OHDSI Home | Forums | Wiki | Github

Trying to find Income quintile

(Micheal ) #1

Hello, all
In Korea, we have each patient’s Income quintile in raw data which is used by hospitals.
I want to know where income quintile is in OMOP CDM, which maps to any table or field.
Does it exist in CDM?
I appreciate it if you tell me.

(Christian Reich) #2


We currently don’t have income quintiles. We don’t have socioeconomics standardized at all. Want to make some good suggestions?

(Anna Ostropolets) #3

It seems that you can do it relatively easy:
put into OBSERVATION table
4269577 Income details - finding
and value_as_concept_id = 4155974 Ninetieth percentile or any other percentile
Wouldn’t it work? Doesn’t seem like something complicated.

(Dmytry Dymshyts) #4

@aostropolets, unfortunately SNOMED has concepts only for 5th, 10th, 90th and 95th percentiles. So it wouldn’t work for most of the cases. And I suppose, it’s more convinient to keep them as value_as_number.

(Anna Ostropolets) #5

Because these are the standard percentiles that are typically used I guess :slight_smile:
Numbers work just fine. Micheal will be pioneering this, so he can choose whatever suits his needs.

(Christian Reich) #6

Yes, @aostropolets’ solution is pragmatic and will do the trick. We still need to standardize. But till then:

observation_concept_id=4073460 “Individual income”
value_as_number={20, 40, 60, 100} (depending on which quintile)
unit_concept_id=4041100 “Percentile”

The latter is ugly. Percentile is not a UCUM unit.

But again. Let’s standardize these data.

(Seng Chan You) #7

You can see how we’ve done it in this repo from line 402 to line 433.
We used Income Status (OMAHA) for the concept_id in the observation table.

(Christian Reich) #8


Is income percentile all socio-economic information you are modeling, or is there more?

(Seng Chan You) #9

Yes, you’re right. Available data for socioeconomic status from Korean National Sample Cohort is deciles of income status. Based on income status, Korean national insurance system charges the insurance fee.

(Christian Reich) #10

Deciles or quintiles? And the income percentile is known for every patient?

(Seng Chan You) #11

@Christian_Reich Deciles. I used Tenth percentile for unit concept id, which seems silly now. :sweat:

Granularity for income status can vary across Korean databases.

In NHIS-NSC, annual deciles of income status is recorded for everyone.
It’s possible because income status is estimated per household.

(Micheal ) #12

@SCYou, Thank you It’s really helpful and clear explanation.
As you mentioned it, income status is estimated per household.
Then how could we classify the deciles of income status for every patient?
I mean, Is it possible that we classify the deciles of income status with only percentile for unit concept_id?

(Seng Chan You) #13

Yes, we need to develop the rule to store income status. I’m not sure I can figure it out quickly (I’ll be absent from 21th March ).
So I posted this issue to the THEMIS.