Site Selection for Decentralized Trials: A 2026 Scoring Framework

Why Traditional Site Selection Breaks for DCTs

Traditional site selection is a mature discipline. Sponsors and CROs build feasibility questionnaires around a familiar set of attributes: principal investigator experience with the therapeutic area, historical enrollment rates, facility quality, patient population access, coordinator bandwidth, and financial stability. Each attribute is scored on a scale, weighted, and combined into a ranking index used to shortlist sites for site qualification visits.¹ The process consumed roughly $170 million in global investigator site time in 2024, translating to about 2,500 hours per site annually.² When the trial is a conventional brick-and-mortar Phase III with all activities on-site, this model works reasonably well, even accounting for the persistent problem that roughly 11% of activated sites still fail to enroll a single patient and about 20% of principal investigators under-enroll.³

Decentralized clinical trials change the underlying question. In a DCT or hybrid trial, the “site” is no longer only the physical clinic where the investigator sits. As the EMA’s recommendation paper puts it, introducing decentralized elements should be considered as an extension of the clinical trial site with the inclusion of trial participants’ homes, resulting in additional oversight obligations for investigators and sponsors.⁴ The FDA’s final guidance similarly frames DCTs as involving remote visits, local healthcare providers, digital health technologies, and direct-to-patient supply chains that all operate outside the walls of the traditional research site.⁵

When the site extends into participants’ homes, the variables that predict operational success change. A principal investigator with a strong publication record and a well-run coordinator team is still valuable, but no longer sufficient. The catchment area’s broadband and smartphone availability starts to matter. The state licensure environment for the investigators who will conduct remote visits starts to matter. Whether the site has a working relationship with a home health nurse network that actually covers the counties where participants live starts to matter. Whether the site’s telehealth platform will exchange data with the sponsor’s electronic clinical outcome assessment (eCOA) platform starts to matter.

The gap between a “traditionally excellent” site and a “DCT-ready excellent” site is often invisible on the standard feasibility questionnaire. A high-performing academic medical center in an urban market can look strong on every legacy metric, then falter on a hybrid protocol because its home health partner sub-contracts to a network that has almost no presence outside the metro area, or because state medical board interpretations create friction for cross-border video visits. Meanwhile, a mid-sized community site with a modest historical enrollment record can outperform expectations because its patient catchment has stronger broadband penetration, its investigators hold licenses in three adjacent states, and its lab courier partner already runs a same-day cold chain to residential addresses across the region.

The 2026 DCT Site Scoring Framework

The framework we use is deliberately additive rather than replacement. Traditional site attributes (PI experience, historical enrollment, coordinator bandwidth, quality history) still count. But we layer five DCT-specific dimensions onto the traditional scorecard, each with its own set of measurable sub-criteria and a weight that can be adjusted for the protocol at hand. The dimensions map to the operational realities the FDA and EMA guidance documents actually describe.⁵⁴

Each dimension carries a default weight that reflects how heavily it tends to drive DCT success in a typical hybrid Phase II or III protocol. In our default weighting, traditional attributes retain 40% of the total score, and the five DCT dimensions share the remaining 60%. That balance shifts significantly for fully decentralized designs, oncology, cardiovascular, and rare disease. We show adjusted weightings later in the article.

Dimension One: Patient Technology Access

Patient technology access is the dimension sponsors most often skip because it feels like something the site cannot control. It is, however, the ceiling on every other DCT capability. If participants in the site’s catchment cannot reliably use a video visit, receive a wearable notification, or complete an ePRO diary on a phone, no amount of telehealth platform sophistication will save the protocol. Adults in rural counties are 42% less likely to use telehealth services than urban adults, driven primarily by broadband availability, and nearly half of low-income Americans lack a home computer while roughly a third do not own a smartphone.⁶⁷ Peer-reviewed telehealth research has also documented that patients relying solely on a smartphone (without a home computer or subscription internet) are significantly more likely to no-show for scheduled visits.⁷

Sub-criteria within this dimension include broadband penetration at the ZIP-code level within the catchment, smartphone ownership benchmarks, digital literacy indicators, and the site’s own experience running remote visits with the target population. FCC broadband data, American Community Survey device ownership tables, and site-level ePRO completion rates from prior studies all feed the score.

What the sub-scores actually measure

• Broadband coverage of catchment: percentage of ZIP codes in the site’s catchment with fixed broadband at 100/20 Mbps or better. Score high when 90%+; score low when large rural pockets fall below 50%.
• Smartphone ownership in the target population: adjusted for age and socioeconomic profile of the therapeutic area. Cardiovascular and diabetes populations typically score lower than autoimmune or oncology.
• Site-provisioned device capability: whether the site has run BYOD trials, provisioned devices at study start, and set up device-lending programs. Sites that have already solved this in one trial can reuse the muscle.
• ePRO/eCOA completion history: completion rates from prior trials, ideally in a comparable population. Below 70% completion should reduce the score materially.

