- Split View
-
Views
-
Cite
Cite
Jacqueline Kathleen Kueper, Mark Speechley, Navena Rebecca Lingum, Manuel Montero-Odasso, Motor function and incident dementia: a systematic review and meta-analysis, Age and Ageing, Volume 46, Issue 5, September 2017, Pages 729–738, https://doi.org/10.1093/ageing/afx084
- Share Icon Share
Abstract
cognitive and mobility decline are interrelated processes, whereby mobility decline coincides or precedes the onset of cognitive decline.
to assess whether there is an association between performance on motor function tests and incident dementia.
electronic database, grey literature and hand searching identified studies testing for associations between baseline motor function and incident dementia in older adults.
of 2,540 potentially relevant documents, 37 met the final inclusion criteria and were reviewed qualitatively. Three meta-analyses were conducted using data from 10 studies. Three main motor domains—upper limb motor function, parkinsonism and lower limb motor function—emerged as associated with increased risk of incident dementia. Studies including older adults without neurological overt disease found a higher risk of incident dementia associated with poorer performance on composite motor function scores, balance and gait velocity (meta-analysis pooled HR = 1.94, 95% CI: 1.41, 2.65). Mixed results were found across different study samples for upper limb motor function, overall parkinsonism (meta-analysis pooled OR = 3.05, 95% CI: 1.31, 7.08), bradykinesia and rigidity. Studies restricted to older adults with Parkinson's Disease found weak or no association with incident dementia even for motor domains highly associated in less restrictive samples. Tremor was not associated with an increased risk of dementia in any population (meta-analysis pooled HR = 0.80, 95% CI 0.31, 2.03).
lower limb motor function was associated with increased risk of developing dementia, while tremor and hand grip strength were not. Our results support future research investigating the inclusion of quantitative motor assessment, specifically gait velocity tests, for clinical dementia risk evaluation.
Introduction
Cognition and motor function decline with ageing and this decline often coexists in the pathway to insidious disability [1–3]. Recently, it has been shown that motor impairments precede cognitive impairment and motor changes have been proposed as potential clinical biomarkers to help predict dementia syndromes [4, 5]. Several motor domains have been associated with cognition, with lower limb motor performance being more commonly studied. For example, changes in gait velocity have been found 12 years before clinical diagnosis of mild cognitive impairment (MCI), a dementia prodromal stage [6].
Previous systematic reviews of potential dementia biomarkers have focused on demographics, genomics, lifestyle behaviours, vascular factors, neurological and neuropsychological factors [7, 8]. One investigated gait, but slow gait classification occasionally required cognitive impairment [9]. A comprehensive systematic review of motor function as a candidate biomarker for dementia has not been performed. Our objective was to systematically evaluate whether older adults with poorer performance on motor function tests are at an increased risk of incident dementia compared to those with better performance.
Methods
Search strategy
On 8 June 2016, literature searches were conducted in MedLine, EMBASE, CINAHL and PsychInfo databases using subject headings and keywords related to motor function, risk or association and dementia. Limits included older adults and documents written in English. Grey literature and hand searching was also performed. Search strategies and document sources are in Supplementary data, Table S1, Appendix 1, available at Age and Ageing online.
Documents were screened for relevance in a two-step process by two independent reviewers. Cohen's kappa coefficient (κ) assessed level of inter-rater agreement at each step [10]. Disagreements were resolved by discussion until consensus. Multiple studies using the same data were included only if they used different subsamples, or were exploring different exposures. Otherwise, only the document providing the greater amount of information was retained.
Eligibility criteria
Eligibility criteria included (i) original research, (ii) assess motor function, (iii) measure incident dementia, (iv) adult human participants and (v) English language. Study quality was assessed using the Newcastle–Ottawa Quality Assessment Scale (NOS) [11].
Data extraction
The primary reviewer extracted from each article the first author, publication year, study design, inclusion/exclusion criteria, population, location, sample size, demographics and follow-up. Motor assessment details, quantification method (i.e. scoring and criteria to classify motor performance), incident dementia cases, effect estimates, statistical analyses, covariates and subgroup analyses on dementia subtypes were also extracted.
Meta-analysis
Random-effects meta-analyses were performed using Comprehensive Meta-Analysis software whenever there were a minimum of three compatible effect estimates for a single motor domain [12]. Effect estimates were manually calculated when needed. When a study reported multiple effect estimates for a motor domain, the result most comparable with other studies, and/or with the longest follow-up, and/or adjusted for the largest number of covariates was selected. Between-study variability was assessed using the I2 statistic, and publication bias with funnel plots.
Results
Study selection
A total of 4,709 documents were located and 2,169 duplicates removed, leaving 2,540 unique documents to screen. Seventy-eight documents passed through title and abstract screening (κ = 0.52) to full-text screening (κ = 0.73) (Figure 1). Multiple analyses of the same cohort were included only if they used different subsamples or a different exposure of interest. Thirty-seven documents were retained for inclusion. Three meta-analyses were conducted using data from 10 studies.
Quality assessment
Studies obtained a variety of ratings on the NOS (see Supplementary data, Table S2 Appendix 2, available at Age and Ageing online). Study quality based on NOS ratings are summarised throughout the results section.
Summary of results by motor domain
Complete study characteristics (Table S3) and results stratified by motor domains (Table S4) are in Supplementary data, Appendix 3, available at Age and Ageing online. Table 1 summarises included studies with key results.
