Concurrent validity and reliability of the Community Balance and Mobility scale in young-older adults

Balance ability generally starts to decline in the third decade of life [1], with an accelerated decline occurring in the sixth decade [2, 3]. Older adults (≥65 years) are more prone to experience a loss of function preventing them to maintain posture and respond to unexpected perturbations caused by slips or trips [4]. Young-older adults of retirement age (60–69 years [5]) generally function at a higher level compared to (old-) older adults. However, their more active lifestyle potentially exposes them to more high-risk balance-challenging situations. Subsequently, the risk for stumbles and near-falls is significantly higher [6]. With a dramatic increase in the proportion of young-older adults (baby boomer generation), a paradigm shift is requested towards early stage innovative population-level efforts to prevent loss of balance [7].

Regular physical activity (PA) is important to maintain independence and prevent functional decline. Current guidelines for older adults aged ≥65 years recommend at least 150 min of moderate intensity or 75 min of vigorous intensity aerobic training per week [8]. Persons with poor mobility should undertake training three or more days per week to improve balance and prevent falls [8]. However, less than 50% of older adults meet the current PA recommended [9] and only 6% complete regular balance training [10].

In order to promote early balance and mobility interventions, adequate assessment strategies are needed to identify subtle balance and mobility deficits in relatively active, high-functioning young-older adults. To date, most balance and mobility assessment tools have been developed to quantify deficits in frail older adults aged ≥70 years [11‐16]. Current systematic reviews focusing on functional balance assessment have shown that several assessment tools developed for older adults are not appropriate for detecting early balance and/or gait deficits in community-dwelling older adults with a more active lifestyle [17, 18]. For example, the Berg Balance Scale (BBS), a widely-used, valid and reliable test of functional balance in frail older adults aged ≥70 years [12, 18]. This test reached ceiling effects when used in community-dwelling older adults aged ≥60 years [15, 17, 18]. With most of the items focusing on basic functional mobility (e.g. transfers, standing unsupported, sit-to-stand), the BBS does not include challenging dynamic balance tasks such as tandem walking, hopping, or climbing stairs. Likewise, the Short Physical Performance Battery (SPPB) was initially developed for community-dwelling older adults aged ≥70 years [19]. This test has also shown ceiling effects in higher-functioning community-dwelling older adults aged ≥60 years [15, 20]. Ceiling effects of these instruments do not only hamper the detection of early balance deficits, but also prevent the detection of intervention-related changes over time in higher functioning older adults [20, 21].

Current systematic reviews focusing on mobility in older adults conclude that tests such as the Timed Up and Go (TUG) test, the Dynamic Gait Index (DGI), or the Performance Oriented Mobility Assessment also suffer from ceiling effects when applied in independently living, higher functioning older adults [13, 17]. They are not challenging enough to adequately assess the performance of older adults who do not display marked mobility deficits, because they lack more demanding mobility components such as turning the head while walking [11, 13, 14, 17, 22].

In summary, several studies have shown that balance and mobility measures developed for older, frailer adults show ceiling effects when applied in high-functioning older adults [13, 15, 17, 18, 20, 23]. The lack of high-challenging balance tasks in the aforementioned scales can result in early signs of balance and mobility decline to remain unidentified. This makes the currently available balance and mobility tests less suitable when the aim is to determine intervention eligibility aimed at preventing decline in balance and mobility at an early stage [13, 24, 25].

In this context, the applicability of the Community Balance and Mobility Scale (CBM) has recently generated significant interest in clinical practice for assessing balance and mobility deficits in community-dwelling older adults, either healthy (mean age 70.3 years [26]) or with knee osteoarthritis (mean age 62.5 years [27]). Unlike commonly used balance and mobility tests such as the BBS [12], SPPB [19] or the Tinetti test [14], the CBM includes several challenging tasks to assess specific aspects of balance and mobility which are necessary to function independently within the community. For example, walking while gaze shifting and turning the head, picking up an object from the floor (crouching) while walking, and complex walking maneuvers, such as forward to backward walking, sideways walking, or suddenly stopping, are included in the CBM [28, 29]. The CBM was initially developed to measure subtle balance deficits in patients with mild traumatic brain injury aged 26.2 years [30] to 31.0 years and is found to be valid and reliable in this population [28, 30].

Recently, the CBM has been validated in a sample of independently living, community-dwelling older adults aged ≥65 years (mean age 73 ± 7), showing excellent correlations with the BBS (ρ = 0.87), good correlations with the Timed Up and Go test (ρ = − 0.69) and self-selected gait speed (ρ = − 0.65) [26]. Reliability of the rating scheme was also analyzed based on videotaped assessments resulting in high inter- (ICC_2,k = 0.95; 95% CI = 0.88–0.98) and intrarater reliability (ICC_3,k = 0.96; 95% CI = 0.93–0.98) [26]. Moreover, the CBM showed no ceiling effects as compared to BBS (23%) and SPPB (33%) [26].

