Figure 4.

Subgroup analysis displaying no differences according to FMD cuff inflation threshold, the presence of age- and sex-matched control group, or a significant difference in obesity prevalence. However, we observed lowered effect sizes in studies using other NAFLD diagnostic methods (multi-detector abdominal computed tomography, magnetic resonance spectroscopy). Effect sizes were pooled according to the random-effects model and the subgroup analysis followed the fixed-effects (plural) model. Results are expressed as standardized mean difference (SMD) with horizontal error bars denoting the 95% confidence intervals (CIs).