TL;DR: Under the U.S. Common Rule, an activity is “human subjects research” only if it clears two separate regulatory tests at 45 CFR 46.102: it must be research (a systematic investigation designed to develop or contribute to generalizable knowledge, 46.102(l)) and it must involve a human subject (a living individual about whom an investigator obtains data or biospecimens through intervention/interaction, or obtains, uses, studies, analyzes, or generates identifiable private information or identifiable biospecimens, 46.102(e)(1)). Failing either prong takes an activity outside the Common Rule’s scope entirely — before any question of exemption, expedited review, or informed consent even arises.
The two-part test, in plain terms
Research administrators and investigators often use “human subjects research” (HSR) as a single phrase, but the regulation actually defines it as the intersection of two independently defined terms. An IRB (or a self-determination checklist built on the same logic) has to answer two separate questions in sequence:
- Is this activity “research”? Defined at 45 CFR 46.102(l) as a systematic investigation designed to develop or contribute to generalizable knowledge. Quality improvement projects, program evaluations, and most public health surveillance are frequently excluded here — not because they lack human participants, but because they aren’t designed to produce generalizable findings.
- Does it involve a “human subject”? Defined at 45 CFR 46.102(e)(1), covered in detail below. An activity can be rigorously systematic and clearly generalizable in intent and still fall outside the Common Rule if it never obtains data, biospecimens, or identifiable private information from or about a living individual.
Only when both conditions are met does the Common Rule apply, and only then do downstream questions — is it exempt, does it need full-board or expedited review, what informed consent elements are required — become relevant.
What makes someone a “human subject” under 46.102(e)(1)
The regulatory text defines a human subject as:
“A living individual about whom an investigator conducting research obtains information or biospecimens through intervention or interaction with the individual, and uses, studies, or analyzes the information or biospecimens; or obtains, uses, studies, analyzes, or generates identifiable private information or identifiable biospecimens.” (45 CFR 46.102(e)(1))
Unpacked, this definition has three load-bearing elements, and the second is structured as an “either/or”:
1. A living individual
The definition is explicitly limited to living people. Research using only information or biospecimens from deceased individuals does not create a “human subject” under 46.102(e) — a common source of confusion for investigators working with autopsy records, tissue repositories, or historical medical records. This does not mean deceased-subject research is unregulated in every respect: other legal frameworks (for example, HIPAA’s Privacy Rule, which protects a decedent’s health information for 50 years after death) and institutional or archival policies can still impose requirements even where the Common Rule itself does not apply.
2. Intervention or interaction, with subsequent use of the data
The first branch of the definition requires an investigator to obtain information or biospecimens through:
- Intervention (46.102(e)(2)) — both physical procedures used to gather information or biospecimens (a blood draw is the standard textbook example) and manipulations of the subject or the subject’s environment carried out for research purposes.
- Interaction (46.102(e)(3)) — communication or interpersonal contact between the investigator and the subject, which is the branch that brings surveys, interviews, and focus groups within scope.
Critically, this branch also requires that the investigator go on to use, study, or analyze what was obtained. Simply drawing a sample or asking a question, with no research use of the resulting information, does not on its own satisfy this branch.
3. Or: identifiable private information or biospecimens, obtained however
The second branch does not require any direct intervention or interaction at all. An investigator who obtains, uses, studies, analyzes, or generates identifiable private information or an identifiable biospecimen — for example, by requesting an existing, identified dataset from a hospital records system, or by re-identifying a supposedly de-identified specimen — has a human subject under this branch, even without ever contacting the individual.
“Identifiable” does the real work in this branch. Per 46.102(e)(5)–(e)(6), information or a biospecimen is identifiable when the subject’s identity is or may readily be ascertained by the investigator, or can be readily associated with the information or specimen. Data or specimens that are genuinely, durably de-identified — and not readily re-linkable by the investigator — fall outside this branch, which is why de-identification is such a common strategy for moving secondary-data-analysis studies outside Common Rule scope (and why an institution’s data-use agreement and code-key retention practices matter as much as the analyst’s own intent).
