Eight RCTs (417 patients) were included.
Study quality: the reporting of studies was generally poor. Four of the 8 RCTS were considered to have adequate allocation concealment and 4 RCTs adequately described blinding.
Change in lean-body mass (5 RCTs, 324 patients): testosterone significantly increased lean-body mass in comparison with placebo. The WMD (random-effects model) was 1.22 kg (95% CI: 0.23, 2.22). A post-hoc subgroup analysis was performed since significant heterogeneity was found (P=0.0002); the heterogeneity was due to differences in the mode of administration. The WMD for i.m. testosterone versus placebo (3 RCTs, 91 patients) was 3.34 (95% CI: 2.07, 4.61); no significant heterogeneity was found (P=0.55). The WMD for testosterone patches versus placebo (4 RCTs, 233 patients) was 0.17 (95% CI: -0.28, +0.61); no significant heterogeneity was found (P=0.25). Omitting the trial in women and the trial that reported body cell mass changed the effect size, but not the direction of effect.
Change in total body weight (8 RCTs, 384 people): there was no significant difference in body weight change between testosterone and placebo. The WMD (random-effects model) was 1.04 kg (95% CI -0.01, +2.10); significant heterogeneity was found (P=0.022). The results were similar for men only data.
Testosterone i.m. plus resistance exercise (2 RCTs, 53 people): there was no significant difference in lean-body mass between testosterone plus exercise and placebo plus exercise. The WMD was 1.28 kg (-0.18, +2.73).
Overall exercise functional capacity or muscle strength: studies used different methods to assess functional capacity and muscle strength. Three of the 7 RCTs that assessed this outcome found that testosterone increased functional capacity or muscle strength compared with placebo; the other 4 RCTs found no significant difference.
Quality of life (6 RCTs): 3 RCTs found that testosterone improved quality of life, while 3 RCTs found no significant difference.
Adverse events: one RCT did not report the number of adverse events, while the other RCTs did not report the adverse effects consistently. In the individual studies, the rates of adverse events ranged from 1 to 86% with testosterone and from 4 to 95% with placebo. The overall rates of adverse events appeared similar with testosterone and with placebo. Adverse events included local reaction to patches, acne, gynaecomastia and breast tenderness.
The funnel plot showed asymmetry so publication bias could not be excluded.