Balance performance and step width in noninstitutionalized, elderly, female fallers and nonfallers.

Balance Performance and Step Width in Noninstitutionalized, Elderly, Female Fallers and Nonfallers Falls and subsequent injuries are common occurrences among the elderly population. One third to one half of the population over 65 years oldwill fall at least once per year. [1,2] In 1983, falls were the leading cause of accidental death in people at least 77 years old in the United States. [3] In 1985, 73% of the people who died from falls in the United States were at least 65 years old, with one half of those falls occurring in the home. [3] A fall might not have been a direct cause of death, but an increase in the number of falls, or "clustering," often indicated a terminal decline in health. [2]

Among the home-dwelling elderly, one third of the falls resulted in no injury and one third resulted in minor injuries. Of the remaining falls, 10% required hospital care and 24% required treatment by the family physician. [4] Among the institutionalized elderly, 45% suffered at least one fall during a five-year period of time, as reported by the institutions' residents or staff members. No injury was sustained in 54% of the falls, 28% required treatment from a physician. [5] Those who had not accommodated to the loss of strength, flexibility, or balance attributable to aging frequently had a high probability of falling. [2]

Several recurrent themes are present in the literature regarding the characteristics of elderly fallers. Women had a greater frequency of falls than men, [1,2] but men had a higher mortality rate. [2] Increased age and living alone were associated with increased rates of falling. [1] Intrinsic factors, such as dizziness, weakness, and loss of balance, contributed more to falls with increasing age, whereas environmental factors became less influential. [1,2] Deteriorations in gait and balance were associated with falls and often provided the causal link between health status and falling. [2]

Although deteriorations in gait and balance are associated with falls, [2] few studies provide quantitative data or use clinically applicable methods. Hageman and Blanke compared the free-speed walking of young and elderly women using high-speed cinematography. [6] Younger subjects had longer step and stride lengths, greater ankle movement and pelvic obliquity, and faster gait than older subjects. No significant difference in stride width, which was extremely variable, was found between the two groups. [6] This study provided quantitative gait data for elderly women, but the equipment used is not practical for clinical practice.

Guimaraes and Isaacs analyzed gait variables by having the subjects walk across paper with ink pads attached to the soles of their shoes. [7] They made comparisons between five groups of subjects: hospitalized fallers and nonfallers, nonhospitalized fallers and nonfallers, and young controls. The groups differed in their respective numbers of subjects and of men and women; only one group--the hospitalized fallers--had an equal number of men and women. The authors concluded that hospitalized fallers had a short and variable step length and slow walking speed compared with the other groups. [7] Although the method used in this study is feasible clinically, the differences in gait variables between the groups may have been a reflection of the unequal distribution of men and women in the groups.

Gabell and Nayak analyzed the effect of age on the variability in gait using an electronic walkway. [8] Both young and elderly subjects demonstrated low variabilities for step length and stride time, which the authors proposed are determined by central gait-patterning mechanisms. The elderly subjects had a shorter median step length than the younger subjects, but no difference in stride time was revealed between the groups. Both young and elderly subjects demonstrated high variabilities for stride width and double-support time, which the investigators proposed are affected by balance control. No significant differences in either stride width or double-support time, however, were revealed between the groups. [8] The variability of step width has not been examined among the elderly who fall, and no research has been reported that has determined the relationship between step width and balance performance.

Balance commonly is measured by recording postural sway. Overstall et al measured postural sway with an ataxiameter and found that sway increased with age in both men and women and that women swayed more than men. [9] No difference was found between the amount of sway of elderly subjects who had not fallen and that of subjects who had fallen because of tripping. The amount of sway, however, was higher for the elderly who fell without warning and without loss of consciousness or from losing their balance than for the elderly who tripped and fell. [9]

