OpenMRS Design Forum 2018-07-25

this call is being recorded Rebekah's
our link that we used to review anything
what yeah please might want to do for
people for the top post to tell okay
guys hey guys you're on I might or
possible I can hear a lot of background
noise
with that did you just hear the
background noise go away it's not really
a constant it's just the hustle and
bustle of the conference room all right
mute for now
hey Brooke hello how are you doing doing
pretty well thanks for joining
might just be good so far it's you and
Newton and me yeah
Oh Newton is the rest of my team that's
working on this project hey Burke hello
I recognize that voice what's going on
so we have a semi large team from IBM
working on oncology here well welcome
thank you
I don't remember if it was confirmed
birth did we confront we confirmed
though assume that we can use this
design call the pack but ontology and
order entry yeah yeah I mean I it was
requested and nothing had been announced
so I put it on talked and announced it
it's the topic so like that I just did
that a few hours ago so so I mean we've
just introduced Center know if you want
to wait for others to join but just to
introduce the call I think as they've
said we've got we've got a really lucky
to have a team and join us for for a
couple weeks and in order to support
building out oncology chemotherapy order
entry and and administration on top
instead of the latest open arrest data
model we've done this several years ago
for everyone implementation and we built
it on the one nine dated model the time
and that resulted in the order extension
module and that has sort of become a bit
of a votes it was never really merged
back into core in the same way and so
that's sort of hanging out there so we
wanted to sort of use the opportunity to
to redo it in a way that's consistent
with the current core data model and the
order groups implementations that have
been added and more most recent versions
of core and so much like those
conversations that we were involved with
a couple years ago with we have we have
near the same questions about how to fit
some of these data elements into the
into this new data model on the right
way to go I think we have you know we
have some proposals and we just wanted
to send it to check them you know with
the group here and see if there any
other other thoughts before we move
forward
it sounds good so and you know great to
hear this I think you know just as a
reminder to all of us let us try to
focus on how we can help them most
efficiently and not fall into some
anyway I know that we can talk for years
about other entry because behind we will
try to be will try to give you answers
that are actionable I appreciate that I
mean you know we've tried to do our
homework at a time and you know read a
bunch of previous talk post and design
topics and on the subject and I think we
are we already largely know I think we
know where most people's opinions will
high in terms of the way that should be
designed and so we've taken that into
account without proposals but I think
there's still two questions that have
come up that we wanted to just validate
and make sure that it was you know it
made sense you know areas where it made
sense to use for example attributes
versus extent in our data model or if
there's more creative ways to think
about modeling something or those who
have experience with other order entry
systems like an aunt ather or registry
or whatever if they have if you guys
have experience with the logician you'll
actually be model those encounters and
observations that are associated with an
order versus on the order of the order
group directly but that's kind of
information that we were hoping together
so I think you know the the team who
were working with they're all from IBM
they're doing a IBM health core
assignment which you know P is applied
to all that many other groups to in
order to try to move forward with with
oncology systems and open em arrests and
so PIH others from around the open
interest community including John black
resolving collaborating on building a
new sort of oncology module and the
first piece of that is chemotherapy
order entry that we're trying to
accomplish and so I don't want to
monopolize the call you guys have heard
me talk with this a lot but I just
wanted to introduce the team Rebecca who
is who's on the call in the room in the
conference room
on the Newton line
isn't doing a lot of thinking and
designing that around do it we should
model this so I want to give her space
to talk and because that and then I'll
chime in from there that make sense that
sounds good to me I can just get started
if everyone's if we've got everyone we
think will have I think this is probably
a forum at this point of course it
requires Google Chrome for me to share
my screen totally we join in and I can
share my screen okay well I've got a
bunch of yeah oh you come out of
different yeah yeah a couple different
diagrams I want to show so okay hello
can you all see my screen oh and it
requires an extension so I'm just I have
to download that now viewing business
but sharing this okay can let me know
when you can see my scoops we're good
okay great um so I've got a couple
different diagrams that I'll kind of
talk through but I'll just start off on
kind of the top post where I posted a
lot of information so like Mike said
we're working on chemotherapy ordering
and kind of where we started with with
trying to figure out what what data
objects we want to use to model out the
different portions of chemotherapy
ordering so the way that we are
intending to order it right now a
regiment template so we like to use
chopped as an example so chopped is
would be an order set the order set then
has order set members and the order set
members would then represent each
medication within the regimen and that
would have the order set member would
then have all
the information on what's the usual dose
how is it prescribed all of all of that
information as well as the order set we
would then want to represent kind of the
normal number of cycles in the regimen
kind of a default for doctors to go by
as well as the usual length of the
regimen Rebecca can you give a like a
view of what chopped actually looks like
in terms of what that orders that
contains because you know there's
elements to it that we're going to ask
about modeling that isn't easy to view
in this design so if for example like
the notion of the notion of pre
