Echocardiographic Predictors of Post-Operative LV Dysfunction

The endpoint was the occurrence of post-operative LV systolic dysfunction defined by an LVEF <50% at ≥6 months after MV repair (27 (link)). Quantitative variables were described as the means ± standard deviations, and qualitative variables were described as numbers and percentages. Non-normally distributed data were reported as median and interquartile ranges. For univariate analyses, comparisons between groups according to the primary endpoint were made using the Chi² test if valid (Fisher's exact test otherwise) for the qualitative variables and using Student's t-test if valid (Mann–Whitney test otherwise) for the quantitative variables. For echocardiographic parameters, a univariate logistic regression model was used to estimate crude odds ratios with their 95% CIs quantifying the excess risk of post-operative LV dysfunction. Multivariate analyses were then performed to assess the independent effect of echocardiographic parameters while taking into account potential prognostic factors selected beforehand according to the literature data and to the results of univariate analysis. Adjusted odds ratios were estimated with their 95% CIs. Firth's correction was applied by performing Firth's penalized-likelihood logistic regression to take into account the small number of post-operative LV dysfunction.
Receiver operating characteristic (ROC) curves were established to determine the optimal cutoff value of echocardiographic parameters to predict post-operative LV dysfunction. Echocardiographic parameters were then dichotomized according to the identified cutoff values. Univariate and multivariate Firth's penalized-likelihood logistic regression models were built to estimate crude and adjusted odds ratios with their 95% CIs related to these dichotomized parameters for predicting post-operative LV dysfunction.
All tests were two-sided. All p-values<0.05 were considered significant. We used R version 4.0.5 for all statistical analyses.