A patient of mine had an actionable mutation sitting in his chart for months before anyone acted on it. The result was there, it just wasn't anywhere a computer could read it. It was a PDF, scanned into the record, indistinguishable to the EHR from a release-of-information form.
That is the ordinary state of cancer genomics in 2026, and it is the problem I want to take apart. We have spent a decade getting better at generating molecular results: broad panels, liquid biopsies, whole-exome and whole-genome sequencing, faster turnaround. We have spent almost none of that decade getting those results into a form the patient's record can actually use. The sequencing got cheaper and richer; the last mile, from the reference lab's report to a field in the chart that can drive a decision, barely moved.
A PDF is not data
Picture how a genomic result usually arrives. A tumor block goes out to a reference laboratory. Weeks later a report comes back, a richly formatted document, often a dozen pages, listing variants, allele frequencies, therapy implications, and trial suggestions. It is faxed or transmitted, and someone scans or attaches it to the EHR. From that moment it is visible and useless in equal measure. A human can open it and read it. Nothing else can.
The distinction that matters here is between a document and a datum. A PDF is a picture of information. It cannot be queried. It cannot trigger a clinical decision support rule when a new targeted therapy is approved for the variant it describes. It cannot be matched against the eligibility logic of an open trial. It cannot be counted in a cohort, fed to a prediction model, or submitted to a quality program. The EGFR exon 19 deletion on page seven might as well not exist, as far as the rest of the system is concerned, and the rest of the system is where care increasingly gets coordinated.
This is not a hypothetical gap. When a multi-institution group I worked with catalogued the real-world barriers to using biomarker results, the failure modes were depressingly concrete: results arriving as unstructured documents, no consistent place in the record for them to land, no way to surface them at the moment of a treatment decision. We wrote it up as Integrating electronic health records to facilitate cancer biomarker testing (Huelsman et al., Journal of Clinical Oncology, 2024). The headline finding is mundane and damning: the testing is rarely the bottleneck. The integration is.
What "structured" actually buys you
When a genomic result enters the record as structured data, discrete, coded variants in defined fields rather than ink on a scanned page, three things become possible that were impossible before, and all three are the difference between sequencing as a line item and sequencing as care.
- The result can warn you. Decision support can watch the patient's structured variants and fire when the evidence changes, a new approval, a label expansion, a newly opened trial for exactly that alteration. The oncologist learns about the option in time to use it, instead of months later or never.
- The result can match. Trial-matching logic and molecular tumor board prep run against a queryable variant, not a human's memory of a PDF they read last spring. The patients who qualify for something get found.
- The result can be counted. Cohorts, outcomes research, real-world evidence, and quality submissions all need the genomics as data. Structured capture is what makes a patient's molecular profile legible to the institution, not just to the one clinician who happened to open the attachment.
None of this requires a new model or a new algorithm. It requires the unglamorous plumbing that turns a report into a record. That plumbing is the work I have spent years on, and it is harder and more valuable than it sounds.
Being eighth, and then being first
Vanderbilt Health started early. In 2019, when exactly one institution in the country had integrated genomic results from a reference laboratory directly into Epic, we became the eighth. That sounds like a modest distinction until you sit with the denominator: in a country with thousands of hospitals and hundreds of cancer programs, fewer than ten had managed to get reference-lab genomics flowing as structured data into the dominant EHR. The hard part was never wanting it. The hard part was the integration.
Going forward was only half the problem. The other half was the back catalog, thousands of patients already on therapy whose molecular reports were sitting in the record as PDFs and scanned attachments, invisible to every downstream system. A cut-over that only structures new results leaves the existing population stranded. So we built an extract-transform-load pipeline that converts that historical EHR genomic data into standardized, queryable profiles built on HL7 FHIR, the same interoperable shape the standards work targets. The result is a unified structured corpus across a patient's entire history, not a clean record from a cut-over date forward and a fog behind it. The upstream workflow is documented in the NCCN abstract Integrating Structured Genomic Data Into the Electronic Health Record (Vento & Osterman, JNCCN, 2023, BIO23-019).
The compounding effect of being early is the part worth internalizing. By the end of 2021 there were roughly twelve thousand tumor genomic reports living as structured data inside our EHR. Today Vanderbilt Health holds more structured genomic data in its electronic health record than any other institution in the United States. That position is not the product of a single clever build; it is the product of starting years before most peers and never stopping. The institutions that begin this work now are not behind by a feature. They are behind by a corpus, and a corpus is the one thing you cannot buy or backfill instantly. You accrue it.
The unlock, not the trophy
The #1 ranking is not the point. The point is what the structured corpus makes possible downstream, and almost everything I care about in precision oncology sits on top of it. Trial matching at scale, molecular-tumor-board governance, the clinical genomics workstream that ties it together, point-of-care alerting when a patient's variant becomes actionable, real-world toxicity studies, industry research collaborations that need an interoperable surface to reach the data, none of them work against a pile of PDFs. They all work against structured fields. The integration is the load-bearing layer; the celebrated applications are what you stack on top once it holds.
This is also why I keep returning to standards. Structured data inside one institution's EHR is necessary but not sufficient, a research collaborator across town still can't reach it without an interoperable shape to read. That is the role of mCODE, the FHIR-based minimal Common Oncology Data Elements, which carries the genomics among the core fields a cancer record should expose. Integration makes the data usable inside the walls; the standard makes it portable beyond them. You want both, in that order: structure first, then portability.
What to do about it
If you run a cancer program, a lab, or the informatics behind one, the actionable conclusion is narrow and stubborn. Stop measuring genomics by tests ordered and start measuring it by structured results landing in the chart where something can act on them. Treat reference-lab integration as a first-class capital project, not an IT ticket. Insist that your labs and your EHR vendor deliver coded, discrete results, not prettier PDFs. And build the backload, not just the cut-over, because the patients already on treatment are the ones a stale record fails first.
The sequencing is the easy half now. The molecular biology that took a generation to make routine is, clinically, the solved part. The unsolved part is the most ordinary-sounding question in medicine: can the chart actually use what the lab just told you? Until the answer is yes, until the variant on page seven is a field the system can read, every dollar spent on the test is spent generating a document no one downstream can act on. The work that turns that document into care is the work worth funding. For the longer arc of how we built it at Vanderbilt Health, see clinical genomics in the EHR.