How Data is Failing Medicine Today & When Data Can Become Actually Useful in Medicine (2023–2024 Update)

Sherri Douville
12 min readAug 19, 2018

The purpose of this post is to help you understand the status of current available data specifically in medicine and how to track where it’s heading. This is updated as of August 10, 2023.

Version 4 of USCDI has been released including “The new data elements — from allergies to health assessments, procedures, vital signs and others.

Changes to certain data elements proposed in Draft USCDI v4 clarify the definitions, uses and standards of data elements.

ONC is also accepting submissions for new data elements for USCDI v5 and feedback on existing data elements until September 20, 2023.” [1]

ONC previously released US Core Data for Interoperability Version 3 which adds data elements for Health Status, Procedures, Health Insurance, and Patient Demographics[2] and Lab elements[3], data classes. Hot tip: if your job doesn’t require you to understand the fine data and technical details in a U.S. context? Scroll straight to the conclusion to see the “what, so what.”

I was inspired to write this post originally based on feedback from one of my favorite health plan CIO’s who agreed to the sentiment that failure of tech in healthcare stems from when a team “doesn’t have perspective, comprehension, or experience for the market or its changes.”

This post is for the person who basically understands how electronic medical records work and who knows that there is a problem today in movement of clinical data between settings (hospital, between two hospitals, clinic) and computer systems to optimally treat patients. You know some evolution is happening, but you’re confused about its status and how to track it. We focus at Medigram on empowering the physician within a team context and we believe that doing so will have the biggest impact on improving quality and reducing cost. We explain exactly where clinical data flow is today and how to track its progress below though first, briefly we describe what this post is not about.

  1. This post is not about consumer generated data:

This post is not about consumer generated health data like wearables and patient-facing wellness apps. While we believe that engaging consumers is important, until fuller health literacy is more widely achieved, we believe that stakeholder efforts should focus predominantly on teaching people how to be healthy prior to attempting specific behavior change. Why? One study showed that when doctors tell heart patients they will die if they don’t change their habits, only one in seven will be able to follow through successfully. [4] Americans still trust their doctors. There’s a lot of work to be done to make all of this data useful and I’d point you to this article in Medscape for a background on that

Further, only 12% of Americans have health literacy. The National Assessment of Adult Literacy (NAAL) provides information on the literacy/health literacy levels of the U.S. adult population. The NAAL is the only large-scale survey of health literacy.”

Who is responsible for peoples’ health literacy? Who’s incentivized to improve this situation? How can it be operationalized and measured or improved in a timely way (our lifetimes). Happy to hear perspectives, especially if they include key strategic elements, operating details, and how it could be measured etc.

Defining Paper on healthcare literacy:

2. This post is not about financial data found in claims which does not include any clinical data. This post is also not about claims data from insurers, a valuable set of data that payers have which incorporate every party who has submitted a bill (claim) to the insurer. Aside from not being available until after 30–90 days, it’s often incomplete or has missing data of variable quality anyway.

3. This post is not about desktop-oriented interoperability efforts. There are many seemingly worthwhile interoperability efforts underway that for a variety of reasons, primarily only work well on a good connection in a desktop context. Because physicians are on the go, we do not focus on this context.

Brendan Keeler, CPO at Flexpa brilliantly points out here a lot of these persistent, unaddressed mainstream gaps even through 2024 policy initiatives. He also demystifies and organizes what to know about data sharing from EMRs here. He was gracious to allow me to post the TL;DR summary here.

Know your use case: Is Interoperability for Desktop or mobile?

4. This post does not describe how to do EMR integration today. Here is an explanation of that. The current way is difficult, labor intensive, custom, and not scalable but possible when necessary.

If you want to know what’s happening in healthcare data for medicine, you need to track the FHIR standard (Fast Health Interoperability Resources). It’s similar to if you really need to understand computing for example, how you would have to understand the TCP/IP protocol, the Transmission Control Protocol/Internet Protocol, which is a suite of communication protocols used to connect network devices on the internet. In plain terms, TCP/IP is what allows data to move. Remember, data is oxygen to apps, they don’t open without them. When I described FHIR (Fast health Interoperability Resources) to one colleague, she said. “that sounds like the TCP/IP for healthcare,” and I said, BINGO you’re right. The problem is that FHIR is not fully built out and that’s why it’s confusing. In order to make a difference in cost and quality in healthcare, you have to be able to help Grandma in her 80’s. She will commonly have co-morbid conditions (two or more such as heart disease AND diabetes).

