A patient's lab result means almost nothing on its own. Its power comes from the range it's compared against—the reference interval that tells a clinician whether a number is reassuring or alarming. Yet that range is also one of the most portable-looking and least portable things in the modern laboratory: manufacturers publish it, labs adopt it, and assays move between analyzers, sites, and continents, and at every one of those handoffs the interval that made a result meaningful can quietly stop being appropriate. That gap between how transferable reference intervals appear and how transferable they actually are was the throughline of a session by Jehan Abukharmah, DHSc, director of laboratory quality assurance & clinical operations at Avantic Medical Lab, at the 2026 Analytical Challenges in Biotech Digital Summit. Abukharmah, who holds a Doctor of Health Science from George Washington University and a Master of Science in Biotechnology from Johns Hopkins, has spent her career translating scientific innovation into clinical practice—and her message to lab leaders was blunt: a reference interval is a claim about a specific population measured a specific way, and it cannot be assumed to survive the trip to your bench.
Jehan Abukharmah, DHSc, director of laboratory quality assurance & clinical operations at Avantic Medical Lab
The definition is deceptively simple. In the clinical laboratory, a reference interval represents the central 95 percent of values from a currently healthy reference population, running from a lower to an upper limit—roughly the 2.5th to the 97.5th percentile. What matters for lab leaders is what it is not. As Abukharmah emphasized, the reference interval is distinct from clinical decision points, from diagnostic cut-offs, and from critical alert values. Conflating them is where interpretation starts to drift.
Crucially, the interval is inseparable from the population and method behind it. Age, sex, ethnicity, geography, environmental exposure, and comorbidities all shape the distribution of a healthy population's results. A manufacturer establishes its published range under a specific population and a controlled environment. The moment your lab serves a different demographic—or runs a different analyzer, reagent, or calibration—that range is a hypothesis, not a fact.
The factors that shift a reference interval's distribution, which a borrowed range may not account for, include:
Age: Physiology changes across the lifespan, and pediatric populations shift rapidly enough to require their own intervals.
Sex: Many analytes carry distinct male and female ranges.
Ethnicity and ancestry: Population genetics can move the central 95 percent of "normal."
Geography and environment: Altitude, climate, and local exposures influence baseline values.
Comorbidities: Chronic conditions such as diabetes can alter what "healthy" looks like for a given marker.
Method and platform: A different analyzer, reagent, or calibration can shift results independently of the patient.
That's the reasoning behind the standards. Under CLSI guidance, a lab adopting an FDA-approved assay should verify the manufacturer's interval using a minimum of about 20 healthy individuals matched to the population it serves. A lab establishing an interval from scratch needs a far larger cohort—on the order of 120 healthy individuals. "Verification confirms local applicability," Abukharmah explained, and establishment "will create new intervals." The tenfold jump in sample size is the price of building something new rather than confirming something borrowed.
Where transfer goes wrong
The risk concentrates at moments of change that many labs treat as routine: a new analyzer, a new reagent, a new platform, a new site, or an acquisition or merger that brings unfamiliar instruments into the fold. Each of these can shift the result distribution—and when it does, the inherited interval may no longer fit. Assay transfer, in other words, is never just a mechanical move; every method transfer carries the possibility that "normal" itself has moved with it.
The failure mode is subtle and dangerous. As Abukharmah described it, skipping verification after a method change can leave a lab reporting results "where normal becomes abnormal, or the opposite." A shifted distribution paired with an unchanged range doesn't announce itself; it just steadily miscategorizes patients, and, in her words, that erosion means "providers will not just trust the lab." Confidence, once lost, is expensive to rebuild.
Abukharmah illustrated the stakes with two scenarios lab leaders will recognize. In the first, a pharmaceutical company develops a novel inflammation biomarker, runs its phase III trials in North America and Western Europe, and then launches commercially into South Asia, the Middle East, Africa, and Latin America. The unavoidable question: can one global reference range legitimately be used across all of them? In the second, a high-sensitivity cardiac biomarker validated at a central reference laboratory is deployed out to regional hospitals running multiple analyzer platforms and serving different patient demographics. Platform comparability, population differences, laboratory information system (LIS) configuration, and staff training all become live risks—and they multiply with every additional site.
Advanced Lab Management Certificate
The Advanced Lab Management certificate is more than training—it’s a professional advantage.
Gain critical skills and IACET-approved CEUs that make a measurable difference.
The last mile: where correct science still fails
One of the more sobering points in the session was that a lab can do the analytical work correctly and still get it wrong at the finish line. Even after sound validation and verification, Abukharmah noted, common errors surface at the LIS stage—incorrect units carried over from a platform change, inaccurate reference limits, middleware that transfers values incorrectly, and auto-verification logic that quietly applies the wrong rule.
Common last-mile failures that undo correct analytical work include:
Incorrect units carried over when a platform changes but the unit field isn't updated.
Inaccurate reference limits transcribed from the instructions for use.
Middleware transfer errors that move values incorrectly between the analyzer and the LIS.
