Marchetti, Gregory F. PT, PhD; Lin, Chia-Cheng PT, MS; Alghadir, Ahmad PT, PhD; Whitney, Susan L. PT, DPT, PhD, NCS, ATC, FAPTA
The measurement of gait in persons with balance disorders is well established and is commonly used to quantify change over time. Gait speed is a very powerful measure of gait performance,1 yet movement of the head while walking, turning, or even changing speeds can often provide additional information about what movements affect gait in patients with balance disorders.2,3 Two commonly used tools that assess dynamic gait include the dynamic gait index (DGI) and the functional gait assessment (FGA).4,5 Both the DGI and the FGA are ordinal measure that are easy to use, reliable, and valid measures of gait performance of persons with balance and vestibular dysfunction.3,5,6 However, there are currently no data available related to how much change in either measure is a meaningful change in performance. Responsiveness is defined as the ability of an instrument to detect clinically relevant change over time.7 When determining responsiveness, the variability of the sample and measurement precision are important considerations. Knowing the amount of change that constitutes a clinically important change could help clinicians better determine whether their interventions was effective.
One important aspect of identifying clinically important change between time points in a measure is the identification of “true” change versus chance or random variation. Minimum detectable change (MDC) has been described as the smallest amount of change in a measure that exceeds measurement error.8 The MDC95 is used to reflect a true change in the measure at 95% certainty. This estimation is based on the distribution of subject measurements and the amount of reliability (test-retest or internal consistency) observed within the sampled measurements.9 The MDC95 as a measurement property has been identified as an important threshold by which to objectively determine the occurrence of patient improvement of a course of intervention.
The purpose of this study was 2-fold: (1) to determine the responsiveness and MDC of the DGI and the FGA in persons with balance and vestibular disorders and (2) to estimate the change in self-reported disability associated with a significant change in gait performance as measured by the DGI and FGA resulting from rehabilitation for dizziness and imbalance.
The study used a retrospective design and was conducted at a tertiary care balance and vestibular clinic. The data extraction methods were approved by the University of Pittsburgh Institutional Review Board. Four-hundred four consecutive patients with balance and vestibular disorders who had had at least 2 physical therapy visits were included in the data extraction. Data were collected via chart review and were systematically recorded into an Excel database. Data extracted included age, sex, and, diagnosis. In addition, initial and discharge DGI, FGA, Activities-Specific Balance Confidence (ABC) scale,10 and dizziness handicap inventory (DHI)11 scores were recorded.
The DGI is an ordinal test of gait function that has been used extensively in persons with balance and vestibular disorders.4 Eight aspects of gait are scored on the basis of observation as the patient walks over a 6.1-m level surface. The rater records an ordinal score that ranges from 0 (unable or done very poorly) to 3 (normal score) for a total point value of 24. Scores of less than or equal to 19 appear to suggest fall risk.2,12,13 Items included in the DGI are ambulation on a level surface, over and around objects, at various speeds, up and down stairs, turning and stopping plus walking with head movements in the pitch and yaw plane. Reliability and validity of the instrument has been previously reported.3,14,15
The FGA is a modification of the DGI with the inclusion of items with enhanced difficulty (tandem walking, walking with eyes closed, walking backward) and an altered scoring algorithm method on several of the original DGI items. These modifications were intended to improve the clinical responsiveness of the DGI and to reduce the likelihood of a ceiling effect that has been reported with the DGI.5,16 The FGA total score ranges from 0 to 30 has been shown to be reliable and valid in persons with balance and vestibular disorders.5,6
External criteria for self-reported improvement with intervention were provided through the DHI and the ABC. The DHI is a tool that was developed to report the handicapping effects of dizziness, and the properties of the DHI are well documented.11,17 The scores on the DHI range from 0 (optimal) to 100 (the worst score possible). The DHI provides a measure of whether the patient felt that they had improved at the end of the intervention. Clinically meaningful change of the DHI has previously been reported as 18 points.11
