Thyroid reference range: enrolling more subjects does not help

564 healthy adults across nine sites. CLSI EP28 governs the demographic mix of the reference population, so the binding constraint is who a site can reach — not how many.

Research & Development Institute, Inc. (RDI Trials) RDI Trials, 16760 Stagg Street, Unit 209, Van Nuys, CA 91406. CLIA + CAP-accredited reference laboratory. Engagement: Jan 2022 – Jun 2023 (closed) · Compiled: 2026 · Correspondence: info@rditrials.com

564 subjects

Across 9 sites
in 18 months

+155 subjects

March 2022:
peak month

446 / 564

Front-loaded:
delivered by Jun 2022

EP28 A3

CLSI guideline driving
demographic mix

Abstract

Background: The sponsor — a top-tier global diagnostics manufacturer¹ — commissioned a thyroid reference-interval establishment program to support analytical performance characterisation of a thyroid immunoassay panel. Reference-interval studies are operationally distinct from method-comparison studies and from open-ended specimen-banking programs: the CLSI EP28 guideline imposes specific demographic distribution requirements (sex, age strata, race/ethnicity, geographic distribution) on the reference population, which drives site selection toward sites that can recruit a representative healthy-adult sample rather than toward sites with the highest general enrollment volume.²
Methods: RDI executed the program as full-service CRO. Nine sites were activated against a site-selection grid that weighted demographic-mix delivery as heavily as raw enrollment throughput. The protocol structured visit flow as a single-visit healthy-adult draw against documented exclusion criteria (no current thyroid disease, no thyroid-affecting medications, no relevant pregnancy or lactation status, no relevant intercurrent illness). Sites delivered against demographic targets per the protocol’s EP28-aligned eligibility envelope rather than against a flat enrollment count; demographic-mix monitoring at the program level redirected enrollment slots across sites as the program progressed.
Results: Cumulative enrollment crossed 60 by end-January 2022, 246 by end-March 2022 (the program’s peak month at +155), 459 by end-June 2022, and reached 564 at program close in June 2023 (Fig. 1). The enrollment curve was strongly front-loaded: 459 of 564 subjects (81%) delivered in the first 6 months; the remaining 105 subjects delivered across the following 12 months as the program transitioned from open-enrollment to demographic-stratum top-up against the EP28-aligned cohort targets. Site distribution was relatively even across the four lead sites (Sites A–D each in the 85–101 subject range), reflecting the program’s deliberate distribution of enrollment slots rather than concentration at a single high-throughput site (Table 1; Fig. 2).
Conclusions: Reference-interval studies are operationally a demographic-distribution problem rather than a raw-volume problem. The site-selection grid weights demographic-mix delivery as heavily as enrollment throughput; the long-tail months at the end of the program reflect demographic-stratum top-up against EP28-aligned targets rather than any failure to maintain enrollment cadence. The same operational frame transfers to other reference-interval establishment programs — reference ranges for any analyte where CLSI EP28 or comparable guidelines apply.

Keywords: in vitro diagnostics; reference interval; reference range; CLSI EP28; thyroid function tests; healthy adults; demographic distribution; analytical validation.

1. Background

The sponsor commissioned a thyroid reference-interval establishment program to support analytical performance characterisation of a thyroid immunoassay panel. Reference-interval studies establish the central 95% range of analyte concentrations expected in a defined healthy-reference population; the resulting reference interval is reported alongside test results to guide clinical interpretation. CLSI EP28-A3c is the governing guideline for reference-interval studies on quantitative laboratory analytes in the U.S.²

Reference-interval studies are operationally distinct from the method-comparison and specimen-banking programs reported elsewhere in this engagement series. The binding constraint is not enrollment throughput but delivery against the demographic distribution requirements of the reference population: sex strata, age strata, race/ethnicity strata, and geographic distribution. A site that delivers high enrollment volume but is demographically narrow contributes less than a moderate-volume site whose demographic mix matches the protocol’s reference-population target.

