KEY WORDS: Exercise intervention, static balance, dynamic balance, aging.

Subjects
Once Institutional Review Board approval was obtained for this study, sixty, sedentary, females, aged 50- 75 years, from the Oklahoma City Metropolitan area, were recruited via flyers, e-mail, and newspaper ads. If potential subjects were classified as healthy (no serious medical illness), sedentary (Baecke physical activity questionnaire), and obtained medical clearance from their own personal physicians, they were then randomly assigned to one of three supervised training groups, ballates, step aerobics, or walking.

Only two of the subjects were premenopausal and 42 subjects were perimenopausal or postmenopausal, with 11 subjects on hormone replacement therapy. Over 50% of the women were taking a one-a-day vitamin or calcium supplements. There were also many subjects who were taking natural herbal supplements as well as other forms of medications for blood pressure, cholesterol, etc, however this was not taken into account for group randomization. Table 1 outlines the baseline characteristics of each exercise group. Groups did not significantly differ in age, height, weight, and body composition (DXA - Lunar Prodigy). The study began with 60 participants and 44 women completed the 13 weeks of training. Reasons for dropping out included time commitment, family emergencies, or excessive absences. None of the women dropped out as a result of injury during the exercise programs.

Pre and Post Test Measurements
Body composition
Dual Energy X-Ray Absorptiometry (DXA - Lunar Prodigy) was used to assess body composition (fat and bone-free fat free mass - BF-FFM) as well as bone mineral density (total body, proximal femur, lumbar spine). A trained DXA technician performed and analyzed each scan using the encore 2002 software (GE Medical Systems, version 6.70.021). Subject height was measured using a wall stadiometer and weight was measured using a TANITA BWB-800 digital scale.

Balance
Four different measurements of static and dynamic balance were analyzed by the NeuroCom Balance Master. Since there are several different components which contribute to balance, the NeuroCom gives an objective assessment of the sensory and voluntary motor control of balance. This system is comprised of fixed 18" X 60" dual force plate which measures the vertical forces exerted by one's feet. These tests were broken into two different categories: 1) Impairment Tests (Modified Clinical Test for Sensory Integration of Balance - mCTSIB) which looked at the effective use of visual, vestibular, somatosensory, automatic and voluntary motor skills that aid in balance and mobility during a variety of changing task conditions; and 2) Functional Limitation Tests (Unilateral Stance - US; Tandem Walk - TW; and Step Quick and Turn - SQT) which looked at the ability of one to safely and efficiently perform mobility tasks in every day activities. In order to maintain consistency and for subject convenience, each of the balance tests were administered in the same order and by the same technician. Subjects were able to watch an instructional video of each required task on a computer screen before each individual test was administered. All tests began by placing the feet according to the foot placement instructions given on the computer screen. The analysis of each test was given in both a numeric version using percentages, ratios, etc, and a comprehensive version using graphs and pictures of movement patterns.

Modified Clinical Test for Sensory Integration of Balance (mCTSIB)
This test measures several components of functional balance by quantifying postural sway velocity while changing the subject's sensory condition. This is accomplished by changing the surface on which the subject stands, from a firm to a foam surface, and by asking the subject to stand on these surfaces with eyes opened and eyes closed. By asking the subject to perform this test with the changing conditions, both the sensory and visual components of balance were accurately assessed. The force plate detected sway patterns outside the subject's center of gravity during a 10 second trial period, therefore giving a measurement of sway velocity in degrees/second. The greater absence of postural sway after being told to hold still would indicate better postural stability and balance. The variables obtained from this test include Sway Velocity Firm Composite score, Sway Velocity Foam Composite score, Mean Center of Gravity (COF) Sway Velocity, and COF Alignment. Composite Sway Velocity is found by adding the two scores for eyes opened and closed and dividing by two for each condition (firm and foam) which creates two variables. The Sway Velocity for each variable is the ratio of the distance traveled by the center of gravity to the time of the trial. The average of the Sway Velocities for each condition gives a Mean COG Sway Velocity, with lower scores indicating greater balance. COG Alignment reflects the subject's center of gravity over their base of support. Three trials were obtained for each condition and an average used in latter analyses.