How do data elements getting included/added?

It’s worth noting that this evolution of data interoperability is not some kind of secret club or process. All healthcare stakeholders are invited to submit suggestions for priority data elements that meet USCDI’s aims of “improving patient care, population health, reducing care costs, or assisting providers.” This is referred to by ONC as a “predictable, transparent and collaborative process.” They can participate via the ONC New Data Element and Class (ONDEC) submission system during the designated cycles. For context, V4 submissions period just ended on September 30.

There are three levels of USCDI data elements, of increasing importance.

To become an element, there needs to be a vocabulary, terminology, content, or structural standard in existence for the data element. All elements must be expressed with standard computer-processable healthcare terminology, like SNOMED CT.

Terminology Examples Source: USCDI v1 July 2020 Errata

ONC posted this updated USCDI v3 Errata for view and download Nov. 14, 2022.

When comparing medical use cases to consumer ones, I like to compare airplanes. For example, fighter jets are on a dangerous, time sensitive mission and Boeing 747 passenger planes have a lot more space and time. To discern using FHIR on iOS for medicine vs. CONSUMERS, you need to really dig down into what kind of data you need, how often you need it, and whether you need to write it back. Some EHRs don’t capture all of the data you might need in a medical context. Some EHRs have bad data, where they don’t use standard codes like SNOMED or ICD-10. And some EHR’s may not support bi-directional data exchange. Apple, for example is doing a good job on and is very motivated by the CONSUMER use case though what we care about at Medigram is enabling truly reliable physician communication for faster, better medicine. Apple is bringing some basic FHIR resources into their app. Patients as consumers are able to see things like allergies, medications, conditions, and immunizations, as well as the sort of things they would check an EHR patient portal for, such as lab results.

25 year old Cassie checks her allergy meds even though Cassie does not require much in the way of medical resources.

By contrast, helping Grandma requires having flow of data related to concepts that are relevant to specialties and genetics and things like that are yet to be fully built out in FHIR. What needs to happen is that the entire FHIR community and clinicians need to work together on one data element at a time, and agree on a model, or how to represent that piece of data. The only mechanism that can push progress on this are provisions from the CURES Act which are described below. You have to have that to really have interoperability across organizations for complex patients with co-morbidities in order to really help health systems improve care.

FHIR is not quite ready to help doctors treat Grandma yet.

Some EMR’s are providing API’s and a sandbox environment. FHIR [Fast Healthcare Interoperability Resources] is technically two things:

  1. Discrete bundles of data explained courtesy of Michelle Currie MS, RN, CPHQ, CPHIMS who explains that literacy for these data models is low and urgent to improve in the industry.

“There is the FHIR data model which enables you to package discrete bundles of data.

The capability to create a bundle of data is enabled by the FHIR data model. The R in FHIR stands for Resources which are the equivalent of a data Class. The Resource defines the attributes and the technical and semantic constraints on the attributes like cardinality, data types, and CodeSets. Data Modeling is not the responsibility of technical staff. Data modeling requires deep expertise in medicine, care delivery, and the industry to accurately model concepts and the relationships between them. This is the work of Clinical Informaticists/Information Scientists/Knowledge Artifact creators.”

Further, Michelle reports that “Semantic Data Quality can help reduce clinician burnout (by increasing the reusability of the data) as well as improving the ability to design more specific and sensitive quality measures to help patients.”

2. Then there is a thing called the FHIR transport standard which enables one application to be able to connect in a physical way with the data repository underneath it, like the EHR.

Healthcare going back decades has been silos of information and care. Treatment has tended to focus on independent specialists treating separate aspects of the patient condition without fully addressing how they impact one another. This makes as much sense as changing the oil on a car with four flat tires. To fully bust these silos, we need interoperability. The Medigram team has built the only system designed to really work on mobile for physicians to collaborate in the context of the conditions in which they practice. Data doesn’t matter if it doesn’t get to the doctor.

How to track FHIR?

The Draft US Core Data for Interoperability (USCDI) and its proposed expansion process aim to achieve the goals set forth in the Cures Act by specifying a common set of data classes that are required for interoperable exchange. Rule making for the act is : “Trusted Exchange Framework” and USCDI is part of that. I would watch that to pace the implementation of clinical data elements into practice of FHIR exchange. (This is when data will become actually available and useful in medicine).

