Resident projects get derailed for one stupid reason over and over: people start collecting data before figuring out whether the work is quality improvement or human subjects research.
That mistake is avoidable.
Here’s the practical difference. IRB approval is the formal pathway for research involving human subjects. QI determination is the institutional decision that your project is meant to improve local care, workflow, or systems rather than create broadly applicable research findings. Those are not interchangeable. And no, calling something “QI” in your slide deck doesn’t make it QI.
The real goal is simple: before you pull charts, send surveys, or change a protocol, you need to know which lane you’re in. Residents need this. Faculty need this. Program leaders definitely need this, because I’ve seen more than one otherwise solid resident presentation get frozen when someone asks, “Wait, did this go through IRB?”
This article gives you the fastest way to sort it out, what usually fits where, and what to do when the answer isn’t obvious.
This article is for education only and isn't legal advice or institutional compliance advice. IRB and QI rules vary by institution, so your local IRB office, compliance team, or QI office gets the final say.
Core Definitions: What IRB Approval Means vs What a QI Determination Means
Start with the clean version.
IRB approval means your institution’s Institutional Review Board has reviewed a project as human subjects research. That review exists to protect people. It looks at risk, privacy, consent, data security, subject selection, and whether the project follows federal regulations and local policy. Depending on the project, the IRB may approve it, require changes, determine it’s exempt, or say it doesn’t qualify as human subjects research at all. The key point: the IRB is the formal oversight body for research.
QI determination means the institution has decided your project is primarily about improving care locally. Think reducing catheter-associated infections on one unit, improving discharge med reconciliation, or fixing a clinic no-show workflow. The purpose is operational improvement, not answering a broad scientific question for external application.
Here’s where people get sloppy: QI determination is not the same as IRB exemption. Exemption is still an IRB-category decision at many institutions. QI determination is a different lane entirely. Also, many hospitals still want QI projects documented through a registration tool, committee review, privacy screen, or departmental approval. “It’s just QI” is not a free pass.
If you remember one sentence, make it this one: IRB is for research oversight; QI determination is for local improvement work that your institution agrees is not research.
How to Decide: The Practical Questions That Determine the Correct Path
Don’t start with jargon. Start with questions.
1) What is the primary intent?
This is the big one. Are you trying to fix a local problem on your service, clinic, or hospital? That leans QI. Are you testing a hypothesis or trying to produce findings that apply beyond your institution? That leans research.
Intent matters more than what you call it.
2) Are you changing care in real time?
If you’re redesigning workflow, updating order sets, or improving a handoff process to make your unit work better now, that’s classic QI territory. If you’re introducing an intervention mainly to study its effects, especially outside routine operations, that starts smelling like research.
3) Are patients exposed to additional risk beyond standard care?
More than minimal added risk, nonstandard procedures, extra blood draws, altered treatment pathways, or vulnerable populations should push you toward IRB review fast. This is where residents get burned by optimism. “It’s low risk” is not a determination. It’s a guess.
4) What data are you using?
Project type matters, but not by itself.
- Retrospective chart review can be QI or research.
- Prospective intervention can be QI or research.
- Survey work can be QI or research.
- Process redesign is often QI, but not always.
A chart review of antibiotic timing to improve your ED sepsis workflow may be QI. A chart review designed to answer a publishable comparative question about treatment outcomes across populations may be research. Same data source. Different purpose.
5) Will the results be shared outside the institution?
Presentation or publication alone does not automatically make a project research. That myth needs to die. I’ve seen excellent QI work presented at regional and national meetings. The issue is whether you set out to generate generalizable knowledge. Still, dissemination plans matter, and many IRB offices will ask about them because they help reveal your true intent.
My rule: if your abstract reads like a local fix, that supports QI. If it reads like a study question first and an operational problem second, expect IRB scrutiny.
Common Resident Project Scenarios: Which Route Usually Fits
Here’s the quick practical guide residents actually need.
Retrospective chart audit
Usually QI or exempt-review territory, depending on purpose. If you’re auditing DVT prophylaxis compliance on your inpatient service to improve adherence, that often fits QI. If you’re asking a broader research question about predictors or outcomes and planning formal analysis for generalizable conclusions, IRB is more likely.
Morbidity and mortality follow-up project
Usually QI if you’re closing the loop on a system issue found in M&M — delayed escalation, handoff failure, sepsis bundle misses. That’s bread-and-butter improvement work. But if you expand it into a structured study of adverse events with broader aims, get IRB input.