Functional Limitation Tests
Unilateral Stance (US)
This test measures balance by quantifying postural sway velocity while the subject stands on either the right or left foot, with eyes opened and with eyes closed, for 10 seconds. The greater absence of postural sway after being told to hold still would indicate better postural stability and balance. Sway Velocity Eyes Open and Sway Velocity Eyes Closed are the two variables used in this analysis. The sway velocity for each variable is the ratio of the distance traveled by the center of gravity to the time of the trial. A mean of three trials was used in latter analyses.

Tandem Walk (TW)
This test quantifies characteristics of gait as the subject walks the length of a force platform, walking heel to toe. The measured parameters include step width, speed, and end sway velocity. A mean of three trials was used in later analyses.

Step Quick and Turn (SQT)
This test measures balance by quantifying turn performance after taking two steps forward and pivoting 180 degrees and taking two steps back to the original starting position. The measured parameters are turn-time and turn-sway velocity. A mean of three trials was used in later analyses.

Interventions
There were three different supervised interventions that the subjects were randomly assigned, Ballates, step aerobics, and walking. Each subject was required to exercise three days per week (Monday, Wednesday, and Friday) from 7:00am to 8:00am for 13 weeks.

Ballates: The Ballates training was a body weight resistance class using a stability and medicine ball. It concentrated on core muscle groups that are imperative for maintaining balance. The subjects performed a combination of aerobic exercises focusing on the lower extremities and abdominal core as well as several forms of abdominal crunches using both the medicine and stability ball. Exercises for this class included a combination of standing and sitting designed to emphasize kinesthetic awareness. Unilateral weight shifts with accompanying rotation of the stability ball to strengthen ankle stability. Resistance moves were performed to increase muscle strength; standing exercises included: side twists, walking side twists, ball sweeps from side to side, ball bounce and shuffle from left to right, ball toss and catch, plie' squats while bringing the ball up and over the head. Seated exercises included: alternating leg lifts, while lying on the ball walking hands out to a bridge and walking back in, ball straddles, and lateral hip moves. As the exercises got easier the subjects were asked to increase the intensity of exercise by means of situating the body/torque position so that more resistance was created.

Step aerobics: Step aerobics was a low to moderate intensity aerobics class, involving both simple and complex step and dance combinations as well as muscle toning with hand- held weights. This class was designed so that the subject could choose their own exercise intensity by choosing the height of the step used (6 to 12 inch or 15.24 to 30.48 cm). The equipment needed for this class was one step bench, two to five pound hand held weights, and a soft mat. Each class began with a 5 minute warm-up. The warm-up consisted of exercises such as side-stepping, hamstring curls, modified jumping-jacks, marching high and low, stretch lunges, calf, hamstring, and quadriceps stretches. After the warm-up was complete there was a 10 minute series of stepping exercises as well as step combinations: basic left/right, curb walking, step up and over, step up and straddle, step and kick, step and lunge, walking around the step. Following the first series of step exercise, time was allotted for a water break. Next, another 10 to 15 minutes of step combinations were performed. After the participants were comfortable with the steps, hand-held weights were added and a series of bicep curls, tricep extensions, lateral raises, etc. were performed while adding in basic lunges for 10 minutes. A cool-down of exercises consisting of low-intensity basic left/right stepping and side steps were performed for five minutes. Finally, there was five minutes of core conditioning, which consisted of crunches, sit-ups, and leg lifts followed by five minutes of stretching and breathing exercises. The intensity of this program was increased by speeding up the cadence of the music over the thirteen weeks of exercise.

Walking: The walking group, who were used as an 'exercise control group', walked around an indoor track (six laps, equaled one mile) at each individual's own pace, but were constantly encouraged to complete as many laps as possible during the exercise session. Heart rates were recorded at the beginning, midpoint and end of each exercise session as well as the total number of laps completed.

Statistical analyses
Statistical analyses were performed using SPSS^® for Windows^® (version 12.0). Descriptive statistics (means ± standard errors) were performed for all variables to describe each group's body composition and balance measurements. Baseline comparisons between the three groups for physical characteristics (age, height, weight, body composition) and pre-training measures of balance were determined by a one-way ANOVA. A two-way repeated measures ANOVA [group (3) × trial (2)] was also performed on all measurements. A Bonferroni post hoc procedure was used if there was a significant group effect. When significant group by trial interaction effects were found, dependent t-tests within each training group were used to determine significant pre to post training differences. The level of significance was set at p < 0.05.

