Re: [Gnumed-devel] GNUMed SOAP vs SOAPless progress notes

[Busser, Jim]

Long post -- lots of mental refactoring! 

On 2011-12-01, at 6:57 AM, Karsten Hilbert wrote:

>> Part of what makes it hard is that it is not a patient complaint but rather 
>> an abstracting of a record of an event.
> 
> Soap does not mean "Patient Complaint". It means "History
> Taken".

Well, 'S' only means subjective.

In spite of living and working in North America, and training at a medical 
school which taught both Weed's POMR as well as SOAP notes (which I cannot 
recall if they were part of Weed's construction) I cannot for the life of me 
recall ever seeing a SOAP note documented in reference to anything other than a 
direct patient interview / assessment. Never in my life have I seen an entry in 
which any part of any other note contained just an

        S: (with entry)

or an

        O: (with entry)

or likewise an A: or P: however what I have seen is

        (free text entry pertaining to new information)
        (for example by phone from the lab or one of the nurses)

followed by

        Assessment
        ----------------
        (writer's diagnosis, differential, reasoning etc)

>> Maybe labelled
>> 
>>      Non-SOAP narrative
> 
> That doesn't make sense.
> "Uncolored color"

No, I did not write or propose

                Non-SOAP SOAP

it is only your insistence that clinical narrative cannot be anything other 
than SOAP, and so cannot be permitted to exist outside a SOAP formalization, 
except to be rendered and treated as invalid or not-yet-valid. It is precisely 
why I wrote Non-SOAP (somewhat facetiously, admittedly, because an actual label 
might better be something else).

Seemingly, however, it is being implied that the doctor should be happy to see 
the label

        Non-clinical notes

and so we have an important problem here! ?

> Clinical narrative is *always* SOAP, except which may not
> have been decided yet.

Above -- where you may mean it is *conceptually* "always SOAP" -- we are 
getting closer to possible agreement …

> If narrative is not SOAP then it is not *clinical* narrative
> and needs NULL as category.

… but now we are disagreeing again.

It isn't right to judge that the reluctance to constrain entries inside a 
choice among {S, O, A, P} makes them 'non-clinical'. That is simply too rigid. 
When you think about it,

        'S' can contain information which is highly precise and stable and 
verifiable

        'O' may be neither accurate nor reproducible
        (except better so with some measurements)

        'A' and 'P' contain narrative which can easily be subjectively
        arbitrary and inconsistent.

I *do* see value in GNUmed looking to *leverage* the value of row 
categorization. I am only questioning what is guiding our choices. Surely the 
creators of SOAP intended to guide an approach to patient problems inside the 
clinical interview. Surely they never intended every clinical entry to (have 
to) be constrained inside one of only four values {S,O,A,P} as part of an 
electronic record implementation.

The value of categorization is in *meaningful differentiation*. So, what is 
*potentially* meaningful to differentiate?

1. patient's descriptions of past and current circumstances:
        activity
        events (including health care interactions)
        environment/exposures/ingestions (meds, vaccines)
        symptoms
        measurements
        resources
        relationships/preoccupations/hopes/dreams/worries ('life')

2. patient's recordings of the above (if available / importable)
        (these are potentially more reliable than memory!)
        suppose a patient provided a record of weights
        --> would we disqualify these as measurements?

3. collateral
        from other observers and documenters
        may originate from both unprofessional and professional sources
        includes records transfers

4. observations and measurements
        these may be generated inside the encounter
        these may be generated after encounter (labs etc)
        but these may also be available from all of the above!
        … the fact that the sources may be unfeasible to parse
                does not alter the reality of their legitimacy

5.  statistics (descriptive and inferential)
        height and weight can yield BMI
        risk factors can generate risks / scores
        - note that, over time, the input values (therefore outputs) can change
        - note that, over time, the formulae can also change

6. by-products of the above

        the clinical analysis itself (clinician's intellectual property)
        the expression of analysis to the patient (diagnostic explanation)
        the reference information provided to the patient (knowledge / 
education)
        the options that are offered to the patient (advice re investigation / 
treatment)
        any agreements reached with the patient (decisions, contracts)
        communication within and outside the praxis to advance any of the above

Looking at the above,

=======================================
1, 2 and 3 - descriptions, recordings and collateral
=======================================

- we are maybe proposing that 1, 2 and 3 are all 'S' despite that these 
actually contain 'O' and 'A' and 'P'

- the objections to *conceding* that they contain 'O' and 'A' and 'P' come from:
        arguments about accuracy
        (and where the info is accurate, the difficulty of
                parsing / splitting / differentially-importing)
        arguments about admissibility
                "at what point do I care about an elsewhere-generated A and P?"

- the arbiter of the above depends on how it is made computationally-relevant:

(a) the capacity to generate statistics does require the inputs to be in a 
certain form
        --> it is a question of what the clinician chooses to trust
                and what the clinician can justify to separately
                input or re-process as 'O' despite its external origin
                example: weights reported by the heart failure patient

(b) the capacity to filter depends on the adequacy of tagging

        --> a clinician might consider to input, into 'O', important abnormal 
findings
                where they had been documented by an outside clinician
                but doing so would mean a clin_when that is outside the 
encounter.
                It may be a dead-end to contemplate inputting collateral 'O' 
into narrative,
                but still-viable to enter trusted collateral measurements 
not-in-notes.

