When performing a study using this information, are you suggesting that you would define a cohort based on the appointment information?
Second question: Does appointment_status in a row change over time? Ie: if you are at the visit, you create an appointment (iâm assuming here the appointment for a visit is for a subsequent visit), it goes into the row as âScheduledâ. But then I call back and change the date. But my phonecall wasnât at a visit, so how does this tie into the visit the appointment was made at?.
Iâm wondering if this bookkeeping of appointments is more administrative than analytical. Can we get a analytical use case?
-Chris
After deliberation/discussion in our group and some pilots at UCI, Ayan suggested a new table. I believe, in part, the rationale was due to an actualized visit might have multiple schedule/reschedules with different attributes. However, our discussion predated OMOP micro visitsâŚ
as CRI researcher - I can supply this use case:
On âOHDSI trial screenerâ app found 15,651 potentially eligible patients for a trial NCT0011223344.
The next app âOHDSI research coordinator helperâ lists for next week, at which clinic and at what time should the research nurse be to talk to the physician and the patient about possibly enrolling the patients.
Scheduling information helps complete the cycle where OMOP CDM is used for streamlining trial enrollment. (the motivation is to save research $) (so use case is not generating evidence but also a noble goal)
So it breaks the dogma that our model is only for data analysis. Which I think deserves to be broken at some point and we should face the real world of CDMs.
Yes, thatâs exactly right Vojtech. We are trying to encourage everyone to adopt OMOP for whatever reason. Including the process of generating evidence. We need to link OMOP IDs to recruiting workflows so that we make the round trip data linkage and getting research data re-integrated into EHRs.
In the case of quality reporting, missed visits for recommended care may sometimes be excluded from denominators.
Hi team,
We are interested in adding appointment information into OMOP CDM tables. So, we can find patients who had Cancelled, Pending, Arrived, No Show, Bumped or Rescheduled status. I read this thread. Is there any update of adding appointment information to CDM? Thanks!
No, there isnât, @gongliangz. Want to propose something?
Generally, the OMOP CDM deals with clinical events that happened, not how it is organized. But I could see a use case where those things could have an effect on outcome. Do you have such use case?
Just a gentle pushback; what happens when a provider puts an order on a cancelled appt, because it was easier to do that than create a new encounter, just for the order? How about chronic conditions, that have no visit in the EHR, gets matched to a no-show? I can think of a few other scenarios, but I am sure you get the idea.
I am sure that neither of these things have ever happened in the history of Homo Sap Sap. 
, but will the this data be burned or will studies now have to look though âbadâ encounters (visits) as well?
@Christian_Reich , No-shows, rescheduled, and cancelled appointments are especially interesting clinically and for health-equity research. They might indicate that a patient is having challenges accessing care, which could lead to poorer outcomes (such as delayed diagnosis or treatment).
Might it make sense to store those in the visit_occurrence table - but somehow flagging that the visit didnât happen? The challenge would be where to add that flag:
- visit_concept_id - not good, since might want to profile which type of visit (inpatient, outpatient, etc.) had a no-show
- visit_type_concept_id - perhaps - might require a few new EHR -related concepts
- admitting_source_concept_id - perhaps, but doesnât feel as appropriate as visit_type_concept_id
How would this affect outpatient only clinics?
How does any of these affect walk-ins, where a patient does not have an appt.?
@Mark , clinically, weâd want to track any visit that is canceled/no-shown, or any situation where a patient leaves without being seen.
Most specifically, CMS has OP-22, which measures what % of patients leave the ED without being seen. Operationally, one would want to know any setting where there is a high % of patients who leave without being seen, or are unable to keep their appointment for some reason.
What kind of order? A procedure? Unlikely happened if the visit was cancelled. A prescription? Doesnât need a visit.
Remember: The OMOP CDM is a (not perfect) recording of the things that happened. If they happened we have a record, if they didnât happen we donât. Closed World model.
Again. If a visit didnât happen we should have no record of it. Otherwise, all existing cohort phenotypes and analytical methods will die. Youâd have to check each record whether or not it happened.
I think your solution is to add the orders as Observations and link them (or not) to events (could be visits or other entities). Find an adequate SNOMED concept and come to the Themis WG to ratify the solution.
Pardon me, i was not clear enough. My first question was specifically about how clinics, that are only outpatient, would use admitting_source_concept_id, if that was the route that is chosen.
Yes, we track those numbers as well, internally, as we service mainly low income and/or special needs, patients.
Thank you for your reply
some other use cases could be:
- forecasting appointment demand and improving resource allocation/utilization
- reducing patient wait times and no show rates to improve Quality of care
No. As @Christian_Reich states: âGenerally, the OMOP CDM deals with clinical events that happenedâ, we donât add potential, missed, canceled or other non-events to the CDM. There are some rare concepts which we do add (non-smokers or negative measurement results), but these are few and far between with solid use cases. The OMOP CDM supports observational medical research. It isnât designed for quality improvement or operational studies for healthcare or insurance systems. I agree with @Thomas_White, there are some health-equity research use cases out there for these data, but how do we distinguish a no-show because a person lacks transportation versus a person that moved or found a better doctor or felt better and decided not to go? We donât have these data in our source encounter tables. These types of things are found in surveys or possibly some type of registry database.
Possibly, but we would have to add a new column to the Visit table and this MUST be incorporated into the CDM tooling, otherwise, researchers wonât know if the person actually had a visit. We really need a strong use case. This is a HUGE change.
RE: other suggestions to place these data in the Visit table. We donât repurpose existing columns. Itâs not a good practice. End users need to trust the data in the field follows the conventions.
Christianâs idea might work:
Themis is a sub-working group of the CDM WG. We define and document conventions on how the data are stored in the CDM. Our meetings are held the 1st and 3rd Thursday of the month at 9:30am Eastern Time on MS Teams. We are just getting started after a 4-5 year hiatus.
If anyone would like to take on this question as an issue for Themis, I will send you version 1 or our Themis issue tracker template. The Themis WG has used this template for one issue, but more feedback will be welcomed. Also to note, the Themis GitHub is a work in progress, but soon the template and other documentation will be located there.
Provider can add No Show Call or other communication message to encounter in which patient did not arrive.
That would distort the noise to signal ratio. I can see putting the missed appt in the observation table, as Christian suggested, but never as a visit. Even in the observation table, I wonder if the juice is worth the squeeze, but in theory, I have no issue with it.
In general, I agree appointment scheduling data should not go in a visit table for an analytic CDM.
Above, a higher level issue arises as to whether OMOP will start to be used for clinical trial workflows, including recruitment. Deciding how and where to store future appointments seems to depend on that decision. There are some minimum requirements for the transition based on functions desired. Probably the most crucial of which is regular - maybe nightly? - updates and an extremely robust system to either snapshot or overwrite previous data (appointment status may change many many times). A new table seems to make sense for this. It may be worth the squeeze ⌠OMOP on FHIR could drive some amazing pragmatic trials if future intervention opportunities could be better defined.