We do several things.
1. When abstracting, each abstractor is expected to run global edit
checks on the case before proceeding to the next case.
2. If the abstractor is either a non-CTR or is a CTR inexperienced
with our caseload, their cases are all visually reviewed by a CTR
who IS experienced with our caseload (many times, this catches
things global edits wouldn't, such as text fields not proving what
was coded or otherwise not meeting text field standards defined in
our abstracting procedure).
3. Our cases are submitted to our regional SEER registry (who in
turn will submit them to the state). There are two reports they
provide back to us which are useful (when we actually have time to
process them, that is). The first is a new cases listing. It shows
how they chose to code their data items - we usually focus on a few
fields rather than trying to verify each and every item, taking into
account that there are still a few little rule differences for SEER
and hospital registries. The second report is a correction report.
Whenever they change a data field on one of our cases, that will
appear in this report.
4. There are several fields I have found over the years are easy to
make mistakes in, and may not be checked with global edits. So I
have made a report to list those fields for groups of cases so I can
review them en mass. This is a fast and easy way to spot common
errors and patterns of errors (sometimes it may even be something
that autocodes but has an imperfect algorhithm - say an unusual
leukemia not getting grouped correctly for SSS grouping or
something).
We haven't set explicit rates, but do submit data to NCDB (and oh
how I wish their genedits program were more user friendly!). Also
the SEER registry in our area HAS set minimum error rates, so that
also is a help to us.
--- In cancerregistrargroup@yahoogroups.com, "corolla_046"
<decoeb@t...> wrote:
> I would like information about the types of quality checks done on
> registry data within your departments and if you have an
established
> error rate that must not be exceeded?