USCDI v4 the latest summary is pictured above in the large white table

USCDI v3 was finalized in July 2022. This version has 94 data elements, up from the initial 52 data elements in USCDI v1.

USCDI v3 “focuses on promoting equity, reducing disparities, and supporting public health data interoperability”, ONC officials wrote. It added two new data classes and 24 new data elements. Here are the new data elements made available.

Draft USCDI V3 Data Elements, image credit: ONC

Update From Monday, July 19, 2021 (What’s live now including May 2022)

For those interested in broader access to electronic health records: (though primarily desktop computing) Learn about TEFCA, an initiative aimed at creating national connectivity of electronic health records and QHIN, a network of organizations working together to share data. here

ONC released these details of United States Core Data for Interoperability (USCDI) v2 along with the Standards Version Advancement Process (SVAP).[5].

“Included data classes and elements are:

  • Care team members: provider name, provider identifier
  • Encounter information: encounter type, encounter diagnosis, encounter time
  • Problems: date of diagnosis, date of resolution
  • Diagnostic imaging: diagnostic imaging order, diagnostic imaging report” [5]

Source: This HITAC meeting presentation

Track progress here:

What are the highlights on interoperability?

“As part of the 21st Century Cures Act, the federal government will require that all healthcare providers grant patients access to their personal health information starting July 1, 2021.”[6]

“Fast Healthcare Interoperability Resources (FHIR) Release 4.0.1 has been determined as the foundational data exchange standard.”[7]

I am a Healthcare Executive, What Do I Do About Complying With the Info Blocking Rules? Medigram CEO, Sherri Douville served on a panel with experts at VIVE who provided the following guidance to health system executives

Great Example From NYU: Policy Against Information Blocking of Electronic Health Information:

Draft USCDI v5 is available for public feedback until April 15, 2024, at 11:59 pm ET. ONC then expects to release the final USCDI v5 in July 2024. V5 will focus on new data elements aimed at driving health data exchange to reduce health disparities and improve patient care.


Prior interoperability efforts satisfied the need to provide limited basic access for patient health information directly to patients. The latest proposed data elements for V4 are intended to support info blocking compliance and reflect priorities for USCDI expansion that include assessment for industry-wide implementation impact. Prior, V3 elements are intended to advance health equity and public health data interoperability functionality. The updates to USCDI V4 and V3 still do not substantially change the clinical relevance specific to the most complex patients with co-morbidities whose care consumes a disproportionate amount of resources and physicians’ time. Think of it more like long term basic FHIR building blocks instead.

However, as of July 19, 2022 it was a significant advance to add the inclusion of health insurance status to V3 As described above.

“Draft USCDI v4 includes data elements that focus on patient care and facilitating patient access while promoting equity, reducing disparities, supporting underserved communities, integrating behavioral health, as well as encounter location information.” The facility is a new class of data.[8]

Though the “big picture” remains that in order for FHIR to gain meaningful traction in health systems, it must increase BOTH security safety and clinical safety profiles while removing any potential of even perception of responsibility and/or liability for MD’s & health systems.

We hope this helps you track where the useful practical application of data interoperability and access for patients and physicians is going.

P.S. What is an API?


By: Sherri Douville, CEO at Medigram

By Sherri Douville, CEO at Medigram, the Mobile Medicine company. Recognized in 8 categories of top CEOs by Board Room Media (Across SMS, mHealth, iOS, IT, Database, Big Data, Android, Healthcare). Top ranked medical market executive worldwide and #1 ranked in mobile technology categories (mhealth, iOS, Android), #1–2 (on any given day) for the cybersecurity market in the U.S. on Crunchbase. Best selling editor/author, Mobile Medicine: Overcoming People, Culture, and Governance & Advanced Health Technology: Managing Risk While Tackling Barriers to Rapid Acceleration, Taylor & Francis; Series Editor for Trustworthy Technology & Innovation + Trustworthy Technology & Innovation in Healthcare.

[1] ONC publishes new USCDI v4 standards



[4] Kegan and Lahey “Immunity to Change”

[5] USCDI Draft v2 and SVP Standards Approved by ONC for 2020 as posted by Healthcare IT Today

[6] What Providers Should Know for CMS Interoperability Rule Compliance

[7] ONC Releases US Core Data for Interoperability Draft Version 2

[8] ONC Releases Draft Version 4 of USCDI “ONC proposes to add 20 data elements across one new (Facility Information) and eight existing data classes”