Estimating Mean Response as a Function of Treatment Duration in an Observational Study where Treatment Duration may be Informatively Censored

• Activity Number: 82
• Type: Contributed
• Date/Time: Monday, August 12, 2002 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section*
• Author(s): Brent Johnson*+ and Anastasios Tsiatis
• Affiliation(s): North Carolina State University and North Carolina State University
• Keywords: Censored Treatment ; Infusion Length ; Confounding ; Inverse Weighting ; Counterfactual

Abstract

After treatment is found to be effective in a clinical study, attention often focuses on the effect of treatment duration on outcome. In many studies, treatment duration is part of a treatment strategy which treats patients for a time {\it t}, or until a treatment-censoring event occurs, whichever comes first. Evaluating mean response for a particular treatment duration strategy from observational data is difficult because treatment duration may not be observed on all patients, because of censoring, and even if it were available for all patients, it may no longer be reasonable to assume patients are prognositcally similar across all treatment strategies. We propose an estimator for mean response as a function of treatment duration strategy, where confounding is assumed present, and treatment strategy is possibly unknown--due to early treatment termination. Our method uses potential outcomes and introduces assumptions that allow us to consistently estimate mean response through a sample of data. We evaluate our estimator through simulation studies and apply the method to the ESPRIT infusion trial from the Duke University Medical Center.