Data synthesis and statistical analyses
The meta-analysis was conducted using R Statistical Software (v3.5.1,
Boston, MA, USA) with the ‘meta’ and ‘metafor’ packages for analysis,
and RevMan (Review Manager Version 5.1, The Cochrane Collaboration,
Copenhagen, Denmark). A two-tailed p-value of < 0.05 was
considered statistically significant. Summary statistics are presented
as weighted mean differences (WMD) with 95% confidence intervals (CIs).
Mean and standard deviation (SD) values were estimated based on the
method described by Hozo et al. [20]. A random-effects model (Der
Simonian and Laird method) was used to estimate the pooled prevalence of
adverse events across studies. The 95% CIs for the prevalence of
adverse events were calculated using the binomial exact method
(Clopper-Pearson) based on the proportion of cases and sample size.
Inverse variance weighting was applied to each study in the
meta-analysis. Results are illustrated in forest plots, the standard way
of presenting individual study and meta-analysis outcomes. Sub-analyses
were conducted to evaluate the effect of different types of
lipid-lowering therapy (LLT) on lipid profiles and safety parameters.
Meta-analyses were performed using the random-effects model.
Heterogeneity across studies was assessed using the Cochrane Q test and
the I² index, with I² < 25% indicating low, 25-50% moderate,
and >50% high heterogeneity. The reduced maximum likelihood
method (tau²) was used to incorporate residual heterogeneity in the
analysis [21]. Publication bias was assessed through visual
inspection of funnel plots and Egger’s test.