Influence of sensory inputs on standing balance in community-dwelling elders with a recent history of falling.

Vision and/or ankle somatosensory inputs often do not correlate with the position of the center of gravity. In this case, visual or somatosensory information ?nay be in conflict with other sensory systems that convey a true sense of body orientation. The purpose of this study was to determine how conflicting visual and ankle somatosensory inputs influenced standing balance in elders with a history of falls. Forty-seven community-dwelling elders (8 male, 39 female), between 65 and 96 years of age (X=80.5, SD=9.0), participated in this project. Subjects with two or more falls in the 6 months prior to study were assigned to a fall group (n= 16), whereas those with no history of falling during the same time interval were assigned to a no-fall group (n =31). In order to remove any bias in the testing procedure, the tester was not aware of group assignments. Subjects were evaluated using a sensory organization test (SOT) for standing balance and a "Get Up and Go" test (GUGT) for general mobility. Analysts of covariance was used to evaluate the SOT scores (by group, viMon, and surface condition) and the GUGT scores. Body sway is known to increase with the normal aging process, and for this reason, age was selected as a covariate. The association between the SOT total score and the GUGT score was evaluated using Spearman rank-order correlation coefficients. The results showed a significant interaction between group and surface conditions, which indicated a decreased stance duration for failers on a compliant surface compared with the stance duration for nonfailers (adjusted mean failer stance duration =53 seconds, SD=42; mean nonfailer stance duration =67 seconds, SD=32). Subjects in the fail group also had significantly higher GUGT scores (which indicated poor mobility function) than did subjects in the nonfall group (adjusted mean failer GUGT score=2.65, SD=1.48; mean nonfailer GUGT score=1.47, SD=O. 77). The Spearman correlation between total SOT scores and the GUGT scores was greater for failers (r= -.67) than for nonfailers (r=-. 44). Orientation input from the ankle appears to have greater importance for preventing fails compared with a visual reference. The SOT and GUGT may be useful in the field to establish criteria for screening elders in a fail-prevention program. [Anacker SL, Di Fabio RP. Influence of sensory inputs on standing balance in community-dwelling elders with a recent history of falling. Phys Ther. 1992;72:575-584.]

Various functional deficits, medications, and environmental hazards can affect the risk of falling in older persons.[1-5] It is known that the occurrence of falls in community-based elders (ie, elders living at home versus in an institution) increases linearly with the number of risk factors (eg, sedative use, cognitive impairment, and disability involving the lower extremities).3 Previous research2 has shown that subjects who receive a post-fall assessment (identifying remediable problems such as weakness, orthostatic hypotension, drug side effects, and gait disorders) are less likely than control subjects to be hospitalized. In addition, elders who receive a comprehensive risk-factor evaluation following a fall will have a tendency for fewer fails in the future compared with those who receive "routine" care.2 Therefore, it is important from a therapeutic standpoint to identify factors that contribute to the risk of falls in elders and to develop therapeutic strategies designed to prevent falling in this population.

Sensory information has an important influence on balance activity in elders,6 and the integration of visual, vestibular, and somatosensory information is necessary to generate appropriate balance responses.7 The inability of an aging neural system to integrate multiple sensory elements may be one factor that contributes to the risk of falling in older persons.8 Distinct sources of sensory input may have a disproportionate influence on balance outcomes under certain conditions.9 In relatively young, nondisabled subjects, postural instability is magnified with ischemic blockade of propdoceptive inputs (from cutaneous, joint, and pressure receptors) in the foot and ankle during balance tasks that utilize sensory feedback to maintain stance on a moving support surface.[10,11] There is a widespread increase in the activation threshold for cutaneous sensation and joint proprioception with advancing age,[12] and it would be reasonable to assume that deficits in ankle proprioception would contribute to the incidence of falls in elders. The weighted contribution of sensory inputs to prevent falls, however, is not known, because visual, vestibular, and somatosensory deficits have not been systematically evaluated with respect to functional balance outcomes in eider subject samples.

Community-dwelling eiders with a previous history of falling are reported to have an increase in postural sway during attempted quiet standing compared with elders who have no recent history of falling.[5,13] Studenski et al[14] reported that elders who experienced a recent fall were more unstable than similar-aged control subjects during a postural stress test, single-leg stance, and turning activities. Heitmann et al[15] noted that elderly "failers" performing the sharpened Romberg test with eyes open had significantly lower stance durations than did eider "nonfailers." In contrast, Briggs et al[16] found that singleleg stance duration and the stance duration observed with a sharpened Romberg test were not different for elders with a prior history of failing. Research on balance function conducted in nursing homes or extended care facilities often involves the frail elderly,[2,17,18]and the results of these studies could not be extrapolated to elders living in the community.

