TWANG is an exceptional tool for estimating the relative effectiveness of two treatments. But many pressing research questions involve more complex settings, such as comparing multinomial (i.e., more than two) treatments and studying the relative effectiveness of time-varying sequences of treatments, which many currently available large-scale substance use observational datasets can support. Also, the R environment may be unfamiliar to some researchers and analysts, creating a barrier to their use of TWANG. This proposal aims to extend the TWANG package to be more versatile and better meet the current and future needs of addiction researchers. It also aims to improve dissemination of the package. Specifically, we will
1. Extend TWANG to estimate propensity scores and assess balance for:
a. Multinomial treatments;
b. Time-varying treatments.
2. Provide access to TWANG via software other than the R environment.
3. Disseminate the updated software through a website, workshops, and webinars.
The contributions from this grant in the short and long term will be to improve both the TWANG package as a health services research tool and the statistical practices of addiction health service researchers. This grant will encourage broader use of propensity score methods through our dissemination efforts which include developing a website, workshop and webinar about TWANG which will include tutorials and case studies that illustrate how addiction researchers can use the TWANG package to draw more robust causal inferences from observational datasets. To date, while readily adopted by many other applied fields like education and behavioral research, causal inference methods have been slow to be adopted in the field of addiction research, likely a result of how difficult it can be to stay abreast of new methodological developments and the software tools available to implement them. This grant will not only improve a promising new causal inference tool but more effectively place it directly into the hands of addiction researchers.
Dr. Almirall’s role on this project will be to provide expert guidance to the project on models and estimation methods for examining the causal effects of time-varying sequences of treatment estimates. This includes providing expert guidance on the use and application of marginal structural models and structural nested mean models and the use of parametric and semi-parametric estimation methods for these models, including the use of inverse-probability-of-treatment weighting techniques. Additionally, he will contribute to all phases of the project by consulting on the design of, and updates to, the new TWANG package and the strategic outreach campaign for dissemination of TWANG. Dr. Almirall will be involved in writing methodological manuscripts and manuscripts on the application of this methodology to substance use data sets. As part of the outreach campaign, he will also be involved in delivering seminar talks and workshops related to dissemination of the methodology.