Test-retest reliability of two tiltboard tests in children.

Key Words: Equilibrium; Pediatrics, general; Posture, tests and measurements; Tests and measurements, functional.

According to Horak,[1] there are at least three major components to postural control that physical therapists and occupational therapists might assess to try to determine the etiology of postural instability in persons with problems in this area. These components are (1) the biomechanical area, including available range of movement and strength; (2) the ability of the sensory system to detect a change of the center of gravity over the base of support; and (3) the precisely timed reactions of the motor system (equilibrium reactions) to keep the person in a stable position.

Pediatric physical therapists and occupational therapists have historically put great emphasis on the evaluation of equilibrium reactions and their facilitation in treatment.[2,3] Equilibrium reactions are thought to provide the background postural set necessary for the development and execution of all

skilled adaptive motor responses.[4,5] The motor responses that occur to maintain a stable posture vary with the environmental conditions both internal and external to the individual.[6-9] Reactions may differ when the person is on a stable versus an unstable base of support.[8,9] These reactions may also differ when the center of gravity is actively displaced by the individual or passively displaced by some external force.[8] The person's internal or central postural set, as related to the person's immediately preceding movement history and past movement experiences, may also affect equilibrium reactions.[6,9] Because the reactions are based on different internal and external conditions, Bundy et al[10] suggest that clinicians use a variety of tests to examine postural control reactions. They developed a series of equilibrium tests for children that attempted to mimic functional situations in which balance might be lost.

Most clinical evaluations of motor reactions to loss of balance (ie, visible responses of the trunk, head, and extremities when the child's balance is disturbed) are relatively subjective and have not been adequately tested for reliability. Therapists commonly isolate and judge equilibrium reactions by placing children on unstable surfaces such as balls or tiltboards and displacing the surfaces. Less commonly used, but perhaps more functionally related, are tests that require the child to reach for an object such that the child must actively displace his or her center of gravity in order to obtain the object.[11]

A few standardized methods for evaluation of equilibrium reactions have been developed. Fisher and Bundy[11,12] developed three tests to evaluate equilibrium reactions in children. These tests involved having the child (1) stand on a small tiltboard or on a stable surface and reach for an object held to his or her side or (2) maintain standing balance on a tiltboard while the board was manually displaced. Criteria for evaluation included not only documenting the angular displacement of the tiltboard, but also recording the qualitative responses (ie, the amount of abduction of the elevated arm and leg, whether the elbow remained extended or was flexed, or whether the knee was extended or flexed) of the trunk, arms, and legs during the equilibrium reaction.

Interrater and intrarater reliabilities of tiltboard angle measurements (expressed as Pearson Product-Moment Correlation Coefficients) and the qualitative descriptions of body part movements (expressed as percentages of agreement) in Fisher and Bundy's three tiltboard tests were greater than .85.[11,13,14] The validity of the tiltboard tests was supported because the three angle measurements could be used to predict age and to discriminate between boys who were developing normally and boys with sensory integrative dysfunction.[12] Test-retest reliabilities, however, were not documented with the three tiltboard tests.

Atwater et al[15] developed two tiltboard tests involving passive tilting in standing with the child's eyes open and closed. These tiltboard tests were originally designed as one component of a group of balance tests including biomechanical screening, sensory interactions for balance, and motor reactions.[16] The tiltboard was used to score a child's postural reaction because tiltboards are easy to use, readily available, and currently used in clinics. The authors thought that a simple scoring of the maximum tilt prior to the child making a gross movement to maintain balance (such as taking a step) would characterize a child's ability to control posture while on a movable surface. Maintaining postural control on the tiltboard could potentially reflect conditions in daily life such as maintaining balance while playing on playground equipment with movable parts or while riding on a bus or a boat. The eyes-closed condition eliminates visual input so that the child's ability to use somatosensory (proprioceptive and cutaneous) and vestibular cues can be evaluated. These tests, if found to provide reliable measurements, could be used to document changes in the ability to maintain posture over time and across therapy sessions. These tests, however, would need to be supplemented by other tests of postural control to determine whether any improvements in functional skills resulted.