Black et al tested the postural control of patients with vestibular deficits using a movable platform and visual surround. [10] The surface or visual surround was adjusted in direct proportion to body sway, with eyes open or with eyes closed, with platform rotations. All subjects could maintain their balance by using somatosensory or visual information under normal conditions. When the surface was adjusted to body sway, thus interfering with somatosensory input, all subjects had increased sway, both with eyes open and with eyes closed, but only subjects who had no vestibular function lost their balance. When both the support and visual input were adjusted to postural sway, thus providing conflicting somatosensory and visual information, one half of the subjects lost their balance and the postural instability was directly proportional to the extent of vestibular deficit. Similarly, the inability to attenuate muscle activity (as measured by electromyography) following conflicting surface inputs caused by platform rotations was directly proportional to the extent of the vestibular deficit. [10] The use of a movable platform, electromyography, ataxiameters, or cinematography, however, may not be feasible in most clinical settings.

Clinically applicable methods of testing balance include the one-legged stance test (OLST) [11,12] and the sharpened Romberg test (SR). [12] Bohannon et al [11] and Briggs et al [13] showed that the amount of time a subject could balance on one leg decreased with age, and they found no difference in standing times on the right versus the left leg. Briggs et al found that performance on the SR also decreased with increasing age, but the decrease was less marked than on the OLST. [13] Subjects could balance longer on one leg [11,13] and on the SR [13] with their eyes open than with their eyes closed. No order effect on balance performance was found, and no difference in performance was shown between shoes-on and shoes-off tests or between fallers and nonfallers. [13]

Briggs et al used the best trials on the balance tests to test their hypotheses, but they did not examine the separate trial measurements. [13] The first trial measurements may have a relationship to falls in the elderly. When an older person is in an unfamiliar situation where a fall is likely, he or she must respond immediately to prevent a fall. The first trial of a balance test similarly is an unfamiliar situation, and performance on the first trial may differ for fallers and nonfallers.

The purposes of this research were 1) to determine differences in age, balance performance, and step width between noninstitutionalized, elderly women with and without a history of falls and 2) to examine the relationship between balance performance and step width. The following research hypotheses were posed: 1) Fallers would be older than nonfallers, 2) the best time on each balance test for fallers would be less than for nonfallers, 3) the time on the first trial of each balance test for fallers would be less than for nonfallers, 4) the time on the best trial of each balance test would be greater than for the first trial for each group, 5) the mean step width would be different for fallers and nonfallers, 6) the step-width variability would be different for fallers and nonfallers, 7) a negative relationship would exist between step-width variability and performance on the balance tests, and 8) a negative relationship would exist between mean step width and performance on the balance tests.

Method

Subjects

This study was approved by the Institutional Review Board for Human Use at The University of Alabama at Birmingham (UAB). The subjects read and signed an informed consent form before participating in the study.

One hundred thirteen healthy women, aged 60 to 89 years, were recruited from the Birmingham, Ala, community (eg, churches, retirement complexes, senior citizen centers, the UAB) and from subjects who were involved in a previous study on balance conducted at the UAB. [13] The subjects had to be able to walk 90 ft * without an assistive device and to be independent in activities of daily living. Institutionalized elderly women and volunteers with a diagnosis of a primary balance disorder, suchas Parkinson's disease or multiple sclerosis, or with residual effects from a cerebrovascular accident were excluded from the study. One subject did not complete the study, another was excluded because of dizziness during the eyes-closed balance tests, and a third was excluded because the ink prints were unclear for measurement of step width. The data analysis, therefore, is based on the test results of the remaining 110 subjects.

Location

The testing was conducted at the UAB's Division of Physical Therapy and at two other facilities easily accessible to different areas of the community. Each testing facility was well lighted and had a walking area measuring at least 20 X 30 ft with a linoleum or tile floor. Both the walls and the floor were a solid color.

Procedure

The subjects were asked to wear supportive, comfortable shoes. Background information and general health questionnaires were completed by each subject. Items of interest from the questionnaires for this project were birth date, health information, and information on falls during the past year. Fallers were defined as those subjects who reported at least one fall during the past year; nonfallers were defined as subjects who reported no falls during the past year.

