Statistical analysis
Normally distributed continuous variables are presented as mean (± standard deviation) and variables not following normal distribution as median (25th - 75th percentile). Nominal variables are summarized as counts and absolute percentages. Distribution of continuous variables was graphically assessed by histograms and quantile-quantile plots. Cardiac measurements were compared between GDM and controls with the independent samples Student’s T Test or the Mann-Whitney U Test and the chi-squared test for continuous and categorical variables, respectively. General linear regression models were used to assess the association between GDM and a range of echocardiographic parameters.
To facilitate the comparison of changes in echocardiographic parameters before and after pregnancy for the two groups of interest, we used linear mixed models with two random effects (random intercept and random slope) and an unstructured variance-covariance matrix. Cardiac parameters which were used as outcome variables included E/A, E/E’, global longitudinal strain, and myocardial performance index. An interaction term (GDM yes/no*time interval between prenatal and postnatal assessment) was introduced in linear mixed models to evaluate the potential differential effect of GDM on changes in cardiac measurements before and after delivery. Analysis was further adjusted for a pre-specified set of confounders, including maternal age, race, time elapsed from delivery to postnatal visit and change in infant weight from birthweight. Statistical analysis was conducted with STATA package, version 13.1 (StataCorp, College Station, Texas USA). We deemed statistical significance at p <0.05. All tests were 2-tailed.