The ABC scale is another self-report measure commonly used with persons with balance and vestibular disorders.10 The ABC scale consists of 16 commonly performed items that are rated on a confidence scale of 0 to 100 that suggests how confident the person is performing the activity. ABC scores between 80 and 100 indicate a high level of functioning in older adults, scores of 50 to 80 indicate a moderate level of functioning, and scores less than 50 suggest low functioning.18 Others have suggested that scores of less than 67% on the ABC scale are predictive of future fall events.19
A cohort was selected from consecutive patients receiving rehabilitation at a tertiary care balance center between September 2004 and April 2009 with primary complaints of dizziness and/or imbalance. Patients were considered for inclusion in the cohort if they were 18 years or older and had been scored on the DGI and FGA at the start of care and at discharge from rehabilitation services by a physical therapist. This cohort of 385 patients had a mean age of 58.4 ± 18.5 years (67% were women). The median symptom duration of the patients in the cohort was 4 (range, 1-462) months. The median number of clinic visits was 6 (range, 2-54). Of the 385 initial patients, 33% (n = 126) had peripheral vestibular disorders other than benign paroxysmal positional vertigo (BPPV), 25% (n = 95) had central nervous system disorders other than mild traumatic brain injury, 15% (n = 57) had mild traumatic brain injury, 15% (n = 59) had BPPV, and 9% (n = 36) had unspecified causes of dizziness or imbalance. Diagnoses in the peripheral grouping included Meniere's disease (4%), unilateral hypofunction (17%), bilateral hypofunction (2%), vestibule-ocular reflex (VOR) dysfunction of uncertain etiology (1%), and status postacoustic neuroma resection (1%), labrynthitis (1%) cholesteatoma removal (1%), and sensory neuropathy (2%). Central diagnoses included cerebellar disorders (1%), anxiety disorders (1%), migraine dizziness (4%), multiple sclerosis (2%), and orthostatic tremor (1%). All subjects were diagnosed by a board certified neurologist who specializes in otology.
From this initial cohort of 385 patients receiving intervention, patients were excluded if they had maximum or near-maximum scores on either the DGI (23 or 24) or the FGA (29 or 30) at the onset of care unless they demonstrated performance regression between the onset of care and discharge on either measure. This resultant cohort consisted of 326 patients with a mean age of 60 ± 18.3 years (range, 18-95 years; 69% female). The median symptom duration of the cohort was 4 (range, 1-264) months. The median number of clinic visits was 6 (range, 2-54). Of the 326 patients in this cohort, 35% (n = 115) had peripheral vestibular disorders other than BPPV, 25% (n = 80) had central nervous system disorders other than mild traumatic brain injury, 15% (n = 48) had mild traumatic brain injury, 14% (n = 47) had BPPV, and 11% (n = 36) had unspecified causes of dizziness or imbalance.
In this patient cohort, dizziness was the primary symptomatic complaint for 95 (29%), imbalance for 131 (40%), and both symptoms for 100 (31%) at the onset of care. Within this cohort, start of care and discharge ABC scores were reported for 274 of the subjects. Start of care and discharge scores on the DHI were reported for 248 of the subjects. Of the 326 subjects included in the analyzed cohort, 25 patients (7.7%) displayed declining scores on the FGA. The range of regressions was 1 to 13 points. A declining DGI score was observed in 33 (10%) of patients. The range of decline was 1 to 21 points.
Effect of Age and Sex on Changes in Gait and Self-Report Scores
The potential effects of age, symptom duration, and sex on changes in DGI, FGA, DHI and ABC scores between initiation of treatment and discharge were tested using analysis of covariance. The mean change score for each measure was compared between male and female subjects with age as a covariate. The effect of sex or age was judged to be significant at P < 0.05.
Minimum Detectable Change
The change score for the 2 measures was obtained by subtracting their discharge score from the initial DGI or FGA score. The minimal detectable change in both the DGI and the FGA was determined using the following formula:
where SEM is a measure of within-subject variability, defined as the standard deviation (SD) in the baseline measures adjusted for the internal consistency (coefficient alpha) for items in both the DGI and the FGA (SEM = SD × √[1 − α]). The value of MDC95 describes the amount of true change in subject status beyond measurement error with 95% certainty.20
Responsiveness of the DGI and FGA was determined via the standardized response mean (SRM). The SRM raw score was defined as the mean raw change score/SD of the change score. The ceiling effect on the 2 measures was described as the proportion of subjects who attained the maximum discharge score on the 2 measures (DGI = 23–24, and FGA = 29–30).
Disability Reduction Associated With MDC95
The reduction in disability was described by comparing mean changes in scores of DHI and ABC obtained at the start of care and discharge between subjects who exceeded the MDC95 and those who did not exceed the MDC95 on the DGI and the FGA using the nonparametric Mann-Whitney U test. A type I error rate of P < 0.05 was used.
