As a simple exercise, for the Structured Treatment Capture, I set up an Excel Spreadsheet with a small reference table based on a source which describes the 10 broad categories of Cancer Treatments: https://www.cancer.gov/about-cancer/treatment/
Then downloaded the v24 NAACCR XML ID Data Dictionary in CSV format and isolated all the Tumor information.
Next, using a broad category (and not entirely correct but good to first order) such as treatment, tumor description, demographic, processing information, etc
Next, isolated all the categories = "treatment"
Next, created a lookup table to match those obvious treatment variables in XML ID with the 10 categories of treatment from the reference above.
The results are provided in the Excel spreadsheet below.
This is a very crude experiment but it provides a quick sort to see how the huge list of Tumor XML ID might be sorted into containers.
The structure of such containers would be the next step, and there are issues in that a number of variables in XML ID map to many or all the categories so there are many-to-many discrepancies. But it's a good exercise to see how one could start to set up containers and to what level the contain sub-levels could be mapped.
I really do like Roger's Contain approach for the very reasons shown here with the Structured Treatment Level 0 capture test provided here.
Long-term we would need to migrate this sort of contain structure into the multi-level structures of MCODE and USCDI, but this is a first crude exercise to see how simple some parts are versus other parts which rapidly become quite complex.
Feedback welcome and appreciated.
PS in Excel, feel free to play around with the results by filtering / sorting into the different treatment types and then ponder - over a cup of coffee / tea / chocolate / ....
------------------------------
Cathy Geiger
NHSCR
------------------------------