Broad consent vs. dynamic consent: what patients actually agree to

A research nurse slides a clipboard across the table. The form says your blood sample and your medical history may be used "for future research, including research that cannot be specified at this time." You sign, because you want to help, and because the alternative is going home. Twelve years later your sample is in a study you would never have agreed to, run by people you have never heard of, and no one is obligated to tell you.

That form was broad consent. It is legal, it is common, and it is the quiet center of an argument about what the word "consent" should mean when the future is the thing being consented to.

What broad consent actually asks of you

Broad consent is a single, up-front agreement to a wide and deliberately undefined range of future research. You sign once; the institution gains durable permission to store your data or biospecimens and reuse them across studies that may not exist yet.

The model exists because biobanks need it. A repository of half a million samples cannot re-contact half a million people each time a new question arises, so it asks for permission in bulk at the moment of collection. Large national efforts work this way, and the United States' All of Us Research Program is built around a broad, up-front permission to be studied over time. The economics are real: re-contact is expensive, response rates decay, and a sample you cannot use is a sample you might as well not have collected.

The cost is informational. Genuine informed consent depends on weighing a specific risk against a specific benefit, and you cannot weigh a study described only as "research that cannot be specified at this time." You are not consenting to a study. You are consenting to a process and trusting the people who run it. That trust may be well placed — but it is trust, not information, and the two are not the same thing.

Dynamic consent: one decision per study

Dynamic consent replaces the single signature with an ongoing relationship. Instead of agreeing once to everything, you make a fresh decision for each study, and you can revisit those decisions as your circumstances or feelings change.

The idea was formalized by researchers building digital consent interfaces in the early 2010s, and the mechanics are now familiar from the rest of your digital life: a standing record of what you have allowed, a notification when something new comes up, a control you can flip. The point is not to nag you. It is to make consent a living state rather than a frozen moment — and, critically, to make withdrawing as easy as granting. We unpack what a real "no" should do, and what it should not undo, in what revocable consent really means.

Dynamic consent has its own failure mode, which is honest to name: consent fatigue. Ask people to adjudicate every micro-decision and they will either rubber-stamp everything or abandon the platform. The design challenge is to surface decisions that matter — a new disease area, a commercial sponsor, a cross-border transfer — without drowning people in prompts about routine, low-stakes reuse.

What the Common Rule actually allows

In the United States, both models live inside the same regulation. The Common Rule — 45 CFR 46, the federal policy governing research with human subjects — was revised in 2018, and that revision added explicit provisions allowing broad consent as a lawful basis for storing, maintaining, and reusing identifiable private information and biospecimens for secondary research.

So broad consent is not a loophole; it is a sanctioned path, with required elements: a general description of the types of research, the types of data, the conditions of storage, and a statement that subjects may not be told about specific future studies. Dynamic consent, by contrast, is not a regulatory category at all. It is a way of satisfying ordinary informed-consent requirements more granularly than the minimum demands. The law sets a floor. How far above it you build is a design choice.

That choice has ethical roots older than the regulation. The Belmont Report — the 1979 foundation of US research ethics — grounds informed consent in respect for persons, the principle that people should govern what happens to them through their own choices. A signature on an unspecified future honors that principle thinly. A decision about a real, named study honors it fully. The same tension runs through European law, where the GDPR's consent standard demands that agreement be specific and freely given, a bar broad consent strains to clear on its own.

The trade-off, laid out

Neither model is simply better. They optimize for different things, and the honest comparison is a table.

The deciding question is what kind of evidence you want to be able to produce later. If a journalist, an auditor, or a participant asks "what exactly did this person agree to?", broad consent answers with a category and dynamic consent answers with a list. The longer a platform expects to hold a relationship with the same people, the more that difference compounds — and chronic-disease research, by definition, is a long relationship.

Why per-study consent earns more durable trust

Mithras Health is built on per-study consent, and the reason is the audit question above. Each study requests a revocable, scope-limited grant against a patient's records; the patient decides on a real study with a real purpose, sponsor, and data scope, using a two-document consent — a plain-language summary and a formal authorization, one decision, recorded and versioned. A grant can be withdrawn at any time, and the withdrawal is a first-class event on the same append-only audit trail as the grant. Because consent is processed per study, the platform can always answer the question broad consent cannot: not "did they agree to research?" but "did they agree to this one?"

This is more expensive than a single signature, and that expense is the point. History is full of samples and records reused in ways the people who provided them never imagined, from immortal cell lines to tribal blood studies — a lineage we trace in from Henrietta Lacks to data dividends, and one that maps cleanly onto how HIPAA treats research authorizations. The cheapest way to avoid repeating that history is also the oldest: ask the actual person about the actual study, and keep a record of what they said. Broad consent saves a conversation. Dynamic consent saves a relationship. Over a lifetime of chronic illness, the relationship is the thing worth keeping.