medications chemotherapy post
medications those categories and the
notion of cycles and things like that
everyone on the call is familiar with
these but just the context I'll bring up
really quick um so so yeah this is the
information that we would want to be
modeling in that regimen template so the
each of these medications as well as
which ones are pre medication so that's
the top box up here and which ones are
post medicate or which ones are the
actual team you know men a little bit
it's it's a little black yeah I
definitely can and being a little slow
so I apologize yeah so working like just
yeah so yes we would want to be
representing each of these medications
as an order set member along with and
for all of the information associated
with them and whether or not they're pre
medications or chemotherapy medications
or if some regimens might have post
medications as well so then in the
actual doctor workflow so all of these
regimen templates would exist in the
database as order set when a doctor goes
into a patient and actually wants to
order one of these regimens for a
patient they would be able to see all of
the
regimens that are in the database select
a regiments a select chop from the list
change anything about that regimen they
like they could so there is you know a
default number of cycles they might want
to increase that or decrease that they
should be able to change that they
should be able to you know remove
medications change the dosage basically
have a lot of flexibility with those
regimens and then actually order that
for the patient the way we would then
represent it is as an order group so the
order group would contain individual
orders which would represent each of
those medications so going back to this
once it's in Stan she ated for a patient
each line in this would be an individual
order grouped within an overall order
group so then we would have information
on the overall regimen such as the any
of the information that they change so
if they did change you know the number
of cycles in the overall regimen we were
intending to store that on the order
group as well as the cycle number and
information like that
any questions there okay so now you kind
of have an idea of how we were intending
to model it looking at kind of our use
case and the information that we need
for our haiti case these are kind of the
fields that we've noticed the data model
doesn't really have a clear place for um
so kind of starting off at the top order
set we would like some some field that
told us what it was usually used to
treat so for our case all of the ones
that all of the order sets we would we
would be representing would be
chemotherapy but if in the future order
sets are someone puts in HIV order sets
or tuberculosis templates then they
would have the flexibility
to then edit that category and it would
be very useful for filtering on the UI
side figuring out which ones are
chemotherapy to display those at the
proper time we would also like I said
like to have kind of default number of
cycles typical and a regimen and the
usual length of cycles in the regimen so
if it's three weeks twenty-eight days
etc um the way we would do it we want to
do it on a stop there and just review
that or is that suggestion first this
weekend yes that was yeah that was kind
of my next spot so um so the way we were
intending to represent that is we were
thinking that a lot of those made sense
as attributes so we were planning on
extending the data model to add you know
an attribute table an attribute type
table for order set so that we could
have category number of cycles length of
cycles as attributes any thoughts yeah
so the amana thing and I think Mike has
heard this from me multiple times and it
this is Berk by the way the thing that I
am I'm trying to picture is how do we
how do we apply you know meet your needs
in a way that doesn't make the order
entry service in openmrs
chemotherapy order entry service a
rather says okay there's a lot of common
functionality amongst how you want to
create order sets and be able to order
them from HIV regimens to chemotherapy
to you know a protocol for someone
coming to the hospital with
cholecystitis or something that there's
there's a myriad of ways people want to
use and typically use order sets and
chemotherapy is one use case a lot of it
is common but things like doing cycles
of therapy is it's fairly specific to
chemotherapy and a few you know a few
other very kind of special specialized
approaches
ordering and so so that's where when I
look at this I think well do we do we
build in cycling into the data model and
all orders have cycles you know as part
of them and you just ignore it for 80 or
90 percent of use cases or do we you
know can we accommodate those in a way
that we say okay we have the notion of
being able to order or do order sets and
and group orders as they're being
ordered and do a lot of this tracking in
terms of you know using attributes etc
and then and then an oncology module
basically takes the kind of base thing
and leverages what's there to meet its
needs and then maybe adds a little bit
of icing around the cake that is
chemotherapy specific that might be just
specific to that module and this is the
part where Darius was saying that we
could talk for hours and so I'll stop I
just to give you my perspective I look
at this and I think something like
categories or general descriptions of an
order set makes total sense as being
kind of a generic feature I think when
you get to number of cycles in a regimen
or how a length of a cycle I think might
you know my gut is that either as a
feature of the module or it's it's we we
leverage that capability of like order
set attributes or something to to allow
for those but there's also the option of
saying well weeks in the model and we
just have you know when we could do
chemotherapy we had a few columns and
those are therefore if you're doing
chemotherapy and then when we get to
other needs we add some more columns and
add some more columns and and then you
just kind of you know end up with an
object that has you know and they
extensive a number of attributes and you
just kind of pick and choose based on
your use case but that to me feels a
little dirtier in the long run one one
question I have this is Henry have we
thought or does openmrs have the concept
of role restricted order sets because
they're all sort of orders with it we
know we don't have we don't have order
sets yeah
so okay so with model demand yeah we