While these findings suggest that the CBM has added value in the assessment of community-dwelling older adults, the measurement properties in the specific population of young-older adults aged 60–70 years are yet to be evaluated. Young-older adults are an extremely heterogeneous population, where some older adults have substantial balance and mobility deficits while others have only minor deterioration in balance performances [31]. The CBM may represent a specific assessment tool for detecting both minor and major balance and mobility deficits in this population, and in turn may allow early interventions to be tailored to prevent functional decline.

In this study, we aimed to examine the concurrent validity and reliability of the CBM in community-dwelling healthy young-older adults (60 to 70 years). The evaluation was performed as preparatory part of the European Commission funded project PreventIT (Horizon 2020 grant no 689238), which aims to develop a lifestyle-integrated training intervention to prevent functional decline in young-older adults.

The first aim of the present study was to examine the concurrent validity of the CBM by comparing its scores to other established balance and mobility measures thought to have related theoretical constructs. We expected a positive association with the Fullerton Advanced Balance Scale [32] as this scale has also been developed to measure balance problems of varying severity in functionally independent older adults. We expected a negative association with the Timed Up-and-Go test [33] based on previous validation studies in older adults [26, 27]. Furthermore, we hypothesized moderate to good associations with balance tests measuring static steady-state balance control (8-level balance scale, comprising the five level balance scale from the SPPB and additional challenging tasks at a higher level, such as “tandem stand eyes closed” [34]) and dynamic steady-state balance control (3 Meter Tandem Walking [34], and gait speed [26‐28, 30, 35]). The second aim was to investigate the ceiling effects of the CBM as compared to other challenging balance and mobility assessments which, based on previous findings, were expected to be lower for the CBM [26, 27, 30]. The third aim was to investigate the intra- and interrater reliability of the rating scheme of the CBM, which was expected to be high based on previous studies in other populations [26, 28]. Finally, we aimed to analyze the internal consistency reliability.

A total of 51 participants aged 66.4 ± 2.7 years (range, 60–70 years; 74.5% female) were tested. Participant characteristics are summarized in Table 1. The number of participants included in the different analyses varied (N = 31–51). For the TUG and gait speed test, the first five participants were not assessed. For the participants in Heidelberg (n = 16), 3MTW performance was rated only by errors, but not by time. Because time was unavailable, these participants were excluded from statistical analysis on the 3MTW test, resulting in a subsample of 31 participants for which information on time and errors was available.

This study is the first to analyze the measurement properties of the CBM in a sample of young-older adults aged 60 to 70 years. As hypothesized, a good correlation with the FAB was found, indicating strong construct validity of the CBM in the target population of young-older adults. Furthermore, moderate to good correlations with other measures suggest that the CBM measures mobility performance (TUG), dynamic steady-state balance control (3 MTW, and gait speed) and static steady-state balance control (8-level balance scale). This is in line with previous studies estimating the measurement properties of the CBM in older adults [26] or those with mild traumatic brain injury [28, 30]. Importantly, the CBM does not show ceiling effects in contrast to other advanced balance scales such as the FAB.

A good correlation was found between the CBM and FAB, showing that both measure a similar construct. Both scales assess performance of more challenging balance tasks, including static, dynamic, proactive, and reactive balance control [28, 30, 32]. The ceiling effect in the FAB may have prevented a higher correlation with the CBM. However, it may also indicate that the tasks within the FAB are not challenging enough to discern difficulties in balance performance in high-functioning older adults [26, 28]. Moreover, the FAB was developed and evaluated to analyze balance impairments in community-dwelling older adults, rather than detecting subtle balance deficits in high-functioning older adults [32]. The correlation with the TUG was moderate (ρ = − 0.42), which was lower than expected and lower than reported in a previous study which validated the CBM in older adults [26]. The lower correlation in our sample of young-older adults might be explained by the fact that the TUG is not a highly challenging assessment tool, but rather measures basic functional performance which is typically applied in older adults or patient populations aged ≥70 years [13, 33, 38]. In the present sample, the average time to perform the TUG was 9.1 ± 1.8 s. A study which validated the CBM in older adults reported an average TUG time of 10.4 ± 2.2 s and found a higher correlation between both measures (ρ = − 0.69) [26]. The poor discriminative ability of the TUG may have prevented the correlation between the TUG and the CBM from being higher. Recent studies confirm this assumption, showing that the TUG is able to discriminate performances in less healthy, lower-functioning populations (e.g. fallers), but not at discriminating performances in healthy, high-functioning groups [13].

The CBM showed good correlation with 3MTW errors (ρ = − 0.61). The 3MTW errors classify a subject based on errors made during a challenging dynamic balance task, which is similar to the classification scheme of the CBM which may explain the good correlation. For 3MTW time, the correlation was lower (ρ = − 0.35) as compared to 3MTW errors. This suggests that the quality of task execution (3MTW errors) is more strongly linked to CBM performance as compared to the time of task execution (3MTW time).

