Statistical analysis
For individual TTE measures of interest, a meta-analysis was only
performed if studies reported continuous variable indices as means and
standard deviations. All indices were meta-analyzed with random-effects
modeling for overall effect size through SMD. The observational design
of the selected studies was considered, and the methodological
differences potentially responsible for a significant component of the
variance within the measures of interest. Pooled estimates were obtained
through a random-effect model. Articles were required to report on a
quantitative estimation including mean, confidential interval (CI), and
standard deviation. 95% confidence intervals (CIs) were calculated for
all indices. Lack of heterogeneity (magnitude of inconsistency) within
the pooled studies was tested using the I2 statistic
and deemed significant if ≥50%. Potential publication bias was assessed
visually from funnel plots (supplementary materials). All statistical
analysis was performed using the Metan package included in the STATA
version 16.1 statistical software. A p value <0.05 was
considered statistically significant for all analyses.
RESULTS