It’s the Quality That Counts

I. Introduction

If you’re reading this, then you already know that healthcare in the US is complicated. Very, very complicated. It hasn’t always been that complicated, but that’s because it hasn’t always tried to put the patient at the center of the industry. We’re all familiar with the concept of “fee-for-service”, that being that if you are sick or injured, you go to the doctor or hospital, and regardless of the outcome, you get sent a bill sometime in the coming weeks asking you to pay whatever balance remains after (hopefully) your insurance has paid their portion. This is a simple transaction, much like any other service we pay for in our daily lives - the hair salon, a lawn service, etc. There is a difference here though - if we are dissatisfied with our haircut or the landscaping work, then we can have those service providers correct it for us as part of the transaction for no additional fee (usually). Can we do that with healthcare? If I go to my primary care provider with some kind of complaint, and they prescribe some form of medication or treatment that doesn’t make me better, can I withhold payment or ask them to fix me for free? The short answer is “no”.

I want to take some time in this blog to explore the transition from this fee-for-service (FFS) model, to a more modern and conscientious fee-for-value (FFV) or value-based care (VBC) model that puts the patient at the center and asks the question “did we provide the highest level of care for this patient, and were the outcomes satisfactory?”. Further to that, I want to take a look at the expanding role of FHIR in transforming these quality payment models, enabling data to move around in a standardized way that makes for a truly interoperable healthcare experience for both healthcare providers and patients, and allowing for accurate feedback and reporting to government agencies and other entities involved in improving patient outcomes and ensuring appropriate reimbursement for providers who participate in these programs.

II. The Need for Transition

The fee-for-service reimbursement model is antiquated and no longer suited to the modern climate. In a world driven by data and in which people have access to multiple different methods of care delivery, we need to ensure that patients’ needs are being met while offering the most value possible at the point of care and beyond. In our hyper-connected world, people are growing increasingly frustrated at dealing with an industry that seems to care more about profits and policies than it does about the health and wellbeing of the US population. A healthy population is a productive one and we appear to be losing the battle against the prevalence of chronic diseases (https://www.cdc.gov/pcd/issues/2024/23_0267.htm), this is why it’s becoming increasingly important to ensure that patients receive the best possible treatment for their issues - both chronic and acute - and that we use all of the data that we have available to ensure that treatment is continually improved. Fee-for-service can’t provide that, it can only provide a pricing model based on aggregation of cost/benefit data. It’s time to focus more of our resources on moving away from the FFS paradigm and into a fee-for-value model that incentivizes higher quality of care.

Value-based payment models come in a variety of shapes and sizes. The most common format we see is a reimbursement model that is based on quality measurement. Participating provider organizations and facilities agree to participate in these models and are granted that participation based on meeting certain criteria. The quality measures are authored by clinical experts as well as industry experts and are based heavily on empirical evidence that shows improved outcomes for specific patient populations based on specific measurable metrics. These can range from safe use of opioids for treatment, to ensuring timely delivery of care in the Emergency Room and so much more. The focus shifts away from ensuring that the patient and their insurance payer is billed for everything possible, to focusing on whether the patient was provided with effective treatment for their conditions. What a novel concept! Further measures and quality programs focus on follow up care and post-acute care, ensuring that patient outcomes are tracked and improved outcomes are rewarded. Providers are incentivized to focus on providing quality care, rather than reimbursed regardless of whether the patient received effective treatment or not.

The measurement of clinical quality based on clearly-defined guidelines is the key to ensuring that the value-based payment model can work. How else can we measure the value of the care being provided by doctors and hospitals across the United States? Quality measures aren’t really “new” in the same way that the latest Marvel movie is “new” - they’ve been around for a while now as our government and particularly CMS (Centers for Medicare and Medicaid Services) have been seeking ways to innovate the reimbursement model for public healthcare in the US for a while now. What is new however is the use of emerging technologies such as FHIR (Fast Healthcare Interoperability Resources) and CQL (Clinical Quality Language) to both exchange the clinical data needed to evaluate measures, and to perform the calculations to produce measure scores for data submitters.

