Lower-extremity resistance training on unstable surfaces improves proxies of muscle strength, power and balance in healthy older adults: a randomised control trial

Participants

Eighty-three (48 female, 35 male) community-dwelling older adults aged 65 to 80 years were included in a stratified-randomised control trial [22]. Recruitment was carried out by placing an advertisement in the local newspaper and during a public information meeting at the local town hall. Eligibility was tested according to the recommendations of Gschwind and colleagues [23]. Inclusion criteria were determined as the ability to walk independently without any gait aid. To safeguard participants and account for possible cognitive and mental health conditions or any other neurological, musculoskeletal or heart-related disease, participants were excluded based on pathological ratings of the Clock Drawing Test (CDT), the Mini-Mental-State-Examination (MMSE, < 24 points), the Falls Efficacy Scale – International (FES-I, > 24 points), the Geriatric Depression Scale (GDS, > 9 points), the Freiburg Questionnaire of Physical Activity (FQoPA, < 1 h) and the Frontal Assessment Battery (FAB-D, < 18 points). Participants’ baseline characteristics are presented in Table 1. Figure 1 shows a flow chart of the study design. The study was approved by the local ethics committee of the University of Kassel (E05201401) and was carried out in accordance with ethical standards of the latest Declaration of Helsinki (WMA, Oct. 2013). Prior to the commencement of the study, written informed consent was obtained after providing information on aims and potential risks of the investigation. The study was designed according to the CONSORT publishing and reporting guidelines [24].

Characteristic	M-SRT (n = 27)		M-URT (n = 26)		F-URT (n = 22)		Baseline difference
	M	SD	M	SD	M	SD	p-value
Age (years)	71.3	4.1	70.0	4.4	69.8	4.3	.409
Body height (cm)	170.1	8.3	172.9	8.6	169.8	7.4	.408
Body mass (kg)	76.7	15.7	77.7	13.1	74.7	13.0	.756
Body mass index (kg/m2)	26.9	3.7	25.9	3.4	25.8	3.8	.952
Sex (f/m)	17/10		13 / 13		13 / 9
Physical activity (h/week)	9.3	4.7	10.5	8.4	9.8	5.0	.820
MMSE	28.4	1.3	28.5	1.1	27.9	1.4	.275
CDT	all participants were classified as non-pathological
GDS	1.2	0.2	0.7	0.2	1.1	0.2	.127
FAB_D	16.0	1.8	16.2	1.9	16.6	1.4	.550

M-SRT machine based resistance training on stable surfaces; M-URT machine based resistance training on unstable surfaces; F-URT free-weight resistance training on unstable surfaces; M mean; SD standard deviation; f female; m male; MMSE Mini Mental State Examination; CDT Clock Drawing Test; GDS Geriatric Depression Scale; FAB_D Frontal Assessment Battery, German Version

Measures

Data was collected in the biomechanics laboratory of the University of Kassel, Germany. Strength, power and balance assessments were conducted by NE and a student assistant. Questionnaires were completed in separated rooms by different student assistants. The allocation of participants to the assessors was carried out randomly. All tests were explained and conducted using standardised verbal instructions regarding the test procedure. All assessors, except NE, were unaware of a participant’s group affiliation during post-testing. A single assessment lasted 90 min per participant.

Maximal isometric leg extension strength, the functional reach test and the chair rise test were considered to be primary measures, as they are thoroughly investigated and therefore reliable measures.

Assessment of strength and power performance

Well established clinical and biomechanical tests were administered to measure outcomes in muscle strength and power. In accordance with the recommendations of Granacher et al. [19], strength and power assessments were performed after balance assessments to reduce interfering effects of muscle fatigue. Strength, power and balance tests were administered in a randomised order within each respective block. One practice trial was provided for every test. Test procedures were conducted according to the recommendations of Gschwind et al. [23] if not stated otherwise. Two test trials were carried out using the mean for further statistical analysis, except for maximal isometric leg extension strength (ILES) and hand grip strength. In this case the better value of two consecutive trials was used for statistical analysis. Sufficient (at least one minute) recovery periods were provided between trials to limit the effects of fatigue.

Maximal isometric leg extension strength (ILES) was examined with a cable pull device (Takei A5002, Fitness Monitors, Wrexham, England) in an upright body posture. To ensure upright posture participants were instructed to maintain contact between their shoulder and the wall and to resist lifting their scapula while pulling. Individual cable lengths were chosen to ensure a knee angle of approximately 135°. Participants were asked to “pull initially with a moderate intensity and slowly increase the intensity to maximum exertion while keeping the upper body extended and upright”. The test was repeated with at least one minute intervals between measurements. The ILES shows excellent test-retest reliability (ICC = .98) for leg extension strength [25].

To measure handgrip strength a Takei hand dynamometer (Takei A5401, Fitness Monitors, Wrexham, England) was used. Participants stood upright with their arm aligned to the body and squeezed the device as hard as they could, using their dominant hand. The width of the handle was adjusted to the participant’s hand size. In order to do this, the intermediate phalanges had to be placed on the inner handle. The Takei handy dynamometer maintains excellent test-retest reliability (ICC = .95) [26].

