QI M I H CARE - Ethics Research Institute · tice of quality improvement in health care. The project, ti-tled “The Ethics of Improving Health Care Quality and Safety,” was funded

MARY ANN BAILY � MELISSA BOTTRELL

JOANNE LYNN � BRUCE JENNINGST H E. . . . . . . . . . . . . . .

HASTINGS. . . . . . . . . . . . . . .

C E N T E R

A HAST INGS CENTER SPEC IAL REPORT

THE ETHICS OF USING

QI METHODS TO

IMPROVE HEALTH CARE

QUALITY AND SAFETY

Rights were not granted to include this image in electronic media.

Please refer to the printed report.

S2 July-August 2006 / HASTINGS CENTER REPORT

George J. Agich, PhDBowling Green State UniversityBowling Green, OH

Mary Ann Baily, PhDThe Hastings CenterGarrison, NY

Paul Batalden, MDDartmouth Medical SchoolHanover, NH

Nancy Berlinger, PhD, MDivThe Hastings CenterGarrison, NY

*Melissa Bottrell, MPH, PhDNational Center for Ethics in Health Care, Veterans Health AdministrationSeattle, WA

*Michael Carome, MDDHHS Office for Human ResearchProtectionsRockville, MD

David Casarett, MD, MACenter for Health Equity Researchand PromotionPhiladelphia VA Medical CenterPhiladelphia, PA

Janet Corrigan, PhDNational Quality ForumWashington, DC

Linda Cronenwett, PhD, RN,FAANUniversity of North Carolina atChapel Hill, School of NursingChapel Hill, NC

Frank Davidoff, MD, MACPInstitute for Healthcare ImprovementWethersfield, CT

Nancy Neveloff Dubler, LLBMontefiore Medical CenterBronx, NY

*Arnold Farley, PhDCenters for Medicare and MedicaidServicesBaltimore, MD

*J. Michael Fitzmaurice, PhDAgency for Healthcare Research andQualityRockville, MD

*Ellen Fox, MDNational Center for Ethics in Health Care, Veterans Health AdministrationWashington, DC

Brent James, MD, MSTATIntermountain Institute for HealthCare Delivery ResearchSalt Lake City, UT

*Stephen F. Jencks, MD, MPHCenters for Medicare and MedicaidServicesBaltimore, MD

Bruce Jennings, MACenter for Humans and NatureNew York, NY

Robert J. Levine, MDYale UniversityNew Haven, CT

Joanne Lynn, MD, MA, MSThe RAND CorporationArlington, VA

Karen Maschke, PhDThe Hastings CenterGarrison, NY

Ethel Mitty, EdD, RNNew York University College of NursingNew York, NY

Robyn Y. Nishimi, PhDNational Quality ForumWashington, DC

Margaret O�Kane, MHANational Committee for Quality AssuranceWashington, DC

Paul Schyve, MDJoint Commission on Accreditationof Healthcare OrganizationsOakbrook Terrace, IL

Theodore Speroff, PhDVanderbilt University Medical CenterNashville, TN

Sean Tunis, MD, MScHealthTechSan Francisco, CA

Matthew K. Wynia, MD, MPHAmerican Medical Association Institute for EthicsChicago, IL

*Note: The authors of this report are re-

sponsible for its content. Statements in

this report should not be construed as

endorsement by the Agency for Health-

care Research and Quality or the U.S De-

partment of Health and Human Services.

The federal government employees list-

ed above participated in the meeting dis-

cussions in an advisory capacity. Howev-

er, in addition to the preceding dis-

claimer, it should be further emphasized

that the statements in this report should

not be construed as endorsement by the

Department of Veterans Affairs or any

other department of the U.S. govern-

ment, nor do they represent the policies

of any federal departments, agencies, or

offiices.

PROJECT PARTICIPANTS Organizational affiliations for identification purposes only; affiliations current as of 2006

This project was funded by theAgency for Healthcare Researchand Quality, grant #1R13HS13369.

S3SPECIAL REPORT / The Ethics of Using QI Methods to Improve Health Care Quality and Safety

PREFACE

Powerful forces of change are at work within theAmerican health care system. The public debateconcerning health care financing and access to in-

surance coverage is intensifying. But below the surface ofthe media and policy debate about cost and access, a qui-eter but perhaps more significant process of change isunder way: the transformation of health care managementand delivery—indeed, health professional work itself—through health care quality improvement.

The innovative, interdisciplinary quality improvement(QI) movement has begun to significantly upgrade deliv-ery of health care in the United States. Taking its cue fromreform approaches in other industries, and driven espe-cially by studies indicating a shockingly widespread inci-dence of medical errors and a striking lack of consistencyin the standard of care patients receive in different facili-ties and from different practitioners, the QI movementhas arrived in health care. Using knowledge gained fromthe disciplines of medicine, nursing, health care manage-ment, and medical and health services research, it at-tempts to mobilize people within the health care system towork together in a systematic way to improve the carethey provide. In this work, discipline-specific knowledgeis combined with experiential learning and discovery tomake improvements.

Ethical issues arise in QI because attempts to improvethe quality of care for some patients may sometimes inad-vertently cause harm, or may benefit some at the expenseof others, or may waste scarce health care resources. Ethi-cal issues also arise because some activities aimed at im-provement have been interpreted as a form of medical re-search in which patients are used as subjects. If this inter-pretation is correct, QI would come under the same com-plex review and regulatory requirements that have beenset up to govern biomedical and other types of research.But is this type of regulation necessary, given what QI in-volves? Is it the most effective and reasonable way to regu-late QI to ensure that it is carried out in an ethical fash-ion? These are important questions, both conceptuallyand practically. Thus far, however, relatively few attemptshave been made to address QI from an ethical perspective,and the interface between research and quality improve-ment has not been adequately explored or defined.

Federal agencies with responsibilities in this area havedisagreed on where the interface between medical researchand QI lies and how it should be handled. (See Box 1 fora particularly dramatic example of such a conflict.) Thestrict ethical rules of oversight, regulation, and patientconsent for human subjects research, including the re-quirement for institutional review board (IRB) approval,have significant implications for the feasibility and cost ofpursuing QI activities. More specifically, the mechanismdeveloped to govern ethical conduct in one importantarea—human subjects research—could have the perverse,if unintended, consequence of interfering directly with anequally important ethical imperative in another area—that is, unceasing efforts by health care professionals tomake clinical care safer and more effective. The currentstate of uncertainty about what is ethically and legally re-quired to safeguard participants in QI activities has al-ready become a disincentive to engage in QI, making itmore difficult to bring about the system transformationurgently needed if health care is to be made better andsafer for patients.

In 2002 The Hastings Center began a project to ad-dress these issues and to investigate more generally theethical and value issues that arise in the theory and prac-tice of quality improvement in health care. The project, ti-tled “The Ethics of Improving Health Care Quality andSafety,” was funded by grant #1R13HS13369 from theAgency for Healthcare Research and Quality (AHRQ).

The Hastings Center project assembled a group of ex-perts from a number of interrelated fields and disciplinesinvolved in health care quality improvement, includingmedicine, nursing, law, social science, health care manage-ment, medical editing and publishing, health policy andregulation, health services research, and bioethics. Theproject’s goals were: (1) to develop an ethical frameworkthat can be applied to ethical issues arising in quality im-provement practice, (2) to make practical policy recom-mendations for quality improvement oversight, and (3) tohelp promote a constructive dialogue both within andoutside the quality improvement community on pertinentethical issues.

Over a two-year period, the project group met fourtimes for two-day working sessions. Project members and


other invited guests presented and debated thefindings of their own research and drafts of com-missioned papers focusing on ethical issues inquality improvement. Those papers will be pub-lished in an edited volume in 2007. In addition toour deliberations with project participants, wehave also made presentations at professional andacademic meetings and communicated electroni-cally with a larger group interested in health caresystem improvement by setting up a moderated e-mail listserv. Following the publication of this re-port, The Hastings Center will continue to serveas a resource in the process of translating the con-clusions and recommendations of the project intopractical policies through the listserv, presenta-tions at professional and academic meetings, pro-duction of additional written materials, and direct

contacts with persons and organizations interestedin the ethics of QI.

This report presents the analysis, findings, andconclusions that emerged from the deliberationsof the project participants and staff research con-ducted for this project. The authors are Mary AnnBaily, Melissa Bottrell, Joanne Lynn, and BruceJennings. This report is not a consensus docu-ment in the sense that each project participantagrees with it in all details, but we have done ourbest to present an accurate and faithful reflectionof the thinking of the group as a whole, and thisanalysis certainly would not have been possiblewithout the benefit of their insights and expertise.

—Mary Ann Baily, Melissa Bottrell,Joanne Lynn, Bruce Jennings

In October 2000, a nephrologist coauthoredan article about a project to improve the dialysiscare delivered to patients in Medicare’s EndStage Renal Disease (ESRD) program.1 Sometime after the article appeared, the university atwhich he held a faculty appointment notifiedhim that an audit of faculty publications hadidentified this project as a quality improvementeffort that met the definition of human subjectsresearch but had not undergone IRB review. Thenephrologist responded that he had participatedin the project as the chair of the local ESRDNetwork’s Medical Review Board, that theESRD Network conducted the project undercontract to the Center for Medicare and Medi-caid Services (CMS), and that the CMS scientif-ic officer overseeing the project had told himthat, as a CMS-directed quality improvementproject, it was not subject to oversight by theuniversity’s IRB. Later, CMS reaffirmed thisopinion in a letter to the university. The univer-sity submitted the dispute to the Department ofHealth and Human Service’s Office for HumanResearch Protections (OHRP) for review.

OHRP responded that the design of the qualityimprovement project met the definition ofhuman subjects research and that the CMS de-termination that the project was exempt fromIRB review “was not made in collaboration withOHRP.” OHRP and CMS have had discussionsand exchanged memoranda as to whether thisand other quality improvement activities meetthe regulatory definition of human subjects re-search and require IRB review; however, these is-sues have not yet been resolved. This issue hascaused concern among other volunteer physi-cians in the eighteen ESRD Networks about theimplications for ongoing quality improvementprojects and the possible impact if their conductis found to violate the regulations for protectinghuman research subjects.

1. P.M. Palevsky et al., “Improving Compliance withthe Dialysis Prescription as a Strategy to Increase the De-livered Dose of Hemodialysis: An ESRD Network 4Quality Improvement Project,” Advances in Renal Re-placement Therapy 7, no. 4, supplement 1 (2000): S21-S30.

1. The Pittsburgh ESRD Case

uality improvement in health care takes manyforms, ranging from changes in financing toreforms in professional education to invest-ments in new facilities and equipment. In this

report, we are concerned with the form of improvementthat occurs through clinical and managerial innovationsand adaptations in the delivery of care. These changeshave always been an integral part of health care delivery,but in the past they were often introduced informally andidiosyncratically, without careful attention to all of theireffects. In recent years, people in health care have begunto use many new formal, explicit methods, some of whichwere developed in other industries, to make the process ofcontinual adjustment more self-reflective and systematic,and thus increase the likelihood that it produces positivechange. In other settings, this approach to improvingquality is often referred to as “QI.”

When we use the term “QI” in this report, we meansystematic, data-guided activities designed to bring aboutimmediate, positive changes in the delivery of health carein particular settings. While QI uses a wide variety ofmethods, they all involve deliberate actions to improvecare, guided by data reflecting the effects. Depending onthe activity, QI can look like a type of practical problemsolving, an evidence-based management style, or the ap-plication of a theory-driven science of how to bring aboutsystem change. Introducing QI methods often means en-couraging people in the clinical care setting to use theirdaily experience to identify promising ways to improvecare, implement changes on a small scale, collect data onthe effects of those changes, and assess the results. Thegoal is to find interventions that work well, implementthem more broadly, and thereby improve clinical practice.Alternatively, a QI activity might begin with a review ofaggregate data at the patient, provider, clinical unit, or or-ganizational level to identify a clinical or managementchange that can be expected to improve care. The changeis made, the effects are monitored, and conclusions aredrawn about whether the change should be made perma-nent. QI can also mean collecting data from multiple or-ganizations, analyzing it to understand what drives posi-tive change, and using the results to design and imple-ment a strategy to achieve a specific improvement acrossorganizations. At its heart, QI is a form of experientiallearning and discovery.1

QI work is data-guided, usually involves human par-ticipants, and sometimes uses methods that are also usedin medical research. Thus, it is not surprising that theissue of its relationship to federal regulations governing re-search with humans has arisen. These regulations were putin place to ensure that federally funded research projects

meet accepted ethical standards for protection of humanresearch subjects. The regulations that are relevant to thisreport are codified at Title 45 CFR 46 as Department ofHealth and Human Services, Protection of Human Sub-jects. The core requirements (Subpart A) are that individ-uals be selected equitably to participate in research; thatresearch subjects give full, voluntary written consent; andthat an IRB review proposed research and approve it onlyif risks to subjects have been minimized and are reason-able in relation to anticipated benefits.2 There are addi-tional requirements (Subparts B, C, and D) for researchthat involves especially vulnerable subjects, such as preg-nant women, prisoners, and children. The DHHS regula-tions do not cover non-federally funded research, but or-ganizations engaged in research are encouraged to provideformal assurances to the government that all their humansubject research will comply with the DHHS regulationswhatever the funding source. Also, although Subpart Aexplicitly exempts some categories of human subject re-

search from the regulations, the DHHS Office forHuman Research Protections (OHRP) has issued guid-ance stating that researchers should not be allowed to de-cide on their own that their projects are exempt. Many in-stitutions require determination of exempt status to bemade by an IRB. In practice, a substantial proportion ofboth federally funded and nonfederally funded humansubject research, especially biomedical research, must besubmitted to an IRB. (For a brief summary of the DHHSregulations, see Box 2.)

Discussion of ethical issues raised by QI has tended tocenter around the legal question, “Is this QI project‘human subjects research’ as defined by federal regulationsand therefore subject to IRB review and the regulatory re-quirements for informed consent?’’ The case outlined inBox 1 is but one example of many in an ongoing contro-versy in the literature, in IRBs, and in organizations doingQI about whether QI is research and therefore covered bythe regulations.

Framing the issue in this way may not be the best wayto proceed, however. The goal is to ensure that QI activi-ties are carried out in an ethical manner and that theirhuman participants are treated appropriately, however theactivities are planned and structured and whatever the ac-


Q

� � �

Although systematic empirical information on

the quality of care is limited�which is itself

surprising�what is available is disturbing.


tivities are called. Focusing on IRB review and the defini-tion of research may implicitly suggest that IRB reviewwould be adequate and that it is the only way to protectQI participants. This is an overly narrow perspective. Inthis report, we explore the similarities and differences be-tween research and QI, but rather than focus entirely on“Is QI research?” and “IRB review, yes or no?” we stepback to look at the bigger picture. We ask: “What is theplace of QI in the operation of the health care system?”“What makes a QI activity ethical?” and “What institu-tional arrangements should be in place to ensure that spe-cific QI activities meet ethical requirements?” Only afterthat do we return to the question of how QI activitiesshould relate to IRB review and other procedures for theprotection of human subjects in research.

Organization of the Report

Section I explains why quality improvement is impor-tant in health care and discusses the role of QI meth-

ods in the management of health care delivery. We thenconsider the underlying ethical question of who has re-

sponsibility for improving the quality of care. We con-clude that engaging in quality improvement is not purelydiscretionary; health professionals, managers, delivery or-ganizations, patients, and government all have an ethicalresponsibility to cooperate with one another to improvethe quality of care. Section II compares and contrasts QIand research in terms of each activity’s goals, methods,role in the health care system, and impact on human par-ticipants. We use the principles of research ethics that un-derlie the current system for protecting human researchsubjects as a base from which to explore and define the re-quirements for ethical conduct of QI activities. We thenconsider the use of IRBs for ethical oversight of the con-duct of QI and conclude that, for a variety of reasons,IRBs as currently constituted are not appropriate for thispurpose. Section III discusses the institutional arrange-ments that should be in place to ensure that QI activitiesmeet ethical requirements. Section IV addresses imple-mentation, and Section V briefly summarizes the conclu-sions and policy recommendations of the report.

Department of Health and Human Services policysets out a complex set of provisions for the protection ofhuman research subjects. It:

� extends to research funded by the Department ofHealth and Human Services and to nonfederally fund-ed research conducted at or in conjunction with an in-stitution that agrees through its federal-wide assuranceto extend the regulations to all research regardless offunding source. Six categories of research activities areexempt from the regulations (45 CFR 46.101(b)(1)-(6))

� requires prior review and approval of nonexempt re-search by a duly constituted institutional review board(IRB), and continuing IRB review of ongoing research(45 CFR 46.103)

� permits expedited review (conducted by IRB chair-person or by one or more experienced reviewers) whencertain criteria are met (45 CFR 46.110 and Guidanceon the Use of Expedited Review Procedures)

� requires seven items to be satisfied for IRB approvalof research: risks to subjects are minimized; risks arereasonable in relation to anticipated benefits; subjectsare selected equitably; informed consent will be ob-tained from prospective subjects or their legally autho-rized representatives; there will be written documenta-tion of informed consent; the trial will receive safety

monitoring, when appropriate; provisions are made forsubject privacy and confidentiality of data, when ap-propriate (45 CFR 46.111)

� requires IRBs to ensure that additional safeguardsare included in a study when participants will likely bevulnerable to coercion or undue influence (45 CFR46.111)

� prohibits institutional officials from approving re-search that has not received IRB approval (45 CFR46.112)

� permits IRBs to suspend or terminate research ap-proval when a study is not conducted in accordancewith IRB requirements or is associated with unexpect-ed serious harm to participants (45 CFR 46.113)

� requires that specific information be provided to re-search participants to satisfy the general requirementsfor informed consent (45 CFR 26.116)

� permits IRBs to approve a consent procedure thatomits or alters some or all of the elements of informedconsent when certain criteria are met (45 CFR 46.116)

� requires additional protections for pregnant women,human fetuses and neonates (45 CFR 46.201), prison-ers (45 CFR 46.301), and children (45 CFR 46.401)

2. Summary of DHHS Policy for the Protection of Human Subjects (Subparts A-D)

Many Americans have a rosy picture of the quali-ty of their health care system. They believe thatit provides a generally high standard of care that

is continuously updated as new research findings comeout, so that it remains the best in the world. They recog-nize, of course, that there are some access problems (espe-cially for the uninsured and underinsured) and somebelow-par facilities and professionals—no health care sys-tem is perfect—but overall, the sense is that the quality ofcare is excellent. Unfortunately, this view is too optimistic.Recent work by individual scholars and by organizationssuch as the Institute of Medicine (IOM) and the Agencyfor Healthcare Research and Quality (AHRQ) has provid-ed detailed, compelling evidence of serious problems withthe quality of American health care.3

Since quality is a complex, multifaceted concept, as-sessing just how well the American health care system isdoing is not easy. It requires the development of a concep-tual framework for understanding quality and the transla-tion of that framework into practical measures that can beapplied in specific contexts. The framework developed bythe Institute of Medicine in several influential reports de-fines quality as “the degree to which health services for in-dividuals and populations increase the likelihood of de-sired health outcomes and are consistent with current pro-fessional knowledge.”4 The core dimensions of quality areidentified as safety, effectiveness, patient-centeredness, andtimeliness.5 Two additional aspects of care that cut acrossthese dimensions are equity and efficiency.6 (See Box 3 fordefinitions of these six quality characteristics.)