have an implement so because I was
thinking for instance at our Medical
Center I can't order a chemo right if I
try to order chemo it would just simply
not exist for me and which obviously
case also typically we would do that
yeah typically we would approach that by
order sets just like any other order
just like any of the individual drugs or
referral or anything else would be or
durables and then you might say well we
want to be able to constrain order bolts
by the person's role or some attributes
of the user we haven't we haven't gone
there but that's we'd probably approach
it at the kind of or durable level which
would work just as well for order sets
as it would for saying that you know
there's an individual chemotherapy drug
but that you know you shouldn't be able
to order but the oncologist case so um
so Burke so I mean yes one question came
out their discussions was you know the
notion of attributes is a way of sort of
extending verses to say sub-classing for
example you know I mean there's no
there's not really a reason why we
shouldn't have like chemotherapy
actually most every order set or the
notion is something that the
chemotherapy you know in in a core
medical you know system data model so I
just I want to you know even if it is a
relatively specialized thing we could be
something stopping a traumatic so I
guess this is very another coup as to I
mean it may not be that you have like
you know 50 columns on one table and you
have to infer which columns are right
for what's use case you could have you
know you could subclass in different
tables that are joined in and only
applicable for those social classes with
objects like you know chemotherapy
orders I could just have the extension
of order sets but I don't know yeah I
mean I know those things can be more
tricky to do and it's a hibernate level
potentially but how do we given any
thought to
that approach and I think for me the
attributes are or have worked well for
us I think but they also introduced a
level of complexity where you still want
a hard code in any particular attributes
that you want so it has to be all this
configuration where you have to specify
well what what are the attributes that
are going to code around and where they
defined in global properties or some
other configuration file etc yeah just
took another step yeah sorry to go back
a step it actually does occur to me that
something like a repeating number of
repeats seems plausible for you know
fitting into the for actually living in
order set I don't know the length of the
cycle repeated annotations layers in
fact yeah I mean if it refers to not
just chemotherapy but there's some other
kind of medical orders this has the same
idea I could see building it in
certainly attributes are fine
I resign one go ahead about category I
was going to mention you might also
consider depending on what your UI use
case well first up I guess I wonder what
what had previously been proposed in a
larger view of order sets to be only a
small bits are implemented but maybe
order set tags there's a little bit more
flexible maybe and if that's good enough
for how you basically I mean category
means a single category you can have a
single category where sleep tags you can
have many and I could imagine maybe why
I might want to be driven by that for
some reason like diabetes a bag like
diabetes like they referred to location
telling you man not like Burke's notion
of tag said it more like a traditional
tag that our user user driven but like
previews location tags where it's more
flexible categorization we can have
multiple multiple search yeah I met my
god not user driven it would be
administrator configured whether it
whether there's like a tag table and a
tank map table I don't really have an
opinion but just yeah that you could
have the administrator whoever's
configuring things could add multiple
tags as opposed to just sending a single
category yeah I mean I I'm curious as to
what people thank you I mean you if you
pointed this at concepts for example and
that concept with the set that could be
a set of category concepts but then you
get translations and you know metadata
management sort of for free with that I
don't mean probably been debated you
know
I'd not give another at other points but
all right yeah I guess I do everything
else have an opinion on that yeah I do
you know there are notions of tests that
repeat but usually there that's like a
frequency and you're saying do this you
know every Monday Wednesday and Friday
or something like that
the I'm sure there's other non
chemotherapy examples of you know
repeating a test like every two months
or something like that
that might fit more into the cycle
notion that cycles is is pretty specific
to chemotherapy but I I think the notion
of saying hey chemotherapy a lot of
people aren't needed let's build the
that as an extension rather than using
attributes is perfectly fine and then
the question is is that something that
that needs to be in core because because
other other parts of the application
need to be able to know
tima therapist if ekat rebukes or is it
something that that really the only
thing that's going to be managing cycles
and doing things with cycles would be in
the in ecology module although certainly
i would say that you know someone who
admits a lot of patients with who are
actively on chemo it's certainly helpful
to know where they are in their cycles
yep right so knowing that a patient's on
cycle five of chopped is certainly
valuable information to a general
internist or hospitalists or whoever's
using this so I mean we are driving we
are good but we are going to capture the
data the question is where we capture
the data in the data model so I think
okay I think that sort of sorry did you
guys have a chance to find check whether
like you do find any examples of
oncology order sets represented in fire
so the only thing I really found was the
the order template which I think Mike
said you also found I didn't really find
I couldn't really find anything around
actually using it to order look at that
template oh yeah so I don't know a whole
lot about fire but they were
representing templates I mean kind of
similarly to how we were with a a plan
definition kind of as the overall object
of the template and that had information
such as the number of cycles and the
length of cycles and then within the
plan definitions was were activity
definition and there was an activity
definition for each medication yeah
