sharing-results — Boyce Data Science

What the draft would require

NIH is proposing that every NIH-supported clinical research study share plain-language, summary-level results with its participants unless there is an approved reason not to. The policy is a draft, and nothing in it is a current requirement.

The table paraphrases the notice and the NIH Office of Science Policy materials. Read the draft policy itself before relying on any line of it; the "what to check" column is where a rare disease, observational, or pediatric perspective is most likely to find gaps.

Element	Draft position, as described in the notice	What to check in the draft text
Who it covers	All NIH-supported clinical research, including clinical trials, regardless of funding level or mechanism, intramural included.	Whether "clinical research" follows the broad NIH definition (patient-oriented research, epidemiologic and behavioral studies, outcomes and health services research), which would put observational and natural history studies in scope. Whether registry and biobank contributors count as participants, and whether people screened but not enrolled do. Who the recipient is when the participant is a child.
What is shared	A concise plain-language overview that may cover objectives, how the study was conducted, methods, aggregate findings, relevance, limitations, and potential implications. The overview is distinct from individual research results.	Whether stratified results (subgroups, genotypes) count as "summary level." Whether exploratory analyses are expected, permitted, or discouraged in participant summaries.
When	For clinical trials, no later than one year after the primary completion date or the applicable ClinicalTrials.gov reporting deadline, whichever is later.	How timing is anchored for studies with no primary completion date, such as longitudinal cohorts, registries, and studies that produce findings over many years. Whether interim summaries are expected. What happens when the outcome families care most about arrives years after primary completion.
Planning	Sharing plans are integrated into study design, and researchers are encouraged to consult participants, IRBs, and institutional officials.	Whether consent language and preference capture are expected. Whether the costs are allowable and where they go in a budget.
Exceptions	NIH approval is needed where sharing is justifiably inappropriate, and is generally requested before the first participant enrolls.	What counts as "justifiably inappropriate," and whether identifiability in small cohorts is a recognized ground. The path for studies already underway when the policy takes effect.
Delays	Studies whose results are not complete within the period provide status updates to participants.	How often and in what form updates are expected, and whether a status update discharges the obligation for very long studies.
Compliance	Extramural recipients confirm through the Research Performance Progress Report (RPPR) closeout that results, or a status update, were shared.	What "shared" means when participants cannot be reached. The documentation burden for small teams.

NIH is explicitly asking for input on implementation, administrative burden, cost, infrastructure, tools, and lower-burden approaches, and says the feedback may shape supplemental guidance. Comments that include concrete estimates and workable alternatives are more likely to be reflected in that guidance.

The draft is Phase 1 of a two-phase effort. NIH held a public listening session in Spring 2026, co-hosted with the Multi-Regional Clinical Trials (MRCT) Center, and opened a written-input portal before drafting. Phase 2, on returning individual research results, is planned to follow.

What already exists

Several mechanisms already push trial results toward the public, and one already requires plain language. Almost all of them were built for interventional trials.

Existing mechanisms for sharing results, from the 2007 federal reporting law through the 2026 NIH draft

2007
Food and Drug Administration Amendments Act (FDAAA) Section 801 and the 2016 final rule (42 CFR Part 11, in the Code of Federal Regulations)Applicable clinical trials must post results on ClinicalTrials.gov, generally within a year of primary completion. The results are technical tables rather than plain language, and observational studies are outside the requirement.
2014
European Union (EU) Clinical Trials Regulation 536/2014Article 37(4) requires, within a year of the end of a trial, both a technical summary and a summary written for laypersons, and Annex V lists the elements a layperson summary must contain. Pediatric trials have six months. The European Commission published writing guidance in 2017, including a suggestion that pediatric studies consider a child-focused version.
2015
MRCT Center Return of Aggregate ResultsA multi-stakeholder workgroup produced a set of principles, a guidance document, and a toolkit with templates, an ethics-committee checklist, and an end-of-study preference form. The guidance notes that it can apply to observational studies and registries, and that some studies will not have discrete "results" or will not know the identities of participants.
2016
NIH Policy on Dissemination of NIH-Funded Clinical Trial InformationThe policy requires registration and results reporting on ClinicalTrials.gov for all NIH-funded clinical trials, including those not covered by FDAAA.
2018
Revised Common Rule, 45 CFR 46.116(c)(8)Informed consent may include a statement about whether clinically relevant research results, including individual results, will be disclosed and under what conditions. It is an additional element rather than a universal one. The same section governs the waivers and alterations of consent that are common in emergency neonatal research.
2018
National Academies report on returning individual resultsThe report examined benefits, harms, and costs of returning individual results and proposed a process-based approach to deciding what to return.
2026
NIH draft Sharing Summary Results PolicyThe draft is the first mechanism to reach all NIH-supported clinical research rather than trials alone, and to require plain language for participants specifically.