        --> a clinician might very well decide to input, into 'A' (or 'P'), an 
outside clinician's
                assessment or plan – even while undecided whether or not to be 
in agreement –
                as the means by which to carry forward the unresolved interim A 
or P

============================
4. observations and measurements
============================

- observations can be labelled 'O' despite we should understand their 
subjectivity.
- measurements can be labelled 'O' but the value in doing so is maybe more to 
facilitate filtering, than in communicating visually 'O' rows

===============================
5.  statistics (descriptive and inferential)
===============================

- these might be considered to be virtual or dynamic except at the point in 
time where they get used in clinical decision-making at which point the 
clinician could reference the data point in a progress note 'O'

- I suppose a script could 'write' a value into defined subsets of patients, 
serving to preserve the values at points in time and maybe leveraging clinical 
reminders and clinical decision support

======================
6. by-products of the above
======================

This, to me, is the biggest value in having worked this posting through. Here I 
have split them up with --- :

        
------------------------------------------------------------------------------------------------------
        the clinical analysis itself (clinician's intellectual property)
        
------------------------------------------------------------------------------------------------------
        the expression of analysis to the patient (diagnostic explanation)
        the reference information provided to the patient (knowledge / 
education)
        the options that are offered to the patient (advice re investigation / 
treatment)
        
------------------------------------------------------------------------------------------------------
        any agreements reached with the patient (decisions, contracts)
        
------------------------------------------------------------------------------------------------------
        communication within and outside the praxis to advance any of the above
        
------------------------------------------------------------------------------------------------------

The first one (the clinical analysis) is an 'A' and is, in the complex patient, 
worth to be able to filter and review distinctly

        what was I thinking?
        what was my colleague thinking

particularly in the deteriorating patient where one needs to seriously and 
properly (effectively) question whether the thinking-to-date remains congruent 
with what is going on.


The next three are related. I am mindful that it is very possible (without the 
patient present) to revise my thinking but which the patient does not yet know 
about. Also, patients receive conflicting advice even within the same praxis 
and sometimes inconsistent advice even from a single clinician. It might be 
decided that for a given disease, it is important to try to standardize the 
education that is provided to patients. It is a reason why I can see value to 
clustering these in a category that, yes, should be respected as a clinical 
category, having the purpose and meaning akin to

         Explanation / Education (including advice, but an 'A' would conflict 
with above)

Now we get to 'Plan" of which the foundation is

        Agreements reached with the patient (decisions, contracts)

despite that in most EMRs the clinicians likely clutter this up with the 
clinicians' to-dos which would better live elsewhere (which GNUmed supports in 
Waitlist). When you think about it, a good

        Plan

should need to take into account the goals, the attainable targets that have 
been realistically set but we have no place for these to live, and so they 
either get re-written from time to time in *some* of the notes and risk getting 
lost, or they remain implicit and maybe not even called to mind when the praxis 
is having a busy day and all hell is breaking loose.

That is why I am now thinking that, in addition to

        Subjective / Hx
        Objective / Px
        Assessment

we should facilitate

        Education

        Goals

                goal-setting (and revision thereof)
                        --> this is dynamic
                        --> suggest support as the episode or Problem level
                                similarly to the episode running summary

        Plan (which, in its was, is a derivative restatement of assessment)

                future education:

                        referral for education, and/or
                        educate further at next visit

                immediate or future investigation / consultation:

                        investigate immediately or after a delay, and/or
                        refer for consultation, and/or
                        assess further at next visit

                        --> the results of investigation become O
                        --> arguably consultants opinions (letters) should be O

                immediate or future prescription:

                        patient tasks re lifestyle / medication and/or
                        referral for treatment and/or

                next visit dependencies

(while contingencies might be considered part of the 'plan' they could be part 
of what the patient was Educated to do.)

We end up with Goals that need to be able to be maintained and visualized, and 
with Plans whose elements can either be active (in some cases needing to be 
arranged), and/or which can be put on hold or kept on reserve:

                education {book | refer for | do at next visit}         <input> 
 
                investigation {initiate | book | refer for}                     
<input>
                consultation {book | refer for}                                 
<input>
                prescription {initiate | refer for}                             
        <input>
                procedure {book | refer for}                                    
<input>

where the above elements each have a potential status or action requirement 
(hinted at between { }) plus of course being 'on hold'

Two final thoughts…

(1) the last of my dashed items under #6

        communication within and outside the praxis…

is really not the plan but derivative of the plan, reflecting for example a 
dependency for referrals, and instead of clinicians needing to type / input 
such things they would really better be handled as a workflow item signalled by 
a status rather than constituting additional 'content'.

(2) given that new S and new O beget evolution in the Assessment over time 
(even between visits), it suggests to me that A should (like the episode 
summary and the goals) be carried forward and dynamically revisable. The only 
question is whether it should overwrite the previous assessments, and cause the 
original to get moved into the Audit table (which I would not suggest) or 
whether it should better be cloned from the most-recently saved version of the 
Assessment and saved as the updated Assessment for the problem episode under 
consideration.

-- Jim