OHDSI Home | Forums | Wiki | Github

Present on Admission vs Admitting Diagnosis

On the CDM Wiki for CONDITION_OCCURRENCE.CONDITION_STATUS_CONCEPT_ID, there are three concepts given for use:

  • Admitting diagnosis: 4203942
  • Final diagnosis: 4230359 (should also be used for discharge diagnosis)
  • Preliminary diagnosis: 4033240

We have a situation where diagnosis could be an admitting diagnosis or present on admission. Right now we are saying they both would use 4203942 but was wondering if we should add this concept to the list. We aren’t sure if we need to worry about the DOMAIN.
46236988 - Diagnosis present on admission

This might need to be a THEMIS ticket, but wanted to start the conversation.

Tagging @anthonysena

We also have it in some of the datasets. It basically means that it’s a final diagnosis, which was also an admitting diagnosis ( or a part of it). So we create two rows with the same disorder concept_id and types: final and admitting diagnosis.

@aostropolets - wait I’m not clear, which one are you talking about would be final diagnosis?

Hi,

In my situation where I’m incorporating the entire spectrum from 911 call (pre-hospital/Emergency Medical Services) through hospital admission/discharge. The Preliminary diagnosis (4033240) would apply to paramedic assessment (linked to clinical impression) as well as “Initial diagnosis” typically documented within an Emergency Department EMR systems.

Admitting diagnosis would represent ED admission as well as ED to Hospital admission.

“Final diagnosis” would be tied to the specific visit_occurrence and EHR’s final/discharge diagnosis as documented by the local institution.

“Diagnosis present on admission” seem to be diagnosis (secondary?) known at the point of admission, but not necessarily the primary reason for admission which is identified by “Admitting diagnosis”. That is a set of diagnosis acquired by nurse screening/triage, derived from presenting symptoms, history, and initial medical device reading. For example the condition of “Right bundle branch block on the EKG” as present upon admission, but the reason for admission was “Acute myocardial infarction”. Or present diagnosis of “wrist sprain” and admitting diagnosis of “Stroke” - which caused the patient to fall and sprained his/her wrist.

So “Diagnosis present on admission” seem like a good way to represent one or more secondary diagnosis - which often is available in both pre-hospital and emergency department documentation.

What we have ( and I thank this is what you have) is a diagnosis with a flag we’re discussing. We assign type Final Diagnosis to it (as it seems to be a discharge diagnosis, which is equal to final). Then we duplicate this row with Type Admission Diagnosis and date of admission. At least this way represents medical state of things.
@mkwong admitting doesn’t necessarily mean primary. That’s why we have status where we can put primary/secondary etc along with Type admitting.

I think we should create a condition_status_concept_id for “Present on Admission”. The Present On Admission (POA) flag in claims is a standardized field, as described by CMS here. There is some variation as to whether claims databases (available to researchers) contain this standardized field. But, if it does, it is data that comes directly from a claim form.

Judging from the previous replies, it would seem that the “final diagnoses” would be found on a claim form, since this form was completed after the patient was discharged from the hospital and the hospital is billing those diagnoses for payment. Claim forms do have a separate field for “Admitting Diagnosis” and that diagnosis could be completely different than the billed diagnoses on the claim form. As far as I know, claim forms do not store “Preliminary” diagnosis fields. Though possibly used in many EHR systems, “Preliminary” diagnosis fields to not refer to a standard like POA fields do.

Quick background on the POA flags. Every diagnosis code billed on a claim form has a POA flag. So the primary diagnosis on the claim form could also have a positive POA flag. Payers are interested in these flags because they have stopped paying for hospital-acquired diagnoses billed (see regulation here). But, if the hospital notes that the hospital-acquired diagnosis was “present on arrival”, and not actually from staying in a hospital bed, the payer will pay on that diagnosis. An example would be a fracture from a fall. The payer will not pay for that fracture unless the POA flag is set on that diagnosis. And, you can imagine a fracture from a fall could also be the final discharge diagnosis and an admitting diagnosis.

So, instead of trying to reuse status_concept_ids to fit standardized data into non-standardized concepts, I would suggest just adding a new condition_status_concept_id to keep things very clear.

1 Like

So does ‘Present on admission’ mean that a patient was diagnosed with a disorder before the admission or on it? It does matter and here’s why:
If it’s an underlying condition/trauma/etc. Then we can’t use this date of diagnosis for specifying the onset date of this disorder. Moreover, it looks like history of a disorder…
This makes me wonder how you would use this status_concept_id in your study. What will be the use case?

POA was implemented when CMS, the US federal agency that pays for most elderly care, announced that they would no longer reimburse hospitals for inhospital events that they felt were preventable (or indicated poor quality care). That triggered hospitals needing a way to identify those diagnoses that did not occur “on their watch” (present at the time of admission) so they didn’t get dinged for something that already existed. Thus the POA flag was born.

You cannot and should not infer anything about the onset of a POA DX. All you know is that it was noted at the time the patient entered the hospital, usually in the physician or nurses admitting notes. Could have started yesterday or last year. Doesn’t matter. Only that it didn’t happen while in MY care…

1 Like

@jenniferduryea and @mgkahn thanks for the input! Didn’t know that. After your clarification, I would strongly agree that we need to add this status. It changes the logic indeed as these diagnoses should probably be treated as ‘history of’ (as we don’t know the actual onset date). The only thing I’m worried about is that CDM specification should state clearly that the date of POA shouldn’t be used for an analysis as it will cause biases.

@aostropolets this might be another forum post, but I would be weary of any diagnosis codes found on claims to represent “actual onset” of a diagnosis. Physicians add diagnoses they treated the patient for on their bills. For acute conditions, you might be able to INFER that the diagnosis treated was the date of confirmed onset. But for chronic conditions (i.e. diabetes), physician will bill for chronic condition diagnoses treated, regardless if it was confirming a new case or treating an old case. The diagnosis codes are the same between new and existing conditions. So generally, to find “onset”, the analyst needs to use some kind of temporal rule (i.e. look for the first diabetes diagnosis in the dataset or look for a diabetes diagnosis with a clean period of 1 year?) to INFER onset. So your concern, while valid, is not limited to POA, but to all claim diagnoses.

I feel this is outside of the realm of CDM standards and should be left up to the analyst to understand, based on the dataset and inherent biases within that dataset.

I’d argue against this: history of says ‘you’ve had this in the past, but not presently’. Maybe the ‘not presently’ is a stretch, but the way I look at CDM condition_occurrence records is not so much ‘the onset date’ but rather ‘was present as of this date’.

Does it somehow correct the suggestion to add new condition_status_concept_id?

@aostropolets no it does not contradict the suggestion to add a new condition_status_concept_id.

However, it does contradict your last sentence below. I believe this sentence is what @Chris_Knoll and I are referring to.

1 Like

All, this is a wonderful thread. Agree to add a new CONDITION_STATUS_CONCEPT_ID for Present on Admission.

1 Like

Good.
As we use SNOMED concepts for this purpose and do not create new ones, it seems like SNOMED
762456 Cinical finding present on admission can be used.

1 Like

I agree, @jenniferduryea. How about this as a definition:

“History of” is any Condition the patient is presently not suffering from (or enjoying for some). Which means diabetes really can never be “history of”, and myocardial infarction very much so.

1 Like

Looks like I just repeated what @Chris_Knoll said. :slight_smile: Oh boy.

1 Like
t