Hello! I’m pretty new to FHIR but have taken on making sure our tech and software is FHIR compliant.
We are starting a clinical trial later this year in which we will have patients enrolling and using a hardware device for monitoring. I’m trying to understand what information we need to be collecting on those patients with regards to organizations and locations to be FHIR compliant going forward. We will eventually be supporting the prescribing of these devices, but are a ways away from that. My product manager is convinced that we need to collect this information and it needs to be using the FHIR models complete with required fields. I believe we only need to collect what we need to do our trial, as I don’t believe we will be exchanging any of this information with anybody else.
Can anybody provide some educated insights or thoughts on this?
The base FHIR standard is extremely loose. It defines the core elements, but exceptionally few of them are mandatory - that’s because the resources might be used in a wide variety of circumstances and the cardinality and other constraints have to allow for all of them. As a result, mandatories are generally limited to "what’s necessary to make use of the resource in any context.
However, FHIR also supports defining implementation guides that impose additional constraints for how the FHIR standard is used in specific contexts. These implementation guides will enforce tighter constraints - both on elements as well as on how data is passed from system to system.
In the research space, implementation guide development is just getting underway, so you’re unlikely to find any that are mandated as yet, but you might still want to track what’s being developed and consider how they might affect you eventually - and also provide feedback to help ensure that the rules they define will be ones that will work for your system(s). The group that’s driving a lot of this work is the Vulcan FHIR accelerator - you can see more about what they’re up to here: Vulcan Accelerator Home - Vulcan Accelerator - Confluence
In terms of how much to bother exposing as FHIR now, vs later, the YAGNI principle says to only build what you know you’re going to need - however you also need to pay attention to what’s cheapest. Certainly if you have a choice between exposing data over FHIR or something custom, I’d plan to expose all your data over FHIR, even if you don’t have FHIR consumers for most of it yet. FHIR is going to give you a set of tools as well as the consolidated expertise of hundreds of experts and will increase the likelihood of your interface standing the test of time. However, if it’s a question of exposing as FHIR or not exposing at all, it may make sense to wait - that way whatever you build will be more likely to meet the expectations of whatever implementation guides or regulations are relevant.