The status of efficiency in the quality framework iscontroversial. In Crossing the Quality Chasm, the Instituteof Medicine Committee on the Quality of Health Careconsiders it a dimension of quality, but another IOMcommittee, charged with translating the IOM’s qualityframework into practical measures, does not, arguing thatefficiency is “an important goal of the health care systemthat is related to, but conceptually different from qualityof care.”7 In this report, we work from the position thatincreasing efficiency is often a goal of quality improve-ment activities, since the developers of QI methods havetraditionally considered efficiency an appropriate QI con-cern, and activities involving changes designed to producethe same quality of care with fewer resources constitute animportant QI category.8

Although systematic empirical information on thequality of care is limited—which is itself surprising, giventhe size and importance of the health care sector—what isavailable is disturbing. In a comprehensive review of theliterature on quality of care from 1987 to 1998, Schusterand colleagues found substantial evidence of overuse, un-deruse, and misuse of care in the United States. Serious

quality problems were seen in preventive, acute, andchronic care, in the care provided in different kinds ofhealth care facilities, in care paid for by different kinds ofhealth insurance, in the care received by different agegroups, and in rural and urban settings.9 The studies theyreviewed document unnecessary surgery; inappropriateuse of medications; failure to perform standard screeningtests; inadequately controlled asthma, diabetes, and hyper-tension; and significant departures from recommendedlevels of care for patients with cardiovascular disease. Amore recent study examined the care received by a largesample of adults in twelve cities and documented a varietyof quality problems, finding that overall, participants re-


I. The Place of Quality Improvement in Health Care

Quality is defined in the Institute of Medicine re-port, Crossing the Quality Chasm, as the degree towhich health services for individuals and populationsincrease the likelihood of desired health outcomes andare consistent with current professional knowledge.1

The report describes the following characteristics ofquality:

Safety: avoiding injuries to patients from the carethat is intended to help them

Effectiveness: providing services based on scientificknowledge to all who could benefit, and refrainingfrom providing services to those not likely to benefit(avoiding underuse and overuse, respectively)

Patient-centeredness: providing care that is respect-ful of and responsive to individual patient preferences,needs, and values, and ensuring that patient valuesguide all clinical decisions

Timeliness: reducing waits and sometimes harmfuldelays for both those who receive and those who givecare

Equity: providing care that does not vary in qualitybecause of personal characteristics such as gender, eth-nicity, geographic location, and socioeconomic status

Efficiency: avoiding waste, including waste ofequipment, supplies, ideas, and energy

1. Institute of Medicine, Crossing the Quality Chasm (Wash-ington, D.C.: National Academy Press, 2001).

3. Quality Framework

ceived only a little over half of recommended care.10 Thereis also ample evidence of equity problems: the quality ofcare varies in ways systematically correlated with ethnicity,geographic location, and socioeconomic status.11

To help remedy the lack of information and make iteasier to monitor quality over time, the IOM and AHRQhave developed a comprehensive set of quality measures,and in 2003, AHRQ published the first in a projected se-ries of periodic national quality reports. The report pro-vided a baseline of quality information, including evi-dence on the tremendous variation in quality of careacross geographic areas and across individual care set-tings.12 The 2004 and 2005 National Healthcare QualityReports found improvement in some measures, but deteri-oration on others, and concluded that the gap between thebest possible care and actual care remained large.13 AHRQalso publishes a companion series, the annual NationalHealthcare Disparities Report. The 2003, 2004, and 2005reports provided baseline and follow-up information onthe pervasive racial, ethnic, and socioeconomic disparitiesin health care access and quality in the United States.14 Fi-nally, for an international comparison, in a recent survey,one-third of patients with health problems in the UnitedStates reported experiencing medical, medication, or testerrors, the highest rate for the six countries included in thesurvey (the others were Australia, Canada, Germany, NewZealand, and the United Kingdom).15

The more we learn about current quality, the clearer itis that safe, effective, patient-centered, timely, equitable,and efficient health care will not happen automatically.Rather, it requires and will continue to require systematicand self-conscious management of health care delivery ex-pressly directed at improving care. This means not onlythe implementation of specific changes, but also the trans-formation of the culture of health care delivery into a cul-ture that is committed to continuous quality improve-ment.

QI Activities: Part of Normal Health CareOperations

Efforts to improve quality take place in a health care en-vironment that is continually experiencing change.

Some of this change occurs in response to fluctuations inmarket conditions as, for example, the age distribution ofthe population evolves, new diseases emerge, or careers inhealth care are considered more or less attractive. Some ofthe change occurs in response to scientific progress as, forexample, a known disease is better understood, a new drugis developed, or new diagnostic technology becomes avail-able. Managing this constant change is a core responsibil-ity of the people and the organizations that make up thehealth care system. The specific patterns and processes ofcare in one clinical setting are the product of thousands ofsmall and large decisions about handling change that

could have been made somewhat differently in that set-ting, and that have been made differently in other settings.

Within this universe of change-related activity is a sub-set of deliberate efforts to make positive changes in the deliv-ery of clinical care. Many of these “designed changes” comeabout by “just doing it”; someone decides that a change inclinical practice or organizational arrangements seems likea good idea and the change is made. Such innovation andadaptation is an intrinsic part of clinical and managerialpractice. Health care practitioners must tailor a generalstandard of practice to individual patients, relying on theirclinical knowledge, their unique knowledge of each pa-tient, and the characteristics and capabilities of the localcontext in which care is being delivered. Managers mustexercise judgment about how resources in a particular set-ting should be organized to carry out treatment plans forthe patients being served.

When this innovation and adaptation is undertaken ina systematic, data-guided way, it becomes what we are call-ing QI. The category of QI includes a wide variety of ac-tivities and cannot be defined by any one method or pro-cedure.

QI is closely related to clinical practice, and in fact,much of QI is simply good clinical care combined withsystematic, experiential learning. Individual practitionersare constantly learning by doing and taking steps to im-prove their own practice. Physicians may develop personaltemplates for entering information in medical records andfor tracking the lab tests they have ordered; surgeons maywork on their surgical techniques to reduce the time pa-tients spend anesthetized; nurses may refine their skin caremethods to reduce the incidence of bedsores.

Very quickly, however, the effort of an individual clini-cian to improve his or her practice becomes an issue forothers working in the same environment. In health carefacilities today, most clinical care is delivered to patientson a team basis, and the ability of the team to deliver goodcare depends on the characteristics of the administrativeinfrastructure and procedures that are in place. In otherwords, both patients and providers are part of systems ofcare. A key insight of those promoting QI methods is thatquality and safety are largely systems issues, and under-standing the interdependencies and relationships withinsystems is at the heart of QI work.16 In systems, manage-ment practices may be as significant to good outcomes asthe clinical practices of individual care providers, so learn-ing by doing in management is also important. Many QIactivities involve clinicians cooperating with each otherand with management and support staff around improve-ments in scheduling procedures, encounter forms, med-ication handling, patient flow within and across clinicaldepartments, communication within patient care teams,record-keeping, and other administrative procedures.

QI methods include a variety of tools for motivatingand structuring the cooperation, changing the process orsystem, monitoring what happens, and evaluating the


change. In addition to local “learning by doing,” QI in-volves deliberate applications in local settings of newknowledge developed elsewhere. The goal of many QI ac-tivities is to introduce a new clinical practice based on anadvance in medical science or a new administrative prac-tice based on managerial experience or social science re-search. Moreover, QI activities often use the science of or-ganizational change to design strategies for changing be-havior in the local setting—for example, to get adherenceto a new practice guideline or a new patient safety proto-col.

In sum, QI is a particular form of the clinical and man-agerial innovation and adaptation that has always been anintegral part of health care operations. To the process ofchange, it adds systematic, self-reflective design, monitor-ing, data analysis, and assessment of the effects of changesto ensure that they are in fact beneficial, and it feeds thisinformation back into the delivery setting. It uses newknowledge developed in the medical and social sciences tohelp identify positive changes and design practical strate-gies to implement them in the local setting. In the process,the QI practitioner implicitly or explicitly uses each at-tempt to improve care to deepen his or her understandingof the content and process of bringing about positivechange in that setting. Ideally, over time, the successful useof QI methods helps to transform the culture of an orga-nization into one in which everyone is committed to con-tinuous quality improvement and has the skills to partici-pate in it. Thus, QI has the potential to make a substantialcontribution to solving the quality problems in Americanhealth care.

The Ethics of Improving Health Care Quality

Before we can discuss the ethical issues that arise whenQI methods are used in health care, we must address

the underlying question of who is responsible for improv-ing the quality of care. Health care is of unusual conse-quence because of the role it plays in relieving suffering,preventing premature death, restoring function, increasingopportunity, providing vital information about an individ-ual’s condition, and giving evidence of a community’s mu-tual empathy and compassion. As a result, health care ac-cess, quality, and cost have always been matters of societalethical concern.17

Since antiquity and across different cultures, healershave been revered as persons with special knowledge of themysteries of life and death, and the relationship betweenpatient and physician has held deep moral and religioussignificance. Physicians, nurses, and other health profes-sionals have been understood to have a special ethical re-sponsibility to serve the interests of their patients, includ-ing the responsibility to maintain and continually strive toimprove the quality of the care they provide.

Over time, the concept of medicine as a profession hastaken hold and science has replaced religion as the basis of

medicine. With scientific progress has come better under-standing of bodily processes and more effective treatmentsfor illness. As these treatments have been embodied in newtechnology, the provision of health care has become amore complex process, involving an array of organizationsand specialized professionals working together in complexsystems of care.

Nevertheless, even as medical practice has changed, themoral and ethical aspects of medicine have remained cen-tral to the understanding of what it means to be a memberof a health profession.18 Since health outcomes are inher-ently uncertain and vary under different treatment op-tions, health care decisions are complicated. Patients findit difficult to understand their choices, especially whenuncertainty and the risk of adverse events are involved,and they must rely on physicians and other health careproviders for help. Given the importance of health to well-being, patients are vulnerable to exploitation in this de-pendent role.

Therefore, society continues to expect physicians to ac-cept a special ethical responsibility to serve each patient’sinterest. As other health professionals and health care or-

ganizations have played increasingly important roles incare delivery, analogous ethical responsibilities to serve thepatient’s interest have been recognized for them.19

The understanding of these ethical responsibilities hassharpened as health care has evolved from a simple pa-tient-physician interaction to today’s complex interactionsbetween patients and caregiving teams and organizations.It is more evident now than in years past that the ethicalimperative to “serve the patient’s interests” does not meanignoring the interests of every patient except the one pre-sent at the moment. Since health care organizations servegroups of patients, clearly they must manage the processof care with policies and procedures that balance the needsof all the patients they serve. Characterizing this as “wetreat populations now, not patients” or “the patient’s inter-est must now be subordinated to the population’s interest”is misleading, however. Physicians have always had multi-ple patients in their practices, and hospital-based nurseshave always managed care for multiple patients simultane-ously. They have thus had to consider the interests of alltheir patients in allocating personal resources of time andenergy to their practices. Moreover, it is in an individualpatient’s interest to have health care providers follow or-


Improvements in the quality of health care

will not happen automatically. Systematic and

self-conscious management of health care

delivery is needed.


derly processes that function well for people in differentsituations. After all, one could be in one of those situa-tions oneself sometime.

It is abundantly clear, therefore, that health profession-als and health care organizations have an ethical responsi-bility to serve the interests of patients, and patients cer-tainly have an interest in the quality of health care. Butwhat is the nature of this interest? What level of quality dopatients want? What level are they entitled to? In the caseof an ordinary commodity, these questions would usuallybe answered through a market process. Buyer preferencesand budgets would interact with seller production costs toproduce the products that buyers wanted and could af-ford—and there would probably be an array of qualitylevels.

But given health care’s complexity, people cannot easi-ly make informed assessments of quality and develop sen-sible and stable preferences for their health care. More-over, even if they could, they would have difficulty gettingthem implemented, because often they are not paying di-rectly for their care. Since future health status is uncertain,people need insurance to make sure they will be able to af-ford the care they want if they get sick. In addition, likemost societies, the United States recognizes a societal eth-ical obligation to provide at least some basic health carewithout regard to ability to pay.20 Third-party paymentadds a social dimension to decision-making. When re-sources are pooled, whether through private insurance orthrough publicly funded social insurance, collective deci-sions must be made about how the pooled resourcesshould be spent. These decisions play a key role in deter-mining both the quality and the quantity of care to whichpatients are entitled.

Within the framework of these collective decisions,health professionals and health care managers have an eth-ical responsibility to meet socially determined quality ex-pectations, which may take the form of a threshold ofminimally acceptable care with a range of permissiblequality variation above it. They have an ethical (and oftena legal) responsibility to cooperate in activities to improvethe performance of the teams and the organizations theybelong to in order to meet those expectations. They alsohave an ethical responsibility to cooperate in societal ef-forts to improve the overall health care system to accordwith those expectations. These obligations follow directlyfrom their underlying professional responsibility to servethe interests of patients.

The responsibility to improve has been stated in termsof meeting current expectations. That is, it is a responsi-bility to reach at least a threshold standard of quality.What makes it a responsibility for ongoing improvement isthe understanding that the societal expectation is one ofcontinuous progress in medicine—progress that producesits benefits when it is incorporated into medical practice.

As part of the system of care, patients also have a re-sponsibility to participate in quality improvement.21 To

meet the quality expectations of patients, health care pro-fessionals and managers must have the cooperation of thepatients themselves, because modern health care deliveryis a collective enterprise. Providing safe, effective, patient-centered, timely, efficient, equitable, and up-to-date careis a health care organization’s mission.22 QI methods allowthe organization to pursue its mission in a systematic,data-guided way. To be effective, these methods require atleast a minimum level of cooperation from patients, suchas permission for the collection and use of data on theirhealth conditions, treatments received, and outcomes. Inother words, ongoing activities to maintain and improvethe quality of care, including activities using QI methods,are an integral part of the normal operations of the orga-nization. This means that someone seeking care from ahealth care organization cannot insist on the freedom toopt out completely from efforts to improve the quality ofcare in that organization without jeopardizing the verybenefits he or she seeks. In fact, it is in the best interest ofpatients to cooperate with QI activities and even to seekout the health care organizations that are the most com-mitted to QI.

As an ethical matter, the responsibility of patients tocooperate in QI activities is justified by the benefits eachpatient receives because of the cooperation of the others inthe collective enterprise. To reap the benefits of such a sys-tem without participating in it—to be a “free-rider”—would be unfair.23 A patient’s responsibility to cooperateis, of course, subject to a standard of reasonableness,which presumes that adequate protections against individ-ual harm and violation of rights are in place. For example,patients can reasonably expect to have the confidentialityof their personal health information protected, and tohave the opportunity to choose whether to participate ina QI activity that exposes them to more than minimal riskcompared to routine medical care. We will address the na-ture of these protections and what constitutes minimaladditional risk over and above that found in routine med-ical care in greater detail later in this report.

In sum, health professionals, health care organizations,and patients have an ethical responsibility to cooperate inmaintaining and improving the quality of health care. Inits traditional role of protector of the health and safety ofits citizens, government also has ethical responsibilitieswith respect to the quality of care. In particular, it shouldhelp to clarify the content of the ethical and legal respon-sibilities of health care professionals and organizations to-ward their patients and to ensure that the responsibilitiesare met.


Although QI is closely related to clinical and man-agerial practice, it also has much in common withresearch. QI uses the kind of reasoning that is in-

herent in the scientific method, it involves systematic in-vestigations of working hypotheses about how a processmight be improved, and it frequently employs qualitativeand quantitative methods and analytic tools that are alsoused in research projects. It is these similarities between QIactivities and research projects that have touched off thedebate about whether QI activities should be subject tothe DHHS regulations on human subjects research.

How does QI differ from research? The definition ofresearch in the DHHS regulations highlights the knowl-edge-seeking aspect of research as the element that sepa-rates it from other activities: “Research means a systemat-ic investigation, including research development, testingand evaluation, designed to develop or contribute to general-izable knowledge.”24 In this definition, research is designedto develop new knowledge, not to implement knowledge;implementation happens later and separately, if it happensat all. Dissemination of research results, such as throughpublication in scientific journals, is thus of fundamentalimportance. Allowing research subjects to assume the bur-dens and risks of research is justified by the expectation ofsocietal benefits from the new knowledge produced; pub-lication is an important step in conveying the new knowl-edge to those who can put it into practice and thereby cre-ate the social benefits.