Reliability research on Atwater and colleagues'[15] tiltboard tests with children who were typically developing indicated that although interrater reliabilities were high (eyes open, r=.98; eyes closed, r=.98), test-retest reliability coefficients (with a 1-week interval between tests) were low (eyes open, r=.45; eyes closed, r=.45). Atwater et al hypothesized that learning (practice effects) between the two tests may not have been controlled for by giving just two trials. Not only were reliability coefficients low, but there was also a significant difference between the mean angle measurements for test and retest, with children having higher measurements on the retest. In addition, because the speed of tilt may influence the sensory system stimulated to trigger a reaction,[17-19] the potential inconsistency in speed of tilt noted by the authors may have affected the results.

Atwater et al[15] recommended that further research incorporate more trials to potentially eliminate any learning effects and a standard speed of tilt. The authors hypothesized that although this "learning effect" was seen with children who were typically developing, it might not occur in children with balance difficulties. Responses of typically developing children might be more stereotypical and not subject to learning over such a short period of time. Thus, Atwater et al suggested that reliability of measurements obtained with the tiltboard tests be examined with children who have balance difficulties.

Our study was designed to follow up on the suggestions made by Atwater et al.[15] The two tiltboard tests - tiltboard tilt-eyes open and tiltboard tilt-eyes closed - were modified in two ways. First, the speed of tilt was standardized through use of a metronome. Second, three additional trials were added to allow for practice or learning effects, resulting in a total of five trials per test. The decision to have a total of five trials was made to allow the child to give his or her best effort without having so many trials that the child lost concentration and tired. The first two trials were regarded as practice trials, which allowed the child to become accustomed to the tiltboard tests.

The test-retest reliability of measurements obtained with these two modified tests of tiltboard postural control were examined for children who were typically developing and for children with developmental delays. Two research questions were addressed:

1. What is the test-retest reliability of measurements obtained with the two tiltboard tests?

2. What are the magnitudes of difference between test and retest angle measurements obtained with the two tiltboard tests?

Method

Subjects

Subjects were 18 children, aged 53 to 81 months (X=64.4, SD=8.3), with no known developmental problems (TD group) and 18 children, aged 50 to 79 months (X=63.3, SD=8.4), with developmental delays (DD group). To minimize differences between the two groups, a matched cohort sample was used for the study. Children were divided into age groups of 4-year-olds, 5-year-olds, and 6-year-olds, with 6 children (4 boys, 2 girls) at each age in the TD and DD groups. The unequal gender distribution reflects the higher incidence of males than females having developmental disabilities.[20]

Children who were typically developing were defined as children with no history of neurologic, orthopedic, or learning disorders as reported on a questionnaire completed by the parents. Children with developmental delays fit the following criteria: They had the ability to ambulate independently with no assistive devices, the ability to understand and follow simple directions, scores greater than 1 standard deviation below the mean on the Peabody Developmental Gross Motor Scale,[21] and placement according to the Washington Administrative Code (WAC) criteria in a category for the handicapping condition of developmental delay. The WAC contains the rules and regulations implementing all applicable federal and state laws regarding the provision of educational services for children of common school age in Washington state. The WAC criteria for the handicapping condition of developmental delay specify that the child have scores on norm-referenced tests greater than 2 standard deviations below the mean in at least one area of development or that the child have scores greater than 1.5 standard deviations below the mean in two or more areas.

Subjects in the DD group demonstrated a variety of types of developmental delays. Diagnoses of the subjects included autism, Down syndrome, pervasive developmental disability, hearing disorders, hypotonia, post-liver transplant, and no known diagnosis. Amount of delay from chronological age in gross motor skins ranged from 5 to 35 months (X=16, SD=8) on the Peabody Developmental Motor Scales.

Subjects were recruited from the Shoreline School District in the Seattle area and the Experimental Education Unit at the Child Development and Mental Retardation Center at the University of Washington. Subject from the Shoreline School District included typically developing children who attended a district-sponsored day-care program. Subjects from the Experimental Education Unit included both typically developing and developmentally delayed children who attended an integrated preschool program. Subjects gave verbal consent prior to testing, and their parents signed consent forms. Parents also completed a short questionnaire accompanying the consent forms. The information from the questionnaire was used to determine whether subjects met the criteria for inclusion in the study. From the returned consent forms, subjects meeting the criteria for inclusion were randomly selected for the study.

All subjects were tested initially and then completed retesting approximately 1 week later. The 1-week test-retest interval was chosen so that each test was administered on approximately the same day of the school week. (Because of a holiday and a school field trip, eight of the subjects were retested either 6 days or 8 days after initial testing.) Changes in test scores were not expected because of maturation or practice from initial to retest sessions. None of the subjects had an opportunity to practice with the tiltboard between testing sessions. In addition, the same time of day was used for testing sessions; that is, the subject was tested and retested in either the morning or the afternoon.

