6. How do I justIFY
spending proposal money for QA?
Is there benefit I can quantiFY?
Will it help me indemniFY my
conclusions?
(indemnify=guarantee)
QA may expose issues,
but does it electriFY?
9. Early discovery of issues
allows time to fix tables
Table SiteB SiteC SiteE SiteG SiteH SiteK
DEMOG 6
PROCEDURE 1 2 9
During
1. ENROLL trend analysis, 7
discovered site was
VITALS 5 8
missing many years of
DEATH 10 3
data.
LAB 4
10. Early discovery of issues
allows time to fix tables
Table SiteB SiteC SiteE SiteG SiteH SiteK
DEMOG 6
PROCEDURE 1 2 9
ENROLL 7
VITALS 5 8
2. When comparing
DEATH 10 3
distribution of procedures,
LAB 4
this site was missing some
major categories.
11. Early discovery of issues
allows time to fix tables
Table SiteB SiteC SiteE SiteG SiteH SiteK
3. When comparing death
DEMOG 6
PROCEDURE
dates to utilization dates,
1 2 9
ENROLL
many discrepancies found. 7
VITALS 5 8
DEATH 10 3
LAB 4
12. Early discovery of issues
allows time to fix tables
Table SiteB SiteC SiteE SiteG SiteH SiteK
DEMOG
4. Site, after looking at median and 6
PROCEDURE 1 2 9
units, realized they had included an
ENROLL 7
incorrect set of labs into one testtype.
VITALS 5 8
DEATH 10 3
LAB 4
13. Early discovery of issues
allows time to fix tables
Table SiteB SiteC SiteE SiteG SiteH SiteK
DEMOG 6
PROCEDURE 1 2 9
ENROLL 7
VITALS 5 8
5. This site had never built
DEATH 10 3
vitals and so could not
LAB 4
participate in this project
unless built.
14. Early discovery of issues
allows time to fix tables
Table SiteB SiteC SiteE SiteG SiteH SiteK
DEMOG 6
PROCEDURE 1 2 9
6. A high percentage of
ENROLL 7
cohort was missing
VITALS 5 8
demographic information.
DEATH 10 3
LAB 4
15. Early discovery of issues
allows time to fix tables
Table SiteB SiteC SiteE SiteG SiteH SiteK
DEMOG 6
PROCEDURE 1 2 9
ENROLL 7
VITALS
7. After viewing 5 8
DEATH
comparisons of rates of
10 3
LAB 4
insurance types site
decided enrollment data
needed to be reviewed.
16. Early discovery of issues
allows time to fix tables
Table SiteB SiteC SiteE SiteG SiteH SiteK
DEMOG 6
PROCEDURE 1 2 9
ENROLL 7
VITALS 5 8
8. DEATH comparing rates of
After 10 3
vital records across sites
LAB 4
decided they were
missing records.
17. Early discovery of issues
allows time to fix tables
Table SiteB SiteC SiteE SiteG SiteH SiteK
DEMOG 6
PROCEDURE 1 2 9
ENROLL 9. Comparing their 7
VITALS
distribution of 5 8
DEATH
px_codetypes with other
10 3
LAB 4
sites, site investigated
and found additional
mapping resources.
18. Early discovery of issues
allows time to fix tables
Table SiteB SiteC SiteE SiteG SiteH SiteK
10. rectifY: this site,
DEMOG 6
PROCEDURE 1 2
after comparing 9
ENROLL with other sites, 7
VITALS reinvestigated, 5 8
DEATH 10 found new 3
LAB sources and
4
doubled the size
of death file.
19. Early QA work makes your
analysis model more accurate
DEMOG
ENCOUNTER
ENROLL
CENSUS
CAUSE
OF DEATH
PHARMACY
20. Early QA work makes your
analysis model more accurate
Produced local individual reports:
% in project cohort vs overall site
DEMOG population.
ENCOUNTER
ENROLL
CENSUS
CAUSE
OF DEATH
PHARMACY
21. Early QA work makes your
analysis model more accurate
DEMOG
Reviewed department and
provider missingness and
distribution to determine if
combination could be used as
ENCOUNTER proxy variable.
ENROLL
CENSUS
CAUSE
22. Early QA work makes your
analysis model more accurate
DEMOG
ENCOUNTER
demystiFY: Asked sites to review
their distribution across different
enrollment plans and comment on
large variances when compared
ENROLL with other sites.
CENSUS
CAUSE
OF DEATH
PHARMACY
23. Early QA work makes your
analysis model more accurate
DEMOG
ENCOUNTER
ENROLL
Discovered percent in census
variables (such as education) have
CENSUS at least three different formats.
CAUSE
OF DEATH
PHARMACY
24. Early QA work makes your
analysis model more accurate
DEMOG
ENCOUNTER
ENROLL
CENSUS
Produced local lists of people who
CAUSE are in COD but do not have
OF underlying causetype or have
DEATH more than one.
PHARMACY
25. Early QA work makes your
analysis model more accurate
DEMOG
ENCOUNTER
ENROLL
CENSUS
CAUSE
OF DEATH
Produced local individual reports:
PHARMACY % with drug coverage
28. CodiFY: Adapt existing QA programs
for your project
*-------------------------------------------------------------------------------------
Program Name: vdw_cod_qa_local_wp01v01.sas
VDW Version: V3
Purpose: Create QA counts and statisitics for a sites VDW Cause of Death file.
Generate reports to assist in verifying or improving data.
Generate datasets to be returned and analyzed.
--------------------------------------------------------------------------------------
Dependencies :
VDW Content Areas
&_VDW_CAUSE_OF_DEATH
&_VDW_DEATH
Other Files
--------------------------------------------------------------------------------------
Folders appearing in the root directory are described below.
document, input, local_only, sas, share
--------------------------------------------------------------------------------------
document:
Contains the workplan for this program.
--------------------------------------------------------------------------------------
input:
qa_macros.sas contain the QA Macros developed by CESR DCC called in this program
--------------------------------------------------------------------------------------
29. clariFY: Early in process,
investigator meets with project staff
to define important/fixable data to QA
Important Important
Not fixable
Fixable
Not important Not important
Fixable Not fixable
32. Funding for health
research is getting tighter . . .
. . . but how can I NOT justiFY
spending proposal money for QA?
33. I vote to commodiFY QA.
I will include QA as a line item in my
proposal.
Each project of mine can do limited
and focused QA.
My project, with minimal amount of QA
funding, will help make the VDW better.
mage of investigator slipping QA into proposal. Bubble1-QA is of valueBubble2-Doing QA give me an advantageBubble3-Focused QA will be part of my project budget.