Calculations of mechanisms for balance control during narrow and single-leg standing in fit older adults: A reliability study

Gait & Posture 34 (2011) 352–357

Contents lists available at ScienceDirect

Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost

Calculations of mechanisms for balance control during narrow and single-leg standing in fit older adults: A reliability study A.C. A˚berg a,b,*, A. Thorstensson a, O. Tarassova a, K. Halvorsen c,d a

The Swedish School of Sport and Health Sciences, Stockholm, Sweden Department of Public Health and Caring Sciences/Geriatrics, Uppsala, Sweden c School of Technology and Health, KTH-Royal Institute of Technology, Stockholm, Sweden d Department of Information Technology, Uppsala University, Uppsala, Sweden b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 7 December 2010 Received in revised form 9 May 2011 Accepted 31 May 2011

For older people balance control in standing is critical for performance of activities of daily living without falling. The aims were to investigate reliability of quantification of the usage of the two balance mechanisms M1 ‘moving the centre of pressure’ and M2 ‘segment acceleration’ and also to compare calculation methods based on a combination of kinetic (K) and kinematic (Km) data, (K–Km), or Km data only concerning M2. For this purpose nine physically fit persons aged 70–78 years were tested in narrow and single-leg standing. Data were collected by a 7-camera motion capture system and two force plates. Repeated measure ANOVA and Tukey’s post hoc tests were used to detect differences between the standing tasks. Reliability was estimated by ICCs, standard error of measurement including its 95% CI, and minimal detectable change, whereas Pearson’s correlation coefficient was used to investigate agreement between the two calculation methods. The results indicated that for the tasks investigated, M1 and M2 can be measured with acceptable inter- and intrasession reliability, and that both Km and K–Km based calculations may be useful for M2, although Km data may give slightly lower values. The proportional M1:M2 usage was approximately 9:1, in both anterio-posterior (AP) and medio-lateral (ML) directions for narrow standing, and about 2:1 in the AP and of 1:2 in the ML direction in single-leg standing, respectively. In conclusion, the tested measurements and calculations appear to constitute a reliable way of quantifying one important aspect of balance capacity in fit older people. ß 2011 Elsevier B.V. All rights reserved.

Keywords: Balance Elderly Assessment Reliability

1. Introduction Balance control in upright positions is a prerequisite for independence in activities of daily living (ADL) [1]. Additionally, the definition of human balance, being the dynamics of postural control to prevent falling [2], indicates that balance performance is necessary to avoid such incidents [3]. Correspondingly, balance impairment has been identified as one of the most reliable predictors of falls among the elderly [4]. Though, due to its complex nature, balance is difficult to assess [5]. In clinical practice, different categories of tests are commonly used to screen for balance problems [6]; for example (I) Timed tests such as standing balance [7] or timed up-and-go (TUG) [8], (II) Reaching tests such as functional reach [9], (III) Stepping tests [10],

* Corresponding author at: Uppsala University, Department of Public Health and Caring Sciences/Geriatrics, Uppsala Science Park, SE-751 85 Uppsala, Sweden. Tel.: +46 18 6117960. E-mail addresses: [email protected], [email protected] (A.C. A˚berg). 0966-6362/$ – see front matter ß 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.gaitpost.2011.05.025

or (IV) Ordinal scales, combining different balance tasks, for example Berg’s balance scale [11]. It has, however, been indicated that the theoretical constructs behind these measures are insufficiently described [6] and that the understanding of balance in clinical environments is largely intuitive. This, however, may lead to a less systematic approach in balance assessment, documentation and intervention [2]. ‘Knowledge translation’ in this area, including theoretical models that are useful for clinical reasoning and practice is, therefore needed. Based on the equations of motion, it is possible to discriminate two balance mechanisms for standing situations when no extra external force (apart from gravity) is applied [12]. The first of these mechanisms (M1) implies moving the Centre of Pressure (CoP) with respect to the Centre of Mass (CoM), to drive it within safe boundaries. This mechanism is commonly described by the inverted pendulum model (IP) for human balance [13], where the body is modelled as a single segment or a ‘stick’ standing on the ground. The second mechanism (M2) identified as ‘counter rotation’ [12] or ‘segment acceleration’ [14] implies that parts of the body are rotated around the CoM, thus causing horizontal reaction forces to accelerate the CoM in

