| Tiotropium + placebo(N = 156 patients) | Tiotropium + fluticasone/salmeterol(N = 145 patients) | Rate ratio (relative risk reduction) (%) | |
| Weighted mean exacerbation rate/patient-year (adjudicated events) | 1.61 | 1.37 | 0.85 (15) |
| Weighted mean exacerbation rate/patient-year (non-adjudicated, independence not assured) | 2.09 | 1.82 | 0.87 (13) |
| Weighted mean exacerbation rate/patient-year with premature exclusion of patients who stop study medications (adjudicated events) | 1.66 | 1.31 | 0.79 (21) |
| Unweighted mean exacerbation rate/patient-year (adjudicated events) | 2.00 | 1.48 | 0.74 (26) |
| Unweighted mean exacerbation rate/patient-year with premature exclusion of patients who stop study medications (adjudicated events) | 4.94 | 2.29 | 0.46 (54) |
Row 1 shows the correct rate ratio for the trial (the maximum likelihood estimate).
Row 2 shows the effects of not adjudicating and assuring independence of individual exacerbation events.
Row 3 shows the effects of prematurely excluding patients from the trial who discontinue study medications early.
Row 4 shows the effects of using an unweighted statistical approach.
Row 5 shows the effects of both prematurely excluding patients from the trial who discontinue study medications early and of using an unweighted statistical approach.