Rare disease registries and data trust requirements for research use
Fewer than 6% of the estimated 10,000 rare diseases have an FDA-approved therapy, and fragmented registry data is a primary reason. Rare disease registries collect critical patient information, but without structured data trust requirements, most of that information cannot reliably support research, regulatory submissions, or AI model training.
Fewer than 6% of the estimated 10,000 rare diseases have an FDA-approved therapy. One of the biggest barriers is not biology. It is data. Rare disease registries exist to close that gap, but the registry itself is only as useful as the trust infrastructure underneath it.
Most rare disease research data fails basic quality checks before it ever reaches a model, a regulatory submission, or a clinical trial protocol. The problem is not collection. It is governance.
What is the rare disease registry framework?
A rare disease registry is a structured system for collecting, storing, and sharing clinical and demographic information about patients with a specific rare condition or group of conditions. The framework typically includes standardized data elements, enrollment protocols, consent management, and governance rules for who can access what data and under what conditions.
The NIH, NORD, and EURORDIS have each published guidance on registry design. Common elements include natural history data, genotypic and phenotypic information, treatment outcomes, patient-reported outcomes, and demographic variables. The Global Rare Diseases Patient Registry Data Repository (GRDR) maintained by the NCATS program at NIH attempts to harmonize elements across registries using Common Data Elements (CDEs).
But frameworks describe structure. They do not enforce trust. A registry can follow every structural recommendation and still contain records with unknown provenance, outdated consent, or unvalidated clinical values.
Key statistics
Rare diseases affect an estimated 25 to 30 million Americans, roughly 1 in 10 people. NORD tracks over 1,200 rare disease patient registries in the U.S. alone. The average time to diagnosis for a rare disease patient is 4.8 years, during which clinical data scatters across 7 or more providers. In SuperTruth's work with imaware, standardizing 105,000 diagnostic records reduced processing time from 3 weeks to 2 hours, a 95% reduction. Over 80% of rare diseases are genetic in origin, making genomic data integrity a non-negotiable requirement for research use.
Can you use PHI for research without any restrictions?
No. Protected Health Information (PHI) used in research is governed by HIPAA's Privacy Rule, the Common Rule (45 CFR 46), and often additional state-level statutes. Research use of PHI requires one of three pathways: individual patient authorization, a waiver of authorization granted by an Institutional Review Board (IRB) or Privacy Board, or the use of a limited data set under a data use agreement.
For rare disease registries, the restrictions are even more consequential. Small patient populations increase re-identification risk. A dataset of 40 patients with a condition affecting 1 in 500,000 people cannot be anonymized the same way a diabetes cohort of 10,000 can. Cell sizes below certain thresholds may need suppression. Geographic identifiers, age ranges, and treatment timelines can all become indirect identifiers when the population is small enough.
This is why consent governance matters even more in rare disease contexts than in general population research. Consent must be granular, versioned, and auditable.
Which research method is most suitable for studying rare diseases?
Natural history studies and patient registries are the most suitable primary research methods for rare diseases. Randomized controlled trials (RCTs) are often infeasible due to small sample sizes. Registries enable longitudinal observational data collection across geographically dispersed patients, which supports natural history characterization, endpoint development, and external control arm construction for clinical trials.
Recent examples demonstrate the value of registry-derived evidence. The T1D Exchange Clinic Registry, referenced in MedPageToday coverage, enabled research across 25,759 participants with Type 1 Diabetes, including analysis of co-occurring autoimmune diseases. Similarly, gene therapy research for Leber hereditary optic neuropathy (LHON) relied on long-term outcome tracking that only registry-like infrastructure can support. MedPageToday reported in December 2024 that patients treated with lenadogene nolparvovec showed sustained vision benefits, data that required years of structured follow-up.
The method works. But the method depends entirely on data quality.
What are at least three things to know about clinical data registries and why they matter?
First, registries are not databases. They are governed data systems with rules about who contributes, what gets collected, how consent is managed, and who can access records for secondary use. A database stores data. A registry enforces standards around that data.
Second, data quality degrades over time. Registries that do not enforce recency checks, validation rules, and concordance monitoring will accumulate stale and conflicting records. A patient's treatment status from 2019 is not reliable for a 2025 analysis. Temporal drift is real and measurable, as we have written about in the context of AI model accuracy.
Third, registries create the foundation for regulatory-grade evidence. The FDA increasingly accepts real-world evidence (RWE) from registries for rare disease drug approvals, but only when the data meets specific provenance, quality, and consent thresholds. Without trust scoring, a registry cannot demonstrate that its data meets those thresholds.
Why rare disease registry data needs trust scoring
SuperTruth's Data Trust Index (DTI) scores every health data record from 0 to 100 across eight dimensions: Provenance (25%), Consent (20%), Recency (15%), Quality (10%), Concordance (10%), Validation (10%), Breadth (5%), and Stability (5%). For rare disease registry data, three dimensions matter most.
Provenance answers: where did this record originate, and can we verify the chain of custody? For multi-site registries collecting data from dozens of clinics across different EHR systems, provenance is the first point of failure.
Consent answers: does the patient's authorization cover this specific research use, this specific data element, this specific time period? Rare disease patients often consent broadly because they want to accelerate research. But broad consent recorded vaguely is legally fragile.
Recency answers: when was this record last validated against a clinical source? A rare disease registry with 10 years of data is valuable only if someone can distinguish which records reflect current patient status and which reflect historical snapshots.
In our work with imaware, applying DTI scoring to 105,000 diagnostic records saved over 200 hours per month and surfaced a patient segment driving 20% of revenue that had been invisible in unscored data. The same principles apply to rare disease registries at any scale.
The path forward for rare disease research data trust
Rare disease registries sit at the intersection of the most sensitive data challenges in healthcare: small populations, genetic information, pediatric participants, multi-site collection, and long time horizons. Every one of these factors amplifies the consequences of poor data governance.
The registries exist. The patients have contributed their data, often at significant personal cost and with genuine hope that it will accelerate treatment. The least we can do is make sure that data is trustworthy before anyone builds on it.
To explore how DTI scoring applies to rare disease registry data, contact Louis Simeonidis, SVP Commercial Operations, at louis@supertruth.ai or (215) 918-4140.
Further reading:
Jason Alan Snyder
Co-founder of SuperTruth and Artists & Robots, and an inventor on the Data Trust Index patents. Twenty-plus years building technology inside Interpublic Group. He writes here nearly every day on data trust, provenance, and what AI should be allowed to act on, and publishes essays on his Substack.
About SuperTruth · LinkedIn · Substack · jasonalansnyder.com
See it in practice
750+ cancer search terms. Live in production.
VIOLET maps behavioral signals 12–18 months before clinical presentation.