitive decline (OR 0.67, CI 0.55–0.78), AD (OR 0.62, CI 0.49–0.75), and a non-protective effect for vascular dementia (OR 0.92, CI 0.62–1.30).
A second systematic review with meta-analysis33 of cohort studies was conducted to assess the dose-response relationship between physical activity and dementia (all-cause, AD and vascular dementia). The systematic review included a total of 16 studies and for all-cause dementia, the sample size was n= 37436, and the follow-up ranged from 3–31.6 years. The meta-analysis reported higher levels of physical activity to be associated with lower risk of all-cause dementia (RR 0.73, CI 0.62–0.87), and AD (RR 0.74, CI 0.58–0.94). The review identified a linear relationship between dementia/AD and leisure-time physical activity, in the range 0–2000 kcal/week or 0–45 metabolic equivalent of task hours (MET-h)/week. For every 500 kcal or 10 MET-h increase per week, an approximate decrease of 10% and 13% in the risk of all-cause dementia and AD was reported, respectively.
Finally, Engeroff and colleagues34 conducted a systematic review assessing the association between physical activity in adult life (18+ years) cognitive functions in late adulthood (60+ years). 14 longitudinal studies and nine cross-sectional studies were included. The review concluded that leisure-time physical activity in early, mid, and late adulthood was associated with better cognitive functioning (including global cognition, executive function, and memory) in older age. Limitations were identified concerning the impact of physical activity in early adulthood and the effect of adherence to current WHO recommendations oh physical activity, due to the fact that only two longitudinal studies have used cut-offs comparable to these recommendations.