Although the definition does not make it explicit, theregulations implicitly reflect a view of research as a knowl-edge-seeking enterprise that is independent of routine medicalcare. Opinions differ on whether a society has an obliga-tion to engage in research, including research that will ul-timately yield important benefits to human health. Buteven if such an obligation exists, there is a presumption inresearch ethics that the research enterprise should rely onvolunteers.25 Investigators choose to do research, and peo-ple should be able to choose whether to be research sub-jects. This ethical paradigm emerged in response to re-search that imposed substantial risks on subjects yet of-fered them no prospect of direct benefit (such as the Naziexperiments and the U.S. Public Health Service Study ofUntreated Syphilis, commonly known as the “Tuskegeesyphilis study”), and it builds on the observation that theinterests of researchers are inherently tangential to, andpossibly even at odds with, the interests of the subjects.Since researchers want to produce successful research, theyhave a strong interest in enrolling subjects in their studies,keeping them in, and getting them to conform to studyrules. People who do not know themselves to be involvedin research, or who do not have the opportunity to discernthe merits of their own involvement, are at risk of being

exploited in the scientist’s search for insight—or morepragmatically, the search for funding, publication, and ca-reer advancement. To counter these pressures to use peo-ple wrongly to further the ends of other people and justi-fy the research, the subject’s decision to participate in theresearch must be voluntary and fully informed.

In health-related research, the independent nature ofthe enterprise is obvious when the research takes placeaway from a health care delivery setting and the humansubjects are healthy. When the research takes place in aclinical setting and involves sick patients who are receivingcare, research and therapy may be mixed, complicatingmatters significantly. Nevertheless, even research on an in-tervention that holds out the prospect of direct benefit tothe subjects is seen to have a different relationship to itssetting when compared to QI. Research with human sub-jects in a clinical setting is usually conceived, funded, and

managed as discrete projects, each led by a principal inves-tigator who is responsible for the project’s design and con-duct. For research in a clinical setting, the resources oftencome from outside the organization in which it is carriedout (from a federal agency, private foundation, or biotechor pharmaceutical corporation, for example); or, if frominside, from a separate research budget, not from clinicalcare resources. Most important, there is significant uncer-tainty about whether the intervention is in fact beneficial,and the activity is designed to produce generalizableknowledge about the intervention, not immediately im-proved care. This usually means that research has a proto-col that is constructed to minimize the effect of specificlocal variables and is maintained unchanged during theperiod of the research. It also means that there is relativelylittle urgency to disseminate the results, and the resultsmay not be made known to the scientific community formany months, even years, after the research is begun. Fi-nally, there is no presumption that the results will be in-corporated into the local care delivery process, and fre-quently little or no attention is paid to the challenges andopportunities for implementation of the findings in anysetting at all.

II. QI and Research: Similarities and Differences

QI is an ongoing process undertaken as a

consequence of health care providers�

responsibility to serve their patients�

interests. This makes it very different

from research.


In contrast, QI is an integral part of the ongoing man-agement of the system for delivering clinical care, not an in-dependent, knowledge-seeking enterprise. QI practition-ers design QI activities to bring about immediate im-provements in care, relying on theory and evidence fromresearch and practical experience to identify changes thatare very likely to be beneficial. QI activities take place in aparticular localized health care setting, their design is ex-pected to incorporate the specific features of the setting,they are led by people who work in that setting, and theyincorporate rapid feedback of results to bring about posi-tive change for the patients in that setting. Instead of afixed protocol implemented for a time period that maylast for years, QI methods often require repeated modifi-cations in the initial protocol as experience accumulatesover time and as the desired changes engage the localstructures, processes, patterns, habits, and traditions. Theterm “continuous quality improvement” (CQI), used al-most interchangeably with the term QI, highlights thefact that QI is not so much the implementation of dis-crete projects as it is an ongoing process of continual, self-conscious change, undertaken as a natural consequence ofhealth care providers’ ethical responsibility to serve the in-terests of their patients. This makes it a very different kindof endeavor from research and generates the prima faciecase for questioning whether the public’s interests wouldbe best advanced by subjecting QI to exactly the sameprocess of review and evaluation that has been designedfor discrete research projects.

QI and research are thus both conceptually and practi-cally distinct and play different roles in health care. Nev-ertheless, in the process of making change, QI does yieldinformation about what works and the way in whichchange can come about. When the results of QI activitiesin various settings are looked at together, patterns mayemerge—and even a single QI endeavor may yield valu-

able insights for a QI practitioner in another setting.Moreover, QI activities that have theory-based designs in-evitably become examples that in the long run tend to in-crease or decrease the level of confidence in the theorieson which they are based. With careful, systematic report-ing, the insights from QI are of use to people in other set-tings, just as case reports on individual patients are useful(although they too are not research). Given this, QI prac-titioners should be encouraged to share informationabout their QI activities with others in the health care sys-tem.

Furthermore, although QI and research are conceptu-ally distinct, in practice they are sometimes combined inone activity. In other words, some systematic, data-guidedactivities designed to bring about immediate local changeare also “investigations designed to develop or contributeto generalizable knowledge”; they are both QI and re-search.

At this point, a diagram is useful to illustrate the rela-tionships among the activities under discussion. In Figure1, the set “Clinical and Managerial Innovation and Adap-tation” is shown as a large oval and consists of activitiesdesigned to bring about immediate local improvements inclinical and managerial practice. The set “QI” is shown asa subset of that oval; it consists of clinical and managerialinnovation and adaptation activities that are designed andcarried out in a systematic, data-guided way.

“Research” is the large circle and consists of systematicinvestigations designed to develop or contribute to gener-alizable knowledge. The “Research” set includes basic andapplied medical research, of course. It also includes othercategories of research with a potential impact on healthcare quality, such as epidemiological research, health ser-vices research, management research, and educational re-search. The activities that are both QI and research are

Research QI Clinical &ManagerialInnovation & AdaptationResearch

on QI QI/Research on QI

QI/Research

Figure 1.

Note: The figures are not drawn to scale.


shown in the diagram as “QI/Research”—the overlappingsection of the “QI” set and the “Research” set.

One important research category, “Research on QI,” isexplicitly depicted in Figure 1. It consists of systematic in-vestigations designed to produce generalizable knowledgerelevant to the design and implementation of QI activities.The application of evidence-based medicine requires thegeneration of new knowledge on the behavior of systems,and research on QI contributes to this knowledge by help-ing to answer questions such as “What are the principlesof change?” “How do these principles work within differ-ent organizational contexts?” and “How can one spreaddesired change across an organization or between organi-zations?”

Research on QI can be independent of the QI activitiesit studies. For example, an investigator could do a retro-spective study of QI activities carried out in different or-ganizations with the aim of testing a hypothesis about theeffects of organizational characteristics on results. Alterna-tively, QI and research on QI can be combined in a singleactivity designed to produce both immediate local changeand generalizable knowledge about the process of change.For example, a health care organization with multiple de-livery sites could conduct an activity in which the sites aredivided into two groups, a different strategy is used ineach group to introduce a new practice, and the results ofthe strategies are compared, with elements included in theactivity’s design to facilitate generalization of the results toother organizations. In Figure 1, the set of such hybrid ac-tivities is the region of overlap between the sets “Researchon QI” and “QI,” and is labeled “QI/Research on QI.”26

Protection of Human Participants in QI andResearch

Both research and QI can adversely affect the peoplewho participate in them. Today’s human research sub-

ject protection system was inspired by research projectsthat offered subjects no direct benefit and exposed them tosubstantial harm. In QI, however, the changes made in theprocess of delivering care are expected to be improve-ments, and given the serious quality and safety problemsin health care, patients are often at greater risk if a currentpractice is allowed to continue than they are if a QI activ-ity goes forward. Nevertheless, any change may have un-expected negative consequences, and even the data collec-tion and monitoring that makes the change a QI activitymay itself impose burdens on the QI participants.

Burdens can take the form of direct physical harms,mental and psychological harms, “hassle” harms such astime consumed in completing surveys or submitting toextra clinical or administrative procedures, or harms relat-ed to loss of privacy and confidentiality. Often, there isuncertainty about whether a harm will occur and how se-vere it will be if it does occur. In the rest of this report, wewill use the broad term “risk” to refer to the various kinds

of burdens that may fall on participants as a result of a QIactivity.

Researchers have a recognized ethical responsibility toensure that the human subjects of their research are ap-propriately protected. The people using QI methods tomanage the quality of care also have an ethical responsibil-ity to participants in their QI activities. The need to meetethical standards creates a potential need for explicit struc-tures and oversight to ensure that research and QI arepracticed appropriately. It does not necessarily follow,however, that the standards and oversight should take thesame form for both activities.

To address this issue, we must first consider the gener-al principles and values that are at stake. We begin withthe principles of research ethics that form the foundationfor the current system for protecting human research sub-jects. The discussion is organized around seven ethical re-quirements: social or scientific value, scientific validity, fairsubject selection, favorable risk/benefit ratio, respect forpotential and enrolled subjects, informed consent, and in-dependent review.27 We discuss the justifications given foreach requirement in the context of research and then askwhether the same reasoning applies to QI work. (The re-sults of the latter portion of this analysis are summarizedin Box 5 near the end of this section.)

Social or scientific value: Is the research worth doing? Do thepotential gains from doing it justify the resources spent and therisks imposed?28

To be ethical, research must be worth doing, becausethere is an ethical obligation to use scarce resources re-sponsibly. Also, it would be wrong to expose human sub-jects to risk without social or scientific benefit. This meansthat researchers have an ethical obligation to share theknowledge gained from research with others, throughpublication in peer-reviewed journals or by other means,so that the benefit can actually be achieved.

Obviously, QI should also be worth doing for the samereasons. The potential gains must justify the resourcesspent on the activity and any risks imposed. The primarygains from QI are the benefits from the local improve-ments that result. Broader social benefits are also possiblewhen reports of QI activities in different settings generateinsights about the nature and process of improvement. Toachieve the social benefits, those who conduct QI activi-ties should be willing to share information about themwith others in the health care system.

This requirement is thus the same for QI and research.Its application is different, however—and in some waysmore difficult—in the QI context. To establish the worthof a research project, one must assess the net benefit to so-ciety from the knowledge it might produce. Usually, thisamounts to ensuring that the research question is appro-priate and the research design adequate. To establish theworth of a QI activity, one must assess the expected net ef-fects of the proposed process of change on present and fu-


ture patients in the local setting, as well as any social ben-efits that may result from sharing the insights gained.Forming an estimate of the impact of the local change canbe a complex professional and management task that re-quires a detailed understanding of the local system of caredelivery.

One must also assess the potential risks to those whoparticipate in the QI activity itself. The chief difficultyhere lies in the identification of the appropriate baselinefrom which to measure the risks to participants attribut-able to the QI activity. In a freestanding research project, aperson is either getting an intervention or not, and thebaseline for measuring harm and risk is ordinary life withno intervention. In a clinical treatment research project,the fact that the subject would be receiving clinical care inany case complicates the assessment, but typically, the re-search has a control group receiving a carefully specifiedstandard treatment that is the appropriate baseline fromwhich to measure the risks to subjects receiving the un-proven treatment.

In the QI context, the risks related to extra visits, ques-tionnaires, or procedures needed for data collection andmonitoring can be distinguished and attributed to the QIactivity. Although these burdens are usually small, theyshould be weighed in deciding on the merits of proceed-ing. But how should one assess the net risks to the partic-ipant associated with the proposed change itself? Here it isimportant to note that the care delivery process to which apatient is entitled is not specified in all its details; in fact,it is always changing. Even in a single clinical setting, theattributes of the care patients receive can vary from day today, depending on the availability of staff, the number ofpatients scheduled, and other factors. Managers and clini-cians have the professional discretion to make changes incare within certain (fuzzy) limits related to the range ofvariation in existing health care delivery. Over time, theyare actually required to make changes in order to remain incompliance with evolving minimum standards of care andto avoid malpractice liability. For many of the changes in-troduced through QI methods, the difference for patientsbetween being in or out of the QI activity may be nogreater than the existing variation in the patterns of careacross organizations, or within the same organization.Moreover, the methods of QI guide practitioners to usetheory, evidence, and practical experience systematically inorder to identify and implement changes that are verylikely to be beneficial. We are not suggesting that it is im-possible to recognize that a specific change has the poten-tial to cause significant harm. Rather, we are noting thatpatients cannot expect that services will only be presentedin one way, or that those services can be specified fully inadvance. This fact is relevant in the assessment of the riskattributable to inclusion in a QI activity.

A related issue arises for staff members, the other majorgroup of human participants in QI activities. For them,the appropriate baseline for measuring QI effects is the

risks in the usual work situation. For example, in the un-likely event that a QI activity exposes workers to extra ra-diation or toxic chemicals, or invades privacy by collectinginformation employers are not normally entitled to have(as might happen with surveys of the use of alcohol, ciga-rettes, or illegal drugs outside the workplace), the potentialeffects on workers should be included in the assessment ofQI-associated risk. On the other hand, one should not in-clude the harm to a worker’s economic security that mightresult if a QI activity reveals that the worker is incompe-tent or the organization can provide quality care withfewer workers. A health care employer has a moral andlegal right—and responsibility—to ensure that employeesare competent and conscientious, and the operations ofthe organization are efficient. The collection of informa-tion on individual and system performance is a feature ofnormal working conditions, and the risk that it will pro-duce information that leads to adverse job actions is partof the baseline job risk.

Scientific validity: Is the research methodologically sound—for example, is it properly structured to achieve its goals?

This requirement is also justified by the scarcity of re-sources and the need to avoid exposing subjects to possibleharm without benefit, and it also applies to both researchand QI. The interpretation of “scientific validity” and“methodologically sound” must be adjusted to QI, howev-er. Since the primary goal of QI is local improvement, QI’smethods and the local knowledge they generate can bequite different from the methods designed to enhance theopportunity to gain generalizable knowledge through re-search.29 QI methods should be chosen on the basis ofknowledge of the context, the requirements for connect-ing evidence to the context, and the requirements of actu-ally “making something happen” in the local setting whilebalancing the importance of the improvement sought andthe clarity needed in the assessment of changes against thepractical costs of monitoring the effects. A randomizedcontrolled trial seeks to eliminate context—key to QI—asa variable and thus would usually be inappropriate for thekinds of questions and changes that QI addresses, whilechanges that are already known to be worth their costs andreadily implemented should in general simply go for-ward.30

Fair subject selection: Are the subjects of the research selectedso that vulnerable individuals are not targeted to bear therisks of the projects while the rich and socially powerful are fa-vored for receiving the benefits?

In the selection of research subjects, there should besimilar treatment of equals, and both the burdens andbenefits of the research should be distributed in a fairmanner. But, to quote Emanuel and colleagues: “Thisdoes not mean that individual subjects and members ofgroups from which they are selected must directly benefitfrom each clinical research project or that people who are


marginalized, stigmatized, powerless, or poor should neverbe included.”31

This requirement applies to both research and QI. Fairsubject selection is particularly relevant to organizationsconducting entire programs of QI activities. The choice ofobjective and the design of the activity will sometimeshave implications for the fairness of the distribution ofbenefits and risks across patients. For example, the distrib-ution of benefits and risks may be affected by choosing todo QI projects on treatment of heart disease rather thantreatment of diabetes or asthma, or introducing improve-ments in some parts of an organization and using theother parts as comparison groups. Such distributional ef-fects should be considered in evaluating the ethical accept-ability of QI. In particular, people who are currently dis-advantaged in the health care system should not have tobear a disproportionate share of the burden of improvingthe system, and in fact should be given priority in the dis-tribution of the benefits of QI activities.

Favorable risk-benefit ratio: Is the research designed to mini-mize the risks and maximize the potential benefits? Are therisks to an individual human subject proportionate to thebenefits to the subject and to society?

This requirement applies to both research and QI. Itrelates to the requirement of social or scientific value, butfocuses directly on the participant’s situation, rather thanthe activity’s overall net benefit. An affirmative effortshould be made to design both research and QI activitiesto minimize the risks and maximize the benefits to theparticipants and society. For example, an activity may havea risk/benefit relationship within an acceptable range, butif the relationship could be improved with minimal effort/cost, both researchers and QI practitioners are ethi-cally obliged to do so.

Respect for potential and enrolled subjects: Is respect for sub-jects demonstrated by the following actions?

� protecting the privacy of individuals and maintainingthe confidentiality of private information

� maintaining the welfare of the subjects

� informing subjects of newly discovered risks or bene-fits associated with participation

� permitting withdrawal

� informing subjects of the results

The first three of these requirements apply to both re-search and QI. Protecting privacy and maintaining confi-dentiality is important in QI, as is maintaining the welfareof the participants and informing them of newly discov-ered risks or benefits associated with their participation.

The other two requirements do not apply to QI as stat-ed. The next section on informed consent explains whypermitting withdrawal from QI activities does not alwaysapply. It also argues that QI participants should receivegeneral information about QI efforts and should be ableto learn more about specific projects if they are interested,but a QI practitioner is not required to provide informa-tion automatically about a QI activity’s existence or resultsto the participants.

Informed consent: Do the subjects receive information aboutthe purpose of the research, its procedures, potential risk, ben-efits, and alternatives, so that each subject understands the in-formation and can make a voluntary decision whether to en-roll and continue to participate?

This requirement differs for QI and research. In princi-ple, research participation is optional for all parties,32 butwe have argued that this is not so for QI. Continuousquality improvement is part of the mission of health care

professionals and health care managers. They owe it totheir patients to be constantly trying to improve theirpractice; QI is an important tool for doing this; and there-fore, participating in QI is not completely optional forthem. Since quality improvement is a critical ingredient inthe creation of the benefits a patient seeks from healthcare, QI is not completely optional for patients, either.

In practical terms, this means that specific informedconsent is not required every time a human participant isincluded in a QI activity. Because QI is an essential part ofnormal health care operations, it is necessary—and ethi-cally acceptable—to have consent to receive health care in-clude consent to a reasonable level of cooperation with theongoing QI activities of the organization providing thecare. Through public education, people should already beaware of this before they seek care. When they presentthemselves for care, they should be reminded of the back-ground expectation and informed in general terms aboutthe QI that occurs in the particular setting and how theycan obtain more information about individual projects.33

By “a reasonable level of cooperation,” we mean thatpatients may be included without their specific consent inQI activities that involve no more than minimal addition-al risks to them compared with the normal risks associatedwith receiving standard medical care. If a QI activity is as-

Because QI is an essential part of normal

health care, it is necessary�and

acceptable�to have consent to receive

health care include consent to a reasonable

level of cooperation with QI activities.


sociated with more than minimal additional risk, patientsmust be asked for their specific informed consent beforeinclusion, and they are free to refuse it. In the latter case,patients must receive full information about the activityand the risks to them personally when they are asked toparticipate, just as for any clinical consent. If later on theywould like information about the activity’s results, theyshould be able to obtain it.34

Organizations should not be required to distribute de-tailed information proactively about the results of individ-ual QI activities. Most people would find the detail super-fluous, and the effort would waste their own resourcessince they ultimately pay the bill for the health care sys-tem. QI practitioners and the organization as a wholeshould, however, willingly provide information to anyonewho asks about any QI activity, to the extent possiblewhile respecting patients’ and providers’ right to privacy.