Dimension Two: Home Health and Community Care Networks

Home nursing has moved from optional to core in DCT protocols. Studies conducted with home nursing (as documented in the pan-European RADIAL trial and in qualitative research on nurse experience in DCTs) show that home visits meaningfully reduce protocol deviations and support retention, particularly for patients who face travel or mobility barriers.⁸⁹ But home health quality is highly local. A site’s ability to actually put a qualified nurse in a participant’s home for a blood draw, vitals check, or IMP administration depends on which vendors it has vetted, which counties those vendors cover, and how quickly they can schedule visits inside the protocol window.

This dimension scores the site on the operational readiness of its home health and community care ecosystem, not just the theoretical availability of vendors. A site that has a signed master service agreement with a national home health CRO covering 40 states is very different from a site whose vendor of record covers three counties and struggles to schedule inside a 72-hour window.

Sub-criteria for home health readiness

1. Vendor coverage across the site’s catchment: percentage of the catchment population that lives within a vendor’s confirmed coverage area. This should be verified by ZIP code, not assumed from vendor marketing maps.
2. Nurse availability and scheduling reliability: historical performance on scheduling remote visits inside the required window. Sites that have used home nursing in prior trials can produce data; sites that have not must be scored on vendor references.
3. Investigator comfort with third-party providers: the site’s principal investigator must be willing to co-manage participants with external nurses and to formally delegate tasks per ICH E6 and the EMA recommendation paper. Sites where the PI is skeptical of home nursing are a warning sign, not a training opportunity, at least on shorter timelines.⁴
4. Language and cultural coverage: availability of nurses matching the demographic profile the diversity plan is targeting. In some markets this is the binding constraint.
5. Vendor concentration risk: whether a single home health CRO carries the entire site’s capability, which creates a single point of failure. Sites with two or more vetted vendors under active MSAs are more resilient when scheduling breaks down on short notice.
6. Documentation and delegation posture: the site’s ability to produce delegation logs, task-level training records, and quality agreements that would survive an inspection. This looks small during feasibility and becomes central during a monitoring visit.

One useful diagnostic during feasibility is to ask the site to describe its process for a specific, made-up scenario: a participant in ZIP code X needs an unscheduled vitals check within 48 hours because a wearable flagged a threshold event. Which vendor gets the call, who confirms availability, how is the visit documented, and how does the site’s principal investigator sign off on the delegation? Sites that answer fluently have thought about this. Sites that pause, ask for a “process document,” or promise to “loop in operations” are telling you where the operational risk lives.

Dimension Three: Telehealth and Data Interoperability

Digital platforms are the connective tissue of any DCT. The Journal of Medical Internet Research analysis of digital platforms in DCTs highlights that fully integrated, unified DCT platforms are still relatively new, with substantial variation in how well sites can connect their existing telehealth infrastructure, EHR, and eSource systems to sponsor-selected platforms.¹¹ Interoperability is not an abstract concern. If the site’s EHR cannot exchange visit summaries with the sponsor’s telehealth vendor via HL7 FHIR APIs, coordinators end up doing manual reconciliation that quickly consumes the operational savings a DCT design was supposed to deliver.

This dimension scores the site on the maturity of its telehealth practice and the readiness of its data environment to interoperate with the sponsor’s chosen platforms. Sites with recent DCT experience, active FHIR integrations, and mature eSource setups earn high scores. Sites with a telehealth platform used only for standard-of-care video visits and no exposure to eCOA or wearable data feeds earn moderate scores. Sites with no telehealth infrastructure at all must be evaluated against a longer setup runway.

Interoperability sub-criteria

• Telehealth platform in production use: platform name, monthly active clinicians, and demonstrated compatibility with sponsor telehealth vendors. Ideally documented in the feasibility questionnaire with specific version numbers.
• EHR interoperability posture: availability of HL7 FHIR APIs, prior integration examples, and IT team responsiveness to sponsor integration requests.
• Wearable and sensor data experience: prior studies where the site ingested wearable-derived data. Between 2001 and 2025, over 1,000 interventional trials incorporated wearable-derived data into study protocols, so this is no longer exotic; sites that have never done it are a lower score.¹²
• eSource maturity: whether the site conducts eSource or continues to rely on paper source that must be transcribed. Sites still on paper source are not disqualified but must be scored down on any protocol that requires rapid remote data verification.
• Identity and access management: the site’s ability to provision study coordinators and investigators into sponsor platforms quickly, revoke access when a team member departs, and produce an audit trail of who accessed what. This is a small operational point that becomes visible during a Part 11 inspection.
• IT team responsiveness: the average turnaround time for the site’s IT function to answer integration questions during startup. Sites with responsive, in-house clinical IT teams score meaningfully higher than sites whose IT function is entirely managed by a health system with a multi-week ticket queue.