Effect size: By examining the means and standard deviations of previous studies (Cosio-Lima et al., 2003; Liu-Ambrose et al., 2004; Shigematsu et al., 2002) the effect size of those studies were calculated. Assuming that a power of 0.80 was needed and the calculated effect size for most standard measures of balance was 1.00, it was determined that a minimum of 13 subjects would be needed for each of the three training groups (Cohen, 1988).

As mentioned earlier in the subject section, there were no baseline differences between the three groups for any physical characteristic (age, height, weight, body composition). Additionally, there were no group differences for any pre-training balance measure with the exception of one value. Only end sway velocity for the Tandem Walk was statistically significant (p = 0.017) between the three groups with the Ballates group starting out significantly worse (5.9 deg·sec^-1) than the other two groups (Step Aerobics = 4.2 deg·sec^-1 and Walking = 3.9 deg·sec^-1).

There were significant trial effects for percent fat variables, with small but significant decreases occurring in total body (p = 0.018), legs (p = 0.001) and trunk (p = 0.029) percent fat (Table 2). Additionally, there were small but significant increases in total bone free fat-free mass (BF-FFM) (p = 0.038) and in trunk BF-FFM (p = 0.019) for all three groups (Table 2). However, since there is a two to three percent measurement error associated with body composition measurements from DXA, most of these changes, although statistically significant, may not reflect a true biological change with training.

There were four pre and post-test measurements of balance, mCTSIB, Unilateral Stance, Tandem Walk and Step Quick and Turn, which looked at both sensory and functional balance. A mean was taken of each of the three trials performed for each test.

Modified Clinical Test for Sensory Integration of Balance (mCTSIB)
"The mCTSIB analyzes one's sensory ability to compensate when one of the contributing factors, from the nervous, visual, vestibular, and somatosensory systems, are taken away. This test was simulated on a firm surface and then on a foam surface with eyes opened and eyes closed. There was a significant group by trial interaction for sway velocity on a firm surface with eyes closed (p < 0.05; Figure 1) and for sway velocity on a foam surface with eyes closed (p < 0.05; Figure 2).

Sway velocity on a firm surface with eyes closed remained unchanged for the walking group after the intervention (0. 291 deg·sec^-1) while the step aerobics group improved their balance as demonstrated by a decreased sway velocity (0.391 to 0.255 deg·sec^-1), whereas the Ballates group had greater sway velocity after the intervention when compared to pre-training values (0.275 to 0.319 deg·sec^-1) indicating a decrease in static balance performance. On the other hand, sway velocity decreased (improved balance) for both the walking and step aerobics groups (1.49 to 1.23 deg·sec^-1 and 1.46 to 1.29 deg·sec^-1, respectively) on a foam surface with eyes closed, while the Ballates group once again increased sway velocity (decreased balance) following the training (1.39 to 1.64 deg·sec^-1).

Unilateral stance
Unilateral stance not only measures one's ability to compensate when senses are taken away, but also one's functional ability to stand on one leg. Sway velocity and length of time on the right foot, with eyes closed both approached statistical significance for the group effect (p < 0.062 and p < 0.074, respectively) but there were no statistically significant group, trial, or group by trial interactions for any of the static balance variables measured during the unilateral stance test.

Tandem walk
There was a significant trial effect (p < 0.01) but no group or group by trial interaction for the distance covered in centimeters during the 10 second tandem walk (Figure 3). All three groups improved in this aspect of dynamic balance (Ballates: 32.11 to 34.69 cm·sec^-1; step aerobics: 27.37 to 34.66 cm·sec^-1); and walking: 27.00 to 36.14 m·sec^-1).

Step quick turn
After analyzing turn time on the right foot, a significant trial effect (p < 0.01) but no group or group by trial interaction was determined indicating that all groups improved in turn time following their respective training intervention (Ballates, 26%; step aerobics, 18.9%; and walking, 6.3%; Figure 4). There was also a significant trial effect for turn sway on the right foot (p < 0.05) but no significant group or group by trial interaction, once again indicating that all three groups improved dynamic balance following the training (Ballates, 19%; step aerobics, 1.8%; and walking, 9.3%; Figure 5).

