Re: [Gnumed-devel] clinician input wanted: how to implement "coding"

[Jim Busser]

On 2010-06-18, at 8:15 AM, Jim Busser wrote:

> it needs to be easy to see, at a later visit, which diseases and conditions 
> (problems in the list) did not yet get coded and for this to be easily 
> identifiably different from those already coded.
> 
> Ideas:
> - prebuilt query to find patients somehow attached (related) to the current 
> user who lack any codes
> - within-patient
>       - (1) visual indicator that is always in view – or is that distracting?
>       - (2) "peek" view that exposes what is present and missing, which 
> disappears on release of mouse or key?
>       - (3) button that *alters* the view to present a different view that 
> relists the problems as
>               problem 1 as original text
>                       1st of n codes attached, plus word equivalent
>                       2nd of n codes attached...
>                       ...
>               problem 2
>                       etc
>       - (4) button or plugin providing a different view
>               problem 1 as code, word equivalent
>               problem 2 as code, word equivalent
>               --> this would have to be understood to show only 1 of n per 
> problem
> - what would a preferred work area look like for the addition and editing of 
> codes?
> - from what (multiple) contexts should such a work area be made accessible?
> - would it be possibly useful / important to make or allow the codes to 
> appear with the problems in the list views and in the request for 
> consultation letter?

Just rethinking earlier thoughts about the purposes (value) of coding… which 
ought to be what directs our development.

*******************************
One obvious practical consideration will be when billing, where one (or more 
codes) may pass into a "billed item" and over to the accounting / billing 
software. A user would want to select from the one (or more) codes inputted 
into i.e. associated with the set of problems for the encounter to be billed.

"Catching" what has "yet-to-be" billed is most easily achieved, on a per-doctor 
basis, by an inbox item aggregating all encounters (or encounters of a certain, 
billable type) for which no associated billed item(s) have been created.

*******************************

Another practical consideration is statistical analysis of clinical aspects of 
the praxis. It may be useful to know, in any interval of time, how many 
encounters were delivered, and for what range of coded problems. OTOH when a 
patient has diabetes and hypertension, and when I had *already* (for that 
patient) associated codes for *each* of those problems into GNUmed, it should 
seem to me enough that I had already done so. When I create new notelets for 
each of those problems, I should not want to have to re-input (every time, into 
new soap rows) the same codes over and over and over again.

I am foreseeing different implications for diagnostic vs therapeutic codes. 
Once a patient with hemochromatosis has been diagnosed, that diagnostic code 
remains applicable at every subsequent encounter. However, the patient does not 
get a phlebotomy at every encounter. A user might thusly input

        diagnostic codes only into the SOA fields, and
        therapeutic codes only into the P fields

Of course, it is possible to plan to do something that is diagnostic, like a 
lab test or an imaging study, but that is still a "thing" that is done on (or 
to) the patient. If the diagnostic *test* later provides a diagnosis, that 
diagnosis can be inputted into the next encounter.

What does this mean for the handling of the codes? Does it mean that

- when inputted into the Notes editor, the codes are associated with
        - the narrative rows for the encounter, and/or
        - the episode

- but that in order to associate these same codes to a health issue, the code 
records replicated, using an fk to the health issue?

- and once the codes have been associated to a health issue, would opening the 
health issue into a notelet pre-populate the notelet with the already-known 
codes?