Participants overwhelmingly want results, and most never get them: in a survey of authors of 1,818 trials that enrolled patients, about a quarter had shared results with participants and a third had no plans to, and fewer than one in five said their funder had suggested it. A 2025 systematic review of 96 studies found that participants expect results regardless of outcome and that mailed lay summaries are the most common method. The guidance for doing this well is also thin: a June 2026 scoping review found 24 guidance documents, none specific to sharing randomized trial results with patients, and only five that had been pilot-tested.

Where summary becomes individual

The draft treats summary-level and individual results as different categories, which holds in a trial of thousands. In a rare disease cohort of 14, 40, or 80 people, the line between them can disappear.

Rare disease researchers have argued for years that coding and de-identification give weaker protection in small populations, while collaboration across borders makes some form of identification unavoidable; the response in the field has been governance and encrypted identifiers rather than silence. A participant summary is also a data release, and it deserves the same disclosure-risk review that a public data set would get.

The word "summary" describes different things at different scales. Rare disease cohorts and many pediatric studies resemble the right side of this picture.

Small cells

A table row that describes three people is a description of those three people. Public-health data releases use minimum cell sizes, commonly between 5 and 11, for this reason, and participant summaries rarely apply them.

Stratified by genotype

"People with variant X declined faster on the functional scale" reads as a prognosis to everyone who knows their variant. In many rare diseases most families do, and the sentence reaches relatives who never enrolled.

Communities that know each other

Families meet through patient organizations, family camps, and group chats, so a summary is read collectively. A phrase such as "one participant experienced" can become a name within days.

Exploratory findings are a second hazard

A statistically significant difference in a subgroup that was not specified in advance is, most of the time, noise. The classic demonstration comes from a 1988 heart-attack trial whose investigators showed that aspirin appeared not to work for patients born under two particular astrological signs, to make the point about chance findings. Credibility criteria for subgroup effects exist and are routinely violated. A post-hoc subgroup effect, translated into plain language and shared with a patient community, can become a funding priority within a fundraising cycle. The therapeutic misestimation literature describes this pattern of reading more benefit into a result than it can support.

One of the most protective sentences a participant summary can contain is a plain statement of which comparisons were planned in advance and which were not.

Results can distress, and people still want them

Receiving results is an intervention in itself. In the qualitative study that accompanied a large pregnancy trial, summary leaflets disappointed or distressed some recipients, and half did not find the leaflet clear; when the childhood follow-up of the same trial later found adverse outcomes for some children, the team had to design the feedback with participants rather than simply send it. Retinoblastoma survivors told about their elevated second-cancer risk mostly wanted the information even though it was upsetting. After a large breast cancer trial mailed results, 95 percent of surveyed participants were glad to have them and 23 percent were more anxious. The evidence argues for choice, preparation, and a route to a human being rather than for withholding.

Who interprets the summary

When a summary includes genetic or prognostic content, participants take it to their clinicians, who did not run the study, are not funded to explain it, and already spend a large share of the day on documentation and desk work. A summary-results policy that does not budget for interpretation can become, in practice, an unfunded mandate on treating physicians and genetic counselors.

Worked examples: infants and very young children

Research with the youngest participants makes the assumptions in the draft visible. The participant cannot consent, may not survive, may barely be remembered as having been enrolled, and may grow up to want a copy of the results in their own right. Both examples below are composites, and neither describes a specific study, site, population, or disease.