Author [Refs] . | Study design . | n . | Mean age (years) . | Follow-up (years) . | NOS score [/9] . | Association between motor function and dementia . |
---|---|---|---|---|---|---|
Aggarwal [24] | Prospective cohort | 189 | 79 | 10 | 8 |
|
Albala [25] | Prospective cohort | 1575 | Range 65–68 | 2 | 6 | Worse Gait Velocity: RR = 1.22 (1.03,1.45), P = 0.022a,b,c,f |
Amieva [26] | RCT | 90 | Dem 73 No-Dem 69 | 2 | 5 | Finger Tapping: P = 0.62 |
Anang [19] | Prospective cohort | 80 | 66 | = 4.4 | 9 | Better Manual Dexterity: OR = 0.67 (0.48,0.94), P = 0.049a,b,g,hWorse Overall Parkinsonism: OR = 0.97 (0.92,1.01), P = 0.21a,b,g,hSlower Gait Velocity: OR = 1.09 (0.84,1.40), P = 0.60a,b,g,h |
Bermejo-Pareja [39] | Case-control within prospective cohort | Cases 206 | Cases 75 | 6.6 | 8 | Tremor Present: RR = 1.50 (0.88,2.54), P = 0.14a,c,i,j,k,l(Note: Cases = Essential Tremor, Control = No Essential Tremor) |
Control 3,685 | ||||||
Control 73 | ||||||
Buchman [28] | Prospective cohort | 877 | AD 79 | = 5.7 | 8 | Better Grip Strength: HR = 0.99 (0.98,1.00)f,i,j,m,n,o,p (AD Only) |
No-AD 74 | ||||||
Buchman [18] | Prospective cohort | 919 | 80 | = 4.7 | 9 | Worse Overall Parkinsonism: HR = 1.35 (1.14,1.59)a,b,c (AD Only) |
Bugalho [34] | Prospective cohort | 61 | 72 | 2 | 6 |
|
Camargo [17] | Prospective cohort | 2046 | 62 | 11 | 9 |
|
Camicioli [40] | Prospective cohort | Wave 1: 538 | 80 | 5, Each wave | 7 |
|
Wave 2: 497 | ||||||
Domellöf [41] | Prospective cohort | 49 | 71 | 5 | 8 |
|
Duara [42] | Prospective cohort | 115 | Range 52–91 | 3 | 7 | Overall Parkinsonism Severity: F[1, 182] = 14.46, P < 0.001 |
Dumurgier [16] | Prospective cohort | 3663 | 75 | 9 | 9 | Worse Gait Velocity: HR = 1.59 (1.39,1.81), P < 0.001a,b |
Gago [35] | Prospective cohort | 24 | Dem 67 No-Dem 62 | 6 | 6 |
|
Grey [43] | Prospective cohort | 2619 | 77 | = 6.5 | 8 |
|
Hobson [36] | Prospective cohort | 51 | 74 | = 4.36 | 6 | Worse Overall Parkinsonism: RR = 1.33 (0.99,1.78) |
Israeli-Korn [29] | Prospective cohort | 111 | 73 | = 3.9 | 6 |
|
Lee [38] | Retrospective cohort | 1775 | No-Dem 73 | 6 | 7 | Poor Balance: OR = 2.27 (1.53,3.37), P < 0.001a |
Dem 75 (Med) | ||||||
Lee [30] | Prospective cohort | 96 | 71 | 13.6 | 6 |
|
Levy [20] | Prospective cohort | 173 | 74 | = 3.6 | 9 |
|
Louis [15] | Prospective cohort | 1028 | 78 | 13 | 9 | Worse Overall Parkinsonism: RR = 1.08 (1.01,1.16), P = 0.02a,b,c,e,j,k,l,n,r,s |
Louis [44] | Prospective cohort | 1851 | 76 | 9 | 8 | Overall Parkinsonism Present: HR = 1.98 (1.37,2.88), P < 0.001a,c,e,i,n,t,u |
Montero-Odasso [45] | Prospective cohort | 252 | 77 | 5 | 7 | Slow Gait Velocity: HR = 4.93 (1.71,14.21), P = 0.003a,b,c,n |
Ramakers [49] | Case-control | Cases 74 | Cases 79 Control 79 | 5 | 6 |
|
Control 125 | ||||||
Shill [46] | Prospective cohort | 507 | ET 80 | 12.1 | 8 | Tremor Present: HR = 0.46 (0.17,1.23), P = 0.12a,b,d |
Control 77 | ||||||
Taaffe [31] | Prospective cohort | 2263 | Dem 79 | = 6.1 | 6 | Global Physical Function: P < 0.001a |
No-Dem 76 | ||||||
Thawani [14] | Prospective cohort | 2056 | 78 | = 3.8 | 9 | Tremor Present: HR = 1.71 (0.97,3.01), P = 0.06a,c,r,t |
Verghese [21] | Prospective cohort | 422 | Range 75–80 | Median = 6.6 | 9 | Abnormal Neurological Gait Pattern: HR = 1.96 (1.30,2.96)a,b,c,m,j,s,e,n |
Verghese [13] | Prospective cohort | 399 | Ab. gait 80 | 5 | 9 | High-Risk Neurological Gait Syndrome Present: HR = 2.66 (1.69,4.18), P = 0.03a,b,c,e (VaD Only) |
Nor. gait 79 | ||||||
Verghese [47] | Prospective cohort | 399 | 77 | 5 | 7 | Slower Gait Velocity: HR = 1.30 (0.95,1.78)a,b,c |
Verghese [48] | Prospective cohort | 767 | 80 | 9.1 | 8 | Slow Gait Velocity: HR = 1.7 (0.8,3.2) |
Verghese [27] | Multiple cohorts | 4550 | Range 60–108 | 12 | 8 | Slow Gait Velocity: HR = 1.77 (1.38,2.27)a,b,c,m,s,u |
Waite [32] | Prospective cohort | 394 | N/A | 6 | 6 | Overall Parkinsonism Present: OR = 1.4 (0.6,3.2) |
Wang [23] | Prospective cohort | 2288 | Dem 79 | 5.9 | 9 |
|
No-Dem 74 | ||||||
Welmer [22] | Prospective cohort | 2232 | 72 | 6 | 9 | Worse Gait Velocity: OR = 1.45 (1.17,1.80)a,c,m,e,w |
Wilson [33] | Prospective cohort | 746 | 75 | 8 | 8 | Worse Overall Parkinsonism: HR = 1.04 (1.02,1.07)a,b,c |
Zhu [37] | Prospective cohort | 261 | 58 | 5 | 6 | Worse PIGD: HR = 1.04 (0.82,1.33), P = 0.72a,c,t,i,g |
Author [Refs] . | Study design . | n . | Mean age (years) . | Follow-up (years) . | NOS score [/9] . | Association between motor function and dementia . |
---|---|---|---|---|---|---|
Aggarwal [24] | Prospective cohort | 189 | 79 | 10 | 8 |
|
Albala [25] | Prospective cohort | 1575 | Range 65–68 | 2 | 6 | Worse Gait Velocity: RR = 1.22 (1.03,1.45), P = 0.022a,b,c,f |
Amieva [26] | RCT | 90 | Dem 73 No-Dem 69 | 2 | 5 | Finger Tapping: P = 0.62 |
