Characteristics that affect score reliability in the Berg Balance Scale: a meta-analytic reliability generalization...

Area of Science:

• Rehabilitation Science
• Clinical Measurement
• Psychometrics

Background:

• The Berg Balance Scale (BBS) is a key tool for assessing balance in neurological conditions.
• Scale reliability is context-dependent, necessitating reporting in each study.
• Systematic reviews and meta-analyses are vital for understanding reliability variations.

Purpose of the Study:

• To determine the mean internal consistency, intra-rater, and inter-rater reliability of the BBS across diverse populations.
• To identify methodological and substantive study characteristics influencing BBS reliability.
• To develop a predictive model for estimating BBS reliability based on study features.

Main Methods:

• A systematic review and meta-analysis of studies published between 1989 and 2015.
• Searched MEDLINE (PubMed), EMBASE, and CINAHL databases.
• Included empirical studies reporting any BBS reliability coefficient, with data extracted by two independent reviewers.

Main Results:

• Eighty samples from 65 studies reported BBS reliability.
• Mean coefficient alpha (internal consistency) was 0.92 (range 0.62–0.98).
• Mean intra-rater reliability (ICC+) was 0.957; inter-rater reliability (ICC+) was 0.97.
• Study characteristics like SD of scores, population type, sample size, and mean scale results were associated with reliability coefficients.

Conclusions:

• The Berg Balance Scale exhibits very satisfactory internal consistency and inter-rater reliability.
• Study characteristics significantly influence BBS alpha coefficients and intra-rater reliability.
• Findings support the BBS as a reliable measure, with identified factors aiding interpretation.