Disability and Rehabilication, December 2007; 29(24) 920-1925 forma RESEARCH PAPER Reliability of four scales on balance disorders in persons with multiple sclerosis DAVIDE CATTANEO, JOHANNA JONSDOTTIR & STEFANIA REPETTI LaRICE: Gait and Balance Disorders Laboratory, Department of Neurorchabilitation, Don Gnocchi Foundation ILRC.O.S., Milan, Iealy Accepted December 2006 Abstract Purpose. Balance rehabilitation is an important component of the retraining program in people sith multiple sclerosis (MS). Measuring balance is fundamental for an accurate assessment and therapy selection, ‘The aim of this study was to ‘Compare interrater and test-retest reliability of the Berg Balance Scale, the Dynamic Gait Index, the Dizziness Handicap Inventory and the Activities-specific Balance Confidence. Method. A group of 25 persons were enrolled in the study. The group consisted of § males and 17 females, mean age ‘41.7 years (12.5 years, SD). The onset of pathology was 8.7 years (8.8 years SD) before the beginning ofthe stuly. To assess, the testeretest reliability nwo consecutive assessments were collected by the same rater. ‘To assess the interrater reliability persons were concurrently assessed by two raters. Results. The Intraclass Correlation Coefficients (ICCs) for interrater reliability ranged berween 0.94 and 0.96, The ICCs for test-retest reliability ranged between 0.85 and 0.96. Conciuion. The interrater reliability of the instruments proved to be satisfactory. Lower but acceptable results were obtained for the test-retest paradigm. ‘The data obtained in this study suggest that these scales are reliable tools for assessing balance function in persons suffering from MS, Keywords: Reliabiliny, balance, rehabilitation, multiple sclerosis Introduction Balance rehabilitation is an important component of the retraining programme in people with multiple sclerosis (MS) [1-3]. Measuring balance is funda- ‘mental to accurate assessments, appropriate therapy selection and the measurement of outcomes, A variety of laboratory techniques and clinical scales have been proposed [4], however, clinical scales are the instruments most commonly used in clinical settings. They provide insight for the plan- ning of rehabilitation programme, are less expensive than laboratory assessments and easily applied in the clinical setting. In addition to scales measuring balance perfor- mance, the use of scales measuring the person's perception of balance and the behavioural conse- quences of balance disorders are important to assess the impact of balance disorders on activities of daily life. The Berg Balance Scale (BBS) [5], the Dynamic Gait Index (DGD) [6], the Dizziness Handicap Inventory (DHI [7] and the Activities-specific Balance Confidence (ABC) [8] have gained popu- larity within the clinical and scientific community. ‘The BBS and the DGI are clinical scales that ‘evaluate balance in standing and during gait activities respectively. They are well known among phy- siotherapists and allow comparisons among groups ‘of persons with various pathologies. The ABC and DHT are questionnaires that measure perception of balance in activites of daily life (ADL) and percep- tion of the level of handicap. Information gained from the two scales measuring balance perception ‘may better account for reduction of daily activities than counting number of falls or measuring clinical balance performance. Taking into account ‘Corespondence: Divide Cataneo. LaRICE: Gait and Balance DsordesLaborston, Seo Ritazione Nearoogca Adult lt. 282); Don Gnowsh Foundaion IRCCS. V. Capea 66 ~ 20145 Mun, lay. Tu 39240305282, Fx: 59240508198. Email dstancordongnchitbehavioural aspects is important for addressing the person's beliefs about his ability to carry out activities of daily life and tuning the person’s behaviour to his actual level of impairment and disability. Informa- tion from all four instruments would make it possible to develop customized rehabilitation programme taking into account both physical impairments and the persons’ own perception of balance, as well as, behavioural consequences of both. To be effective for clinical and/or research pur- poses the scales have to show good performance in three psychometric characteristics: Validity, reliabil- ity and responsiveness. Validity of these scales for a population with MS has been reported [9] but no study has reported their reliability for a population with MS, ‘The aim of this study was to compare reliability of the four balance scales in a cohort of persons suffering from MS. Method Subjects Forty-five persons suffering from MS attending the Department of Multiple Sclerosis of Don Gnocchi Foundation were assessed. ‘To be eligible for the study, the persons had to meet the following inclusion criteria: Clinically or labora- tory definite relapsing-remitting or secondary pro- gressive multiple sclerosis, ability to stand independently in upright position more than 30 