that's why I would assume that that you
know plan the care plan stuff or
something like that would would be where
fire would put this kind of stuff handle
handling protocols because I mean
chemotherapy regimens like this are the
bracelet protocols yeah so to summarize
what I think we are so far certainly you
would be you could add attributes for
order sets and that would be shortly
fine for this purpose or another it
seems like if you wanted to create you
know if you if it turns out it's
actually straight forward and
hibernating it doesn't like cause
to have a subclass of order set that
adds the cycle stuff that sounds like
something you could do either in court
or module and then just like that in
your category suggesting like a cyclical
order set there is to mean subclass
three times there's no Union or
chemotherapy or - or something I mean I
think either I had originally I'm taking
cyclical order stead of repeating order
set but cyclical order said you know
clear my score but there's just putting
them in who wanted but versus just
putting them in order set to begin right
again I would assume that cyclical
orders that would extend the order set
with another table who just adds not a
couple of columns um so it's still be an
order set yep okay it's - I don't and we
haven't done a lot with polymorphism and
hibernate really in open LS so if you
guys want to take that path I think it's
perfectly fine I don't know what the
technical implications so much but you
could basically you can do that you
could do attributes I'm for category I
think that either category cut away or
something like that is fine it would
help to find Burke's original proposal
for WaterSense which might have been a
wiki page long ago and I thought he put
indication on there but I can't I can't
I have to look back okay I'm sure I had
everything I ran that kitchen sink
yeah but you know he might have had an
opinion about what that category or that
column should be called but the
alternative would be to do tags you know
in the original design for order sets we
also allow daughter
we also allowed for sections we there
was the notion that you might have when
you look at a typical order says you
it's not just a list of orders but
there's informational you know
components in between and so the notion
was the members would could be orders or
they could be a you know a section of
text or something and so I don't you
know we haven't talked about that
anytime recently but that was in the
original design and also the notion of
the hierarchy
of order sets that you could do because
often you end up in these order sets
saying here's four things and you can
either you need to do them all or none
or you need to be able to do pick one of
them but only one of them or you can do
any combination of them and that level
of control but again I think that if I
was either a rabbit holder we'll get
there in a moment so I actually think
that is in the kraang communication
board a sub member to some level so
you'll get there we'll get there as soon
as we can move on from this yeah we
actually don't otherwise let's move on
to it numbers yes okay I'd like you had
more on this we don't examine we got our
information you're happy with either
extending the data model or adding it or
adding attributes and adding category
directly to the order set as it is or
cat I think I directly over shipping is
what you're proposing yeah if we're
gonna say we want to create something
that that is more like location tag I'd
prefer not to call it tags just because
something's a tag and you can have and
then we treat it more like a token I
don't think that's what it means then
where does it necessary I'm just I'm
just I just wanted to pay you in a multi
a hybrid x animal thing yeah I guess
what I'm saying is that just because you
create tags doesn't mean that anybody in
the system can create the tag so the
notion that users can create tags is not
implied by their being DAGs I wouldn't
expect it you know I wouldn't expect
users for example to be able to tag
concepts or things like that even though
no we're not suggesting that just
calling it just calling a tag to
establish with location tags but if you
think that that was a bad pattern B
don't mark in any way well well I guess
what I don't think what to rename it
what I'm pushing for is if it's a if
it's a token that's essentially you know
attached to you know like chemotherapy
or HIV that you want to attach to order
sets in order to categorize them and
you're okay with the notion of leaving
it up to the administrator to the side
if they want to you know put
chemotherapy and HIV or just limit to
one if you're saying that by design you
only want to
have one of these things for any order
set and that would be something
different but if you're saying it we
just want to be able to label these
things as chemotherapy or HIV and if
someone came along and said hey this is
used in both situations and I wanna
label it with both that feels like a tag
to me and then you say just limit who
can edit those we don't we don't really
have the multi use case Thursday we're
just trying to think through the future
flexibility of the model and so I mean
really you just need to be able to say
in a UI like these are your you know HIV
regimens these are your tens therapy
residences here TV regimens yeah and to
lead on those or search on those
whatever to me tagging seems like the
most natural way to do that but right if
it's something where we're gonna say you
have to first go and define a set of
tags and then you can and then you can
go and use them yeah then maybe we
should call it something else
that's what that is what we're
suggesting I thank you yeah yep so what
is that other thing to call it you know
whether that's
labels categories category seems fine to
me compared to I mean label and tag are
almost synonymous there's not really a
difference between those now I didn't
suggest you probably don't care enough
to spend the night oh yeah we call on
that yeah yeah well then we're building
it we're gonna be we're gonna build it
so we're gonna do something and it's
going to get into core so that's why we
need inside it now if we don't want if
we don't care enough to talk about it
and we can do it I think I think it's
here's even beside something good
something up on it we'll talk yeah yeah
think we're all on time and we'll call
the others and talk about the orders on
time yeah all right so add tags or yeah
okay cool um so yeah sounds like like
