Twenty-five studies (n=248) were included in this review: 4 RCTs, 2 controlled studies without randomisation and 19 uncontrolled before-and-after studies. The number of patients per study ranged from 1 to 20.
Overall, the methodological quality was judged to be acceptable in 14 studies (scoring better than 50% on the checklist) and low in 11 studies. The results were grouped according to study design, outcome measure and quality of the relevant trials, as shown below.
Two of the RCTs compared exercise versus no exercise and were described as being of low but acceptable quality, one looked at supine versus sitting training position and was of relatively high quality, and the fourth compared two intensities of training (quality not reported). Of the non-randomised studies, one compared two training intensities and the other assessed training versus no training.
VO2peak (21 studies): 18 out of 21 studies reported a significant improvement in VO2peak following training of some kind. Of these 21 studies, 13 were judged to be of acceptable quality and, overall, these showed a mean improvement of 17.6% in VO2peak (SD=11.2).
POpeak (20 studies): 16 of the 20 studies reporting these data found a significant increase in POpeak following training and the remaining 4 studies found no increase. Twelve studies were of acceptable quality and, overall, reported a mean change of 26.1% (SD=15.6).
VO2peak (paraplegia): 9 studies of acceptable quality (including 2 RCTs) assessed this outcome and found improvements of between 7% and 30%.
POpeak (paraplegia): 8 uncontrolled studies of acceptable quality assessed this outcome and found improvements of between 10% and 30%.
Four studies of acceptable quality reported results on VO2peak and POpeak for tetraplegic patients with mixed findings.
Type of training was considered in a further analysis, but was hindered by the lack of clear descriptions of the intervention used in the primary studies.