sec, ability to walk for 6 m with or without an assistive Cognitive impairment of all persons had been clinically assessed a priori by the medical staff. Persons with cognitive impairments that might hinder understanding of the tasks to be accomplished were excluded from enrolment. A sub-group of 25 persons ‘met the inclusion-exclusion criteria and were enrolled in the study. The group consisted of 8 males and 17 females, mean age 41.7 years (12.5 years SD). The onset of pathology was 8.7 years (8.8 years SD) before the beginning of the study. Four persons used a walking aid in their daily activities. After informed consent was obtained, persons completed a ques- tionnaire providing information about theit age, the onset of pathology and other demographic character- istics. Then the battery of tests was administered, To assess the test-retest reliability, ewo consecutive assessments were collected by the same rater, To avoid influences due to change in the pharmacolo- gical and physiotherapy treatments and behavioural consequences of hospital stay, the data were collected three days apart at the beginning of the hospital stay; no rehabilitative sessions, drugs or invasive exams were provided between the first and the second assessment, Self-administered tests were evict Balance assessment in MS 1921 collected at both assessments. To reduce the person’s recall of the first assessment of the scores assigned to self-administered tests, the order of the items of the second test was varied. The tests were divided in to two parts and the second half was administered first and then the first half, To assess the inter-rater reliability of the BBS and DGI scales persons were concurrently assessed by two raters who were physiotherapists with long experience in neurorehabilitation. Inter-rater relia- bility for the ABC and the DHI was not assessed because the scales are self-administered, “The assessment was carried out in one session, the testing protocol took 3040 min to be administered, persons were allowed to rest during testing if necessary. Persons were tested wearing their normal shoes, Because five of the auto-administered tests were not completed the sample for these scales consisted in 20 subjects, Instruments Berg Balance Scale (BBS). This scale rates perfor- ‘mance from 0 (cannot perform) t© 4 (normal performance) on 14 items. The items explore the ability to sit, stand, lean, cum and maintain the upright position on one leg [5]. The psychometric properties of the scale have been assessed on a population of elderly subjects. For that population of subjects, the scale proved to be a valid and reliable instrument, In another study on elderly subjects the test retest and interrater reliability of BBS was found to be very high with the ICCs ranging from 0.98-0.99 for intrarater reliability and 0.98 for interrater reliability [10]. In our previous study on the validity of the BBS for subjects with MS the seale ‘was found to have good concurrent validity but not a good discriminant validity, that is, because of a low level of sensivity it did not distinguish well between fallers and non-fallers [9]. However, a cut-off score of 44 was established as a criterion to identify MS subjects with a high risk of fall based on results in the literature [9]. Performance on both the BBS and the DGI was rated by the first two authors (DC and JD). Dynamic Gait Index (DGD. This scale measures mobility function and dynamic balance, The eight tasks of this scale include walking, walking with head tums, pivoting, walking over objects, walking around objects and going up stairs [6,11]. The performance is rated on a 4-point scale. McConvey and Bennett [12] found an inverse correlation of 0.80 between the scores of the DGI and the time to walk 6.1 m in persons with MS. Data on reliability were published in the same study using videotaped recordings of the person’s performances. The study showed an1922, Cattaneo et al. interrater reliability of 0.98 and test-retest reliability (Pearson's correlation coefficients) ranging fom 0.76-0.99. However, their study did not permit an evaluation of day-to-day variability of performance which is one of the reasons the present study was undertaken, Aaivities-specific Balance Gonfidence (ABC). This is a scale in which the person rates his self-perceived level of confidence while performing 16 daily living activities (8). In a study by Miller et al. [13] the test- retest reliability of the ABC Scale among people who have a lower-limb amputation was 0.91. Moreover, correlations with the 2 Meters Walking test and ‘Timed Up and Go test were observed to be 0.72 and —0.70 [13]. Another study that aimed to assess convergent validity between the ABC Scale and Falls Efficacy Scale (FES) in a population of elderly people indicated that the ABC and FES were highly correlated [14]. Dizziness Handicap Inventory (DHD. This scale is ‘multidimensional self-assessment scale that