Patient registries: the quiet workhorses of chronic-disease research

In 1966, the Cystic Fibrosis Foundation started keeping a list. Care centers reported each person with CF they treated — survival, lung function, treatments — year after year, in a common format. Six decades later, that registry follows most Americans living with the disease, and it is part of how a diagnosis that once rarely let a child reach adulthood became one where predicted survival extends well into middle age. The list showed which care centers were getting better results, and it let everyone else study why.

That is a patient registry: not a breakthrough, not a molecule — a list, kept with discipline, for a long time.

What a registry actually is

A patient registry is an organized data system built around a defined population and a stated purpose. The standard reference is the registry user's guide published by AHRQ, the US Agency for Healthcare Research and Quality, whose definition is worth quoting because every word is load-bearing:

An organized system that uses observational study methods to collect uniform data (clinical and other) to evaluate specified outcomes for a population defined by a particular disease, condition, or exposure, and that serves one or more predetermined scientific, clinical, or policy purposes.

Organized — not a pile of charts. Uniform — the same fields, defined the same way, at every site and every visit. Predetermined — built to answer something, not hoarded in case something turns up. A registry is not a trial: it observes routine care and assigns nothing. It is not a biobank: it holds data rather than specimens, though the two are often paired. It is closer to a long-running census of one disease, designed by people who knew what they would need to know in twenty years.

Three jobs registries do

Most registries are doing one of three jobs, and the best do two at once.

The quality job deserves a story, because it is the least glamorous and the most quietly effective. Sweden has run a national hip-arthroplasty register since the late 1970s, and national joint registries in Scandinavia, the UK, and Australia flagged underperforming hip implants ahead of regulators — thousands of revision surgeries avoided because someone kept uniform records of every operation, for decades, and looked. No single hospital could have seen the pattern. The registry existed so the pattern had somewhere to appear.

Registries are also a primary source of real-world evidence: when regulators weigh how a product behaves outside trials, registry data is routinely what they are weighing.

For rare diseases, the registry is the field

When a condition affects a few thousand people, the registry often is the research field — the only systematic evidence in existence. Natural-history data from registries determines what a feasible trial looks like: how fast the disease progresses, which outcomes move, how much variation to expect. In some rare-disease decisions, registry data has served as the comparison against which a therapy's effect is judged, because a placebo arm was impossible or unethical. And when a therapy finally arrives, the registry is trial-readiness itself — it knows where the patients are.

Many of these registries are run not by institutions but by patient organizations, sustained by the communities they describe — a pattern with a long track record that we trace in how patient communities move research. The NIH supports rare-disease research networks for the same reason: without standing infrastructure, every study starts from zero.

Why registries die

Registries usually fail for infrastructure reasons, not scientific ones. The mission is measured in decades; the funding arrives in three-to-five-year grants. Each disease community commissions its own database, its own consent forms, its own governance — paying full price for plumbing that is identical to the registry next door. Consent collected for yesterday's purpose cannot stretch to tomorrow's question, so data sits unusable or gets used uneasily. Data entry is the first thing unfunded, so quality erodes until the registry is a list of names with stale fields.

Every one of these failures orphans years of patient effort — thousands of people who filled in forms, returned for visits, and trusted that the record would outlive the grant cycle. Often it does not.

One spine, many diseases

The parts of a registry that are hard — consent management, privacy enforcement, audit, data standards — are the same for every disease. Only the clinical content differs. That is the economic argument for sharing the spine instead of rebuilding it per condition, which we make in full in what 28 diseases on one platform makes possible.

It is also what Mithras Health is: registry-grade infrastructure built once and shared. Twenty-eight chronic conditions across twelve categories run on one stack — records kept as FHIR R4 / US Core with OMOP views for research, per-study revocable consent rather than a blanket signature, a hard k ≥ 5 release floor, and an append-only audit trail retained for six years. The flagship is an interstitial cystitis / bladder pain syndrome cohort built with a patient community whose accumulated knowledge spans roughly thirty years — the kind of longitudinal substance no funder can commission retroactively, as we describe in why IC research has been so hard. On a shared spine, the marginal cost of the next disease's registry falls toward zero — which is the only arithmetic under which underfunded conditions ever get one.

The list the CF Foundation started turns sixty this year. The diseases still waiting for their version of it are not waiting for a scientific breakthrough. They are waiting for a list — and for infrastructure cheap enough that no community has to spend a decade building one before the science can start.