[Gnumed-devel] Fwd: Medications plug-in, refactored - *Managing* medications

[Busser, Jim]

==================
Pulling these together:
==================

1. Let us default clin_when (Started) to today(), and not now(), because the 
patient does not actually consume it now(). Often, they do not even obtain and 
start the medication for another day, and sometimes two.

2. Let us however -- with 1.4x -- *allow* clin_when to be NULL, and without a 
prompt, because it already defaults to a non-NULL value, and so if a clinician 
goes to the trouble of removing the default value, this is different than to 
have overlooked to input a value.

3. Let us allow a value for "Duration" even when the medication is "Long-term" 
because a medication intended to be long term can beg to be reassessed. 
Patients have reported symptoms that I am not yet convinced were due to the 
drug, or there may be a drug which for other reasons I am considering to 
discontinue, or replace, but about which I have not yet decided. Sometimes it 
is a question of dosage adjustment, for example in some patients with adrenal 
insufficiency or giant cell arteritis it can be important for years (or 
lifetime) that they remain on *something* and allowing a value in Duration at 
the same time as allowing the checkbox "Long-term" to remain selected provides 
dual value.

4. Whenever a value is entered for Duration / until

- let the GUI accept intervals and dates

- let the result be stored in a new column "until_date" populated as follows:

        where Duration / until is an interval

                where Started is non NULL
                --> until_date = Started + interval

                where Started is nuLL
                --> until_date = today() + interval

        where Duration / until is a date
                --> until_date = date inputted

5. Let's replace what is currently shown as "Duration / until" with two 
information "parts" separated by right-pointing arrow →

The left part would need suitable labels, but would express (when known)

        Time on therapy (Time since therapy)

and the right part would express (when known)

        Until

resulting in these examples for a drug started 7 days ago (July 29) prior to a 
visit "today" which, for me, is August 4

6.      At the start of today's encounter, seven days later, what the cell 
would display would vary with whether and what had been set for

                Duration / until_date.

        Some scenarios:

        (6a) 7 days i.e. therapy is complete
        (6b) 14 days i.e. 7 days will remain to complete
        (6c) 21 days
        (6d) no duration set

        … the display could be:

                (6a)    7 d
                (6b)    7 d → 7 d
                (6c)    7 d → 14 d
                (6d) 7 d → ?

Now let us say that, within the visit, the patient reports that they took only 
two days' worth for reasons which may or may not warrant to go back on the 
medication. If we postpone to change the plan, and first deal only with what 
this will look like if we would input this "date discontinued" and make no 
other changes, we could then see

                (6a)    2 d (- 5 d)
                (6b)    2 d (- 5 d) → 7 d
                (6c)    2 d (- 5 d) → 14 d
                (6d)    2 d (- 5 d) → ?

so this tells us that the patient took only 2 days, and last took it 5 days 
ago, and in cases (b) and (c) would have still been due to take another 7 and 
14 days.

In the case of (6a) the above did not make apparent that there was a deviation 
by 1 or more days from the plan, and so I suggest an additional indicator which 
can remain apparent at later review of inactive medications:

                (6a)    2 d (- 5 d) X
                (6b)    2 d (- 5 d) X → 7 d
                (6c)    2 d (- 5 d) X → 14 d
                (6d)    2 d (- 5 d) → ?

Now, in the case where clin_when was null, the first term in the expression

                2 d (- 5 d)

would devolve to a question mark, since we could not compute for how long they 
have been taking it:

                ? d (- 5 d)

Next, where a drug was check-boxed "Long-term", I suggest to concatenate "space 
ellipsis infinity"

                (6a)    2 d (- 5 d) X … ∞
                (6b)    2 d (- 5 d) X → 7 d … ∞
                (6c)    2 d (- 5 d) X → 14 d … ∞
                (6d)    2 d (- 5 d) → ? … ∞

which can help the clinician to recognize in every case of parentheses

                                (-5 d)

that there is a problem, whether or not the discontinuation was "per plan" 
since that plan has yet to take into account a possible precaution that if not 
*this* dosage (or if not this particular medication), the patient may be judged 
to need to be on *some* insulin (say, for their type 1 diabetes) or *some* 
prednisone (say, for their adrenal insufficiency or giant cell arteritis) or 
*some* dosage anti-retroviral or anti-viral (say, for their HIV or chronic 
relapsing herpes zoster outbreaks).

Additionally, re-purposing the data from (6) in the case where the patient did 
*not* yet discontinue a medication which was marked for long term use, we can 
see from

                (7a)    7 d … ∞
                (7b)    7 d → 7 d … ∞
                (7c)    7 d → 14 d … ∞
                (7d)    7 d → ? … ∞

that the clinician can easily see, in the case of (7a)'s missing arrow, that 
there is a problem because whatever was the planned duration of therapy has 
decided on some already-expired until_date without dealing with the fact that 
this medication remained connected to a plan for long term use. Some decision 
about what do may have failed to be taken care of. I see the ability to detect 
this as a safety advantage.

By contrast, all cases of ∞ which *have* an arrow include a plan to reassess 
this medication in 7 or 14 days, or no current plan to reconsider the 
medication.

This leaves uncovered the case where a patient was *intended* to have already 
stopped a medication. Suppose the patient had only been intended to take 5 days 
of medication. Well, at the start of the visit, we cannot make assumptions 
about what they did. Therefore, until we can confirm that they did in fact 
discontinue it, the cell should display

                (8a) 5 d + 2 d

and here is how it can be handled …

i)  they did as directed … input date discontinued …  refresh
        5 d (- 2 d)

ii)  they did as directed, except they started one day late … you alter 
Started, you input Discontinued, and you decide whether or not you wish to 
adjust your intended until_date which you might decide to do, in which case …  
display shows the following with, or without, an appended X

        5 d ( -1 d)

iii) they did *not* do as directed because they did not understand that they 
were supposed to discontinue it. Therefore you explain to them, and if they say 
they understand, you could leave until_date as it was. You might even decide to 
postpone to enter a value into "discontinued" because this patient did not yet 
prove that they had discontinued it. I might rather wait until the next visit.

-- Jim