The most common examples of QI activities that donot require consent are those activities that use routinepersonal health information to implement a change in thecare process and impose no additional risk on participantsbeyond that associated with the information use. For ex-ample, a group practice might decide to modify schedul-ing practices to reduce the time patients spend waiting tosee their physicians when they come for an appointment.The QI team might begin by reviewing patient visitrecords to develop data on the types of visits, the averagetime needed for each type, the pattern of visit types over atypical month, and so on, and then use the data to devel-op new scheduling practices, implement them, monitorthe effect on waiting time, and refine them until the de-sired result is achieved. If confidentiality is appropriatelyprotected, then using patient information for QI activitieswithout specific consent is ethically acceptable on thesame grounds that its use is acceptable in other normalhealth care operations. Since QI is part of the care process,the people doing QI have the same status as others in thecare process. Often this is literally true, since QI teams areoften made up of the people who provide direct care; how-ever, the rationale is the same when access to protectedhealth information is granted to someone more removed,such as an epidemiologist abstracting information frommedical records for statistical analysis.

The Privacy Rule of the Health Insurance Portabilityand Accountability Act (HIPAA) has established standardsfor the protection of the confidentiality of personal healthinformation in normal health care operations, and has in-cluded QI (but not research) within that category.35 Con-sent to the use of protected health information for QI ac-tivities is provided as one item in the list of routine usesthe patient agrees to as a condition of treatment. Accord-ing to HIPAA standards, a QI practitioner’s access to pro-tected health information is limited to the informationneeded for the activity, persons engaged in QI have thesame confidentiality obligations as other health care work-ers, and the same precautions must be in place for the use

of protected health information in both clinical care andQI. (See Box 4.)

When the human participants in a QI activity are staff,consent requirements play out a little differently. Healthprofessionals, management, and other workers have a gen-eral ethical obligation to cooperate to improve the qualityof care, and a specific contractual obligation to cooperatewith their organization’s QI program (since it is part ofnormal health care operations). The obligation is not un-limited, however. Specific consent is required when the QIrepresents more than minimal risk to the worker, as mea-sured from the baseline of normal working conditions.Therefore, consent would be required for a QI activitythat exposed workers to more than minimal additionalrisk of physical or mental harm compared to their currentworking conditions (exposure to radiation or toxic chemi-cals, for instance) or collected information about workersthat was outside the category of information employers arenormally entitled to have about their employees (such astheir use of tobacco, alcohol, or illegal drugs outside of theworkplace). Consent is not required, however, for QI thatis risky to the worker simply because it might generate ev-idence of incompetence on the job or lead to a reductionin force for efficiency reasons. The formal contracts withworkers and the informal workplace expectations shouldreflect the understanding that cooperation with minimalrisk QI activities is part of the job. As with patient partic-ipants, confidentiality of personally identifiable informa-tion about workers should be appropriately protectedwithin the organization.

Does any of this analysis change if those leading the QIproject publish or otherwise share their process and ac-complishments with others? Do participants have to givespecific consent to the publication of QI results? QI prac-titioners should be encouraged to share QI results throughpublication and other means, since sharing yields benefitsto individual patients and to society as a whole. The peo-ple who receive care from health care organizations pay asa group for the development of this knowledge, and over alifetime, people receive care from many different organiza-tions. Although some organizations may consider QI re-sults proprietary information and wish to maintain secre-cy for competitive advantage, patients are better off ifhealth care providers cooperate by sharing QI results.

Even though most QI can be carried out ethically with-out explicit patient consent, published results must be pre-sented in a form that preserves patient confidentiality;otherwise, the team must have the patients’ specific con-sent to publication. Similarly, individual worker informa-tion must be nonidentifiable or worker consent to publi-cation must be obtained, unless there are overriding con-tractual or legal rules relating to the provision of informa-tion on worker performance.

The ethical acceptability of the QI activity itself andthe ethical acceptability of the form in which the resultsare disseminated are conceptually distinct, and the deci-


sion about whether to publish and in what form may rea-sonably be made after the results are in. Therefore, it is ac-ceptable and may be desirable to separate ethical assess-ment of the activity itself and ethical review of the form inwhich the results are disseminated. Thus, organizationsmight well have different procedures for ensuring ethicalimplementation of QI and ethical dissemination of thefindings.

Independent review: Is there review of the design of the re-search, its proposed subject population, and its risk-benefitratio by individuals unaffiliated with the project?

In research, the purpose of ethical review by individu-als unaffiliated with the proposed research activities is toensure that the researcher has understood the require-ments for ethical research and applied them properly indeveloping the proposed activities. Applying requirements

takes judgment calls, and conflicts of interest may distortthe researcher’s judgment (consciously or unconsciously),potentially causing harm to participants or taking awaytheir opportunity for voluntary consent. Review can pro-tect against such distortion and help potential researchsubjects have confidence in the honesty and integrity ofthe offer to volunteer to participate. In the United States,this ethical review takes the form of review of research pro-tocols and relevant documents by the institution’s IRB.

In the discussion of the first six categories of ethical re-quirements for human research subject protection, wehave concluded that QI activities with human participantsshould meet similar ethical requirements, but the require-ments must be interpreted and applied somewhat differ-ently. (See Box 5.) As with research, some considerationsmight justify review of QI activities to ensure that the re-quirements for ethical conduct of QI are met. Review

� defines health information as “any information,whether oral or recorded in any form or medium that:(1) is created or received by a health care provider, healthplan, public health authority, employer, life insurer,school or university, or health care clearinghouse; and (2)relates to the past, present, or future physical or mentalhealth or condition of an individual; the provision ofhealth care to an individual; or the past, present, or fu-ture payment for the provision of health care to an indi-vidual” (45 CFR 160.103)

� defines protected health information (PHI) as “indi-vidually identifiable health information that, subject toexceptions, is (i) transmitted by electronic media; (ii)maintained in any medium described in the definition ofelectronic media; (iii) transmitted or maintained in anyother form or medium” (45 CFR 164.501)

� extends to a defined set of covered entities that trans-mit health information electronically in connection withcertain defined HIPAA transactions (45 CFR 160.13):1

• health plans• health care clearinghouses• health care providers• nonemployee business associates of covered entities• researchers who are health care providers engaging

in any of the covered electronic transactions

� prohibits a covered entity from using or disclosingPHI about an individual without the individual’s specif-ic written permission (“authorization”) unless:

• PHI will be used/disclosed for treatment, payment,and health care operations (45 CFR 164.506(c))

• PHI will be disclosed to business associates provid-ed certain arrangements in place (45 CFR 164.502 (e))

• the disclosure is required by law, authorized by lawfor public health purposes, or required by the FDA foradverse event reporting (45 CFR 164.512)

• an institutional review board (IRB) or privacy boardwaives or alters the authorization requirement to use ordisclose PHI for research, which is possible when certainconditions are met (45 CFR 164.512(i))

• a researcher satisfies certain conditions to use PHIfor review preparatory to research (45 CFR164.512(i)(1)(ii))

• a researcher satisfies certain conditions for researchsolely on information of decedents (45 CFR164.512(ii))

• the PHI has been deidentified in accordance withspecific Privacy Rule standards (which makes the data nolonger PHI [45 CFR 164.514(a)-(c)])

• the covered entity releases the information in theform of a limited data set that includes the removal ofcertain identifiers and a data use agreement between theresearcher and the covered entity (45 CFR 164.514(e))

� QI (but not research) is included in health care oper-ations and specifically exempted from the above require-ment for project-specific written permission (“authoriza-tion”)

1. For additional information see U.S. Department of Healthand Human Services, OCR Privacy Brief, Summary of the HIPAAPrivacy Rule, at http://www.hhs.gov/ocr/privacysummary.pdf.

4. Summary of the Privacy Rule of the Health Insurance Portability and Accountability Act

45 CFR 160 and 45 CFR 164, Subparts A and E


might be justified, for example, if the QI might pose sub-stantial risks or waste substantial resources, and if reviewby a nonparticipant could raise those issues and stop theactivity. As QI becomes a desirable skill and care systemsinvest in QI budgets, the leaders of QI might find them-selves invested in QI work that has become unresponsiveto the real needs of patients and families—a conflict of in-terest that would echo that which gave rise to research re-view. Furthermore, the focus on QI is relatively new, andthe practices that support it are still in flux. Thus theboundaries between QI and research and between QI androutine treatment are not well understood. Review by anonparticipant would be useful if a project that shouldhave counted as research is not being managed as such, orif a project that offers no particular opportunity for im-provement over usual practice is being touted as QI.

Since these arguments justify some review, at least forsome projects, the question is what form it should take.Should QI also receive IRB review, or would some otherapproach to ethical oversight be more appropriate? In thenext section, we will discuss why we have concluded thatIRB review is not ethically required and would not evenbe generally beneficial or effective for QI activities.

IRB Review of QI: Why Not?

Much of the concern expressed about IRB review ofQI has focused on the time and effort associated

with the preparation of documents for submission to anIRB and the time required for the IRB to complete its re-view.36 Researchers have been dissatisfied with these as-pects of IRB review for some time, and improving IRB ef-ficiency is an acknowledged policy goal.37 This issue iseven more important for QI than for research, since thepeople who initiate QI are already employed in the deliv-ery of care, and the resources used for QI are part of thehealth care system’s cost structure. Even with a more effi-cient IRB process, many valuable QI projects would beunable to muster the necessary resources and simplywould not be done if IRB review were required. For thoseQI activities that did go forward, the process would im-pose high transaction costs on improvements that areoften small in scale and represent little burden or risk toparticipants. This would be a poor use of scarce resourcesand would in itself be a kind of harm to the patients,whose resources are thereby wasted.

There is a more basic problem, however: the currentstructure of IRB review is inappropriate for QI. To explainwhy, we will describe the process and discuss its limita-tions in the QI context. The IRB’s role is to ensure that re-search carried out under an institution’s auspices complieswith federal regulations and conforms to accepted ethicalprinciples. Usually, the IRB is part of the institution inwhich the research is done, but it is always independent ofthe research itself. The IRB meets at intervals to review re-search protocols, consent forms, and other relevant docu-

ments. The research protocol specifies the goals of the pro-ject, the subject population, methodology, and time peri-od for the research. After reviewing the documents, theIRB either approves the research project or requires theprincipal investigator to make changes to the protocol, theassociated consent form, or other documents until all areacceptable. Then the project begins and continues un-changed unless the PI obtains IRB approval for modifica-tions. In addition, the PI is required to report any unan-ticipated problems involving risks to subjects or others tothe IRB. Although the research is carried on under theauspices of the institution and receives continuing IRB re-view (at least once per year), the PI takes full responsibili-ty for its day-to-day conduct.38

Most QI activities are unlike the activities that IRBsroutinely approve and monitor. Instead of a fixed protocolwith fixed goal, methodology, population, and time peri-od, QI activities ordinarily entail frequent adjustments inthe intervention, the measurement, and even the goal overtime as experience accumulates. Even in cases where a QIactivity produces insights useful in other settings as well aslocal change, the activity is not a knowledge-seeking en-terprise that is independent of ongoing clinical care. It isdesigned to produce local change and is closely linked tothe normal operations of the institution in which it takesplace. QI occurs in the context of the professional obliga-tions of the health professionals involved, and the peoplewho initiate or lead a QI activity are often part of the teamthat provides care. All parties affected by the QI activityfunction within a supervisory and management structurethat already bears responsibility for clinical care, includingits quality and safety. As we have shown above, the ethicalrequirements for QI are in some ways similar to those forresearch, but they must be modified to the particular localsetting and the QI goal of immediate improvement. Inparticular, the desirability of the change itself and the risksto participants often depend heavily on the characteristicsand reactions of the system in which the QI activity oc-curs. This means that assessing the activity’s potential ef-fects and understanding how risks can be minimized re-quires intimate knowledge of that system.

Given these characteristics of QI and its role in healthcare, the procedures used to ensure its ethical conductmust allow QI activities to remain flexible and fully inte-grated into the ongoing management of care delivery.Since the current IRB review process does not readilyallow this, requiring QI to undergo IRB review would cre-ate perverse incentives for professionals and organizationstrying to improve care. Change is already an intrinsic fea-ture of health care delivery. Solving the quality and safetyproblems in American health care will require manychanges—and these changes are more likely to be genuineimprovements if done in a systematic, data-guided way,using QI methods. If the “price” of making change in thisway is a cumbersome, costly process of review, however,


managers may opt to make change without using QImethods. Worse, they may simply leave things as they are.

Of course, not all clinical and managerial changes areethically acceptable. The underlying questions here are:What standards should be applied to determine whether achange is ethical? How should the agents of change inhealth care be held accountable for their actions? Current-ly, to the extent that accountability for clinical and man-agerial change occurs, it is through the system of account-ability that has been established for all ongoing clinicalcare. Admittedly, the clinical accountability system hasshortcomings, and concern about these shortcomings fuelsthe concern for the welfare of QI participants. Neverthe-less, desirable procedures for ethical oversight of QI shouldencourage, not discourage, the collection and evaluationof information on the consequences of any change thatcould harm someone, and should avoid acting to slow thepace of improvement. These ends cannot be accomplishedby exporting QI into the IRB-based research review sys-tem.

Instead, in our view, ethical review to protect humanparticipants in QI should be selectively imported into theaccountability system for clinical care. By this, we meanthat ethical oversight of QI should be fully integrated intothe routine management and supervision of health careoperations.39 The responsibility for ensuring that this inte-gration occurs and results in ethically conducted QIshould belong to the entities that already have responsibil-ity for the quality of care provided by individual profes-sionals and health care organizations.

Protection of human participants in QI belongs in theclinical accountability system because ethical decision-making is an intrinsic part of everyday clinical manage-ment. Clinical and managerial decisions in health careoften—perhaps always—have ethical dimensions. SendingQI to an IRB for ethical review implies that protection ofparticipants is an ethical question that is separate from theother ethical questions that arise in managing patientcare—or even worse, that ethical decision-making in gen-eral is something to outsource. When an activity is asclosely tied to clinical practice as QI is, the responsibilitiesfor its conduct should remain closely aligned. The specialcircumstances of widespread conflict of interest and risksof wrongful use of human subjects without consent thatgenerated the strong requirements of separate ethical re-view do not apply to QI.

Some might challenge the last statement by arguingthat change in health care is inherently biased towardmaking patients worse off for the sake of “the bottomline.” From this perspective, QI is often just a cover forchange motivated by concern for profits, not patients. Theunderlying issue here is that of the relationship of costcontainment to quality improvement, and the relationshipof both to professional and organizational ethics. To re-spond to the challenge, one might point out that in today’shealth care system, it is often possible to make changes

that improve quality and lower cost at the same time.There is ample evidence of waste in the system, in thesense of care that is unnecessary, harmful, or inefficientlyproduced and delivered. Reallocating the resources devot-ed to this care to beneficial uses improves care without in-creasing cost.40

Nevertheless, sometimes a tradeoff between quality andcost is unavoidable. Managers must make changes forbudgetary reasons, and the changes may reduce the quali-ty or quantity of care for some people. Is it always unethi-cal to make patients worse off than they are under existingpractices—or even to expose them to a small chance ofbeing worse off—without their explicit consent? The an-swer is no. Patients have an interest in maintaining andimproving quality, but they also have a stake in ethicalstewardship of the resources devoted to health care. Whenchanges must be made and budgets are tight, sometimes itwould be unethical not to reduce the quality or quantity ofcare for some people. For example, suppose a costly newdiagnostic test becomes available that allows diagnosis of adeadly disease at a treatable stage. The medical staff mem-bers of a nonprofit health plan think the plan should beginusing it immediately in patients at risk for the condition.

How should the plan find the resources to cover the test-ing and the effective treatment it makes possible? The plancould raise premiums, free up resources by producing thesame care more efficiently, reallocate resources by reducingthe quality of care slightly for plan members in a domainwhere the current quality level is well above the acceptablethreshold level, or do some combination of the three. Eachalternative has ethical implications, all should be on thetable, and sometimes, reducing the quality of care in an-other domain is the solution that best meets ethical re-quirements. Similar examples could be constructed for ahospital, a group practice, or an individual physician.

One might acknowledge that change must be man-aged, that decisions about change have ethical implica-tions, and that a decision to trade quality for cost can beethically acceptable, yet still argue that, given the currentstate of quality and safety in health care, these decision-making processes cannot be trusted. Many now agree thata transformation in the organizational culture of healthcare is needed to improve the quality of care. This is exact-ly where QI methods make their contribution, however.

The procedures used to ensure that QI is

conducted ethically must allow QI activities

to be flexible and integrated into the

management of care delivery. The current IRB

review process does not allow that.


Optimal oversight of change requires learning how to as-sess the dangers of specific changes to patients, and QI is asystematic approach to improving decisions about change.By providing evidence on the effects of changes, QI meth-ods are valuable tools in holding health care providers ac-countable for the care they deliver and can be used to re-spond effectively to past failures to live up to quality stan-dards. We conclude that to the extent that health caremanagers have a conflict of interest in managing cost-re-ducing change in health care, QI methods are not theproblem, but an integral part of the solution.