A.C. A˚berg et al. / Gait & Posture 34 (2011) 352–357

a safe direction. The hip-strategy and arm and leg motions belong to this mechanism, which is mainly seen when the base of support (BoS) and hence available space for CoP displacement, is limited. In the original publication of M1 and M2 [12], including a force plate experiment, it was concluded that considerations of both M1 and M2 are relevant in assessment of single-leg standing. In addition to this, the possibility to alternatively calculate these mechanisms from kinematic (Km) data was highlighted. However, the most reliable method for M1 and M2 calculations – based on Km data only or a combination of kinetic (K) and Km data (K–Km) [12,14] – has not yet been demonstrated. The aims of the present study are to investigate inter- and intrasession reliability of such calculations, to compare calculation methods based on K–Km or Km data only concerning M2, and to examine the proportional usage of M1 and M2, during narrow and single-leg standing in older persons. 2. Methods 2.1. Participants and data collection Data from nine physically fit persons aged 70–78 (mean = 74.5) years, of whom six were women, are presented (Table 1). Their self-perceived fitness was confirmed by the use of a protocol including (in)dependency in ADL. Fall efficacy was evaluated by the use of the Falls-Efficacy Scale (FES) [15]. Additionally, the performance based tests – TUG [8], five-chair rises [16] and hand grip strength (Gripmeter, Sagitta Pedagog AB, Mariestad, Sweden) were performed. Anthropometric measures of height from the floor to the trochanter major, foot width, body mass, and height were measured prior to the experiment. Comfortable gait speed was tracked in the laboratory on a 6 m long, 1 m wide and 0.09 m high runway. 2.2. Instrumentation Synchronised Km and K data were collected with a sample frequency of 100 Hz. Km data were obtained with a 7-camera motion capture system (ProReflex, Qualisys AB, Go¨teborg). Thirty six reflecting markers allowed construction of a 13 segments body model; 1 head/neck; 2 upper trunk; 3 pelvis; 4–5 each upper arm; 6–7 each forearm; 8–9 each thigh; 10–11 each shank; and 12–13 each foot. Six extra markers, positioned medially at the elbow, knee and ankle joints, were used during reference measurements in neutral standing. To improve reliability, the same person attached all markers in all experiments. Ground reaction forces were measured by two force plates (50 cm  25 cm, Model FP2550-06, Bertec, Columbus, USA). 2.3. Procedure The participants wore tight sport shorts, a brassiere (in women), and ordinary walking shoes. They stood with one force plate beneath each foot with eyes focused on a stationary target at approximately eye height on the wall 3 m in front, for 3 repeated trials of narrow standing and single-leg standing on each leg, respectively. Each trial lasted for 30 s and free arm movements were allowed. For narrow standing, their feet were placed close together, while maintaining one foot on each force plate. For the single-leg standing they were asked to place their feet in the middle of each force plate, and lift one (alternately right and left) foot up and keep balanced in single-leg stance for as long as possible. If necessary, they were allowed to put down the uplifted foot and then lift it again as soon as balance control was recaptured. This allowed a maximal amount of data collection in single-leg stance during each 30 s-trial. Re-test measurements with the same laboratory set-up were conducted within a two weeks period. 2.4. Data processing The data were exported to Visual3D (C-Motion Canada) for further processing, including definition of the body model, according to marker positions from the reference trial, with the subject in a neutral standing position. The mass of the individual segments was defined as fractions of the total body mass using results from Dempster [17]. The moment of inertia and CoM were determined from the shape of the segment model, assuming a uniform distribution of mass. The shapes of the segments were represented by truncated cones, except the head, which was represented by an ellipsoid. The shapes were determined from the positions of the reflective markers in a neutral standing position. All data were low-pass filtered using a fourth-order zero-phase Butterworth filter with a cut-off frequency of 6 Hz. Built-in functions of Visual3D were used to compute the position, orientation, acceleration, and angular acceleration of each body segment.

