Predictors of decline in walking ability in community-dwelling Alzheimer’s disease patients: Results from the 4-years prospective REAL.FR study

Longitudinal studies of subjects with Alzheimer’s disease (AD) have mainly explored the gradual decline of cognitive performance and the changes in behavioral domains. The decline of ability to perform activities of daily living, such as walking, has received less attention. However, AD is also characterized by an early deterioration in mobility [1] of significant practical importance [2‐4].

Mobility limitation in AD patients results in approximately two or three times higher risk of falling than a similar mobility limitation in cognitively intact older adults [5]. Moreover, the ability to walk is also a key determinant of an AD patient’s quality of life [6], institutionalization [7], risk of death [8] and burden for the caregiver and the community [9].

Numerous factors may explain the decline in walking ability during the course of AD. Poor executive function has been associated with poor physical performance. In severe AD patients, extrapyramidal symptoms can be observed [5, 10, 11], potentially leading to fall and fracture events. Psychotropic treatment, weight loss, depressive symptoms [12], behavior disturbance and poor balance due to prolonged bed-rest, may also be associated with gait disorders and loss of walking ability. To our knowledge, no study has previously reported the longitudinal change in mobility using the walking item of the Activities of Daily Living (ADL) scale in people with AD and investigated their predictors in community-dwelling AD patients.

The identification of risk factors associated with future mobility impairment represents a first step for the optimization of preventive intervention in this vulnerable population. Moreover, given the widespread use and clinical relevance of the ADL scale, the present analyses may provide results that may be easily implemented in routine clinical practice. In fact, by showing that specific risk factors may predict future decline of walking ability in AD patients, we may hypothesize about the design of effective interventions aimed at reversing this deterioration. Thus, the aim of the present prospective study is to investigate predictors of decline in walking ability in a large cohort of AD patients, followed-up for four years.

The flow chart of the four-year REAL.FR prospective study is shown in Figure 1. Among the 684 AD subjects recruited at baseline, 207 attended the final follow up visit at four years. During this period of observation, 103 participants died, 88 declined continuation in the study, 75 were institutionalized and 70 were lost to follow-up. Compared to the patients followed over the four years, the patients who did not completed the four-year follow up (Figure 1) were, at baseline, significantly older (78.51 years; SD 6.74 versus 76.53; SD 6.81), were less frequently women (68.72% versus 76.09%), were more disabled (ADL = 5.33; SD 0.96 versus 5.67; SD 0.60), more severely demented (MMS = 19.56; SD 4.34 versus 20.94; SD3.84), had more behavior disturbance (NPI = 16.70; SD 16.35 versus 12.69; SD 12.57), had more cardiovascular disease (29.52% versus 17.84%) and had more frequently an abnormal one-leg balance test (19.17% versus 7.31%) (all P values <0.05). Baseline characteristics of the study sample are summarized in Table 1. The incidence decline in walking ability was 12.76% (95%CI, 10.86% to 14.66%) per year during the four years of follow-up.

In the first model (Table 2), (after adjustment for confounders without the variable ADL), the following independent factors remained significantly associated with decline in walking ability: older age (RR = 1.05 (1.02 to 1.08)), length of time from diagnosis of dementia (RR = 1.20 (1.05 to 1.37)), painful osteoarthritis (RR = 1.81 (1.17 to 2.78)), MNA score (RR = 1.45 (1.01 to 2.08)), institutionalization (RR = 1.83 (1.03 to 3.25)), hospitalization for lower limb fracture (RR = 5.80 (2.85 to 11.81)), MMSE score between 16 to 20 versus >20 (RR = 1.07 (0.66 to 1.74)) and MMSE score <15 versus >20 (RR = 2.41 (1.54 to 3.75)), abnormal one leg balance test (RR = 2.25 (1.44 to 3.51)) and use of AChE inhibitors (RR = 0.48 (0.26 to 0.89)) (Table 2). In the second model (after adjustment for all confounders including the variable ADL), the following independent factors predicted the decline of the ability to walk: older age (RR = 1.05 (1.02 to 1.08)), length of time from diagnosis of dementia (RR = 1.16 (1.01 to 1.33)), painful osteoarthritis (RR = 1.84 (1.19 to 2.85)), hospitalization for lower limb fracture (RR = 6.35 (3.02 to 13.37)), higher ADL score (walking item excluded) (RR = 0.49 (0.43 to 0.56)) and use of AChE inhibitors remained in this final model (RR = 0.52 (0.28 to 0.96)).

This prospective study reports a high rate of decline in walking ability in a community-dwelling AD population. In this population, we reported a 12.76% annual incidence of walking ability decline. After four years, about one third (30.20%) of the remaining sample needed assistance or had lost the ability to walk. We also show that high risk of decline in walking ability is associated with age, time since diagnosis of dementia, poor baseline functional score and orthopedic factors but treatment with AChE I appears protective.