Mike said sounds like you're okay with
us either doing attributes using using
attributes to extend it or sub classing
it so well
which one more capable doing in our
timeframe and then we'll look at doing
some sort of tag or label for the
category field so moving on we actually
didn't want to talk about order set
members because that is that particular
table it's flexible enough that we can
make do with what we've got well I think
we'll end up by talking about it when we
get that when we get to drug around
dosing reduction okay
the kind of next time we wanted to talk
about what the next step says the order
group so once the doctor selects the
order set instantiate sit for a user
then we would create this order group so
this would represent in our model it
would represent one cycle of a
chemotherapy treatment so kind of the
fields we feel like we're missing here
the cycle number is there on cycle 205
the again the category some sort of
label to tell us what this order group
is being ordered to treat the number of
cycles in the regimen so that way we can
represent you know this is we have the
cycle number two and we want to know the
total number of cycles that the doctor
has maybe um changed and also the length
of the cycles in the regimen and we
wanted to add some sort of concept of a
prior order group so being able to link
cycles of a regimen together in some way
having having some sort of linkage we
were we also thought that similarly to
order sets category or tag or label
could go on the order group table itself
go directly on the object and the other
fields we could represent as attributes
so yeah the prior order group I would
just try to mimic what we did with
previous order yeah you know to me it
seems natural that that we we have the
notion of when you're ordering and
you're revising orders we make that
linkage and I think if we try to mimic
that same pattern it would be intuitive
instead of doing the same thing but in a
different way but as that's fine the
various order groups even yeah something
like that
yeah I mean I call it a previous order
group because the way previous order
seems to be used it's more of a we're
discontinuing this one and in the case
of like prior order group it's not that
we're discontinuing one order and
starting another it's that we've we've
completed the first cycle and this is
the next one and to me discontinued and
complete don't really mean the same
thing and I did I so I didn't want to
generate confusion there but I could be
wrong
yeah I yeah it's a fair point I think
when I think of previous order it's it's
it's just creating a linked list and so
previous order group I would probably
just you know I don't know that's just a
detail I would just kind of try to keep
naming similar so that when people are
trying to guess oh there's previous
order where how do I find that for the
order group the are you want to store
the category if you think the category
of the order sets might change over time
and you want to know what how is it
categorized at the time it was ordered
um I don't know if the category will
change over time I think we're just
we're nervous about relying on an order
set to look up kind of the treatment
that explore because order sets are
technically mutable we're hesitant to
rely on that linkage for any sort of
information okay that said it would be
pretty wrong to change edit a
chemotherapy order seven turn it into an
HIV or yeah I mean that seems like items
you should leave to be inferred from the
foreign key relationships like yeah I
mean I think I personally think that
category is probably the wrong thing in
this context and that category can be
inferred toward as a category I was
thinking this was more like a reason I
mean more more closer to the reasons for
order column on an order like increase
in indication individual or each
individual order has a reason for reason
on it
the reason coded or whatever and so you
may want that at the order group level
just for for organization and getting
getting back all where if there's a
reason it certainly makes sense that if
we have indication of a reason on order
to just have the same thing as optional
on ordered group that says plausible to
me okay so you're saying like what our
group order group reason only otherwise
like if we wanted to get the you know
use the reasons to organize things you
have to like look at all the members and
yeah I don't know I mean you could do
that as well it's inspected members and
see the reasons for each member's yeah
the usually the reason is water reasons
are pushed for billing purposes as much
as anything but that's yeah I think if
you you can make that call if you decide
you want an indication then you can use
a like an order of reason if you decide
that we need to check something else
then we can hash that out over talk
okay um so what about on the cycle
number does that make sense as an
attribute of an order group yeah to me
this is where it's like you know that's
very that's it might as well be
chemotherapy cycle number class so so
the question then is is this we're gonna
have the same conversation set where we
can either add attributes or we could
extend order group to be cyclical order
group yeah to me the difference I mean
yeah I mean in some ways I guess it's
ultimately it's maybe attributes if you
want to be able to classify them in some
ways I just like to say you know here's
like your you know your set value
properties associated with this order
group and let they use the use of them
kind of but I guess in our model at
using attributes it's it is so you
probably don't have time to explore this
now but you know it would be cool to
actually explore whether we can use and
really I mean I know that Postgres
supports good tonight recently and I
suppose my people probably does also but
of having like juicing columns that you
can basically listen you can do like the
equivalent of attributes although they
wouldn't have to be tight it would be
neat to just I mean as potentially where
orders that members easier and more
lightweight also member has it order set
member and whatever yet had it has
template in a template type and the
template type is goodnes other parts
template and to something use it yeah
yeah but I understand that Postgres
allows you to actually query like sequel
queries as its dad my pickle like my
pickle five seven I highlight people had
lovely snow too
so that would be neat although that
would be a longer exploration yeah like