2. Constraints encountered

Three constraints shaped the execution plan:

EP28-aligned demographic distribution as the binding constraint

The protocol specified target enrollment ranges across the EP28-aligned demographic strata. Hitting the program target on raw subject count is straightforward at the lead sites; hitting the demographic distribution targets requires active enrollment-slot management at the program level — closing enrollment to over-represented strata at one site while keeping it open to under-represented strata at another. The site-selection grid weighted demographic-mix delivery as heavily as raw enrollment throughput.
Healthy-adult eligibility verification at screening

The protocol’s exclusion criteria (no current thyroid disease, no thyroid-affecting medications, no relevant pregnancy or lactation status, no relevant intercurrent illness) had to be verified at screening. Sites with on-staff medical assessment capability could verify eligibility in the screening visit; sites without on-staff capability scheduled an additional visit for medical-history confirmation, which extended the consent-to-draw window and increased operational losses at the screening-to-draw transition.
Sample integrity for thyroid analytes

Some thyroid analytes have known pre-analytical sensitivities (sample handling temperature, time-to-spin, hemolysis interference). Sample-handling SOPs were tightened across the program network and validated against the sponsor’s pre-analytical specifications before site activation. This work front-loaded program timeline at site activation but had no ongoing per-month overhead once sites were running.

81% of program volume delivered in the first 6 months. The remaining 19% delivered across the following 12 months as the program transitioned from open-enrollment to demographic-stratum top-up against EP28-aligned cohort targets.

3. Methods

RDI executed the program as full-service CRO. Three operational moves are reported here in order of implementation.

3.1. Demographic-mix-driven site selection

Nine sites were activated across the program window, selected against a site-selection grid that weighted demographic-mix delivery as heavily as raw enrollment throughput. The four lead sites (A–D) were chosen so that their combined demographic mix approximated the EP28-aligned reference-population target; the five supplementary sites (E–I) were activated to fill specific demographic gaps the lead sites could not deliver against. Site selection was not "highest enrollment volume wins" — it was "highest enrollment-volume-against-the-target-demographic-mix wins."

3.2. Single-visit healthy-adult draw with on-screening eligibility verification

The protocol structured visit flow as a single-visit healthy-adult draw against documented exclusion criteria. Sites with on-staff medical assessment capability completed consent, eligibility verification, and draw in a single visit; sites without on-staff capability scheduled an additional visit for medical-history confirmation. Visit-flow design at the lead sites prioritised single-visit completion to compress the consent-to-draw window and minimise operational losses.

3.3. Demographic-stratum monitoring with mid-program enrollment-slot redirection

Demographic-stratum monitoring at the program level ran continuously through the enrollment window. As the program approached the protocol’s stratum-level targets in the high-volume strata (e.g., adult women in the 20–40 age band), enrollment slots were redirected toward under-represented strata at sites with stronger demographic reach in those strata. The 12-month long-tail at the end of the program is the visible signature of this stratum-level top-up — sites running open against under-filled strata while strata that had hit their target were closed.

Raw-volume site-selection (alternative frame)

"Pick the highest-enrollment-volume sites, run them at full capacity until target is hit." Delivers raw subject count quickly; produces a demographically skewed reference sample. Not EP28-compliant.

Demographic-mix site-selection (this program)

"Pick sites whose combined demographic mix approximates the reference-population target; redirect enrollment slots across sites as strata fill." Delivers EP28-aligned demographic distribution. Slower long-tail by design as under-filled strata are filled.

4. Results

Cumulative enrollment crossed 60 by end-January 2022 (program open), 91 by end-February, 246 by end-March 2022 (the program’s peak month at +155), 343 by end-April, 406 by end-May, and 459 by end-June 2022 — 81% of program volume delivered in the first 6 months. The program then transitioned from open-enrollment to demographic-stratum top-up: the next 12 months delivered 105 subjects at an average cadence of approximately 9 per month, with a closeout in June 2023 at 564 (Fig. 1).

Site distribution was relatively even across the four lead sites: Site A 101 subjects (18%), Site B 88 (16%), Site C 87 (15%), Site D 85 (15%); the four lead sites combined: 361 of 564 (64%). The five-site supplementary network contributed 203 subjects (36%) with stronger demographic-mix coverage of strata under-represented at the lead sites (Table 1; Fig. 2). The lead-site share is structurally lower than typical for the engagement series — a deliberate consequence of demographic-mix-driven site selection rather than maximum-throughput site selection.

