Effectiveness of exercise interventions for adults over 65 with moderate-to-severe dementia in community settings: a systematic review

To review the literature on the effectiveness of exercise interventions for people with moderate-to-severe dementia in community settings. The literature was of low quality, but suggested exercise programmes may improve physical function of people with moderate-to-severe dementia. There was no evidence that exercise programmes improve mood. More research is needed to improve the quality of the evidence to better understand the effectiveness of exercise programmes in community-dwelling older people with moderate-to-severe dementia. To conduct a systematic review of the literature to evaluate the effectiveness of exercise interventions for people with moderate-to-severe dementia in community settings. Eight electronic databases (MEDLINE, Embase, CINAHL, AMED, PsycINFO, PEDro, The Cochrane Library and BNI) were searched from inception to July 2018. Snowball searching identified additional articles not identified initially. Articles were included if they: reported randomised or quasi-randomised controlled trials comparing exercise with usual care or no treatment; and involved people over 65 with moderate-to-severe dementia in community settings. Outcome measures of interest were strength, endurance, mobility, mood and quality of life. Titles and abstracts of all studies were screened by one reviewer. Two reviewers independently screened full text articles for all eligible studies, extracted data and assessed quality and risk of bias. Eight studies with 819 participants were included. Interventions were variable in terms of content, duration and frequency. There was some evidence exercise programmes may improve physical function of people with moderate-to-severe dementia, with significant effects seen for gait speed and endurance, and a trend towards improvement in strength. There was little evidence to suggest exercise programmes improve mood. Most studies were of low quality. Exercise was associated with improvements in gait speed and endurance for older people with moderate-to-severe dementia living in the community, but the quality of evidence was low. There was no conclusive evidence regarding effect on strength or mood. Findings are limited by the quality of the available evidence.


Introduction
Dementia encompasses a range of cognitive and behavioural symptoms including memory loss, judgement and changes in personality that can lead to decline in function and difficulties with activities of daily living [1]. 47 million people with dementia worldwide were affected by activity limitation in 2015, a figure predicted to increase to 75 million by 2030 [2].
People with dementia have increased risk of falls and fractures, whilst some of the morbidity of dementia is related to declining performance status associated with loss of muscle strength and enduranceendurance [3]. Exercise, by improving muscle strength and endurance and reducing risk of falls, has the potential to mitigate against this morbidity and provide physical and mental wellbeing benefits for people living with dementia [4].
The Physical Activity Guidelines Advisory Committee [5] stated the benefits of exercise and physical activity include the potential to increase strength, balance, mobility and cardiovascular fitness, which may lead to subsequent improvements in function and therefore independence [6]. Previous systematic reviews [7] have found evidence to support these physical benefits in older people without cognitive impairment, whilst separate research has suggested that it may also improve psychological wellbeing and health-related quality of life. [8,9] A previous Cochrane review [6] found evidence that exercise interventions in people with dementia were associated with improvement in activities of daily living but that there was no clear evidence of benefit for cognition, neuropsychiatric symptoms or depression. The studies included showed wide heterogeneity and many of the trials focused on people with mild to moderate cognitive impairment. The authors concluded that more work was needed to understand what type of exercise would be most beneficial in people with dementia, at what dose, and whether specific subgroups of dementia patients demonstrated particular benefit. [4,6] Patients with more advanced dementia are differently able to engage with exercise programmes. There may be need to modify or attenuate interventions to account for this. [11,12]. It is also possible that the efficacy and effectiveness of exercise interventions are attenuated by the more advanced frailty found in people with advanced dementia. This group therefore needs to be considered separately [4,6]. This review set out to consider the effectiveness of exercise interventions for people living with moderate to severe dementia focusing on physical benefits such as muscle strength and endurance, and mental wellbeing benefits, such as mood, and quality of life.

