R software

All statistical analyses were conducted using Stata version 11.2 (StataCorp, College Station, TX, USA) and R software (version 4.1.3). A p value less than 0.05 was considered statistically significant. To compare the differences in the antimicrobial resistance rates between VSSA and VISA/hVISA strains, we used both the Chi-square test and Fisher’s exact test, with Fisher’s exact test being utilized for any cell counts less than 5. To compare the proportion of bacterial strains showing synergy and antagonism between two antibiotic combinations, we used McNemar’s test, which is a statistical test used to compare paired proportions. The viable bacterial counts did not follow a normal distribution, as assessed by Shapiro–Wilk test. Therefore, non-parametric statistical tests were used to compare the medians of viable bacterial count between groups. We performed the Friedman test, followed by the Wilcoxon signed-rank test with Bonferroni adjustment for multiple groups. Multiple logistic regression analysis was performed to identify predictors for synergistic actions of ciprofloxacin– and levofloxacin–rifampin combinations. This analysis included variables such as year of isolation, vancomycin susceptibility, sequence type, SCCmec type, spa type, and levels of resistance to rifampin, ciprofloxacin, and levofloxacin.

Heterogeneity was assessed using Cochran's Q statistic and I² method. The strength of the association between the NSCL/P and BMP4 rs17563 polymorphism risk was assessed by crude ORs with 95% confidence intervals (CIs), which comprised the following models: allele contrast (C versus T); codominant (TC versus TT and CC versus TT); dominant (TC + CC versus TT); and recessive (CC versus TT + TC). If the P value < 0.10 or I² was greater than 40%, the summary OR estimation was calculated with a random-effect model (DerSimonian and Laird method). Otherwise, the fixed-effect model (Mantel-Haenszel method) was adopted. Cumulative meta-analyses and sensitivity analysis were conducted to evaluate the stability of the results for each model. Potential publication bias was evaluated by Egger's linear regression and Begg's funnel plots. To adjust for multiple comparisons, we applied Bonferroni method with R software. Statistical analysis was performed using STATA version 11.0 (Stata Corporation, College Station, TX, USA) and R software (version 3.1.1). All P values were two-sided. P < 0.05 was considered significant.

This study used STATA/SE 15.1 (StataCorp, 2017) and R software (version 4.0.1) to conduct data analysis and figures generation. We used WMD and their associated 95% CIs to summarize results. We took into account the existence of heterogeneity among different RCTs; thus, the random-effects model was selected to combine effect sizes in this network meta-analysis. We used the node-splitting model to assess inconsistency between direct and indirect comparisons. The bias in publication and small-scale study effects were evaluated with comparison-adjusted funnel plots, which were generated using “netfunnel” command. The network geometry of three different traditional Chinese fitness exercises was shown and described with network evidence plots, which were generated using “networkplot” command. We calculated the SUCRA and likelihood of being the best and the worst for each intervention to predict the curative effect ranking of each traditional Chinese fitness exercise. The significance level for all data analyses of this network meta-analysis was predetermined at 0.05.