Guide · Registry Demo

Start here: two kinds of data

This is the distinction that causes the most confusion in registry planning, so it comes before the glossary.

Two different kinds of data

Survey & participant-reported data

What a person tells you directly: enrollment questions, survey modules, quality-of-life instruments, treatments they report taking. You design it, you own the instrument, and the person is the source.

Sometimes called patient-reported outcomes, or PROs.

Electronic health record (EHR) data

What clinicians recorded during care: diagnoses, medications, lab results, visits, clinical notes. You do not design it and you cannot change it. It arrives from a health system, and getting it is a separate project with its own cost, consent, and technical work.

Also called health records, clinical data, or real-world data.

Most registries need both, and they are not interchangeable. Survey data tells you what daily life is like and what someone believes happened. EHR data tells you what was documented, coded, and billed. Where they disagree, that disagreement is often the finding.

Throughout the site, sections that apply to only one carry a tag: Survey or EHR. Sections that apply to both carry Both.

Registries and studies

Patient registry	An organized collection of data about people sharing a condition, exposure, or characteristic, gathered for a defined purpose. An established term of art, kept as-is.
Contact registry	A maintained list of who your community is and how to reach them. Not a research dataset, but the foundation of one.
Natural history study	A study describing how a condition unfolds over time in the absence of, or alongside, intervention. Usually the evidence a sponsor needs before designing a trial.
Trial readiness	Whether a community's data and infrastructure would let a trial be designed, powered, and recruited.
External control arm	Using existing observational data in place of a concurrent control group. Places very high demands on data quality and provenance.

Standards and models

C-CDA	Consolidated Clinical Document Architecture. A document-based exchange standard; each file carries a header identifying the person and a body of sections, each with both narrative text and coded entries.
FHIR	Fast Healthcare Interoperability Resources. A resource-based standard where discrete records reference each other. Easier to load into tables; references can point outside the file you were sent.
OMOP CDM	A common data model that gives source codes a standard identity so the same analysis can run at many sites without rewriting it.
Common data element	A variable defined once and reused across studies so that the same question means the same thing everywhere.
Phenopacket	A GA4GH exchange format for deep phenotyping, used in rare disease matchmaking and discovery.
Ontology	A structured vocabulary of concepts and their relationships, SNOMED CT, LOINC, RxNorm, HPO, Mondo. What makes a code mean something outside the system that issued it.

Getting records: EHR only

Electronic health record (EHR) data	What clinicians documented during care: diagnoses, medications, labs, visits, notes. You do not design it and cannot change it. Acquiring it is a separate project with its own cost, consent, and technical work. Everything in this section concerns it.
Survey / participant-reported data	What a person tells you directly, through instruments you design and own. Costs nothing to acquire beyond the platform. Not covered in this section, see the modules in the participant demo instead.
Patient portal	The interface through which a person accesses their own records. An established term of art. The download options are usually the cheapest acquisition pathway available.
HIPAA authorization	A signed permission allowing a third party to retrieve a person's records across the places they have been seen. More comprehensive than a portal download, and considerably more expensive.
TEFCA	The Trusted Exchange Framework and Common Agreement, a US framework for nationwide health information exchange between participating networks.
Individual access services	A route under network exchange through which a person requests their own records across participating providers, subject to identity verification.
Patient matching	Determining that two records describe the same person. An established term of art. Imperfect in both directions.

Governance

Data use agreement	The terms a data requester signs before receiving anything.
Data access committee	The group that decides who gets data, against a written rubric.
Broad consent	Consent covering future unspecified research within stated limits, rather than one named study.
De-identification	Removing or transforming identifiers so a dataset no longer identifies individuals under an applicable standard.
Pseudonymization	Replacing identifiers with consistent surrogates, keeping the mapping somewhere controlled. Not the same as de-identification.
FAIR	Findable, Accessible, Interoperable, Reusable. Compatible with controlled access, FAIR does not mean open.

Quality and methods

Computable phenotype	A rule, expressed in code, that decides who counts as a case.
Conformance, completeness, plausibility	The three families of data quality check in common use: does it match the expected structure, is it there, and could it be true.
Patient-reported outcome	A measurement reported directly by the person, without clinician interpretation. An established term of art.
Representativeness	Whether the people in your registry resemble the people with the condition. A design problem, not a limitation to note afterward.
Provenance	The record of where a value came from and what was done to it. The difference between a number you can defend and a number you cannot.

Implementation

Implementation science	The study of how evidence and interventions actually get adopted and sustained in real settings.
RE-AIM	Reach, Effectiveness, Adoption, Implementation, Maintenance. A planning checklist.
Maintenance	Whether the work survives the launch, the grant, and the founder. Reliably the hardest dimension for small organizations.
Advocacy group	A community-led organization representing people affected by a condition. Preferred here over 'patient advocacy group', though the longer phrase is retained in literature searches because that is how the field is indexed.

A note on language

This site prefers people with lived experience and participants over "patients", and advocacy group over "patient advocacy group". The exception is established terms of art , patient registry, patient portal, patient-reported outcome, patient matching, patient-directed data, where changing the word would make the term unrecognizable or unsearchable.

The Live Monitor follows the same rule with one deliberate wrinkle: filter labels use the preferred term while the underlying queries also include the indexed phrase, so nothing is lost from the search. See content/monitor_topics.py.