UPMC report here.
See subscription coverage at Genomeweb here. (Alexandra Byrne.)
###
AI CORNER (Chat GPT)
##
UPMC Precision Medicine Report:
When “Access” Means More than Just the Test
Precision medicine has spent two decades being described as the future of care. The new UPMC/University of Pittsburgh report offers a more practical update: in many health systems, the future has arrived, but it is still stuck in the plumbing.
The issue is no longer just whether genetic tests exist, or whether they can be sent to a reference lab. The harder question is whether a health system can order the right test, manage payment, return the result, place it usefully in the EHR, interpret it, and act on it.
That is the central message of Operationalizing Precision Medicine: How U.S. Health Systems Are Implementing Genetic Testing, a 22-page report from UPMC’s Center for Connected Medicine, the University of Pittsburgh Institute for Precision Medicine, and KLAS Research. The report is based on interviews with 21 U.S. health system leaders conducted between October 2025 and February 2026. A GenomeWeb/Precision Medicine Online article by Alexandra Byrne highlighted the same theme: programs are maturing, but reimbursement remains a persistent drag.
The headline: implementation, not invention
The report frames precision medicine as having moved from a science problem to a health-system problem. Genomics has advanced. The next barrier is scale.
UPMC puts it directly: scaling genetic testing into routine care requires more than “access to tests.” Mature programs need governance, workflows, interpretation support, clinical-system integration, and measures that matter to clinicians, patients, and payers.
That is a useful correction. A hospital does not have a precision medicine program just because a physician can order a commercial panel from a national lab. But it also does not need to perform every assay locally. The mature model is not “local lab or no access.” It is testing plus infrastructure.
Does UPMC mean “genetics” or “genomics”?
UPMC uses “genetic testing” broadly. It includes traditional genetics, but clearly extends into what many readers would call clinical genomics.
The report lists pharmacogenomics, oncology/immunotherapy, behavioral health, maternal/fetal health, cardiology, neurology, rare disease, preventive/population health, pediatrics, and ophthalmology as areas where organizations apply genetic testing. The “other” category includes CDC Tier 1 genomic screening, hereditary cancer risk assessment, NICU/inpatient rapid genome sequencing, transplant-related uses, prenatal genetics, and primary care genomics.
Oncology is not a minor add-on. The report ties oncology testing to targeted therapy selection, immunotherapy decisions, and clinical trial matching. UPMC’s own examples include sequencing cancers from both blood and tissue to monitor disease and direct therapy.
For diagnostics and reimbursement readers, that matters. The report does not provide a coding-level taxonomy of CGP, MRD, liquid biopsy, germline testing, pharmacogenomics, and polygenic risk. It is not a MolDx or CPT document. But in ordinary health-system language, its “genetic testing” includes much of what the industry calls genomics.
Pharmacogenomics has moved to the front
One notable change from the 2020 survey is the rise of pharmacogenomics. In 2020, oncology was the dominant use case. In the new survey, pharmacogenomics is first, with 19 of 21 organizations reporting use, followed closely by oncology/immunotherapy.
That makes operational sense. Pharmacogenomics fits into medication workflows. It can be supported by pharmacy, EHR alerts, drug-gene guidance, and prescribing decision support. It also lends itself to the “once tested, repeatedly useful” model health systems like for genomics.
Oncology genomics remains deeply adopted, but it is messier. Tumor type, specimen adequacy, payer policy, FDA labeling, companion diagnostics, clinical trial matching, tumor boards, MRD timing, and reference-lab logistics all intersect. UPMC groups oncology within the broader precision medicine family, but does not unpack those coding and reimbursement complications.
What does “access” mean?
The report’s most useful contribution may be its implicit definition of access.
Access is not merely the physical availability of a test. Access means the patient reaches a system that knows when to test, can manage coverage and cost, can order correctly, can receive results promptly, can integrate them into the EHR, and can act on them.
A reference lab may offer a test nationally. But if the local oncologist does not order it, the payer denies it, the result returns as a buried PDF, or no one knows what action should follow, then “access” is mostly theoretical.
At the same time, UPMC does not define access as local testing only. The report discusses vendor partnerships, centralized teams, external partners, and reliable data exchange when organizations use multiple labs. It also notes that some institutions cannot fund in-house testing and may follow larger regional peers rather than build duplicative programs.
