Statistical analysis
The continuous variables are expressed as mean values ± SD. The intra- and inter-observer reproducibility of the LA strain measures were evaluated in 20 patients by two operators (M.I., N.D.) by means of the intraclass correlation coefficient (ICC), which measures the strength of the association between the two baseline recordings. The data were considered reproducible if the ICC was >0.60, and the reproducibility was considered almost perfect if the ICC was between 0.81 and 1.00 [10].
Pearson’s linear correlations were used to assess the relationship between atrial strain measures and the other variables studied. Cox’s proportional hazard model was used to verify the associations between the variables and the endpoints. Hazard ratios (HRs) are given with their 95% confidence intervals (CIs) and refer to a unit increase in the variable. The variables used in the forward stepwise multivariable models were selected based on their statistical significance during univariate analysis and the most relevant risk parameters for HF patients. Discrimination of the univariate and multivariate models was determined using a modified concordance statistic (C-index) for censored data.
Receiver operating characteristic (ROC) curves and the area under the curve (AUC) of the ROC were calculated to determine the associations between the studied variables and 1-year events. The best cut-off values for the analysed events were defined based on the highest sum of sensitivity and specificity. The event-free curves were based on Kaplan–Meier analysis stratified by the best cut-offs of LASr and LASr/Ees and were compared using a log-rank test. The proportional hazards assumption was formally assessed. Statistical analyses were performed using STATA, version 12 (StataCorp, College Station, Texas, USA), or Statistica 6.1 (StatSoft Inc., Tulsa, Oklahoma, USA).P -values <0.05 were considered statistically significant.