Table 1 Estimated difference in 5-year mortality risk (95% confidence interval) when comparing 0% vs. 100% of follow-up being at least 80% adherent, Coronary Drug Project
From: Improved adherence adjustment in the Coronary Drug Project
Method | Unadjusted | Standardized by baseline covariates | + IP weighting for post-randomization covariates |
|---|---|---|---|
Censoring when binary adherence level deviates from baseline | 9.6 (– 7.0, 28.9) | 1.5 (– 11.0, 16.4) | 0.01 (– 12.2, 13.2) |
Dose-response hazards models | |||
Quadratic cumulative adherence | 6.7 (− 3.2, 16.5) | 2.6 (− 5.8, 11.1) | 0.2 (− 8.2, 8.6) |
Quadratic cumulative adherence, with time interaction | 6.0 (− 5.8, 18.0) | 1.6 (− 8.4, 11.6) | − 0.7 (− 12.2, 10.7) |
Binary current adherence, plus quadratic cumulative adherence up to previous visit | 11.6 (− 1.0, 24.2) | 6.4 (− 3.9, 16.8) | 4.5 (− 6.3, 15.3) |
Binary current adherence, plus quadratic cumulative adherence up to previous visit, with time interaction | 10.3 (− 2.3, 22.9) | 4.0 (− 6.0, 14.1) | 2.4 (− 9.3, 14.1) |