Statistical Analysis
All statistical analyses except inter- and intra-rater reliability
analyses were carried out in SAS 9.4 (SAS Institute, Cary, N.C.). To
assess inter- and intra-rater reliability, Pearson correlation
coefficients were calculated for edited autostrain GLS measurements
within and between raters. Demographics were summarized by pre/post
group using means and standard deviations for numeric variables and
frequencies and percentages for categorical variables. Mean pre-post
changes in LVEDV, LVESV, LVEF, LAVI, and RVSP were calculated and were
tested for significance using paired t-tests. The effect of
pre-procedural GLS (Pre-GLS) on the odds of a 10% change in EDV was
modeled using logistic regression. This was done for the entire sample
and for several sub-groups. Finally, the effect of GLS on the numeric
change in EDV was modeled using linear regression, as were the effects
of LVEDV, LVESV, LVEF, LAVI, and RVSP on numeric change in EDV. For this
final set of linear regression models, crude (single-predictor) models
were fit for each predictor first. Linear and quadratic models were fit
to assess the best functional form. GLS and any other predictors that
were significant in their respective crude models were included as
predictors in a final multivariable model.