Using a Bayesian latent growth curve model to identify trajectories of positive affect and negative events following myocardial infarction.