How best should I map values that are the result of composite scores for a particular condition?

Implementers
1

In my OMOP ETL, I have run into a bit of a conundrum. I currently have a few columns whose values reflect a composite score (e.g. for scoring a patient’s alcohol risk or calculating the atrial fibrillation stroke risk, a.k.a. CHAD score) but I’m not sure how best to map these values to concepts in the OMOP vocabularies.

For example, in the case of the CHAD score, the values in this column are integers that range from 0 to 9, where higher values indicate a greater stroke risk. These integers, in turn, are calculated by asking 9 different ‘yes/no’ questions to the patient (e.g. is hypertension present, is patient over 75 years old), where a ‘yes’ value increases the score by 1. Because the integer in this column represents a composite score, I am unable to easily break down which part of the composite value is derived from which question. Two patients with the same CHAD score of 4 might have had a different set of ‘yes/no’ questions fulfilled in order to attain the same composite score.

So, I am unsure what I should do then in mapping these composite scores to their pertinent OMOP concepts. Has anybody else experienced a similar problem and figured out a solution?

1 Like
2

@PhilipePholop:

Two questions:

  1. Are you doing it in your own environment for your own research, or is this something that will be used in the community and in other institutions?

If the former - don’t bother putting it into the OMOP CDM. It is for standardizing patient data across the community. If the latter:

  1. Are you calculating these values from the OMOP CDM records, or is that done elsewhere and you want to use them as patient data?

If the former - this is not part of OMOP. It’s the result of your analysis. OMOP data are the input. If the latter - you probably have to create a concept for this, either your own private one (so-called 2-Billionaire), or you add one through the community contribution mechanism.

Hope this helps.

1 Like
3
  1. This OMOP ETL is being down for our own research, and will not be used in the community, although there is a small possibility that at least one other institution may need access to this data as well.

So at this point, I will create a set of private concepts for this.

Thanks so much.