Habitual gait speed, a less challenging measure of dynamic balance, showed a moderate correlation with the CBM (ρ = 0.46). This suggests that a simple assessment of gait speed, commonly applied in older adults aged ≥70 years [47], may not be sufficient to detect subtle balance deficits in a sample of young-older adults. However, these measurements were intentionally included for comparing the CBM to commonly applied clinical assessment tools and because it has been used in previous validation studies with the CBM in samples of older adults and knee osteoarthritis patients [27, 28].

As expected, a moderate correlation was found between the CBM and the 8-level balance scale (ρ = 0.32). The 8-level balance scale is a measure of static steady-state balance control whereas the CBM primarily evaluates dynamic aspects of balance during complex mobility tasks. In line with the present findings, previous studies have reported moderate associations between static and dynamic steady-state balance control, suggesting that both aspects of balance control are partly interrelated, but represent distinct aspects of balance control (e.g. Functional Reach Test vs. gait speed, r = 0.08–0.39 [48] or one-leg stand vs. jumping over a hurdle, r = 0.05–0.23) [49].

An excellent inter- and intrarater reliability of the CBM total score was found, exceeding the recommended standards of 0.90 to 0.95 for clinical assessments [42]. For the first time, the reliability of the scoring of the single items of the CBM were also evaluated, showing good to very good intrarater reliability [46] for all 19 items. This finding suggests that if the same rater evaluates a participant’s performance on the CBM scale on two separate occasions, high reliability can be expected. For interrater reliability, only five out of the 19 test items had a moderate and two a fair agreement (i.e., “Forward to backward walking” and “Walking and looking right”) [46]. Possible explanations for these two items might be that raters rated individual’s performance differently, such as maintaining straight path versus veering during walking (e.g., “Forward to backward walking”) as well as difficulties to determine for how long the participant’s eyes focused on a point (e.g., “Walking and looking”).

The Cronbach’s alpha as a measure for internal consistency was 0.88. Although it does not exceed the value of 0.90 suggesting redundancies among items [50], further studies should analyze if there are redundant items to design a shortened version of the CBM. As indicated by the results (Table 4), each individual item correlated > 0.20 with the total score, indicating satisfactory internal consistency [45]. On the same note, our findings indicate that future studies with adequate sample sizes should perform a more detailed analysis to purify the CBM. As indicated by Table 4, item-scale correlations for seven items were < 0.50 (“Tandem walking”, “180° Tandem pivot”, “Crouch and walk”, “Lateral dodging”, “Running with controlled stop”, “Descending stairs”, and “Step ups × 1 step right”) which may suggest that their additional value is limited as the cut-off points for internal consistency vary [51‐53]. Future studies could determine the underlying factors that represent the CBM construct and eliminate items which cannot be assigned to a factor for purification of the assessment tool. Such factor analyses require a sample size of at least 10 participants per item in the scale [54], which would be 190 participants for the CBM. The development of a shortened CBM has been requested previously [26] and could be of significant benefit as the original version takes 20–30 min to complete.

A limitation of this study is that the sample consists of participants from three countries. While beneficial, cross-national research has limitations. It might be that variation in the performance across the countries could have occurred, despite standardized operating procedures.

Additionally, females were overrepresented in our sample (75%) as compared to the general population aged ≥60 years (56% [55]). However, the sample was too small to perform a stratified analysis for gender. Additionally, the posthoc exploratory analyses for the ability of the CBM to discriminate young-older fallers (mean score 58.3 ± 14.6) from non-fallers (mean score 66.3 ± 11.8) did not reveal statistically significant differences (p = .09). A larger sample is needed to evaluate the validity for discriminating fallers from non-fallers.

This cross-sectional study did not allow the determination of responsiveness. Further studies are needed to evaluate the responsiveness of the CBM in the target population.

This study provides evidence that the CBM is a suitable tool for the assessment of challenging balance and mobility performances in healthy, young-older adults. The CBM tasks represent meaningful everyday performances which are specifically required to ambulate safely within an everyday environment. With trained assessors, the scale is easily administered, requires little equipment, and most importantly, is valid and reliable in the studied target population. Based on the present results, the CBM has been selected as an end point within the EU project PreventIT and is currently used within a randomized controlled trial evaluating a lifestyle-integrated training intervention for preventing functional decline in healthy, young-older adults (registered online; https://​clinicaltrials.​gov/​ct2/​show/​NCT03065088). The CBM may help to better understand the mechanisms of early balance and mobility decline in young-older adults and inform the development of treatments and intervention programmes aimed at improving early deterioration in balance and mobility, which is in line with the recently updated guidelines for early implementation of neuromotor exercise training in public health approaches [7].