Table 2.
Unmatched Original Cohort | Propensity-Score-Matched Cohort | |||||
---|---|---|---|---|---|---|
Non-User Group | User Group | E Value | Non-User Group | User Group | E Value | |
Total stroke | ||||||
No. of events (%) | 1110 (8.6%) | 169 (6.9%) | 244 (9.9%) | 168 (6.8%) | ||
IR per 1000 person-years (95% CI) | 18.96 (17.86, 20.11) | 13.80 (11.80, 16.05) | 23.09 (20.28, 26.18) | 13.73 (11.73, 15.97) | ||
Unadjusted HR (95% CI) | 1.00 (reference) | 0.71 (0.61, 0.84) | 2.17 | 1.00 (reference) | 0.59 (0.48, 0.72) | 2.78 |
Multivariable-adjusted HR (95% CI) | 1.00 (reference) | 0.69 (0.59, 0.81) | 2.26 | 1.00 (reference) | 0.60 (0.49, 0.73) | 2.72 |
Propensity-score-adjusted HR (95% CI) | 1.00 (reference) | 0.68 (0.58, 0.80) | 2.30 | 1.00 (reference) | 0.59 (0.49, 0.72) | 2.78 |
Types of stroke | ||||||
Ischemic stroke | ||||||
No. of events (%) | 877 (6.8%) | 139 (5.6%) | 200 (8.1%) | 138 (5.6%) | ||
IR per 1000 person-years (95% CI) | 14.98 (14.00, 16.00) | 11.35 (9.54, 13.40) | 18.92 (16.39, 21.73) | 11.28 (9.48, 13.33) | ||
Unadjusted HR (95% CI) | 1.00 (reference) | 0.75 (0.63, 0.90) | 2.00 | 1.00 (reference) | 0.60 (0.48, 0.74) | 2.72 |
Multivariable-adjusted HR (95% CI) | 1.00 (reference) | 0.71 (0.60, 0.85) | 2.17 | 1.00 (reference) | 0.61 (0.49, 0.76) | 2.66 |
Propensity-score-adjusted HR (95% CI) | 1.00 (reference) | 0.71 (0.59, 0.85) | 2.17 | 1.00 (reference) | 0.60 (0.49, 0.75) | 2.72 |
Hemorrhagic stroke | ||||||
No. of events (%) | 275 (2.1%) | 39 (1.6%) | 50 (2.0%) | 39 (1.5%) | ||
IR per 1000 person-years (95% CI) | 4.70 (4.16, 5.29) | 3.19 (2.27, 4.35) | 4.73 (3.51, 6.24) | 3.19 (2.27, 4.36) | ||
Unadjusted HR (95% CI) | 1.00 (reference) | 0.64 (0.46, 0.90) | 2.50 | 1.00 (reference) | 0.63 (0.42, 0.96) | 2.55 |
Multivariable-adjusted HR (95% CI) | 1.00 (reference) | 0.66 (0.47, 0.92) | 2.40 | 1.00 (reference) | 0.63 (0.42, 0.96) | 2.55 |
Propensity-score-adjusted HR (95% CI) | 1.00 (reference) | 0.63 (0.45, 0.88) | 2.55 | 1.00 (reference) | 0.64 (0.42, 0.97) | 2.50 |
In the unadjusted model, the only treatment was included as a covariate. In the multivariable model, all additional covariates were included. Using the same covariates, the individual propensity score was calculated. IR, incidence rate; HR, hazard ratio; CI, confidence interval.