CMS Risk AdjustmentCMS implemented the Hierarchical Condition Categories (HCC) model in 2004 to adjust capitation payments to Medicare Advantage plans for the health expenditure risk of their enrollees. CMS' Risk Adjustment materials include a lot of information such as an
The 2012 Version 12 model expects two datafiles: PERSON and DIAG.
The variables needed in the PERSON dataset from the enrollment file are the following:
The variables needed in the DIAG dataset from the MedPAR, outpatient, and carrier data are the HICNO and DIAG, the ICD-9-CM diagnosis code from (1) principal hospital inpatient (2) secondary hospital inpatient (3) hospital outpatient (4) physician, and (5) clinically-trained non-physicians (e.g. psychologist, podiatrist). CMS delivers the SAS code model files with ".TXT" extensions. By convention, SAS code usually has a ".sas" extension. The 2012 Version 12 main program V1212F1P.sas calls the main macro program V1212F1M.sas which in turn calls AGESEXNV.sas EDITICD9.sas V12H70L1.sas V12H70H.sas V12H70M.sas and SCOREVAR.sas SAS can create new variables using formats. The models use the diagnosis codes to create Condition Categories (CCs). Then the program creates hierarchical condition codes by imposing hierarchies on the CCs. CMS delivers the formats as SAS transport files and (usually) also as a text file for reference. The SAS transport files can (and have been!) converted to regular SAS data files using SAS' PROC CIMPORT. A worked example using the HCC 2014 Version 12 final model on made up input data and the data the model outputs is available. The 2014 HCC 12 model uses ICD9 codes. Made up input data files for the HCC 2016 Version 22 model are available too. The HCC 2016 22 model can use ICD9 or ICD10 codes. Jean Roth set up the Risk Adjustment materials here to make it a bit easier to get familiar with the CMS Risk Adjustment software model materials.
CMS Risk Adjustment Model Materials
The risk adjustment software and files often have names like the following:
The files beginning with "R" usually are RxHCC model files.
Contact data@nber.org with questions, comments, or suggestions.
|

Contact Us