So the practical answer is: reference lab testing can count as access, but only if the result works inside the health system. If it is ordered, paid for, returned, structured, interpreted, and acted on, it is real access. If it is just a test menu item somewhere in the cloud, it is not.
Reimbursement remains the wall
The familiar villain is still reimbursement. In the new survey, 11 respondents named cost or lack of insurance coverage as a main barrier to genetic testing. Provider education and lack of clear guidelines followed. In the 2020 survey, reimbursement/ROI was also the top barrier.
The GenomeWeb article quotes Adrian Lee of UPMC’s Institute for Precision Medicine saying that payers want evidence of clinical utility, and that personalized care is “not cheap.” It also notes that programs are funded through a mixture of fee-for-service or bundled payments, philanthropy, pharma funding, grants, and internal investment budgets.
That mix tells its own story. Precision medicine is routine enough to need operations, but not routine enough to have stable economics everywhere. Many programs are still patched together with clinical revenue, institutional investment, philanthropy, and research support.
The EHR is necessary, but not sufficient
UPMC gives appropriate weight to the EHR. Respondents describe getting genetics into the EHR as a major operational shift. Epic genomics modules can support electronic ordering, discrete results, and embedded decision support, including drug-gene alerts and hard stops.
But the more important point is that EHR integration alone is not enough. The valuable state is structured, findable, actionable data linked to clinical decision support. A scanned PDF may be legally present in the chart, but operationally half-invisible.
This is especially important for reference labs. If outside labs want to be part of mature health-system precision medicine programs, the competitive edge may not be just assay menu, turnaround time, or price. It may be clean data return, workflow support, payer support, and integration with care pathways.
Equity is framed in practical terms
The report’s equity section is operational rather than rhetorical. Organizations describe equity work as financial assistance, payer navigation, multilingual education and consent, virtual genetics access, targeted outreach, and community partnerships. Six of 21 respondents reported no formal equity strategy.
This again broadens access. In precision medicine, equity is not just whether a test exists. It is whether patients are identified, counseled, protected from surprise bills, and followed after the result.
AI enters, mostly around the edges
The report includes a new section on AI. Most organizations are already using AI in adjacent areas such as clinical decision support, imaging, diagnostics, administrative tasks, predictive analytics, and workflow automation. GenomeWeb quotes Lee saying that digital infrastructure and genomics integration have accelerated, and that the next five years may show whether AI can enable precision medicine.
The realistic reading is that AI is not yet the main event in genomic medicine. It is becoming part of the support structure: interpretation, decision support, imaging, documentation, risk stratification, patient outreach, and workflow automation. That may be less glamorous than “AI discovers cure,” but more relevant to adoption.
Bottom line
The UPMC report is less about a new genetic test than a new maturity level. Precision medicine is no longer a boutique research aspiration. It is becoming a normal health-system capability. But normal capabilities require budgets, governance, workflows, IT, staff, and evidence.
For diagnostics companies, the message is pointed: the test is only part of the product. The rest is getting the result into care.
For health systems, the message is equally pointed: a precision medicine program is not a press release, a reference lab contract, or a few enthusiastic specialists. It is an operating model.
For payers, the report shows that the evidence question is becoming more practical. Health systems are trying to measure what testing changes, what it prevents, and whether it produces value. The next stage of precision medicine may be won not by the most elegant assay, but by the best-integrated one.
Sidebar: Ten Surprising Findings
Pharmacogenomics now leads oncology as the most commonly reported use case.
Nineteen of 21 organizations reported using genetic testing in pharmacogenomics.
Oncology/immunotherapy remains near the top, reported by 18 of 21 organizations.
Behavioral health is also near the top, a reminder that precision medicine is no longer only cancer and rare disease.
Seven organizations had been offering genetic testing for six or more years.
Four organizations started within the past year, showing that new entrants are still arriving.
Cost or lack of insurance coverage remains the leading barrier.
Provider education is almost as important as payment, cited by 10 respondents.
Sixteen of 21 organizations reported a formal precision medicine structure, such as a department, institute, service line, or program office.
Access is becoming operational, not geographic. Reference lab testing can count, but only if ordering, payment, result return, EHR integration, interpretation, and follow-up actually work.