A receiver operating characteristic (ROC) curve was used to identify the level of disability reduction associated with the MDC95. The ROC plots the sensitivity versus 1-specificity and shows the trade-off between true-positive and false-positive errors as each of several cut points in the change scores for the DHI and ABC is considered. The ROC curve allowed determination of the sensitivity and specificity for DHI and ABC change in classifying patients who had improved gait performance by the MDC95. The area under the curve (AUC) calculates the ability of the change scores in the DHI and ABC scores to discriminate between improving and nonimproving patients per the DGI/FGA. An omnibus nonparametric statistical comparison of the AUC under both curves against the null hypothesis of AUC = 0.50 for the DHI and ABC change scores was computed for both the DGI and the FGA. Values of improvement in DHI and ABC scores were identified at the point the DHI/ABC changed that maximized the sum of sensitivity and specificity.
Mean baseline and discharge DGI and FGA and change scores are presented in Table 1. Analysis of covariance demonstrated that there was no significant effect of age, symptom duration, or sex on mean changes in DGI, FGA, DHI, or ABC scores between initiation of care and discharge.
The SRM as an index of responsiveness was greater for the FGA (1.25) than for the DGI (0.72). Both measures demonstrated good internal consistency with baseline measures. The amount of pre- to posttreatment change that exceeds chance variation was estimated at 4 points for the DGI and 6 points for the FGA. This degree of change from baseline to discharge was met or exceeded by 42.5% of subjects on the DGI and 47.2% on the FGA. A maximum score was demonstrated by 52% on the DGI (score, 23-24) and 25% on the FGA (score, 29-30), indicating that the DGI demonstrates a greater possibility for a ceiling effect during treatment than the FGA.
Subjects who scored 23 or 24 on the DGI at discharge evaluation had significantly lower discharge mean DHI (28 ± 23) and higher ABC (76 ± 19) scores compared with those who did not achieve the maximum DGI scores (P < 0.001). Similarly, subjects who attained a final FGA score of 29 or 30 had significantly lower discharge mean DHI (26 ± 22) and higher ABC (78 ± 21) scores compared with those who did not achieve the maximum FGA scores (P < 0.007).
Subjects who exceeded the MDC95 on the DGI and the FGA demonstrated significantly greater improvements in self-reported disability as measured by the DHI and ABC than those who did not exceed the MDC95 (Table 1). Changes by at least the MDC95 in both the DGI and the FGA were significantly associated with improvements in self-reported disability as measured by the ABC and DHI.
Analysis of area under the ROC curve showed that changes in both the DHI (AUC = 0.61; 95% confidence interval [CI], 0.54-0.68) and the ABC (AUC = 0.66; 95% CI, 0.59-0.73) identified subjects who made at least a 6-point change in the FGA (Table 2). An improvement in DHI of at least 13 points best identified subjects who made an FGA change of at least 6 points (sensitivity = 50%; specificity = 70%). The point of optimal discrimination in ABC change was 12 points or more (sensitivity = 53%; specificity = 63%).
Changes in both the DHI (AUC = 0.64; 95% CI, 0.57-0.71) and the ABC (AUC = 0.68; 95% CI, 0.61-0.75) identified subjects who made at least a 4-point change in the DGI (Table 2). A DHI change of 11 points or more optimally identified subjects making a 4-point change in the DGI (sensitivity = 61%; specificity = 66%). An improvement of at least 10 points in ABC was associated with at least 4-point change in the DGI (sensitivity = 63%; specificity = 69%).
A subgroup analysis by quartiles, as suggested by Dye et al,16 revealed that persons with greater gait challenges at the start of care improved more on both the DGI and the FGA than those who are closer to the ceiling when the gait tests are first utilized (Figures 1 and 2). Mean changes in DGI scores were significantly different between all quartile groups (P < 0.002) except between those in the second and third quartiles of initial scores (P = 0.19). Mean changes in FGA scores differed significantly between all of the quartile groups (P < 0.026) except between the first and second quartiles (P = 0.534), and between the second and third quartiles (P = .109) of initial scores.
The results of our analysis indicate that the minimum detectable change in the FGA is 6 points and 4 points for the DGI in persons with balance and vestibular disorders. On the basis of the SRM, the FGA seems to be more responsive to changes because of treatment than the DGI. Subjects who change by at least the MDC95 on both the DGI and the FGA demonstrate greater improvements in self-reported disability as measured by the DHI and ABC. Improvements in both measures of self-reported disability discriminated patients who improved by at least the MDC95 on the FGA and DGI.