Instrumentation and Procedures

Each child was tested on two tiltboard tasks. The order of the two tests was randomized for each subject. In 56% of the subjects, the eyes-open condition came first; in 44% of the subjects, the eyes-closed condition came first. For both tasks, five trials for each side were completed, alternating right and left sides. A 1-minute rest period was given between the two tests. Prior to testing, the child removed his or her shoes and socks. For each trial, the child stood on a 45.7x45.7-cm (18 x 18-in) tiltboard with a nonskid surface (Figure). Consistent foot placement was achieved by markers placed on the tiltboard. The tiltboard had angle markers that extended 10.2 cm (4 in) from the board to the wall, where a grid was located. The grid had degree lines from 0 to 90 degrees, in increments of 5 degrees. Because of the tiltboard design and the placement of the angle markers, degrees noted on the grid did not correspond exactly with the actual degrees that the tiltboard was tilted. For example, at a grid reading of 90 degrees, the tiltboard was actually at a 50-degree angle relative to horizontal plane.

Tiltboard tilt-eyes open test. The child was positioned in the middle of the tiltboard with feet together and medial malleoli touching one another. The child was instructed to remain with hands on hips and feet in the original position as long as possible while the examiner manually tipped the board to the right or to the left at the rate of approximately 15 degrees per second. Timing was controlled by use of a metronome, which registered each second. A bold line at each 15-degree line on the grid was used to assist in controlling movement of the platform. A second examiner guarded for falls and watched for any postural adjustment that signaled the end of the trial. Postural adjustments included the child removing hands from hips, stepping, or beginning to fall (eg, the spotting examiner felt she must catch the child to prevent a fall). The child was told in which direction the board would tilt. The board was tipped in each direction five times, alternating sides after each change in direction (ie, right-left, right-left, and so on). Direction (right or left) of initial tilt alternated for each successive subject. For each trial, angle displacement of the board at the time of any postural adjustment was recorded.

Tiltboard tilt-eyes closed test. The procedure was identical to the eyes-open procedure, except that the child was instructed to close the eyes during the passive tilt. Opening of the eyes during the trial signaled a postural adjustment and the end of the trial.

The two examiners remained consistent throughout the testing. Percentage of agreement for angle measurements, within 5 degrees, was established at 90% or greater prior to data collection and was rechecked on six occasions by an independent rater during the data collection. Percentage of agreement for angle measurements, within 5 degrees, was 97%. All subjects were retested approximately 1 week (6-8 days) after initial testing. Order of the two tests, testing conditions, and the tester remained constant between testing sessions, because these were variables that potentially could affect reliability.[22] In addition, during the initial testing session, the rater noted whether the subjects appeared to understand the directions given for testing and whether they cooperated with the testing.

From the right and left tilt measurements for the five trials, a single angle measurement was determined by taking the median of the last three of five angle measurements. This was done for both initial and retest measurements. The median measurement was chosen for two reasons. First, the median measurement is less subject to extreme scores,[23] such as when a child immediately falls during a trial. Second, this measurement is easy for clinical use because it can be determined quickly.

For each subject, angle measurements were obtained separately for right-and left-sided tilt. This was done because of the potential clinical relevance of being able to examine right and left angle measurements separately to detect asymmetries in performance.

Data Analysis

All statistical analyses were completed using the median angle measurement of the last three trials. Descriptive statistics for test and retest measurements were calculated and distributions were plotted for the TD group and the DD group separately for each of the four conditions (eyes open-tilt right, eyes open-tilt left, eyes closed-tilt right, and eyes closed-tilt left). Intraclass correlation coefficients (Shrout and Fleiss formula 2,1,[24] a two-way, random-effects, repeated-measures model) were used as indexes of reliability. The two-way, random-effects, repeated-measures model was used because it considers differences between test and retest measurements to be sources of error. Reliability coefficients generated by this model also reflect a situation comparable to the typical clinical situation in which a person's measurement is based on one administration of the test rather than being based on the average of two or more administrations of the test.[25-27] Test-retest reliability coefficients in this study were not tested for significance because angle measurement distributions were moderately skewed. Significance testing requires angle measurement distributions that approximate normality and sampling distributions for all possible time intervals.[27,28]

The next step in the data-analysis process involved examination of the stability of angle measurements from test to retest using a nonprobabilistic index of agreement. The magnitudes of difference between individual test and retest angle measurements (ie, retest measurement minus test measurement) were calculated in increments of 5 degrees to determine the actual differences between the measurements.

