OHDSI Home | Forums | Wiki | Github

Classifying surgical complications


(Andreas Weinberger Rosen) #21

Dear @Alexdavv, that sounds amazing!

@ericaVoss, I’m sorry I didn’t notice you tagged me. Regarding how The Clavien-Dindo (CD) scores are derived and how they might be used. The score grades the severity of complications by how they are managed, so if a patient would have a post-operative complication say a urinary tract infection, it could be graded as a CD II if it is managed by antibiotics or CD IVa if it leads to an ICU stay due to single organ dysfunction. So they are derived based upon the management of the conditions. For analytics, it would probably mainly be an outcome we were interested in predicting or assessing the effect of different interventions on the risk of getting an, e.g.>= grade IIIa complication.
However, the CD scores could also provide valuable insight for predicting the risk of readmission for patients admitted to a department or using it as an exposure for late-term outcomes such as the risk of recurrence after a surgery for cancer.
I hope this answered your question, if not please don’t hesitate to reach out.

(Erica Voss) #22

@Alexdavv & @Christian_Reich, what are we settling on here pre-coordinated or post-coordination? If I got a vote I think I would vote pre-coordination now because it is easier from the tool/analytics side, but I see how this is a nightmare on the Vocab side.

(Alexander Davydov) #23

I’d agree with pre-coordination only if we come to the decision (for the whole community, in fact) to leave only one parent concept for each of these scales in the Condition domain and create/use the pre-coordinated Conditions placed below in hierarchy.
It means that people would not be technically able to post-cordinate using these concepts and we’ll get rid of ugly pre/post-coordinated mixture, at least in these 2 scales. So we need to conclude that things like “II or III”, “between II and III”, “nearly III” go to trash here :blush:

And again, what if someone will need to use them as Condition modifiers? It should be Measurement/Observation domain then.

(Sebastiaan Van Sandijk) #24

I am not sure, but this plea for pre-coordination seems to apply to the OMOP-CDM as a whole, not just for the Clavien-Dindo complication grades. In that context, the other day I read a proposal by the WHO Working Group that works on the harmonization of the ICD, ICF and ICHI. Basically, they work on a unifying ontology or Content Model for this Family of International Classifications. Thing is,this WG calls post-coordination ‘key’ and ‘critical’ to the Content Model. I am not fully versed in this domain, but would this pose a problem? Can OMOP-CDM function on a preferred or fully pre-coordinated model with some of the main (standard) vocabularies choosing for a post-coordinated model? I can imagine that this complicates things, either for data custodians or on the side of the CDM.

(Christian Reich) #25

Can you show what they are saying?