quanti- fies the level of disability and handicap in three subscales: physical, emotional and functional {7}. Scores range from 0-100; where 100 means high level of disability and handicap. DHI demonstrated good intemal consistency (0.91) and rest retest reliability (0.97) [15]. Since in the other tests evaluated here higher scores mean better perfor- mances, the scores of the DHI were transformed: DHL_MODIFIED — 100-DHI, so 100 points indi- cates a high level of performance. Dasa analysis ‘Total scores of the tests were used to compute absolute agreement Intraclass Correlation Coetfi- cients (C3), using a two-way, random ellects ‘model. In that way raters are conceived as being a random selection from among all possible raters, and subjects are conceived as being a random factor too [16]. Standard errors of measurements for total scores were calculated for test-retest and interrater reliability [17]. Standard error of measurement of a test (for example in a test-retest condition), refers to the variability of test scores that would have been obtained from a single person had that person been tested multiple times. Change scores higher than 2#SEM points (for example 3 points in the BBS) cannot be due only to exposition to two consecutive assessments, Bland-Altman scatter plots [18] were used to illustrate the consistency of measurements and the presence of tends. Since itis possible thatthe exposition to two con~ secutive executions of the test might improve sub- jects" performance, we compared the first baseline assessment with the retest assessment to verify the presence of a leaming effect, Results ‘Three persons reported | fall and 1 person reported 10 falls in the month before the beginning of the study; the mean number of falls was 0.68 (2.29 SD), Table I shows the test scores for the whole group. ‘The group of subjects reported mild impairment in static balance (on the average reaching 84% of the ‘maximum score of BBS scale) and more pronounced impairment in dynamic balance (on the average reaching 66% of the maximum score of DGI scale), ‘The self-administered tests showed moderate (about 50% of the maximum score of the scales) confidence of persons in their balance skills. ‘The effect of two consecutive assessments ac- counted for less than 1 point of improvement on the retest both on the BBS and the DGI scales, those improvements corresponded to a 1.6% and 3.5% of improvement respectively for BBS and DGI. Self- administered scales did not show a learning effect. ‘The scores decreased 4 points for the ABC scale and 1 points for the DHI scale, corresponding to a 4% and 1% of decrement respectively for ABC and DHI. ‘Table II shows the Intraclass Correlation Coefli- cients for the testretest condition. The highest coeflicient was observed for the scale rating static balance (BBS), conversely for the DGI the coeffi- cient was lower. No relevant differences of ICC values were observed between self-administered scales. ‘Table ITI shows the Intraclass Correlation Coeffi- cients for the interrater condition. In contrast to the test-retest condition where ICCs of BBS and DGI were different in the interrater condition the tests Table 1. Destine satis foe the whole group of perms Deserve stat ” Min Mex Men SD SEM BRS-te 2531564727772 EES—Im 531364792751 BBS-ret 25285648. 765 La DGI-te 25333 15.96 Glin 25 3285.90 Dol-wt 25733 16.8058 ABC—bas 20999 56.08 25.20 ABC—ret 2014 9952.94 2674 TAH DH—be 2026100 55.10 22.48 DHI-ree 20221005440 2605 8.12 BBS, Berg Balance Scale; DGL, Dynamic Gait Index: ABC ‘Actviy-specifc Balance Confidence; DHT, Dizaers Handicap Inventory. Bas, baseline assessment, fst rater, Int assessment of second rater; Ret, second assessment of the fist rater. SEM, standard error of measorements‘Tale Ml. Intraciss Correlation Cooficients for the test retest condition, Tevcretet C ce a ns 0.96 oot 0.98 Dol 085 ot 093 ABC 092 080 097 pat 0.90 a7 0.96 BBS, Borg Balance Scale; DGL, Dynamic Gait Index; ABC, Actiiy-speciic Balance Confidence; DHT, Digeness Handicap Inventory. ICC, Intralass Correlation Coeficiets Cl, 95% cnnfidence interval ‘Tale TL, Ineraclass Correlation Coeiclents for the interrater condition, = rec ctr a BBS 0.96 0.90 097 Dol 08 ose 097 BBS, Berg Balance Seale; DG, Dynamic Gait Index; ICC, Inustdass Corelation Covfficint; Cl, 95% confidence interval. showed similar coellicients, Note the differences in ICC values between interrater and test-retest condi- tions for the DGI scale, Bland-Altman scatter plots of the tests in the interrater and test-retest conditions are depicted in Figure 1 ‘A visual evaluation of Bland-Altman scatter plot of the BBS demonstrated an apparently better reliability at the higher level of scores in the interrater condition as demonstrated by a closer position of the points to the X axis; the same was true for the ABC scale in the test-retest condition. No other trends were detect- able for the other tests in the two experimental