Training Years: 2014-2015
Training Site: Universidad Peruana Cayetano Heredia (UPCH)
Mentors: Robert Gilman, MD; Patty Garcia, MD; Bryan Weiner, PhD
Title: Complex systems: An innovative approach to improve MDR-TB treatment adherence
Aim 1: Develop theoretical framework: This aim will include a literature review, meta-analyses (if appropriate), and causal loop diagraming (CLD) to develop a theoretical framework for MDR-TB diagnosis and treatment periods. CLD is a map of a system including its components, interactions, and dynamic behavior. The objective is to identify recurring themes for barriers to treatment adherence that appear during MDR-TB diagnosis and treatment periods. Data from my MPH thesis will be revisited to strengthen the theoretical framework. System boundaries will be set and themes/constructs will be organized. The result of this aim will be a theoretical framework with identified individual and organizational level constructs while also incorporating system interrelatedness and time as integral parts. Results will be used in Aims 2 and 3.
Aim 2: Testing the theoretical framework using an observational time-variant longitudinal multilevel study to identify trends in barriers to DOT fidelity throughout MDR-TB diagnosis and treatment periods. We will use linear regression and structural equation modeling (SEM) to test the theoretical framework developed in Aim 1. SEM is a multivariate statistical method that allows for the simultaneous evaluation of direct, indirect, and total effects of qualitative causal assumptions from theory-based models using empirical data while accounting for measurement error. We will develop data collection instruments based on the findings from Aim 1 and the literature from organizational behavior and implementation sciences. The units of analysis will be the patient and the organization (TB clinic). We will be on-site to collect data monthly, for 6 months, from 10 TB clinics, and 200 patients in Lima, Peru. Sample size power will be reevaluated once Aim 1 is complete based on theoretical framework assumptions and Monte Carlo simulations. Additional data for patients and TB clinics will be collected as control variables.
Aim 3: Modeling treatment adherence using agent-based modeling (ABM) techniques and discrete rate methods to conduct simulation experiments of treatment adherence. ABM is computational modeling for simulating the actions and interactions of autonomous agents to assess their effects on the system as a whole, and can be used as a platform to test exogenous changes to the model, such as interventions designed to improve treatment adherence. The simulation will be based on the findings from Aim 1 and 2 and will be composed of patients and providers interacting with each other over time based on rules or operating policies of the system. We will begin this with a simple approximation of the process of adherence that will serve as the foundation for future model refinements as the understanding of the adherence evolves. The advantage of this simulation is that as data are added these approximations will improve and the model will more closely reflect real-life processes. We plan to build on this initial model in my future career as more data become available.
NIH Support: Fogarty scholars doctoral training award