Assessment of Psychometric Properties of Various Balance Assessment Tools in Persons With Cervical Spondylotic Myelopathy

STUDY DESIGN: Clinical measurement, cross-sectional, repeated-measures design. BACKGROUND: Persons with cervical spondylotic myelopathy (CSM) are known to have balance impairments. The psychometric properties of various balance assessment tools have not been evaluated in this population. OBJECTIVE: To examine the floor and ceiling effects, item difficulty and item discrimination index, internal consistency, reliability, and validity of the Balance Evaluation Systems Test (BESTest), Mini-BESTest, Brief BESTest, and Berg Balance Scale (BBS) in persons with CSM. METHODS: Seventy-two individuals with CSM were evaluated with the above balance scales. Thirty-two were assessed by a second rater on the same day to establish interrater reliability, and by the same rater 1 to 2 days later to assess test-retest reliability. RESULTS: Of the 4 balance tools, only the BBS showed a substantial ceiling effect (skewness, gamma(1)<-1.0). The BBS also had the highest proportion (greater than 80%) of easy items. The item discrimination index was greater than 0.4 for all items in the 4 balance scales. All balance tests showed good internal consistency (Cronbach a>0.8), with excellent test-retest (intraclass correlation coefficient [ICC]z1>0.80) and interrater reliability (ICC2.1>0.80). In addition, all balance tests were strongly correlated with one another (ie, concurrent validity), and with the modified Japanese Orthopaedic Association score (ie, convergent validity). Their correlations with the Abbreviated Mental Test were low (ie, discriminant validity). All 4 balance tests could adequately identify fallers and users of mobility aids (area under the curve, >0.8) (ie, known-groups validity). CONCLUSION: The Brief BESTest is the most preferred tool for persons with CSM because of its excellent reliability, validity, and short administration time, whereas the BBS is the least preferred due to its substantial ceiling effect.