A number of clinical assessments have been used to measure balance function in older individuals.[14-19] In order to assess the interaction of peripheral and central nervous system influences on standing balance, studies have incorporated trials with vision and no vision[20,21] single-leg Stance,[22] Or weight shifts to the right or left.[23] These maneuvers were done to manipulate sensory inputs or to mechanically stress the balance control system. Shumway-Cook and Horak[24] developed a tool that systematically altered visual and ankle somatosensory inputs so that sensory influences on standing balance could be evaluated. Lord et a1[25] adapted this approach and demonstrated that visual acuity and visual-contrast sensitivity were not associated with body sway when eiders (residing in a long-term care facility) maintained stance on a firm surface. When stance was attempted on a foam base (designed to provide ankle orientation inputs to the central nervous system that conflicted with the true position of the center of mass), however, the correlations between body sway and each visual parameter were statistically significant compared with correlations for stance on a firm surface.25 These results indicated that visual influences on postural stability became more important during stance on a compliant surface because the surface dynamics reduced the internal association between ankle proprioception and the position of the center of mass. The lack of correlation between ankle afference and the orientation of the body will theoretically require the use of sensory cues from other modalities (eg, visual and/or vestibular) to control body sway.

In populations of community-based elders, the inability to maintain balance when presented with misleading sensory information has not been clearly related to a prior history of failing. Most studies using stance time as a measure of balance function in samples of community-based eiders with a history of falls have not examined the influence of conflicting sensory inputs on falling.[5,9, 13-16] The purpose of this study was to determine how visual and ankle somatosensory inputs influenced standing balance in community-dwelling elders who recently fell. Given the reliance on sensory feedback for conscious control of stance,[6,10,11,25] the age-related changes in sensory perception, and the increase in spontaneous body sway for elders with a history of falling described above, it was hypothesized that conflicting visual and/or ankle somatosensory inputs would decrease stance duration (ie, the amount of time balance could be maintained without assistance) in eiders who had a prior history of falling compared with elders with no recent history of failing. In order to evaluate the relationship between quasi-static standing balance (ie, standing balance that appears overtly "static," but that actually involves subtle dynamic corrections in the position of the center of mass) and dynamic functional performance (ie, movement with a functional purpose that is not static), the association between sensory organization test (SOT) scores and scores obtained with a test of general mobility was assessed. The secondary hypothesis was that a substantial correlation (>.50) between SOT and general mobility scores would be found for both eider failers and nonfailers.

Method

Subjects

Subjects were recruited through community and church organizations and through local home health agencies in the Minneapolis-St Paul metropolitan area. An effort was made to solicit subjects from all racial and socioeconomic classes. Fifty-four potential subjects were given a medical history questionnaire in order to determine whether they could participate in the study. Forty-seven communitydwelling eiders (8 male, 39 female), between 65 and 96 years of age (X=80.5, SD=9.0), met the criteria for participation. These subjects were all relatively healthy elders, because the screening protocol excluded subjects with recent fractures, surgeries, vestibular deficits, acute heart ailments, multiple sclerosis, Parkinson's syndrome, stroke, or neurological problems. Subjects were also excluded from the study if (1) they were living in an institution, (2) they used assistive devices for standing, (3) there was a recent history (ie, within 6 months prior to testing) of lowerextremity surgery or use of lowerextremity prosthetic or orthotic devices, or (4) there was only one fail reported within 6 months prior to testing. Subjects who reported only one fall were excluded from the study because we wished to establish conservative criteria for assignment to the fall group. It was felt that subjects assigned to the fall group using the "two-fails" criterion would represent a high-risk sample. Informed consent was obtained from all subjects prior to data collection.

The medical questionnaire was used to determine the group assignment for each subject. Specifically, subjects who had fallen two or more times within 6 months preceding the study were assigned to the fall group (n= 16), whereas those who had no falls during that same interval of time were assigned to the no-fall group (n=31). The relevant characteristics of each group are summarized in Table 1. A "fall" was defined as a person's trunk, knee, or hand unintentionally coming to rest on the ground or some lower level (below the waist). This meant that losing balance and coming to rest on a stair was considered a fall, but catching the body with hands on the kitchen counter was not a fall. A fall attributable to an overwhelming hazard was not counted. An "overwhelming hazard" was defined as "a hazard that would result in a fall by most young healthy persons,"26 such as slipping on ice.