Anang [19] | Prospective cohort | 80 | 66 | = 4.4 | 9 | Better Manual Dexterity: OR = 0.67 (0.48,0.94), P = 0.049a,b,g,hWorse Overall Parkinsonism: OR = 0.97 (0.92,1.01), P = 0.21a,b,g,hSlower Gait Velocity: OR = 1.09 (0.84,1.40), P = 0.60a,b,g,h |
Bermejo-Pareja [39] | Case-control within prospective cohort | Cases 206 | Cases 75 | 6.6 | 8 | Tremor Present: RR = 1.50 (0.88,2.54), P = 0.14a,c,i,j,k,l(Note: Cases = Essential Tremor, Control = No Essential Tremor) |
Control 3,685 | ||||||
Control 73 | ||||||
Buchman [28] | Prospective cohort | 877 | AD 79 | = 5.7 | 8 | Better Grip Strength: HR = 0.99 (0.98,1.00)f,i,j,m,n,o,p (AD Only) |
No-AD 74 | ||||||
Buchman [18] | Prospective cohort | 919 | 80 | = 4.7 | 9 | Worse Overall Parkinsonism: HR = 1.35 (1.14,1.59)a,b,c (AD Only) |
Bugalho [34] | Prospective cohort | 61 | 72 | 2 | 6 |
|
Camargo [17] | Prospective cohort | 2046 | 62 | 11 | 9 |
|
Camicioli [40] | Prospective cohort | Wave 1: 538 | 80 | 5, Each wave | 7 |
|
Wave 2: 497 | ||||||
Domellöf [41] | Prospective cohort | 49 | 71 | 5 | 8 |
|
Duara [42] | Prospective cohort | 115 | Range 52–91 | 3 | 7 | Overall Parkinsonism Severity: F[1, 182] = 14.46, P < 0.001 |
Dumurgier [16] | Prospective cohort | 3663 | 75 | 9 | 9 | Worse Gait Velocity: HR = 1.59 (1.39,1.81), P < 0.001a,b |
Gago [35] | Prospective cohort | 24 | Dem 67 No-Dem 62 | 6 | 6 |
|
Grey [43] | Prospective cohort | 2619 | 77 | = 6.5 | 8 |
|
Hobson [36] | Prospective cohort | 51 | 74 | = 4.36 | 6 | Worse Overall Parkinsonism: RR = 1.33 (0.99,1.78) |
Israeli-Korn [29] | Prospective cohort | 111 | 73 | = 3.9 | 6 |
|
Lee [38] | Retrospective cohort | 1775 | No-Dem 73 | 6 | 7 | Poor Balance: OR = 2.27 (1.53,3.37), P < 0.001a |
Dem 75 (Med) | ||||||
Lee [30] | Prospective cohort | 96 | 71 | 13.6 | 6 |
|
Levy [20] | Prospective cohort | 173 | 74 | = 3.6 | 9 |
|
Louis [15] | Prospective cohort | 1028 | 78 | 13 | 9 | Worse Overall Parkinsonism: RR = 1.08 (1.01,1.16), P = 0.02a,b,c,e,j,k,l,n,r,s |
Louis [44] | Prospective cohort | 1851 | 76 | 9 | 8 | Overall Parkinsonism Present: HR = 1.98 (1.37,2.88), P < 0.001a,c,e,i,n,t,u |
Montero-Odasso [45] | Prospective cohort | 252 | 77 | 5 | 7 | Slow Gait Velocity: HR = 4.93 (1.71,14.21), P = 0.003a,b,c,n |
Ramakers [49] | Case-control | Cases 74 | Cases 79 Control 79 | 5 | 6 |
|
Control 125 | ||||||
Shill [46] | Prospective cohort | 507 | ET 80 | 12.1 | 8 | Tremor Present: HR = 0.46 (0.17,1.23), P = 0.12a,b,d |
Control 77 | ||||||
Taaffe [31] | Prospective cohort | 2263 | Dem 79 | = 6.1 | 6 | Global Physical Function: P < 0.001a |
No-Dem 76 | ||||||
Thawani [14] | Prospective cohort | 2056 | 78 | = 3.8 | 9 | Tremor Present: HR = 1.71 (0.97,3.01), P = 0.06a,c,r,t |
Verghese [21] | Prospective cohort | 422 | Range 75–80 | Median = 6.6 | 9 | Abnormal Neurological Gait Pattern: HR = 1.96 (1.30,2.96)a,b,c,m,j,s,e,n |
Verghese [13] | Prospective cohort | 399 | Ab. gait 80 | 5 | 9 | High-Risk Neurological Gait Syndrome Present: HR = 2.66 (1.69,4.18), P = 0.03a,b,c,e (VaD Only) |
Nor. gait 79 | ||||||
Verghese [47] | Prospective cohort | 399 | 77 | 5 | 7 | Slower Gait Velocity: HR = 1.30 (0.95,1.78)a,b,c |
Verghese [48] | Prospective cohort | 767 | 80 | 9.1 | 8 | Slow Gait Velocity: HR = 1.7 (0.8,3.2) |
Verghese [27] | Multiple cohorts | 4550 | Range 60–108 | 12 | 8 | Slow Gait Velocity: HR = 1.77 (1.38,2.27)a,b,c,m,s,u |
Waite [32] | Prospective cohort | 394 | N/A | 6 | 6 | Overall Parkinsonism Present: OR = 1.4 (0.6,3.2) |
Wang [23] | Prospective cohort | 2288 | Dem 79 | 5.9 | 9 |
|
No-Dem 74 | ||||||
Welmer [22] | Prospective cohort | 2232 | 72 | 6 | 9 | Worse Gait Velocity: OR = 1.45 (1.17,1.80)a,c,m,e,w |
Wilson [33] | Prospective cohort | 746 | 75 | 8 | 8 | Worse Overall Parkinsonism: HR = 1.04 (1.02,1.07)a,b,c |
Zhu [37] | Prospective cohort | 261 | 58 | 5 | 6 | Worse PIGD: HR = 1.04 (0.82,1.33), P = 0.72a,c,t,i,g |
Ref, Reference; n, Sample Size; NOS, Newcastle–Ottawa Quality Assessment Scale; RR, Relative Risk; PIGD, Postural Instability Gait Difficulty; RCT, Randomised Controlled Trial; Dem, Dementia; , Mean; OR, Odds Ratio; AD, Alzheimer's disease; HR, Hazard Ratio; ET, Essential Tremor; Ab, Abnormal; Nor, Normal; VaD, Vascular Dementia; N/A, not available. Covariates adjusted for: a = Age; b = Sex or Gender; c = Education; d = APOEe4 allele; e = Stroke; f = Physical activity; g = duration or severity of PD; h = duration of follow-up; i = Mental Health (e.g. Depression); j = Vascular Factors (e.g. hypertension); k = smoking; l = alcohol consumption; m = Cognition; n = Other Comorbidities (e.g. diabetes); o = late-life social networks; p = early socio-economic status; q = BMI or waist-to-hip ratio; r = Race or Ethnicity; s = Cardiovascular Diseases; t = Medication; u = cohort source; v = Family history of AD; w = pain.
