Best (but oft-forgotten) practices: identifying and accounting for regression to the mean in nutrition and obesity research.