In addition to isometric strength assessment, power tests were conducted. Supplementary to the standard Chair Rise Test (CRT) on stable surfaces, test trials were also recorded while standing on a foam pad (AIREX^©). Participants were instructed to stand up and sit down five times as quickly as possible without using their arms. They were advised to fold their arms across the upper body. Time was measured by a stopwatch to the nearest .01 s. After the countdown “ready-set-go”, testing time was started and stopped when participants were sitting down for the fifth time. For the CRT, high test-retest reliability has been shown (ICC = .89) [27].

In addition, a stair climb power test (SCPT) was administered. This test has shown meaningful associations with mobility performance and strength measures in older adults [28]. Participants were instructed to walk quickly but safely up and down a nine step flight of stairs (step height: 17 cm). Time was started after the cue to go and stopped when the second foot reached the top step and/or the floor, respectively. Use of the handrail was permitted for safety reasons. Time was measured with an ordinary stopwatch to the nearest .01 s. Ascent and descent times were recorded separately and power was calculated with the following formula: \( \mathrm{P}=\frac{\left(\mathrm{M}\ \mathrm{x}\ \mathrm{D}\right) \times \mathrm{g}\ }{\mathrm{t}} \), where P = power (Watt), M = body mass (kg), D = vertical distance covered (m), t = time (s) and g = 9.8 m/s² (constant of gravity). Test-retest reliability is shown to be excellent (ICC = .99) [28].

Assessment of balance performance

Proactive balance was tested using the timed-up-and-go test (TUG) and the functional reach test (FRT). For the TUG, participants were asked to rise from a chair and walk three meters at their normal walking speed around a cone, return to the chair and sit down. Time for the TUG was recorded with a stopwatch to the nearest .01 s on the command “ready-set-go” and stopped as soon as the participants sat down. The TUG has shown excellent test-retest reliability (ICC = .99) in older adults [29]. The FRT measures the maximal distance participants were able to reach forward while standing. For this test participants were instructed to lift their dominant arm and to reach forward as far as they could without taking a step forward. Maximal reach distance (cm) was recorded. The FRT showed excellent test-retest reliability with older adults (ICC = .92) [30]. In addition to the standard FRT on stable surfaces, test trials were recorded while standing on a foam pad (AIREX^©).

Dynamic steady-state balance was tested while walking on a 10-m walkway, measuring temporal-spatial gait variables, i.e., stride length (cm), stride velocity (m/s), stride width (cm), and double support (%), using a two-dimensional OptoGait^© system (Microgate, Bolzano, Italy). In addition, the coefficient of variation (CV) was calculated as follows: CV (%) = (SD/mean) × 100. The gait variables showed high concurrent validity between the OptoGait^© system and a previously validated system (ICCs = .93 to .99) [31]. Participants were asked to walk for 10 m in their own footwear at a self-selected pace three times to calculate test-retest reliability. A three-minute interval was given between individual trials to rest, save the data as well as to prepare for the next trial. At the start and the end of the walkway, sufficient distance was provided to accelerate and decelerate safely. In addition, the first and last step was excluded from the analysis to eliminate possible acceleration and deceleration bias. Each trial was recorded at 1000 Hz using the manufacture approved OptoGait^© software, running on a laptop computer (Lenovo ThinkPad T530).

To test reactive balance the Push and Release Test (PRT) was used. The PRT rates the postural response to a sudden perturbation. Participants were instructed to push backwards against the examiner’s hands and to regain their balance after the examiner releases his hands. The number of steps required to regain balance was counted and the corresponding score was recorded (0 = 1 step, 1 = 2–3 small steps backwards with independent recovery, 2 = ≥4 steps with independent recovery, 3 = steps with assistance for recovery, 4 = fall or unable to stand without assistance). The PRT has shown high test-retest reliability (ICC = .84) [32].

Cognitive measures

Psychosocial functions were assessed using several questionnaires. Global cognition was tested using the 30-point MMSE, which is a test for assessing cognitive function and has shown high test-retest reliability (ICC = .89) [33]. The MMSE tests seven cognitive domains: orientation, registration, attention and calculation, recall, language and simple command following [33]. The CDT and FAB-D [34, 35] were used to assess executive function. The CDT is a screening tool for cognitive impairment and a measure of spatial dysfunction. Participants are asked to draw a clock with all its components and a self-selected time. Inter-rater reliability of the CDT has been shown to be excellent (IRR = .92) with sensitivity and specificity values of 50 and 84%, respectively [36]. The FAB-D consists of six neuropsychological tasks, evaluating executive function. Excellent inter-rater reliability has been found for the FAB-D too (IRR = .87). Furthermore, internal consistency has been found to be good (r = .78).