Procedure

In order to prevent bias in the testing procedure, a research assistant reviewed the medical questionnaires so that subjects could be screened and then assigned to the appropriate group. The researcher testing the subjects was not aware of group assignments. Subjects were evaluated using an SOT for standing balance24,27 and a "Get Up and Go" test (GUGT) for general mobility.[19] The order of test conditions was identical for all subjects. The SOT consisted of quiet standing on a firm support surface during three visual conditions (eyes open, eyes closed, and visual stabilization using a helmet dome). The dome was made from a white 50.8-cm-diameter (20-in-diameter) oriental-style paper lantern. A hole was cut in the back to enable the subject to put the dome over the head. The dome restricted vision from the top, bottom, and sides. Several vertical black stripes were fixed to the inside surface of the dome and provided a visual fixation point 20.3 cm (8 in) from the eyes. These lines were 5.1 cm (2 in) apart at the top rim and 15.2 cm (6 in) apart at the center (see ShumwayCook and Horak[24). When wearing the visual dome, the subjects were instructed to keep their eyes open and to look forward. The dome moved in synchrony with head movement and therefore "neutralized" vision (or reduced the meaning of vision) as a reference for body sway. The effect of the dome on visual inputs has been referred to as "visual stabilization," because the visual field moves in proportion to body sway.28

Each visual condition was repeated while the subject stood on a compliant foam surface in order to reduce the meaning of orientation information from the ankle. During stance on a compliant surface, the change in ankle position may no longer correlate with movement of the center of gravity. Under these circumstances, the subject may rely less on information from ankle proprioceptors and more on information from the sensory receptors in the head and neck to control body sway.

The subject stood with arms crossed over the chest and with hands on the elbows. Three trials (30 seconds/trial) were administered for each of the six stance conditions. The SOT scores were summed by stance condition (n=3 trials/condition/subject) and in aggregate (n=18 trials/subject) to calculate a condition score as well as a total score. If a subject was able to stand for 30 seconds during a given trial, no additional testing for that condition was needed and a score of 30 was assigned to each remaining trial.27 A maximum score of 90 seconds was possible for each condition, and the maximum total score was 540 seconds. A trial was terminated if any of the following occurred:

1. The subject's arms moved off the chest or the hands moved off the elbows.

2. One or both knees were flexed.

3. Heels or toes were lifted from the floor.

4. A step was initiated to compensate for instability.

Sway about the ankles while maintaining balance was not considered a reason for terminating the trial. The GUGT required subjects to stand from the sitting position without using their arms, walk several steps, turn, and then return to sitting without using their arms for support. The chairs were standard kitchen or dining-room chairs without arms. The subjects sat with their back against the chair prior to initiating stance. A fivepoint scale was used to provide one score for the entire sequence of movement Crab. 2). This scale was used as a broad measure of general mobility and was not intended to precisely define impairments underlying balance dysfunction.

Statistical Design and Data Analysis

The SOT scores were evaluated using a multifactorial analysis of covariance (ANCOVA) with three factors: group (failers versus nonfallers), surface condition (firm versus compliant), and vision (eyes open, eyes closed, and visual stabilization). Age was selected as a covanate because there is a linear relationship between the amount of body sway and age.20 A one-way ANCOVA was used to determine whether the GUGT scores for failers and nonfailers were different. Tukey's post hoc studentized range test was used to assess multiple pair-wise comparisons at an experimentwise alpha level of .05. In addition, Spearman rank-order correlation coefficients were calculated to evaluate the degree of association between the total SOT scores and the GUGT scores. The Spearman coefficient was selected because the rank order of the scores from the SOT and the GUGT provides a descriptive measure of the degree that these two different clinical tests measure similar functions.

Reliability

Previous work has demonstrated the reliability of measurements obtained with the SOT when applied to subjects with cerebrovascular disease,29 but to our knowledge reliability has not been specifically evaluated using eider subject samples. Test-retest reliability of the SOT and GUGT scores was examined in our investigation using the first 10 subjects (1 male, 9 female; 66-% years of age) who agreed to participate in the study. The retest was completed by a single rater within 7 days of the initial test session. Correlations of the SOT total scores (r=.75) and the GUGT scores (r--.96) across test-retest sessions were statistically significant (P<.05). There was 95% agreement between the two test sessions for SOT total scores and 96% agreement for the GUGT scores. A kappa coefficient was calculated to further characterize the agreement in GUGT scores (k= .52). In situations in which agreement between experimental conditions is high (ie, there is little variability in the categorizations), the value of kappa may not reflect the consistency between ratings.[30] The small magnitude of the kappa coefficient in this reliability substudy was probably due to minimal variability in the GUGT scores.

