Multivariable survival models showing the relationship between statin adherence and ischemic stroke, according to atrial fibrillation (AFib) status. Three multivariable Cox regression models are shown, displaying the hazard ratio for the relationship between the natural logarithm of percentage days covered (PDC) and ischemic stroke >3 y. Solid triangles represent the hazard ratio point estimates and horizontal error bars show the 95% confidence interval. The top row shows the results of the model for the subgroup without AFib (n=4669). The middle row shows the results of the model for the subgroup with AFib (n=1446). The bottom row shows the results of the mode for the subgroup with AFib on warfarin anticoagulation, after controlling for time in the therapeutic range (TTR) for international normalized ratio values (n=1010). All models controlled for age, sex, race/ethnicity, hypertension, diabetes mellitus, hypercholesterolemia, coronary artery disease, congestive heart failure, history of prior ischemic stroke, and hospital center.
Model estimation of the 3-y cumulative hazard of ischemic stroke according to the adherence (PDC) and atrial fibrillation (AFib) status. A, Plot of the model estimates of the cumulative hazard of recurrent ischemic stroke over 3 y (y-axis) according to the adherence (PDC). Estimates of the hazard of stroke rose sharply at low levels of PDC for both subjects with AFib (gray curve, 95% confidence interval in flanking dotted lines) and subjects without AFib (No AFib, black curve, 95% confidence interval in flanking dotted lines). B, Similar plot of the model estimates of stroke hazard according to PDC, but in this case the AFib group is the subgroup on warfarin anticoagulation with additional control for time in the therapeutic range for international normalized ratio values (AFib [Control for TTR], gray curve, 95% confidence interval in flanking dotted lines). Underlying models for estimation were prediction of ischemic stroke >3 y by natural logarithm of PDC in multivariable Cox regression, controlling for age, sex, race/ethnicity, hypertension, diabetes mellitus, hypercholesterolemia, coronary artery disease, congestive heart failure, and history of previous ischemic stroke. For plotting purposes, natural logarithm of PDC was transformed back to PDC.
Statin adherence by percentage days covered (PDC) and change in low-density lipoprotein (LDL). A, Distribution of PDC in the cohort shows number of subjects (y-axis) in each bin on PDC, by 10% (x-axis). B, Relationship between PDC and change in LDL in the subset of subjects who were not taking a statin at the time of their initial ischemic stroke. PDC is plotted on the x-axis against change in LDL in mg/dL over the period of observation on the y-axis (negative numbers indicate a decrease in LDL over time). C, Percentage of subjects with a decrease in LDL by ≥30 mg/dL over the observation period, according to PDC. On the left, the difference is shown according to PDC<50 vs PDC 50+ (adherence level less than or greater than/equal to half the time), and at right, the difference is shown for PDC<85 and PDC 85+ (adherence level less than or greater than/equal to the mean PDC for the cohort). P values are from the Fisher exact test.
Adjusted ischemic stroke-free survival curves as a function of statin adherence, according to atrial fibrillation (AFib) status. Adjusted ischemic stroke-free survival curves are displayed for a cutpoint of percentage days covered (PDC) <85 vs PDC 85+ (cut at the mean PDC for the overall cohort). A, Survival curves for the subgroup without AFib. B, Survival curves for the subgroup with AFib. C, Survival curves for the subgroup with AFib on warfarin anticoagulation, after control for time in the therapeutic range (TTR) for international normalized ratio values. All survival curves are adjusted for age, sex, race/ethnicity, hypertension, diabetes mellitus, hypercholesterolemia, coronary artery disease, congestive heart failure, history of previous ischemic stroke, and hospital center.
Alexander C. Flint, Carol Conell, Jeff G. Klingman, Vivek A. Rao, Sheila L. Chan, Hooman Kamel, Sean P. Cullen, Bonnie S. Faigeles, Steve Sidney, S. Claiborne Johnston
Journal of the American Heart Association. 2016;5:e003413, originally published July 29, 2016