Methods
The protocol for this systematic review has been registered on PROSPERO (CRD42018096194) [13]. Randomised and quasi-randomised, including cluster randomised, controlled trials of adults over the age of 65 with moderate to severe dementia in community settings were included. Community settings were defined as being all settings other hospital, including participants' own homes, care homes and nursing homes. Studies were included where the dementia severity for the study population was stated as being moderate or severe.
Where the severity of dementia in the study population was mixed, studies were included where 70% of the participants were classed as being affected by moderate to severe dementia. We accepted author classifications of dementia severity. Where no classification of severity was specified in the articles, we used the previously published cut-offs, used by Forbes, and colleagues [6] of <10, 10-17 and 17-26, to represent severe, moderate and mild dementia respectively. Interventions included were those involving any form of physical activity or exercise, such as walking or strength exercises. Articles which did not report the severity of dementia in the study population were excluded. There were no exclusion criteria for the type, frequency, intensity or duration of exercise. The comparator was usual care or no treatment control.
Primary outcomes were physical outcome measures, including timed up and go, 30 second s t a n d a n d a 4 m a n d a  independently to determine whether they met the inclusion criteria.
Data were extracted using the Cochrane data extraction form [14], which was piloted before revision to extract data on population characteristics, settings, study methods used, intervention and controls used, outcome measures and any effect sizes shown. Two reviewers (AL and KR) extracted data independently and subsequently came together to discuss results and findings.
Each included study was assessed for risk of bias by the two reviewers using the Cochrane Collaboration Tool for assessing Risk of Bias [14]. Blinding was only assessed in relation to outcome assessors, because the nature of exercise interventions characteristically makes blinding of subjects and researchers delivering the intervention difficult.
The raw data on treatment effect was extracted as means for the intervention groups and compared with the control groups with 95% confidence intervals (CIs).  Table 1  in all studies to establish criteria for judging level of dementia. There was variability in the cut-off points used in different studies so, for consistency, the cut-off points for mild (17 -26), moderate (10 -17) and severe (<10) dementia were taken from a previous Cochrane Review [6].
A summary of risk of bias isare presented in Figures 2 and 3 below. The methods used to generate the allocation sequence were well described in five trials [15,16,, allocation concealment was adequate in four trials [15,1718,1819,2021] and outcome assessors were blinded in five six trials. These attributes were unclear for the remaining studies. Attrition rates varied from 0% to 46% in the included trials. One trial did not specify the drop-out rates for control and intervention arms separately [2122]. The dropout rates were higher in the control group than the intervention group in five six trials [15--1198, 2021]. Kemoun [1920] was the only study where attrition was higher in the intervention group. All studies provided reasons for attrition including: death, medical reasons, no longer resident, not adhering, no longer interested/declining to participate further, and hospitalisation.
Intention-to-treat analysis was used in two trials [15,1819]. There was no selective reporting bias with all included trials reporting on all planned outcome measures.

Primary Outcomes
Eight physical and four mood outcome measures were used in the included studies but none of the reported outcome measures were the same between studies. Six Seven studies looked at the physical effects of the intervention with two also looking at the effect on mood. One reported using the 6-minute walk as a measure of functional exercise capacity, but reported on the effect of the intervention on mood only.

Physical Effects
The results of the 6 7 studies  which looked at the physical effects of the intervention are summarised in table 2.
Three studies [16,-18,19] used an endurance intervention -either walking or recumbent cycling. There was a mixed effect on endurance outcomes with one study [187] demonstrating an improvement in the six minute walk test (20% increase in walking distance, p <0.05), whilst another found no effect on the 2-minute walking test performance [16]. One study used the timed up-and-go test, which mainly assesses gait and balance, as an outcome and did find a significant difference in favour of the intervention (F = 5.43, p = 0.03) [18]. There was risk of bias in all three studies, and not all studies included an endurance outcome measure. We consider the evidence with regard to endurance to be of low quality. Two Three studies [15,17,1920] looked at multi-component exercise programmes which included strength, endurance and balance exercises. Both Two studies [15,20] used walking speed as an outcome measure (10m walk and 6m walk respectively) and both found that there was a significant increase in gait speed associated with the intervention (increase of 1.02 m/s, p< 0.01 and increase of 0.41 m/s, p = 0.02 respectively). One of these [15] also used the timed get up-and-go outcome measure and found no improvement. The third study [17] used an endurance outcome measure (6 min walk test) and found no significant differences between the groups. Both All studies used appropriate outcome measures for their interventions but there was a significant risk of bias in one study [1920]. Overall, the evidence here was, again, of low quality.

