Skip to main content

Table 4 Summary table of the Berg Balance Scale test as a falls assessment tool

From: Predicting falls in older adults: an umbrella review of instruments assessing gait, balance, and functional mobility

Review Review characteristics Risk of bias Summary of key findings Interpretation
Lee [29] Systematic review without meta-analysis
Study (n = 4)
Mixed settings
High Community dwelling older people (n = 1):
- 61% sensitivity, 53% specificity, 49% positive predictive value, AUC = 0.59
Outpatient stroke clinics (n = 3):
- 69% sensitivity, 65% specificity, 64% positive predictive value, 70% negative predictive value, AUC = 0.69 (0.58–0.80)
- 85% sensitivity, 49% specificity, 55% positive predictive value, 83% negative predictive value
Cut off < 49, 83% specificity, 91% specificity, 71% positive predictive value, 95% negative predictive value
Favourable for outpatient stroke population
Nakamura [20] Narrative review
No details on characteristics
High No data to extract Unclear
Neuls [33] Systematic review without meta-analysis
(n = 9)
4 studies with adults with neurological disorders
High Sensitivity ranges from 25% t0 95.5%
Specificity ranged from 20.8 to 100%
Calculated cut-off scores ranging from 33 to 54.
Not favourable
Park [12] Meta-analysis
(n = 5, 427 participants)
Community-dwelling
High Pooled sensitivity was 0.73 (95% CI 0.65–0.79). Heterogeneity among studies was high (82.7%; χ2 = 23.09, P = .0001).
Pooled specificity was 0.90 (95% CI 0.86–0.93), and heterogeneity among articles was low (31.9%; χ2 = 5.87, P = .21).
sROC AUC was 0.97 (standard error [SE] = 0.02)
Favourable
Scott [35] Systematic review without meta-analysis
(n = 4)
Mixed settings
High Community (n = 3):
- reported in one study as 53% sensitivity and 96% specificity
Supportive housing (n = 1):
- significant predictor with score < 45 indicating a relative risk for multiple falls over the next 12 months.
Acute: no data to extract
Inconsistent
Dolatabadi [26] Systematic review without meta-analysis
(n = 1)
Older adults with diagnosis of dementia
Unclear One study reported on BBS and no significant findings reported. Not favourable
Lusardi [9] Meta-analysis
(n = 4)
Community-dwelling
Unclear BBS < 50 points, positive likelihood ratio 3.4, negative likelihood ratio 0.7, posttest probability with a positive test 59%, posttest probability with a negative test 23%.
Sensitivity 41% and specificity 88%
Favourable
Lima [11] Systematic review without meta-analysis
(n = 8)
Mixed settings
Low BBS low to moderate sensitivity achieving its best value of 67% for 6-month using a cut-off score of 45 points for any falls, and 69% for 12-month follow-up, using a cut off score of 53 points for multiple falls. Not Favourable
Muir [42] Meta-analysis
(n = 1*)
Community-dwelling
Low One study with non-significant results on fall prediction, meta-analysis not completed for this measure. Not favourable
  1. Abbreviations: AUC Area under the curve, BBS Berg Balance Scale, CI Confidence interval, n number of included studies, OR Odds ratio, SROC Summary receiver operating characteristic. *This study did meta-analyses, but not on BBS, which was only reported in one paper