Author [Refs] . | Study design . | n . | Mean age (years) . | Follow-up (years) . | NOS score [/9] . | Association between motor function and dementia . |
---|---|---|---|---|---|---|
Aggarwal [24] | Prospective cohort | 189 | 79 | 10 | 8 |
|
Albala [25] | Prospective cohort | 1575 | Range 65–68 | 2 | 6 | Worse Gait Velocity: RR = 1.22 (1.03,1.45), P = 0.022a,b,c,f |
Amieva [26] | RCT | 90 | Dem 73 No-Dem 69 | 2 | 5 | Finger Tapping: P = 0.62 |
Anang [19] | Prospective cohort | 80 | 66 | = 4.4 | 9 | Better Manual Dexterity: OR = 0.67 (0.48,0.94), P = 0.049a,b,g,hWorse Overall Parkinsonism: OR = 0.97 (0.92,1.01), P = 0.21a,b,g,hSlower Gait Velocity: OR = 1.09 (0.84,1.40), P = 0.60a,b,g,h |
Bermejo-Pareja [39] | Case-control within prospective cohort | Cases 206 | Cases 75 | 6.6 | 8 | Tremor Present: RR = 1.50 (0.88,2.54), P = 0.14a,c,i,j,k,l(Note: Cases = Essential Tremor, Control = No Essential Tremor) |
Control 3,685 | ||||||
Control 73 | ||||||
Buchman [28] | Prospective cohort | 877 | AD 79 | = 5.7 | 8 | Better Grip Strength: HR = 0.99 (0.98,1.00)f,i,j,m,n,o,p (AD Only) |
No-AD 74 | ||||||
Buchman [18] | Prospective cohort | 919 | 80 | = 4.7 | 9 | Worse Overall Parkinsonism: HR = 1.35 (1.14,1.59)a,b,c (AD Only) |
Bugalho [34] | Prospective cohort | 61 | 72 | 2 | 6 |
|
Camargo [17] | Prospective cohort | 2046 | 62 | 11 | 9 |
|
Camicioli [40] | Prospective cohort | Wave 1: 538 | 80 | 5, Each wave | 7 |
|
Wave 2: 497 | ||||||
Domellöf [41] | Prospective cohort | 49 | 71 | 5 | 8 |
|
Duara [42] | Prospective cohort | 115 | Range 52–91 | 3 | 7 | Overall Parkinsonism Severity: F[1, 182] = 14.46, P < 0.001 |
Dumurgier [16] | Prospective cohort | 3663 | 75 | 9 | 9 | Worse Gait Velocity: HR = 1.59 (1.39,1.81), P < 0.001a,b |
Gago [35] | Prospective cohort | 24 | Dem 67 No-Dem 62 | 6 | 6 |
|
Grey [43] | Prospective cohort | 2619 | 77 | = 6.5 | 8 |
|
Hobson [36] | Prospective cohort | 51 | 74 | = 4.36 | 6 | Worse Overall Parkinsonism: RR = 1.33 (0.99,1.78) |
Israeli-Korn [29] | Prospective cohort | 111 | 73 | = 3.9 | 6 |
|
Lee [38] | Retrospective cohort | 1775 | No-Dem 73 | 6 | 7 | Poor Balance: OR = 2.27 (1.53,3.37), P < 0.001a |
Dem 75 (Med) | ||||||
Lee [30] | Prospective cohort | 96 | 71 | 13.6 | 6 |
|
Levy [20] | Prospective cohort | 173 | 74 | = 3.6 | 9 |
|
Louis [15] | Prospective cohort | 1028 | 78 | 13 | 9 | Worse Overall Parkinsonism: RR = 1.08 (1.01,1.16), P = 0.02a,b,c,e,j,k,l,n,r,s |
Louis [44] | Prospective cohort | 1851 | 76 | 9 | 8 | Overall Parkinsonism Present: HR = 1.98 (1.37,2.88), P < 0.001a,c,e,i,n,t,u |
Montero-Odasso [45] | Prospective cohort | 252 | 77 | 5 | 7 | Slow Gait Velocity: HR = 4.93 (1.71,14.21), P = 0.003a,b,c,n |
Ramakers [49] | Case-control | Cases 74 | Cases 79 Control 79 | 5 | 6 |
|
Control 125 | ||||||
Shill [46] | Prospective cohort | 507 | ET 80 | 12.1 | 8 | Tremor Present: HR = 0.46 (0.17,1.23), P = 0.12a,b,d |
Control 77 | ||||||
Taaffe [31] | Prospective cohort | 2263 | Dem 79 | = 6.1 | 6 | Global Physical Function: P < 0.001a |
No-Dem 76 | ||||||
Thawani [14] | Prospective cohort | 2056 | 78 | = 3.8 | 9 | Tremor Present: HR = 1.71 (0.97,3.01), P = 0.06a,c,r,t |
Verghese [21] | Prospective cohort | 422 | Range 75–80 | Median = 6.6 | 9 | Abnormal Neurological Gait Pattern: HR = 1.96 (1.30,2.96)a,b,c,m,j,s,e,n |
Verghese [13] | Prospective cohort | 399 | Ab. gait 80 | 5 | 9 | High-Risk Neurological Gait Syndrome Present: HR = 2.66 (1.69,4.18), P = 0.03a,b,c,e (VaD Only) |
Nor. gait 79 | ||||||
Verghese [47] | Prospective cohort | 399 | 77 | 5 | 7 | Slower Gait Velocity: HR = 1.30 (0.95,1.78)a,b,c |
Verghese [48] | Prospective cohort | 767 | 80 | 9.1 | 8 | Slow Gait Velocity: HR = 1.7 (0.8,3.2) |
Verghese [27] | Multiple cohorts | 4550 | Range 60–108 | 12 | 8 | Slow Gait Velocity: HR = 1.77 (1.38,2.27)a,b,c,m,s,u |
Waite [32] | Prospective cohort | 394 | N/A | 6 | 6 | Overall Parkinsonism Present: OR = 1.4 (0.6,3.2) |
Wang [23] | Prospective cohort | 2288 | Dem 79 | 5.9 | 9 |
|
No-Dem 74 | ||||||
Welmer [22] | Prospective cohort | 2232 | 72 | 6 | 9 | Worse Gait Velocity: OR = 1.45 (1.17,1.80)a,c,m,e,w |
Wilson [33] | Prospective cohort | 746 | 75 | 8 | 8 | Worse Overall Parkinsonism: HR = 1.04 (1.02,1.07)a,b,c |
Zhu [37] | Prospective cohort | 261 | 58 | 5 | 6 | Worse PIGD: HR = 1.04 (0.82,1.33), P = 0.72a,c,t,i,g |
Author [Refs] . | Study design . | n . | Mean age (years) . | Follow-up (years) . | NOS score [/9] . | Association between motor function and dementia . |
---|---|---|---|---|---|---|