A useful early test during feasibility is to send the site’s IT contact a specific FHIR question and measure how quickly and precisely the answer comes back. Sites that respond within a few business days with a concrete answer (including any known limitations) are almost always faster to integrate than sites that respond with a marketing overview of their platform. This is a proxy for something harder to measure directly: the operational velocity at which the site can absorb sponsor requirements.

Dimension Four: Regulatory and Consent Environment

Regulatory readiness in DCTs is heavily state-specific in the United States and highly variable across EU member states. Investigators and practitioners providing clinical care as part of a DCT must be licensed in the state where the trial participant is located unless an exception applies.¹³ Cross-state licensure compacts (IMLC for physicians, NLC for nurses) reduce but do not eliminate the friction, and cross-state licensing exceptions for telehealth-only interactions are inconsistent by state and by clinician category.

Sites in states that are members of major licensure compacts and whose principal investigators hold multi-state licenses can support broader participant catchments. Sites in restrictive states, or whose investigator teams are single-state, will constrain enrollment to a smaller geographic footprint. Similarly, sites with mature remote informed consent workflows (electronic signature capture, identity verification, comprehension assessments) require less protocol customization than sites still relying on wet-ink consent at a physical visit.

Regulatory environment sub-criteria

1. Multi-state licensure profile: number of states covered by investigator team licenses; compact memberships (IMLC, NLC, PSYPACT).
2. Remote informed consent maturity: production experience with electronic consent under FDA’s guidance, including identity verification and comprehension checks. Local healthcare providers cannot obtain informed consent per FDA, so the site must own this process.⁵
3. Central IRB comfort: the site’s ability and willingness to work under a central IRB, which EMA and FDA both prefer for DCTs. Sites that require local IRB review are slower and more expensive.¹⁴
4. State telehealth regulatory posture: flags for states with restrictive interpretations of telehealth practice or with prescribing limitations that could affect the trial. This changes over time and should be re-checked at the start of every program.

Dimension Five: Specimen and Investigational Product Logistics

The final dimension covers the physical supply chain. In a DCT, cold chain does not stop at the site loading dock; it extends to residential addresses across the catchment, and it operates against tighter time windows because samples have to move between a home visit and a central lab without the buffer of an on-site freezer. Contract Pharma’s analysis of DCT supply chain considerations notes that maintaining the cold chain is arguably the most difficult aspect of DCT logistics.¹⁵ World Courier’s direct-to-patient model and similar programs from other logistics vendors handle drug delivery, specimen pickup, and ancillary supply returns at the participant level, but the site must still coordinate scheduling and handle exception cases.¹⁶

Sub-criteria include: existing relationships with direct-to-patient logistics vendors, courier density in the catchment, cold chain performance history in prior trials, medical waste handling for residential pickups, and whether the site has run a home-based drug administration protocol before. Sites with courier partners that operate a same-day service across the catchment score high; sites that would need to build the supply chain from scratch score low, particularly for temperature-sensitive investigational products.

Logistics sub-criteria

• Direct-to-patient vendor coverage: the geographic footprint of the site’s or sponsor’s DTP logistics vendor at the ZIP-code level. A vendor that “covers the United States” may still have cold chain gaps in specific rural corridors that matter for the trial’s target population.
• Cold chain performance history: temperature excursion rates from prior trials that used residential pickups and drop-offs. Excursion rates in the 3-5% range are typical for well-run programs; anything above 8% is a warning sign.
• Home administration experience: whether the site has previously overseen at-home dosing of investigational product, including any post-dose safety monitoring windows. For biologics and injectables, this is a step change from oral at-home dosing.
• Ancillary supplies and returns: whether the site can efficiently coordinate the return of unused study drug, wearables, and other ancillary equipment. Return logistics get systematically under-planned in first-time DCT programs.
• Medical waste handling: the site’s process for coordinating biohazard pickup at residential addresses. This is often the operational detail that surprises sponsors who are new to DCTs.

The logistics dimension is where hybrid trials most frequently discover the gap between the site’s stated capabilities and its actual operational reach. A site whose feasibility questionnaire confidently claims a home visit and specimen collection capability may, in practice, be relying on a home health vendor whose courier partner does not operate a validated cold chain in the counties where the target patient population lives. Verifying this requires ZIP-code-level coverage confirmation, not vendor marketing collateral.