Putting the two prongs together
| Scenario | “Research” prong (46.102(l))? | “Human subject” prong (46.102(e))? | Common Rule applies? |
|---|---|---|---|
| Interview study on medication adherence, results intended for publication | Yes — systematic, generalizable | Yes — interaction, data used | Yes |
| Internal chart-review quality-improvement project, findings used only to change local clinic workflow | Typically no — not designed for generalizable knowledge | Would have been yes | No (case-by-case IRB/QI determination still often documented) |
| Secondary analysis of a public, pre-existing, de-identified federal dataset | Yes | No — not identifiable | No |
| Oral history project not designed to test a hypothesis or produce generalizable conclusions | Often no | Would have been yes | Often no, but institutional policy varies |
These examples are illustrative of how the two-prong logic is commonly applied, not a substitute for an actual IRB determination — the “research” prong in particular involves institution-specific judgment calls (quality improvement and program evaluation are the most frequently litigated edge cases) that a research office or IRB, not the investigator alone, should formally resolve and document.
Scope: who has to follow this definition
45 CFR 46 Subpart A — the Common Rule proper — applies to research that is conducted or supported by any of the roughly 20 federal departments and agencies that have formally adopted it (HHS among them), per 46.101. FDA-regulated research (drugs, biologics, and medical devices) is governed by a parallel but not identical set of human-subjects regulations at 21 CFR Parts 50 and 56, harmonized with but not identical to the Common Rule in places (for example, FDA’s exemption categories differ from 46.104’s). Many universities and research institutions also hold a Federalwide Assurance (FWA) with HHS’s Office for Human Research Protections, under which they commit, by institutional policy, to apply Common Rule standards to all human subjects research they conduct regardless of the funding source — which is why an unfunded or industry-funded study at such an institution can still be subject to IRB review even though 46.101’s federal-support trigger, read narrowly, wouldn’t otherwise reach it.
Determining exempt vs. non-exempt status
Once an activity is confirmed to be human subjects research under both prongs, the next question is whether it qualifies for one of the eight exemption categories at 46.104(d) — see Exempt human subjects research for the categories and the critical point that exemption is a determination made by an IRB or authorized reviewer, never a self-assessment by the investigator. Non-exempt human subjects research proceeds to full-board or expedited IRB review, and, for either exempt or non-exempt studies where consent is required, the specific disclosure elements set out in The Components of Informed Consent (45 CFR 46.116).
Frequently asked questions
Does a survey count as human subjects research?
Usually yes, if it meets the “research” prong (systematic, designed for generalizable knowledge) — a survey is a textbook example of “interaction” under 46.102(e)(3). Some survey research on adults, using no sensitive categories of data, may qualify for exemption under 46.104(d)(2), but exemption is a separate question from whether the Common Rule applies at all.
Is secondary analysis of an existing dataset human subjects research?
It depends entirely on identifiability. If the dataset is identifiable (or readily re-identifiable by the investigator), it meets the human-subject definition’s second branch even without any direct contact with participants. If it is genuinely de-identified and not linkable back by the investigator, it does not.
Does the Common Rule protect deceased individuals?
No — 46.102(e)(1) is explicitly limited to living individuals. Other protections (notably HIPAA’s 50-year post-mortem privacy protection for health information) can still apply to decedents’ data even where the Common Rule does not.
Who decides whether something is human subjects research?
An institution’s IRB, research office, or another individual formally authorized to make that determination — not the investigator’s own judgment, and not this or any other reference page. Institutional human research protection programs typically provide a determination worksheet or intake form for exactly this question.
Is “human subjects research” the same thing as “clinical research”?
No. Clinical research (interventional trials, observational clinical studies) is very often human subjects research, but human subjects research is the broader regulatory category — it also covers non-clinical work such as social-science surveys, educational research, and secondary analysis of identifiable records that have nothing to do with a clinical trial.
Related CASRAI resources
- Research (45 CFR 46.102(l)) — the “research” prong on its own, including the quality-improvement/program-evaluation screening test
- Common Rule (45 CFR 46)
- Exempt human subjects research
- Belmont Report
- The Components of Informed Consent: Required Elements Under 45 CFR 46.116
- The 4 Principles of Informed Consent
- CITI Human Subjects Research (HSR) Training
- FWA Lookup: How to Find an Institution’s Federalwide Assurance Number
- IRB/REC Approval Process: What Needs Review and What Doesn’t
