Figure 1.

PPI administration (%) for patients receiving the clopidogrel, prasugrel or ticagrelor antiplatelet (AT2) or dual antiplatelet (DAT). The total number of patients in the subgroups are given inside the bars. If no censoring was required (left panel – Internal Section patients that received steroids), proportions of PPI administration calculated from the synthetic sets were essentially identical to the proportions in the real data, and their range across 1000 sets (minimum and maximum in red dotted lines) was very narrow. If censoring was required, as in the case of the Surgery Section, results varied across the synthetic sets, and their ranges were wider (right panel – Surgery Section patients older than 65 years). The means across 1000 sets (purple lines) show small biases.