Our discussion has brought us back to the questionraised in Section I of this report: should increasing effi-ciency by providing the same quality of care at lower costcount as an improvement in quality? One way to answer isto say: “No, it is a good thing, but it is cost containment,not quality improvement.” Another way to answer is tosay: “Yes, ‘value for money’ is a quality characteristic ofhealth care.” Both positions have conceptual merit, but ac-

cepting “value for money” as a quality characteristic haspractical advantages. In considering the ethics of using QImethods to evaluate a designed change, it is best to avoidcreating a bright-line distinction between efforts to im-prove quality and efforts to reduce cost. When assessingquality, people often disagree about what weights to applyto different dimensions of care. If several dimensionschange at once, some may consider the result higher qual-ity care and others lower quality care. Thus, when a pro-ject reduces cost and also changes quality characteristics,there may be no consensus on whether the change is qual-ity improvement that also lowers cost or cost containmentachieved at the expense of quality. Either way, manage-ment should know what the consequences of the changeare, and QI methods can be used to find out.

Note that the methods used in QI are also useful forthe study of changes that everyone agrees are not good forpatients but that are made anyway: for example, changesforced by budget deficits, such as cuts in nursing staff or

Social or scientific valueThe gains from a QI activity should justify the re-

sources spent and the risks imposed on participants.

Scientific validityA QI activity should be methodologically sound—

properly structured to achieve its goals.

Fair subject selectionParticipants should be selected to achieve a fair distri-

bution of the burdens and benefits of QI.

Favorable risk/benefit ratioA QI activity should be designed to minimize risks

and maximize potential benefits, and to ensure that risksto an individual human participant are proportionate tobenefits to the participant and to society.

Respect for participantsA QI activity should be designed to protect the priva-

cy of participants through confidentiality.Participants in a QI activity should receive informa-

tion about findings from the activity that are clinicallyrelevant for their own care.

All patients and workers in a care delivery settingshould receive basic information about the program ofQI activities.

QI results should be freely shared with others in thehealth care system, with participant confidentiality pro-tected by putting results into nonidentifiable form or ob-taining specific consent to sharing.

Informed consentPatients should give background consent to inclusion

in minimal risk QI activities as part of consent to receivetreatment.

Patients should be asked for informed consent to beincluded in a specific QI activity if the activity imposesmore than minimal risk.

The risk-harm ratio for patients is measured relativeto the risk associated with receiving standard health care.

Workers (employees or nonemployee professionalswho provide care within an organization) are expected toparticipate in minimal risk QI activities as part of theirjob responsibilities.

Workers should be asked for their informed consentto inclusion in a QI activity that imposes more thanminimal risk.

The risk to workers is measured relative to the risk as-sociated with the usual work situation and does not in-clude any risk to economic security that might result if aQI activity reveals that the worker is incompetent or thatthe organization can provide quality care with fewerworkers.

Independent reviewAccountability for the ethical conduct of QI should

be integrated into the system of accountability for clini-cal care. Each QI activity should receive the kind of eth-ical review and supervision that is appropriate to its levelof potential risk and project worth.

5. Ethical Requirements for the Protection of Human Participants in QI Activities


To ensure that QI meets ethical requirements, theremust be a set of regular procedures, understoodobligations, and clear standards to hold health care

professionals and organizations accountable.Many organizations and policies within the health care

system intertwine to create the existing system of account-ability for the quality and safety of health care.41 First, ofcourse, is the widely understood social contract, articulat-ed in statements of professional ethics, which calls onpractitioners to be allegiant to the needs of their patients.Formal accountability organizations include state licensingbodies, private sector accrediting bodies, the Centers forMedicare and Medicaid Services, and certification andcredentialing organizations. Managed care organizationsand large employers often impose quality performance re-quirements on the professionals and organizations withwhom they contract. Medical staff in hospitals and grouppractices undertake responsibility for credentialing theirmembers and awarding staff privileges. Many hospitalsnow support clinical ethics committees. Under certain cir-cumstances, health care organizations are subject to the ac-countability requirements included in the HIPAA PrivacyRule and the DHHS regulations for the protection ofhuman research subjects. The malpractice litigation sys-tem allows patients to pursue compensation from healthcare providers for harms caused by negligence in the deliv-ery of care. Within this accountability system, the variousparties may hold each other accountable over differentcontent areas. For example, a hospital may evaluate the ac-tivities of its physicians, but then in turn be accountable to

government regulators, contracting health plans, and thepublic.

Within the clinical accountability system, the arrange-ments for ensuring that QI is conducted ethically shouldbe designed to be flexible. The specific practices should beable to vary across organizations to accommodate varia-tion in the local environment and in the QI activities con-ducted. The interaction between the characteristics of aQI activity and the characteristics of the local setting iscritical in determining whether the activity is ethical andwhat kind of review and supervision it should receive. Ingeneral, if participants are at more than minimal risk,there must be more clarity about review and more super-vision. However, for a given level of risk, an organizationwhose employees are very familiar with QI methods andethical requirements can allow QI to be done with a lowerlevel of supervision than an organization whose employeeshave lesser skills. A large organization may choose to havea specialized QI oversight system that a small organizationcannot afford; however, the small organization may notneed any formal review if its QI activities are simpler andits size allows everyone in the organization to know whatis going on. For example, an HMO with multiple deliverysites might have formal procedures, while a small nursinghome might simply require an informal review by thenurse executive.

The practices should be able to vary over time. SinceQI is part of a process of continual change in health care,it would be a mistake to impose an ethical oversight struc-ture that cannot respond quickly and appropriately to

III. Institutional Arrangements Needed to EnsureEthical Conduct of QI

elimination of evening clinic hours. In fact, it may be es-pecially important to understand the effects of purelycost-driven changes, since their effects may be more far-reaching and more negative than anticipated. When suchchanges are made in an incremental, data-guided, moni-tored way, management can understand better what is ac-tually at stake and more easily minimize the negative im-pact. Obviously, these activities are not quality improve-ment and should not be referred to as QI, since doing sowould foster cynicism about the whole QI enterprise.Nevertheless, the practices that provide ethical review ofQI should be designed to encourage rather than discour-age the use of data collection and monitoring methods tominimize harm in these situations.

In sum, we argue that ethical review of QI activities forpurposes of protection of human subjects should be inte-grated into the accountability system for managing thequality of care rather than delegated to the IRB-based re-search review system. Is this a proposal to give QI a “free

pass” compared to research? No. We believe that reviewcan be cheaper, less bureaucratically frustrating, and—most important—at least as likely to ensure ethical con-duct of QI if designing and implementing it is seen as acore management function.

This approach will be most effective if it occurs as partof a systematic transformation of the existing clinical ac-countability system. The transformation should includeclarification of the ethical responsibility of clinicians andmanagers to take the lead in improving the quality of care,ongoing guidance on the requirements for the ethical useof QI methods, and change in professional and internalorganizational cultures to make continuous quality im-provement routine. Established external clinical care ac-countability mechanisms should then be used to makesure that professionals and organizations have proper su-pervision of the conduct of QI, including the extent towhich ethical requirements are met and whether humanparticipants are adequately protected.


changing conditions. If the QI regulatory procedures erectartificial barriers to new ways to organize care delivery andreimbursement in health care, they may prevent desirablechange, or even create perverse incentives that encourageundesirable change.

The practices should be designed to ensure that ethicaloversight produces its benefits at affordable cost. The ad-ministrative complexity of doing QI should be as low aspossible so that people employed in direct care are not dis-couraged from initiating QI projects. This considerationreinforces the conceptual arguments for building ethicaloversight of QI into the ongoing management of care de-livery.

We recommend that the primary responsibility for theethical conduct of QI be lodged in individual organiza-tions, where it should be seen as a normal clinical careobligation and integrated into normal supervision andmanagement, with the organizations’ leaders responsiblefor seeing that the integration occurs and is effective. Theorganization doing QI could be, for example, a practice ofone or a few physicians and support personnel, a largemultispecialty physician group practice, a hospital, a nurs-ing home, a hospice, a staff model HMO, or a more loose-ly configured managed care plan. It could also be a hospi-tal chain, a nursing home chain, or an integrated, compre-hensive multisite managed care organization like Kaiser orIntermountain Health Care. It could be the VeteransHealth Administration. Whatever the organization is, itmust have a set of practices in place to ensure that QI ac-tivities are done, and that they are done in conformitywith ethical requirements. The organization must alsohave a procedure in place to provide basic information topatients and staff about the organization’s QI activities.We call this internal accountability for the ethical conductof QI.

We recommend that there also be procedures to ensurethat individual health care organizations and their leadersare held accountable for having well-designed, functioninginternal accountability practices in place to carry out theirQI-related responsibilities. We call this external account-ability for the ethical conduct of QI, and we argue belowthat it should be integrated into the overall system of ac-countability for the quality and safety of clinical care.

Internal Accountability for Ethical Conduct of QI

Activities in a health care facility require managementand supervision for many reasons besides the protec-

tion of the human participants in QI. Sometimes the rea-sons are directly related to other ethical issues in healthcare delivery. For example, management and supervisionare required to ensure that an organization fulfils its ethi-cal responsibilities to employees, ethical responsibilities ofstewardship of the organization’s resources, and ethical re-sponsibilities to patients who have been given implicit or

explicit promises about the kind of care they will receive.Sometimes the reasons are not directly related to ethical is-sues but to other organizational concerns. For example,the organization may be seeking to maintain market posi-tion and reputation, remain financially solvent, avoid law-suits, or prevent bad publicity. The ethics-related oversightof QI, including the oversight of the protection of humanparticipants, must be incorporated into the overall systemof management and supervision of health care delivery.

In discussing how this can be done, it is useful to dis-tinguish three broad categories of oversight: professionalresponsibility for QI; local management review and super-vision of QI; and QI-IRB review for QI that is also humansubjects research. In addition, there should be proceduresin place for ethical oversight of QI activities that are car-ried out through collaborations across organizations.

Professional responsibility for QISome QI activities are so closely tied to clinical practice

that no additional oversight beyond normal managementsupervision and professional ethics is needed. Activitiesappropriate for this category are of minimal risk, in thesense that confidentiality is protected and no additionalrisk or harm is imposed on patients compared to that ex-pected from clinical care in the absence of the activity.(And often, a priori evidence exists that suggests that sub-stantial benefits are likely for the patients or staff in-volved.) Therefore, specific informed consent is not re-quired. The activities are simple in design, so there is noneed for methodological review. The effects of the activi-ties are very local, in the sense that their success or failurewill have no repercussions on other parts of the organiza-tion. The activities do not use significant additional orga-nizational resources beyond the usual resources for clinicalcare.

This category includes most of the routine perfor-mance-enhancing improvement activities undertaken byindividual health professionals. Here are two examples ofit:

� A doctor wants a test order tracking system thatmakes it easy for him to remember what tests have beenordered for a patient, access the test results, and maketimely and appropriate adjustments in the patient’s care.He develops his own patient tracking system, monitorsits performance, and refines it over time in response tohis observations until he is satisfied with it.

� Several nurses find that they are each using a differentmethod to perform a routine nursing procedure. Themethods are within standard nursing practice and areused interchangeably in various clinical settings; howev-er, the nurses wonder whether the methods differ in thetime they take and the effect they have on the patients.The nurses try each method in sequence over a fewmonths, keep track of how long the procedure takes


with each method, and ask patients whether there is anydifference in comfort. At the end of this QI activity, theyconclude that one method works better than the othersin their local setting and all decide to use it.

Activities in this category do not receive any addedoversight other than discussion among the participants,and they may not even be written down in any formalmanner; however, the clinicians who engage in them areresponsible for protecting the participants by following theethical requirements of their profession and their organi-zation. It is worth noting that the monitoring initiated bythe clinicians carrying out the QI activity ensures that pa-tients’ interests are protected. If anything were going awry,the altered process would quickly be abandoned.

Local management review and supervision of QIThis category consists of activities designed to improve

care in the local setting that require at least some monitor-ing by management. The minimum management elementin this category is a method for tracking the QI activitiesin progress. In a small organization, this could be as sim-ple as an informal discussion with a QI manager or otherdesignated executive staff member. In a larger organiza-tion, it could take the form of standardized reportingabout QI activities to a quality management staff member.

To establish a physical record of QI activities, a simpleregistration process might be used. Such a process couldalso serve as an educational tool to help people doing QIthink through the design of the activity and recognize theethically important aspects, and as a screen to identify ac-tivities with attributes that indicate the need for moreelaborate review and approval. In a small organization,registration could take the form of answers to a set ofstructured questions kept in a locally available file. QI ac-tivity registration that met the requirements outlined bythe questions in the registration form would constituteapproval to perform the activity. In a large organization,registering and informing management might be accom-plished through an interactive, Web-based registrationform. The initiator of the activity would enter a short de-scription on the form and answer a set of questions. Aseach question is answered, the individual would be direct-ed to carry out the activity or to obtain a specific form ofoversight, depending on activity characteristics and poten-tial for risks. After completion, the form would be storedin a database to be used as a tool for ongoing monitoringof QI activities, as a source of information about past andpresent QI activities for members of the organization, andas a basis for systematic organizational reflection on thepatterns and outcomes of its improvement work.

For some activities, registration is sufficient. If a regis-tered QI activity uses few organizational resources, impos-es minimal additional risks on patients besides the use ofprotected health information, and has confidentialitymeasures in place, it would be unusual to require further

review by supervisors to verify that it meets ethical re-quirements. Usually, the activity could simply go forwardwithin the ordinary supervisory framework.

Other activities in this category may need explicit re-view and supervision by someone other than the peopleengaged in them. When a QI activity takes more re-sources, has the potential to impose more disruption orrisks, or makes use of unusually sensitive protected healthinformation, supervisors and immediate managers shouldbe aware and should take the opportunity to be sure thatthe endeavor is prudent and that the participants arebrought in on a fair and appropriately informed basis.Again, two examples:

� After doctors develop patient test order tracking sys-tems, they ask the nurses who care for their patients touse them. The nurses point out that using a dozen dif-ferent doctors’ systems would create complexity forthem and lead to mistakes. They recommend that testorder tracking be made into a QI activity to develop acommon tracking system incorporating the best featuresof the individual systems. The QI activity is then carriedout under the supervision of the management level that

covers all parts of the organization affected by thechange, with one supervisory task being review of the ac-tivity’s design for conformity with ethical requirements.

� An activity that imposes no more than minimal addi-tional risks on patients does not require specific in-formed consent on ethical grounds, but sometimes aninformal consent to additional data gathering or minorannoyances is appropriate. “We’re working on improv-ing the way we move patients through the radiology de-partment. Would you mind filling in a brief surveyabout your experience during today’s visit?” For such QIactivities, management may wish to introduce measuresto ensure that any written surveys and patient interviewsare properly structured and that the activities are coordi-nated so that even minor burdens are spread fairly acrosspatients.

As a QI activity changes along the dimensions of in-creasing levels of organizational resource use, methodolog-ical complexity, potential health risk, burden of participa-tion, and confidentiality risk, more intense review andoversight becomes necessary in order to ensure that the QIpractitioners understand the ethical requirements and that

Organizations should establish specialized

Quality Improvement IRBs to review activities

that are both QI and research on human

subjects.


the QI activity meets them. When QI activities have sub-stantial resource or staffing implications, major changes inthe providers’ or patients’ expectations, or more than min-imal additional risk to patients compared to ordinary clin-ical care, then they require active approval by managementabove the level of the immediate managers involved with-in the organization. The registration process outlinedabove could identify activities with characteristics thatmake additional management approval necessary. De-pending on an organization’s structure, form, or size, ap-proval could be provided by a quality management staffmember, QI director, facility manager, or other staff mem-ber who is schooled in QI methods and the ethical re-quirements for QI. Clinical leaders most responsible forthe patient population affected (a department chair ornurse manager, for example) should also be part of theprocess of approval.

Activities that pose more than minimal risks comparedto those in ordinary clinical care require specific informedconsent and explicit review by at least one person who isnot directing the project and who is knowledgeable aboutboth ethical issues and management perspectives. This istrue even if the proposed project also holds the potentialfor substantial benefits to the subjects. The possible harmsmake it desirable to create “distance” between the organi-zational units that favor the proposed changes and theoversight structure. A patient safety committee or clinicalpolicy committee could perform a structured review. Inany case where an outsider might think the internalprocess could involve a conflict of interest, it would be ap-propriate to refer the proposed activity to a committeewith representation from outside the organization. This

committee could be the organization’s IRB or the QI-IRBdescribed in the section below. An outside perspectivewould also be useful in cases where an organization’s staffmembers lack the technical skills to assess the benefits andburdens of the activity; in such a case, the organizationmight contract with a group such as a quality improve-ment organization.

We believe that an organization should develop its ownapproach to bringing the QI activities in this category intoconformity with ethical standards. QI methods themselvesattach considerable importance to identifying the compo-nents of the existing system that must be involved in a QIactivity in order for it to be done well. In each organiza-tion, the categories that managers use will be defined bycombinations of attributes, some relevant to ensuring eth-ical conduct of QI and others relevant to other superviso-ry goals. A wide variety of QI management configurationsare likely to emerge. As long as these different configura-tions are effective in ensuring that QI activities meet ethi-cal requirements, they should be acceptable.

It is worth noting that, from an ethical perspective, ex-ternal review of individual QI activities is not generally in-herently better than internal review. On the one hand, anoutsider might provide a useful community perspective onthe acceptability of the risks involved, and for some kindsof risks, this perspective should definitely be sought. Onthe other hand, an outsider might not understand the in-ternal environment well enough to understand the truenature of the risks and benefits and the relationship of theactivity to the organization’s current standard of care andallocation of resources. As we will discuss below, there is aneed for external oversight, but it should be primarily fo-

Research

Exempt HuSR

Human SubjectsResearch (HuSR) Activities Involving

Human Data Sources

Figure 2.



cused on review of the structure and functioning of the or-ganization’s procedures for ensuring the ethical conduct ofQI, rather than on the review of individual QI activities.

QI-IRB review and supervisionAs noted earlier, some activities are both QI and re-

search involving human subjects. The DHHS regulationsrequire any activity meeting the definition of human sub-jects research to be treated as such even when it is also an-other kind of activity, such as QI, education, or publichealth surveillance. Therefore, these hybrid activities mustcomply with the regulations if they are supported by fed-eral funds or are covered by the assurance many organiza-tions give that all their human subjects research will com-ply with the DHHS regulations whatever the fundingsource.

