医学
药物依从性
心力衰竭
队列
药物流行病学
比例危险模型
内科学
逻辑回归
人口
药理学
药方
环境卫生
作者
Xiwen Qin,Joseph Hung,Matthew Knuiman,Tom Briffa,Tiew‐Hwa Katherine Teng,Frank Sanfilippo
摘要
Abstract Purpose There is no gold standard method to calculate medication adherence using administrative drug data. We compared three common methods and their ability to predict subsequent mortality in patients with heart failure (HF). Methods Person‐linked population‐based datasets were used to identify 4234 patients (56% male, mean age of 76), who survived 1 year (landmark period) following hospitalization for HF in Western Australia from 2003 to 2008. Adherence was estimated by the medication possession ratio (MPR), MPR modified (MPRm), and proportion of days covered (PDC) in patients dispensed a renin‐angiotensin system inhibitor (RASI) and/or β‐blocker within the landmark period. Adjusted Cox regression models that fitted restricted cubic splines (RCS) assessed the relationship between medication adherence and 1‐year all‐cause death postlandmark period. Results In the landmark period, 87% and 68% of the HF cohort were dispensed RASI and β‐blockers, respectively. Mean adherence estimates for RASI and β‐blockers were 90% and 79% for MPR, 96% and 86% for MPRm, and 82% and 73% for PDC, respectively. In RCS models, MPRm was not associated with subsequent 1‐year death in either the RASI or β‐blocker group, while MPR was independently associated with death in the RASI group only ( P ≤ .01). However, PDC as a binary variable (PDC <80% or ≥80%) or continuous variable was independently associated with 1‐year death in both RASI and β‐blocker groups (all P ≤ .02). Conclusion Proportion of days covered calculated from administrative drug data provides a more conservative estimate of adherence than MPR or MPRm and was the most consistent predictor of subsequent mortality in an HF cohort using RCS analysis.
科研通智能强力驱动
Strongly Powered by AbleSci AI