2.6. Analyses and measures of treatment effect:
For each trial, mean differences and standard deviation were calculated
for continuous variables. The differences of mean changes in cortisol
and testosterone concentration and their correspondence SD were
extracted as the effect size. In trials that the standard error (SE)
measure was reported, SE was converted to SD applying this formula: SD=
SE×√n (n=number of subjects in each group). Between-study heterogeneity
was assessed using the chi-squared (χ2) test and quantified applying the
I2 statistic, which shows the total variation
percentage across studies that is ascribable to heterogeneity rather
than to chance. Significant heterogeneity was defined with a P-value of
<0.05.
The random effects model was applied to calculate the weighted mean
differences (WMDs) with 95% confidence intervals (CIs) for estimating
the overall effect. To evaluate whether the outcomes could have been
altered by a single study distinctly, a sensitivity analysis was
performed (28). Also, subgroup analysis was conducted, based on the
sports type (resistance training or aerobic and anaerobic combined
training) and duration of studies (over six weeks and 6 weeks or lower).
Publication bias was evaluated by Egger’s regression asymmetry test and
Begg’s rank correlation test. Funnel plots also depicted the effect
sizes against their corresponding standard errors. Statistical analyses
were conducted applying STATA 11.2 software (StataCorp, College Station,
Texas, USA).