Unfortunately, it can be difficult to determine whetheran activity is only QI or both QI and human subjects re-search, given the ambiguity of the phrase “designed to de-velop or contribute to generalizable knowledge” in the reg-ulatory definition of research. The matter is of some con-sequence, given the sanctions that may be imposed on or-ganizations that violate the human subjects protection reg-ulations. This section considers approaches to review andsupervision of the activities that belong in this hybrid cat-egory. The section that follows discusses approaches totheir identification.

We begin with a brief discussion of the activities towhich the regulations apply (Figure 2). The regulationsdefine a human subject as “a living individual aboutwhom an investigator (whether professional or student)conducting research obtains data through intervention or

interaction with the individual, or identifiable private in-formation.” In the figure, the long oval is the set “Activi-ties Involving Human Data Sources,” where a human datasource is defined as “a living individual about whom a per-son conducting the activity obtains data through interven-tion or interaction with the individual, or identifiable pri-vate information.” (The awkward term “human datasource” is used rather than “human subject” because, asshown, the regulations define a human subject as someoneabout whom data is obtained for research purposes.) Theoverlap between this long oval and the circular Researchset (defined earlier for Figure 1) is the subset “HumanSubjects Research.” Within this subset, the smaller oval la-beled “Exempt” consists of human subjects research thatfalls into any of the categories of research that have beenofficially declared exempt from the regulations. (These in-clude, for example, some research in educational settingsusing normal educational practices; some research involv-ing tests, surveys, interviews, and public behavior observa-tions; some research using existing data and specimens;and some research by public benefit programs.42) The re-maining area is the subset of human subject research ac-tivities covered by the regulations.

Figure 3 combines the diagrams in Figures 1 and 2 toshow the cross-hatched subset “Nonexempt QI/HumanSubjects Research Overlap (Nonexempt QI/HuSR),” theset of QI activities that must be reviewed by an IRB andconform to other regulations, such as the requirements re-lating to informed consent. Although the figure does notshow it, the boundaries of this subset are currently fuzzy,given the lack of clarity about precisely which QI activitiesalso meet the definition of human subjects research.

Research QI Clinical &ManagerialInnovation & AdaptationResearch

on QI

QI/Research

on QI

Exempt HuSR

Activities InvolvingHuman Data Sources

QI/Research

Human Subjects Research (HuSR)

QI Involving Human Data Sources

NonexemptQI/HuSR

Figure 3.



How do we ensure that all of the hybrid activities con-form to the regulations, especially given the fuzziness ofthe boundary? Since the current IRB review process is notsuitable for most QI, simply sending all QI activities to ex-isting IRBs would not be a good approach. A second ap-proach would be to send no QI activities to IRBs, and in-stead authorize the development of a robust but separateinternal accountability system. This separate system wouldbe based on the ethical requirements developed in this re-port (Box 5) and designed for all types of QI, includingactivities that are both QI and human subjects research.The hybrid activities could then be exempted from theDHHS regulations but required to submit to this alterna-tive pathway. Unfortunately, modification of the existingDHHS regulations to allow this is a project that wouldtake years and is not assured of a satisfactory outcome.

A third approach, and the one we recommend, is fororganizations to establish specialized Quality Improve-ment IRBs (QI-IRBs) to review the hybrid activities. TheQI-IRB would be a committee that meets the minimumregulatory requirements for IRB composition and hasstandard operating procedures that conform to IRB regu-lations; however, it would design its process specifically tomeet the review and supervision needs of an activity thatis both QI and human subjects research. This approachseems compatible with the current regulatory framework.Although many IRBs have chosen to operate under rela-tively rigid rules and procedures, there is actually substan-tial flexibility in the regulations. OHRP’s Guidance onWritten IRB Procedures, July 11, 2002, summarizes the ac-tivities IRBs must carry out (the distribution of materialson research projects to IRB members, initial and continu-ing review of the materials, issuance of approval or disap-proval of the project, review of changes to the researchprotocol, and so on) but also says:

OHRP has not developed a model written IRB proce-dures document for institutions to adapt because proce-dures appropriately can vary significantly among institu-tions as the result of differences in institution size, thetype of research activities, institutional administrativepractices, number of IRBs, and local and state laws andregulations. For each required element, the written IRBprocedures should provide sufficient step-by-step opera-tional details so that an independent observer can un-derstand how an IRB operates and conducts its majorfunctions.

The minimum regulatory requirements for IRB composi-tion (shown in Box 6) allow IRBs to have as few as fivemembers and also to incorporate considerable flexibilityinto the specification of member qualifications. Thus thespecialized IRB—that is, an IRB whose members are cho-sen to have the expertise needed to review the specifictypes of research activities usually submitted to thatIRB—is compatible with the regulations.

� A minimum of five members with varying back-grounds must sit on the board.

�Members must not all be from one profession.

� At least one member must have primary concernsthat are “scientific” (a term the regulations do not de-fine).

� At least one member must have primary concernsthat are “nonscientific” (also left undefined).

� At least one member must be unaffiliated with theinstitution in which the IRB resides and not an im-mediate family member of a person affiliated withthe institution.

�Members cannot have a conflicting interest in anyproject that comes before the IRB for initial or con-tinuing review.

� The IRB may invite individuals with competencein special areas to assist in the review of issues that re-quire expertise beyond or in addition to that availableon the IRB, but these individuals may not vote withthe IRB.

� The IRB shall be sufficiently qualified through itsmembers’ experience, expertise, and diversity, includ-ing consideration of race, gender, and cultural back-grounds and sensitivity to such issues as communityattitudes, to promote respect for its advice and coun-sel in safeguarding the rights and welfare of humansubjects.

� The IRB shall be able to ascertain the acceptabili-ty of proposed research in terms of institutional com-mitments and regulations, applicable law, and stan-dards of professional conduct and practice, and shalltherefore include persons knowledgeable in theseareas.

� If an IRB regularly reviews research that involves avulnerable category of subjects, such as children, pris-oners, pregnant women, or handicapped or mentallydisabled persons, consideration shall be given to theinclusion of one or more individuals who are knowl-edgeable about and experienced in working withthese subjects.

6. Requirements for IRB MembershipFrom 45 CFR 46.107


Organizations could make use of the flexibility in theregulations to tailor the membership and process of theQI-IRB to the QI/HuSR activities conducted within theirorganizational environment. A QI-IRB should be inte-grated with the managerial and professional supervisorystructure of the organization and should have memberswho are thoroughly familiar with the local environment.A QI-IRB should also have members with expertise in thetheory and practice of QI and at least one nonaffiliatedmember who can provide an outside perspective. (Ofcourse, all members must also have a basic understandingof both the ethical requirements for QI and the human re-search protection regulations.) With the right member-ship and a flexible process for review and supervision inplace, it should be relatively straightforward for the QI-IRB to ensure that the QI/HuSR activities it reviews aredesigned to meet ethical requirements. As the discussionin Section II summarized in Box 5 indicates, the ethicalrequirements for QI and human subjects research havemuch in common.

The one significant difficulty lies in the strong pre-sumption of voluntariness on the part of researcher andresearch subject. When human subjects research is com-bined with an activity in which participation cannot becompletely optional, problems tend to arise in applyinginformed consent requirements. In the case of QI/HuSRactivities, the proper course is clear if the combined activ-ity presents more than minimal risk; specific informedconsent is ethically required for either QI or human sub-jects research activities of more than minimal risk. If thehybrid activity presents no more than minimal risk, how-ever, the research regulations and QI ethical requirementsseem to conflict, since specific informed consent is notethically required for a QI activity of that degree of risk.

In fact, informed consent is not quite an absolute re-quirement for research under the DHHS regulations. AnIRB may waive or alter the specific informed consent re-quirement if these four conditions all hold: (1) the re-search “cannot practicably be carried out” without thewaiver; (2) the subjects’ rights and welfare will not be ad-versely affected; (3) the research involves no more thanminimal risk; and (4) when appropriate, subjects will beprovided with additional pertinent information after par-ticipating. Unfortunately, we believe that at this time, re-search IRBs are likely to interpret these requirements nar-rowly (particularly the “not practicable” requirement) andbe unwilling to grant waivers of informed consent forQI/HuSR activities.

We believe that a waiver would generally be appropri-ate in cases in which the QI/HuSR activity represents nomore than minimal risk (and especially where strong a pri-ori evidence suggests that substantial benefits to subjectsare likely). First, remember that consent to QI is required,but takes the form of background consent to inclusion inroutine minimal risk QI activities, given as part of consentto receive health care from (or work for) the organization.

This consent is given with the understanding that confi-dentiality is protected and no identifiable participant in-formation from the QI activity is shared outside the orga-nization without participant authorization. Also, generalinformation about the QI program is available, and it isexpected that participants in a QI activity will be in-formed of any findings that are clinically relevant to theirown care. In other words, participants are already aware ofand have consented to their possible inclusion in minimalrisk QI with their rights and welfare protected. In thesecircumstances, the IRB might reasonably determine that itis unnecessary and even confusing to patients to requirespecific informed consent to a QI project that could bedone ethically without it, simply because the QI activityalso contributes to a minimal risk research endeavor.

In fact, the IRB could reasonably decide that requiringinformed consent would impose additional burdens, sinceobtaining specific informed consent has a cost in patienttime and energy. The costs to the organization might alsomake it impractical. Often, requiring specific informedconsent to a hybrid activity would be a poor use of re-sources needed for patient care, as well as a disincentive toan organization to add low cost, minimal risk design ele-ments to a QI activity that would allow it to serve a re-search purpose as well as bring about a local improvement.All in all, it seems reasonable for the IRB to interpret theterm “not practicable without the waiver or alteration” toapply, and thus to permit specific informed consent to bewaived for most minimal risk QI/HuSR activities inhealth care settings.

As in the local management review and supervision ofthe QI category, this category of QI-IRB review and su-pervision can—and should—be constructed and used indifferent ways in different organizations. The membershipcomposition and the operating procedures of the QI-IRBshould be appropriate to the nature of the organizationand the QI activities it conducts.

An organization that carries out traditional humansubjects research activities as well as QI/HuSR activitieswill probably find it worthwhile to establish both a QI-IRB and a separate IRB for research. The research IRBwould not need to be as integrated into ongoing manage-ment of the organization as the QI-IRB, could meet lessfrequently to review the fixed protocols typical of research,and would normally have a different composition in termsof scientific and statistical methodological expertise. Ifthere are separate IRBs, there should be excellent lines ofcommunication between them to ensure that their mem-bers have a basic understanding of the similarities and dif-ferences between the roles of the QI-IRB and the researchIRB within the organization.

Finally, each organization should have at least one per-son who is trained and authorized to sort activities—todecide which activities are required to go to the QI-IRB,which are required to go to the research IRB (if there isone), which are not covered by or are exempt from the


DHHS regulations altogether, and which do not fallunder the regulations but should nevertheless go to theQI-IRB because the organization deems them to need thekind of review and supervision the QI-IRB provides. It isnot an efficient use of an IRB or QI-IRB to use it to per-form this sorting task. The proposal to designate and trainone or more specific persons to perform the task is com-patible with the DHHS regulations.

Separating QI from Research and Defining the OverlapWe have proposed a specialized QI-IRB to review the

QI activities that are also human subjects research and aresubject to the DHHS regulations. Now we must addressthe vexing question of how to determine whether a QI ac-tivity falls into this category.

This report has defined QI as “systematic, data-guidedactivities designed to bring about immediate positivechanges in the delivery of health care in particular set-tings.” So understood, QI is just a systematic, data-guid-ed form of the clinical and managerial innovation andadaptation that has always been an integral part of clinicaland managerial practice. The fact that QI is a normalhealth care operation focused on improving local care hasbeen critical to our argument that ethical review of QIshould be incorporated into the system of accountabilityfor clinical care.

We have noted that in the process of making local im-provements, QI may produce information that is of use topeople in other settings. The regulations define research as“a systematic investigation, including research develop-ment, testing and evaluation, designed to develop or con-tribute to generalizable knowledge.” Depending on one’sinterpretation of the last clause, one might argue eitherthat nearly all QI or that no QI should count as research.We are persuaded that neither extreme is appropriate. QIreports can provide useful insights, just as individual pa-tient case reports, success stories, and cautionary tales pro-vide useful insights. Nevertheless, like these examples,most QI activities are not research as understood by thosewho framed the human subjects protection regulations.Nevertheless, some QI activities are genuine hybrids: sys-tematic investigations designed to bring about local im-provement and develop generalizable knowledge simulta-neously. Such activities are an extremely useful and cost-effective way to deal with quality and safety problems,and they must be encouraged if we want to make realprogress in improving the American health care system.

Drawing lines separating categories of activities is a dif-ficult task in a regulatory system. We have tried to shapeour proposals for ethical oversight of QI to minimize theneed for bright lines separating categories of activitieswhile remaining in accordance with current DHHS regu-lations. The category called local management and super-vision covers QI that requires oversight but is not humansubjects research (either because it does not meet the reg-ulatory definition of research or because it qualifies as re-

search but does not involve any people who meet the reg-ulatory definition of human subjects). The organizationitself sets the boundary between this category and the cat-egory of professional oversight of QI, and it draws what-ever lines it needs within the category to sort activities anddirect them to receive appropriate management and su-pervision.

Specialized QI-IRB review should eliminate some ofthe problems in having QI/HuSR activities reviewed byresearch IRBs; nevertheless, clarifying which QI activitiesare subject to the regulatory requirement of IRB review isstill important. Dithering over ambiguous classificationswastes time and tends to produce inconsistent decisionsthat undermine respect for the review system. Moreover,if organizations cannot tell which activities are in theQI/HuSR category, they may overload the QI-IRBs withactivities that don’t need to go there, just to be safe, whenthese activities could receive effective management andsupervision at lower cost within the purely QI oversightsystem.

Developing a clear rule for separating QI fromQI/HuSR can be seen as a conceptual task or as a practi-cal regulatory task, but of course it is both. What shouldone look for in a sorting rule? The rule should use easilyobserved aspects of an activity to determine whether it be-longs to the hybrid category—avoiding, for example, re-liance solely on the intent of the person initiating the ac-tivity and focusing on concrete elements in the activity’sdesign or context. The rule should be as consistent as pos-sible with the use of the word “research” in both commonlanguage and the regulatory definition, while openly ac-knowledging that some arbitrariness is inevitable in inter-preting the word to devise a practical rule for regulatorypurposes. Finally, the rule’s arbitrary lines should bedrawn so as to best serve the end goal of protectinghuman participants—from both the harm that might becaused by the activity and the harm caused by quality andsafety deficits in the health care system.

Some aspects of the history and purpose of human re-search protection shape the possibilities. The origins ofthe regulation of research to protect human subjects lie inrisky biomedical research and public outrage over inci-dents in which researchers conducted research on patientsand members of the public without their knowledge orconsent, or wasted public or shared resources, or took un-reasonable risks with little prospect of gain. What was be-fore the National Commission in the late 1970s was a se-ries of scandals and a growing research enterprise that hadshown itself to allow, at times, threats to the wellbeing ofits human subjects. In responding to this situation, theNational Commission saw as the domain of research notjust insights that were “generalizable” in that they wouldbe true of some other situations, but facts that are essen-tial to the nature of the human body and its reactions todrugs and treatments, as well as other facts that would beexpected to be true for all time and in all places. In its re-


port on the provision of health services by the Depart-ment of Health, Education, and Welfare, the Commissionhighlighted the importance of audit and improvement ac-tivities and contrasted these activities with research, not-ing that the practitioners in improvement activities didnot have the conflicts of interest that make human sub-jects vulnerable to abuse in research.43

Over time, this understanding of research has altered.The IRB-based human subjects protection system em-bodied in the DHHS regulations has been the only sub-stantial professional or governmental activity that aims toprotect human subjects. When questions on the scope ofresearch have been posed to the regulators, virtually anyactivity that might provide insights of use in another situ-ation has ended up being included in the category of re-search potentially subject to the regulations. This processhas engendered long-running controversies over, for ex-ample, the status of ethnography, oral history, and report-ing on public programs. The shift toward a very broad in-terpretation of research is also seen in recent OHRP asser-tions that public health surveillance is generally research,and likewise in tentative OHRP determinations that re-search includes administering surveys to clients as re-quired under the Government Paperwork Reduction Act,which establishes practices to limit fraud and to ensuregovernment contractors’ accountability.

Given the tendency to interpret research broadly andthe severity of the penalties available to OHRP and tothose who enforce the HIPAA Privacy Rule, organizationshave been afraid to develop their own guidelines for sort-ing activities into the categories of QI, research, andQI/HuSR. Though many parties know that improvementactivities are increasingly important and that impositionof IRB requirements is a major barrier to their implemen-tation, each organization is best served by avoiding anydispute with OHRP and any risk of penalty underHIPAA. At the same time, OHRP has shown little incli-nation to provide highly specific guidance on how to dothe sorting.

Given the great variety and the significant change cur-rently under way in the kinds of QI and research beingconducted in clinical settings (not just clinical research,but organizational, managerial, and health services re-search), this may be a prudent decision on OHRP’s part.It would be very difficult to produce consistent, practicalguidance at the regulatory agency level. We believe thatsuch guidance would be more easily developed by com-bining reflection on the nature of research and the aims ofthe human research protection system with the practicalexperience in managing QI and QI/HuSR activitiesgained by a variety of organizations. The goal would be aninterpretation of the definition of human subjects researchthat enabled QI and QI/HuSR activities to receive thetype of review and supervision needed to ensure ethicalconduct without imposing excessive costs.

Below, we give an initial delineation of the divides, dis-cussing both the kinds of QI activities that should countpresumptively as research, and those that should not. Wewant to underscore that learning about these dividesshould entail some deliberate innovation with reportingcases in the professional literature and having discussionsin policy settings. It is not likely that we will have hit uponjust the right dividing characteristics, or that we have ar-ticulated them well enough for implementation withoutcase studies.

