Fig. 2.

Linear regressions of trends in utilization of steroid-sparing agents. a Linear regressions modeling proportions of patients with a drug exposure, with year as a predictor; regression lines are labelled with their slopes (rituximab’s slope of β = 0.005 indicates that with each additional year, the proportion of patients with evidence of rituximab utilization increases by 0.005. Cyclophosphamide’s slope of β = − 0.002 indicates that with each additional year, the proportion of patients with evidence of cyclophosphamide utilization decreases by 0.002)