Proposed Revision of Process for Creation of Health Care Priority Status in VAM3

(Jon Jacobson, 18 November 2003)

A. The program PROCESS_INPUT will use the parameter file CPSTAT_CENSDMDC to assign frequencies of CP_STAT (CPSTAT_i for i = 1 to 22, where the sum of CPSTAT_i = 1) by RACE (2 categories, white/non-Hispanic and other), VAHUD (2 categories, low-income and other), and possibly AGE (under 65 and over 65) to veterans in the merged Census/DMDC database. CPSTAT_CENSDMDC parameter values will be estimated from 2001 National Survey of Veterans data (aggregating categories where needed to deal with small cell sizes in the NSV). CP_STAT itself will take up to 22 values, defined as follows:

1. COMP_SC_65PL = receiving comp (not pen), 65+ % degree of SC disability

2. COMP_SC_45_64 = receiving comp (not pen), 45-64 % degree of SC disability

3. COMP_SC_25_44 = receiving comp (not pen), 25-44 % degree of SC disability

4. COMP_SC_05_24 = receiving comp (not pen), 5-24 % degree of SC disability

5. COMP_SC_00_04_P3 = receiving comp (not pen), 0-4 % degree of SC disability but indications (such as POW status) placing veteran in HCP group 3

6. COMP_SC_00_04_P4 = receiving comp (not pen), 0-4 % degree of SC disability but indications (such as A&A status) placing veteran in HCP group 4

7. COMP_SC_00_04_P5_P6 = receiving comp (not pen), 0-4 % degree of SC disability and indications (such as no POW/A&A status) placing veteran in HCP groups 5 or 6

8. PEN_SC_65PL = receiving pen, 65+ % degree of SC disability

9. PEN_SC_45_64 = receiving pen, 45-64 % degree of SC disability

10. PEN_SC_25_44 = receiving pen, 25-44 % degree of SC disability

11. PEN_SC_05_24 = receiving pen, 5-24 % degree of SC disability

12. PEN_SC_00_04_P3 = receiving pen, 0-4 % degree of SC disability but indications (such as POW status) placing veteran in HCP group 3

13. PEN_SC_00_04_P4 = receiving pen, 0-4 % degree of SC disability but indications (such as A&A status) placing veteran in HCP group 4

14. PEN_SC_00_04_P5 = receiving pen, 0-4 % degree of SC disability and indications (such as no POW/A&A status) placing veteran in HCP group 5

15. OTH_SC_65PL = receiving neither comp nor pen, 65+ % degree of SC disability

16. OTH_SC_45_64 = receiving neither comp nor pen, 45-64 % degree of SC disability

17. OTH_SC_25_44 = receiving neither comp nor pen, 25-44 % degree of SC disability

18. OTH_SC_05_24 = receiving neither comp nor pen, 5-24 % degree of SC disability

19. OTH_SC_00_04_P3 = receiving neither comp nor pen, 0-4 % degree of SC disability but indications (such as POW status) placing veteran in HCP group 3

20. OTH_SC_00_04_P4 = receiving neither comp nor pen, 0-4 % degree of SC disability but indications (such as A&A status) placing veteran in HCP group 4

21. OTH_SC_00_04_P5_P6 = receiving neither comp nor pen, 0-4 % degree of SC disability and indications (such as no POW/A&A status) placing veteran in HCP groups 5 or 6

22. OTH_SC_00_04_P5_P8 = all other veterans not identifiable in augmented C&P data, and assumed to be in HCP groups 5, 6, 7, or 8 on the basis of VAHUD and PERSVC

HCP_STAT will take on the same values as CP_STAT, except: using VAHUD, CP_STAT = 7 veterans will be split into COMP_SC_00_04_P5 and COMP_SC_00_04_P6 veterans, CP_STAT = 21 veterans will be split into OTH_SC_00_04_P5 and OTH_SC_00_04_P6 veterans; using VAHUD and PERSVC, CP_STAT = 22 veterans will be split into OTH_SC_00_04_P5, OTH_SC_00_04_P6, OTH_SC_00_04_P7, and OTH_SC_00_04_P8 veterans. So the number of possible values for HCP_STAT will be 25, versus 22 possible values for CP_STAT.