Aggarwal [24] | Prospective cohort | 189 | 79 | 10 | 8 |
|
Albala [25] | Prospective cohort | 1575 | Range 65–68 | 2 | 6 | Worse Gait Velocity: RR = 1.22 (1.03,1.45), P = 0.022a,b,c,f |
Amieva [26] | RCT | 90 | Dem 73 No-Dem 69 | 2 | 5 | Finger Tapping: P = 0.62 |
Anang [19] | Prospective cohort | 80 | 66 | = 4.4 | 9 | Better Manual Dexterity: OR = 0.67 (0.48,0.94), P = 0.049a,b,g,hWorse Overall Parkinsonism: OR = 0.97 (0.92,1.01), P = 0.21a,b,g,hSlower Gait Velocity: OR = 1.09 (0.84,1.40), P = 0.60a,b,g,h |
Bermejo-Pareja [39] | Case-control within prospective cohort | Cases 206 | Cases 75 | 6.6 | 8 | Tremor Present: RR = 1.50 (0.88,2.54), P = 0.14a,c,i,j,k,l(Note: Cases = Essential Tremor, Control = No Essential Tremor) |
Control 3,685 | ||||||
Control 73 | ||||||
Buchman [28] | Prospective cohort | 877 | AD 79 | = 5.7 | 8 | Better Grip Strength: HR = 0.99 (0.98,1.00)f,i,j,m,n,o,p (AD Only) |
No-AD 74 | ||||||
Buchman [18] | Prospective cohort | 919 | 80 | = 4.7 | 9 | Worse Overall Parkinsonism: HR = 1.35 (1.14,1.59)a,b,c (AD Only) |
Bugalho [34] | Prospective cohort | 61 | 72 | 2 | 6 |
|
Camargo [17] | Prospective cohort | 2046 | 62 | 11 | 9 |
|
Camicioli [40] | Prospective cohort | Wave 1: 538 | 80 | 5, Each wave | 7 |
|
Wave 2: 497 | ||||||
Domellöf [41] | Prospective cohort | 49 | 71 | 5 | 8 |
|
Duara [42] | Prospective cohort | 115 | Range 52–91 | 3 | 7 | Overall Parkinsonism Severity: F[1, 182] = 14.46, P < 0.001 |
Dumurgier [16] | Prospective cohort | 3663 | 75 | 9 | 9 | Worse Gait Velocity: HR = 1.59 (1.39,1.81), P < 0.001a,b |
Gago [35] | Prospective cohort | 24 | Dem 67 No-Dem 62 | 6 | 6 |
|
Grey [43] | Prospective cohort | 2619 | 77 | = 6.5 | 8 |
|
Hobson [36] | Prospective cohort | 51 | 74 | = 4.36 | 6 | Worse Overall Parkinsonism: RR = 1.33 (0.99,1.78) |
Israeli-Korn [29] | Prospective cohort | 111 | 73 | = 3.9 | 6 |
|
Lee [38] | Retrospective cohort | 1775 | No-Dem 73 | 6 | 7 | Poor Balance: OR = 2.27 (1.53,3.37), P < 0.001a |
Dem 75 (Med) | ||||||
Lee [30] | Prospective cohort | 96 | 71 | 13.6 | 6 |
|
Levy [20] | Prospective cohort | 173 | 74 | = 3.6 | 9 |
|
Louis [15] | Prospective cohort | 1028 | 78 | 13 | 9 | Worse Overall Parkinsonism: RR = 1.08 (1.01,1.16), P = 0.02a,b,c,e,j,k,l,n,r,s |
Louis [44] | Prospective cohort | 1851 | 76 | 9 | 8 | Overall Parkinsonism Present: HR = 1.98 (1.37,2.88), P < 0.001a,c,e,i,n,t,u |
Montero-Odasso [45] | Prospective cohort | 252 | 77 | 5 | 7 | Slow Gait Velocity: HR = 4.93 (1.71,14.21), P = 0.003a,b,c,n |
Ramakers [49] | Case-control | Cases 74 | Cases 79 Control 79 | 5 | 6 |
|
Control 125 | ||||||
Shill [46] | Prospective cohort | 507 | ET 80 | 12.1 | 8 | Tremor Present: HR = 0.46 (0.17,1.23), P = 0.12a,b,d |
Control 77 | ||||||
Taaffe [31] | Prospective cohort | 2263 | Dem 79 | = 6.1 | 6 | Global Physical Function: P < 0.001a |
No-Dem 76 | ||||||
Thawani [14] | Prospective cohort | 2056 | 78 | = 3.8 | 9 | Tremor Present: HR = 1.71 (0.97,3.01), P = 0.06a,c,r,t |
Verghese [21] | Prospective cohort | 422 | Range 75–80 | Median = 6.6 | 9 | Abnormal Neurological Gait Pattern: HR = 1.96 (1.30,2.96)a,b,c,m,j,s,e,n |
Verghese [13] | Prospective cohort | 399 | Ab. gait 80 | 5 | 9 | High-Risk Neurological Gait Syndrome Present: HR = 2.66 (1.69,4.18), P = 0.03a,b,c,e (VaD Only) |
Nor. gait 79 | ||||||
Verghese [47] | Prospective cohort | 399 | 77 | 5 | 7 | Slower Gait Velocity: HR = 1.30 (0.95,1.78)a,b,c |
Verghese [48] | Prospective cohort | 767 | 80 | 9.1 | 8 | Slow Gait Velocity: HR = 1.7 (0.8,3.2) |
Verghese [27] | Multiple cohorts | 4550 | Range 60–108 | 12 | 8 | Slow Gait Velocity: HR = 1.77 (1.38,2.27)a,b,c,m,s,u |
Waite [32] | Prospective cohort | 394 | N/A | 6 | 6 | Overall Parkinsonism Present: OR = 1.4 (0.6,3.2) |
Wang [23] | Prospective cohort | 2288 | Dem 79 | 5.9 | 9 |
|
No-Dem 74 | ||||||
Welmer [22] | Prospective cohort | 2232 | 72 | 6 | 9 | Worse Gait Velocity: OR = 1.45 (1.17,1.80)a,c,m,e,w |
Wilson [33] | Prospective cohort | 746 | 75 | 8 | 8 | Worse Overall Parkinsonism: HR = 1.04 (1.02,1.07)a,b,c |
Zhu [37] | Prospective cohort | 261 | 58 | 5 | 6 | Worse PIGD: HR = 1.04 (0.82,1.33), P = 0.72a,c,t,i,g |
