4.11.2023 By Sam Schifman, Chief Architect for Clinical Solutions at Availity & Adam Cole, Senior Solutions Architect and AVP Sales Engineering at Smile Digital Health
Fast Healthcare Interoperability Resources (FHIR®) is celebrating 10 years of development and use. This is partially due to the people at the non-profit Health Level Seven International (HL7®), who created an incredible tool for interoperability by applying lessons learned from earlier efforts and successfully convening broad stakeholder participation. In terms of healthcare data standards, FHIR is still the relatively new kid on the block.
Before FHIR, achieving interoperability was notoriously difficult and brittle because healthcare data would be stored—and effectively locked behind—proprietary systems across multiple vendor organizations. Therefore, achieving interoperability was notoriously difficult and brittle.
Fast forward to today, FHIR has enjoyed a widespread adoption from a global community of health leaders, as well as recent requirements of the Centers for Medicare & Medicaid Services (CMS) and the Office of the National Coordinator for Health Information Technology (ONC).
Why is data interoperability needed in healthcare?
A short answer would be to empower patients, increase the quality of care, and reduce information silos. But it would be helpful to step back and consider some drivers propelling us forward.
Thanks to sources like electronic health records (EHR), personal fitness applications, and home genome test kits, the growth rate of health-related data is growing substantially faster than other industries such as finance or manufacturing. Data researchers estimate approximately 30% of the world’s data volume is generated by the healthcare industry.1 However, as data about our health piles up—the healthcare industry and patients should collectively understand a lot more than we used to about the human body, disease, and health outcomes. But having a lot of data is not enough.
Patients, providers, and payers must be aware of the quality of data they have access to, understand what it means, and act on that understanding. While the challenges are in some ways more acute in the United States because of its fragmented system of care, they exist in health care across the globe. According to a report conducted by the Commonwealth Fund, the U.S. has the lowest life expectancy at birth, the highest death rates for avoidable or treatable conditions, and the highest rate of people with multiple chronic conditions.2
The healthcare system will continue to face challenges as more and more people with chronic diseases age into Medicare every year. Currently, an estimated 61.5 million people living in the United Stats are enrolled in Medicare3 and 86% of them have one or more chronic conditions.4 This number will continue on an upward trajectory with two-thirds of the world’s population expected to be over the age of 60 by 2050.5
The growing complexity of providing comprehensive care has been a driving force behind enhanced medical research and documentation. However, researchers found that practices that implemented EHRs saw an increase in stress.6 The cumbersome process of data entry, government regulations, and lack of autonomy, among other variables, have led to a rise in practitioner burnout. Today, research shows nearly two-thirds of healthcare professionals are experiencing at least one symptom of burnout.7 This will lead to an inevitable increase in demand while the industry is heading straight for a decrease in supply.
There’s no doubt that patients will be directly impacted, but the burden will also be felt by caregivers and health plans. Middle-aged adults, who often do not live near their older relatives, will have no other choice but to step in and act as caregivers in increasing numbers. In fact, a recent study done by the Centers for Disease Control and Prevention (CDC) found that 22.3% of adults reported providing care for or assistance to a friend or family member in the past 30 days (about four and a half weeks).8 Alongside the ability for health plans and providers to leverage clinical data as a strategic asset, providing patients and their loved ones with reliable healthcare data is becoming more and more critical for delivering quality care and improved outcomes.
In this regard, health plans can play a significant role by developing a long-term data strategy centered around high-quality and interoperable clinical data assets, providing support for caregivers, and other non-traditional clinical data consumers. This includes family caregivers who may lack the medical expertise required to comprehend the intricate coding variations that constitute raw clinical data assets from EHRs. Non-traditional clinical data consumers will need reliable and comprehensive data, including reference ranges and meaningful descriptions, with the help of technology.
Although current technology-based solutions to improve data quality and interoperability may not prioritize non-traditional clinical data consumer needs at the moment, it's essential to pivot towards a consumer-focused approach for the future. AI can play a critical role in this convergence of supply (access to high-quality FHIR data) and demand (the shift of daily care responsibilities away from medical professionals). While AI will never replace human medical professionals, it can assist non-medical professionals in managing care between medical encounters. By creating a comprehensive data strategy that prioritizes data quality and interoperability, healthcare organizations can lay the groundwork for success in the ever-changing healthcare landscape. To learn more about how clinical data can be a powerful, strategic asset when it’s actionable, accessible, and prepared for use, download Availity’s 2023 Clinical Data Integration Buyer's Guide.
Why does data quality matter for FHIR?
FHIR combines three powerful concepts: small document/resource size, a query-able API, and extensibility. Together, these aspects are designed to revolutionize how people in all roles interact with healthcare data.
