Investigating the clinical pattern of motor relearning in people with multiple sclerosis.

The two treatment groups were not randomised and the assessors were not blinded to group allocation or to earlier results, so the difference in learning speed between balance-specific and non-specific training should be read with caution. Scoring the same Berg Balance Scale items every session may have produced practice effects, so people may have got better at the test rather than at balancing. No sample-size calculation was done beforehand, so a larger study could give different results. The Berg Balance Scale has ceiling effects, does not assess dynamic balance in depth, and is measured on an ordinal scale. The model describing session-to-session variability fitted the data poorly and none of its individual parameter estimates were statistically significant. The group comparison models only converged after the within-subject correlation structure was removed, which the authors say may have introduced bias. The study cannot show which sensory, motor or cognitive processes drive the improvement; that would need physiological or imaging evidence.

The authors declare no conflict of interest. The work was supported by the Italian Ministry of Health (Ricerca Corrente).

Do not take this as proof that balance-specific training is better than non-specific training. The two groups were not randomly assigned, the assessors were not blinded, and the authors themselves state the study cannot say which option is most effective for improving balance; only a faster rate of change was seen in the specific group.