Results

Table 1 details the descriptive statistics for the angle-of-tilt measurements for the two tiltboard tests, for the right and left sides separately. As evidenced by examination of test and retest means and medians, as a group, retest measurements were higher than test measurements.

Table 1. Descriptive Statistics for Angle-of-Tilt Measurements (in Degrees)

Test                     X       Median   SD    Low Value/
TD group(a)                                     High Value
Eyes open-tilt right
Test                     67.78   70.00   16.47   20/90
Retest                   75.00   77.50   10.00   50/90
Eyes open-tilt left
Test                     71.94   72.50   11.13   50/90
Retest                   74.72   80.00   11.44   50/90
Eyes closed-tilt right
Test                     48.06   47.50   16.90   15/80
Retest                   56.39   55.00   15.61   30/85
Eyes closed-tilt left
Test                     50.56   47.50   13.05   30/70
Retest                   59.17   65.00   15.54   30/75
DD group(b)
Eyes open-tilt right
Test                     36.39   30.00   20.57    5/75
Retest                   47.78   45.00   22.38   20/80
Eyes open-tilt left
Test                     37.78   35.00   18.49    5/75
Retest                   47.78   47.50   21.57   20/85
Eyes closed-tilt right
Test                     20.00   20.00   10.43    5/40
Retest                   23.06   22.50   13.19    5/60
Eyes closed-tilt left
Test                     20.56   15.00   16.35    5/75
Retest                   24.17   30.00   12.97    5/55
(a) TD group=typically developing subjects (n=18).
(b) DD group=developmentally delayed subjects (n=18).

Intraclass correlation coefficients (two-way, random-effects, repeated-measures model) are reported in Table 2. These values ranged from .49 to .54 for the TD group and from .52 to .82 for the DD group.

Table 2. Intraclass Correlation
Coefficients (Model 2,[24]) for Test-Retest
Measurements

                         TD         DD
Test                     Group(a)   Group(b)

Eyes open-tilt right     .54        .61
Eyes open-tilt left      .53        .52
Eyes closed-tilt right   .49        .82
Eyes closed-tilt left    .54        .53

(a) TD group=typically developing subjects
(n=18).
(b) DD group=developmentally delayed subjects
(n = 18).

Magnitudes of difference between the retest and test angle measurements, in increments of 5 degrees, are presented in Table 3. This table reports the number of subjects having retest angle measurements a specific number of degrees higher or lower than test angle measurements, as well as the cumulative percentage of subjects whose test and retest measurements were within that specific number of degrees. For example, in the second row [+ or -] 5) for the eyes open-tilt left test, the 7" indicates that seven subjects had angle measurement differences of 5 degrees. The next line (2/5) indicates that two of the seven subjects had angle measurement differences of +5 degrees between retest and test and that five of the seven subjects had angle measurement differences of -5 degrees between retest and test. The next line is a cumulative percentage and indicates that 55.6% of the TD group subjects had test-retest angle measurement differences of less than or equal to [+ or -] 5 degrees for the eyes open-tilt left test.