We conducted the Harris test before the balance and gait activities to determine the subjects' lower limb dominance. [14] Performance on the balance tests was timed to the nearest hundredth of a second with a digital stopwatch. *2 The subjects were first tested for balance wearing their shoes, and they then performed the SR, followed by the OLST. Each balance test was performed first with eyes open and then with eyes closed. The investigator (DKH) stood to the side of the subject to provide contact guarding when necessary. Subjects who requested help to assume the testing positions were allowed to use the investigator's arm to steady themselves prior to starting the timed trials. The subjects were given a one- to two-minute rest period between the SR and OLST conditions, and they were permitted to rest between trials as desired.

The subjects performed the SR in a standing heel-to-toe position with the dominant foot behind the nondominant foot. The subjects began testing with

their arms by their sides, but they were allowed to move their arms during the testing period. Timing started when the subject assumed the proper position and indicated she was ready to begin the test. Timing stopped if the subject moved her feet from the proper position, opened her eyes on the eyes-closed trials, or reached the maximum time of 60 seconds or if the investigator provided contact to prevent a fall. Three trials were performed if the maximum time was not reached in either of the first two trials. The first time measurement and the best time measurement of the recorded trials were used for analysis.

The subjects performed the OLST in the standing position with their arms by their sides. The subjects were allowed to move their arms during the testing period. Timing began when the subject raised the nondominant foot off the floor. Timing stopped if the subject repositioned the support foot, touched the floor with the suspended foot, used the suspended foot for support on the weight-bearing foot, required support by the investigator, or reached the maximum time of 30 seconds. Three trials were performed if the maximum time was not reached in either of the first two trials. As in the SR, the first time measurement and the best time measurement of the recorded trials were used for analysis.

A permanent record of each subject's gait pattern was made from moleskin triangles with a 1.5-in *3 base and 1.5-in moleskin squares attached to the sole of the subject's shoes at the midline of the toes and the heels, respectively. Red ink was applied to the triangles and black ink to the squares. [15] The subject was instructed to walk, at her typical speed, on a sheet of paper measuring 2 X 30 ft. She then walked back to the starting position on another sheet of paper of equal size.

The investigator measured step width for 10 subjects and then remeasured the same subjects to determine intrarater reliability. Intrarater reliability was determined by calculating the Pearson product-moment correlation coefficient (r = .99).

A T square was used to determine the distance, to the nearest millimeter, from the edge of the paper to the midpoint of the back edge of each heel square (Figure). Step width was represented as the difference in the measurement between two successive footprints. [15] This method was used to provide consistency in measurement and to decrease the variability attributable to the foot angle. Step width was measured between footprints one and two, two and three, three and four, and so on. The data from the middle 20 ft of each paper walkway were used to calculate a mean and standard deviation (variability) for each subject. The first and last 5 ft of each paper walkway were not measured to eliminate potential differences in step width caused by acceleration and deceleration.

Data Analysis

Descriptive statistics were calculated for age, first and best trials of each balance test, and mean and variability (standard deviation) of step width for fallers, nonfallers, and the total group. Differences between the means for fallers and nonfallers were tested for statistical significance using the Student's t test. For the balance tests, a maximum time was imposed and reached by some of the subjects, which resulted in non-normal distributions. Because the t test assumes normality, the Mann-Whitney U test, a nonparametric test that does not assume normality, was also performed for the balance tests. Differences between first and best balance test trials were tested for statistical significance using a two-factor analysis of variance (ANOVA) for repeated measures. The two factors were group (fallers, nonfallers) and trial (first, best). Again, because of the non-normality of the balance test data, a nonparametric test, the Wilcoxon matched-pairs signed-rank test, was used to compare the first and best trials. The Spearman rank-order correlation coefficient was used to determine the relationship between the balance measurements and the step-width variables. An alpha level of .05 was used as the criterion for all tests of statistical significance.