Results

A summary of the multifactorial ANCOVA for SOT scores is shown in Table 3. Age was found to be a significant predictor of the total SOT score. When SOT scores were adjusted for age, there was no statistically significant difference between failers and nonfailers (the main effect for group in Tab. 3). A significant interaction between group and surface conditions (Fig. 1), however, indicated a reduction in stance duration on a compliant surface for failers compared with nonfailers (adjusted mean failer stance duration=53 seconds, SD--42; adjusted mean nonfailer stance duration=67 seconds, SD--32). The largest reductions in stance duration also were observed during stance conditions using a compliant surface with eyes closed (Fig. 2),

There was a significant difference in the GUGT scores between failers and nonfailers (F1,44--' 11.37, Pc .01; adjusted mean failer GUGT score-- 2.65, SD--1.48; adjusted mean nonfailer GUGT score= 1.47, SD--0,77). The Spearman correlation between total SOT scores and the GUGT scores was greater for failers (r=- .67) than for nonfailers (r--- .44). The negative association indicated that higher total SOT scores were correlated with lower GUGT scores, Lower GUGT scores reflected better standing and dynamic balance performance.

Discussion

The results of this study showed that somatosensory input from the ankles provides a critical source of information for controlling the degree of body sway in elders who fall. Elders in the fall group had a lower stance duration than did nonfailers when attempting to stand quietly on a compliant foam surface (Fig. 1). In contrast to failers, the nonfailers were able to compensate for "conflicting" ankle orientation information. The mechanism by which a dependence on ankle proprioception is created in elderly failers is not dear, Possibly the loss of lower-limb strength in elderly failerss,14 and the increase in activation threshold for cutaneous and joint proprioception12 predispose the ankle to chronic excessive displacements during stance on a compliant surface. Therefore, a reduction in the ability to generate torques about the ankle joint and the decline in proprioception (ie, the input from cutaneous, joint, and pressure receptors) may alter the motor strategy used for standing balance by substituting inefficient patterns of movement that do not depend as heavily on proprioceptive feedback. This view is supported by the findings that ischemic blockade of ankle joint and cutaneous afferents in nondisabled subjects altered the balance strategy from use of stabilizing ankle torques to use of excessive hip and trunk movements to maintain stance on a moving platform?

In addition, a "fear of falling" may have been a factor in the reduction of stance duration on a compliant surface or in the hesitation and slowness of movement evidenced by the GUGT scores for the fall group. Maki et al[31] reported that subjects who expressed a fear of falling demonstrated a reduction in stance duration during attempted quiet stance with and without visual cues. These authors noted that they could not determine whether the fear of falling and poor balance performance were due to a true deterioration of postural control or whether the association was merely an artifact generated by apprehension during the test.

Decreased stance duration on a compliant surface was observed for all visual conditions (Fig. 2). Previous work reporting SOT scores for subjects with stroke demonstrated a reduction of stance duration on a compliant surface with eyes closed and with visual stabilization.32 In the current study, stance duration with conflicting visual information could not be used to differentiate between failers and nonfailers, because both groups showed a significant reduction in stance duration for each visual condition while standing on the foam base (Tab. 3, Fig. 2). Our results indicated that the accuracy of visual cues played a secondary role in preventing fails, because there was no statistically significant interaction between groups (failers versus nonfailers) and visual conditions (Tab. 3).

Although proprioception is likely to be the primary modality used to control balance in the elderly, the integration of proprioception with visual and vestibular inputs provides a mechanism for adaptation to changing environmental conditions in which proprioception alone would not provide enough information. For example, Woollacott et al,8 using a moveable platform, found that balance control deteriorated in elderly subjects when proprioceptive input was incongruent with the direction of body sway. Several authors6,8 have suggested that the mechanisms responsible for the interaction of visual, vestibular, and proprioceptive information deteriorate with increasing age. The degeneration of this hypothetical integrative mechanism cannot fully explain our results, because there was no difference between failers and nonfailers with respect to stance duration across visual conditions. Furthermore, our analysis controlled for balance deficits observed through the normal aging process by using age as a covariate. Therefore, if an intact vestibular mechanism can be assumed, the reduction of somatosensory inputs or the inability to correctly interpret somatosensory information probably contributed to the recent falls reported for the eiders in the experimental group.