However, healthcare data is uniquely complex, ranging from administrative data, such as claims, to clinical data, which includes patient charts documented in EHRs. While FHIR covers all facets of data, clinical data is arguably the most important because it contains rich information about an individual's health journey and includes timely, precise indicators that are not always present in other data sources, such as claims data.
Clinical data, in raw form, often contains inconsistent or missing codes, duplicate entries, and information documented in incorrect places. This is due to the lack of standardization caused by non-conformant codes, details buried in text, redundancies, and other systemic data quality issues as well as the variation in formats, from Consolidated Clinical Document Architecture (C-CDAs) to flat files to Admit, Discharge, and Transfer (ADT) messages. Challenges with data quality issues in this area will only grow more complex as more data sources, such as patient-generated information from wearables, need to be collected.
There is significant variability in how EHRs code and represent values with the lack of data standardization. This can lead to essential data being either incomplete or absent.
Furthermore, most EHRs assume the consumer of the data is a trained medical professional who understands particular jargon, has specific knowledge, and can fill in the gaps. For example, it is a reasonable expectation that a doctor will know if a drug is an opioid or at least know how to find out if it is in a timely manner. Unfortunately, this assumption is not true for EHR users that increasingly include administrative staff, patients, and their caregivers, and even computers/AI.
Healthcare institutions also have their own unique way of documenting patient information. For example, one clinician might document a flu shot in the EHR’s procedures section while another might use a narrative note. This lack of standardization can lead to gaps in care and decision making based on fragmented information. Other factors, like licensing code systems and the need for abbreviations, sometimes conflict across specialties. The result is that there is little conformity in clinical data.
Certainly, FHIR addresses all these problems, right? Well, it does move the bar significantly by providing standard structure and requiring consistency in terminology via code systems in specific places. However, if the original data was coded in the EHR in another code system, then there needs to be logic to crosswalk to the one specified by FHIR. But what if there is no code, only a description? If the original data was recorded in a nonstandard way, the resulting FHIR data is often flawed unless other steps are taken. This lack of standardization makes it difficult for public health departments, for example, to determine appropriate responses in a timely fashion. For example, there are wide discrepancies in both the code and descriptions of COVID-19 test results. So, if a developer writes an app that pulls FHIR data to check for COVID-19 testing and care recommendations, their app could populate erroneous information.
Example of the wide variation in test data from labs
There are over 40 ways that a negative COVID-19 result is coded within these tests, not to mention the hundreds of ways the test itself is coded.
Example of coding variation across multiple EHRs
Below is an example of the wide variation in how a hemoglobin A1c (HbA1c) test might be documented in a medical record. For analytics, this data should map to a single LOINC code (LOINC 4548-4) and description.
FHIR Alone is Not Enough
FHIR addresses some of these issues by sharing data in a standard structure, but it does not address the meaning of data and how it is represented and interpreted within a given context or system. Fixing the semantics is key to making it actionable. Many of the remaining problems can be solved by applying some simple rules to the data. When codes are available, but semantically incorrect in the context, these rules can crosswalk the codes from one coding system to another, but they can also use text descriptions to look up codes. For example, a "systolic blood pressure" description maps easily to a code and even implies the units used to measure it. All of which can be inserted, if missing.
At the same time, the data can be tagged with meaningful metadata that can help improve its usability. For example, suppose all opioids are labeled as such during this process. In that case, downstream systems can consistently and accurately report on Morphine Milligram Equivalence (MME) or similar drug classes or even broader clinically relevant contextual knowledge. Other rules can be applied to the data to reorganize it by clinical intent. For example, finding future-dated procedures and moving them to a plan of care. Duplicated data can also be fixed. It requires certain complex logic, often supported by the metadata found earlier. For example, metadata about the classification of a condition can be used to understand its chronic nature and, if it is episodic, the window of time that the condition lasts. Rules can be written using this information to understand that all flu diagnoses within a 10-day period are the same, but a diagnosis two months later is not. But all the diagnoses of hypertension, a chronic condition, are likely to repeat the first one.
At Availity, we have coined the phrase Upcycling Data™ to refer to our automated, API, and cloud-based solution for generating high-quality, semantically interoperable data assets.
How Upcycled Data and Clinical Data Repositories Help
Availity’s automated data transformation engine, Availity Fusion, produces Upcycled Data—data that is normalized to national standards, interoperable, deduplicated, consolidated into a longitudinal health record, and available in fit-for-purpose packages for flexible deployment at scale. It can process an average-sized continuity of care document (CCD) in less than a second. That data is converted to high-quality FHIR data and can power multiple downstream systems. This work only needs to be done once, and then it can be stored in a clinical data repository (CDR) and repeatedly used with confidence, forming the basis of a high-quality longitudinal health record. Even in situations where different display output is required, having the data in the CDR semantically standardized, or upcycled, means that downstream display logic can count on consistent representations. This dramatically reduces the complexity of the translation process.