Results

The total group had a mean age of 73.6 years (s = 7.2). The mean age of the fallers (n = 26) was 75.1 years (s = 7.7), and the mean age of the nonfallers (n = 84) was 73.1 years (s = 7.0). The difference in age between the fallers and the nonfallers was not statistically significant (t = -1.22, df = 108, p = .113).

The best trials of the fallers on the SR in the eyes-open condition were significantly lower than those of the nonfallers (p [is less than].05) (Tab. 1). No statistically significant differences were found between fallers and nonfallers on any of the other balance tests or on the step-width variables (Tabs. 1-3). Because the probability values for the Student's t test and the Mann-Whitney U test yielded the same results, only the t-test results are presented (Tabs. 1-3).

For the total group, the best trial on each balance test was significantly higher (p [is less than].05) than the first trial. These differences were consistent across fallers and nonfallers, as indicated by the nonsignificant interaction terms in the two-factor ANOVA. The two-factor ANOVA and the Wilcoxon matched-pairs signed-rank test yielded the same results, so only the ANOVA results for the main effect of trial are presented in Table 4.

All balance test results, except the first trial of the OLST in the eyes-closed condition, showed a significant negative correlation (p [is less than].05) with step-width variability. Only the best trial of the SR in the eyes-closed condition showed a significant negative correlation (p [is less than].05) with mean step width (Tab. 5). Only 109 subjects are represented by this negative correlation because timing was inadvertently stopped for one subject. Similarly, one subject refused to perform the OLST, and one trial was recorded incorrectly for another subject on the OLST in the eyes-open condition, thus accounting for the totals of 108 and 109 subjects shown in Table 5.

Tables 6 and 7 present the distribution of the subjects on the two balance tests. The SR was easier to perform than the OLST in that 50% of the fallers and 67% of the nonfallers were able to reach the maximum time (60 seconds) in the eyes-open condition (Tab. 6), whereas only 23% of the fallers and 28% of the nonfallers reached the maximum time (30 seconds) in the same condition on the OLST (Tab. 7). No subjects were able to reach the maximum time on the OLST in the eyes-closed condition, whereas 18% were able to reach the maximum time in the same condition on the SR.

Discussion

The SR and the OLST are easy, quantifiable, and quick methods of testing balance in a clinical setting. Because the subjects' first trial was significantly lower than their best trial for each balance test, practicing these test positions may enhance balance performance. Because of this possible practice effect, we did not retest those subjects who had been tested previously for balance, but instead used their previous test results for the data anaylsis. Roberts and Fitzpatrick found no improvement in balance in elderly women after the women rocked 30 minutes in a rocking chair, but rocking stimulates only the vestibular component of balance. [16] The OLST and the SR require several components of balance, including the vestibular system, strength, flexibility, proprioception, and vision. Fansler et al found that elderly women increased their balance performance on the OLST after five days of practicing this test position. [17]

Perhaps a way to increase the likelihood of showing a difference between fallers and nonfallers on balance-performance and step-width variables is to better define a faller. This study did not separate fallers according to the cause of falls. More differences may be revealed if falls are grouped into categories of intrinsic or extrinsic causes. Intrinsic falls are caused by such events as dizziness or loss of balance, whereas extrinsic factors include causes such as slipping on the ice and tripping over objects. [1,2] Overstall et al measured balance by recording postural sway and found that subjects who fell because of intrinsic factors swayed more than nonfallers or those who fell because of extrinsic factors. [9]

Information on activity levels may be another method of further defining fallers. Overstall described the cycle of an elderly faller as follows: The individual falls; becomes fearful of falling again, which reduces his or her mobility; falls again; and so on. [18] Differences in balance-performance and step-width variables may exist between the person who does not fall, the person who occasionally falls but maintains his or her activity level, and the type of faller Overstall described. [18] A difference in balance performance may be found between those groups because activity causes a person to use balance skills and increases his or her confidence in safe mobility. [16,18]

The subjects in this study were asked to report their relative frequency of falls within the preceding year. This method of collecting the history of falls is problematic because it is based on long-term recall. We reognize that the person who falls once during the year may have a different profile from those who fall several times, but we chose to treat them the same because of the possibility of inaccurate frequency information. Linking the recall period to memorable dates or events (eg, holidays) may provide more accurate information. [19] Methods that do not rely on subject recall are needed to collect this information for research and clinical purposes.