353

Table 1 Distribution of anthropometrical measures and clinical assessments concerning the investigated 9 subjects of whom 5 were females. Group (n = 9) Mean Age (years) Height (m) Mass (kg) Clinical assessments Fall efficacy (0–130) Timed up-and-go (s) Five-chair rise (s) Hand grip strength (N) Female (male) (N) Gait speed (m/s)

Median

74 1.73 76.8 129.5 9.2 9.6 349 292 (421) 1.21

74 1.71 76.5 130 9.8 9.2 370 289 (412) 1.22

Range 70–78 1.59–1.92 61.5–94.9 127–130 7.8–10.1 8.0–11.5 209–490 209–414 (370–490) 0.63–1.47

Half a second of data was discarded from the beginning and end of each trial and/ or interval in which the subject succeeded to accomplish the task. All subjects managed to maintain balance for the entire trials of 30 s in narrow standing. In single-leg standing, one or more intermittent intervals of two-leg support occurred (and were discarded) in 18 out of 32, 30 s-trials. In each such trial, the discarded intervals of successful single-leg standing exceeding three seconds were concatenated into one time series. Calculations of the contribution of the two balance mechanisms, which were determined by the relative magnitude of the terms M1 and M2 [12,14,18], were carried out in Matlab (Matlab R14, The MathWorks, Natick, MA, USA) according to the following equation [14]: d H ¼ ðr CoM  r CoM0 Þ  maCoM þ Ibody aCoM ðr CoP  r CoM0 Þ  F G  |fflfflfflfflfflfflfflfflfflfflfflfflfflfflffl ffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |ffl{zffl} dt |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} M1

M2

(1)

Mtot

where rCoP is the CoP, rCoM is the position of CoM, r CoM0 is the vertical projection of the CoM onto the horizontal support surface, FG is the gravitational force, d/dt(H) is the rate of change of the angular momentum with respect to the CoM, m is the mass of the body, aCoM is the acceleration of CoM, Ii is the moment of inertia of the ith segment, aCoM is the angular acceleration of the line connecting rCoM and rCoM. Here the M1 term was computed from both Km data (to obtain r CoM0 ) and K data (to obtain rCoP and FG) whereas the M2 term was computed from Km data only. On the right hand side of the equation, Mtot was computed in two different ways, using either Km data only or K–Km data. The latter way makes use of a fixed, equivalent pendulum length which is 1.1 times the height of CoM, as in Hof [12]. The contribution of each of the mechanisms over the duration of a task was calculated as the slope of linear regressions, computed separately from the horizontal components of the vector equation above using the total least squares (TLS) method [19]. The slope coefficients were multiplied by 100 to obtain percentages [14]. Since it is not possible to compute M1 based on Km data only, the comparisons between methods (Km vs K–Km) exclusively concerned M2. 2.5. Statistical analyses Descriptive statistics were provided for the clinical assessments. Reliability was analysed concerning the percentage usage of the two mechanisms of balance control M1 and M2 [12], which were calculated in two ways (see above). The mean, SD for each test (T) and retest (Rt) sessions, including between session (T  Rt) differences were calculated for M1 and M2 in anterio-posterior (AP) and mediolateral (ML) directions during the tasks of narrow standing, single-leg standing on right and left leg, respectively. A repeated measures ANOVA with four within group factors; that is task (narrow and right and left single-leg standing), direction (AP and ML), balance mechanisms utilization (M1 and M2) and repeated sessions (T and Rt), was used to compare M1 and M2 usage. Tukey’s post hoc tests were performed to detect differences between narrow and single-leg standing; AP and ML directions; single-leg standing on right and left leg and T- and Rt-sessions. Both inter- and intra-session reliability, referring to the three repeated trials per T and Rt sessions, respectively, were estimated by ICC. F-test ANOVA was performed to verify the presence of any systematic bias. As no such bias was detected, analyses were made by ICC3,k for inter- and ICC3,1 for intra-session reliability [20]. The absolute reliability indicators, standard error of measurement (SEM) including its 95% CI and minimal detectable change (MDC) between T  Rt measurements were calculated within the estimated variance components [21]. To compare the obtained ICCs, their 95% CIs were analysed for any overlapping. Pearson’s correlation coefficient (rP) was used for verification of possible agreement between the two methods of calculation (based on Km or K–Km data) applied for M2 calculations, and then paired t-tests and a Bland–Altman plot were performed to detect any systematic bias. Due to occasional low data quality, the reliability analyses were performed on nine subjects for narrow standing and on eight subjects for single-leg standing,

354

A.C. A˚berg et al. / Gait & Posture 34 (2011) 352–357

respectively. Statistical analyses were conducted using the Statistical Package for the Social Sciences (SPSS Inc., Chicago, IL, USA).