B. In the program PROCESS_COMPPEN, we will have merged onto the C&P frozen mini master file for each year any other available data that can distinguish veterans by CP_STAT for a given combination of GENDER, AGE, PERSVC, POSTVNE, and STATE. The program PROCESS_COMPPEN will use the new parameter file CPSTAT_CP to attribute counts of certain non-comp, non-pen veterans who are not currently being merged onto the C&P file. This “shadowing” approach assumes that non-comp, non-pen veterans in health care priority groups 1-4 exist only for combinations of GENDER, AGE, PERSVC, POSTVNE, and STATE for which there exist some veterans with the same health care priority who are receiving comp or pen benefits. CPSTAT_CP parameters, which will be obtained from the Sullivan file, corresponding C&P data, and possibly other sources, will include, by AGE, the following ratios:

OTH_SC_65PL vets to the sum of COMP_SC_65PL, PEN_SC_65PL vets

OTH_SC_45_64 vets to the sum of COMP_SC_45_64, PEN_SC_45_64 vets

OTH_SC_25_44 vets to the sum of COMP_SC_25_44, PEN_SC_25_44 vets

OTH_SC_05_24 vets to the sum of COMP_SC_05_24, PEN_SC_05_24 vets

OTH_SC_00_04_P3 vets to the sum of COMP_SC_00_04_P3, PEN_SC_00_04_P3 vets

OTH_SC_00_04_P4 vets to the sum of COMP_SC_00_04_P4, PEN_SC_00_04_P4 vets

OTH_SC_00_04_P5_P6 vets to the sum of COMP_SC_00_04_P5_P6, PEN_SC_00_04_P5 vets

Applying these ratios will produce totals of vets, by GENDER, AGE, PERSVC, POSTVNE, and STATE, with CP_STAT = 1 to 21. These counts will be called CPVETS_i_CP for i = 1 to 21, and will be saved in the augmented C&P dataset produced by PROCESS_COMPPEN. Veterans in the CP_STAT = 22 group will not be identified in the augmented C&P dataset, but will be “netted out” from Census / DMDC totals of veterans, as indicated in step D, below.

C. The program PROCESS_INPUT will merge the Census and DMDC data by GENDER, AGE, PERSVC, and POSTVNE. The program will then use the sum of vets by GENDER, AGE, PERSVC, and POSTVNE from both the Census (DMDCVETS_CENSUS) and the DMDC (DMDCVETS_DMDC). The program will create the number of vets per cell defined by GENDER, AGE, PERSVC, POSTVNE, RACE, VAHUD, and STATE as NUMBERVETS = CENSUSVETS (= the Census count) for cells contained in the Census but not the DMDC, and as NUMBERVETS = CENSUSVETS * DMDCVETS_DMDC / DMDCVETS_CENSUS for cells contained in both the Census and DMDC. The program will then apply CPSTAT_i (created in step A, above) to the number of vets (NUMBERVETS) defined by GENDER, AGE, PERSVC, POSTVNE, RACE, VAHUD, and STATE, and sum for groups defined by GENDER, AGE, PERSVC, POSTVNE, and STATE to obtain CPVETS_i_CENSDMDC for i = 1 to 21.

D. PROCESS_INPUT will then merge onto the Census/DMDC data, by GENDER, AGE, PERSVC, POSTVNE, and STATE, the augmented C&P dataset containing CPVETS_i_CP for i = 1 to 21. The number of vets per CP_STAT category, CPVETS_i for i = 1-21, will be coded to equal CPSTAT_i * NUMBERVETS * CPVETS_i_CP / CPVETS_i_CENSDMDC for each cell defined by GENDER, AGE, PERSVC, POSTVNE, RACE, VAHUD, and STATE. CPVETS_22 will be defined as [NUMBERVETS – sum(CPVETS_1,…, CPVETS_21)]. If CPVETS_22 is negative, CPVETS_i for i = 1 to 21 will be rescaled so their sum equals NUMBERVETS.

E. PROCESS INPUT will then use CP_STAT, VAHUD, and PERSVC to split the database by the health care priority status variable HCP_STAT with the 25 possible values described above. CP_STAT and, if necessary, VAHUD will be removed from the database in order to reduce the size of the file. HCP_STAT will be used to tabulate HCP group totals for group 1A, group 1B, and groups 2 through 8, without the need for a separate attribution variable in the model database.