Table 3. Magnitude of Difference Between Retest and Test Angle-of-Tilt Measureme
nts(a)
                        Eyes    Eyes    Eyes      Eyes
Differences             Open-   Open-   Closed-   Closed-
in Degrees              Tilt    Tilt    Tilt      Tilt
of Tilt                 Right   Left    Right     Left
TD group(b)
  No. of subjects         3       3        2        7
  ([+ or -]) 0            3       3        2        7
  Cumulative percentage   16.7    16.7     11.1     38.9
  No. of subjects         6       7        4        4
  ([+ or -]) 5            (3/3)   (2/5)    (2/2)    (4/0)
  Cumulative percentage   50.0    55.6     33.3     61.1
  No.of subjects          8       4        3        1
  ([+ or -]) 10           (8/0)   (3/1)    (3/0)    (1/0)
  Cumulative percentage   94.4.   77.8     50.0     66.7
  No. of subjects         0       3        4        2
  ([+ or -]) 15           (0/0)   (2/1)    (3/1)    (2/0)
  Cumulative percentage   94.4    94.4     72.2     77.8
  No. of subjects         0       0        1        0
  ([+ or -]) 20           (0/0)   (0/0)    (1/0)    (0/0)
  Cumulative percentage   94.4    94.4     77.8     77.8
  No. of subjects         0       0        1        1
  ([+ or -]) 25           (0/0)   (0/0)    (1/0)    (1/0)
  Cumulative percentage   94.4    94.4     83.3     83.3
  No. of subjects         0       1        0        3
  ([+ or -]) 30           (0/0)   (1/0)    (0/0)    (3/0)
  Cumulative percentage   94.4    100      83.3     100
  No. of subjects         1       0        3        0
  >([+ or -]) 30          (1/0)   (0/0)    (2/1)    (0/0)
  Cumulative percentage   100     100      100      100
DD group(c)
  No. of subjects         1       3        7        7
  ([+ or -]) 0            1       3        7        7
  Cumulative percentage   5.6     16.7     38.9     38.9
  No. of subjects         7       5        8        4
  ([+ or -]) 5            (7/0)   (1/4)    (5/3)    (4/0)
  Cumulative percentage   44.4    44.4     83.3     61.1
  No. of subjects         1       1        1        3
  ([+ or -]) 10           (1/0)   (1/0)    (1/0)    (3/0)
  Cumulative percentage   50.0    50.0     88.9      77.8
  No. of subjects         3       2        1         1
  ([+ or -]) 5            (2/1)   (2/0)    (1/0)     (1/0)
  Cumulative percentage   66.7    61.1     94.4      83.3
  No. of subjects         2       3        1         1
  ([+ or -]) 20           (1/1)   (3/0)    (1/0)     (1/0)
  Cumulative percentage   77.8    77.8     100       88.9
  No. of subjects         1       1        0         1
  ([+ or -]) 25           (1/0)   (1/0)    (0/0)     (1/0)
  Cumulative              83.3    83.3     100       94.4
  No. of subjects         1       1        0         0
  ([+ or -]) 30           (1/0)   (0/1)    (0/0)     (0/0)
  Cumulative percentage   88.9    88.9     100       94.4
  No. of subjects         2       2        0         1
  >([+ or -]) 30          (2/0)   (2/0)    (0/0)     (0/1)
  Cumulative percentage   100     100      100       100
(a) Number before slahs denotes number of subjects who had higher retest measure
ments compared
with test measurements; number after slash denotes number of subjects with lower
 retest measurements
compared with test measurements.
(b) TD group=typically developing subjects (n=8).
(c) DD group=develomentally delayed subjects (n=8).

Wide variation was seen between the two groups of subjects and the two tests on agreement of test and retest angle measurements. For the TD group, 94.4% of test and retest measurements were within 15 degrees of one another for the eyes open-tilt right and eyes open-tilt left tests; however, only 72.2% of the measurements were within 15 degrees of each other for the eyes closed-tilt right test, and only 77.8% of the measurements were within 15 degrees of each other for the eyes closed-tilt left test. For the DD group, the test and retest measurements were within 15 degrees of each other for 66.7% of the subjects for the eyes open-tilt right test, for 61.1% of the subjects for the eyes open-tilt left test, for 94.4% of the subjects for the eyes closed-tilt right test, and for 83.3% of the subjects for the eyes closed-tilt left test. The largest difference between test and retest was 50 degrees for the TD group and 45 degrees for the DD group.

Discussion

All of the subjects in the TD group appeared to understand the directions given to them and cooperated with the testing. In the DD group, 12 of the children (67%) appeared to understand the directions well and cooperated throughout all trials during testing. The other 6 children (33%) required extra cues or limits to complete the test. Although a total of 20 developmentally delayed children were tested, 2 children failed to cooperate and their data were not included in the analyses.

When comparing the results of this study with those of the original study by Atwater et al,[15] the test-retest reliability of measurements obtained with the two tiltboard tests showed slight improvement. Reliability coefficients in the Atwater et al study were .45 for both the eyes-open and eyes-closed tests using the Spearman Rank-Order Correlation Coefficient. When the tiltboard tests were modified by adding trials and standardizing the speed of tilt, reliability coefficients ranged from .49 to .82 using the intraclass correlation coefficient, a more conservative estimator of reliability. The test-retest reliability coefficients, however, are still low for most measures (six of eight were less than .55). Also, when magnitudes of difference between test and retest angle measurements were examined, a clear practice effect, similar to that found in the original study,[15] was demonstrated.

