Premium
Measuring persistence to hormonal therapy in patients with breast cancer: accounting for temporary treatment discontinuation
Author(s) -
Huiart Laetitia,
Ferdynus Cyril,
Dell'Aniello Sophie,
Bakiri Naciba,
Giorgi Roch,
Suissa Samy
Publication year - 2014
Publication title -
pharmacoepidemiology and drug safety
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.023
H-Index - 96
eISSN - 1099-1557
pISSN - 1053-8569
DOI - 10.1002/pds.3631
Subject(s) - discontinuation , medicine , tamoxifen , persistence (discontinuity) , breast cancer , confidence interval , hormonal therapy , pharmacoepidemiology , medical prescription , demography , cancer , pharmacology , geotechnical engineering , sociology , engineering
Purpose Several studies have been conducted to estimate persistence to hormonal therapy among women with breast cancer (BC). Most studies focus on first treatment discontinuation. Patients, however, can have numerous periods of treatment discontinuation or treatment exposure. Our objective is to estimate persistence to tamoxifen in patients with BC while accounting for temporary treatment discontinuations and this by using multi‐state (MS) models. Methods A cohort of 10,806 women with BC having received at least one prescription of tamoxifen between 1998 and 2008 was constituted from the UK General Practice Research Database. We fitted a semi‐Markov model with three states to estimate the probability of being off treatment over a 5‐year period while accounting for temporary treatment discontinuations (transition between on treatment and off treatment) and competing risks (recurrence of BC or death). Results Non‐persistence, as estimated from the MS model, ranged from 12.1% (95% confidence interval [95%CI]: 9.2–15.1) at 1 year to 14.9% (95%CI: 11.7–18.1) at 5 years. Estimations of non‐persistence based on the Kaplan–Meier model were higher, i.e., 29.3% (95%CI: 28.1–30.6) at 5 years, as well as those obtained from a competing risk model, i.e., 24.0% (95%CI: 22.9–25.1). Most temporary discontinuations (94.7%) lasted less than 6 months. Temporary treatment discontinuations are frequent and should be accounted for when measuring adherence to treatment. MS models can provide a useful framework for this sort of analysis insofar as they help describe patients’ complex behavior. This may help tailor interventions that improve persistence to hormonal therapy among women with BC. Copyright © 2014 John Wiley & Sons, Ltd.