The correlations between SOT total scores and the GUGT scores indicated that there may be some "transfer" of stance stability that accounts for the variance in performance of dynamic balance activities. The ability to maintain stance in a quasi-static balance task, however, has not been widely established as a predictor of the ability to perform dynamic functional tasks. For example, Winstein et a1[33] reported that improvements in hemiplegic standing balance do not necessarily lead to quantifiable changes in gait. Future research should focus on a prospective study of the value of the SOT (a quasi-static stance test) for predicting falls in the eider population.

Summary

We concluded that orientation inputs from the ankle are important for preventing falls in elders. Vision appears to have a secondary, rather than primary, influence on the occurrence of falls in older persons. Use of the SOT and the GUGT in the field to identify somatosensory dependence for balance and general mobility status would seem to be appropriate. Further research is needed to identify the critical factors needed to screen "at-risk" elders in a fall-prevention program.

Acknowledgments

We would like to express our appreciation to Beruadine M Feldman, PhD, and Louis R Amundsen, PhD, PT, at the University of Minnesota for their consultation regarding the implementation of this study.

References

1 Hadley E, Radebaugh TS, Suzman R. Falls in the elderly: biologic and behavioral aspects. Clin Gertair Med. 1985; 1:497-500.

2 Rubenstein LZ, Robbins AS, Josephson KR, et al. The value of assessing fails in elderly population. Ann Intern Med. 1990;113:308--316.

3 Tinetti ME, Speechley M, Ginter S. Risk factors for falls among elderly persons living in the community. N EngIJ Med. 1988;319: 1701-1707.

4 Campbell AJ, Reinken J, Allan BC, Martinez GS. Fails in old age: a study of frequency and related clinical factors. Age Ageing. 1988;10: 264-270.

5 Gehlsen GM, Whaley MH. Falls in the elderly, part II: balance, strength, and flexibility. Arch Phys Med Rebabil. 1990;71:739-741.

6 Stelmach GE, Teasdale N, Di Fabio RP, Phillips J. Age related decline in postural control mechanisms. Int J Aging Hum Dev. 1989;29: 205-223.

7 Dichgarks J, Diener HC. The contribution of vestibulo-spinal mechanisms to the maintenance of human upright posture. Acta Otolaryngol (Stockh). 1989;107:338-345.

8 Woollacott MH, Shumway-Cook & Nashner LM. Postural reflexes and aging. In: Mortimer J, Pirozzolo FJ, Maietta JG, eds. Aging Motor Syston. New York, NY: Praeger Publishers; 1983: 98-119.

9 Lichtenstein MJ, Shields SL, Shiavi RG, Burger MC. Clinical determinants of biomechanics platform measures of balance in aged women. J Am Geriatr Soc. 1988;36:996-1002.

10 Horak FB, Nashner LM, Diener HC. Postural strategies associated with somatosensory and vestibular loss. Exp Brain Res. 1990;82: 167-177.

11 Dienet HC, Dichgans J, Guschlbauer B, Mau H. The significance of proprioception on postural stabilization as assessed by ischemia. Brain Res. 1984;296:103-109.

12 Kokmen E, Bossemeyer RW, Williams WT. Quantitative evaluation of joint motion sensation in an aging population. J Gerontol. 1978; 33:62-67.

13 Overstail PW, Exton-Smith AN, Imms FJ, Johnson AL. Fails in the elderly related to postural imbalance. BMJ. 1977;1:261-264.

14 Studenski S, Duncan PW, Chandler J. Posrural responses and effector factors in persons with unexplained fails: results and methodolgic issues. JAm Gertair Soc. 1991;39:229-234.

15 Heitmann DK, Gossman MR, Shaddeau SA, Jackson JR. Balance performance and step width in noninstitutionalized, elderly, female failers and nonfailers. Phys Ther. 1989;69: 923-931.

16 Briggs RC, Gossman MR, Birch R, et al. Balance performance among noninstitutionaiized elderly women. Phys Ther. 1989;69:748-756.