The fact that the modified clinical tests for sensory integration of balance (mCTSIB) reflected improved measures of static balance for the walking and step aerobics groups but not for the Ballates group may be due to the fact that the walking and step aerobics interventions focused on leg strength and endurance whereas the Ballates intervention, which utilized stability ball exercises, focused on strengthening the abdominal core. These results are similar to findings of Urbscheit and Wiegand, 2002 but contrary to a study by Rogers et al., 2001 who reported a significant decrease in sway velocity and improved static balance following training with a stability ball. Possible differences between the current study and the Rogers et al., 2001 study included the size, gender, and ages of the subjects. The previous study reported results based on 17 subjects (5 men and 12 women) between the ages of 61 and 77 years, whereas the current study utilized 44 females between the ages of 50 and 75 years (mean age approximately 58 years). Although, subjects were screened using the Baecke physical activity questionnaire the younger subjects in the current study may have been more fit and or healthy than the subjects in the Rogers et al., 2001 study, prior to the interventions, which would make improvements in balance more difficult to detect.

We did not find any improvements in the variables associated with unilateral stance following all three exercise interventions. Cosio-Lima et al., 2003 reported significant improvements in unilateral stance scores with eyes open and eyes closed following a program of sit-ups and back extensions using the stability ball compared to performing these same exercises on the floor, however, their subjects had a mean age of 23 years, whereas the mean age of the current study was 58 years.

Even though all three exercise groups significantly improved the distance covered during the tandem walk test following the training, the walking group and the step aerobics groups had larger improvements when compared to the Ballates group. This finding might be expected considering the nature of the training that each group completed, for example the walking group was encouraged to walk during training at a faster pace than normal as well as for the stepping group to increase the intensity or speed of cadence in which they were stepping. On the other hand, even though all three groups improved their turn sway during the step quick and turn test, the Ballates group now had the largest improvements again emphasizing the specificity of the training programs (Cosio-Lima et al., 2003; DiBrezzo et al., 2005), which in this case would be the effects of the core strengthening of the training exercise and use of core muscles when making a 180 degree turn.

One possible limitation of this study was that the younger age of the subjects, which ranged from 50 to 75 years (11 subjects were older than 63 years and only 2 subjects were greater than 70 years of age), with a mean age of 58, when compared to many of the other studies which recruited men and women aged 65 years and above. This could explain why there were no group differences or group by trial interactions following the training since these subjects already had good postural stability and balance even prior to training. Another limitation of the current findings when trying to compare the results to previous studies is the fact that most other studies measured balance using the Berg Balance Scale as well as functional field tests.

There are almost 400 potential fall risk factors that have been identified, but the most important of these are the intrinsic factors. This study and other studies have demonstrated that exercise can have a positive effect on the risk factors associated with balance. It seems as though no particular type of exercise is superior to the other forms of balance training, with the exception of prescribing a combination of these exercises, i.e. stability ball plus weight training. Additional research needs to investigate the effectiveness of stability ball training in combination with dance based or step aerobics, or in combination with walking exercises. If similar improvements in balance can be realized when combining home based exercise using the stability ball with walking compared to the traditional resistance based weight training programs, then more elderly individuals may be able to enhance their balance and reduce their risk of falling without having to train at a gym. Before 2001, no studies were found that looked at the efficiency of stability ball training. In 2002, Schlicht et al. examined the safety and/or injury risk of using the stability ball in an exercise routine. Since there is a limited number of research papers that have used the stability ball in balance training and the results that have been presented are somewhat contradictory (Cosio-Lima et al., 2003; Rogers et al., 2001; DiBrezzo et al., 2005) more research is needed that compares the stability ball to balance based exercise programs like Tai Chi or to traditionally based resistance training programs.

In conclusion, the Ballates training program was effective for improving dynamic balance (TW and SQT) but not for measures associated with static balance (mCTSIB tests involving the measure of sway velocity on firm and foam surfaces). In contrast, the step aerobics and walking programs improved measures of both static and dynamic balance. Therefore, our findings suggest that increasing physical activity levels, in general, may improve measures of both static and dynamic balance.

• Exercise training can improve balance
• Need to consider both static and dynamic aspects of balance individually
• Improved balance can reduce the risk of fall