Quantitative characterization of the relationship between levels of extended corticosteroid use and related adverse events in a US population