III. The Current State

As mentioned, there have been quality measurement efforts underway at CMS for some time - beginning in 2001 (See this paper from the National Institute of Health for a history of these efforts!). Programs and models that seek to improve healthcare for patients and ensure that providers have accurate and evidence-based guidelines have existed for over two decades, with standards such as HEDIS (Healthcare Effectiveness Data and Information Set) existing long before that. While the measure guidelines do a good job of describing what data is required in order to accurately measure healthcare quality, the how of exchanging that data and evaluating the data isn’t very well defined. The current methods are varied, and will differ depending on the program, the model, the submission period, and a host of other factors. A provider organization might have to submit their data in three or four different ways - all to CMS, but to different areas within CMS for different payment models or value-based payment programs. 

If all of the measures and submitters and programs leveraged the same methods for submitting data and evaluating data, then the ecosystem could be viewed as somewhat efficient, however this is not the case. There still exist many other ways to submit the necessary data depending on the program or model, and many different ways to evaluate the data and calculate the measures - leading to inefficient processes and a resource-intensive burden on CMS or other entities performing the calculations and providing reporting to submitters and other systems. The need for standardization of both the data model and the exchange mechanism is clear as we move into an era where consolidation of effort and increasing efficiency are paramount.

IV. eCQMs and dQMs: The Potential of FHIR

Many of the most recent measures come in the form of eCQMs (electronic Clinical Quality Measures), these being computable measure logic libraries that do not require manual interpretation and encoding into some sort of business logic engine. The idea behind this being that the data requirements and measure calculation logic can be expressed in a computable, standardized form allowing both the submitters and the system calculating the measure to have the same computable logic. The first Trump administration pledged to move all of the eCQM measures into dQM (digital Quality Measures) by 2025. DQMs differ slightly from eCQMs in that they are intended to be a wholly self-contained downloadable package containing not only the computable measure logic and data requirements, but also any needed value sets and related libraries to perform the measure evaluation. DQMs also leverage data from a wide variety of sources, including claims, registries, patient-submitted data, etc. eCQMs’ scope is generally limited to data available in an EHR (Electronic Health Record) system.

WIth both of these measure types, the potential for leveraging the FHIR standard as a modern, standard, and efficient method of exchanging data and evaluating measures over that data is as exciting as it is important. With a standard way of submitting/collecting data, both provider, payer, and vendor systems can align around a holistic standard that provides built-in validation and addresses the need for system interoperability across the US healthcare landscape. Disparate systems using proprietary data models and non-standard measure evaluation methods lead to massive inefficiencies in value-based incentive models and long lead times for evidence-based medicine to update clinical practice guidelines so that patient outcomes can be improved. Much of the momentum behind adopting FHIR for quality measurement is centered around shortening the cycle timescale between data collection -> research -> measure evaluation -> evidence-based medicine updates -> clinical practice guideline updates -> clinical decision support. Clinicians need clear, up-to-date guidelines on providing the best possible care to their patients, they need clinical decision support systems based on the most current research that leverages the data we use to evaluate the quality of care being provided. 

FHIR provides not only a direct, like-for-like replacement of the QRDA formats for individual and summary reporting, but it provides ways to encapsulate the measure libraries and related value sets, it also provides standardized operations for obtaining needed knowledge resources and actually evaluating a set of FHIR resources for one or more measures. FHIR APIs allow for simple, modern, secure, and efficient data ingestion from multiple sources by using the REST API paradigm and the Bulk FHIR exchange implementation. In the future, FHIR subscriptions can be created and leveraged to allow submitters to automatically push new or updated data to receiving systems such as CMS to increase efficiency even further, lowering the burden on providers and allowing them to focus time, money, and effort on what they do best - caring for their patients.

V. CQL and Interoperable Quality Measures

Historically, quality measures were evaluated using a data model known as QDM (Quality Data Model), and this standardized data model allowed for measure evaluation to leverage multiple different rules engines or other programmatic ways to calculate a measure score by sorting and filtering QDM rows based on various parameters to create a denominator (eligible population to be evaluated), and a numerator (number of patients in that population that had the “quality” care provided to them). Generally, a simple calculation of numerator divided by denominator will provide a “measure score”. The higher this score, the better quality of care being provided to patients. 

We’ve already looked at some of the reasons that FHIR can improve efficiency and standardization of processes within this ecosystem, but FHIR is primarily a data model - therefore we need a way to evaluate measures using FHIR instead of QDM. Enter CQL - the Clinical Quality Language. This is a simple declarative, functional, domain-specific language that is designed for evaluating FHIR data for quality measurement, though it is capable of defining measures using the QDM also. Computable CQL libraries can be embedded within FHIR Measure resources to support both eCQMs and dQMs, thereby ensuring an interoperable quality measures ecosystem.

