Developing an anticoagulation composite measure: a stronger predictor for warfarin associated complications and a more comprehensive performance measure for anticoagulation clinics
MetadataShow full item record
BACKGROUND: Percent time in therapeutic range (TTR) and INR variability are both used to measure anticoagulation control with warfarin. TTR measures anticoagulation intensity, while INR variability measures anticoagulation stability; both predict definitive clinical outcomes such as stroke, major hemorrhage. Here, we examine whether an intermediate composite measure (ICOMO) predicts warfarin associated complications better than each measure separately. We also examine how the choice of measure changes the ranking order of anticoagulation clinics (ACCs) in the Veterans Health Administration (VHA) healthcare system. METHODS: We calculated TTR and INR variability for the study sample (N=130,897 patients) from 100 VHA ACCs. We constructed ICOMO using an equally weighted method, adding standardized TTR to standardized log-transformed INR variability. We used a subset of patients anticoagulated for atrial fibrillation (N=40,404) and divided them into quintiles based on their level of control, for each anticoagulation measure. We calculated the Hazard ratios for ischemic stroke and major bleeding and compared the ability of our independent variables (TTR, log INR variability, ICOMO) to predict each outcome. We measured mean observed value (O) and mean expected value (E) for each clinic, after adjusting for important clinical and demographic variables, for each anticoagulation measure. We identified outlier anticoagulation clinics if O was one standard deviation different from its corresponding E. We measured Kappa score and Pearson correlation coefficients when ranking sites according to each anticoagulation measure. RESULTS: ICOMO predicted ischemic stroke better than TTR and log INR variability in all quintiles. ICOMO and TTR predicted major bleeding similarly except in the second-best quintile; but both measures were better than log INR variability in all quintiles. Kappa scores identifying outlier and non-outlier clinics among our three profiling measures were moderate between ICOMO and its components (0.59 for TTR and 0.54 for log INR variability) but was weak between TTR and log INR variability (0.025) CONCLUSION: ICOMO predicts ischemic stroke better over TTR and log INR variability alone but it is only better than the latter in predicting major bleeding. The choice of which measure to use for clinic profiling changes clinic rankings considerably.
Thesis (M.S.H.P.) PLEASE NOTE: Boston University Libraries did not receive an Authorization To Manage form for this thesis or dissertation. It is therefore not openly accessible, though it may be available by request. If you are the author or principal advisor of this work and would like to request open access for it, please contact us at firstname.lastname@example.org. Thank you.