This approach is recommended by others (Senn 2002). Power calculations were not conducted because there were no previous studies upon which to base a sensible estimate of the likely SD for urine output or with which to set a minimally worthwhile treatment effect. Therefore, a pragmatic approach to

determining the sample size was adopted. That is, we selected a sample size that was realistically achievable within a 2-year recruitment period even though ultimately we recruited within a 1.5-year period. We reasoned that an estimate of treatment effect even if imprecise from a trial with minimal bias would progress knowledge in this area and help sample size calculations for future trialists. Fourteen participants entered and completed 5-FU chemical structure the study. Their median (interquartile range) age was 25 years (22 to buy PCI-32765 32) and time since injury was 118 days (64 to 135). All participants had motor complete lesions (AIS A, B) with neurological

levels ranging from C4 to T10, as presented in Table 1. Figure 1 demonstrates the flow of participants through the trial. Primary and secondary outcomes were attained for every participant with no drop outs. The assessors remained blind for all aspects of the trial. Participants received a median of 8 FES cycling sessions (IQR 8 to 9) over a mean of 2 weeks (SD 0.5). There was some variation because the FES cycling was continued until the assessment at the end of the 2-week FES cycling phase could be completed. These assessments were sometimes delayed for a day or more because

of difficulties with scheduling. The results for all outcomes are presented in Table 2, with individual participant data presented in Table 3 (see eAddenda for Table 3). The mean between-group difference for urine output was 82 mL (95% CI –35 to 199), where a Oxygenase positive value favours the experimental intervention because it indicates an increase in urine output with FES cycling. The other mean between-group differences were –0.1 cm (95% CI –1.5 to 1.2) for lower limb swelling, –1.9 points (95% CI –4.9 to 1.2) on the 32-point Ashworth Scale, and –5 points (95% CI –13 to 2) on the 164-point PRISM. Here, negative values favour the experimental intervention because they indicate a decrease in swelling and spasticity with FES cycling. All but two participants reported improvements with the FES cycling on the Global Impression of Change Scale with a median improvement of 3 points (IQR 3 to 4) on the scale from –7 to +7. The median perception of inconvenience of the FES cycling was 0.3 points (IQR 0 to 3.8) on the 10-point Visual Analogue Scale. There were two reports of adverse effects. One related to an increase in spasticity and the other related to precipitation of a bowel accident.