2.4. Statistical Analysis

Statistical analyses were performed using RevMan 5.3 [15]. The means of each OxSM were analyzed using the DerSimonian and Laird method to obtain a pooled mean difference (MD) with 95% confidence intervals (95%CI). If necessary, standard deviations were calculated from the mean, standard error, and number of participants using GraphPad QuickCalcs (GraphPad Software, Inc, San Diego, CA, USA). Studies with insufficient information to be included in the meta-analysis were excluded. For the enzyme superoxide dismutase, a sub-analysis was performed in terms of Tai Chi sessions per week. For the heterogeneity assessment, I² statistical analysis was performed. I² values ≤25%, 25–50%, 50–75%, and >75% indicated no, small, moderate, and significant heterogeneity, respectively. Leave-one-out sensitivity analysis was utilized to determine if the pooled effects were robust, and the study with the greatest effect was excluded. A sensitivity analysis was conducted only when three or more studies were included in the comparison. Publication bias was assessed using the Egger regression asymmetry test when the comparison contained at least 10 studies.