The study populations were required to be primarily

aged 60 or older. Trials that included younger participants were considered eligible if the mean age of participants minus one standard deviation was over 60 years. Eligible interventions included strength and balance training, and physical training such as dance, Tai Chi and other complementary therapies. Comparisons in eligible studies were between the intervention group and either a usual care or control group, and studies with factorial designs comparing more than one intervention were also included. Included studies measured physical function with performance tests or questionnaires and/or falls with calendars or incident reports. Eligible aspects of physical function were mobility, balance, Fulvestrant clinical trial strength and proprioception. Random-effects meta-analyses were conducted using commercial softwarea to compare the impact on the outcomes of interest of programs designed to enhance physical function or prevent selleck products falls with control programs or usual care. The weighted mean difference (WMD) was calculated using the pre-intervention and post-intervention means

and standard deviations. Statistical heterogeneity was quantified with the I2 and Q statistics. The electronic database search identified 3451 records after removal of duplicates. After screening by title and abstract, full articles were then obtained for 10 trials and their eligibility assessed against the inclusion criteria. After more detailed investigation, three papers were excluded because

they were not randomised controlled trials, one because the participants MRIP were not visually impaired, one because there was no physical intervention and one because it was another report of an included trial. Four trials were deemed to fit the inclusion criteria and results from two trials were combined in a meta-analysis. Figure 1 shows the flow of search results through to the selection for meta-analysis. The four studies included in the review were randomised controlled trials published in English. Their quality scores are presented in Table 1, and their designs, participant characteristics, interventions and outcome measures are summarised in Table 2. The VIP trial by Campbell and colleagues20 was a 12-month, 2 x 2 factorial community-based trial involving men and women over 75 years of age with visual impairment. The remaining three trials were undertaken in residential care settings. The trial by Chen and colleagues21 ran for 16 weeks and stratified the randomisation based on gender, age and level of visual impairment. Cheung and colleagues22 assessed women over 70 years of age in a 12-week trial, and Kovács and colleagues23 assessed women over 60 years of age in a 6-month trial. There were 522 participants in total in the included studies, but data from only 91 participants could be pooled for meta-analysis. Three trials21, 22 and 23 measured physical function as the primary outcome.