Example A. A randomized trial in a neonatal intensive care unit

A multicenter pragmatic trial compares two respiratory support strategies already in routine use for infants born before 30 weeks of gestation. About 800 infants are enrolled over four years at 20 sites. Parents consent within hours of birth, and at some sites an approved alteration of consent allows treatment to begin while consent is completed. The primary outcome is survival without a serious lung complication at 36 weeks of corrected age, and a pre-specified secondary outcome is neurodevelopmental status at two years of corrected age. Roughly one infant in ten does not survive to discharge. The trial is open-label, so parents know which strategy their infant received.

How the draft applies

Covered, with a timing problem: An interventional trial is squarely in scope. The clock starts at primary completion, when the last infant reaches 36 weeks. A year later the two-year neurodevelopmental results, the outcome most families care about, will not exist. As described, the draft would require a first summary, a status update, and then a second summary.
The recipient is not the participant: The participant is a child, and the recipient is a parent or guardian who consented in the first hours after birth. By the time two-year results are ready, some families have separated or moved, and about one in ten has lost the child.
Consent alterations: Where consent was altered or deferred under 45 CFR 46.116, some parents recall enrollment only vaguely, and a results summary may be the first moment participation fully registers.
Allocation is known: A summary favoring the other strategy is read differently by parents who know their infant received the one that fared worse. Parents of critically ill newborns interviewed after receiving trial results wanted them, and their reactions were shaped by the outcome of their own child.
Small cells inside a large trial: Gestational-age strata are standard, and the most immature infants are few at any single site, so mortality by arm and stratum produces small cells even in an 800-infant trial.
Exploratory signals: A post-hoc suggestion of benefit in the smallest infants would move quickly through parent networks and could shift practice before replication.

What good practice looks like

Capture preferences at consent, including whether to be contacted about long-term results; re-confirm at the two-year follow-up visit; keep contact details current through the follow-up clinic.
Prepare two versions of each summary: a standard version and a bereavement-sensitive version that opens with acknowledgment and lets parents decide whether to read outcome data. Offer a call with the follow-up team or a social worker.
State which comparisons were planned in advance, report by arm with uncertainty, and suppress site-level and very small stratum counts.
Use the two-year follow-up visit as the primary channel for families who attend and a mailed or emailed summary for those who do not, with review by the family advisory council of the unit before release and translation into the languages families use.
Budget for tracking families over three to four years, translation, two versions of two summaries, and a one-page brief for community pediatricians.

What a comment can request

Timing anchored to the outcomes families care about, with status updates when long-term follow-up results come later.
Guidance for bereaved families, including opt-in and a separate version.
Guidance for studies enrolled under consent waivers or alterations.
Allowable costs for multi-year family tracking.

Example B. A natural history study of infants with a rare, gene-defined condition

Infants identified in the first months of life with a rare, gene-defined neurodevelopmental condition, many through newborn screening or early genetic testing, are enrolled in a multi-site natural history study. About 40 infants are followed with clinical visits every six months for developmental milestones, feeding, seizures, and growth, and blood samples support biomarker work. Families are connected through one or two patient organizations and an active online group. Analyses describe developmental trajectories, frequently by variant type. Papers appear over several years, and the study has no single end point.

How the draft applies

Probably covered, with no clock: Observational research with clinical visits is clinical research in the usual NIH sense, so the policy likely applies, but there is no primary completion date and results arrive paper by paper. The draft, as described, does not say when the summary obligation begins for a study like this.
Genotype strata of three to eight children: A sentence such as "children with variant type X had not achieved independent sitting by age two" is a prognosis for every family that knows its variant, and most do before symptoms appear. It also reaches carrier parents and siblings who never enrolled. Here the summary functions as individual results.
A community where families know each other: Forty families in two organizations are likely to map any statement onto specific children, so a phrase such as "one infant experienced" can become a name.
Exploratory signals: An apparent association between an early supportive therapy and better milestones could drive fundraising and off-label demand within a fundraising cycle.
The right not to know: Some families deliberately avoid trajectory information in the first years, and preferences captured at enrollment may not hold two years later.
Participants grow up: Some children will later be able to receive an age-appropriate version of the results themselves, and the notion of "participant" in the draft should anticipate that.