conditions. ‘Standard errors of measurements, calculated from ‘mean scores, are shown in Table I. For the BBS and DGI scale the SEMs were in the range of 1.5-2 points, For self-administered scales the SEMs were in the range of 7-8 points, Discussion ‘The nim of this study was to assess interrater and test-retest reliability of 4 scales tapping balance disorders in MS population. ‘The BBS showed good pesformances both in terms of interrater and test-retest reliability. The ICC values proved to be satisfactory though lower than those of elderly adults [19] and traumatic brain injury persons [20], Comparison with persons with Balance assessment in MS 1923 post stroke hemiplegia is dificult because of different statistical indices used [21]. Ceiling effects may have influenced the ICC values in a couple of ways. These effects may have decreased the amount of true variance thus leading to a reduction of the ICC values, or they may have improved the ICC values because higher agreement is observed at high scores of the scale as apparent in the Bland-Altman scatter- plots (see Figure 1). Both the DGI scale and the BBS showed very good performance in terms of interrater reliability ‘This outcome is surprising since in general the BBS hhas a more accurate definition of items with time intervals to discriminate berween the levels of each item. The BBS also requires fewer judgements con- ceming the ‘normality’ of performances. The varia- bility of gait performances on the DGI between sessions was demonstrated by a decrease of the ICC value in the test-retest condition, In our opinion this variability was due to an inconsistency of perfor- ‘mances rather than to inconsistency of the rater. ICG values obtained in this study were lower than the data obtained by McConvey and Bennett [12] who reported an ICC value of 0.98 for inter- rater reliability. The differences in scores were likely due to different experimental procedures. since ‘McConvey used video recorded performances, thus nullifying the variance due to time effect. Because of the different statistical indices used, the comparisons with other studies were difficult [22]. ‘The self-administered tests had similar ICC values. The ICCs for the test-retest condition were 0.92. and 0.90 for the ABC and the DHT test respectively. The ICC values obtained in this study for the ABC scale were similar to those obtained for persons with lower limb amputation [13] but higher than those obtained by Botner etal. for persons with, stroke [23]. Scores seemed less stable for lower levels of perceived balance impairment as apparent from the Bland-Altman plots. No such tends were detected for the DHI test. For the BBS and DGI scale the slight improve~ ‘ment in scores from the first to the second con- secutive assessment was not clinically relevant, In contrast to tests based on performance, self adminis- tered scales showed a small reduction of values from first to second assessment. SEMS refer to the variability of test scores obtained from a single person had that person been tested ‘multiple times. The SEMs were quite low among the exception of the DGI in the retest paradigm where the level of uncertainty was 4 points, fn error that could be clinically relevant. ‘The variability of symptoms reported by persons with MS may have accounted for a small increase ‘of variability in the clinical balance tests when compared to persons with lower limb amputation1924 _D. Cattaneo et al. 8 -nterrter- 05 -Testestast- 10 10 3° . 1 o @ Bo Fo Bo pip » wa oe Fob nm » @ Rta a : ° 40 40 Meare Moana at nter rater 01 Tostetet- 0 0 ° ge +. oe Bole go mo * tt ge B. 5 i sh OS SuxMos 0 2% “0 “0 Ac -Testeetest- it -Testetet- 6 Co 20 ” Bn ] | an boc be, 0% 4 3» Enf Dg ra % o 0 ao “0 “ Figure 1. Bland-Aluman scatter plots ofthe tests inthe interrater and txe-retest conditions, BBS, Berg Balance Scale; DGI, Dynamic Gait Index; ABC, Actvi-specifie Bilance Confidence; DHI, Dizaness Handicap Inventory. or stroke. This variability in symptoms may be the cause of the higher ICs in the interrater condition than the test-retest condition, A limitation of this study was the small number of, persons enrolled and the clinical characteristics of the group that showed only mild to moderate balance impairments. The data should thus be generalized with caution to persons with different levels of impairment, ‘The raters were experienced neurological phy- siotherapists, the same level of reliability cannot be assumed with less experienced physiotherapists or other health care professions. A stronger evaluation of interrater reliability could have been gained with more raters having a wider range of expertise or with the adoption of more complex studies (¢.g., D studies) to assess the reliability of an assessment carried out by different therapists in different occasions, for example wo consecutive assessments rated by two different therapists [17]. The historical effects on the