yeah yeah but I guess just fundamentally
the notion of the the cycle number a
number of cycles in the regimen
length of those cycles those properties
are pretty chemotherapy specific and so
no I at the end of the day if it was my
friendship for Mike forces with Mike
versus knitting these attributes in I'd
say and throw them wherever you want to
it those are you know fairly
chemotherapy specific and that's where
you know knowing that we're trying to
create an order service that is you know
used for all types of orders you know
like you know a cycle number for you
know a order set on you know like
treating someone coming into the
hospital with appendicitis would would
be kind of nonsensical and so and my
friendship luckily is not is not staked
work I appreciate that
well yeah we don't really care I mean I
think we're just trying to come up with
the best thing for them rather
stairwells to move forward yeah I know
I'm I don't think we were ever I don't
think you were suggesting you should be
properties on our group yeah I think
maybe just the data we need to capture
and so I think yeah the designs were
thinking about are either and attributes
for all the cyclical stuff or subclass
called cyclical order group which could
either go into core if you wanted at the
living core or go it had to go into an
oncology module and I think either the
either those are fine and if you wanted
to go rogue and say we want to have
order group you know property or order
group you know something like that
that's just like a JSON string and we'll
I think leave it up to the user I mean
until we need any of those sections any
of those feel better than saying we're
gonna hard code into the model that you
know every order group are gonna have
this property that belongs to like a 5%
use cases orders say well I can bother
like I would have ordered route dot like
I look like additional data or something
like that and that would allow you to
put whatever you want in there it's in
text field and yeah I'd probably I
probably named it to something that's a
little less of an afterthought
maybe it sounds bill official and what
our group purgative properties okay
okay I'm not sure how that's different
from attribute other than it's just it's
just it's just less effort on your part
and then more pain down the road then
but when everybody decides to go at it
differently and I just want to be
consistent I mean if we're going to do I
mean if attribute if if that's the new
model that we should choose over adding
like attributes yeah now anything extra
bit will be the proper way to do it I
just don't want to be a blocker for you
guys in terms of you're like oh we can
do attributes but it's gonna take it's
like three more months and we won't get
this done well personally I mean this
working discuss somewhere in that room
and she's already you know probably
because the way I was building both
these attributes in two core so just
assuming that's the way we're gonna go
now that doesn't have to be we have to
use that so that Berardi sort of suing
that's the past okay okay great
so thumb sounds like attributes are good
to go
ding-ding-ding-ding okay um yeah so
we're gonna move on to drug order and
we're going to start with the less
confusing one so the second one which is
chemotherapy administration group so
that's something again that like Burke's
been saying this is pretty chemo
specific but we want some way to know
which medications within the order group
are for pre medication for chemotherapy
or for post medication so if we wanted
to fit it into the current data model
there is there is this field in order
called order reason that we have that we
could use create a pre medication post
medication order reason order resonance
is intended for like the indication for
the order and so I wouldn't try to
repurpose it for
or a classifier okay so then yeah that's
not a so pre-med a chemotherapy free
medication is not is not a indication I
guess I mean I guess you could you could
kind of try to you have a module that
can I like I don't know what else would
I don't know is the indication for those
orders would be other than chemotherapy
premedication what would you normally
put for those yeah I guess I guess I'm
coming at it from the standpoint at
least from when when you're saying like
what you're trying to accomplish in
terms of categorizing things in an order
set or I guess are we are we in order
group now or order said I can't remember
what we're in hello an order yeah so
like there's for example there's a bunch
of orders that you've placed and they
are the pre medications oncology and you
want them to show up
oh so where were the heavy metal cutting
of pre medications I was and then I was
I was thinking can you go back to the
previous page like the yeah so okay so
you're actually in the drug order itself
yeah yeah so we need that once we
instantiate the order we have all of
these order objects within our order
group they would all be drug orders
these since they're all medications and
we we want to I want to know that this
drug order that was placed was actually
in this group I mean the the problem
you're gonna have that might work with
pre medication with chemotherapy it's
the indications most likely gonna be
what you're treating or it would be a
competent competition with that and then
for post medication it would be in
competition with actually treating
whatever symptom came up like you know
sometimes post medication might be for
protocol but a lot of times post
medication is nausea stuff or you know I
can't you know things things for
side-effects of chemotherapy in which
case you might want to have the reason
for this order is because they started
throwing up yeah they
or because they you know start having
chills or whatever one question I have
so is there another thing you can think
of work that might be a it's basically
it's like a category within the order
loop that would decide on the order so
the thing is this is going to happen to
be in the order this this data is going
to have to be in the order set because
we're going to have somehow and the
member or the order said or there was
herself gonna have to be able to show
these are clear indications you care
that's where I guess where it was that's
where I was originally and and in terms
of my initial reaction and and I imagine
I having a point of from the drug order
or the order to the order set member so
much it came right exactly that's that's
a possibility the one question is orders