Ref, Reference; n, Sample Size; NOS, Newcastle–Ottawa Quality Assessment Scale; RR, Relative Risk; PIGD, Postural Instability Gait Difficulty; RCT, Randomised Controlled Trial; Dem, Dementia; , Mean; OR, Odds Ratio; AD, Alzheimer's disease; HR, Hazard Ratio; ET, Essential Tremor; Ab, Abnormal; Nor, Normal; VaD, Vascular Dementia; N/A, not available. Covariates adjusted for: a = Age; b = Sex or Gender; c = Education; d = APOEe4 allele; e = Stroke; f = Physical activity; g = duration or severity of PD; h = duration of follow-up; i = Mental Health (e.g. Depression); j = Vascular Factors (e.g. hypertension); k = smoking; l = alcohol consumption; m = Cognition; n = Other Comorbidities (e.g. diabetes); o = late-life social networks; p = early socio-economic status; q = BMI or waist-to-hip ratio; r = Race or Ethnicity; s = Cardiovascular Diseases; t = Medication; u = cohort source; v = Family history of AD; w = pain.
Global physical function
Two studies created composite scores (range 0–16) of physical function based on timed walk, chair stand, grip strength and balance tests. One high-quality cohort study found every one-point increase was associated with an estimated 7% reduced short-term risk of incident global dementia and 6% reduced short-term risk of incident Alzheimer's Disease (AD) (n = 2,288, follow-up = 5.9 years) [23]. A second study also found a significant association between composite score and incident dementia (n = 2,263, follow-up = 6 years) [31].
Manual dexterity
Two studies with overall high-quality used the Purdue Pegboard Test (PPT) to assess manual dexterity [50]. One PD cohort study found each one-point PPT increase was associated with 33% decreased odds of developing dementia (n = 80, follow-up = 4.4 years) [19]. An MCI study did not find an association with incident dementia (n = 189, follow-up = 10 years) and may have lacked representation of the reference population [24].
Finger-tapping
An MCI study did not find a significant association between finger tapping speed and incident dementia (n = 90, follow-up = 2 years) [26]. No adjustment for potential confounders was made.
Grip strength
Studies showed mixed results. Two generally high-quality studies (n = 2,619, follow-up = 6.5 years [43], and n = 2,046, follow-up = 11 years [17]) failed to find significant associations between low grip strength and incident dementia. A separate high-quality cohort study (n = 2,288, follow-up = 5.9 years) found each quartile increase in grip strength was associated with an estimated 13% decreased risk of incident dementia [23]. Another cohort study (n = 877, follow-up = 5.7 years) found each point increase in grip strength was associated with an estimated 1% decreased risk of incident AD, and may have lacked representation of their reference population [28].
Global parkinsonism
Six studies assessing parkinsonism (presence versus absence) and incident dementia showed mixed results. Three studies including community-dwelling older adults (n = 394, follow-up = 6 years) [32], MCI (n = 111, follow-up = 3.9 years) [29] and PD (n = 61, follow-up = 2 years) [34] failed to find significant associations. Significant associations were found by two PD studies (n = 1851, follow-up = 9 years [44] and n = 51, follow-up = 4 years [36]) and one study in cognitively healthy older adults (n = 538, follow-up = 5 years) [40]. Representation of reference population was a concern in two of the above studies [29, 32], attrition bias may have occurred in three [36, 40, 44] and four did not control for potential confounders [29, 32, 34, 36].
Nine studies evaluated severity of parkinsonism as a continuous variable. Significant associations with incident dementia were found in studies of older adults with MCI (n = 189, follow-up = 10 years) [24], PD (n = 173, follow-up = 3.6 years) [20], MCI and PD (n = 49, follow-up = 5 years) [41], and four studies of community-dwelling older adults (n = 115, follow-up = 2.6 years [42], n = 746, follow-up = 4.8 years [33], n = 919, follow-up = 4.7 years [18], and n = 1028, follow-up = 5.7 years [15]). In contrast, two studies including only PD patients did not find statistically significant associations with incident dementia (n = 80, follow-up = 4.4 years [19] and n = 24, follow-up = 6 years [35]). Two of these studies may have lacked representation of the exposed target population [24, 33], one had lower quality methods of outcome ascertainment [42] and one had potentially biased comparison groups [35]. Two were overall high-quality [15, 18, 20].
A meta-analysis combining studies which defined overall parkinsonism as a dichotomous exposure [32, 40] found parkinsonism presence was associated with an approximately 200% increased odds of incident dementia (Figure 2A). Publication bias was not a concern (see Supplementary data, Figure S1, available at Age and Ageing online).
