This application seeks continued support for a team of statisticians, economists, and clinicians to collaborate on the development and application of longitudinal hierarchical discrete choice models for understanding diffusion of mental health technologies and for causal inference. By developing better statistical models for understanding the dynamics of treatment adoption, or innovation, and of treatment rejection, or exnovation, researchers will gain greater insight into longitudinal patterns of usual care treatments.
In Specific Aim 1 we will extend likelihood-based approaches to accommodate a flexible family of dynamic discrete choice diffusion models. This will permit the study of the effects of patient, provider, product, and market characteristics on technology adoption or rejection.
Specific Aim 2 will extend Aim 1 to include geographic variation in the diffusion of mental health treatment technologies. This will permit estimation of geographic specific diffusion effects.
In Specific Aim 3 we will extend Aim 1 to study the causal effect of patient, provider, product, or market characteristics on diffusion. This involves the use of estimators that unconfound covariate effects while accounting for within- and between-patient, provider, product, and market variation, thereby permitting unbiased inferences. We will apply these methods to cohorts of patients with depression, bipolar affective disorder, and schizophrenia using multiple sources of data. An Advisory Board comprised of leaders in statistics, economics, and psychiatry will convene annually to validate methods and ensure integration of techniques into mental health services research. The methodological advances from this research will enable researchers, policy makers, and methodologists to better characterize factors impacting technology innovation/exnovation and to expand the inferences for usual care.
Showing the most recent 10 out of 22 publications
