Falls in Persons With Spinal Cord Injury: Validity and Reliability of the Berg Balance Scale

Background. Persons with spinal cord injury who are able to walk are at risk for falls. Objective. The objectives were to investigate if the Berg Balance Scale (BBS) can discriminate those with a propensity to fall; to determine whether the BBS is associated with mobility measures, fear of falling, and muscle strength; and to assess interobserver reliability. Methods. The measurement tools used were the BBS, the Spinal Cord Independence Measure,the Falls Efficacy Scale (FES-I), the Walking Index for Spinal Cord Injury, the 10-m walk test, and the standard neurological classification including motor scores (MS). Falls were recorded retrospectively for the previous month and prospectively for the subsequent 4 months. To determine interobserver reliability, BBS performance was videotaped and analyzed by additional physical therapists. Associations between BBS and the number of falls, measures of mobility, FES-I, and MS were calculated using Spearman correlations. The interobserver reliability was quantified using Kendall's coefficient of concordance and intraclass correlation coefficients (ICCs). Results. Forty-two participants were included of whom 26 sustained I or more falls. BBS performance correlated with measures of mobility, FES-I, and MS (r(s) = -83 to .93; P <.001) but not with the number of falls (r(s) = -.17; P = .28). The interobserver reliability was excellent, both for single items (.84-.98, P < .001) and for the total score (ICC = .95; 95% confidence interval = 0.910-0.975). Conclusions. The BBS proved to be reliable and to relate well with other mobility measures, fear of falling, and muscle strength. However, it was unable discriminate between people who did fall and people who did not fall.