Alternative Statistical Analysis Shows Exercise Training-Induced Improvements in Peak VO^sub 2^ are Clinically Significant

• Published in: Sports medicine (Auckland) 2015-05, Vol.45 (5), p.763
• Main Authors: Li, Sheyu, Li, Qianrui, Lyu, Xiafei, Tian, Haoming, Ran, Xingwu
• Format: Article
• Language: English
• Subjects: Confidence intervals; Exercise; Handbooks; Heart attacks; Intervention; Meta-analysis; Mortality; Physical fitness; Sensitivity analysis; Studies
• ISSN: 0112-1642; 1179-2035

Noting its small sample size and flawed study design, the dominance of this study largely increases the risk of bias in overall outcome. [...]because the meta-analysis did not provide a sensitivity analysis, we excluded this highly weighted study [12] from the overall analysis and observed a very different outcome, even when using a fixed-effects model (fixed-effects model: MD 1.11 mlkg-1min-1 [95 % CI 0.87-1.35, p\0.00001]; random-effects model: MD 1.38 mlkg-1min-1 [95 % CI 1.02-1.74, p\0.00001]). [...]the conclusions drawn by the investigators and the results they obtained seemed contradictory. [...]when introducing a third group of study duration 24 weeks or longer (two studies, two intervention groups; MD 0.78 mlkg-1min-1 [95 % CI -0.37-1.93, p = 0.18]), a tendency was observed that when the duration elongated beyond 52 weeks, peak VO2 improvement decreased abruptly, suggesting that a study duration between 12 and 52 weeks might be optimal, instead of a duration of at least 24 weeks, as concluded by the authors [1]. [...]when we re-analysed the data reviewed in the meta-analysis [1], we found that the overall improvement in peak VO2 in the PAD exercise training group was in fact of clinical significance, and that a study duration of 12-52 weeks was preferred.