Under federal regulations, research is defined as “a systematic investigation, including research development, testing, and evaluation, designed to develop or contribute to generalizable knowledge.” This definition, found in 45 CFR 46.102, has two core criteria that must both be present for an activity to qualify as research: the activity must be a systematic investigation, and it must be designed to produce generalizable knowledge. Understanding these two criteria is essential because any activity meeting this definition triggers oversight requirements, including potential review by an Institutional Review Board (IRB).
The Two Core Criteria
The regulatory definition hinges on a two-part test. First, the activity must involve a systematic investigation. Second, that investigation must be designed to develop or contribute to generalizable knowledge. Both elements must be present. A project that collects data in a structured way but has no intent to produce broadly applicable findings doesn’t meet the threshold. Likewise, stumbling onto a generalizable insight through random, unstructured observation doesn’t qualify either.
This definition applies regardless of what a program calls itself. A demonstration project, a service program, or a quality improvement initiative can all constitute research if the activity meets both criteria. The label doesn’t matter; the design and intent do.
What “Systematic Investigation” Means
The Office for Human Research Protections (OHRP) describes a systematic investigation as a methodical approach to gathering and analyzing information. In practice, this means the activity would typically involve a hypothesis or research question, a plan for collecting data, and a method for analyzing that data. Think of it as the difference between casually noticing a pattern among your patients and formally designing a protocol to test whether that pattern holds up.
The “systematic” part also encompasses research development, testing, and evaluation. So pilot studies, feasibility assessments, and program evaluations can all qualify as systematic investigations if they follow a structured methodology. You don’t need to be running a randomized controlled trial. Any commonly accepted scientific method, applied in a deliberate and organized way, can satisfy this criterion.
What “Generalizable Knowledge” Means
The second criterion asks whether the investigation is designed to produce conclusions that apply beyond the specific people or setting being studied. The Office of Research Integrity puts it plainly: a project is research if it draws conclusions that have “some general applicability” using a commonly accepted scientific method. Randomly collecting information about individuals with no intent to draw broader conclusions is not research under this definition.
Intent matters here. If you design a project to answer a question whose findings could inform practice, theory, or policy for others beyond your immediate context, the activity is designed to contribute to generalizable knowledge. The intent to publish results in a journal or present them at a conference is often a strong signal, though it’s not the sole determining factor. An activity can meet this criterion even if publication isn’t planned, as long as the design aims to produce broadly applicable findings.
Activities Explicitly Excluded
The regulation carves out four categories of activity that are deemed not to be research, even if they share some surface features with systematic investigation:
- Scholarly and journalistic activities such as oral history, journalism, biography, literary criticism, legal research, and historical scholarship, particularly when focused directly on the specific individuals involved.
- Public health surveillance conducted or authorized by a public health authority, including monitoring disease outbreaks, tracking risk factors, and providing situational awareness during public health emergencies like natural disasters.
- Criminal justice data collection by or for a criminal justice agency, when authorized by law or court order and used solely for criminal justice or investigative purposes.
- National security operations including authorized activities supporting intelligence, homeland security, defense, or other national security missions.
These exclusions exist because while such activities may involve structured data collection, their primary purpose is something other than producing generalizable knowledge for the broader scientific community.
Where Case Reports and Small Analyses Fall
One of the most common practical questions is whether a case report qualifies as research. Many institutions, including Johns Hopkins, treat a retrospective analysis of one, two, or three clinical cases as a medical or educational activity rather than research. The reasoning is that examining a small number of individual cases doesn’t constitute a systematic investigation designed to produce generalizable knowledge. Once you move beyond three cases into a larger case series, the activity starts to look more like research and typically requires IRB review.
This is a useful illustration of how the two criteria work together. A clinician writing up an unusual case for a journal is describing a specific clinical experience, not testing a hypothesis across a population. But if that same clinician designs a protocol to systematically review 50 similar cases to identify treatment patterns, the activity crosses the line.
How the FDA Definition Differs
The FDA uses a related but distinct term: clinical investigation. Under 21 CFR 56.102, a clinical investigation is any experiment involving a test article (a drug, device, or biologic) and one or more human subjects, where the results are intended for submission to the FDA as part of an application for a research or marketing permit, or where the study requires prior FDA submission. The FDA considers the terms “research,” “clinical research,” “clinical study,” “study,” and “clinical investigation” synonymous within its regulations.
The key difference is scope. The HHS definition in 45 CFR 46 is broad and applies to any systematic investigation designed to produce generalizable knowledge, whether or not it involves a regulated product. The FDA definition is narrower, centered on experiments with test articles whose results will be reviewed by the agency. A behavioral study with no drug or device involvement would fall under the HHS definition but not the FDA’s. Many federally funded studies involving drugs or devices must satisfy both frameworks simultaneously.
Quality Improvement vs. Research
The distinction that trips people up most often is the line between quality improvement (QI) and research. A QI project typically aims to improve a specific process or outcome within a particular institution or practice setting. It becomes research when the design shifts toward producing findings that apply beyond that local context.
Consider a hospital tracking surgical infection rates to improve its own protocols. That’s quality improvement. But if the same hospital designs a structured comparison of two sterilization techniques with the goal of publishing findings that could change practice at other hospitals, the project now has both a systematic methodology and an intent to produce generalizable knowledge. It meets the regulatory definition of research.
The practical takeaway is that the determination isn’t about the topic or the methods alone. It’s about whether the activity, as designed, combines a systematic approach with the goal of producing knowledge that extends beyond the specific individuals or setting being studied. When both elements are present, the activity is research under federal regulations and falls within the scope of IRB oversight.