Leveraging Multiple Study Designs and Statistical Methods to Evaluate Comparative Effectiveness of Asthma Medications

• Activity Number: 258 - SPEED: Causal Inference and Related Methodology
• Type: Contributed
• Date/Time: Monday, July 30, 2018 : 2:00 PM to 2:45 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Tebeb Gebretsadik* and Pingsheng Wu and Rees L Lee and Amber M Evans and Tan Ding and Nicholas M Sicignano and Ann Wu and Carlos Iribarren and Butler Melissa and Chang Yu and William Dupont and Christina Fox and Tina V Hartert
• Companies: Vanderbilt University Medical Center and Vanderbilt University and U. S. Navy and Health ResearchTX LLC and Vanderbilt University Medical Center and Health Research Tx and Harvard Medical School and Kaiser Permanente Division of Research and Kaiser Permanente and Vanderbilt University Medical Center and Vanderbilt University Medical Center and Health ResearchTx and Vanderbilt University Medical Center
• Keywords: pharmacoepidemiology; comparative effectiveness; propensity score methodology; health care systems; treatment effect

Abstract

To comprehensively evaluate the comparative effectiveness of different asthma controller medications (inhaled corticosteroids [ICS], leukotriene receptor antagonists (LTRA), long acting beta agonists [LABA] and ICS with LABA) on asthma health-care outcomes, we used integrated medical, pharmacy and eligibility administrative data files and designed three distinct asthmatic cohort studies; subjects had to have at least one year of continuous enrollment prior to the index date of interest and follow-up to one year unless they died or disenrolled. Design 1 was a new-user cohort requiring a full year of non-use of controller medications to eliminate prevalent user bias; Design 2 required an entry criteria of asthma hospitalization or ED visit to define a cohort population with asthma; Design 3 was a cohort with an index date for establishing medication use exposure and follow-up. We conducted Poisson regression and Cox proportional hazard models with i) covariates adjustment and ii) propensity score methods to evaluate treatment effects on asthma exacerbations or time to first event. Our findings elucidate the strengths and limitations of each study design and analytical approach.