OHDSI Home | Forums | Wiki | Github

Why are supplements under device instead of drug domain?

(Selva) #1


During vocabulary mapping, I observed that nutritional supplements such as glucerna liquid/powder, jevity liquid, and pulmocare liquid/powder are under the device domain?

Shouldn’t they be under the drug domain?

Can someone please help me understand why is it assigned under the device domain?

(Oleg Zhuk) #2


According to current OHDSI documentation, all nutritional products and supplements are devices. I strongly believe that was done due to questionable effects of nutritional products on the body and lack of formulas’ standardization.

During the custom mapping process, you could spot similar substances from different vocabularies classified as devices in one case and drugs in another. That results from semi-automatic approach to drug-device separation, historical changes of the convention, and vagueness of its current version.

We are slowly moving towards clean reliable unambiguous vocabulary documentation and I hope one day we will stop arguing on this topic :slight_smile:.

(Selva) #3

Hi @zhuk,

Thanks for your response. May I check with you on a follow-up question?

Let’s say if our researchers want to find diabetes people who had been on some nutritional supplements (not sure whether it is a valid use case but let’s consider it valid for the sake of discussion).

So in Atlas, we should ask them to use the device domain for selecting supplements. But that doesn’t sound intuitive. Even when they search for drugs in the vocabulary tab, they might apply a filter for the drug domain to select these supplements.

So, in future OHDSI will move all these concepts to drug domain? will the concept id remain the same when all those concepts (for supplements) are transferred to drug domain?

(Christian Reich) #5

“Intuitive” is a matter of perspective. As @zhuk said, most people would find that supplements are not drugs, and don’t even have much of an effect. They are just purified food components sold at extraordinary prices. If the “researcher” knows they are devices - where is the problem?

@zhuk meant to say that all these definitions are prominently displayed and all concepts have a correct domain assignment.

(Selva) #6

Okay. Understood @Christian_Reich

(Piper Ranallo) #7

@Christian_Reich, probably an spot-on statement for a lot of the supplements out there! :slight_smile: But not all - see this article by by Zhang: Mining biomedical literature to explore interactions between cancer drugs and dietary supplements

(JD Liddil) #8

As @piper points out There are drug/supplement interactions. That is why with Hep C meds they ask patient about their supplements

(Christian Reich) #9


Now what do we do? If you think you have a solid use case for studying effects of supplements they would need to be modeled. Hopefully from some existing repository of such information. And you need data to support the research: rich information about supplement use as well as clinical data.

Does that exist?

For the former, we have a Vocab group who can help with the concept creation. But they need to start from something.