Jan 2022 Program open

Mar 2022 Peak month (+155)

Jun 2022 Cum 459 (81%)

Jul 2022 Stratum top-up phase begins

2023 Top-up at ~9/month

Jun 2023 Program close (564)

Table 1 Site-level enrollment across the 18-month program window. Lead-site distribution is structurally even (Sites A–D each in the 85–101 subject range) reflecting demographic-mix-driven site selection rather than maximum-throughput selection.
Site	Setting	Enrolled	Share
Site A	Multi-physician clinical-research-organization with broad demographic reach	101	18%
Site B	OB/internal-medicine practice with adult clinical-research apparatus	88	16%
Site C	OB/internal-medicine practice with adult clinical-research apparatus	87	15%
Site D	OB/internal-medicine practice with adult clinical-research apparatus	85	15%
Lead-site subtotal	4 sites with combined demographic-mix approximating the reference target	361	64%
Site E	Coastal multi-specialty research clinic; geographic-mix contribution	70	12%
Site F	Multi-specialty clinical research center	53	9%
Site G	Standalone clinical-research organization	47	8%
Site H	Standalone clinical-research organization with under-represented-stratum reach	30	5%
Site I	Multi-specialty research clinic; small-volume top-up site	3	<1%
Supplementary network subtotal	5 sites filling demographic gaps from the lead-site mix	203	36%
All sites at program close (Jun 2023)	9 sites total across the 18-month window	564	100%

Fig. 1 Cumulative enrollment (orange line) and monthly enrollment (gray bars) across the 18-month program window. The enrollment curve is strongly front-loaded: 459 of 564 subjects (81%) delivered in the first 6 months at an average cadence of ~76/month; the remaining 105 subjects delivered across the following 12 months at ~9/month as the program transitioned from open enrollment to demographic-stratum top-up against EP28-aligned cohort targets. Source: Diagnostica™ at program close, June 2023.

Fig. 2 Site contribution across the 18-month program window. Lead-site distribution is structurally even (Sites A–D each in the 85–101 subject range) reflecting demographic-mix-driven site selection rather than maximum-throughput selection. The five-site supplementary network (E–I) contributed under-represented demographic strata against the EP28-aligned reference-population target. Source: Diagnostica™ at program close, June 2023.

5. Discussion

Three observations bear noting. First, reference-interval studies are operationally a demographic-distribution problem rather than a raw-volume problem. CLSI EP28 imposes specific demographic distribution requirements on the reference population; site selection and enrollment-slot management at the program level have to deliver that distribution, not just the raw subject count. The "highest-enrollment-volume sites running at full capacity" approach that works in many specimen-banking programs would produce a non-EP28-compliant demographic skew here.

Second, the front-loaded enrollment curve followed by a 12-month stratum-top-up tail is the visible signature of demographic-stratum management. The first 6 months ran open enrollment across the full demographic envelope; the following 12 months ran selective enrollment as individual strata approached their targets and were closed. The 12-month tail is the program working as designed, not the program failing to maintain cadence.

Third, the structurally even lead-site distribution (Sites A–D each in the 85–101 subject range) reflects deliberate enrollment-slot allocation rather than competitive enrollment-volume dynamics. Lead sites in this engagement series typically show a 35–45% top-site concentration; the 18% Site A concentration here is a design choice driven by demographic-mix considerations.

6. Conclusion

Reference-interval establishment programs are operationally a demographic-distribution problem against CLSI EP28 or comparable guidelines. A nine-site network with demographic-mix-driven site selection delivered 564 healthy adult subjects across an 18-month window, with 81% of program volume in the first 6 months and a 12-month stratum-top-up tail to fill EP28-aligned cohort targets. The same operational frame — demographic-mix-driven site selection, enrollment-slot management at the program level, structurally even lead-site distribution — transfers to other reference-interval establishment programs.

Notes & References

Sponsor identity withheld per confidentiality terms. Specific protocol code, target counts, analyte panel composition, and assay platform names omitted. Sponsor descriptor reflects the sponsor’s category.
CLSI EP28-A3c, "Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory; Approved Guideline." The EP28 guideline is the standard governing reference-interval studies on quantitative laboratory analytes in U.S. clinical-laboratory practice.
Demographic-stratum monitoring at the program level is implemented through the RDI Salesforce data layer with continuous reporting of per-stratum enrollment against per-stratum targets; mid-program enrollment-slot redirection is operationalised through site-level enrollment-status updates.
Endpoint counts (564 subjects across 9 sites at program close June 2023; +155 March 2022 peak; 459 subjects in first 6 months) are taken from the Diagnostica™.
Site labels in Table 1 and Fig. 2 are anonymised. Site descriptors reflect the site’s clinical setting category.
No human subjects data, identifiers, or laboratory results are reported. This is an operations report, not a clinical-results publication.

Suggested citation. RDI Trials. Thyroid reference range: enrolling more subjects does not help. RDI Trials & Operations Reports, 2026;13. RDItrials.com.