Tremor
Five studies, using Unified Parkinson's Disease Rating Scale (UPDRS) variants, did not find significant associations between tremor and incident dementia. These studies included older adults with PD (n = 61, follow-up = 2 years) [34], MCI (n = 189, follow-up = 10 years [24], and n = 111, follow-up = 3.9 years [29]), MCI and PD (n = 49, follow-up = 5 years) [41] and community-dwelling status (n = 3,891, follow-up = 6.6 years) [39]. Incident dementia was also not significantly associated with clinician classification of resting tremor in two PD samples (n = 96, follow-up = 4.9 years [30], and n = 2,056, follow-up = 3.8 years [14]), nor with essential tremor in community-dwelling older adults (n = 84, follow-up = 5.4 years) [46]. Two of the tremor studies did not control for potential confounders [29, 34], three may have lacked representation of the target population [24, 29, 30], and attrition bias was a concern for three [39, 41, 46].
A meta-analysis combining studies with dichotomous tremor definitions [14, 30, 46] did not find a statistically significant overall association with incident dementia (Figure 2B). Publication bias was not a concern (see Supplementary data, Figure S1, available at Age and Ageing online).
Bradykinesia
Five studies assessed bradykinesia using relevant sections of the UPDRS or mUPDRS. One PD study found each one-point increase in severity of bradykinesia was associated with an estimated 9% increased risk of incident dementia (n = 173, follow-up = 3.6 years) [20], but a second PD study failed to find a significant unadjusted association (n = 61, follow-up = 2 years) [34]. An MCI study found every 1% increase in bradykinesia score was associated with an estimated 2% increased risk of dementia (n = 189, follow-up = 7.2 years) [24]. A separate MCI study failed to find a significant unadjusted association between bradykinesia and incident AD (n = 111, follow-up = 3.9 years) [29]. Representation of the target population was a concern for both MCI studies [24, 29]. A study including MCI and PD found the association between bradykinesia and Parkinson's Disease Dementia (PDD) was mitigated after controlling for age, sex and education (n = 49, follow-up = 5 years) [41]. Attrition bias may have occurred [41].
Rigidity
Three studies quantified severity of rigidity using the UPDRS or mUPDRS. Higher rigidity scores were associated with incident dementia in a PD study (n = 61, follow-up = 2 years) that did not control for potential confounders [34] and with PDD in a study of MCI and PD (n = 49, follow-up = 5 years). An MCI study did not find a significant association between rigidity and incident dementia (n = 189, follow-up = 7.2 years) [24], nor did a 4th study of PD where rigidity was defined clinically (n = 96, follow-up = 4.9 years) [30]. Representation of the target population was a concern for two studies [24, 30], and two may have contained attrition bias [30, 41].
Postural instability gait difficulty
An MCI study found every 1% increase in PIGD score on the mUPDRS was associated with an estimated 2% increased risk of incident dementia (n = 189, follow-up = 10 years) [24]. A study of cognitively healthy adults found PIGD presence was associated with 3.17 and 4.41 times the odds of developing dementia (n = 538, follow-up = two 5-year waves) [40]. The median score of PIGD relevant items on the UPDRS III was not significantly higher for those who developed PDD in a study of PD and MCI (n = 49, follow-up = 5 years) [41]. Two PD studies using the UPDRS found those who developed dementia had significantly higher PIGD scores at baseline (n = 61, follow-up = 2 years [34] and n = 24, follow-up = 6 years [35]). A third PD study did not find a significant association between increasing PIGD score on the Short Parkinson's Evaluation Scale and incident dementia (n = 261, follow-up = 4.8 years) [37]. Two of the PIGD studies may have lacked target population representation [24, 40], attrition bias may have occurred in three [37, 40, 41], and three did not adjust for potential confounders [34, 35, 37].
Balance
One study of cognitively healthy adults found the inability to perform a ‘one leg balance’ test was associated with an estimated 127% increased odds of incident dementia (n = 1,775, follow-up = 6 years) [38]. A high-quality study of community-dwelling adults found each one-point (range 0–4) increase on a ‘standing balance test’ was associated with an estimated 13% decreased short-term risk of incident dementia (n = 2,288, follow-up = 5.9 years) [23]. A PD study found failing the ‘pull test’ was associated with an estimated 245% increased risk of incident dementia (n = 96, follow-up = 4.9 years). The PD study may have lacked representation of the target population [30]. Two studies may have contained attrition bias [30, 38].
Quantifiable gait
One PD study found slow gait velocity, measured by the Timed Up and Go Test, was not significantly associated with incident dementia (n = 80, follow-up = 4.4 years) [19]. A non-PD study found each standard deviation (SD = 0.20 metres/second (m/s)) decrease in gait velocity was associated with an estimated 59% increased short-term risk of incident dementia and 47% increased short-term risk of incident AD (n = 3,663, follow-up = 9 years) [16]. Another non-PD study did not find slow gait velocity (<0.6 m/s) was significantly associated with incident dementia or AD, but was associated with an estimated 113% increased short-term risk of incident non-AD dementia (n = 2,619, follow-up = 6.5 years). A cohort study found slow gait velocity categorisation based on age and gender was not significantly associated with incident overall dementia, but was associated with an estimated 350% increased short-term risk of incident Vascular Dementia (VaD) (n = 767, follow-up = 3 years) [48]. Six cohort studies using different methods to quantify slow gait velocity (m/s, cut-off values, quartiles) consistently found significant associations with incident dementia (see Supplementary data, Table S4, available at Age and Ageing online) [17, 22, 23, 25, 27, 45]. One cohort study summarised quantitative gait performance into three factors: pace, rhythm and variability (n = 399, follow-up = 2 years) [47]. Pace was not significantly associated with incident dementia, but it was the only factor associated with increased risk of VaD specifically. Gait rhythm was associated with incident dementia unless baseline memory was controlled for. Gait variability was associated with incident dementia unless executive function was controlled for [47]. Three quantitative gait studies obtained the maximum quality rating on the NOS [16, 17, 19, 22, 23], one may have lacked representation of the target population [25], and potential attrition bias was a concern for five [25, 27, 43, 45, 47, 48].
Five studies combined in a meta-analysis [16, 17, 43, 45, 48] estimated slow gait velocity (dichotomous definition) increases the short-term risk of incident dementia by 94% (Figure 2C). The pertinent funnel plot suggests some publication bias towards larger studies reporting positive associations, and thus, the summary HR may be an overestimate (see Supplementary data, Figure S3, available at Age and Ageing online). Study heterogeneity precludes combining all gait velocity studies in a single funnel plot, so we do not know whether this publication bias is present for all studies assessing gait velocity and incident dementia or if it is only for the subset included in this meta-analysis.
