The boundary between quality improvement (QI) projects and clinical research requiring IRB oversight remains nebulous with finalization of the Common Rule on Jan. 18, 2017. There was some attempt to address this situation in the proposed new rule, but ultimately the solutions were deemed problematic and the issue was left unresolved, explains Joshua Rolnick, MD, JD, a clinical scholar in the National Clinician Scholars Program at the University of Pennsylvania.

Among the proposed changes in the Notice of Proposed Rulemaking (NPRM) were an exclusion for QI activities and interventions designed to change an accepted practice. The exclusion reflected a desire to create a new framework that addressed the need for QI intervention in a “learning healthcare system.” As noted, the final rule discarded the proposal, and Rolnick and colleagues agreed with that particular move in a recent analysis. However, the overall problem remains.

“We believe that HHS was correct to reject in the final rule the proposed QI exclusion in the NPRM, which provided a flawed attempt to distinguish QI from research,” they concluded.1 “However, the final rule carries forward the problems of the original Common Rule for oversight of QI activities. First, uncertainty regarding the need for IRB review discourages careful efforts to understand the impact of individual QI efforts, exerting a chilling effect on the conduct and publication of QI evaluation, when ethical oversight of QI should result in incentives to improve healthcare delivery. Second, reliance on the Common Rule creates an inverse relationship between methodological rigor and oversight: systematic evaluation of QI interventions receives the administrative oversight of the IRB, while less systematic QI efforts are usually subject to little or no ethical review.”

This conclusion would certainly suggest that hospitals may be tempted to dilute the rigor of a QI project rather than risk facing the scrutiny of IRB review.

“I think that temptation does exist,” Rolnick says. “To be fair, there are many excellent QI practitioners using innovative methods to evaluate interventions in a rigorous fashion without employing the traditional techniques of clinical research. And sometimes practitioners will decide to use a technique like randomization, fold the evaluation under the umbrella of clinical research, and go through the usual IRB channels. However, I think that uncertainty about IRB status and the implications of classifying an evaluation as clinical research does discourage the use of techniques like randomization when they might be appropriate.”

Generally speaking, a hospital trying to improve compliance with hand hygiene or implement a sepsis prevention protocol as QI projects might approach the realm of clinical research if, for example, patients who receive intervention and those who do not are randomized.

“The use of randomization is often treated as an indication that the evaluation should be considered research rather than quality improvement, although specific practices vary by institution,” Rolnick says. “However, randomizing by design should be distinguished from the kind of pseudorandomization that sometimes occurs through institutional practice. Interventions are often used first in certain units for a host of operational reasons, which can sometimes create a sort of pseudorandomization. This is less likely to be regarded as an indicator that the evaluation is clinical research.”

Rolnick and colleagues strongly question whether this randomization threshold between QI and clinical research is appropriate.

“We question if this reasoning makes sense, on either legal or ethical grounds,” says Rolnick, who is both a physician and an attorney. “On legal grounds, the implicit case is that randomization makes the results more generalizable to other medical settings. However, it is not clear that this is usually the case. Randomizations may improve the ability to tease out cause and effect — i.e., internal validity. However, whether randomization improves external validity is a different question. In fact, sometimes randomization involves the use of protocols that depart from routine clinical care, making results more difficult to generalize than those from a purely observational evaluation.”

On ethical grounds, it is unclear why using randomization to determine if a patient receives an intervention has a different ethical status from the myriad nonclinical factors that can influence who receives an intervention, he adds.

“Staffing considerations, for example, may determine whether an intervention is piloted in one ward or three,” he notes.

To address this issue, Rolnick proposes locating ethical oversight of QI outside the IRB system, in a different locus that better connects quality improvement with clinical practice. This approach, at the onset, would recognize the close connection between QI and clinical care. One option is creating a separate “QI-IRB” to address ethical issues specifically raised by QI as distinct from clinical research, he says.

“I think that separate panels for QI would enable hospitals to better embed ethical oversight of QI in operations, thus allowing smoother integration with the other aspects of overseeing a new QI intervention,” he says. “For example, many QI interventions now involve the electronic health record, and many EHR systems offer randomization capabilities. Clinical informatics is well poised to offer oversight. It’s already establishing itself as a distinct area of practice. Many institutions have informatics committees, and clinical informatics has its own fellowship training and board certifications. These committees may have the experience and capacity to take on this oversight function alongside operational oversight of EHR interventions.”

Even with QI-IRBs in place, Rolnick and colleagues warn that the “problems of consent and the inverse relationship between methodological rigor and oversight still apply. A better system would be one operating outside the IRB framework and, thus, designed more flexibly to accommodate the unique features of QI. Is such a system possible without regulatory reform? To a degree, we believe only better federal rules would bring clarity.”

In the interim, clinical decision support (CDS) committees also could help overcome obstacles to improving care and patient outcomes.

“I think clinical decision support committees could offer guidance to practitioners on when IRB oversight is required, how to navigate that process, and how to ensure ethical practice when the IRB is not involved,” he says. “I would also like CDS committees to encourage rigorous evaluation of interventions. They might nudge practitioners to use techniques like randomization when suitable, and help liaise with the IRB to determine the best form of ethical oversight for a given case. Part of that, in my opinion, would include noting when randomization may not necessarily prompt the need for IRB review.”

In any case, it goes without saying that healthcare systems have an ethical obligation to conduct QI efforts in the name of patient safety and improved outcomes. Regardless of IRB oversight, ethical principles apply to QI projects, including the following basic tenets summarized from a study cited by Rolnick.2

  • Social or scientific value: Benefit should justify resources and risks.
  • Scientific validity: Grounded in science and appropriately designed.
  • Fair subject selection: Burdens and benefits should be fairly distributed.
  • Favorable risk-benefit ratio: Protocol should emphasize minimal risk and maximum benefit.
  • Respect for participants: Protect privacy and confidentiality. Inform participants of clinically relevant findings.
  • Informed consent: Patients should give “background consent” to minimal risk QI as part of agreement for treatment.
  • Independent review: Accountability should be a requisite feature of QI. Ethical review should match the level of risk and project benefit.


  1. Rolnick J, Downing NL, Shieh L, et al. Ethical Oversight of Quality Improvement and the Research-QI Boundary: A New Common Rule Changes Little. IRB: Ethics & Human Research 2017;39(3):1-10.
  2. Baily MA, Bottrell M, Lynn J, et al. The Ethics of Using QI Methods to Improve Health Care Quality and Safety. Hastings Center Special Report 2006;36:(4):S1-S40.