for procedures are they combine a Belen
this because I'm thinking in a lot of
chemo regimens and other orders set type
things you want to include a procedure
like an echocardiogram at a specific oh
yeah I mean order sets are intended to
be include every kind of kind of order
so orders referrals diet procedures any
any type of order we might come way up
with in the future there's easily a
dozen or more that are in common use we
want we want to be able to put those in
an order set I think um my concern with
you know maybe having a pointer from the
order back to its order set member is
I'm not sure if there's a use case when
a doctor orders a chemotherapy regimen
when they might add a completely new
medication to the pre medication then we
don't have any sort of way to have a
pointer back to the orders that member
could it wasn't originally in the the
regimen template and we'd still need
some way to mark that this is a pre
medication yeah and there's another
reason and that is we wouldn't want
foreign keys from our data to our made
it metadata in a way that
like people started ordering these and
then you go in and say oh we need to
correct this order set that I can't
change it because yeah - but it feels to
me that at this point where you're
saying whether it's drug order or really
any type of order coming from an order
said you actually want to be able to
categorize it based on some feature of
the order set and you have a certain
order set that has these categories of
maybe I shouldn't say category pickles
but in any case you have the you know
the pre medication section that
chemotherapy section of the post
medication section you're trying to to
capture that because order sets can be
multiple you know ideally we want to get
to the point that they can be embedded
what's in each other and you know you
could have it you know I could get could
get kind of messy we have played around
with the idea of having premedication as
an order set with and chemotherapy has
an order set both within yeah I taught
me the most the most natural way for
this would be as if you had if you had
an order set with your pre medications
and order set was your chemotherapy and
an order set was your post medication
and then you created an order set that
includes those three sections then your
order groups would give you the
information you're trying to capture
here
I guess one one we did consider having
you know an order group for the pre
medications an order group for the
chemotherapy medications etc but then
there's no no good way with the data
model to then tie those order groups
together
there's no nested ordered route coño
so that's something that I think is a
viable attribute would be if you had an
order if because it's I think we we we
do want to end up in a place where you
can nest these things and if you have an
order set that is you've created over
two sets a B and C and then you come
along and create a D that combines those
and and someone orders you know pulls up
order set D and places the orders
capturing that
and then we want to persist not just the
lower level order groups but that
relationship so an order group what a
group group like Parenthood a group
that's knowable yeah yeah I think that I
mean effectively what we're saying is
yeah you have an order says probably
supposed to be in the original design
but if you have enough nested orders set
members then you should have nested
order groups I mean the problem with
order set design also is that there's no
there's nothing in it that why is that
well there's no way to really represent
a nesting I mean because it's just an
order set can have a bunch of order set
members which are just and there is have
a template so that those templates I
guess could have a way to have to
represent other order sets but it's not
clear how that there's no structure for
it in the data model well okay there too
let's say if you end up and want if you
go that path let's have another let's
have a deeper design discussion on that
yeah I think the intent is that orders
fashionable and I think it would be
perfectly reasonable to capture that
path you know like if you're you know
Ted Cheney to put in order group
something like a parent order group of
something that would allow you if you
had an order group that was just all I
had was three order groups in it I mean
an order set that only had three order
sets in it and then there's an orders
for the nature those and someone places
orders across that you you would lose
information if you just stored the
lower-level order groups and you didn't
have a parent or a group to link them so
like we should consider adding that in
order life yeah yeah okay good to know
that that sounds reasonable
um so we got four four minutes left
there's this maximum lifetime dose
there's another talk post out there
right now on that one um just one that
should be I think dosing reduction is a
bit gnarly so I'm hesitant
bring that up with three minutes left
one that should be hopefully pretty
quick so on drugs there is a concept of
a maximum daily dose and a minimum daily
dose there's no concept of a unit for it
or any sort of indication of what that
number means would it be reasonable to
add oh because we removed unit and made
strengthen to strain fusions ago okay
why my idea I think yeah I just couldn't
separate dose and dose unit on that I
think it was for combinations well just
finally did you make that change anyway
it's a fair to point
just like quick very quick reaction is
that should probably though should
probably actually be constraints but
then it's not so obvious how to
interpret them but in the same way the
dose is actually a string so that one
lesson a charity rings ring testicle no
drinks refers to the strength of the
formulation and daily dose refers to how
much you can give in a 24-hour period so
they're a little bit different like if
you have a 10 milligram pill that
doesn't necessarily mean whether you can
you know maximum daily strength would so
you know imply that there's like some
you know if you can take one hundred
milligrams of the day there's one
hundred milligram pill even if there
might be just tense yeah the reason for
the reason that it is like the way it is
because of combination drugs okay oh
yeah put combination going so 800
milligrams of one thing on hundreds just
you know again to the other yeah and the
fact that we took out units to address
that or to when we turn strength into a
string to try to adapt to some of those