A change of at least 6 points on the FGA was associated with at least a 13-point improvement in the DHI and a 12-point improvement in the ABC. A change of at least 4 points on the DGI was associated with at least an 11-point improvement in the DHI and a 10-point change in the ABC. The MDCs on the DGI and FGA are associated with a relatively smaller improvement in perceived dizziness handicap as measured by the DHI.11 Significant change in balance confidence has not been established in this study population. However, Talley et al21 have reported a smaller SRM (0.05) for the ABC and a moderate correlation with measures of gait function (timed up-and-go, gait speed) in older adult females in a fall prevention program. These finding suggest that significant improvements in self-perceived limitations are complex and require more than meaningful clinical changes in gait function.
Dye et al16 suggest that the DGI has a considerable ceiling effect in community-living older adults who participated in a balance and gait program, especially for those who had higher scores at the onset of the exercise program. In addition, they suggest that more difficult gait tasks may need to be designed for older adults, as there seems to be little possibility of improvement if participants begin with scores closer to the maximum at the beginning of an exercise program. A 2.9-point change on the DGI has recently been reported as the MDC for adults over the age of 65 years with a history of falls or near falls.22 The mean age of our sample was 58 years, which may have accounted for the difference in MDC values. Fifty-two percent of our subjects achieved the maximum score on the DGI, leaving no potential to note additional improvement beyond the optimal score of 24/24. Even though subjects maximized their DGI scores, their DHI and/or ABC scores were not optimized. Gait, dizziness, and perceived balance confidence are related but not 100% correlated.23,24
The goal of the FGA was to develop a tool that was more difficult for the older adult.5 The attainment of this goal was observed in our subjects because fewer subjects reached the optimal FGA score at the end of rehabilitation compared with their scores on the DGI. The mean score of 18/24 on the DGI versus 19/30 by the participants on the FGA at the onset of care suggests that the patient has greater opportunity to change on the FGA versus the DGI, which is evident on the basis of reviewing Figures 1 and 2. The FGA demonstrates a smaller percentage of subjects achieving the maximal score. Therefore, when choosing a gait assessment tool, it is imperative that scores associated with the tool have potential for change. It seems that for some persons with balance and vestibular disorders, regardless of age, diagnosis, or sex, more difficult gait assessment items may be necessary to increase sensitivity to change over time.
The range of abilities measured by the gait measures may be insufficient to capture the skills required to be fully functional (ie, the tests are too easy). It is also possible that the psychological constructs underlying self-perceived limitations are not fully addressed even though gait function has normalized. The domains underlying self-efficacy may be more complex and may not improve in a linear fashion with respect to changes in gait performance. Having identified the MDCs for the DGI and FGA in persons with balance and vestibular disorders, it is hoped that clinicians can better determine when patients have truly “changed” from the provided interventions. If the patient comes close to maximizing their score on the DGI at the start of the episode of care, use of the FGA might be indicated to better identify meaningful change.
The DGI and the FGA are responsive to change over time in persons with balance and vestibular disorders. Significant changes in both gait performance measures are associated with smaller but potentially meaningful reductions in self-reported disability. Both measures seem to be responsive to change in patients with lower levels of initial gait and balance performance. The FGA is less subject to ceiling effects in patients who may still be experiencing limitations compared with the DGI, and may be a better choice of clinical assessment to measure gait performance in patients with higher levels of function. The ceiling effect rate in both measures suggests that measures with more complex gait tasks may be needed to increase test sensitivity, particularly for testing of performance in patients with higher levels of function who still perceive functional limitations.
1. Studenski S, Perera S, Patel K, et al. Gait speed and survival in older adults. JAMA. 2011; 305:(1):50–58.
2. Whitney SL, Hudak MT, Marchetti GF. The dynamic gait index relates to self-reported fall history in individuals with vestibular dysfunction. J Vestib Res. 2000; 10:(2):99–105.
3. Hall CD, Herdman SJ. Reliability of clinical measures used to assess patients with peripheral vestibular disorders. J Neurol Phys Ther. 2006; 30:(2):74–81.
4. Shumway-Cook A, Woollacott M. Motor Control: Theory and Practical Applications. 1st ed. Baltimore: Williams & Wilkins; 1995; .
