Welcome back to Out of the FHIR. This isn’t just any episode. This is a reunion. FHIR Data guy brings together two heavy-hitters from his past, Eric Melymuk Principal Engineer at Progyny and John Dobak Senior FHIR Analytics Engineer at b.Well a trio of veterans who all came up in the “bad old days” of HEDIS and traditional quality measurement. This is the story of a technology stack (and a mindset) that defined healthcare data for decades, and the new stack that’s poised to replace it. They dive deep into the messy, complicated, and fascinating transition from the world of custom SQL, claims data, and regulatory checklists to the new frontier of FHIR, CQL, and true clinical interoperability. Hot Takes * The “Mind Shift” Thesis: Perfectly stated by Eric, is that the move from traditional SQL-based reporting to FHIR + CQL (Clinical Quality Language) isn’t just a syntax change. It’s a fundamental mind shift away from “logic customized to my organization” and toward “universal clinical intent.” This is the unlock. * Beware the “Gray Water”: The biggest risk of the new FHIR-based world is creating “dirty FHIR data.” It’s data that is technically compliant (it passes the validator!) but is clinically useless. He calls this “gray water,” and it’s the next great crisis. Organizations treating FHIR as a low-cost, check-the-box “cost center” are just propagating this gray water downstream, poisoning the well for everyone else. * The “Consumer Reports” Blind Spot: For decades, HEDIS and Star Ratings have been the “Consumer Reports for healthcare” (John’s analogy). The problem? We’ve been hiding the magazine from the consumers. The real promise of FHIR isn’t just better back-end reporting; it’s finally putting that quality data into the hands of the patients making the decisions. * LLMs as the Co-pilot, Not the Pilot: ”Do we even need measures in the age of AI?” Yes. But AI and LLMs aren’t the replacement for clinicians or measures. They are the co-pilot (Eric’s term) that helps builders create the logic and test cases, and they may become the translator that finally makes this data understandable to a patient. * The Cultural Battle: This isn’t a technology problem; it’s a culture problem. The entire panel agrees: if your organization views FHIR and interoperability as a regulatory “cost center,” you’ve already lost. The winners will be the ones who find executive champions (Eric’s point) who understand this is a new product and the strategic foundation for everything that comes next. From Brittle SQL to a FHIR Future This conversation is a journey from the past to the future of healthcare data. Act 1: The Old World (The HEDIS Grind) The guys reminisce about the “hardest problem” in health tech: traditional quality reporting. It was a world of brittle, custom-built SQL engines, siloed (and non-technical) quality departments, and a total dependency on claims data not because it was the best data, but because it was the only consistent data. Act 2: The New Stack (The FHIR/CQL Unlock) Gene pivots the conversation to the “new stack.” The difference is night and day. * Old Way: NCQA hands you a 200-page book (Volume 2) and says, “Good luck.” You have to design the schema, the logic, the whole thing. * New Way: NCQA hands you the code. The schema is the FHIR Implementation Guide (IG). The logic is the executable CQL file. The entire low-level, error-prone interpretation is abstracted away. Act 3: The Messy Middle (The “Gray Water” Problem) But this transition is where the real drama is. The panel identifies two massive risks: * The Data Itself: We’re moving from clean claims data to messy, unstructured clinical (EHR) data. The old logic (built on claims “artifacts” like Place of Service codes) simply doesn’t map. * The “Cost Center” Trap: Because of regulatory mandates, organizations are scrambling to create FHIR data. They’re doing it cheap, fast, and without clinical governance. This creates John’s “gray water”FHIR bundles that are valid but meaningless, which will break everything downstream. Act 4: The Real Goal The episode concludes by refocusing on the why. Eric makes the passionate case that we’ve “lost sight” of the goal. It’s not about checking boxes for payment; it’s about improving care. The new stack (FHIR, CQL) is the first real chance to build systems that give data back to providers and patients in real-time, finally fulfilling the original promise of quality measurement. The Playbook: Learnings from the HEDIS OGs * 1. Don’t Replace the Warehouse, Feed It. This is a key strategic insight from John. CQL and FHIR do not replace your SQL data warehouse. They work together. Use the CQL engine to process the messy, real-time FHIR data, and then use the output (the FHIR MeasureReport resources) to feed your traditional SQL warehouse for population-level analytics. It’s the best of both worlds. 2. Your FHIR Implementation Is Your Product. The panel’s biggest warning: do not treat your FHIR implementation as a “cost center” or an IT project. It is a product. It requires governance, clinical input, and executive champions. If you don’t, you’ll spend millions building a “dirty data” engine that provides no value. * 3. Use the EOB as Your Bridge. How do you get from a claims-based world to a clinical-data world? Eric points to the bridge: the ExplanationOfBenefit (EOB) FHIR resource. It’s designed to model adjudicated claims. This allows you to “dip your toe in” by mapping your existing, clean claims data into a FHIR structure first, before you tackle the mess of raw EHR data. 4. Get Your Feet Dirty (The Community is Open). The single best way to learn is to do. The HL7 FHIR community (on Zulip) is famously open. As John notes, you can even feed an entire Implementation Guide into an LLM and start asking it questions. The barrier to entry has never been lower. Get full access to FHIR IQ playbook at evestel.substack.com/subscribe