and so there was a units in there that
disappeared and maximum daily dose and
minimum daily dose miss it but I are you
looking to capture daily because I
thought your what was your unopposed or
next time I thought you're like we we
are looking at maximum lifetime dose
there are certain drugs as well that in
the templates you know specifically say
only give them two milligrams per day um
so we were we were also wanting to
capture that information somewhere and
since this field was already there it
seems like the reasonable place if only
we had a unit um but I may be
misunderstanding something
no it's historic and you're not
misunderstanding anything official open
em rest statement on this is that it the
unit is thingy all you have to do is add
the word thingy to whatever the double
is and you've got the correct yeah so
yeah that's I think I don't know if
anybody is using maximum minimum daily
doses it's in the model but its its
crust certainly for chemo drugs it's
useful yeah yeah yeah so I think the
challenges have what you know and you
guys could make any you know a
reasonable proposal of like how do we
how do we add
to the model so that we can capture
what's needed for chemo and if you do
need to know daily doses then I would
suggest well let's make some change to
the model that that it would would be at
least give a migration path for people
that are already have data and that
could be just you know take whatever is
in those fields and if it's not null
then take the parse out that uniform
string and and turn each of those into a
string it could be yeah or it could be
make new attributes with a different
name and deprecated the ones that are
there and I think if you wanted to have
like a lifetime like maximum lifetime
dose as a I will probably wouldn't make
it a double without units would be my
recommendation lifetime is a whole
separate yeah yeah and worms that that
we're not proposing or talking about
here we just yeah we all we we were
proposing was putting that unit field on
here so we could tell you know what the
maximum daily dose is being measured in
but if that's not the correct route it's
andrew by historical reasons we'd love
any any thoughts on what we should do
yeah I mean just just off the top of my
head like a daily dose units might be
the simplest approach okay sorry
the other possibility actually is the
master this is actually standing
I mean it's unit um to the drug so like
if the drug of the 400 milligram tablet
and one the max is one name is 400
milligrams I don't know if that's right
or wrong but that's the other of course
is to actually just interpret it that
way
Turpin is it's this is a formulation how
many can you take
yeah it gets kind of ugly though when
you say you know well while the
challenges in key mode or oncology drugs
like what are gonna do with biologics
right yeah they're crazy oncologic
therapy I mean Carty like with
so we have to be I think having a
separate unit saying hey this
calculation is done on the following
unit I'm a big fan of doing this because
the it's so easy to make lethal errors
with chemo drugs that anytime you can be
more specific as a good thing so so
aren't there any issues are there any
concerns with us just adding a daily
dose unit field to drugs oh okay
how silence is good doctors consent
vehemently we're disagreeing okay um so
we're we're a little over time um I can
post in talk kind of on this I'll post
you know a summary of our call and kind
of the decisions we made on this
particular thread and then I love any
thoughts on dosing reduction if you have
questions on it I'll I I tried to
explain it quite a bit in the talk
thread but you know if you have
questions please post them and any
feedback is very much appreciated so you
give the one sentence dosing reduction
explanation just yes refers to yes so
one sentence explanation at time of
ordering a drug or even seeing the
patient the doctor might look at the
dose and say oh this patient is sickly
this patient isn't reacting well I want
to reduce the dosage by 10 percent the
nurse then will will do her calculations
of the drug and then after calculating
the dosage will then do the 10 percent
reduction and that ten percent reduction
is not only important if that's 10
percent reduction that's the 10%
reduction off of like the thing you'd
calculate based on milligrams per yeah
so you're actually you're actually
trying to capture that order of fooling
physical reduce this drug by ten ten
percent right it's the problem is this
is all based off the old paper era
orders which ironically in this case is
the current paper era orders where we're
talking about deploying this thing but
in the old paper era right you had a
protocol that was literally printed and
so you couldn't easily alter your
protocols so the easy way was to tell
the nurse oh just dose reduce it by ten
percent because you couldn't change the
original order because it was
pre-printed and so but yeah that's the
way they're used to thinking about it
and and we want to capture the reduction
percent because not only is it important
for this particular you know cycle but
it's something that often doctors will
choose to persist to future cycles and
they'll want to know you know oh I've
already dosed reduced this by ten
percent
do I want to dose reduce it more do I
want to leave it at the reduction do I
want to take off the reduction so that's
important not just in the current cycle
but in subsequent cycles as well yep so
my one sense responses that feels to me
like it's that's dosing instructions
that are like chemotherapy system in
there specific to that type of dosing
yeah yeah and there's yeah so if you if
you look at the post there is I do think
you know there in the field that we
could put it in but you know we've
considered also extending it in some way
so yeah any thoughts that you want to
put there would be great okay so thanks
a lot for all your feedback today guys
um and we came to some good decision so
that's very helpful crap yeah no I
appreciate all your cases due diligence
and and making the effort to try to to
build this within the community with the
community spirit we appreciate that so
next would be will will revise some
designs and post our intentions and then
also zenithal questions to talk and
proceed that way that makes sense yes
any gray hair you see on Mike is
probably due to most of the conversation
okay great well thanks a lot guys all
right take care
well talk to you later beyond the
instrument cyberspace
you