5. Wrisley DM, Marchetti GF, Kuharsky DK, Whitney SL. Reliability, internal consistency, and validity of data obtained with the functional gait assessment. Phys Ther. 2004; 84:(10):906–918.
6. Wrisley DM, Kumar NA. Functional gait assessment: concurrent, discriminative, and predictive validity in community-dwelling older adults. Phys Ther. 2010; 90:(5):761–773.
7. Stratford PW, Binkley JM, Riddle DL. Health status measures: strategies and analytic methods for assessing change scores. Phys Ther. 1996; 76:(10):1109–1123.
8. Finch E, Brooks D, Stratford PW, Mayo NE. Hamilton: Canadian Physiotherapy Association. Physical Rehabilitation Outcome Measures: A Guide to Enhanced Clinical Decision Making. 2nd ed. Hamilton, CA: Canadian Physiotherapy Association, BC Decker Inc; 2002; .
9. Stratford PW, Binkley J, Solomon P, Finch E, Gill C, Moreland J. Defining the minimum level of detectable change for the Roland-Morris questionnaire. Phys Ther. 1996; 76:(4):359–365; discussion 366–358.
10. Powell LE, Myers AM. The Activities-Specific Balance Confidence (ABC) Scale. J Gerontol A Biol Sci Med Sci. 1995; 50A:(1):M28–M34.
11. Jacobson GP, Newman CW. The development of the Dizziness Handicap Inventory. Arch Otolaryngol Head Neck Surg. 1990; 116:(4):424–427.
12. Herdman SJ, Blatt P, Schubert MC, Tusa RJ. Falls in patients with vestibular deficits. Am J Otol. 2000; 21:(6):847–851.
13. Bishop MD, Patterson TS, Romero S, Light KE. Improved fall-related efficacy in older adults related to changes in dynamic gait ability. Phys Ther. 2010; 90:(11):1598–1606.
14. Whitney S, Wrisley D, Furman J. Concurrent validity of the Berg Balance Scale and the Dynamic Gait Index in people with vestibular dysfunction. Physiother Res Int. 2003; 8:(4):178–186.
15. Wrisley DM, Walker ML, Echternach JL, Strasnick B. Reliability of the dynamic gait index in people with vestibular disorders. Arch Phys Med Rehabil. 2003; 84:(10):1528–1533.
16. Dye DC, Eakman AM, Bolton KM. Assessing the validity of the dynamic gait index in a balance disorders clinic: an application of Rasch analysis. Phys Ther. 2013; 93:(6):809–818.
17. Jacobson GP, Newman CW, Hunter L, Blazer GK. Balance function test correlates of the dizziness handicap inventory. J Am Acad Audiol. 1991; 2:253–260.
18. Myers AM, Fletcher PC, Myers AH. Discriminative and evaluative properties of the activities-specific balance confidence (ABC) scale. J Gerontol. 1998; 53A:M287–M294.
19. Lajoie Y, Gallagher SP. Predicting falls within the elderly community: comparison of postural sway, reaction time, the Berg balance scale and the Activities-specific Balance Confidence (ABC) scale for comparing fallers and non-fallers. Arch Gerontol Geriatr. 2004; 38:(1):11–26.
20. Haley SM, Fragala-Pinkham MA. Interpreting change scores of tests and measures used in physical therapy. Phys Ther. 2006; 86:(5):735–743.
21. Talley KM, Wyman JF, Gross CR. Psychometric properties of the activities-specific balance confidence scale and the survey of activities and fear of falling in older women. J Am Geriatr Soc. 2008; 56:(2):328–333.
22. Romero S, Bishop MD, Velozo CA, Light K. Minimum detectable change of the Berg Balance Scale and Dynamic Gait Index in older persons at risk for falling. J Geriatr Phys Ther. 2011; 34:(3):131–137.
23. Whitney SL, Wrisley DM, Brown KE, Furman JM. Is perception of handicap related to functional performance in persons with vestibular dysfunction? Otol Neurotol. 2004; 25:(2):139–143.
24. Legters K, Whitney SL, Porter R, Buczek F. The relationship between the Activities-Specific Balance Confidence Scale and the Dynamic Gait Index in peripheral vestibular dysfunction. Physiother Res Int. 2005; 10:(1):10–22.
© 2014 Neurology Section, APTA.