NIH’s Proposed Trial Form Puts Feasibility Where It Belongs: Before the Study Starts

The important change is not the form itself

The March 2026 NIH proposal outlined a 23-page, semi-structured application. Within the proposed “Approach” section, applicants would address the study setting and sites, single-site versus multisite organization, intervention availability and operational capacity, eligibility, outcome collection, recruitment, retention, participant burden, contingency planning, and statistical analysis. Human-subject protections, timelines, demographic plans, data management, and biospecimen handling would remain identifiable components. ([grants.nih.gov](https://grants.nih.gov/grants/guide/notice-files/NOT-OD-26-058.html))

That structure makes operational dependencies harder to hide in separate attachments. A recruitment projection cannot be evaluated responsibly without the eligibility criteria, visit schedule, intervention logistics, site capabilities, and retention demands sitting nearby. Respondents nevertheless asked NIH for clearer treatment of site readiness, multisite organization, community engagement, recruitment, retention, and inclusion. They also questioned whether one structure could accommodate complex, evolving, behavioral, psychosocial, and early-stage designs. ([grants.nih.gov](https://www.grants.nih.gov/news-events/nih-extramural-nexus-news/2026/09/heres-what-we-heard-about-the-proposed-clinical-trial-specific-application-form-and-whats-next))

The practical signal is straightforward: reviewers increasingly need to understand not only whether a trial is scientifically sound, but whether the proposed delivery system can execute it.

Feasibility should be auditable, not adjectival

Terms such as “large database,” “experienced investigator,” “diverse population,” and “strong referral network” are descriptions, not feasibility evidence. A useful assessment connects a defined source population to executable protocol criteria and then applies realistic losses at each step.

For example, an electronic health record count may identify patients with a diagnosis code, but the protocol may also require a narrow laboratory range, a specific treatment history, stable concomitant medication, specialist confirmation, willingness to stop an existing therapy, and availability for repeated daytime visits. Each requirement reduces the reachable cohort. The remaining patients may already be assigned to another study, managed by physicians who do not refer, or unable to meet transportation and scheduling demands.

ICH E6(R3) reinforces this upstream approach by calling for quality to be designed into trials, identification of factors critical to quality, and proportionate management of risks. CTTI’s recruitment framework similarly recommends evidence-based feasibility, realistic milestones, adequate resources, an ideal site profile, and early analysis of competition, participant burden, seasonal effects, and access to the target population. ([fda.gov](https://www.fda.gov/media/169090/download))

• Define the source population and the exact period represented by the data.
• Apply major eligibility criteria rather than relying on diagnosis-level counts.
• Separate patients directly controlled by the investigator from external referral potential.
• Document assumptions for contactability, interest, prescreen qualification, consent, and screening success.
• State data limitations, including missing laboratory, medication, or treatment-history fields.
• Identify competing trials, standard-of-care alternatives, and anticipated changes in the local treatment environment.

Five questions that should be answered before the forecast is accepted

A stronger feasibility package does not need to become a second protocol. It does need to answer several operational questions with enough specificity to support a decision.

First, where will participants actually come from? Trial teams should distinguish investigator practice, institutional databases, referring physicians, community partners, paid media, advocacy organizations, and sponsor-generated referrals. These channels have different lead times, costs, conversion rates, and ownership responsibilities.

Second, can the site execute the participant journey? Map consent, screening, washout, randomization, dosing, procedures, safety follow-up, reimbursement, and long-term retention. Include who performs each task, how long it takes, and what happens when visits fall outside normal clinic hours.

Third, what capacity is genuinely available? Investigator experience alone does not establish coordinator capacity, pharmacy capability, laboratory throughput, procedure-room availability, backup coverage, or the ability to manage several overlapping trials. Capacity should be assessed by time period, not as a permanent site characteristic.

Fourth, what will participation require from the community? NIMH’s planning guidance asks investigators to consider transportation, time away from work, childcare, language, literacy, out-of-pocket costs, trust, and relationships with local organizations. It also recommends using local data for enrollment estimates and planning resources across both recruitment and retention. ([nimh.nih.gov](https://www.nimh.nih.gov/funding/grant-writing-and-application-process/points-to-consider-about-recruitment-and-retention-while-preparing-a-clinical-research-study))

Fifth, does the inclusion plan describe access or merely demographics? In bilingual and culturally varied markets such as South Florida and Puerto Rico, translating a flyer does not create an enrollment pathway. Sites may need bilingual consent capacity, culturally and linguistically appropriate outreach, relationships with treating physicians, flexible scheduling, and processes that allow family members or caregivers to participate appropriately. These requirements affect budget, staffing, timelines, and vendor selection.

Multisite feasibility has an aggregation problem

A common study-level mistake is to add every site’s optimistic enrollment estimate and treat the total as independent capacity. Sites may draw from overlapping referral networks, compete for the same participants, activate at different times, or depend on the same central advertising campaign. Ten sites forecasting two participants per month rarely create a dependable 20-participant monthly rate on the first day of enrollment.

The NIH feedback specifically identified multisite organization and site readiness as areas needing clearer treatment. NIMH’s operational guidance asks whether each site can access the target population, how many participants it can screen and follow simultaneously, how staff will be trained, whether backup sites are planned, and how unbalanced enrollment will affect the study. ([grants.nih.gov](https://www.grants.nih.gov/news-events/nih-extramural-nexus-news/2026/09/heres-what-we-heard-about-the-proposed-clinical-trial-specific-application-form-and-whats-next))

A credible multisite model should therefore include activation curves, expected zero-enrolling sites, ramp-up time, concentration of enrollment among stronger sites, replacement-site decision points, and the coordinating center’s ability to detect and address problems. The goal is not to make the forecast pessimistic. It is to make the forecast manageable.

• Model enrollment by site and month rather than dividing the total target evenly.
• Separate startup assumptions from post-activation recruitment assumptions.
• Define early indicators such as leads contacted, prescreens completed, eligible decliners, screenings, and screen failures.
• Set decision dates for added support, retraining, enrollment reallocation, or replacement sites.
• Protect geographic and demographic objectives when shifting enrollment to high-performing sites.

What sponsors, CROs, investigators, and sites should do now

No organization needs to wait for NIH to finalize the form. Sponsors and CROs can involve sites before the protocol and enrollment model are effectively locked. Investigators can insist that recruitment projections account for visit burden, competing care options, staff capacity, and local referral behavior. Sites can replace unsupported claims with ranges, assumptions, source dates, and clearly stated constraints.

Just as important, trial leaders should stop rewarding the most aggressive feasibility answer. If conservative sites repeatedly lose awards to optimistic sites, the process selects for overstatement. Better feasibility depends on giving sites permission to identify risks without assuming that every limitation is a disqualifier.

The strongest lesson from NIH’s September update is not administrative. Trial feasibility belongs inside trial design. When site readiness, participant access, retention burden, and multisite execution are reviewed as core components of the study, teams have a better chance of correcting weak assumptions before they become amendments, rescue recruitment campaigns, or missed milestones.