17 Fernie GR, Gryfe CI, Holliday PJ, Llewellyn A. The relationship of postural sway in standing to the incidence of falls in geriatric subjects. Age Ageing. 1982;11:11-16.

18 Wolfson L, Whipple R, American P, Tobin JN. Gait assessment in the elderly: a gait abnormality rating scale and its relation to falls. J Gerontol. 1990;45:MI2-M19.

19 Mathins S, Nayak USL, Isaacs B. Balance in elderly patients: the "Get-up and Go" test. Arch Phys Med Rehabil. 1986;67:387-389.

20 Sheldon JH. The effect of age on the control of sway. Gerontologia Clinica. 1963;5: 129-138.

21 Brocklehurst JC, Robertson D, JamesGroom P. Clinical correlates of sway in old age: sensory modalities. Age Ageing. 1982;11: 1-10.

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

23 Murray MP, Seireg AA, Sepic SB. Normal postural stability and steadiness: quantitive assessment. J Bone Joint Surg [Am]. 1975;57: 510-516.

24 Shumway-Cook A, Horak FB. Assessing the influence of sensory interaction on balance= suggestion from the field. Phys Ther. 1986;66: 1548-1550.

25 Lord SR, Clark RD, Webster IW. Visual acuity and contrast sensitivity in relation to falls in an elderly population. Age Ageing. 1991;20: 175-181.

26 Tinetti ME. Performance-oriented assessment of mobility problems in elderly patients. JAm Geriatr Soc. 1986;34:119-126.

27 Horak FB. Clinical measurement of postural control in adults. Phys Ther. 1987;67: 1881-1885.

28 Vidal PP, Berthoz A, Millanvoye M. Difference between eye closure and visual stabilization in the control of posture in man. Aviat Space Environ Med. 1982;53:166-170.

29 Di Fabio RP, Badke MB. Relationship of sensory organization to balance function in patients with hemiplegia. Phys Ther. 1990;70: 542-548.

30 Haley SM, Osberg JS. Kappa coefficient calculation using multiple ratings per subject: a special communication. Phys Ther. 1989;69: 970-974.

31 Maki BE, Holliday PJ, Topper AK. Fear of falling and postural performance in the elderly. J Gerontol. 1991;46:MI23-M131.

32 Di Fabio RP, Badke MB. Stance duration under sensory conflict conditions in patients with hemiplegia. Arch Phys Med Rehabil. 1991; 72:292-295.

33 Winstein CJ, Gardner ER, McNeal DR, et al. Standing balance training: effect on balance and locomotion in hemiparetic adults. Arch Phys Med Rehabil. 1989;70:755-762.

Commentary

It is attractive to try to find a simple clinical measure of balance or mobility that can be used to predict which elderly patients are likely to fall, because fails are the leading cause of injury-related deaths in the elderly) The article by Anacker and Di Fabio attempts to determine whether two simple, reliable clinical measures of balance and mobility can differentiate elderly persons who are prone to falls from those who are not. Unfortunately, data obtained with even relatively simple and reliable clinical measures can be difficult to analyze and interpret, so therapists must be cautious in applying results from studies such as this to clinical settings.

Interpretation of the Sensory Organization Test

There seems to be a common confusion in interpretation of data obtained with the sensory organization test (SOT), or the Clinical Test of Sensory Interaction on Balance, as it is sometimes called.2 Patients who have significant loss of somatosensory information are expected to be more prone to falls than are individuals without somatosensory deficits.3 Is it reasonable to expect patients with frequent fails attributable to somatosensory loss to have reduced stance time on compliant foam? No, probably not. If a patient cannot use somatosensory information for orientation, further attempts to disrupt somatosensory inputs by placing the patient on compliant foam should have no further impact. This expectation is confirmed by the observation of normal sway on a sway-referenced platform by patients with profound somatosensory loss secondary to multiple sclerosis.4 It is therefore not reasonable to conclude that patients who perform poorly on foam have somatosensory losses.

It is more likely that the elderly subjects in the study who had trouble with compliant conditions had vestibular loss rather than somatosensory loss.5 Patients who sway excessively or fail on compliant surfaces are more likely to be excessively dependent on reliable somatosensory information from the surface because other sensory information, such as information from vestibular and visual systems, is no longer adequate. Moreover, there is an exponential decline in vestibular function in the elderly population that is just as pervasive as somatosensory lOSS.6

Statistical Analysis and Clinical Significance

A major difficulty in analyzing timed measures of balance and mobility is that the data are usually not normally distributed, which is a basic assumption necessary for using common parametric statistics such as the analysis of covariance. Data from timed tests of standing balance show skewed distributions because the range of possible scores is limited both at the upper (90-second maximum) and lower ends. When basic assumptions of statistical tests are violated, statistical significance may not reflect real differences. Regardless of the type of statistic used, however, therapists should not assume that statistical significance implies clinical significance.

