ESTIMATING TREATMENT IMPORTANCE in MULTIDRUG-RESISTANT TUBERCULOSIS USING TARGETED LEARNING: AN OBSERVATIONAL INDIVIDUAL PATIENT DATA NETWORK META-ANALYSIS

• Activity Number: 179 - Emerging Methods for Complex Biomedical Data
• Type: Contributed
• Date/Time: Monday, July 30, 2018 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Guanbo Wang* and Mireille Schnitzer and Andrea Benedetti
• Companies: McGill University and University of Montreal and Respiratory Epidemiology and Clinical Research Unit, McGill University Health Centre
• Keywords: MDR-TB; multiple treatments; drug resistence; TMLE; semiparametric estimation; transportability

Abstract

Multidrug resistant tuberculosis (MDR-TB) is defined as strains of TB that do not respond to at least the two most powerful anti-TB drugs. MDR-TB is often treated with multiple antibiotics. Our data consist of individual patient data from 31 international observational studies with varying prescription practices, access to medications, and the antibiotic resistance. In this study, we develop identifiability criteria for the estimation of a generalized treatment importance metric in the context where not all medications are observed in all studies. The importance metric can be interpreted as the effect of adding an additional medication to the existing treatments. We then use this metric to rank 15 observed drugs in terms of their estimated add-on value. Using the concept of transportability, we propose an implementation of targeted maximum likelihood estimation (TMLE), a doubly robust and locally efficient estimator, to evaluate the treatment importance metric. A clustered sandwich estimator is adopted to compute the variance.Simulation studies are conducted to assess the performance of our estimator and its variance estimation, and verify the double robustness property.