Patient satisfaction or staff surveys
Often QI with review, especially if the survey is meant to improve one clinic or service line. But survey projects turn into research fast when they become hypothesis-driven, target sensitive topics, or include vulnerable groups. Anonymous doesn’t automatically mean harmless.
Protocol changes or workflow redesign
Usually QI determination if you’re implementing standard improvement methods to improve local care — new discharge checklist, revised triage algorithm, faster consult process. But if you randomize units, compare competing interventions, or test something nonstandard, IRB likely belongs in the conversation.
Educational interventions
This one is a classic gray zone. A residency curriculum change meant to improve your own program may be QI or educational operations review. A structured study comparing resident performance across intervention groups to draw broader educational conclusions may need IRB. I’ve seen people get this wrong a lot.
Borderline cases that should make you stop and ask
- Randomization
- New or nonstandard interventions
- Extra procedures or data collection
- Vulnerable populations
- Multi-site design
- External collaborators
- Hypothesis-first framing
- Plans that outgrow the original local problem
If your project hits even two or three of those, don’t freestyle it. Escalate it to the IRB office, QI office, or your mentor.
What to Submit, Who Decides, and How to Avoid Delays
Here’s the workflow that saves time.
First, write a one-page project summary. If you can’t explain the project clearly in one page, you’re not ready to submit it anywhere. Include:
- Project title
- Problem being addressed
- Primary aim
- Why you’re doing it
- Methods
- Data source
- Who is affected
- Any change to patient care or workflow
- Privacy/data handling plan
- Consent plan, if applicable
- Dissemination plans
Then send it through the right screen.
Who usually decides
- IRB office for research review, exemption determinations, or human-subjects questions
- QI office or QI committee for improvement determinations
- Privacy/compliance staff if PHI, data sharing, or security issues are involved
- Department research leadership if your program has pre-review
- Faculty mentor for the first reality check
And yes, your mentor matters. A good mentor can smell trouble in three sentences. A bad one says, “Just collect the data and we’ll figure it out later.” That is terrible advice.
How to avoid delays
- Submit early. Not a week before abstract deadline.
- Use the institutional screening tool. Most places have one.
- State the primary intent plainly. Don’t muddy it with buzzwords.
- Be honest about publication plans.
- Document the determination. Email, letter, portal confirmation — something you can show.
- Do not start data collection first. Ever. Not even “just a few charts.”
That last one matters. If you collect data before approval or determination, you may make the project unusable. I’ve seen residents lose months that way.
Special Situations Residents Ask About Most
A few edge cases come up constantly.
Multi-site projects
The moment you involve more than one institution, things get more complicated. Fast. Even if the idea started as local QI, multi-site work often raises research and data-sharing issues. Assume you need formal review somewhere and ask early.
Resident-led or student-led projects
Being trainee-led changes nothing about the oversight requirement. A resident project can absolutely be human subjects research. “It’s just for residency” is not a valid category.
De-identified data
Helpful, but not magic. If the institution gives you a truly de-identified dataset and the project doesn’t involve identifiable human subjects data, the analysis may fall outside some IRB requirements. But local policy still controls the determination. Don’t self-certify.
Quality dashboards and registry data
These often support QI, especially if you’re using existing operational metrics to improve local performance. But if you’re mining registry data to answer a broader scientific question, expect IRB involvement.
Conference presentations
No, presenting at a conference does not automatically require IRB approval. But conference submission forms often ask whether the project had IRB review or QI determination. You want a real answer there, not a panicked email chain at midnight.
Projects that evolve
This is the trap. A project can start as QI and later become research. Maybe the scope expands. Maybe you add comparison groups. Maybe you shift from “fix our readmission workflow” to “test a hypothesis about discharge interventions.” When that happens, stop and get a new determination. Don’t keep going under the old label.
Local policy wins. Every time. If your institution has an official pathway, use it.
Conclusion: The Best Rule of Thumb for Resident Projects
Here’s the rule I trust: if the project is designed to create generalizable knowledge or involves human subjects research, go to the IRB. If it’s designed to improve local care or operations within your institution, get a QI determination and document it.
Simple. Not always easy, but simple.
Ask early. Write the project clearly. Let the institution make the call. Then stay inside the lane you were given.
That protects patients, protects you, and protects the project from becoming a useless mess after months of work.