3. Results All subjects were independent in ADL. Anthropometrical measures and the results of the clinical assessments are presented in Table 1.

was indicated by significant (p < 0.05) rP in the range 0.82–0.97 for T and 0.73–0.97 for Rt sessions. However, significant systematic bias (p  0.05) was also detected, indicating that the Km calculations showed slightly lower values as illustrated by Fig. 2 and by a Bland–Altman plot (Fig. 3). Absence of significant systematic bias was only seen (in both directions) in T-sessions during narrow standing (Fig. 2). 3.3. Proportional contribution of the balance mechanisms

3.1. Reliability It is to be noted that in single-leg standing, the number of successful reiterations per trial and hence, duration per trial varied within and between subjects (Appendix 1). It is also noteworthy that the obtained contributions of M1 and M2 are not necessarily in the range 0–100%, and may, hence, not sum to 100% [14] (Fig. 1). Additionally, some results in Fig. 1 may appear to indicate a significant shift towards lower values at the retest sessions. However, analysis by repeated measures ANOVA did not show any significant individual between-sessions differences. The raw individual data are presented in more detail in Appendix 2A and 2B. All comparisons between the ICCs revealed some degree of CI overlap (Table 2). The investigation by the F-test ANOVA of inter-session reliability showed no significant bias. For narrow standing, the ICCs were 0.73–0.92, indicating good reliability for the M2 scores in both directions (Table 2). However, for M1 (in both directions) calculated from K–Km data, the ICCs were only 0.35 and 0.45, and some T  Rt differences exceeded the estimated MDCs (Table 2, Appendix 3). For the total sample of single-leg standing, the ICCs of 0.75–0.96 indicated good reliability, though some of the T  Rt differences were higher than the MDCs for three subjects in standing on the right leg and two subjects in standing on the left leg (Table 2, Appendix 3). Correspondingly, the majority of ICCs for the intra-session reliability tests of 0.61–0.94 indicated good reliability, but poor reliability (0.42 and 0.58, respectively) was shown for M2 in ML (from Km data) and M1 in AP direction calculated (from K–Km data) for the Rt session. 3.2. Agreement between methods An overview of the percentage usage of M2 based on Km and K– Km calculations, respectively, is shown in Fig. 2. A strong correlation between Km and K–Km calculation methods for M2

Analysis of the usage of M1 and M2 (based on Km and K–Km data) showed no significant differences between T and Rt sessions for any standing situation or between right or left leg in single-leg standing. Comparisons between narrow and single-leg standing showed higher M1 and lower M2 in narrow standing (p < 0.05). Significant differences (p < 0.05) between AP and ML directions were shown only in single-leg standing (right and left) for both M1 and M2, that is, M1 was more used in AP than in ML direction, whereas M2 dominated in the ML direction (Fig. 1). The average M1:M2 proportion was approximately 9:1 in both AP and ML directions in narrow standing, and about 2:1 in AP and 1:2 in ML directions for single-leg standing.