Clinical gait
A case-control study found those with gait abnormalities had 3.5 times the odds of developing dementia in 5 years and 6.1 times the odds of developing dementia in 1 year compared to those without gait abnormalities present (n = 189) [49]. This study selected controls in a way that limited generalisability [49]. One high-quality non-PD study found any abnormal gait pattern (unsteady, ataxic, frontal, parkinsonian, neuropathic, hemiparetic or spastic) was associated with an estimated 96% increased short-term risk of dementia (n = 422, follow-up = 6.6 years) [21]. A second high-quality non-PD study found older adults with high-risk neurological gait syndrome had an estimated 170% increased short-term risk of VaD (n = 399, follow-up = 5 years) [13].
Discussion
Poor motor function in older adults is associated with higher risk of progression to dementia syndromes, even after adjusting for important covariates. The two most consistent findings for specific motor domains were positive associations of balance and gait abnormalities with risk of dementia syndromes in non-PD samples, and no associations between tremor and dementia syndromes in any population.
Upper limb motor domains, bradykinesia and overall parkinsonism showed mixed results. However, the heterogeneity in findings for overall parkinsonism might simply reflect varying proportions of individual parkinsonism subtypes with different predictive validity across samples. For example, samples that happened to have a greater proportion of subjects with essential tremors contributing to their global parkinsonism score would be less likely to find an association with dementia than studies with a greater proportion of subjects with non-benign gait disturbances contributing to their overall parkinsonism score. Rigidity was significantly associated with incident dementia in some but not all PD studies, and not significantly associated in an MCI sample. PIGD appeared to indicate an increased risk of incident dementia in MCI and community-dwelling older adults, but not in four of the five studies that included exclusively older adults with PD. More objective measures of balance maintained strong associations with conversion to dementia in community-dwelling older adults, but significant associations in PD samples were not found. This discrepancy whereby PD samples tend to have weaker associations than other samples between parkinsonism motor subtypes and incident dementia was suggested in other motor domains as well.
Impairments in gait performance emerged as one of the motor domains most strongly associated with incident dementia. Previously, it has been suggested that clinical gait abnormalities were most consistently associated with future progression to VaD, and this was supported by our systematic review. Importantly, when gait was assessed quantitatively, either by measuring gait velocity manually or with an electronic walkway, all but one study which included an exclusively PD sample [19] found significant associations with at least one incident dementia syndrome. Future studies need to assess whether the lack of association between gait velocity and future dementia in PD samples is an exception to the otherwise seemingly strong connection between gait velocity and dementia, or if it is merely obscured by other motor deficits present in PD.
Mechanistically, our findings align with the hypothesis that cognitive and motor function may share brain regions and networks which can be affected by both neurodegeneration and vascular factors [5]. The close proximity of frontal subcortical networks that control both motor and cognitive functions and their watershed vascularisation may explain the susceptibility of these brain regions and networks to brain microvascular disease, and not only to neurodegenerative processes [5, 52]. This mechanistic plausibility provides a rationale to propose that simple motor tests can be used as markers to detect patients at higher risk of dementia.
Strengths and limitations
To the authors’ knowledge, this is the first systematic review to provide a comprehensive evaluation of different motor domains and their association with incident dementia. A large number of studies were captured through the literature search, and the results of these studies were both reviewed qualitatively by motor domain and combined in meta-analyses when possible. However, the decision to report multiple effect estimates from the same study for multiple motor domains increases the probability that individual participants’ data may appear multiple times throughout the review, amplifying any bias present in a single study. Furthermore, the inability to include all studies which assessed any given type of motor function in a single meta-analysis may have biased the pooled effect estimates. In particular, inspection of study results not included in the meta-analyses suggests that the meta-analysis for overall parkinsonism and incident dementia may be an overestimate. Additional limitations include searching only English literature and having a single person perform data extraction.
Implications for practice and research
Motor function decline affecting, gait, balance and motor composite measures was associated with an increased risk of incident dementia syndromes in non-PD populations. In a clinical setting, these motor biomarkers are accessible with minimal cost and time, and can help to detect the risk of cognitive decline and dementia syndromes among older adults [51]. From a practical diagnostic perspective, the results from our syntheses suggest the prediction of dementia can be augmented by adding simple motor assessments, including gait velocity testing. Future studies should evaluate practical cut-offs for clinical applicability. In research settings, motor biomarkers can be included in longitudinal studies assessing dementia incidence. Ideally, motor tests will follow a clear protocol and the number of participants with and without motor abnormalities who convert to dementia be reported. This will allow larger meta-analyses to be conducted in the future, which may shed light around some of the mixed findings in this systematic review.
Poorer motor function was associated with higher risk of progression to dementia in older adults.
Poor lower limb motor performance, particularly slow gait, was consistently associated with incident dementia in older adults without Parkinson's Disease.
Tremor was not significantly associated with dementia syndromes in any population.
Parkinsonism and upper limb motor domains showed mixed results across older populations.
More consistent assessment of motor function domains and reporting of results is needed for future studies.
Supplementary data
Supplementary data mentioned in the text are available to subscribers in Age and Ageing online.
Acknowledgements
We would like to thank Dr John Costella, MLIS, for his assistance with search strategies.
Conflicts of interest
None declared.
Funding
Financial sponsors played no role in the design, execution, analysis and interpretation of data, or writing of the study. Miss J.K.K is supported by the Alzheimer Foundation London and Middlesex Master's Student Award. Dr Montero-Odasso's program in ‘Gait and Brain Health’ is supported by grants from the Canadian Institute of Health and Research (CIHR MOP 211220), the Ontario Ministry of Research and Innovation (ERA Award; grant number ER11-08-101), the Ontario Neurodegenerative Diseases Research Initiative (ONDRI; grant number 34739), the Canadian Consortium on Neurodegeneration in Aging (CCNA; grant ‘FRN’ CNA 137794) and by Department of Medicine Program of Experimental Medicine (POEM) Research Award (award number: 768915), University of Western Ontario. He is the first recipient of the Schulich Clinician-Scientist Award and holds the Ontario Premier's Research Excellence Award, and the CIHR New Investigator Award.
References
Please note that 30 of the most important references for this paper are listed here, and indicated by bold font in the text. The complete list of 52 references can be found in Appendix 4 of the journal's online supplementary material.
Comments