2015 Q4 Quality Results
This quarter we added data to get at the quality of our alignment with the GPC 
standard ontology mapping and our ability to show results in two key areas, the 
data used for the breast cancer cohort within the NAACCR registry and the data 
we had predicted would be useful as a part of the height and health cohort. In 
the scripts, there were two categories of results that were available for any 
given value.

*         Either the specific value was going to be found within the GPC 
standard ontology:

o   And the value was found within the exact matching GPC ontology

o   And the value was not found within the matching GPC ontology

*         Or the site overrode the default standard ontology and specified a 
different location for the value:

o   And the value was found at a site specified non-matching ontology spot

o   but the value was not found at the site specified non-matching ontology 
spot.

Given the data pre-processing R scripts that Kansas wrote helped to point out 
ontology matching errors we would expect good matching in terms of ontology 
alignment within the NAACCR data, and in fact we did. The term matching rate 
was over 99% among the subset of fields measured and sites responding to the 
quality query. Just because the terms match, however, doesn't mean that the 
data was found there, and only 50% of the results were actually found in the 
expected spots.

For the measures selected as being generally useful for the height and health 
cohort, things were much different. The ontology matching within the databases 
dropped to 35%. However, the individual sites were better and understanding 
their data and results were found almost 72% of the time at the reporting sites.

Note: One of the latest upgrades to the GPC REDCap site made the existing data 
dictionary incompatible (a hard limit on variable name lengths) and it can no 
longer be exported and reimported without some work. This will mean some 
additional changes will have to be accommodated in the next quality run.
___________________________________________________
Thomas F. Mish Jr.  -   Informatics Systems Specialist
SMPH-IT    Biomedical Informatics Services
School of Medicine and Public Health, UW-Madison
[email protected]  -  (608)616-0362

_______________________________________________
Gpc-dev mailing list
[email protected]
http://listserv.kumc.edu/mailman/listinfo/gpc-dev

Reply via email to