For the purpose of the DHHS regulations, we proposethat the category of research be made up of activities thatare designed to learn something enduring about the natureand function of human beings and their environment. Re-searchers might also be clinical service providers, but theirresearch role is to be kept distinct and identifiable to pa-tients. Although the word “research” is used in more in-formal ways, such as “library research for a high schoolterm paper” or “research by an investigative journalist,” wethink this definition is appropriate for implementation ofthe DHHS regulations. The proposed narrative definitionaccords with the regulatory definition but gives voice to

what could be meant by “designed to develop or con-tribute to generalizable knowledge.”

Under this interpretation, most QI is not research. QIis designed to bring about the immediate improvement ofcare in local settings, and most of the urgently needed QIactivity involves changes in practice that are clearly withinthe standard of care—often moving from “dangerouslysubstandard” and “barely acceptable” practices to “better”and “best” practices. As a result, QI activities are generallybased on knowledge about the enduring “nature and func-tion of human beings and their environment,” rather thandesigned to create new understandings in this regard.Therefore, even though the activities represent “systemchanges,” these changes are within the discretion of clini-cians and managers to make. In fact, they have an ethicaland sometimes also a legal duty to make such changes.Changes of this type include adoption of evidence-basedpractice guidelines and best practices issued by respectedmedical authorities, introduction of process modificationsdesigned to reduce the probability of medical error, andreplacement of a practice that is dangerously below thestandard of care with one that is acceptable.

�Research� refers to activities designed to

learn something enduring about the nature

and function of human beings and their

environment. Under this interpretation, most

QI is not research.


Introducing these changes may require adjustments inpractice over and above what is specifically mentioned ina description of the new practice, and the adjustments arelikely to differ from one organization to another. QI isabout “moving fast but checking on progress as you go”—identifying the adjustments that may be required, trackingwhat happens as practices change, and being alert for un-expected effects—and thus may be the best way to protectpatients as the system adjusts to the change. Ensuring thatthis very specific and local knowledge is developed in amanner that appropriately protects patients is the essenceof QI. It is not an activity “designed to develop or con-tribute to generalizable knowledge,” and in fact, the ideathat implementing a new evidence-based guideline in alocal delivery setting is a research project requiring IRB re-view seems indefensible.44 This type of activity would notrequire review by a QI-IRB or a research IRB; it wouldgenerally belong in the category of “local management re-view and supervision of QI.”

Nevertheless, some QI activities really do have a re-search component in the sense that they are designed tolearn something enduring about the nature and functionof human beings and their environment in addition toimproving care in the local setting. Perhaps a project is de-signed to improve compliance with guidelines regarding aparticular treatment, but includes design elements thatallow the formal testing of a few strategies to gain thatcompliance, such as having some sites use QI (while oth-ers use other methods). Perhaps a project employs a newtreatment that has not yet been fully accepted as the stan-dard of care, but embeds that treatment in a QI processaimed at bringing about compliance with other aspects ofcare. Projects like these might well require review as re-search, as well as review as QI.

What characteristics of projects would make it likelythat the project is both QI and research and should be re-viewed by a QI-IRB? Here is a starter set.

� Randomization of patients into different interventiongroups in order to enhance confidence in differencesthat might be obscured by nonrandom selection (butnot randomization done to achieve equitable allocationof a scarce resource)

� Testing of issues that are beyond current science andexperience, such as new treatments

� The involvement in key project roles of researcherswho have no ongoing commitment to improvement ofthe local care situation (and who may well have conflictsof interest with the patients involved), even if others onthe team do have professional commitments

� Delayed or ineffective feedback of data from moni-toring the implementation of changes, especially if feed-

back is delayed or altered in order to avoid biasing theinterpretation of data

� Funding from an outside organization with a com-mercial interest in the use of the results

As outlined below in the section on strategies for imple-mentation, this initial list deserves to be tested in somepractical settings, aiming to modify it to fit the actual risksand projects that arise. Organizations with substantial in-vestment in overlap projects might well be among thosebest situated to develop QI-IRBs, which are describedabove.

Ethical Oversight for Collaborations across OrganizationsMulti-organization QI activities, such as QI collabora-

tives, often combine QI activities focused on bringingabout change in individual organizations into a larger pro-ject with an overarching goal of sharing informationabout the experience of bringing about positive change.The collaborative structure may provide for the initialtransfer of technical information about a new practice to aQI team in each organization, group discussion of howbest to introduce it using QI methods, and periodic groupmeetings to boost morale and exchange informationabout failures and successes.

The initiators of the collaboration should take the re-quirements for ethical conduct of QI into account in thedesign of the entire enterprise. In deciding whether to par-ticipate, each organization should subject the QI activityat the core of the collaboration to its internal QI account-ability process (assigning it to the appropriate category ofreview and supervision). The aim is to ensure that the ac-tivity is consistent with organizational goals and is de-signed so that participants within the organization areproperly protected.

The main additional issue raised by collaborating withother organizations concerns the sharing of informationabout the QI activity outside the organization. Care mustbe taken to ensure patient confidentiality and conformityto HIPAA rules. The initial transfer of technical informa-tion to participants can include assistance in clarifying therequirements for ethical conduct of the activity, includingthe management of protected health information in thesharing of results. If patient information might sometimesbe shared, or if follow-back to identify patients is plausi-ble, then the affected participants should be bound tohonor confidentiality through a Business Associate Agree-ment or an Organized Health Care Arrangement.45

Typically, QI collaboratives generate substantial insightabout the effective strategies for QI around a given topic,but this does not automatically mean that the collabora-tive project amounts to research involving human subjectsand is therefore subject to IRB review. In general, it doesnot even make it research, since the insights are still tied toplace and time and are acquired while pursuing improve-


ment in the clinical services. Even when the design doeshave some characteristics of research (such as a startingagreement to implement some key changes in a staggeredfashion so that their effects can be identified), the fact thatthe organizations cooperate to share insights about theprocess of change within each organization would usuallymake any research component an instance of research onorganizational behavior, not research on human subjects.

External Accountability forEthical Conduct of QI

The established accountability approaches used by thehealth care system to measure, identify, and demon-

strate quality and safety in the delivery of clinical careshould be used to hold health care professionals, man-agers, and organizations accountable for the ethical con-duct of QI.46

A first line of accountability is through the profession-al organizations of those who work in health care—physi-cians, nurses, health care managers. These organizationsshould inform their members about their professionalobligation to improve quality, identify the basic QI skillsthey should have, and ensure that they understand anduphold the standards for ethical conduct of QI. Qualityimprovement can be incorporated into professionalCodes of Ethics. For example, the American Nurse Asso-ciation Code of Ethics states that nurses should “partici-pate in establishing, maintaining, and improving health-care environments and conditions of employment con-ducive to the provision of quality health care and consis-tent with the values of the profession through individualand collective action.” The Code of Ethics of the Ameri-can Medical Association offers as one of nine core princi-ples the rule that “A physician shall continue to study,apply, and advance scientific knowledge,” and on the basisof this statement, AMA policy calls on physicians andtheir organizations to “(1) strive continuously to improvethe quality of health care; (2) encourage the ongoing eval-uation of continuous quality improvement models; (3)promote implementation of effective quality improve-ment models; and (4) identify useful approaches for as-sisting physicians in implementing quality improvementprocedures in their medical practices and office manage-ment.”47

Accountability at the organizational level can be fos-tered through the accrediting bodies that set standards forhealth care organizations, assess compliance with thosestandards, and in some cases focus on the operation andeffectiveness of internal quality improvement systems. Insome areas, state and federal governments rely on or rec-ognize private accreditation to ensure compliance with li-censure or regulatory requirements. To encourage ac-countability for the ethical practice of quality improve-ment, the standards of the major private organizationssuch as the National Committee for Quality Assurance

(NCQA), which accredits managed care plans, and theJoint Commission on Accreditation of Healthcare Orga-nizations (JCAHO), which accredits most types of healthcare organizations, could be modified to add to their re-view of the conduct of QI in organizations a review of theextent to which the organization has procedures in placeto ensure that the QI meets ethical requirements. Eventu-ally, outcome and process indicators for ethical qualityimprovement could be part of a set of regularly collectedquality measures.

Public sector mechanisms for accountability includethe federal HIPAA Privacy Rule and other federal andstate statutory privacy laws that regulate the protection ofidentifiable health information. The HIPAA Privacy Rulerequires that “covered entities”—persons and organiza-tions that acquire, use, disclose, or store health informa-

tion—protect the individually-identifiable health infor-mation they create or receive by establishing and adheringto privacy protections. These protections include provid-ing notice to patients about their privacy rights and theentity’s policies regarding disclosure of identifiable infor-mation, implementing internal privacy policies and pro-cedures, establishing safeguards to protect data privacy,training employees to understand privacy laws, and assist-ing health consumers to exercise their rights under thePrivacy Rule. The Privacy Rule allows covered entities touse protected health information to perform activities de-fined as normal health care operations provided specifiedprocedures are followed and includes QI in this category.Research is not included in normal health care operationsand is subject to different standards regarding the use anddisclosure of identifiable health information.

Some authors have argued that privacy risks may bethe key potential harm to patients from quality improve-ment projects. The broad applicability of the Privacy Ruleacross the health care system could protect patients fromprivacy breaches when their data are used for quality im-provement initiatives. The HIPAA Privacy Rule includesprovisions for the protection of data from unauthorizedreview, requires that only the “minimum necessary” databe used, and introduces HIPAA privacy boards to moni-tor the use of protected health information within organi-zations. It has been suggested that a HIPAA privacy boardcould adequately review QI projects and even research

The established approaches used by the

health care system to ensure quality and

safety in clinical care should be used to

establish accountability for the ethical

conduct of QI.


Improving the quality and safety of American healthcare is of vital importance. Health professionals, man-agers, health care delivery organizations, patients, and

government all have an ethical responsibility to cooperatewith one another in pursuing this goal. In this report, wehave discussed the ethical issues raised by the use of QImethods to improve health care. Using the principles ofresearch ethics that underlie the current human researchprotection system as a point of departure, we have definedthe requirements for ethical conduct of QI activities andthen considered the institutional arrangements needed toensure that QI conforms to them. We have discussed theuse of IRBs for ethical oversight of the conduct of QI andconcluded that IRBs as currently constituted are not ap-propriate for this purpose. Instead, we argue that ethicalreview of QI activities for purposes of protection ofhuman subjects should be integrated into a trustworthyaccountability system for managing the quality of clinicalcare.

We recommend that the primary responsibility for theethical conduct of QI be lodged in individual organiza-tions, where it should be seen as a normal clinical careobligation and integrated into supervision and manage-ment, with the organizations’ leaders responsible for over-seeing the integration. We call this “internal accountabili-ty” for the ethical conduct of QI, and we distinguish threebroad categories of oversight: professional responsibilityfor QI; local management review and supervision of QI;and QI-IRB review for QI that is also human subjects re-search. We recommend that there also be external ac-countability for the ethical conduct of QI in the form ofprocedures to ensure that individual health care organiza-tions have well-designed, functioning internal practices inplace to carry out their QI-related responsibilities. Theseprocedures should be part of the overall system of holdinghealth care providers accountable for the quality and safe-ty of clinical care.

We have recommended a variety of specific steps bywhich the ethical conduct of QI might be ensured, butthese recommendations bear summarizing.

� Clarify professional and organizational obligations to-ward QI

Organizations of health professionals (physicians, nurs-es, health care managers, and so on) should inform mem-bers about their professional obligation to improve quali-ty. They should also identify the basic QI skills theirmembers should have, educate members about standardsfor ethical conduct of QI, and even incorporate qualityimprovement into professional codes of ethics. Organiza-tions of health care provider organizations (hospitals,nursing homes, health plans, and so on) should educatetheir members about their quality improvement obliga-tions, the need to ensure that their employees have basicQI skills, and the standards for ethical conduct of QI.Leaders in professional education should press for greateremphasis on the obligations of health professionals towardthe quality of care and the development of QI skills in ed-ucational curricula.

� Clarify patients’ obligations toward QIWhen people seek health care, they should be told

about the importance of QI activities for maintaining andimproving the quality of their care and informed that con-sent to receive care includes consent to a minimum levelof cooperation with ongoing QI. They should be givenbasic information about the organization’s QI programand told how they can obtain more information if theywant it.

Health care organizations should develop educationalmaterials for their patients on the conduct of QI withinan organization and on patient rights and responsibilitieswith respect to QI. Organizations, such as JCAHO andthe Institute for Healthcare Improvement (IHI), and asso-ciations of organizations, such as the American HospitalAssociation (AHA), might assist with this by promotingthe development and dissemination of model educationalmaterials that individual organizations could adapt fortheir own use.

IV. Implementation

projects for which the only significant risk to patients wasfrom a breach of privacy and/or confidentiality.

For those QI activities that fall into the QI/HuSRoverlap category, the federal human subject protectionregulatory system acts as an external accountability mech-anism for the specialized QI-IRBs proposed as part of theinternal accountability structure for the conduct of QIwithin organizations.

Finally, the health care market may enforce a degree ofaccountability through the ability of purchasers to makethe observance of ethical standards in the conduct of QI a

factor in their choice of provider. This effect is limited,however, since the information available to patients is alsolimited, and they rarely have full control over decisionsabout the care that is purchased on their behalf. Marketforces are most likely to have an effect when they actthrough third party purchasers. Health care purchasinggroups and other payers could encourage quality im-provement and its ethical practice by including outcomeand process standards for the ethical practice of QI in thecontracts they make with providers.

� Develop guidance on QI methodologyQI uses—and should use—a somewhat different range

of methodologies from traditional research. Moreover, as arelatively newly rediscovered field of endeavor, QI’s meth-ods are in a state of rapid evolution. QI practitioners needongoing guidance on appropriate methodological stan-dards in order to meet the ethical requirement that QI ac-tivities be properly structured to achieve their goals.Groups that are already active in developing and promot-ing QI and educating people about QI methods, such asIHI and AHRQ, are well placed to perform this function.

� Develop new models of management and supervision ofQI and QI/HuSR

We have given a rough outline of the internal account-ability arrangements that should be in place to ensure thatQI and QI/HuSR activities receive appropriate oversight.The next step is to translate this rough outline into mod-els that will work in real health care settings. We suggestthat one promising way to move this forward is throughcollaborative efforts to develop such models by organiza-tions that are leaders in introducing QI methods into theiroperations.

OHRP could take the lead in encouraging some lead-ership organizations to undertake the task, or the leader-ship organizations could take the initiative themselves.IHI might be able to play a coordinating role, perhapsalong the lines of the role it plays in its “BreakthroughCollaboratives.” The goal would be for the individual or-ganizations to develop systems of internal managementand supervision for their QI activities, establish QI-IRBsfor activities that are both QI and human subjects re-search, devise practical rules for identifying the activitiesthat should be reviewed by QI-IRBs, and develop QI-IRBreview processes tailored to the types of QI/HuSR carriedout in their organizations. As they launched these efforts,the organizations could share their experiences with oneanother regularly, engaging in robust discussion of the eth-ical issues they encounter and the practical wisdom theygenerate. Ideally, organizations such as OHRP, AHRQ,and JCAHO would also be involved in this endeavor.

A federal agency could also undertake to lead an effortto develop practical rules and models, under the exemp-tion in the DHHS regulations for public benefit pro-grams. This effort could be instead of, in addition to, oreven in collaboration with an effort by private organiza-tions. The obvious candidate would be the Centers forMedicare and Medicaid Services, since it sponsors QI andresearch and has accepted the mission of improving carefor Medicare and Medicaid patients. However, the HealthResources and Services Administration (HRSA), the De-partment of Veterans Affairs, and the Department of De-fense could also carry out such efforts under the publicbenefit agency exemption.

The consensus of this work should be clearly articulat-ed from time to time, as the categories become more set-

tled and the interpretations of the language of the DHHShuman subjects regulations become clearer. This workwould be greatly enhanced by robust discussion in theprofessional literature and at meetings. The topics dis-cussed would include the criteria used to sort projects, themerits of particular methods of review and supervision,and actual case studies that illuminate the work.

The cooperation of OHRP would be very important,especially for efforts to develop practical guidance on theclassification of activities as QI or QI/HuSR. One mightexpect that OHRP would be especially anxious about sucha process, since it has been a long practice to extend thescope of research broadly. However, keeping OHRP in-volved and reflecting upon the actual situations that arisewould be helpful in developing sound public policy. Hav-ing AHRQ or another agency sponsor some ongoing de-scriptive research into the nature of the projects and thereview approaches that are implemented would help guidethe process. If some adverse occurrences arise during thisprocess of exploration and consensus building, then a co-ordinated response among the concerned parties wouldhelp to ensure that all the issues are carefully weighed andthat overly rigid rulings are not made and transformedinto permanent precedent.

As consensus develops on promising models for themanagement and supervision of QI and QI/HuSR and onrules for sorting activities into the appropriate category,IHI, JCAHO, OHRP, the various trade organizations,and professional journals should disseminate the results.Public Responsibility in Medicine & Research(PRIM&R) could also help to educate IRBs and the re-search community in general on the similarities and dif-ferences between QI and research, and the standards forethical conduct of QI. In particular, PRIM&R could helpthe IRB community to understand the developing con-sensus on the management of activities that are both QIand human subjects research, and ensure ongoing consid-eration of the issues such hybrid activities raise through itsAnnual Human Research Protection Program (HRPP)Forum.

� Develop and expand external accountability for QIAccrediting bodies such as JCAHO and NCQA have

already begun to include quality improvement require-ments in the accreditation process. They should continueto develop and expand these efforts to include review ofthe extent to which organizations have effective mecha-nisms in place for managing QI and ensuring that it meetsethical requirements in the accreditation process.

� Disseminate QI resultsThe ethical analysis in this report supports the conclu-

sion that the intent to publish (or actual publication) doesnot make a QI activity into a research project, nor does itestablish that IRB review is ethically required. The analy-sis also implies that QI results can have social value outside



Although the findings and conclusions we have ar-rived at during the course of our research on ethi-cal issues and health care quality improvement are

complex and nuanced, they can be summarized as follows:

� QI is any systematic, data-guided activity that is de-signed to bring about the immediate improvement ofcare in a local setting.

� QI is both appropriate and vital to health care.