For example, Figure 1 in the article reveals the large variability in stance duration on a compliant surface by both failers (53-+42 seconds) and nonfailers (67-+32 seconds). Obviously, some failers were able to stand for a full 90 seconds on the compliant surface, and some nonfailers were limited to a much shorter stance time. Thus, this test, by itself, has little predictive value for identifying a failer in the clinic. The likelihood that a given failer and a given nonfailer would have similar scores on both the SOT and the "Get Up and Go" test (GUGT) is actually relatively large, which leads to problems of clinical sensitivity and specificity. A sensitive clinical test would be positive for every failer, and a specific clinical test would be negative for every nonfailer.

The poor predictive value of a single measure of balance or mobility is certainly not surprising, given the many other variables that may contribute to falls. For example, falls may be related to factors not directly differentiated by these tests, such as the propensity of some patients to put themselves in precarious situations, poor strength, fear of falling, and many other musculoskeletal or sensomotor deficiencies. A history of falls may actually correlate poorly with balance ability, because those patients who are most vulnerable to falling may severely limit their mobility and thus prevent falls. In addition, self-reports of falls have been shown to be a poor reflection of actual falls.

In condusion, therapists need clinical tests that help them differentiate whether a patient requires treatment, help focus the treatment on the problem, and help quantify the effectiveness of the treatment. Although the SOT and the GUGT may be very useful clinical measures of specific aspects of balance and mobility that can aid in the design of treatment programs and monitoring of progress, they have not yet been shown to be sensitive and specific predictors for falls in the elderly.

Fay B Horak, PhD, PT Associate Scientist Robert S Dow Neurological Sciences Institute Good Samaritan Hospital and

Medical Center 1120 NW 20th Ave Portland, OR 97209-1595

References

1 Baker S, O'Neill B, Karpf R. The Injury Fact Book. Lexington, Mass: Lexington Books; 1984.

2 Crowe TK, Dietz. JC, Richardson PK, Atwater SW. Interrater reliability of the Clinical Test of Sensory Interaction on Balance. Physical and Occupational Therapy in Pediatrics. 1990;10: 1-27.

3 Stelmach CE, Worringham CJ. Sensorimotor deficits related to postural stability: implications for falling in the elderly. In: Radebaugh TS, Hadley E, Suzman R, eds. Falls in the Elderly: Biologic and Behavioral Aspects. Philadelphia, Pa: WB Saunders Co; 1985;1(3): 67%694.

4 Pratt C, Horak FB, Hem don R. Differential effects of somatosensory and motor system deficits on postural dyscontrol in multiple sclerosis. In: Woollacott M, Horak FB, ed& Posture and Gait: Control Mechanisms. Eugene, Ore: University of Oregon Press; 1992:118-122.

5 Horak FB, Mirka A, Shupert CL. The role of peripheral vestibular disorders in postural dyscontrol in the elderly. In: Woollacott M, Shumway-Cook A, eds. The Development of Posture and Gait Across the Lifespan. Columbia, SC: University of South Carolina Press; 1989:253-279.

6 Bergstrom B. Morphology of the vestibular nerve, II: the number of myelinated vestibular nerve fibers in man at various ages. Acta Otolaryngol (Stockh). 1973;76:173-179.

Author Response

Dr Horak is credited with the development of a "simple" clinical test of balance used in our study.[1] We believe, however, that the interpretation of the sensory organization test (SOT) recommended in the commentary by Dr Horak is highly speculative. For example, Dr Horak suggests that the SOT results (see Fig. 1 in our article) indicated that eiders in our study had a vestibular loss rather than a deficit in the integration of somatosensory information. We disagree with this interpretation for several reasons:

1. No direct measure of vestibular function can be made from the SOT. One can only observe balance during various stance conditions designed to manipulate visual and/or somatosensory information. When supporting otologic tests are not available or feasible, the role of the vestibular system must be inferred. Therefore, the conclusion that "elderly subjects [likely] ... had vestibular loss" on the basis of SOT scores overstates the scope of this clinical test.