4. Discussion The main findings of this study were that in the investigated subjects, standing balance in terms of utilization of M1 and M2 can be measured with acceptable inter- and intra-session reliability; that both Km and K–Km based calculations may be useful for M2 estimation; and that more of M2 is used in the ML than in the AP direction in single-leg standing. The tested measurements and calculations might, hence, be a constructive way of quantifying the declining balance capacity in older people, which is particularly evident in single-leg standing [22] as well as in ML balance control [23,24]. The participants were of an older age (mean = 74.5 years) and physically fit. They all reported independency in ADL and excellent fall efficacy, showed physical performances corresponding to normative data for their ages in the TUG [8] and self selected gait speed [25], along with five-chair rises [16] and hand grip strength exceeding age and gender matched normative data [26]. The overall results point towards acceptable inter- and intrasession reliability of M1 and M2 measurements in both AP and ML

Fig. 1. Mean values of the percentage usage of the balance mechanisms ‘moving centre of pressure’ (M1) and ‘segment acceleration’ (M2), in anterio-posterior (AP) and mediolateral (ML) directions relating to both test (T) and retest (Rt) sessions during the tasks of narrow standing and single-leg standing on the right and left leg, respectively. Calculations for the presented M1 and M2 results were based on a combination of both kinetic and kinematic data.

A.C. A˚berg et al. / Gait & Posture 34 (2011) 352–357

355

Table 2 Results of reliability analyses of the percentage usage of the balance mechanisms ‘moving centre of pressure’ (M1) and ‘segment acceleration’ (M2), in anterio-posterior (AP) and medio-lateral (ML) directions of both test (T) and retest (Rt) sessions, including between session differences (T  Rt) during the tasks of narrow standing and single-leg standing on the right and left leg, respectively. The balance mechanisms were calculated based on either a combination of both kinetic (K) and kinematic (Km) data (K–Km) or the Km data only concerning M2. Narrow standing (n = 9)

Single-leg standing, right leg (n = 8)

Single-leg standing, left leg (n = 8)

M1,AP (%)

M2,AP (%)

M1,ML (%)

M2,ML (%)

M1,AP (%)

M2,AP (%)

M1,ML (%)

M2,ML (%)

M1,AP (%)

M2,AP (%)

M1,ML (%)

M2,ML (%)

K–Km

Km

K–Km

K–Km

Km

K–Km

K–Km

Km

K–Km

K–Km

Km

K–Km

K–Km

Km

K–Km

K–Km

Km

K–Km

10 (9)

75 (7)

4 (2)

4 (4)

65 (20)

25 (13)

44 (19)

40 (29)

48 (21)

67 (25)

70 (16)

25 (14)

41 (19)

37 (25)

51 (19)

69 (22)

9 (6)

74 (6)

6 (2)

4 (3)

56 (22)

36 (18)

56 (22)

33 (24)

53 (22)

73 (21)

61 (15)

30 (14)

49 (17)

31 (20)

51 (19)

73 (13)

3 (3)

7 (5)

2 (1)

2 (1)

12 (8)

13 (13)

13 (14)

9 (10)

8 (9)

11 (12)

11 (8)

7 (4)

9 (10)

8 (9)

6 (4)

11 (12)