Mood
The results of the 3 studies which looked at the effects on mood [15,1819,2122] are presented in table 3 below.
Two studies [15,1819] found no significant difference in mood between intervention and control using the Cornell Scale for Depression in dementia and the Montgomery-Asberg Depression Rating Scale respectively. One study [2122] showed a mixed pattern with no significant differences seen between the groups for the positive domains (p = 0.38) but some differences for the negative domains of the Alzheimer's Mood Scale (p = 0.004).
There were also some differences, favouring the intervention, in Dementia Mood Assessment scores (p = 0.007). Means and standard deviations were not provided for this study. There was risk of bias in all three of these studies with incomplete data for one, [2122] with a range of different measures used. Hence, this was considered to be very low-quality evidence.

Quality of Life
No included studies used quality of life as an outcome measure.

Heterogeneity
The included studies were clinically and/or methodologically heterogeneous, hence neither a meta-analysis nor tests for statistical heterogeneity were appropriate.

Discussion
This systematic review has extended our understanding of the evidence on the effectiveness of exercise in people with dementia by focusing specifically on studies which aimed to support those with moderate to severe dementia. Only seven eight studies were suitable for inclusion in the review. The interventions were extremely variable in both content and duration and, with one exception [1718], were targeted at all people with dementia without respect to severity. There was some evidence that exercise programs may have a role in improving the physical function of people with moderate to severe dementia [15, but there was little evidence to suggest that exercise programs may improve mood. Overall, the evidence wasMost of the studies retrieved were of either very low, or low, quality.
The strengths of this review are that it was conducted systematically according to the guidelines laid out in the Cochrane Handbook for Systematic Reviews of Interventions [14] to minimise bias during the review process. Two reviewers independently extracted the data and assessed risk of bias and the protocol was registered on PROSPERO prior to searches commencing. Hand searching of reference lists was used to maximise coverage.
As with all such reviews, there may be additional research classified under alternate search headings that was not included. It is also possible that important research conducted using methodologies other than randomised controlled trials was missed. A limitation is that different types of physical outcome measure were included in the analysis, including those, such as endurance, which would be described by some authors as physical performance rather than physical function measures. We have not, however, conflated results using different types of measures and presenting these studies together here serves to underscore how little work has been done regarding exercise in people with moderate to severe dementia. There are acknowledged limitations of using the MMSEwhich was used in most of the included studies -to classify severity of dementia, in part because of its lack of sensitivity to change and in part because it focusses exclusively on progression of cognitive symptoms [223]. We are, however, as limited as reviewers to what is available in the published literature. Including up to 30% of patients with mild dementia may have skewed our findings somewhat. A broader limitation across the literature is that most studies did not consider whether statistically significant improvements were clinically meaningful -more work is required as part of empirical studies with patient, family carer and professional consultees, to work out what constitutes clinically meaningful change.
All previous reviews have focused mainly on people with mild to moderate cognitive impairment with very few trials containing participants with moderate to severe dementia.
Depression was considered by two previous reviews [6,2342] and found no clear evidence of the benefit of exercise for depression in people with dementia. Three reviews looked at physical function  and again showed similar results to those identified here, with no clear evidence of benefit. One review [2342]

Author Contributions
All authors wrote the protocol for the review, contributed to the analysis and wrote the manuscript. AL and KR conducted the review and led the analysis.     Other bias 0% 25% 50% 75% 100%

Funding Tables
Low risk of bias Unclear risk of bias High risk of bias