Yinzhu Jin1, Joan E Landon1, Whitney Krueger2, Alexander Liede2, Rishi J Desai1, Seoyoung C Kim3,4. 1. Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women's Hospital and Harvard Medical School, 1620 Tremont Street, Suite 3030, Boston, MA, 02120, USA. 2. AbbVie Global Epidemiology, Pharmacovigilance and Patient Safety, North Chicago, IL, USA. 3. Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women's Hospital and Harvard Medical School, 1620 Tremont Street, Suite 3030, Boston, MA, 02120, USA. sykim@bwh.harvard.edu. 4. Division of Rheumatology, Inflammation, and Immunity, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA. sykim@bwh.harvard.edu.
Abstract
BACKGROUND: Previous observational studies utilizing administrative claims data have largely been unable to consider clinical factors that may be related to patterns of drug use among patients with rheumatoid arthritis (RA). OBJECTIVE: To understand predictors of treatment changes following initiation of a tumor necrosis factor inhibitor (TNFi) using nation-wide electronic health record (EHR) data in the USA. METHODS: The Optum Immunology Condition EHR data (01/01/2011-09/30/2019) was used to identify a population of adult patients with RA initiating a TNFi as the first line biologic disease-modifying anti-rheumatic drug (DMARD). The primary outcome was any treatment change during the 1-year post-index period defined as cycling to a different TNFi or switching to non-TNFi biologic or targeted synthetic DMARDs. Secondary outcomes were the individual components of TNFi cycling and switching, examined separately. To identify predictors of DMARD treatment changes, we used a least absolute shrinkage and selection operator (LASSO) regression model. Model c-statistics and odds ratios (ORs, 95% confidence intervals (CIs)) of predictors were reported. RESULTS: We identified 24,871 patients with RA who initiated a TNFi. The mean age was 55.5 (± 13.7) years and 77.2% were female. Among the TNFi initiators, 22.2% experienced TNFi cycling or switching during the 1-year follow-up time. Predictors that are associated with higher likelihood of TNFi cycling or switching included female gender (OR: 1.26, 95% CI: 1.16-1.36) and glucocorticoid use (OR: 1.30, 95% CI: 1.21-1.40). In contrast, inflammatory bowel disease (OR: 0.62, 95% CI: 0.48-0.78), psoriasis (OR: 0.82, 95% CI: 0.70-0.95), recent use of methotrexate (OR: 0.89, 95% CI: 0.81-0.97), and vitamin D intake (OR: 0.92, 95% CI: 0.85-0.99) were negatively associated with TNFi cycling or switch. CONCLUSIONS: Gender, glucocorticoid use, inflammatory bowel disease, psoriasis, and vitamin D intake were identified as significant predictors of TNFi cycling or switching for TNFi initiators in the RA population. Predicting treatment change remains challenging even with large detailed EHR data. This study aimed to identify key determinants of treatment changes among patients with rheumatoid arthritis (RA) initiating a tumor necrosis factor inhibitor (TNFi) as their first-line biologic disease-modifying antirheumatic drug (DMARD) in routine care settings using a US nation-wide longitudinal electronic health record (EHR). Among 24,871 patients with RA who initiated a TNFi, 22.2% experienced TNFi cycling or switching during the 1-year follow-up time. Female patients and those who used glucocorticoids were more likely to experience TNFi cycling or switching, whereas inflammatory bowel disease, psoriasis, recent methotrexate use, and vitamin D intake were negatively associated with the outcome. However, predicting treatment change remains challenging even with larger detailed EHR data potentially due to unmeasured factors such as prescriber's preference or patient's belief in the medication.
BACKGROUND: Previous observational studies utilizing administrative claims data have largely been unable to consider clinical factors that may be related to patterns of drug use among patients with rheumatoid arthritis (RA). OBJECTIVE: To understand predictors of treatment changes following initiation of a tumor necrosis factor inhibitor (TNFi) using nation-wide electronic health record (EHR) data in the USA. METHODS: The Optum Immunology Condition EHR data (01/01/2011-09/30/2019) was used to identify a population of adult patients with RA initiating a TNFi as the first line biologic disease-modifying anti-rheumatic drug (DMARD). The primary outcome was any treatment change during the 1-year post-index period defined as cycling to a different TNFi or switching to non-TNFi biologic or targeted synthetic DMARDs. Secondary outcomes were the individual components of TNFi cycling and switching, examined separately. To identify predictors of DMARD treatment changes, we used a least absolute shrinkage and selection operator (LASSO) regression model. Model c-statistics and odds ratios (ORs, 95% confidence intervals (CIs)) of predictors were reported. RESULTS: We identified 24,871 patients with RA who initiated a TNFi. The mean age was 55.5 (± 13.7) years and 77.2% were female. Among the TNFi initiators, 22.2% experienced TNFi cycling or switching during the 1-year follow-up time. Predictors that are associated with higher likelihood of TNFi cycling or switching included female gender (OR: 1.26, 95% CI: 1.16-1.36) and glucocorticoid use (OR: 1.30, 95% CI: 1.21-1.40). In contrast, inflammatory bowel disease (OR: 0.62, 95% CI: 0.48-0.78), psoriasis (OR: 0.82, 95% CI: 0.70-0.95), recent use of methotrexate (OR: 0.89, 95% CI: 0.81-0.97), and vitamin D intake (OR: 0.92, 95% CI: 0.85-0.99) were negatively associated with TNFi cycling or switch. CONCLUSIONS: Gender, glucocorticoid use, inflammatory bowel disease, psoriasis, and vitamin D intake were identified as significant predictors of TNFi cycling or switching for TNFi initiators in the RA population. Predicting treatment change remains challenging even with large detailed EHR data. This study aimed to identify key determinants of treatment changes among patients with rheumatoid arthritis (RA) initiating a tumor necrosis factor inhibitor (TNFi) as their first-line biologic disease-modifying antirheumatic drug (DMARD) in routine care settings using a US nation-wide longitudinal electronic health record (EHR). Among 24,871 patients with RA who initiated a TNFi, 22.2% experienced TNFi cycling or switching during the 1-year follow-up time. Female patients and those who used glucocorticoids were more likely to experience TNFi cycling or switching, whereas inflammatory bowel disease, psoriasis, recent methotrexate use, and vitamin D intake were negatively associated with the outcome. However, predicting treatment change remains challenging even with larger detailed EHR data potentially due to unmeasured factors such as prescriber's preference or patient's belief in the medication.
Authors: Theresa M Hunter; Natalie N Boytsov; Xiang Zhang; Krista Schroeder; Kaleb Michaud; Andre B Araujo Journal: Rheumatol Int Date: 2017-04-28 Impact factor: 2.631
Authors: Ahva Shahabi; Jason Shafrin; Lauren Zhao; Sarah Green; Tammy Curtice; Alexander Marshall; Damemarie Paul Journal: J Med Econ Date: 2019-02-12 Impact factor: 2.448
Authors: Abiola Oladapo; Jamie C Barner; Kenneth A Lawson; Suzanne Novak; Karen L Rascati; Kristin M Richards; David J Harrison Journal: J Manag Care Spec Pharm Date: 2014-07