2. The subjects in both groups (failers and nonfailers) were equivalent with respect to all relevant demographic and health characteristics (see Tab. 1 in the article). The only factor used to categorize eligible subjects into groups

was a previous history of falling within 6 months of the study. We found no significant interaction between visual conditions, support-surface conditions, and groups Crab. 3 in the artide). This means that stance duration for test conditions that supposedly target vestibular mechanisms (ie, visual stabilization coupled with the foam base) were not different for failers compared with nonfailers.

3. The argument that all subjects had vestibular loss is not plausible because subjects in the nonfall group scored significantly better than failers on a separate test of mobility function (the "Get Up and Go" test). In other words, if all subjects had vestibular deficits, why were the mobility scores of eider failers inferior to the mobility scores of the control subjects?

Dr Horak maintains that patients who perform poorly during stance on a foam base do not have somatosensory loss. A distinction must be made between somatosensory loss and the inability to utilize existing proprioceptive inputs to maintain balance. We did not conclude that eider failers had somatosensory loss. In fact, subjects who had neurological deficits resulting in somatosensory impairment were excluded from the study (refer to the "Method" section in the article). We did conclude, that for elders with a history of falling, orientation input from the ankles played a greater role in stance stability compared with visual reference. When ankle somatosensory inputs did not correspond perfectly with the position of the center of mass (ie, stance on a foam base), eider failers lost their balance more quickly than did nonfailers (Fig. 1 in the article).

It was difficult to evaluate Dr Horak's concern about the statistical analysis because no alternatives were suggested. It should be noted that the validity of analysis of covariance (ANCOVA) is not dependent on a normal distribution.2 Hays states that "relaxation of assumptions is often possible in the two- or multiway analysis... [and] the requirement of a normal distribution of errors seems to be rather unimportant."2 In addition, the analysis of age as covariate served as an important control. The ANCOVA adjusted SOT scores in the relatively younger control group to reflect the expected deterioration of balance with advancing age. The ANCOVA allowed us to evaluate real differences in balance scores between failers and nonfailers.

Dr Horak suggests that our findings lack "clinical significance," irrespective of the statistical significance reported in the artide Crab. 3). The argument presented by Dr Horak is focused on the potential overlap of individual subject scores rather than on group data. The question our study addressed was: Do failers and nonfailers come from the same population with regard to balance ability? We concluded that failers and nonfailers perform differently on selected tests of balance and therefore do not come from the same population. Our conclusion was based on statistical probability. The difference between failers and nonfailers was not likely due to chance. In our view, these results are clinically meaningful.

We disagree with the assertion that "a sensitive clinical test would be positive for every failer." Most clinical tests are not absolute, and there is a known and acceptable level of error. We did not intend to predict which eiders were likely to fall or to characterize the error associated with that prediction. Others have made predictions of elders at risk for failing,3 but the contribution of SOT scores in determining the likelihood of falling has not been delineated. Therefore, additional research is needed to evaluate the association between SOT scores and other factors contributing to falls among elders living at home. Dr Horak has suggested that selfreports of fails may be a poor reflection of actual falls. We attempted to minimize the problem in our study by (1) excluding subjects reporting only one fall within 6 months of the study and (2) selecting subjects who were not institutionalized. The subjects in our study were functional elders with the mental facilities to be independent at home.

The remainder of Dr Horak's comments deal with her concern about the importance of considering multiple factors when evaluating falls in elders. We believe that this issue was addressed in the article. Our discussion deals with potential balance problems related to lower-extremity strength and fear of failing. The introduction also identifies other possible "risk" factors including sedative use, cognitive impairment, and lower-extremity disability (as outlined by Tinetti and others'). The summary states, "Further research is [required] to identify the critical factors needed to screen 'at-risk' elders in a fail-prevention program." Therefore, we do agree with Dr Horak that the SOT may be very useful as a measure of specific balance attributes, but it has not yet been shown to be a sensitive and specific predictor for falls in the elderly.

Susan L Anacker, PT

Richard e Di Fabio, PhD, PT

References

1 Shumway-Cook AS, Horak FB. Assessing the influence of sensory interaction on balance: suggestion from the field. Phys Ther. 1986;66: 1548-1550.

2 Hays WL. Statistics. 4th ed. New York, NY: Holt, Rinehart & Winston Inc; 1988:470.

3 Tinetti ME, Speechley M, Ginter S. Risk factors for fails among elderly persons living in the community. N EnglJ Med. 1988;319: 1701-1707.