The ability to walk has been reported to be a major factor affecting quality of life in patients with dementia [6]. Once a patient with AD has lost their ability to walk, he/she is generally unable to regain it [22]. Delaying loss of the ability to walk during the disease course for as long as possible should be a goal in care planning. Previous studies have reported the significant changes that occur in gait characteristics and gait speed [1]. Physical performance tests [23] and specific scales, for example the Unified Parkinson's Disease Rating Scale (UPDRS), have been used to measure the changes in motor signs in people with AD in longitudinal studies [24]. We believe that the reported rate of decline in walking ability during a four-year follow-up, using the ADL score, provides clinically relevant data for potential interventional studies aimed at slowing the decline in walking ability.

Age, time since diagnosis of dementia and poor baseline functional score were expected results from this study. Osteoarthritis [25] and hip fracture [26] are also well-known and strong risk factors for disability, especially for the mobility domain and in the context of dementia [27].

AChE inhibitors were found to be associated with a lower risk of decline in walking ability and this factor remained statistically significant in the different models. A statistically significant improvement in global function of patients treated by AChE inhibitors was previously reported in some studies [28‐30] but not in others [31]. Reviews and meta-analyses on the beneficial effect of the AChE inhibitors on global functioning support only a modest benefit in slowing decline in global function [32]. These previous studies have addressed all ADL domains that include heterogeneous items (urinary incontinence, transfer, dressing, eating and walking). Then, from an analytic perspective an improvement in just one area might be clinically significant but statistically masked by no effect in other functional domains. The previous randomized clinical trials on global functional score do not allow specific conclusions on the ability to walk.

In our study, the suggested protective effect of AChE inhibitors on decline in walking ability raises several hypotheses. Maintaining the ability to walk relies on many factors, including cognitive function, balance and strength but also physical training, behavior disturbance and environment. Some of these factors can be, at least in part, directly or indirectly improved by AChE inhibitors. AChE inhibitors have demonstrated a positive effect on cognition [33, 34] and studies suggest that galantamine may more specifically improve attention and executive function [35]. Since AChE inhibitors improve gait velocity and reduce gait variability [36], it is plausible to think that participants on AChE inhibitors may develop a better gait strategy improving their ability to adapt their gait pattern to unexpected situations, hence reducing their risk of falls [36]. Moreover, AChE inhibitors may impact walking ability indirectly through a reduction in behavioral disturbance. Indeed, AChE inhibitors have positive effects on a variety of behavioral and psychiatric symptoms in AD, particularly for apathy, depression and anxiety [37]; these symptoms are often associated with low physical activity and physical deconditioning, which may impact on the ability to walk in the longer-term. The effect of AChE inhibitors on peripheral neuromuscular mechanisms has been poorly explored compared to the central effect and requires further study and validation. However, the cholinergic neuromuscular junction of skeletal muscle is potentially a rich source of amyloid-beta (Aβ) peptides and AD might have systemic manifestations [38]. Moreover, AChE inhibitors can generate fasciculation in rats [39] and experimental electrophysiological data support an improvement in the neuro-muscular properties of AD patients treated with AChE inhibitors [40]. Finally, AChE inhibitors may attenuate the negative effect of the systemic inflammatory responses through direct or indirect mechanisms. Increased peripheral concentrations of inflammatory markers have been reported in AD patients [41], and acetylcholine can inhibit the production of pro-inflammatory cytokines [42]. Basic research supports the theory that donepezil has neuroprotective and cardioprotective effects via anti-inflammatory mechanisms but other organs, such as the muscle, may be affected [43].

This study has several limitations. First, discontinuation of AChE inhibitors is a frequent event and may be related to the acceleration of the functional decline. Cognitive decline, anxiety, weight loss and hospitalization have been reported in this cohort to predict discontinuation of the AChE inhibitors [44]. To reduce this limitation (AChE inhibitors discontinuation may be the consequence of the functional decline or the decline of the walking ability), we took into account discontinuation of the AChE inhibitors at least six months prior to the decline in walking ability. Despite this, AChE inhibitors remained significantly associated with the decline in walking ability in the final model which included the ADL score. However, we cannot ignore the fact that treatment with AChE inhibitors was only a marker of better walking ability. Second, many factors, such as pain, low level of physical activity and deconditioned balance because of prolonged bed-rest, were not assessed in our study and may account for the decline in walking ability. However, hospitalization and painful osteoarthritis were taken into account in our analyses. Third, selection bias could also have occurred because the recruitment of the participants was performed via specialized memory centers in France. The attrition rate was high although similar to other cohorts of patients with dementia. Characteristics of the patients lost to follow up compromise the generalization of our results. However, patients who refused to participate, were institutionalized or lost to follow up may have experienced a higher rate of functional decline, probably leading to the under-estimation of incident walking impairment. Finally, the choice of the ADL scale to assess the ability to walk is open to criticism. The ADL is useful in the management of AD but the walking item may not be sensitive to small but significant changes. Additional data on physical performance such as the gait speed test would have reinforced our results.