In terms of standing, our finding is in contrast to Barclay-Goddard et al (2009) and van Peppen et al (2006) who both reported no effect of biofeedback (force information via visual feedback) on standing, with Berg Balance Scale effects of MD –2, 95% CI –6 to 2 (2 trials) and SMD –0.20, 95% CI –0.79 to 0.39 (2 trials). It is possible that some of the positive effect of biofeedback could Bcl-2 inhibitor be explained by the amount of practice

carried out by the experimental group compared with the control group. When analysing only those trials where the control group practised the same activity for the same amount of time as the experimental group, with the only difference being the substitution of

biofeedback for therapist feedback in the experimental group, the effect of biofeedback was still clinically and statistically significant (SMD 0.51, 95% CI 0.20 to 0.83, I2 = 47%, fixed-effect model of 8 trials, see Figure 9 on eAddenda for detailed forest plot) and of a similar magnitude to the original analysis (SMD 0.49, 95% CI 0.22 to 0.75). This suggests that improvement in lower limb activities is due to the type of feedback selleckchem (ie, biofeedback compared with therapist feedback during usual therapy) rather than the amount of practice. Why might biofeedback be more effective than therapist feedback? An observational study of therapist-patient interactions during therapy found that the content of feedback was motivational rather than informative, with specific feedback rarely given (Talvitie 2000). As early as 1932, Trowbridge and Casen demonstrated that the content of feedback is important, with feedback containing specific information regarding ways to improve future practice, enhancing learning more than motivational feedback. By its very nature, biofeedback provides specific information that can be used to adapt

the next attempt at the task. This review has some potential limitations. Several of these limitations may have led to an overestimate of the effect of biofeedback. First, there old was a lack of blinding of participants and therapists since this is not always possible in trials of biofeedback. Second, even after including only high quality trials in the meta-analysis, the results are potentially affected by small trial bias, with an average number of 27 participants per trial (range 13–54 participants). Third, when multiple measures were reported, the measure used in the meta-analyses was the measure most congruent with the aim of the intervention, which may have introduced selection bias. On the other hand, the inclusion of trials that compared biofeedback only with usual therapy only does not distinguish the effect of biofeedback precisely, making the result from this systematic review a more conservative estimate of the effect.