Although changes in the tiltboard tests seem to have improved the test-retest reliability slightly, we believe reliability is still not adequate to justify using the tests to assess change over time.

The reliability of these measurements can be compared with that of other pediatric balance tests that have documented test-retest reliability. The Standing Balance Eyes Open Test, part of the Southern California Sensory Integration Tests, has Pearson Product-Moment Correlation Coefficients ranging from .55 to .72, depending on the age of the children.[29] On the Standing Balance Eyes Closed Test, test-retest reliability coefficients ranged from .16 to .51.[29] Ayres' newer test, the Sensory Integration and Praxis Test, documents test-retest reliabilities of .80 to .86 on the Standing and Walking Balance Test.[30] Test-retest reliability for one-foot standing balance was documented by Atwater and colleagues; Spearman coefficients were .95 for the eyes-open tests and .77 for the eyes-closed tests.[15] The reliability coefficients in our study were somewhat lower than those of Ayres' newer test and the one-foot standing balance tests as described by Atwater and colleagues. Comparison of reliability data is problematic, however, because of the different statistics used by researchers.

When examining magnitudes of difference between test and retest measurements and comparing the right- and left-sided tilt for each test, no differences were noted between the two sides. In post hoc analysis of each subject's data, no clear relationship was seen between the subject's age and the consistency of response. Likewise, no clear relationship was seen between any particular disability and extreme inconsistency, although no subset of subjects with a specific disability included enough subjects to draw any firm conclusions. Some of the largest gaps between test and retest measurements were seen in the eyes-closed tests for a few subjects in both the TD and DD groups. This may be because the measurements for the eyes-closed tests were typically lower than for those for the eyes-open tests, therefore giving more potential for improvement on retest. The eyes-closed tests also represent tasks that many children of this age do not experience in everyday functional activities, so they may be fearful at first. With several practices, however, improvement seems to occur.

Five of the eight measurements showed a mean change from test to retest of approximately 10 degrees. The reliability of these measurements might have been improved had the data been analyzed in 10-degree increments instead of 5-degree increments. One would expect the balance of children with postural problems, however, to progress in very small increments. We would view 10 to 15 degrees as a meaningful change, and would want to be able to discern such a change secondary to treatment. For example, if a child initially achieved 50 degrees on the tiltboard test, the range for improvement would only be 40 more degrees (90 [degrees] maximum). Using an increment of more than 5 degrees would decrease the sensitivity of the test.

The two tiltboard tests used examined only the children's reactions to passive tilting. Postural control is a multifaceted ability that involves not only reacting to externally induced changes, but also actively controlling and moving within the environment, including situations in which vision, proprioception, or somatosensation may be absent or distorted. Tilt reactions during active movement were not examined in this study. In addition, adaptation to tilt at only one specific speed was tested. Because little is known regarding the response of the vestibular system to varying speeds of movement, other speeds of tilt could potentially produce varying results.

Directions for Future Research

The test-retest reliability of these tests might be improved by additional trials, until a plateau in the learning curve occurs. Adding too many trials, however, may make the test too long and time consuming to be useful as a screening tool.

Aspects of tilt reactions such as the qualitative motor patterns used, if objectively quantified as done by Fisher,[13] may show better test-retest reliability, as well as differentiate between TD and DD groups. Further research in this area is warranted.

Conclusions

Measurements obtained for the two tiltboard tests studied, for both children with and without developmental delays, were not stable across testing sessions. As a result, clinicians should be cautious in using these tests to measure treatment effects. These tests might be useful as a screening tool if used in conjunction with other balance assessments. Further research may be merited to refine the protocol in order to achieve higher levels of test-retest reliability and to determine the extent that these tests can be used diagnostically.

Methods of assessing postural control abilities in children vary widely among therapists. This study examined only two tests many others are used clinically. Some tests involve adaptation to passive movement, whereas others examine postural changes during active movement. Further research should explore the reliability and validity of measurements obtained with such tests before the tests are used in clinical documentation and decision making.

