Table 2

Utilities, medical resource use and long-term life expectancy

ParameterBase case valueRange of variation in sensitivity analysis*Basis of variation; distribution in PSA
Life expectancy (years)15–17
 1-year survivors, at end of year 111.268.45–14.08±25%; triangular
 3-year survivors, at end of year 39.377.03–11.71
Utility scores18
 First year after STEMI event0.680.60–0.76Utility
 Subsequent years0.720.65–0.79Decrement ±25%; triangular
Anticoagulant use, % of patients
 GPI (H-GPI; bivalirudin)95.3%; 7.6%NA
 Bivalirudin (bivalirudin)96.9%
Anticoagulant phial numbers, mean§
 Abciximab 10 mg (H-GPI; bivalirudin)3.07; 2.803.03 to 3.11; 2.63 to 2.9695% CI; γ
 Eptifibatide 20 mg (H-GPI; bivalirudin)1.88; 1.641.84 to 1.92; 1.50 to 1.78
 Tirofiban 12.5 mg (H-GPI; bivalirudin)1; 1
 Bivalirudin 250 mg (bivalirudin)1.231.21 to 1.26
Initial hospitalisation days, mean
 Normal ward (H-GPI)2.452.28 to 2.65 **95% CI; γ
 ICU/CCU (H-GPI)1.951.76 to 2.16 **
 Normal ward (Δ bivalirudin–H-GPI)0.06(−0.11) to 0.3295% CI;
 ICU/CCU (Δ bivalirudin–H-GPI)−0.32(−0.48) to (−0.17)lognormal††
Estimates for 42.5% radial arterial access use as assumed in the base case analysis§
 Initial hospitalisation days, mean
  Normal ward (Δ bivalirudin–H-GPI)0.05(−0.21) to 0.3395% CI;
  ICU/CCU (Δ bivalirudin–H-GPI)−0.25(−0.49) to 0.01lognormal††
  • * Only covering sensitivity analyses addressing the impact of parameter uncertainty. Additional scenario analyses are described in the text and in the online supplement.

  • In the alternative model (with detailed modelling of adverse events until the end of year 3), overall survival in the comparator strategy was made identical to that seen in the main model.

  • Varied in scenario analysis.

  • § Mean per patient in whom substance was used. The means shown originate from integer phials numbers at the patient level; non-integer values were increased to the next integer. For example, in the case of bivalirudin, this resulted in 1–3 phials per patient; in the vast majority of patients one phial was used.

  • Based on bias-corrected bootstrap results using 1000 replications.

  • ** CIs do not take into account uncertainty in the H-GPI strategy overall length of stay assumption of 4.4 days.19

  • †† In the model, these length of stay differences between strategies were implemented using multiplication factors that were assigned lognormal distributions.

  • Δ, difference; H-GPI, heparin and glycoprotein IIb/IIIa inhibitor; ICU/CCU, intensive care unit/coronary care unit; NA, not applicable; PSA, probabilistic sensitivity analysis; STEMI, ST-segment elevation myocardial infarction.