Questionnaires

Fall self-efficacy was measured using the German version of the FES-I [37, 38]. This test has shown excellent internal validity (Cronbach’s alpha = .96) and test-retest reliability (ICC = .96) when assessing the level of fear of falling [37, 39]. To assess health-related physical activity, exercise and the amount of energy expenditure, FQoPA was conducted [40]. Frey and colleagues [41] have demonstrated that FQoPA score correlates with maximum oxygen uptake, indicating high validity (r = .42).

Exercise interventions

Participants were stratified into three intervention groups based on equal distribution of age and gender ratio and baseline values. The allocation to one of the three training programmes occurred randomly, using a random generator [42]. Intervention group one followed a ‘traditional’ machine-based stable resistance training programme (M-SRT). Intervention group two (machine-based unstable resistance [M-URT]) followed a similar training programme with exercise-machines, but with additional unstable devices placed between the participant and the exercise-machine or floor respectively. The third intervention group conducted free-weight resistance training on unstable devices (F-URT) using dumbbells instead of exercise-machines. According to Behm and colleagues [12, 43, 44], free-weight resistance training inherits a certain degree of instability in comparison to machine-based resistance training. In consequence implementing different degrees of instability is achieved by distinguishing machine-based resistance training using unstable devices (i.e., a moderate degree of instability) and free-weight resistance training using unstable devices (i.e., a high degree of instability).

All intervention groups trained for 10 weeks, twice per week on non-consecutive days for 60 min each. The 10-week intervention period consisted of a 1 week introductory phase and three major training blocks lasting 3 weeks each. Training intensity was progressively and individually increased over the 10-week training programme by modulating load and sets for all groups and level of instability for groups M-URT and F-URT. After week one, four, and seven the training load (weight) was increased following one repetition maximum (1-RM) testing for each major exercise. Since the load of a 1-RM is too heavy for untrained elderly, training load was calculated using the prediction equation provided by Epley (as cited in Reynolders, Gordon, Robergs, [45]), showing .03% deviation of actual achieved 1-RM in squats with a correlation of r = .97 [46]. Instructors ensured that repetitions did not exceed 15–20, because 1-RM predication accuracy is higher with fewer repetitions [46]. Training under unstable surface conditions, especially with additional weight, implies a certain degree of accident risk. Due to this factor all instability exercises were observed by instructors and made secure with additional aids like boxes. Training was supervised by skilled instructors at all times. For the first two weeks the participant to instructor ratio was 5:1, thereafter 10:1.

Since effectiveness of resistance training when applied as a single means, in comparison to a non-exercising control group on measures of lower-extremity strength, power and balance has been frequently reported in randomised controlled studies [47, 48], reviews [49–51], and meta-analyses [3, 52, 53] M-SRT served as an active control group.

Statistics

An a priori power analysis using G*Power 3.1 [55] with an assumed type I error of .05 and a type II error of .10 (90% statistical power, correlation among groups: .5, nonsphericity correction: 1) was computed to determine an appropriate sample size to detect medium (.50 ≤ d ≤ .79) interaction effects. The calculations were based on a study assessing the effects of core strength training using unstable devices in older adults [19]. The analysis revealed the requirement of 54 participants (18 per group) to obtain medium “time × group” interaction effects. Considering the likelihood of dropouts, 83 participants were recruited to compensate for a possible dropout rate of ~20%.

Prior to the main analysis, normal distribution was checked by visual inspection and tested with the Shapiro Wilk test for each dependent variable. In addition, Levene’s test for equality of variance was conducted. Baseline differences were tested between groups with a one-way ANOVA or a Kruskal-Wallis test depending on distribution and homogeneity. No differences were found (p ≥ .067). A 3 (group: M-SRT, M-URT, F-URT) × 2 (time: pre-test, post-test) ANOVA with repeated measures on time was conducted. Ryan-Holm-Bonferroni [56] adjusted post-hoc tests (dependent t-tests, Wilcoxon tests) were used to detect statistically significant pre- to post-test differences within each group. Ryan-Holm-Bonferroni corrected p-values were reported. In the case of distribution or homogeneity violations, Kruskal-Wallis one-way analyses of variance and Friedman tests were performed for non-parametrical variables and to control results of parametrical tests. If differences were detected, non-parametric test results would be expressed. In addition, differences in absolute training intensity within the last training block were analysed. Therefore, the training load of the squat movement, which was a common exercise to all groups, was used. Differences were computed using a one-way ANOVA. Post-hoc applied independent t-tests were used to identify differences between groups. Changes for all variables within groups were calculated with the formula ∆% = ((Mean_pre/Mean_post) – 1) × 100.

To improve readability and homogeneity, effect sizes (Cohen’s d) were calculated for all statistical tests [57]. Following Cohen [57], d-values ≤ .49 indicate small effects, .50 ≤ d ≤ .79 indicate medium effects, and d ≥ .80 indicate large effects. Significance level was set at α = 5%. All analyses were performed using SPSS version 21.0 (SPSS Inc., Chicago, IL, USA).