References

[1] Horak FB. Clinical measurement of postural control in adults. Phys Ther. 1987;67:1881-1884. [2] Bobath B. The very early treatment of cerebral palsy. Dev Med Child Neurol 1967;9:373-390. [3] Bobath K, Bobath B. The neurodevelopmental approach to treatment. In: Pearson PH, William CE, eds. Physical Therapy Services in the Developmental Disabilities, Springfield, Ill: Charles C Thomas, Publisher; 1972:114-185. [4] Gilfoyle EM, Grady AP, Moore JC. Children Adapt. Thorofare, NJ: SLACK Inc; 1981. [5] Bobath B. Postural Reflex Activity Caused by Brain Lesions. London, England: William Heinemann Medical Books Ltd; 1971. [6] Horak FB, Diener HC, Nashner LM. Influence of central set on human postural responses. J Neurophysiol. 1989;62:841-853. [7] Di Fabio RP, Badke MB, McEvoy A, Breunig A. Influence of local sensory afferents in the calibration of human balance responses. Exp Brain Res. 1990;80:591-599. [8] Nardone A, Corra T, Schieppati M. Different activations of the soleus and gastrocnemius muscles in response to various types of stance perturbation in man. Exp Brain Res. 1990;80: 323-332. [9] Nardone A, Giordano A, Corra T, Schieppati M. Responses of leg muscles in humans displaced while standing: effects of types of perturbation and of postural set. Brain. 1990; 113: 65-84. [10] Bundy AC, Fisher AG, Freeman B, et al. Concurrent validity of equilibrium tests in boys with learning disabilities with and without vestibular dysfunction. Am J Occup Ther. 1987;41:28-34. [11] Fisher AG. Equilibrium: Development and Clinical Assessment. Boston, Mass: Boston University; 1984. Dissertation. [12] Fisher AG, Bundy AC. Equilibrium reactions in normal children and in boys with sensory integrative dysfunction. Occup Ther. J Res. 1982;2:171-183. [13] Fisher AG. Objective assessment of the quality of response during two equilibrium tasks. Physical and Occupational Therapy in Pediatrics. 1989;9:57-78. [14] lzraelevitz TE, Fisher AG, Bundy AC. Equilibrium reactions in preschoolers. Occup Ther. J Res. 1985;5:154-169. [15] Atwater SW, Crowe TK, Deitz JC, Richardson PK. Interrater and test-retest reliability of two pediatric balance tests. Phys Ther. 1990;70: 79-87. [16] Crowe TK, Deitz JC, Richardson PK, Atwater SW. Interrater reliability of the Pediatric Clinical Test of Sensory Interaction for Balance. Physical and Occupational Therapy in Pediatrics. 1990;10:1-27. [17] Weisz S. Studies in equilibrium reaction. J Nerv Ment Dis. 1938;88:150-162. [18] Haas G, Diener HC, Bacher M, Dichgans J. Development of postural control in children: Short-, medium-, and long-latency EMG responses of leg muscles after perturbation of stance. Exp Brain Res. 1986;64:127-132. [19] Haas G, Diener HC, Rapp H, Dichgans J. Development of feedback and feedforward control of upright stance. Dev Med Child Neurol. 1989;31:481-488. [20] Behrman RE, Vaughan VC. Nelson Textbook of Pediatrics. 13th ed. Philadelphia, Pa: WB Saunders Co; 1987. [21] Folio MR, Fewell RR. Peabody Developmental Motor Scales and Activity Cards Allen, Tex: Developmental Learning Materials; 1984. 22 Deitz JC. Reliability. In: Miller LJ, ed. Developing Norm-Referenced Standardized Tests Binghamton, NY. The Haworth Press Inc; 1989: 125-147. [23] Cermak S. Norms and Scores. In: Miller LJ, ed. Developing Norm-Referenced Standardized Tests. Binghamton, NY: The Haworth Press Inc; 1989:91-123. [24] Shrout PE, Fleiss J. intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420-428. [25] Berk RA. Generalizability of behavioral observations: a clarification of interobserver agreement and interobserver reliability. Am J Ment Defic. 1979;83:460-472. [26] Suen HK, Ary D. Analyzing Qualitative Behavioral Observation Data. Hillsdale, NJ: Lawrence Erlbaum Associates Inc; 1989. [27] Tinsley HE, Weiss DJ. Interrater reliability and agreement of subjective judgements. J Counseling Psychol. 1979;22:358-376. [28] Haggard EA. Intraclass Correlation and the Analysis of Variance. New York, NY: Dryden Press; 1958. [29] Ayres AJ. Southern California Sensory Integration Tests Manual Los Angeles, Calif: Western Psychological Services; 1980. [30] Ayres AJ. Sensory Integration and Praxis Test Manual Los Angeles, Calif Western Psychological Services; 1988.