What good practice looks like

Agree on a schedule with families: an annual update plus a summary when each paper is published, with interim summaries only for pre-specified analyses.
Set a minimum cell size before drafting, collapse variant types into groups where needed, and describe ranges of trajectories rather than individuals.
Treat genotype-stratified findings as individual-level: a genetic counselor joins the family webinar, and a one-to-one call is offered.
Co-design the summary with the patient organizations, brief them on what any exploratory finding does and does not mean before release, and send a one-page brief to community neurologists and pediatricians.
Plan a child-friendly version for later, and align summaries with any patient registry the study feeds.

What a comment can request

Minimum cell-size guidance for participant-facing materials.
Individual-results safeguards for genotype-stratified summaries.
A defined timeline for observational and longitudinal studies.
Co-design with patient organizations, and recognition of registries as return channels.
Allowable costs for genetic counseling time, translation, and hosting.

Both examples can be inserted into a comment as illustrative scenarios from the comment builder, and either can be replaced with a scenario from your own experience.

Generate considerations for a study

Describe a study by choosing the options that fit it. The output is a tailored set of guardrails and questions, grouped the way an IRB or a comment letter would group them. Nothing you enter is stored or sent anywhere.

Ways to return results

No single channel does everything. The MRCT Center principles put participants or their designees at the center and ask sites to support dissemination, and the practical question is which combination fits the study and the community.

Channel	Where it works	Cautions	Rare disease and pediatric note
Plain-language letter or email	It is scalable and documented, and it can be layered. It is the most common method in practice.	Reading level, no way to ask a question, and the "leaflet effect" of disappointment or distress. Lay summaries are often not fit for purpose.	Assume it will be forwarded, and write for the whole community as well as the recipient. Prepare a bereavement-sensitive version where any participants have died.
Public posting (study site, ClinicalTrials.gov results, registry page)	It reaches people who were screened out and the wider community, and it satisfies transparency.	Posting does not by itself share results with participants, and registry results tables are technical.	Apply minimum cell sizes to public tables, and consider a plain-language companion.
Community webinar or meeting, recorded	It is two-way, clinicians can join, and questions get answered.	Attendance skews, and recordings persist and travel.	Co-host with patient organizations and prepare for subgroup questions in advance.
Conversation with a clinician or genetic counselor	It suits prognostic or genetic implications.	It is unfunded and not scalable, and the clinician may not know the study.	Send clinicians a one-page brief before participants receive anything. In pediatric studies, the follow-up clinic visit is often the natural moment.
Patient organization newsletter or registry portal	It is a trusted, familiar channel that reaches families.	Framing drift and organizational priorities.	Agree on wording before release and decide who answers follow-up questions.
Short video or infographic	It supports comprehension across varied literacy; one trial team tested a one-page text, a one-page infographic, and a video with participants.	Oversimplification and a single-number takeaway.	Avoid headline numbers for small subgroups. A child-focused version can follow later.
Layered approach	A notice that results exist comes first, then a summary, then deeper material, then a conversation on request, which honors preferences.	There are more moving parts to plan and document.	It is a reasonable default for small cohorts with prognostic content, and for families who consented in a crisis.

Whatever the channel, the end-of-study preference form in the MRCT toolkit is a good one to adopt: preferences captured at consent may be years old by the time results exist, and people change their minds in both directions. As of 2025, tools for communicating results were beginning to be tested head-to-head, and a rare disease cohort is running a series of randomized comparisons of formats.

Points for your comment

Each theme below pairs a prepared case with concrete asks, written from the position of someone who wants the policy to exist and can see where it is likely to break. The comment builder turns them into a referenced comment you can edit and download.

Build your comment

Walk through the themes, weigh each one, add your own experience, and download a referenced comment as a Word document, a PDF, or plain text. Nothing is stored, and progress is kept only while this page is open, so save a progress file or download the comment before closing it.

Step 1

References

Links go to the article, notice, or document where a stable page was confirmed on September 17, 2026, and every link opens in a new tab. Where a link is labeled "find on PubMed," it opens a PubMed search for the citation; confirm the digital object identifier (DOI) before citing. The comment builder renumbers the references it uses in order of first appearance.