Comparative Effectiveness Studies Using Large Administrative Data: Are Designs and Methods Important?

• Activity Number: 82
• Type: Contributed
• Date/Time: Sunday, July 29, 2012 : 4:00 PM to 5:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Bassam Dahman*+
• Companies: Virginia Commonwealth University School of Medicine
• Keywords: comparative effectiveness ; propensity ; matched cohort ; large data ; rare events ; colonoscopy

Abstract

Pooled cross sectional and matched cohort designs are commonly used in observational studies. Statistical methods used in comparative effectiveness studies using large administrative data include propensity score matching, and fixed effects models. We assess the underlying assumptions of these designs and methods in evaluating the causal effects of factors and covariates on the probabilities and counts of rare outcomes. We illustrate the similarities and differences in the empirical estimates of these methods by evaluating the effects of types of colonoscopies on serious complications. We extracted a study sample with over 2 million colonoscopies from the ambulatory surgery and inpatient hospital discharge datasets from the Florida Agency for Health Care Administration (1997-2004). Although with large datasets, these methods might not differ substantively empirically, understanding the conceptual differences between them is important for applying the proper method to address the relevant research question. The choice of model in specific applications would depend on the relevant questions being addressed, which in turn informs the type of design that would be relevant.