The subjects' ability to reach the maximum score on the balance tests was better on the SR than on the OLST. Bohannon et al [11] and Briggs et al [13] reported similar times in balance performance in this age group. The OLST is more difficult for subjects to perform than the SR, in part because the base of support is diminished in the OLST. The difficulty of the OLST, combined with the diminished balance ability associated with aging, narrowed the subjects' timed performance to the extent that possible differences between fallers and nonfallers were not uncovered. The descriptive data provide values the clinician can expect from women who are representative of the study sample.

The majority of women is this age group obtained the maximum score on the SR in the eye-open condition, whereas fewer than 20% were able to reach the maximum score in the eyes-closed condition. Fewer than 30% of the women were able to reach the maximum score on the OLST in the eyes-open condition, and none of the subjects were able to reach the maximum score with their eyes closed. Age-adjusted values for these tests would be useful to the clinician.

Studies have shown that 28% to 68% of the falls in the elderly occur during some form of activity (eg, walking, ascending and descending stairs). [4,9] The method of testing balance in this study, however, involved static balance tests. These tests might not be sensitive enough because a significant difference in balance performance between the fallers and nonfallers was found only on the best trial of the SR in the eyes-open condition. Future research should be directed at using a dynamic balance test to compare balance performance of fallers and nonfallers because greater than 50% of the fallers reached the maximum of 60 seconds on the static SR in the eyes-open condition.

Negative relationships between performance on the balance tests and the mean and variability of step width were found. Gabell and Nayak stated that an increase in the variability of step width indicates a lack of compensation for instability and a possible predisposition to falls. [8] They suggested an increase in mean step width as a compensation for instability. [8] Thus, those subjects with increased step-width means and variability should have decreased balance performance. Our study demonstrated this negative relationship between balance performance and step-width means and variability, thus providing some evidence in support of Gabell and Nayak's statement. [8] The correlation values (Tab. 5), although significant, were low, which indicates that much more than step width accounts for balance performance in the elderly. These data support the complexity of balance or postural control described by other investigators. [10,20,21]

The method used for measuring step width in this study is a simple, reliable, and appropriate quantitative technique for clinical use. Boenig [15] compared measurements of step width obtained by the paper-and-ink method with the results obtained by Murray et al, [22] who used interrupted-light photography, and found the mean values of step width to be similar. The mean values of step width for the total group, fallers, and nonfallers in this study were similar to the results of Boenig [15] and Murray et al. [22]

Perhaps by analyzing the pattern of step width, more differences could be found between fallers and nonfallers. For example, fallers may have more negative measures of step width, where the midpoints of the heels actually cross each other, than nonfallers. Because we compared means of step widths, the effect of these negative measures was minimized.

Conclusions

The mean value of the best timed trial on the SR in the eyes-open condition was less for fallers than for nonfallers in our sample of elderly women (p [is less than].05), whereas no significant differences in the other conditions on the balance tests or in the mean and variability of step width were revealed. The mean values on the first trials were lower than the mean values on the best trials of the balance tests for the total group (p [is less than].05). Small, but statistically significant (p [is less than].05), negative relationships existed between balance-performance and step-width variables. More sensitive balance tests, a more stringent definition of fallers, and prospective studies are needed to determine characteristics of fallers, predictors for falling, and cause-effect relationships.

(*1) ft = 0.3048 m.

(*2) Model 1000, Heuer Time 3 Electronics Corp. 960 S Springfield Ave, Springfield, NJ 07081.

(*3) 1 in = 2.54 cm.

