question about patient/turmor model

[Bruce Riddle]

We are seeing a small but increasing number of patients who have different information for certain demographics between the 1st tumor and the 2nd tumor sometimes a decade or more later. The XML model is patient, tumor, tumor, etc. So where will different street address, city, zip, etc. go between tumor one and tumor two (or two and three)? Add in all the geocoding stuff, it gets complicated. And then, for even a smaller group, what happens when the patient changes sex and race? The group this affects the most are patients who had cancer as a child and then 20 years later pick up a second tumor.

[Bruce Riddle]

More research and figured out Part 1 of my question. Part 2 is harder. Like the SAS conversion issue,
the challenge remains on how to create an accurate analytic record that contains the correct patient
and correct tumor info.

[Fabian Depry]

Bruce,

I think your question is related to transmission versus persistence. The purpose of transmission is just to move data around (either from hospitals/labs to the Central Registry, or from the Central Registry to the Standard Setters, etc…). The purpose of persistence is to keep a consolidated view of the data at the Central Registry; I would assume that most Registries if not all use some kind of database for that. Both mechanism have a data model but those models do not have to be the same and in many registries, the persistence data model is more complex than the transmission data model because the Registry wants to keep track of more data than can be transmitted via NAACCR data files.

The primary purpose of the NAACCR XML Exchange standard is to transmit data and its data model was design with that in mind. A Registry is welcome to use the same data model for persistence if that works for them, but if not, they probably need to move to a more complex persistence data model.

Taking the address as an example: the standard defines it as the “current address” and so the transmission data model only allows one (the address at the time the abstract was created). If you want to keep track of the addresses over time in your database, you would have to use a more complex data model where you consolidate all the incoming current addresses into a list of addresses.

I understand having different models is not as convenient than using the same model for transmission and persistence because it requires some type of conversion, but in general those conversion should be fairly simple.

[Fabian Depry]

To tie your second comment to mine, I would say that your analytic record should be based on the persistence data model (the consolidated view) and not the transmission data model. If the new NAACCR XML data model works for your persistence model, that’s great; if it’s not exactly what you need, then you might need to tweak it and use some type of conversion from one to the other.

[Bruce Riddle]

A very insightful comment. The conversion will not be simple.

[Fabian Depry]

Unfortunately, I agree 🙂

Things are going to be rough for the next few years, but I am convinced in the long run, these changes will help the community moving forward in the right direction.

[Author]

Posts