We obtain Medicaid participation records from the Centers for Medicare and Medicaid Services (CMS). States submit their data to the CMS quarterly. Individuals are included in the file if Medicaid covered them for at least one day during the quarter. No personal identifiers are included on the records used by the SAHIE program.
We aggregate the data to the state and county levels by age and sex.
The counts of individual participants in the Medicaid data seem erroneous in some counties. Some counties have missing data and some counties have data that seem implausible. In either of these cases the Medicaid count is rejected. The rejection criterion relies on a regression-based approach. The predictors are: the population, the numbers of tax poor exemptions, and the numbers in poverty as measured by Census 2000. To form estimates for these counties, it is necessary to use a reasonable substitute for the rejected data. A statistical match is used to replace the rejected Medicaid counts; these values are treated in the same way as the original data. This approach works well because there are few enough counties for which data are rejected.
We combine the Medicaid and Children's Health Insurance Program (CHIP) participation data together into one variable for use in our models.
For more information see the CMS Medicaid page.