That’s where Smile Digital Health comes in. FHIR provides standards and definitions for data structure, data access, identity management, tracking updates and data provenance, patterns of use (i.e., implementation guides), robust RESTful APIs, etc.; and Smile Digital Health is recognized the world over as the preeminent FHIR server. As the reference implementation of FHIR in Java, Smile Digital Health provides the most robust implementation of the FHIR specifications.
Smile Digital Health's platform recognizes all the benefits of FHIR alongside myriad of industry-leading data-sharing features such as a FHIR Gateway, MDM, security and consent management, federated IAM, etc., in a single solution. In coordination, these features underpin a Health Data Fabric. The combination of quality Upcycled Data in a Health Data Fabric empowers health enterprises to provide genuine coordinated care; and, equally exciting, rise above the details of interoperability to focus on modernizing healthcare and innovation to recognize true business value. On the other hand, if the first uses of FHIR rely on sub-par data and a sub-par data repository, there is a real chance of failure. This is not because FHIR is flawed but because the data quality, comprehensiveness, and access methods have issues.
The story will be completely different with Upcycled Data and a best-of-breed. The most meaningful data can be made available at scale. The positive outcomes are too numerous to count, and they will help FHIR become the foundation of a modernized, future-oriented health system. FHIR, in conjunction with the Patient Access API, is a game changer. The data is now available through public endpoints in an open, structured, and normalized format. And developers/implementers love that this data is accessible via well-defined RESTful APIs. Leading FHIR vendors such as Smile Digital Health are CMS and ONC compliant, in addition to providing:
- Multiple security layers and subscriptions in an Event Driven Architecture.
- Provenance, FHIR operations, CQL, support for automatically loading and validating against unlimited Implementation Guides (IGs).
Effectively providing futureproofing against forthcoming Da Vinci IGs and CMS and ONC mandates, SMART on FHIR support, real-world high performance at web scale, as well as a host of other benefits collectively allow for the enterprise to transition to rapid interoperability with ad hoc connectivity to other certified FHIR systems. For Smile Digital Health customers, this translates to increasingly lower-cost implementations and maintenance, faster go-lives, and ROI recognition.
As we have seen in many other industries, such as airlines, finance, planning and logistics, etc., innovation is unlocked when data is shared through open standards and APIs. Activities that we can only dream of today become the reality of tomorrow. With FHIR, healthcare is truly on the eve of tomorrow. Rapid and increasingly seamless interoperability has only as much value as the quality of the data. Ultimately, FHIR’s success will be internally assessed by how effortlessly data can be exchanged within a network. The best implementations of FHIR will mean nothing if plagued by data interoperability issues.
Bringing high-quality, Upcycled Data together with a best-of-breed CDR such as Smile Digital Health’s open-standards FHIR platform will enable users to enjoy a single architecture that optimizes the value of their data assets - interconnected within the ecosystem of their peers. This can be game-changing and opens the possibility of truly addressing the challenges for the future of healthcare. Reliable data quality in healthcare is good for one and all. How will your organization adapt and recognize the transformative value of this new reality?
1 “Healthcare's Data Tsunami.” Brunswick, https://www.brunswickgroup.com/healthcare-data-i20729/#:~:text=As%20a%20result%2C%20approximately%2030,health%20data%20will%20be%2036%25.
2 “U.S. Health Care from a Global Perspective, 2022: Accelerating Spending, Worsening Outcomes.” U.S. Health Care from a Global Perspective, 2022 | Commonwealth Fund, 31 Jan. 2023, https://www.commonwealthfund.org/publications/issue-briefs/2023/jan/us-health-care-global-perspective-2022.
3 “Research, Statistics, Data & Systems.” CMS, https://www.cms.gov/Research-Statistics-Data-and-Systems/ Research-Statistics-Data-and-Systems.
4 “Health Policy Data Requests - Percent of U.S. Adults 55 and over with Chronic Conditions.” Centers for Disease Control and Prevention, 6 Nov. 2015, https://www.cdc.gov/nchs/health_policy/adult_chronic_ conditions.htm.
5 Ageing and Health.” World Health Organization, World Health Organization, 1 Oct. 2022, https://www.who.int/news-room/fact-sheets/detail/ageing-and-health.
6 “Physician Burnout.” Agency for Healthcare Research and Quality, July 2017, https://www.ahrq.gov/ prevention/clinician/ahrq-works/burnout/index.html.
7Meyer, Holly. “The Great Resignation, Covid-19 and Physician Burnout.” Providertech, 1 Feb. 2023, https://www.providertech.com/great-resignation-covid-19-physician-burnout/.
8 “Caregiving for Family and Friends - A Public Health Issue.” Centers for Disease Control and Prevention, https://www.cdc.gov/aging/agingdata/docs/caregiver-brief-508.pdf.