Applying the Framework: Scorecard, Weighting, and Case Examples

The mechanics of the framework are straightforward. Each site is scored on each sub-criterion using a 1-5 scale, sub-criteria are averaged within their dimension, and dimension scores are weighted according to the trial’s design and therapeutic area. The result is a single index that can be sorted and paired with the traditional attribute score for a combined ranking. What matters is the weighting.

Default and adjusted weightings

These are starting points. A pediatric protocol with an at-home IMP administration will lean harder on home health and logistics. A cardiovascular outcomes trial enrolling ambulatory patients will benefit from stronger patient technology access weighting because wearable-derived endpoints are increasingly common. A CNS or psychiatric trial may weight regulatory and consent higher because state-level telehealth practice rules significantly affect remote assessments. The point is that every program should re-derive its weights, not inherit them.

Case example: a hybrid Phase II cardiometabolic protocol

Consider three candidate sites for a hybrid Phase II cardiometabolic protocol with an at-home dosing schedule, weekly wearable data collection, and mid-study home visits for laboratory draws.

Scorecard template structure

A working scorecard has three sections: identifier and traditional attributes (weighted per the therapeutic area), DCT dimensions with sub-criteria (weighted per the trial design), and a comments column with references to specific evidence supporting each score. Every score should be traceable to something documented in the feasibility questionnaire, a site qualification visit note, or an external data source (FCC, ACS, licensure database, prior study performance). Scoring without a paper trail is scoring the site’s confidence, not its capability.

Common Site Selection Mistakes When DCTs Became the Default

Over the past several years, as decentralized elements have moved from pilot to default in many programs, we’ve seen a consistent set of site selection mistakes across sponsors, especially those transitioning from traditional operations into hybrid designs. Seven mistakes come up repeatedly.

Mistake 1: Bolting a “DCT readiness” attribute onto the legacy scorecard

The most common pattern. The traditional weighted rubric stays untouched, and a single new attribute worth 5-10% of the score is added. Because traditional attributes still dominate, the site selection is functionally identical to the pre-DCT process, and the sponsor is puzzled when enrollment underperforms.

Mistake 2: Treating home health capability as binary

“Yes, we can do home visits” is not a score. Home health capability is a function of vendor MSA coverage, scheduling reliability, PI comfort, language coverage, and confidentiality documentation. Treating it as binary produces false positives.

Mistake 3: Assuming interoperability from vendor certifications

A site telling you it “supports FHIR” often means one integration in one system with one narrow use case. Sponsors should ask for prior integration examples and, where possible, request access to a test environment early in feasibility.

Mistake 4: Ignoring state licensure until initiation

State licensure environment should be part of the initial screening, not a last-mile check. Sites in states with restrictive telehealth interpretations can add weeks to startup and force protocol amendments for the affected participant population.

Mistake 5: Scoring the site instead of the catchment

The site is high-performing. The catchment is a broadband desert. This produces trials that look fine on paper and fail to reach the diverse populations the FDA Diversity Action Plan requires. Break every score by catchment segment where equity matters.

Mistake 6: Ignoring the courier map

DCT logistics fall apart when the courier network the sponsor selected does not have same-day service in the site’s catchment. This is knowable at feasibility. Sponsors that select their central logistics vendor before finalizing site list frequently discover this at site initiation, well after they can adjust.

Mistake 7: Failing to re-weight for the therapeutic area

A generic DCT scorecard applied identically to an oncology trial, a rare disease trial, and a cardiometabolic trial will produce systematically wrong rankings. Weightings must be re-derived by therapeutic area, ideally validated with two or three retrospective examples before being used prospectively.

Conclusion

Site selection for decentralized and hybrid clinical trials is no longer a variation on traditional site selection; it is a different problem. The variables that determine operational success have shifted from the four walls of the investigator’s clinic to the patient’s home, the courier’s route, the state’s licensure environment, and the interoperability posture of the site’s EHR. A scoring framework designed for the 2018 clinical trial landscape will systematically select sites that were excellent then and are only average now, while overlooking sites whose infrastructure fits the protocol at hand. The five-dimension framework we’ve described (patient technology access, home health and community networks, telehealth and data interoperability, regulatory and consent environment, and specimen and IMP logistics) is intended as scaffolding that clinical operations teams can adapt to their own portfolios, weightings, and therapeutic areas.

Sakara Digital works with pharma and biotech organizations building this kind of clinical operations capability, particularly in decentralized and hybrid trial design, site feasibility, and the data quality practices that make DCT platforms trustworthy for regulatory submissions. If you are re-thinking site selection for a hybrid or fully decentralized program and want an independent perspective on where to start, we are happy to have that conversation.