Lisa J Herrinton1, G Thomas Ray2, Jeffrey R Curtis3, Jashin J Wu4, Bruce Fireman5, Liyan Liu6, Robert Goldfien7. 1. Research Scientist for the Division of Research in Oakland, CA. lisa.herrinton@kp.org. 2. Senior Data Consultant for the Division of Research in Oakland, CA. tom.ray@kp.org. 3. William J Koopman Endowed Professor in Clinical Immunology and Rheumatology at the University of Alabama at Birmingham. jcurtis@uab.edu. 4. Director of Dermatology Research and Associate Residency Program Director for the Department of Dermatology at the Los Angeles Medical Center in CA. jashin.j.wu@kp.org. 5. Statistician for the Division of Research in Oakland, CA. bruce.fireman@kp.org. 6. Data Scientist for the Division of Research in Oakland, CA. liyan.liu@kp.org. 7. Chair of the Chiefs of Rheumatology for The Permanente Medical Group in Richmond, CA. robert.goldfien@kp.org.
Abstract
BACKGROUND AND OBJECTIVES: Comparative safety studies typically use hierarchical treatment categories that lump monotherapy and combination therapy. The consequence of this approach on study results is not clear. For example, studies of tumor necrosis factor inhibitors usually lump users regardless of whether they are using the drug alone or in combination with other agents. This study explored the importance of lumping vs splitting users of monotherapy and combination therapy. We also explored whether the timing of disenrollment from Health Plan membership was informative as an outcome variable when interpreting unmeasured, time-varying confounding. METHODS: This observational cohort study included Kaiser Permanente Northern California 2003 to 2013 members with rheumatoid arthritis who started methotrexate. The study end point was a major cardiovascular event. In Cox proportional hazards analysis, we compared treatment classifications using five lumped categories with treatment classification using nine split categories. We also studied disenrollment as an outcome. RESULTS: Among 5885 patients, 238 experienced serious cardiovascular events during an average follow-up of 4.25 years. Analysis of drug treatments using 5 lumped categories was difficult to interpret because treatment effects and drug users were mixed. In contrast, analysis of 9 drug categories that split monotherapies from combination therapy was easier to interpret, although confidence intervals were wider. Analysis of drug treatment in relation to disenrollment provided useful information with which to assess study validity, although the power of the analysis was limited. CONCLUSION: In comparative safety studies, we recommend greater transparency in classifying treatment and evaluating disenrollment.
BACKGROUND AND OBJECTIVES: Comparative safety studies typically use hierarchical treatment categories that lump monotherapy and combination therapy. The consequence of this approach on study results is not clear. For example, studies of tumor necrosis factor inhibitors usually lump users regardless of whether they are using the drug alone or in combination with other agents. This study explored the importance of lumping vs splitting users of monotherapy and combination therapy. We also explored whether the timing of disenrollment from Health Plan membership was informative as an outcome variable when interpreting unmeasured, time-varying confounding. METHODS: This observational cohort study included Kaiser Permanente Northern California 2003 to 2013 members with rheumatoid arthritis who started methotrexate. The study end point was a major cardiovascular event. In Cox proportional hazards analysis, we compared treatment classifications using five lumped categories with treatment classification using nine split categories. We also studied disenrollment as an outcome. RESULTS: Among 5885 patients, 238 experienced serious cardiovascular events during an average follow-up of 4.25 years. Analysis of drug treatments using 5 lumped categories was difficult to interpret because treatment effects and drug users were mixed. In contrast, analysis of 9 drug categories that split monotherapies from combination therapy was easier to interpret, although confidence intervals were wider. Analysis of drug treatment in relation to disenrollment provided useful information with which to assess study validity, although the power of the analysis was limited. CONCLUSION: In comparative safety studies, we recommend greater transparency in classifying treatment and evaluating disenrollment.
Authors: Lennart T H Jacobsson; Carl Turesson; Anders Gülfe; Meliha C Kapetanovic; Ingemar F Petersson; Tore Saxne; Pierre Geborek Journal: J Rheumatol Date: 2005-07 Impact factor: 4.666
Authors: Vijay U Rao; Andrey Pavlov; Micki Klearman; David Musselman; Jon T Giles; Joan M Bathon; Naveed Sattar; Janet S Lee Journal: Arthritis Rheumatol Date: 2015-02 Impact factor: 10.995
Authors: Jeffrey D Greenberg; Joel M Kremer; Jeffrey R Curtis; Marc C Hochberg; George Reed; Peter Tsao; Michael E Farkouh; Adeel Nasir; Soko Setoguchi; Daniel H Solomon Journal: Ann Rheum Dis Date: 2010-11-24 Impact factor: 19.103
Authors: Sarah L Westlake; Alexandra N Colebatch; Janis Baird; Patrick Kiely; Mark Quinn; Ernest Choy; Andrew J K Ostor; Christopher J Edwards Journal: Rheumatology (Oxford) Date: 2009-11-27 Impact factor: 7.580
Authors: Ernest Choy; Kandeepan Ganeshalingam; Anne Grete Semb; Zoltán Szekanecz; Michael Nurmohamed Journal: Rheumatology (Oxford) Date: 2014-06-06 Impact factor: 7.580