re: Real-World Data

We have all considered how good it will be to use Real-World Data in our research and analytics. There are many things to recommend it. With the addition of FHIR Bulk Data Export we anticipate that more data will made available from more sources. That data will be structured in an interoperable fashion and won’t suffer from as many “translation errors” because a lot of that data will have come from FHIR-based sources in the first place. Thus we should be able to combine BodyStructure with BodyStructure and MedicationRequest with MedicationRequest. This is all to the good.

Having worked in the realm of Real-World Data - for more decades than I care to admit - I have seen a lot of issues with the Real-World Data I’ve encountered. I believe there are a number of things that show up when we start processing Real-World data that make it quite challenging when we start wading through it. Following are some observations and concerns from the front line of the Real-World Data battlefield:

• Long Problem Lists: One of the things we’ve seen is that a list of conditions, or a problem list, can become rather heroic for any patient that has been in a system for any length of time. This actually raises a number of concerns:
• Ownership: It seems common practice that no provider wants to edit or delete a problem they didn’t contribute to the list. If another provider thought it important enough to document, would it be in this provider’s interest (or workflow) to evaluate each condition and determine if it is still in play? If not, whose job is it to keep that problem list current/complete/correct?
• Timeliness: The patient may have had a broken ankle last year, but should that still be in the problem list today? Whose job is it to resolve that particular problem so it no longer shows up on the list? That problem may have been transferred to, or from, another EHR via interoperability, so who would update that problem in the distant system?
• Provenance: Which provider contributed this problem to the list? Is that important (see next)?
• Ownership (again): Who (really) owns the data? We may receive data that is stored in electronic health record (EHR) systems, but is that data truly owned by the EHR company, the provider(s) managing that data in the EHR, or the organization which gainfully employs the provider(s)? This becomes apropos when it comes to question of accuracy and provenance. If there is an issue with Real-World Data, does it need to be resolved by the EHR company, the provider, or the hospital/health system in which the provider conducts his/her practice? Is it the IT department’s responsibility to clean that data up? What if the issue with the data is entirely clinical in nature?
• Timeliness (again): Take medications for example… The date on which a patient was prescribed a certain medication is well documented, as is the provider who wrote the script. But *when* did the patient *stop* taking that medication? Usually the patient doesn’t inform the provider when they stopped, and even if they did, the end date is probably not put into the EHR next to the prescribed-on date. One of the things I’ve seen is that for analytical purposes a medication end date is calculated on a formula like: “Date of last refill + number of pills that were given to the patient”. That’s a pretty sophisticated maneuver, and usually not available to providers providing primary or referral care.
• Provenance (again): As interoperability continues to increase (which I am happy about) it is likely that the data documented in this EHR may have come from any number of sources (not just EHRs!), and it may be difficult to see who/what is responsible for the observation, medication, etc.
• Focus of the data: My work is currently focused in oncology. When we receive data from an EHR (even an Oncology-focused EHR) there are often a rather-large number of diagnoses, medications, procedures, and reports that have nothing to do with the patient’s cancer diagnosis or treatment. There is usually no way to “filter out” the data that is germane to the patient’s cancer journey.
• Important data hidden in notes: There may be a place where a provider can put an important data element… However EHR usability, and time pressures on providers, often mean that important data often ends up in providers’ notes (only). That way the provider knows where it is when they need it. Besides, it may not be apparent which window/sub-window, tab/sub-tab, pane/sub-pane, menu/sub-menu this field should be placed in, and who has time to go look for it? (I don’t hold it against the EHRs, by the way, they have to organize and manage a tremendous amount of data)
• Dirty data: Often the data stored in the databases of our vaunted health-IT systems becomes an accumulation of years-worth of data that has not been curated (to say the least). It may be that there are known issues with data that has been entered, but does the person who spots the error have admin rights to fix it? Who does?
• Duplication: If I send a diagnostic report to a particular provider, and then another person in my practice sends the same report… will that second (redundant) report be deleted? By whom? I admire the nurse navigators and support people that keep the EHR updated and relevant. It’s a difficult and thankless task.

These are some of the issues of Real-World Data that I’ve encountered in my career. I believe we are all doing our very best to manage the tremendous amount of data collected in our healthcare system. But it is a tremendous amount of data! It's been said that if we have too-much data we really don't have any data at all. I just want to be sure we recognize that the myriad ways in which data gets into our healthcare IT systems, and the lack of any proper data governance, often mean that Real-World Data, as such, can be very challenging to gather and difficult to use when we need it. The data may be there, but can you get to it? Is the quality of the data good enough to use? Will the Real-World Data that we gather allow us to conduct the analysis, and gather "actionable insights", from the data that we’ve worked so hard to gather?

#realworlddata #electronichealthrecords #ehr #interoperability #healthcare #healthcareit