Geocoded County

[Jim Hofferkamp]

3 new data items related to the geocoded county are requested (by NAACCR not NPCR) for this years call for data. These are for U.S. registries only.

Data Item 94 County at DX Geocode1990
Data Item 95 County at DX Geocode2000
Data Item 96 County at DX Geocode2010

I’m hoping Recinda (our resident epidiologist with phD in this stuff) will respond to this post with an explanation of why these fields are so important. Evidently a lot of the variables that are calculated are based on County at Diagnosis. Things like Census Tract Poverty Indicator, RUCA, URIC, etc need a county code to be calculated.

What we worry about is an incorrect county code reported by the registrar. She has done some research on this comparing geocoded county to reported county and found a significant number of incorrect reported counties.

I probably shouldn’t admit this, but i may have guessed on the county code a time or two while abstracting cases. It can be a pain to have to stop and look-up a county code when abstracting.

Anyway, we do understand that these data items are new for v16. If your registry hasn’t converted to v16, you may not be able to populate them. We also know that some registries may not have time to populate these data items in time to submit for the call for data. That is disappointing, but a reality.

If this is the case, you should consider replacing the reported county (item 90) with the geocoded county based on the year of diagnosis.

I would love to hear from central registries on the pro’s and con’s of doing this.

[Recinda Sherman]

Howdy!

There are 2 reasons for being so fussy about using geocoded county codes:

1–to ensure correct calculation of county-level rates–a standard statistic reported by most central registries.
Geocoding is very useful in ensuring the correct county. It is more precise–unlike the USPS system, a geocoded county reflects the actual county of an address when a city or zip crosses county lines. And geocoding can correct miscodes stemming from a mistype or a common error like keying in the facility county instead of the patient’s county. Please note, that for small counties, using the geocoded county instead of the reported county will result in changes in rates. In general, small counties without cancer facilities will gain cases and can see higher rates using the geocoded county. Large, metro counties with large hospitals will lose cases but generally not enough to impact the rates.
NOTE: it is important to use the geocoded county for county rates only when the case has been geocoded. A registry does not want to overwrite a county reported by the facility with a blank for geocoded county–else we lose information.
2–to ensure correct tract is used for tract-level poverty code and urban/rural codes.
The tract number is only unique when combined with state and county. Tract number 001100 is a common tract number that is valid for most counties in the US. So if the reported county is wrong, the poverty codes and other SES area-based measures are also wrong. The use of these SES area-based measures have become a critical component to our understanding of the health of communities.

[Author]

Posts