performance of the tests could be better evaluated by the comparison of videotaped performances. A better description of several items and more accurate definitions of the scoring system within each item would further improve the reliability of those scales. Conclusion, ‘The results of this study support the concept that the test scores for persons suflering from multiple sclerosis are not less reliable with respect to other pathologies. The inter-rater reliability of the instruments proved to be satisfactory. Lower but acceptable results were obtained for the test-retestparadigm suggesting that these scales are reliable tools for assessing aspects of balance function in persons suflering from MS. ‘We assessed the test-retest and inter-rater relia- bility of a couple of scales that measure balance performance and a couple of scales that measure the perception of balance, both types of scales demon- strated excellent reliability when applied to persons with MS. The comparison between balance perfor- ‘mance, person’s balance perception and the assess- ment of subject’s behaviours in the activities of daily life may improve the estimation of fall risk and allow the implementation of a more tailored treatment plan, References 1. Lod SE, Wade DT, Halligan PW. A comparison of avo physiotherapy treatment approaches to improve walking in ‘multiple sclerosis: A pilor randomized controlled stad. Clin [Rehabil 1998;12:477 ~ 86. 2, Wiles CM, Nevecombe RG, Faller KJ, Shaw S, Furnival- Doran J, Pckersgil TP, Morgan A. Controlled randomised crossover tial ofthe effets of physiotherapy on mobility in chronic multiple sclerosis. J Neurot Neurosurg Psychiouy 2001;70:174-179, 3. Armudu K, Karabudak R, Nusa G. Physiotherapy ap. proaches in the treatment of atte mainte sleross: A pilot seady, Neurorchabil Neural Rept 2001;15:203~211 4. Berg KO, Maki BE, Willams Jl, Holiday Pf, Wood- Daaphinee SL. Clinical and Inborstory messes of postr balance in an elderly population. Arch Phys Med Rehabil 1992573:1073 1080, 5. Berg KO, Wood-Dauphinee SL, Williams Jl, Gayton D. ‘Measuring balance inthe eldery: Preliminary development of an inswument. Physiouher Canada 1989;41:304—311 6. Whitey SE, Hudsk MT, Marchets GF. The dynamic gut index relates t sel reported fall history in individuals with ‘esfbular dysfunction. J Vesib Res 2000;10:09 105, 7. Jason GP, Newman CW. The development of the Divziness Handicap Inventory. Arch Otolaryngol Flea! Neck Surg 1900;116424— 427. 8, Powell LE, Myers AM, The Actvtesspecific Balance Contidence’ (ABC) Seale. J Gerontol A Biol Sci Med Sci 1995508: M283, 9. Cattaneo D, Regola A, Meotd M. V fonts seales in persons with multiple. sclerosis. Rehabil 2006 15:28:78 795, diy of six balance is Disabil 10. a 12, 1. 14 15. 16. 1. 18. vs. 2, En 2 Balance assessment in MS 1925 Miyamoto ST, Lombardi Junior I, Berg KO, Ramos LR, Natour J. Bewilian version of the Berg Balance Seal. Brae J (Med Biol Res 2008557:1411~ 1421 ‘Whiney S, Weitley D, Furman J. Concurent validity of the Berg Balance Scale and the Dynamic Gait Index in people with vestibular dysfunction, Physiother Res Int. 20038 178-186, McConvey J, Bennett SE. Reliability of the Dynamic Gait Tex in individuals with multiple selerois. Arch Phys Med Rehabil 2005:80:130- 133. Miller WC, Deathe AB, Speschley M. Psychometic pr erties of the Activitiesspectic Balance Confidence Seale mong individuals witha lower-lim’ amputation. Arch Phys ‘Med Rehabil 2003:84:656-661 Howhkiss A, Fisher A, Roberson R, Ruuencuter A, Schuller J, Basker DB. Convergent and predictive validiy of thoe scale related ofl in the elders. Am J Occup Ther 2004:58:100—103, Jacobson GP, Newman CW. The development of the Divziness Hanlicap Inventory. Arch Ocolaryngol Head Neck Surg. 1990511624427, Rankin G, Stokes M. Reliability of assessment tools in rehabilitation: An illustration of appropiate stattial ana Iyses. Clin Rehabil 1993:12:187—199, Suciner DL, Norman GR. Healdh measurement scales A practical guide to their development and use. 2nd ed ‘New York! Oxford University Press, 1996 Bland JM, Akman DG. Stasis! methods for assessing agreement benvean evo methods of dinical measurement Lanoet 19861307 ~310. Miyamoto. ST, Lombardi Junior 1, Berg KO, Ramos LR, Natous J Bewzlian version of the Berg Balance Seale Braz J (Med Biol Res 2008;37:1411— 1421 Newstead AH, Hinman MR, Tombedin JA. Relibily ofthe Berg Balance Seale and balance master ims of stably ets for individuals with brain injury. J Neurol Phys Ther 2005 Mars20:18-23, Tyson SF, DeSousa LH. Reliability and validity of fanc- sional balance tests poststroke. Clin Rehabil 2004sDeci18 915-92. Wrisley DM, Walker ME, Fchtemach Jl, Swasnick B. Reliability ofthe dynamic gut index in people with vestibular disorders. Arch Phys Med Rehabil 20033841528 1533, Bower EM, Miller WC, Eng I Measurement properties of the Actviiesspecfic Balance Confidence Scale among individuats with stoke, Distbil Rehabil 2005;18(27) 156-163,