Table 2

Crude and adjusted incidence rate ratios (IRRs) of upper gastrointestinal bleeds per subgroup

n eventsFollow-up time (person-years)Incidence rate (%/year)Crude IRR (95% CI)IPW IRR (CI)LRT significant*
Age2 out of 3
0–645343 5420.122.55 (1.33 to 4.88)1.09 (0.48 to 2.48)
65–74252101 0120.251.50 (1.06 to 2.11)0.99 (0.65 to 1.49)
75–8429487 9540.330.71 (0.49 to 1.03)0.58 (0.37 to 0.89)
≥8520740 0630.520.74 (0.49 to 1.12)0.67 (0.42 to 1.07)
Sex0 out of 3
Female457149 5970.311.02 (0.77 to 1.34)0.66 (0.47 to 0.92)
Male349122 9740.281.12 (0.83 to 1.51)0.88 (0.62 to 1.24)
HAS-BLED3 out of 3
Low (0–2)472209 5530.231.14 (0.86 to 1.52)0.95 (0.7 to 1.29)
High (≥3)33463 0180.530.76 (0.57 to 1.03)0.54 (0.36 to 0.8)
Concomitant NSAID1 out of 3
No691249 5200.281.01 (0.81 to 1.27)0.74 (0.57 to 0.96)
Yes11523 0500.501.15 (0.72 to 1.86)0.84 (0.46 to 1.54)
Concomitant AP2 out of 3
No585239 3390.241.10 (0.87 to 1.4)0.80 (0.61 to 1.06)
Yes22133 2320.670.79 (0.54 to 1.16)0.63 (0.38 to 1.04)
NOAC2 out of 3
Apixaban28293 5660.300.93 (0.67 to 1.31)0.67 (0.45 to 1.01)
Dabigatran240100 1050.241.06 (0.71 to 1.58)0.64 (0.39 to 1.07)
Rivaroxaban27876 8420.361.29 (0.92 to 1.80)1.03 (0.71 to 1.50)
  • Number of events, follow-up time, incidence rate, crude IRR and IPW IRR of PPI versus no PPI exposure in different subgroups.

  • *The number of databases in which the LRT was significant. If this test was significant in two or more databases, we considered a subgroup as a relevant effect modifier.

  • AP, antiplatelet; IPW, inverse probability weighted; LRT, likelihood ratio test; NOAC, non-vitamin K antagonist oral anticoagulant; NSAID, non-steroidal anti-inflammatory drug; PPI, proton pump inhibitor.