A computerised search was conducted of the PubMed database using the search terms: ((urinary AND incontinen*) OR pelvic floor) AND (Yoga OR Tai Chi OR Pilates OR breathing OR posture OR transversus abdominis OR fitness). The advanced search on PEDro used the terms ‘incontinence’ and ‘clinical trial’. In PubMed the search was limited to randomised controlled trials reported in the English, Scandinavian, or German languages. The final searches were conducted on 4 January 2013. Studies were

included in the review if they were randomised controlled trials investigating the effectiveness of exercise regimens other than specific

pelvic floor muscle training. Pelvic floor muscle training could be carried learn more out with or without biofeedback, electrical stimulation, vaginal SB203580 nmr cones, and resistance devices (Dumoulin and Hay-Smith 2010, Hay-Smith et al 2011, Herderschee et al 2011, Parsons et al 2012). The inclusion criteria for the review are presented in more detail in Box 2. Exclusion criteria were: studies on women with other forms of urinary incontinence or lower urinary tract symptoms, studies on women with neurological diseases, and studies on bladder training. Design • Randomised trial The included trials were classified according to many preset criteria: type of alternative exercise regimens, comparison intervention, participants and diagnoses, interventions, primary outcome measures, and results.

We considered methodological limitations of each of the trials. The PEDro scale for rating quality of randomised controlled trials was used to score methodological quality (Maher et al 2003). Two researchers classified and scored each trial independently. Disagreements were resolved by discussion. The results are presented in the following way. Each alternative exercise regimen is considered in turn. First we provide a brief description of the theoretical justification for the therapy. Then the evidence supporting the intervention is presented, beginning with the evidence from laboratory studies and observational (epidemiological) studies and concluding with randomised trials. We did not attempt to systematically search for laboratory or epidemiological studies as this would have been very difficult and the focus was on randomised trials.