Logrank Tests with the Inverse Probability of Treatment Weighting to Assess the Impact of Beta-Interferon Treatments in Delaying Disability Progression in Multiple Sclerosis

• Activity Number: 491
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Mohammad Ehsanul Karim*+ and Paul Gustafson and John Petkau and Afsaneh Shirani and Yinshan Zhao and Elaine Kingwell and Mia van der Kop and Joel Oger and Helen Tremlett
• Companies: University of British Columbia and University of British Columbia and University of British Columbia and University of British Columbia and University of British Columbia and University of British Columbia and University of British Columbia and University of British Columbia and University of British Columbia
• Keywords: causal inference ; observational study ; inverse probability weighting ; survival analysis ; multiple sclerosis ; longitudinal

Abstract

The inverse probability of treatment (IPT) weighted Kaplan-Meier (KM) estimators have been developed to evaluate treatment effects in the presence of confounders. This approach does not require fitting of a Cox model and can therefore provide an alternative approach to the marginal structural Cox models. When weighted by unstablized IPT, the resulting KM curves have a direct causal interpretation. However, the resulting survival curves may be sensitive to the high variability of the estimated IPT weights and consequently the confidence intervals may be wide. Variability of the survival estimates can be reduced by truncation or minimal stabilization of the IPT weights. The logrank test with IPT weighting can be used to compare such survival curves. To allow emphasis on specific portions of the curves (such as, early or late differences), different relative weight functions and corresponding variance estimation techniques were explored. These methods are illustrated by investigating the association between beta-interferon treatment and disability progression in MS using longitudinal cohort data from the British Columbia Multiple Sclerosis (MS) database (1995-2008).