03 · What You Need to Know
A Project Can Change Purpose Without Rewriting Its Past
Evaluation and Research Can Coexist
Evaluation and research are not necessarily mutually exclusive. A project can serve a practical operational purpose while also being designed to answer a research question.
U.S. OHRP guidance makes this point explicitly for quality improvement. Many activities undertaken solely to improve local care are not research under the HHS regulatory definition, but some projects are designed both to improve care and to accomplish a research purpose. When the research component involves human subjects and is not exempt, human-subject protection requirements may apply.
The practical question is therefore not whether you must choose one label forever. It is whether the project, as it now exists, contains activities that meet the applicable definition of research.
The Moment of Change Matters
Suppose a program evaluation begins entirely for local decision-making. Six months later, the team develops a new hypothesis and proposes additional data collection specifically to test it.
The important point is not to pretend that the first six months were secretly research. Nor should the team continue indefinitely under the original evaluation classification simply because that was accurate when the project started.
The new research purpose should trigger reassessment before the new research activities begin.
Original evaluation activity
Should be characterized according to its actual purpose and design when it was conducted.
New research activity
Should be assessed prospectively according to its new purpose, procedures, participants, data use, and applicable research-ethics requirements.
What Changes Can Signal That Reassessment Is Needed?
There is no single feature that universally transforms evaluation into research. Several changes, however, should prompt researchers to look closely at the project's status.
Examples include developing a research question that was not part of the original evaluation, collecting additional information specifically to answer that question, introducing an intervention to test its effects, adding research-specific comparison groups, linking datasets for a new analytic purpose, collecting identifiable information not required for the evaluation, or systematically repurposing existing records for a separate research analysis.
The significance of each feature depends on the governing framework. A methodological change is a signal to reassess, not an automatic verdict.
A New Research Question Can Be Enough to Change the Purpose
Imagine a university evaluating whether students are satisfied with a mentoring program. The original question is operational: should the institution modify the program next year?
Later, investigators ask whether a particular mentoring model improves student retention across different demographic groups and design a systematic analysis intended to produce broader evidence about mentoring effectiveness.
The underlying dataset may be similar, but the purpose has changed. That new purpose should be assessed rather than hidden inside the original evaluation label.
Adding Research-Specific Data Collection Is Particularly Important
A project may originally collect only information needed for administration. Researchers later decide they need additional questionnaires, interviews, biological measurements, behavioral tasks, or follow-up contacts solely to answer a research question.
Those activities did not exist in the original evaluation. They should not begin merely because participants or records are already available.
OHRP has similarly distinguished follow-up undertaken as part of ordinary care from follow-up initiated for data collection in a research study. When the added activity serves a research purpose and the project meets the regulatory definition of research, human-subject regulations may apply.
An Untested Intervention Can Move a Project Toward Research
Quality improvement often involves changing practice, so novelty alone is not a universal definition of research. Still, introducing an untested intervention partly to establish scientific evidence about whether it works is a strong reason to reassess the project.
OHRP gives essentially this example: a QI project introducing an untested clinical intervention for purposes that include both improving care and collecting patient outcomes to establish scientific evidence about effectiveness may also constitute nonexempt human-subject research.
Randomization Does Not Automatically Turn Evaluation Into Research
Random assignment often looks research-like, but it should not be used as a one-factor test.
U.S. advisory guidance has noted that a cluster-randomized design can sometimes be used for a genuine QI purpose without the activity necessarily becoming research under the Common Rule. The project's purpose and use of the resulting information remain central.
Conversely, a project does not need randomization to constitute research. Observational and qualitative studies can plainly be research.
Publication Does Not Mark the Exact Moment Evaluation Becomes Research
A team may decide that its evaluation findings deserve publication. That decision alone does not establish that the project has transformed into research.
OHRP expressly states that intent to publish is insufficient to determine whether a QI activity meets the regulatory definition of research. Nonresearch improvement work may be publishable, while research can exist without any intention to publish.
The better question is whether the team is now proposing a research activity, not whether someone has opened a manuscript template.
Existing Evaluation Data May Become Secondary Research Data
One of the most common transitions does not involve collecting anything new. The evaluation ends, and someone realizes that the dataset could answer an interesting research question.
At that point, the proposed analysis may become secondary research using existing data. Identifiability, original permissions, privacy, consent arrangements, authorized access, and applicable exemption criteria may all become relevant.
The fact that the information was legitimately collected for evaluation does not automatically authorize every later research use.
The Original Participants May Not Have Agreed to Research Use
Participants may have provided information because their hospital, school, employer, government agency, or community organization was evaluating a service. They may not have been told that their information would later be used for research.
Whether new consent is necessary depends on the applicable framework, the nature of the data, identifiability, original notices or permissions, available exemptions or waivers, and the proposed research use.
Do not solve this question informally by assuming that “the data belong to the institution.” Institutional custody of records and ethical authorization for secondary research are not necessarily the same thing.
Do Not Quietly Add Research Procedures to an Evaluation
A problematic transition occurs when researchers recognize that the project has developed a research purpose but continue operating under the original evaluation classification because ethics review would delay the work.
Prospective ethics oversight exists so that participant protections can be considered before research procedures occur. Once a new research component is identifiable, the appropriate time for reassessment is before implementing it.
Watch Out
Do not wait until manuscript submission to disclose that an evaluation gradually became a research study. If the project's purpose, procedures, participants, or use of data changes in a way that may create human-participant research, seek reassessment before the new research activity begins.
A Change in Status Does Not Necessarily Mean Full Committee Review
Recognizing a research component does not automatically mean the entire project must undergo full-board review.
The new research may fall outside the applicable human-participant research definition, qualify for exemption, or be eligible for another review pathway depending on its design and governing framework.
OHRP likewise notes that QI activities constituting human-subject research may sometimes qualify for exemption or expedited review rather than convened IRB review.
The first task is accurate classification. The appropriate review pathway comes afterward.
Keep the Operational and Research Components Conceptually Separate
When feasible, describe which activities are being conducted for service delivery or evaluation and which are being performed specifically for research. This can make ethics assessment considerably clearer.
OHRP has applied this logic to activities containing simultaneous quality-improvement and research purposes, treating particular components according to what they are designed to accomplish.
Clear separation also helps explain to participants, data custodians, collaborators, and reviewers why particular information is being collected and under which authority.
Reassessment Is Also Needed When the Research Component Expands
Suppose the REC or IRB approves a limited research use of evaluation data. The team later adds new sites, collects new identifiers, contacts individuals represented in the records, or introduces an intervention.
The original research determination should not be treated as unlimited permission for whatever the project later becomes. Follow institutional procedures for amendments and reassessment when material changes occur.