VI. Addressing the Barriers to Using FHIR

While all of this sounds promising - and it definitely is - there are some significant barriers to realizing the potential of FHIR and CQL in becoming THE standard for quality measurement in the US. 

• Need for mature and consistent tooling - There is a current lack of maturity in CQL/clinical reasoning software. Much in the same way that there are multiple open source and closed source FHIR server technologies available to healthcare entities wishing to support FHIR-based data exchange, there are also a few different options for measure evaluation software. This can of course lead to differences in measure scores owing to minor differences in how the CQL libraries are executed over the FHIR data set. Work is ongoing within the Digital Quality Implementers Community (DQIC) to standardize both CQL and the FHIR resource profiles used to perform measure evaluations, and this work will be important in enabling FHIR quality measures at scale.
• Insufficient data requirements - Current baseline core data requirements for interoperability are insufficient for quality measurement. The USCDI v1 (current baseline standard) and even USCDI v3 (upcoming baseline standard as of 01/01/2027) do not contain all of the necessary data points needed to satisfy the requirements of almost all of the eCQM/dQMs at CMS. There is a need to raise that baseline standard for any Certified Health IT who want to participate in the quality payment programs to either the QI Core (Quality Improvement Core) data set, or the USCDI+ Quality data set. This would require updates to existing legislation in order to make this a reality.
• Questions around scalability - Bulk FHIR operations are as-yet not performant enough to ensure that large data sets can be exchanged for the purpose of quality measurement. When looking at scales of 100,000+ patients with potentially 1000+ resources per patient, the timescale to validate and intake that amount of data, even at a throughput rate of 1000 resources per second, is much too long to be practical (> 27 hours!). The Argonaut project is looking at ways to increase the efficiency of bulk data transfer by limiting the data elements exchanged to only what is essential for the task at hand, but even at that there is still a significant issue with scale. 
• Lack of governance framework - There is currently a lack of governance structure or trust framework for exchanging data in a way that is both interoperable and secure. Having a framework where data can be exchanged in a standardized way, with requirements on security, trust, visibility, and availability, as well as requirements on responsiveness and data format based on the use case would greatly reduce the burden both on submitters and the systems managing the measure evaluation process. The TEFCA framework would allow for this, but currently lacks some needed enforcement and governance structure for the quality measurement use case. There is also the matter of cost to access the TEFCA QHIN network either directly with a QHIN as a participant, or via a current participant as a subparticipant. There is an opportunity for CMS to lead the way in ensuring that TEFCA is seen as the shortest and best path to making the quality measurement ecosystem interoperable and standardized.

VII. What is Bellese doing?

Bellese has partnered with our customers at CMS many of the FHIR initiatives that seek to improve the quality measurement workflow and standardize many of the processes that are still in heavy use today. We have been active participants at HL7 and CMS FHIR Connectathons in the Clinical Reasoning/Clinical Quality Improvement tracks as we aim to help further the standard and aid in testing of the relevant Implementation Guides used in quality measurement.

As a CMS contracting Application Development Organization, Bellese has overseen the Hospital Quality Reporting program for the past 8 years, and have provided advice and performed research into how the transition to FHIR can be achieved for quality measurement - not only for HQR, but for all of the CMS quality programs and beyond - for Alternative Payment Models at the Innovation Center and other quality-based initiatives. We have developed a FHIR sandbox that we use to test measure evaluation, and through our research we have learned where the improvements need to be made in order to foster an environment wherein FHIR-based eCQMs and dQMs can become a reality, rather than some nebulous goal.

Bellese has recently joined the FHIR At Scale Taskforce (FAST) - a FHIR accelerator program originally created by ONC and now overseen by HL7, seeking to help accomplish FAST’s goals with regards to enabling FHIR data exchange at scale in a secure and interoperable way, including digital identity, computable consent, and the creation of a National Directory of Healthcare providers and services. We have been participants in the DQIC, to promote valuable, consistent measure evaluations across a variety of commercial and non-commercial CQL measure engines. We continue to evaluate TEFCA as a possible channel to enable large-scale data exchange in a well-defined governance structure. Lastly, we welcome the opportunity to provide feedback to CMS on their recent proposed rules for quality measurement - we were excited to see that CMS are still very much interested in leveraging FHIR for quality measurement use cases, and Bellese intends to help them accomplish their goals in that arena.