References

[1] Perry BC: Falls among the elderly: A review of the methods and conclusions of epidemiologic studies. J Am Geriatr Soc 30:367-371, 1982

[2] Nickens H: Intrinsic factors in falling among the elderly. Arch Intern Med 145:1089-1093, 1985

[3] Accident Facts. Chicago, IL, National Safety Council, 1986, pp 11, 17

[4] Gabell, A, Simons MA, Nayak USL: Falls in the healthy elderly: Predisposing causes. Ergonomics 29:965-975, 1985

[5] Gryfe DI, Amiess A, Ashley JJ: A longitudinal study of falls in an elderly population: Incidence and morbidity. Age Ageing 6:201-210, 1977

[6] Hageman PA, Blanke DJ: Compariso of gait of young women and elderly women. Phys Ther 66:1382-1387, 1986

[7] Guimaraes RM, Isaacs B: Characteristics of the gait in old people who fall. Int Rehabil Med 2:177-180, 1980

[8] Gabell A, Nayak USL: The effect of age on variability in gait. J Gerontol 39:662-666, 1984

[9] Overstall PW, Exton-Smith AN, Imms FJ, et al: Falls in the elderly related to postural imbalance. Br Med J 1:261-264, 1977

[10] Black FO, Wall C, Nashner LM: Effects of visual and support surface orientation references upon postural control in vestibular deficient subjects. Acta Otolaryngol (Stockh) 95:199-210, 1983

[11] Bohannon RW, Larkin PA, Cook AC, et al: Decrease in timed balance test scores with aging. Phys Ther 64:1067-1070, 1984

[12] Graybiel A, Fregly AR: A new quantitative ataxia test battery. Acta Otolaryngol (Stockh) 61:292-312, 1966

[13] Briggs RC, Gossman MR, Birch R, et al: Balance performance among noninstitutionalized elderly women. Phys Ther, to be published

[14] Harris AJ: Harris Tests of Lateral Dominance. New York, NY, The Psychological Corp, 1958, p 10

[15] Boenig DD: Evaluation of a clinical method of gait analysis. Phys Ther 57:795-798, 1977

[16] Roberts BL, Fitzpatrick JJ: Improving balance: Therapy of movement. Journal of Gerontological Nursing 9:151-156, 1983

[17] Fansler CL, Poff CL, Shepard KF: Effects of mental practice on balance in elderly women. Phys Ther 65:1332-1338, 1985

[18] Overstall PW: Prevention of falls in the elderly. J. Am Geriatr Soc 28:481-484, 1980

[19] Cummings SR, Neirtt MC, Kidd S: Forgetting falls: The limited accuracy of recall of falls in the elderly. J Am Geriatr Soc 36:613-616, 1988

[20] Nashner LM, McCollum G: Organization of postural human movements: A formal basis and experiments synthesis. Behavioral and Brain Sciences 8:135-172, 1985

[21] Woollacott MH, Shumway-Cook A, Nashner LM: Aging and posture control: Changes in sensory organization and muscle coordination. Int J Aging Hum Dev 23:97-112, 1986

[22] Murray MP, Kory RC, Sepic SB: Walking patterns of normal women. Arch Phys Med Rehabil 51:637-650, 1970

D Heitmann, MS, PT, is Physical Therapist, Department of Physical Therapy, Penrose Community Hospital, 3205 N Academy Blvd, Colorado Springs, CO 80917. She was a graduate student, Division of Physical Therapy, The University of Alabama at Birmingham, Birmingham, AL 35294, when this study was completed in partial fulfillment of the requirements for her Master of Science degree.

M Gossman, PhD, PT, is Associate Professor, Division of Physical Therapy, The University of Alabama at Birmingham.

S Shaddeau, MMSc, PT, is Associate Professor, Division of Physical Therapy, The University of Alabama at Birmingham.

J Jackson, PhD, is Assistant Professor, Office of Educational Development, School of Medicine, The University of Alabama at Birmingham.

Address all correspondence to Dr Gossman, Division of Physical Therapy, School of Health Related Professions, The University of Alabama at Birmingham, RM B41, Birmingham, AL 35294 (USA).