Faulty auto-verification logic that applies the wrong rule or formula to results.
Abukharmah’s example is worth internalizing. A lab moves to a new platform, updates the reference interval from the new instructions for use—but forgets to update the corresponding units. The science was right; the transcription wasn't; the patient still gets a wrong flag. That's why she stressed validating and verifying "every step" before results reach the provider and the patient, not just the analytical portion. The risk is highest for high-risk markers—troponin, PSA, CA-125, BNP, reproductive hormones, and novel oncology markers—where an abnormal result triggers immediate diagnostic or treatment decisions.
Interested in lab leadership?
Register for a FREE Lab Manager account to subscribe to our Lab Leadership Digest Newsletter.
Much of this guidance runs into a practical wall: recruiting matched healthy cohorts, especially across sex, age, and ethnicity, is challenging—and for pediatric populations, where physiology changes rapidly and sample sizes are small, it can be nearly impossible. Here Abukharmah pointed to indirect approaches as a legitimate, regulation-acceptable path. Rather than recruiting a fresh cohort, a lab can mine the data already sitting in its LIS and apply statistical methods—such as the refineR or Hoffmann approaches—to derive or verify an interval. The practical sequence she described: adopt the manufacturer's interval initially, then, once sufficient real-world data has accumulated, use it to verify or refine the range for the population you actually serve.
This is also where AI enters—as an accelerant, not a replacement. Statistical verification that once took days, she suggested, could be compressed into hours. But her caution was firm and echoes a theme running through clinical quality work: "AI is made by humans," she said, and even a system that's correct 99 percent of the time leaves a margin that demands "another set of eyes" from someone who understands the regulatory requirements before anything is deployed.
A risk-based framework leaders can act on
The organizing idea for leaders is to stop treating reference-interval transfer as a checkbox, but rather as a risk assessment across three axes: population risk, method risk, and platform risk. Different populations, different methods, and different platforms each move the distribution, which is precisely why a patient's result from one lab can differ from another's a few months later—not because either is wrong, but because the measurement systems and reference populations differ.
Abukharmah's recommendations track the lifecycle. Plan the reference-interval strategy early, during assay development. Assess transferability during validation. Perform verification during implementation—and then, critically, don't treat it as "find and done." As the population that a lab serves shifts over time, intervals should be re-evaluated periodically, a practice Abukharmah framed as "dynamic" reference intervals. Looking ahead, she pointed toward precision medicine approaches that weigh a patient's full history and genetics rather than a single range, with AI expediting the statistical legwork underneath.
For lab leaders, a practical reference-interval checklist across the assay lifecycle looks like this:
During development: Plan the reference-interval strategy early rather than treating it as an afterthought.
During validation: Assess how transferable the interval is across the platforms and populations it will reach.
During implementation: Perform verification—a minimum of about 20 matched healthy samples under CLSI guidance—before reporting.
Ongoing: Re-evaluate intervals periodically as the served population changes; treat them as dynamic, not fixed.
Across all stages: Assess population risk, method risk, and platform risk, and keep a knowledgeable human reviewing any AI-assisted output.
Abukharmah's final takeaway during the presentation was that transferability cannot be assumed, verification builds the trust that makes a result usable, and lifecycle thinking—not one-time sign-off—is what makes clinical implementation succeed. A reference interval, in the end, is only as good as the population and method standing behind it. Move either one, and the burden is on the lab to prove the range still holds.
Add Lab Manager as a preferred Google source to see more of our trusted coverage.
Frequently Asked Questions (FAQs)
What is a reference interval?
A reference interval is the central 95 percent of values from a currently healthy reference population, typically spanning the 2.5th to 97.5th percentile. It is distinct from clinical decision points, diagnostic cut-offs, and critical alert values, and it is tied to the specific population and method used to establish it.
Why can't a manufacturer just provide one universal reference interval?
Because patient demographics—age, sex, ethnicity, and health status—affect analyte concentrations, and manufacturers establish their intervals under a specific population and controlled environment. Labs serving different populations, or running different methods, need to verify that the interval applies locally.
What's the difference between verifying and establishing a reference interval?
Verification confirms that a manufacturer's published interval applies to your population and generally requires a minimum of about 20 matched healthy individuals under CLSI guidance. Establishing a new interval from scratch requires far more—on the order of 120 healthy individuals—because you are creating the range rather than confirming it.
When does a reference interval need to be re-verified?
Whenever something changes the measurement: a new analyzer, reagent, platform, or laboratory site, or an acquisition that brings new instruments. Method changes can shift the result distribution, making the inherited interval inappropriate until it's re-verified—an expectation reflected in CLIA regulations and accreditation standards.
Sarah Bauder is the senior editor at Lab Manager. She possesses a diverse background spanning editorial, digital marketing and film and television production. She brings over 15 years of experience in editorial writing, B2C and B2B content creation. A student of history, she graduated from York University in Toronto, Ontario, Canada. She can be reached at sbauder@labmanger.com.