0.92 0.66 0.98 3 (6) 8

0.35 0.00 0.85 6 (12) 17

0.73 0.00 0.94 1 (3) 4

0.91 0.59 0.98 1 (3) 4

0.92 0.62 0.98 8 (15) 22

0.78 0.00 0.96 9 (18) 26

0.83 0.13 0.96 11 (22) 31

0.95 0.73 0.99 8 (17) 24

0.93 0.65 0.99 8 (15) 22

0.87 0.37 0.97 11 (21) 30

0.87 0.36 0.97 7 (15) 21

0.95 0.74 0.99 4 (9) 12

0.89 0.46 0.98 8 (16) 22

0.95 0.74 0.99 7 (14) 20

0.96 0.82 0.99 5 (10) 14

0.75 0.00 0.95 11 (22) 32

0.91 0.75 0.98 3 (5) 8

0.78 0.48 0.94 4 (7) 10

0.68 0.31 0.91 2 (3) 5

0.86 0.63 0.96 2 (3) 5

0.71 0.33 0.93 10 (19) 27

0.64 0.22 0.90 8 (16) 23

0.67 0.27 0.91 10 (20) 29

0.94 0.81 0.99 7 (14) 20

0.89 0.69 0.98 7 (14) 20

0.86 0.62 0.97 10 (19) 28

0.75 0.40 0.94 9 (17) 24

0.83 0.54 0.96 6 (13) 18

0.83 0.55 0.96 9 (17) 24

0.82 0.53 0.96 11 (23) 32

0.79 0.47 0.95 10 (19) 27

0.77 0.43 0.94 12 (23) 33

0.85 0.61 0.96 2 (4) 6

0.73 0.39 0.92 4 (7) 10

0.42 0.01 0.80 2 (4) 5

0.78 0.48 0.94 2 (3) 4

0.82 0.53 0.96 10 (19) 27

0.85 0.59 0.96 7 (14) 20

0.70 0.31 0.92 13 (25) 36

0.90 0.71 0.98 8 (16) 22

0.88 0.66 0.97 8 (15) 21

0.86 0.61 0.97 9 (17) 24

0.58 0.15 0.88 12 (23) 33

0.78 0.45 0.95 7 (15) 21

0.64 0.23 0.90 12 (23) 33

0.85 0.60 0.97 9 (17) 25

0.88 0.65 0.97 7 (14) 19

0.61 0.19 0.89 12 (23) 32

Balance mechanism utilization T Mean 89 8 (5) (7) (SD) Rt Mean 84 7 (SD) (6) (4) T  Rt Mean Diff 6 2 (5) (2) (SD) Intersession reliability ICC3,3 0.45 0.91 95% CI 0.00 0.60 +95% CI 0.84 0.98 5 2 SEM (95% CI) (9) (4) MDC95% 13 6 Intrasession reliability T ICC3,1 0.64 0.69 95% CI 0.25 0.32 +95% CI 0.89 0.91 SEM 3 4 (95% CI) (7) (7) MDC95% 10 10 Rt ICC3,1 0.67 0.84 95% CI 0.30 0.59 +95% CI 0.90 0.96 SEM 4 2 (95% CI) (8) (3) MDC95% 11 5

Diff = difference; ICC = intraclass correlation coefficient; SEM = standard error of measurement; CI = confidence interval; MDC = minimum detectable change.

Fig. 2. Mean values of the percentage usage of the balance mechanism ‘segment acceleration’ (M2) calculated based on either a combination of both kinetic (K) and kinematic (Km) data (K–Km) or Km data only. Results are given for M2 usage in anterio-posterior (AP) and medio-lateral (ML) directions concerning both test (T) and retest (Rt) sessions, during the tasks of narrow standing and single-leg standing on the right and left leg, respectively. Significant differences (p  0.05) between K–Km and Km, implying a systematic bias, are indicated by *.

directions. This was indicated by the ICCs of single-leg standing, although poor reliability (ICCs < 0.45) for narrow standing was shown for M1 (both directions) according the K–Km calculations. However, when considering the fact that ICC is a relative measure of the between-subject and the within-subject variation [20], low

ICCs of M1 for narrow standing can be explained by the low between-subject variation [21]. The inter-session reliability, hence, appears to be good, despite the low ICCs for M1 for narrow standing. This was further supported by the comparisons of SEM [27] and MDCs – the absolute indicator of reliability [28] – showing

356

A.C. A˚berg et al. / Gait & Posture 34 (2011) 352–357

Fig. 3. Bland–Altman plot concerning the agreement between the two calculation methods – based on either kinematic (Km) or a combination of both kinetic (K) and Km data (K–Km) – for the percentage usage of the balance mechanism ‘segment acceleration’ (M2). Data points are presented for each subject in anterio-posterior and medio-lateral directions, related to test and retest sessions during the tasks of narrow standing (n = 9) and single-leg standing on the right and on left leg (n = 8), respectively. LOA = limits of agreement.

