Eleven RCTs were included (n=474 according to the text and n=455 according to the data extraction tables). These studies provided 38 effect sizes.
The mean unadjusted effect size using one effect size per study was 0.58 (95% CI: 0.22, 0.93). The mean effect size using all 38 effect sizes was 0.53 (95% CI: 0.21, 0.86).
There appeared to be evidence of heterogeneity but it was not clear to which analysis this applied (all 38 effect sizes, or the 1 effect size per study analysis). Other analyses were based on all 38 effect sizes.
Effect sizes were larger for measures of lower extremity muscle strength (0.63, 95% CI: 0.03, 1.28; based on 20 effect sizes) compared with upper extremity strength (0.47, 95% CI: 0.12, 0.84; based on 17 effect sizes).
Effect sizes were larger for studies using injected testosterone (0.95, 95% CI: 0.33, 1.58; based on 18 effect sizes) compared with topical (0.26, 95% CI: 0.08, 0.42; based on 16 effect sizes) or oral testosterone (-0.21, 95% CI: -1.40, 1.02; based on 4 effect sizes. Effect sizes from studies with no attrition were larger than those from studies with attrition levels greater than 10% (1.27 versus 0.10). Effect sizes from studies using ITT analysis were smaller than those from studies that did not use ITT analysis (0.15 versus 0.89).
Effect sizes from higher quality studies were smaller than those from lower quality studies (0.30 versus 0.64).
The removal of 1 study (n=12, provided 5 effect sizes) substantially decreased the treatment effect size from 0.53 to 0.23 (95% CI: 0.09, 0.38). This study administered an individually determined injected dose of testosterone.
All 11 studies monitored adverse events but definitions varied widely. Three studies reported high prostate-specific antigen or prostate disease. Four studies reported no adverse events.