I am working on the analyses for a manuscript I hope to submit to a leading health policy journal. I am analyzing prescription drug data associated with the U.S. Medicaid program and its expansion under the Affordable Care Act. I have prescription drug data (converted to morphine milligram equivalents) for 50 states by quarter from 2008 through 2017. I also have details on which year and quarter a state did or did not expand their Medicaid program per the ACA. This variable can be represented as either a grouping (4 categories) or a time varying covariate (I am not sure which is the better way).
The dependent variable is prescription opioid fatality rates over the same time period but available only annually.
I want to see if variation of the proportion of prescriptions reimbursed for one of two specific classes of opioids: oxycodone or hydrocodone (a separate regression model for each) is associated with changes in the fatality rates over time and if Medicaid expansion status affected this relationship.
I am not sure how to set up what looks to me to be a conditional process model with proportion of prescriptions as a mediator and Medicaid expansion as a moderator. The time varying nature of Medicaid expansion is what is throwing me the biggest curve. I need help with the statistical model setup and post-estimation analysis of the best fitting model (e.g., is a quadratic term needed to represent time).
I don't need help setting up the data. I can set up the data set any way needed. I program in Stata and it would be easiest to use that but I have StatTransfer and can convert to any major statistical package (SPSS, SAS, Mplus, or R) if you're more fluent in one of those programs.
About the recuiterMember since Mar 14, 2020 Jasola Janta Fl
from Ida-Virumaa, Estonia