that SEM of M1 (both directions) in narrow standing was lower than in single-leg standing, indicating a lower within-subject variation for narrow standing. Moreover, the ICCs (range 0.87–0.95) for singleleg standing of M1 (both directions) showed excellent reliability. This however, may be somewhat overestimated due to the higher between-subject variability of this task. Likewise, the ICCs for intra-session reliability mostly indicated good reliability with low reliability (ICCs < 0.58) only for two measurements in Rt sessions. It should be considered that during the single-leg standing tasks, some individuals did not manage to maintain balance for the entire trial of 30 s, which resulted in comparisons of trials with different durations and number of reiterations per trial, which should have a negative impact on the reliability. As previously shown, sample duration per trial and number of trials per session impact the stability and reliability of the measures [29]. Hence, increasing the number of trials per session should improve the reliability [29], though in an elderly sample like ours the risk of fatigue must be considered. According to Pearson’s correlation coefficient, there was a strong correlation between the two methods Km and K–Km for M2 calculations. However, a systematic bias was detected, indicating that the Km calculations showed slightly lower M2 values. This bias may be explained by a negative correlation between errors in the calculated terms of Eq. (1) – when these are calculated from identical raw data – which influences the estimated slope in the linear regression. As expected, our results revealed significant differences between performance in narrow and single-leg standing, which could be explained by the diverse characteristics and balance challenges of the tasks. In narrow standing, M1 was the most dominant mechanism, while M2 was almost negligible, whereas in single-leg standing the average M1:M2 proportion was about 2:1 in AP, and 1:2 in the ML direction. This can be compared to the results obtained in the initial M1 and M2 experiment involving ML balance of a young man [12], showing a proportional usage with about 2:1 in favour of M1 in single-leg standing, followed by a clear dominance of M2 in standing on a 4 cm wide bar. In line with previous studies [30], no balance performance differences (p < 0.05) were found between right and left single-leg standing. However, M1 usage dominated in AP direction whereas M2 was more used in ML direction during single-leg standing. This

is probably due to loading of the whole body weight on one foot, that is on a narrow BoS, entailing restricted possibility of M1 utilization in the ML direction. This interpretation corresponds to previous research showing that by gradually narrowing the BoS in single-leg standing on a beam, the participants switched from M1 to more M2 usage in the ML direction, which was most evident among elderly subjects [30]. Limitations of the present study are the small sample size and number of trials per sessions along with comparisons of trials with different durations and number of reiterations per trial which means that conclusions must be drawn with some caution. There are also three major sources of errors in the calculation of the balance mechanisms: (1) soft tissue deformation and movement causing extraneous movements of the markers (M1 and M2); (2) approximations implicit in the model of the human body (M1 and M2); and (3) twice differentiation of positions to achieve accelerations (M2, Km only). Calculating M2 using kinematic and force platform data (K–Km) does not require assessment of segment acceleration, and is therefore recommended. However, this restricts the use of the method to movements on force platforms. It is most likely that systematic errors have occurred, which remain unknown. Random errors are however assessed by the reliability measures presented here. Taking this into consideration, it is still concluded that the proportional M1:M2 usage seem to change in favour of M2 in more balance challenging standing tasks, which is most evident in the ML direction, and that these measures appear to constitute a reliable way of quantifying one important aspect of balance capacity in fit older people. Conflicts of interest There are no conflicts of interest concerning this study. Acknowledgements This study was supported by grants from the Swedish School of Sport and Health Sciences, Stockholm, Sweden. The authors thank PhD, RPt Anna Bjerkefors for contributing to this study and Statistician PhD Lars Berglund for valuable statistical advises. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.gaitpost.2011.05.025. References [1] Frank JS, Patla AE. Balance and mobility challenges in older adults: implications for preserving community mobility. Am J Prev Med 2003;25(3 Suppl. 2):157–63. [2] Pollock AS, Durward BR, Rowe PJ, Paul JP. What is balance? Clin Rehabil 2000;14(4):402–6. [3] Lyons RA, John A, Brophy S, Jones SJ, Johansen A, Kemp A, et al. Modification of the home environment for the reduction of injuries. Cochrane Database Syst Rev 2006;18(4):CD003600. [4] Ganz DA, Bao Y, Shekelle PG, Rubenstein LZ. Will my patient fall? JAMA 2007;3/ 297(1):77–86. [5] Huxham FE, Goldie PA, Patla AE. Theoretical considerations in balance assessment. Aust J Physiother 2001;47(2):89–100. [6] Tyson SF, Connell LA. How to measure balance in clinical practice. A systematic review of the psychometrics and clinical utility of measures of balance activity for neurological conditions. Clin Rehabil 2009;23(9):824–40. [7] Bohannon RW. Standing balance, lower extremity muscle strength, and walking performance of patients referred for physical therapy. Percept Mot Skills 1995;80(2):379–85. [8] Pondal M, del Ser T. Normative data and determinants for the timed ‘‘up and go’’ test in a population-based sample of elderly individuals without gait disturbances. J Geriatr Phys Ther 2008;31(2):57–63. [9] Duncan PW, Weiner DK, Chandler J, Studenski S. Functional reach: a new clinical measure of balance. J Gerontol 1990;45(6):M192–7.

