Figure 2.

Association between long-term immunosuppression and in-hospital death, by medication class

Analyses were done in the propensity score matched cohort, with doubly robust adjustment for any remaining covariate imbalances after matching.