Introduction
‘Motor capacity’ refers to an individual’s motor function assessed in a standardized laboratory environment whereas ‘mobility performance’ depicts enacted mobility in real-life situations [
1]. The International Classification of Functioning, Disability, and Health (ICF) differentiates between these two measures: what a person can do (capacity) and does do (performance). Understanding the association and complementarity of motor capacity and mobility performance could help to understand gait, balance, and mobility disabilities in older adults. If a relationship exists, one could use motor capacity measures as a surrogate marker of mobility performance. Likewise, a causal relationship would imply that improving a person’s motor capacity would result in increased mobility performance. However, research has shown that the association between motor capacity and motor performance is not straightforward, and several studies have shown that in-lab measured gait differs from real-life measured gait in community-dwelling older adults [
2‐
6]. One reason may be that younger or healthy older adults need a lower relative effort compared to impaired older individuals who perform near their maximal capacity to execute daily motor tasks [
7]. In support of this idea, laboratory studies have shown this for muscle function [
8] and walking [
4,
9], which led us to the assumption that the association between motor capacity and mobility performance might become evident with increasing impairment in older persons. This assumption has not been tested, yet. It also remains unclear whether this has implications for everyday life.
Frailty incorporates both muscle function and walking and is a widely used, accepted cumulative measure of age-related, gradual multisystem impairment [
10,
11]. We therefore use frailty status as a distinguishing criterion, categorizing older persons into different stages of impairment, which allows to further explore the association of motor capacity in mobility performance in different subsamples. Specifically, we investigated whether the association between motor capacity and mobility performance is moderated by frailty status in older adults. We hypothesized that the association becomes more evident with transition into frailty status.
Discussion
To our knowledge, this is the first study that has investigated whether the association between motor capacity and mobility performance is moderated by frailty status. We have confirmed our hypothesis that frailty status indeed acts as a moderator of this relationship. Against the background of inconclusive results of previous studies showing that these factors are weakly associated or not associated in different samples [
2,
3] we have enhanced this line of research with the important finding that they in fact are associated when using one of the most accepted categorization—frailty status according to Fried et al.—as a distinguishing criterion. More precisely, motor capacity is only associated with gait-related mobility performance in daily life if a certain degree of physiological impairment is given, in our case pre-frailty and frailty. No significant associations were found in the non-frail group in the final models (Step 3), with trends in the counterintuitive direction that higher capacity is associated with lower performance. One explanation could be that there is a ‘performance threshold’ where higher capacity does not yield any performance enhancement; or that persons with higher motor capacity have larger variability in their mobility performance, which would lead to lower correlations between both measures.
Unlike in the other two mobility performance outcomes, the moderation effect was not evident in CumPA, that is, the association between motor capacity and mobility performance was not related to frailty status. Simply put, the moderation effect is present with regard to
how someone enacts her/his mobility, but not with regard to
how much she/he is physically active during daily life in general. This is backed by findings from van Lummel et al. [
27] who found that physical activity and physical performance described two different domains of physical function. Others have shown that slower walking speed is associated with less physical activity according to activity monitoring [
28]. This inconsistency may be explained with the fact that factors affecting mobility are complex [
29,
30], which may be the case especially when a more behaviour-oriented aspect of mobility such as physical activity is investigated.
Our findings hold important practical implications and suggest laboratory-based gait assessment better represents walking-related everyday performance in pre-frail and frail older persons. For example, being able to walk longer distances and to perform several walking bouts of a somewhat longer distance may have a strong impact on the degree of autonomy and self-determination in everyday life: Participation in recreational activities within the neighbourhood or upholding social contacts within the community would be doable without assistance. In clinical practice, motor capacity assessments (e.g., timed up-and-go, gait speed) are used to draw conclusions on subjects’ mobility performance and functionality in real life, reflecting their performance beyond the time of the assessment. This is critical as clinical decisions or subsequent therapy prescriptions are often based on such laboratory-based test results.
Based on our findings we can only speculate about a causal relationship between both factors, that is, the question whether actual mobility performance may be improved via increasing motor capacity (walking speed) in frail persons warrants further research. A confirmed causal relationship would allow tailored interventions depending on frailty status.
Important strengths of our study are the objective, sensor-based assessment of mobility performance and motor capacity using validated systems as well as the innovative character of our investigation. Still, our results should be interpreted with respect to potential limitations. Since walking speed is a frailty criterion, there is a theoretical contamination in the data. To explore a potential circular reasoning a modified version of the frailty phenotype was calculated as previously done by Blodgett et al., [
31]. When analyses were rerun with this modified frailty phenotype, no moderation effects were found. However, sample sizes were severely changed by this modification: for NWS models, 45% (FWS: 31%) more subjects were categorized as non-frail than with the ‘normal’ frailty phenotype, that is, many of those who were by definition pre-frail were now labelled as non-frail. Moreover, analysis of multicollinearity between NWS/FWS and frailty dummies showed that neither variance inflation factor nor tolerance of the models were above generally accepted limits [
32]. We suggest this issue be investigated in future research, e.g., by measuring motor capacity using multi-dimensional mobility tasks. Another limitation is that in our sample, only 19 subjects were categorized as frail. Hence, statistical power to detect a moderating effect may have been limited in this sample. For FWS, we have solved this matter (frail:
n = 3) by dichotomizing frailty status into non-frail and pre-frail/frail, which appears to be justified given how close the coefficients were for pre-frail and frail in the NWS models 1 and 2. Another limitation is that our sample was predominantly women, which is why we have controlled for sex in the models of NWS, but not FWS in order not to ‘overload’ the models due to the smaller sample size. Regarding frailty, there is no overall consensus on an operative definition [
33]; possibly a broader frailty concept or a different measure could have altered the results. We also want to highlight that the effect sizes of the interactions are rather small in most models. In some models, the direct effect explains far more variance than the interaction whereas in others the proportion of variance explained by the interaction terms is rather large (e.g., Table
2, 2nd model, more than a quarter of the overall variance explanation), and no other direct effects are observed. This is no unexpected result because one can expect the lab-measured capacity to explain a fair amount of variance of real life performance. However, the significant interaction shows that the grouping variable (frailty status) significantly impacts on the amount of variance explained, that is, we could confirm our hypothesis that the issue of frailty–or maybe other impairments as well–moderating the relationship between motor capacity and mobility performance is worth considering in future research. A next step could be to examine a moderation effect of the associations between the exact same parameters when measured in the laboratory and during real life, as performed by Hillel et al. [
5]. Qualitative outcomes such as gait variability, symmetry, regularity, and other outcomes that represent gait quality may hold promising potential for future research as well.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.