Sensitivity analysis
As presented in Table 2 , we re-analyzed all outcomes using fixed-effects models, with subgroup analyses based on study design. For the primary outcomes of SSE and major bleeding, the results of the fixed-effects model and subgroup analyses were similar to the random-effects model analyses described above. However, fixed-effects models for major bleeding outcomes showed an advantage in using NOACs in both moderate and severe polypharmacy patients. In the sensitivity analysis, the corresponding HR values were not changed substantially after excluding one study at a time.