.png){kind=logo}
NEST has opened a public comment period on a new tool that changes when, and how, you should be scrutinizing a real-world data source. The NEST Mark: Real-World Data Source (Public Comment Draft version 0.1, May 4, 2026) is a structured, study-agnostic evaluation of whether an RWD source has the foundational quality needed to support regulatory-grade real-world evidence. It comes from the National Evaluation System for health Technology Coordinating Center (NESTcc), the MDIC-run public-private partnership established under a 2016 FDA cooperative agreement, and was developed with input from CDRH and industry partners. The comment window runs from May 4 through July 3, 2026. Below is what it says and what it means if you are building a device program that will lean on RWE.
What the NEST Mark is
FDA has, for several years, encouraged device manufacturers to submit real-world evidence to support regulatory decisions, from label expansions to post-market surveillance. The persistent problem is that the quality of the underlying real-world data varies enormously, and most teams only discover a source's limitations after they have committed to a study question and started the analysis.
The NEST Mark inverts that sequence. It evaluates the data source itself, independent of any specific study, design, or submission. The question it answers is not "is this dataset fit for my study" but "does this source have the accuracy, traceability, completeness, and governance to be a credible foundation for regulatory evidence at all." A data source holder that demonstrates mature, well-governed systems and processes may be granted the NEST Mark: Real-World Data Source designation. Once a source carries the mark, anyone deriving a dataset from it inherits a completed foundational assessment, which reduces redundant due diligence and provides a transparent baseline for the study-specific work that still has to happen.
Why this tool exists now
The NEST Mark is a direct response to hard evidence about what goes wrong with RWD. In 2024, a RAND team led by Justin Timbie published a synthesis of 18 NEST Test-Cases conducted between 2020 and 2022, drawing on study reports and 49 semi-structured interviews with participating organizations. The Test-Cases were pilot projects designed to test whether existing RWD could actually answer the regulatory questions manufacturers wanted to ask.
The recurring failure points were specific and familiar to anyone who has worked with claims or EHR data:
- The lack of unique device identifiers (UDIs), making it hard to find a specific device in the data at all.
- Limited reliability of diagnosis and procedure codes in structured data.
- Key clinical details trapped in unstructured EHR notes rather than structured fields.
- Thin capture of long-term study endpoints.
- Missing data and friction around data sharing.
The same study identified what worked: leveraging manufacturer and supply chain data, using clinical registries and their reporting processes, querying standardized EHR data models, applying natural language processing to unstructured text, and staffing multidisciplinary teams. The NEST Mark takes those lessons and turns them into an upfront, repeatable quality bar.
How the tool is structured
The questionnaire runs to eleven sections. Each one targets a dimension of data quality that has historically undermined RWE submissions, and each asks the holder to reference real documentation (SOPs, flow diagrams, QA reports, audit results) rather than offer assurances.
- General data holder and data source information. What the source is, where the data originate, how often it updates, the lag to research readiness, and the device identification elements available (UDI, product code, brand, model, manufacturer).
- Governance and quality management system. Data governance, a functioning QMS, SOP coverage, standards alignment and certification, and audit and inspection history.
- Data consistency and stability and quality control. The QC and QA framework, plus tracking of every extract and query run for research.
- Data accrual, traceability, and versioning. Provenance of the data, the ability to trace a value back to its origin, and disciplined versioning of both the system and the data source.
- ETL and technical controls. The extract-transform-load process, and critically, how algorithmically derived fields, unstructured data, and AI/ML methods are handled and validated, along with privacy-preserving transformations.
- Data integration and linkage. How records from different sources are joined and the integrity of that linkage.
- Data access and sharing, privacy and security, and regulatory transparency. Access controls, storage and retention, correction processes, privacy, and de-identification.
- Representativeness, continuity of care, and longitudinality. Whether the population reflects the target, whether patients are observable across settings, and how deep the longitudinal follow-up runs.
- Quantitative data source characteristics. The hard numbers on size, coverage, and completeness.
- Documentation package and signature. The supporting evidence and an attestation.
What this means for developers and manufacturers
Vet the source before the study, not after. The single biggest shift here is timing. If your RWE strategy depends on a particular registry, claims aggregator, or EHR network, the NEST Mark gives you a way to pressure-test that source against a regulatory-grade rubric before you commit a study design to it. A source that cannot trace provenance or version its data is a risk you want to surface early.
Treat AI/ML-derived fields as a first-class quality concern. The tool explicitly probes algorithmically derived fields, NLP outputs, and AI/ML methods used to populate the data. If your evidence depends on variables that were inferred rather than directly recorded, expect to document how those methods were validated. This is consistent with where FDA is heading on AI in regulated contexts generally.
Device identifiability is still the gating issue. The Timbie findings, and the tool's detailed device-identification matrix, both reflect that you cannot generate device evidence from data in which you cannot reliably find the device. If UDIs are absent, plan for the supply chain and registry strategies that the Test-Cases showed can work.
A designated source is not a finished study. The NEST Mark is foundational, not study-specific. Even with a marked source, you still owe a fitness-for-use assessment once the study question and curated dataset are defined. The value is that you start that work from a known, transparent baseline instead of from scratch.
A few caveats worth holding
This is a draft. NEST has said the tool will evolve through the comment period and continued engagement with stakeholders including FDA, and that the final version is expected to be digitized. The designation process, rubric, and thresholds are still being shaped. That is precisely why the comment window matters: if you hold or rely on an RWD source, this is the moment to influence what "regulatory-grade" will mean in practice. Comments are open through July 3, 2026.
The bigger picture
The NEST Mark is part of a broader move to make real-world evidence dependable enough to carry regulatory weight rather than merely supplement it. It pushes data quality upstream, standardizes what "good" looks like, and creates reusable infrastructure that compounds across studies. For teams who have watched promising RWE efforts stall on data they could not trust, that is a meaningful change. For more on how quality frameworks are reshaping evidence generation, see our related write-up on the Cosm blog.
How Cosm Can Help
Cosm helps medical device, SaMD, and AI/ML companies build regulatory and quality strategy that holds up under FDA scrutiny, including real-world evidence planning, data source evaluation, and the quality systems that underpin them. If you are weighing whether an RWD source can support your submission, or want to position a source for a NEST Mark assessment, we can help you scope it. Reach us at info@cosmhq.com or visit cosmhq.com.
You can download the full NEST Mark draft tool from our resources library, linked above.
Disclaimer - https://www.cosmhq.com/disclaimer