A.C. A˚berg et al. / Gait & Posture 34 (2011) 352–357 [10] Cho BL, Scarpace D, Alexander NB. Tests of stepping as indicators of mobility, balance, and fall risk in balance-impaired older adults. J Am Geriatr Soc 2004;52(7):1168–73. [11] Berg KO, Wood-Dauphinee SL, Williams JI, Maki B. Measuring balance in the elderly: validation of an instrument. Can J Public Health 1992;83(Suppl. 2):S7–11. [12] Hof AL. The equations of motion for a standing human reveal three mechanisms for balance. J Biomech 2007;40(2):451–7. [13] Winter DA. ABC of balance during standing and walking. Waterloo, CA: Waterloo Biomechanics; 1995. [14] Halvorsen K. Comments on ‘‘The equations of motion for a standing human reveal three mechanisms for balance’’ (A. Hof, vol. 40, pp. 451–457). J Biomech 2010;43(16):3244–7. [15] Hellstrom K, Lindmark B, Fugl-Meyer A. The falls-efficacy scale, Swedish version: does it reflect clinically meaningful changes after stroke? Disabil Rehabil 2002;15/24(9):471–81. [16] Lindemann U, Claus H, Stuber M, Augat P, Muche R, Nikolaus T, et al. Measuring power during the sit-to-stand transfer. Eur J Appl Physiol 2003;89(5):466–70. [17] Dempster WT. Space requirements of the seated operator WADC Technical Report. Dayton, OH, USA: Wright-Patterson Air Force Base, OH; 1955. Report No.: TR-55-159. [18] Hof AL. Reply to letter to the editor about ‘‘The equations of motion for a standing human reveal three mechanisms for balance’’ by K. Halvorsen. J Biomech 2010;43(16):3247–8. [19] Van Huffel S, Vandewalle J. The total least squares problem: computational aspects and analysis. Philadelphia: SIAM-Society for Industrial and Applied Mathematics; 1987. [20] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull 1979;86(2):420–8.

357

[21] Weir JP. Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. J Strength Cond Res 2005;19(1):231–40. [22] Vereeck L, Wuyts F, Truijen S, Van de Heyning P. Clinical assessment of balance: normative data, and gender and age effects. Int J Audiol 2008;47(2):67–75. [23] Aberg AC, Frykberg GE, Halvorsen K. Medio-lateral stability of sit-to-walk performance in older individuals with and without fear of falling. Gait Posture 2010;31(4):438–43. [24] Schrager MA, Kelly VE, Price R, Ferrucci L, Shumway-Cook A. The effects of age on medio-lateral stability during normal and narrow base walking. Gait Posture 2008;28(3):466–71. [25] Bohannon RW. Comfortable and maximum walking speed of adults aged 20–79 years: reference values and determinants. Age Ageing 1997;26(1): 15–9. [26] Bohannon RW, Bear-Lehman J, Desrosiers J, Massy-Westropp N, Mathiowetz V. Average grip strength: a meta-analysis of data obtained with a Jamar dynamometer from individuals 75 years or more of age. J Geriatr Phys Ther 2007;30(1):28–30. [27] Looney MA. When is the intraclass correlation coefficient misleading? Meas Phys Educ Exerc Sci 2000;4(2):73–8. [28] Shrout PE. Measurement reliability and agreement in psychiatry. Stat Methods Med Res 1998;7(3):301–17. [29] Lafond D, Corriveau H, Hebert R, Prince F. Intrasession reliability of center of pressure measures of postural steadiness in healthy elderly people. Arch Phys Med Rehabil 2004;85(6):896–901. [30] Curtze C, Postema K, Akkermans HW, Otten B, Hof AL. The narrow ridge balance test: a measure for one-leg lateral balance control. Gait Posture 2010;32(4).