�QI is marked most distinctly by the prompt feedbackof the effects of deliberate changes to the same care de-livery setting that is making the changes.

� QI is intrinsic to health care delivery and obligatoryfor both professionals and patients.

� Though QI is often driven by a priori evidence thatsuggests substantial benefits are likely for the patientsand/or staff involved, QI can pose some risks to somepatients.

� Not undertaking QI in the face of recognized qualitydeficiencies also puts patients at risk.

� QI should itself be implemented ethically.

� Low-risk QI should generally have the same reviewand standards as routine health care delivery.

� Higher-risk QI should undergo routine and orderlyreview within the usual arrangements for clinical super-vision or by an advisory group.

� Some projects are correctly counted both as QI and asresearch involving human subjects and should meet therequirements for review of protection of human subjectsin research.

� Meeting those requirements might be more readilyaccomplished with a QI-IRB that met regulations, butwhose policies and procedures were also tailored to theneeds and expectations of QI.

� Certain issues might trigger the requirement for for-mal review of a proposed QI project: randomized de-signs, novel treatments, involvement of researchers, de-layed feedback of monitoring, or external funding.

� Federal agencies and voluntary organizations shouldcooperate in further developing and implementing theseideas.

Resolving the uncertainty about what is ethically andlegally required to safeguard participants in QI activities isessential if QI methods are to be available to help trans-form the culture of American health care delivery into aculture committed to continuous quality improvement.In this report, we have outlined what must be done toprotect patients’ and staff members’ rights to privacy, in-formed consent, and justice when carrying out QI activi-ties. The challenge now is for the regulatory communityand health care management and professionals to put theappropriate standards and practices for QI in place. Thegoal—a better and safer health care system—is worth theeffort.

the setting in which they were developed, and therefore,dissemination by a variety of means should be encour-aged. There has been considerable confusion in the pastabout the relationship of publication to IRB review andthe designation of QI as research; however, at this point,OHRP and others now seem to be in agreement withthese conclusions.48

The next step is to develop strategies that facilitate thesharing of QI results. For example, publication of QI re-sults should be encouraged, and the results should be pub-lished in a standardized format so that it is easier to com-pare different QI activities and evaluate the extent towhich the conduct of the activities met ethical require-ments.49 By allowing wider scrutiny of the methods usedin QI, removal of the current bias against publication of

QI articles would also promote the development and clar-ification of QI methodological standards.

Editors of medical and other health care-related jour-nals could adopt a standardized format for the reportingof QI activities for publication, encourage the submissionof articles on QI methods and results, and become edu-cated about what falls outside of the QI/HuSR range sothat they do not reject articles that report QI activities justbecause they were conducted without informed consentor IRB review. AHRQ, CMS, and other federal agenciescould develop and implement ways of providing financialsupport for the conduct of QI, the dissemination of re-sults, and the development of QI methods, and provideongoing guidance on the application of HIPAA privacyrules to QI activities that involve collaboration across or-ganizations.

V. Conclusion


References

1. F. Davidoff and P. Batalden, “Improv-ing Improvement: The Educational Chal-lenge for Medicine,” manuscript submittedfor publication.

2. Because fourteen other federal depart-ments and agencies have codified Subpart Ain their own departmental regulations, it isknown as the “Common Rule.” An addi-tional agency (the Central IntelligenceAgency) is required by executive order tocomply with all parts of 45 CFR 46. In ad-dition to these human subjects protections,some federal agencies have issued regula-tions specific to human research conductedunder their jurisdiction (for example, theEnvironmental Protection Agency regardingpesticide research), and several states havestatutes and/or regulations governing re-search with humans (such as the CaliforniaHealth and Safety Code).

3. See M.P. O’Kane, “Do Patients Needto be Protected from Quality Improve-ment?” background paper prepared for TheHastings Center project, “The Ethics of Im-proving Health Care Quality and Safety.”

4. K.N. Lohr, ed., Medicare: A Strategy forQuality Assurance (Washington, D.C.: Na-tional Academy Press, 1994), quoted in In-stitute of Medicine, Crossing the QualityChasm (Washington D.C.: National Acade-my Press, 2001), 232; see also Institute ofMedicine, Envisioning the National HealthCare Quality Report (Washington D.C.: Na-tional Academy Press, 2001).

5. See IOM, Crossing the Quality Chasm,5-6.

6. Ibid.7. See IOM, Envisioning the National

Health Quality Report, 9. The committeenotes that “specific causes of inefficiency,such as repeat procedures due to error,overuse, fragmentation of care, and unneces-sary delays, are included under the appropri-ate component of quality.”

8. The issue of changes that involve trade-offs between quality and cost—that is, ac-cepting a lower level of quality in exchangefor lower resource costs—is discussed laterin this report.

9. M.A. Schuster, E.A. McGlynn, andR.H. Brook, “How Good Is the Quality ofHealth Care in the United States?” MilbankQuarterly 76 (1998): 517-63.

10. E.A. McGlynn et al., “The Quality ofHealth Care Delivered to Adults in theUnited States,” New England Journal ofMedicine 348 (2003): 2635. Results basedon data collected between 1998 and 2000.

11. B.D. Smedley, A.Y. Stith, and A.R.Nelson, eds., Unequal Treatment: Con-fronting Racial and Ethnic Disparities inHealth Care (Washington, D.C.: NationalAcademy Press, 2003).

12. Agency for Healthcare Research andQuality, 2003 National Healthcare QualityReport (Rockville, Md.: U.S. Department ofHealth and Human Services, 2003).

13. Agency for Healthcare Research andQuality, 2004 National Healthcare QualityReport (Rockville, Md.: U.S. Department ofHealth and Human Services, 2004); Agencyfor Healthcare Research and Quality, 2005National Healthcare Quality Report(Rockville, Md.: U.S. Department ofHealth and Human Services, 2005).

14. Agency for Healthcare Research andQuality, 2003 National Healthcare Dispari-ties Report (Rockville, Md.: U.S. Depart-ment of Health and Human Services,2003); Agency for Healthcare Research andQuality, 2004 National Healthcare Dispari-ties Report (Rockville, Md.: U.S. Depart-ment of Health and Human Services,2004); Agency for Healthcare Research andQuality, 2005 National Healthcare Dispari-ties Report (Rockville, Md.: U.S. Depart-ment of Health and Human Services,2005).

15. C. Schoen et al., “Taking the Pulse ofHealth Care Systems: Experiences of Pa-tients with Health Problems in Six Coun-tries,” Health Affairs 24 (2005): w509-w525(published online November 3, 2005;10.1377/healthaff.w5.509).

16. Professional societies have recentlymade this clear. See ABIM Foundation,ACP-ASIM Foundation, and the EuropeanFederation of Internal Medicine, “MedicalProfessionalism in the New Millennium,”Annals of Internal Medicine 136 (2002):243-46; Institute of Medicine, Health Profes-sionals Education: A Bridge to Quality (Wash-ington D.C.: National Academy Press,2003); and the description of “the six com-petencies” developed by the AccreditationCouncil for Graduate Medical Education(ACGME) and endorsed by the AmericanBoard of Medical Specialties (ABMS) at:http://www.acgme.org/outcome/comp/compFull.asp.

17. See President’s Commission for theStudy of Ethical Problems in Medicine, Se-curing Access to Health Care, Volume One: Re-port (Washington D.C.: U.S. GovernmentPrinting Office, 1983).

18. See M. Wynia et al., “Physician Ethicsand Participation in Quality Improvement:Renewing a Professional Obligation,” back-ground paper prepared for The HastingsCenter project, “The Ethics of ImprovingHealth Care Quality and Safety.”

19. N. Lang, “Healthcare Quality Im-provement: A Nursing Perspective,” and G.Agich, “Healthcare Organization Responsi-bility for Quality Improvement,” back-ground papers prepared for The HastingsCenter project, “The Ethics of ImprovingHealth Care Quality and Safety.”

20. We acknowledge that the UnitedStates has so far failed to actually fulfill thisobligation. Nevertheless, the ongoing debateabout health care reform and the existenceof programs like Medicare, Medicaid, StateChildren’s Health Insurance Programs, andother programs that provide assistance tovarious categories of the poor are evidence ofrecognition of an obligation, although thereis no consensus about the extent of it or thepublic policies that should be used to fulfillit.

21. See N. Dubler et al., “Informed Par-ticipation in the QI Process,” backgroundpaper prepared for The Hastings Centerproject, “The Ethics of Improving HealthCare Quality and Safety.”

22. See IOM, Crossing the Quality Chasm,5-6.

23. The argument here contrasts with theposition in the ethics of QI in which greatimportance is attached to the extent towhich an individual QI participant, or a ma-jority of participants, benefits directly fromthe specific QI activity. See D. Casarett, J.H.Karlawish, and J. Sugarman, “Determiningwhen Quality Improvement InitiativesShould Be Considered Research: ProposedCriteria and Potential Implications,” Journalof the American Medical Association 283(2000): 2275-80.

24. See Department of Health andHuman Services, “Code of Federal Regula-tions: Protection of Human Subjects,” June23, 2005, http://www.hhs.gov/ohrp/humansubjects/guidance/45cfr46.htm#46.102, section 46.102(d). Emphasis added.

25. See Dubler et al., “Informed Partici-pation in the QI Process.”

26. Research on the quality improvementprocess is sometimes referred to as “QI Re-search,” but we prefer the terminology of“Research on QI” to distinguish this setclearly from the set, “QI/Research,” theoverlapping region of the QI and Researchsets.

27. See E.J. Emanuel, D. Wendler, andC. Grady, “What Makes Clinical ResearchEthical?” Journal of the American Medical As-sociation 283 (2000): 2701-2711. We usedthe framework of this article because it is abrief, readily accessible, recent synthesis ofgenerally accepted ideas in research ethicsand includes all the topics we wish to discussin this report. We recognize that it is a varie-gated list of topics consisting of fundamen-tal ethical principles and both substantiveand procedural norms. For an extended,more rigorous discussion and classificationof these topics, see R.J. Levine, Ethics andRegulation of Clinical Research, second edi-tion (New Haven, Conn.: Yale UniversityPress, 1988).

28. The questions in italics with each re-quirement paraphrase the questions posedfor research in Emanuel, Wendler, and

Grady, “What Makes Clinical Research Eth-ical?” 2701-2711.

29. See F. Davidoff, “Publication and theEthics of Quality Improvement Projects,”background paper prepared for The Hast-ings Center project, “The Ethics of Improv-ing Health Care Quality and Safety.”

30. For an entertaining discussion relatedto this point, see the satirical article byG.C.S. Smith and J.P. Pell, “Parachute Useto Prevent Death and Major Trauma Relat-ed to Gravitational Challenge: SystematicReview of Randomised Controlled Trials,”British Medical Journal 327 (2003): 1459-61, and responses by M. Diaz and D.Neuhauser, “Pasteur & Parachutes,” Quality& Safety in Health Care 14 (2005): 140-43.

31. Emanuel, Wendler, and Grady,“What Makes Clinical Research Ethical?”2705.

32. In practice, the DHHS regulationsallow informed consent to be waived forminimal risk research provided that severalother stringent requirements are met. Suchwaivers are considered incompatible withethical principles by some ethicists, howev-er.

33. See Dubler et al., “Informed Partici-pation in the QI Process.”

34. See discussion of “notification” inLevine, Ethics and Regulation of Clinical Re-search, 178-81.

35. See K. Maschke, “Health Care Qual-ity Improvement Activities and the HIPAAPrivacy Rule,” background paper preparedfor The Hastings Center project, “TheEthics of Improving Health Care Qualityand Safety.”

36. See J. Lynn, “When Does QualityImprovement Count as Research?” Qualityand Safety in Health Care 13 (2004): 67-70;and R.J. Levine, “Institutional ReviewBoards: A Crisis in Confidence,” Annals ofInternal Medicine 134 (2001): 161-63.

37. See Institute of Medicine, ResponsibleResearch: A Systems Approach to ProtectingResearch Participants (Washington D.C.:National Academy Press, 2002).

38. The IRB is required to regularlymonitor the research. For example, it must

“receive and review reports of local, on-siteadverse events and unanticipated problemsinvolving risks to subjects or others and anyother information needed to ensure that itscontinuing review is substantive and mean-ingful. In addition, institutions and IRBsmay require additional information for con-tinuing review at their discretion.” See Of-fice for Human Research Protections,“Guidance on Continuing Review,” July 11,2002, http://www.hhs.gov/ohrp/human-subjects/guidance/contrev2002.htm.

39. One might object that a review car-ried out by the organization doing the QI isnot an independent review. Note, however,that the research ethics requirement is forreview by individuals unaffiliated with theresearch project, not review by individualsunaffiliated with the organization in whichthe research is carried out. Under currentrules, IRBs must have at least one memberunaffiliated with the organization, but therest can be affiliated. Moreover, the name“institutional review board” reflects theoriginal concept of IRBs as entities responsi-ble for carrying out an institution’s obliga-tion to protect the human subjects in the re-search conducted under its auspices.Human subjects protection was deliberatelylocated at the institutional level so that re-viewers would have knowledge of local con-ditions and the opportunity for ongoingoversight (limited though it is). In otherwords, locating primary responsibility forthe protection of human participants in QIwithin the organization doing the QI is nota significant departure from independentreview as it currently occurs in research.

40. A. Gauthier and M. Serber, A Need toTransform the U.S. Health Care System: Im-proving Access, Quality, and Efficiency (NewYork: The Commonwealth Fund, 2005).

41. M. Bottrell, “Accountability for theConduct of Quality Improvement Pro-jects,” background paper prepared for TheHastings Center project, “The Ethics of Im-proving Health Care Quality and Safety.”

42. The regulations are complicated anddifficult to summarize briefly or reduce to asimple diagram. For more details, see Officeof Human Research Protections, “Human

Subjects Regulations Decision Charts,” Sep-tember 24, 2004, http://www.hhs.gov/ohrp/humansubjects/guidance/decision-charts.htm.

43. The National Commission for theProtection of Human Subjects of Biomed-ical and Behavioral Research, Ethical Guide-lines for the Delivery of Health Services byDHEW: Report and Recommendations(Washington, D.C.: U.S. GovernmentPrinting Office, 1978).

44. Note, however, the survey in P.K.Lindenauer et al., “The Role of the Institu-tional Review Board in Quality Improve-ment: A Survey of Quality Officers, Institu-tional Review Board Chairs, and JournalEditors,” The American Journal of Medicine113 (2002): 575-77, showing that manychairs of IRBs thought such an activityshould receive IRB review.

45. See K. Lawlor, “OHCAs and Collab-orative Quality Improvement Projects,”background paper prepared for The Hast-ings Center project, “The Ethics of Improv-ing Health Care Quality and Safety.”

46. Bottrell, “Accountability for the Con-duct of Quality Improvement Projects.”

47. See American Nurses Association,Code of Ethics for Nurses with InterpretativeStatements, June 30, 2001, http://www.nursingworld.org/ethics/chcode.htm;American Medical Association Council onEthics and Judicial Affairs, Code of MedicalEthics (Chicago, Ill.: American Medical As-sociation, 2004-2005).

48. Statement by Michael Carome, page158 of transcript, Secretary’s Advisory Com-mittee on Human Research Protections(SACHRP) Meeting, October 5, 2004,h t tp : / /www.hhs . gov /ohrp / s a ch rp /mtgings/mtg10-04/oct0405.txt.

49. See Davidoff, “Publication and theEthics of Quality Improvement Projects,”and F. Davidoff and P. Batalden, “TowardStronger Evidence on Quality Improve-ment. Draft Publication Guidelines: TheBeginning of a Consensus Project,” Quality& Safety in Health Care 14 (2005): 319-25.


ABIM Foundation, ACP-ASIM Foundation, and theEuropean Federation of Internal Medicine, “Medical Profes-sionalism in the New Millennium,” Annals of Internal Medi-cine 136 (2002): 243-46.

Agency for Healthcare Research and Quality, NationalHealthcare Quality Report (Rockville, Md.: U.S. Departmentof Health and Human Services, 2003).

Agency for Healthcare Research and Quality, NationalHealthcare Disparities Report (Rockville. Md.: U.S. Depart-ment of Health and Human Services, 2003).

Agency for Healthcare Research and Quality, NationalHealth Care Quality Report (Rockville, Md.: U.S. Depart-ment of Health and Human Services, 2004).

Agency for Healthcare Research and Quality, NationalHealthcare Disparities Report (Rockville, Md.: U.S. Depart-ment of Health and Human Services, 2004).

Agency for Healthcare Research and Qualit,. NationalHealthcare Quality Report (Rockville, Md.: U.S. Departmentof Health and Human Services, 2005).

Agency for Healthcare Research and Quality, NationalHealthcare Disparities Report (Rockville, Md.: U.S. Depart-ment of Health and Human Services, 2005).

Amoroso, P.J., and J.P. Middaugh, “Research versus Pub-lic Health Practice: When Does a Study Require IRB Re-view?” Preventive Medicine 36 (2003): 250-53.

Bellin, E., and N.N. Dubler, “The Quality Improve-ment-Research Divide and the Need for External Over-sight,” American Journal of Public Health 91 (2001): 1512-17.

Bottrell, M.M., “The Ethics of Quality Improvement:Practitioners’ Perspectives. A Paper Commissioned by theEpistemology and Ethics of Quality Improvement Project atthe RAND Center to Improve Care of the Dying,” 2003,http://www.innovation.cc/peer-reviewed/bottrell-ethics.pdf.

Byers, J.F., and E.D. Aragon, “What Quality Improve-ment Professionals Need to Know About IRBs,” Journal ofHealthcare Quality 25 (2003): 4-9.

Casarett, D., J.H. Karlawish, and J. Sugarman, “Deter-mining When Quality Improvement Initiatives Should beConsidered Research: Proposed Criteria and Potential Im-plications,” Journal of the American Medical Association 283(2000): 2275-80.

Choo, V., “Thin Line Between Research and Audit,”Lancet 352 (1998): 337-38.

Cretin, S., et al., “Should Patients in Quality-Improve-ment Activities Have the Same Protections as Participants inResearch Studies?” Journal of the American Medical Associa-tion 284 (2000): 1786-88.

Davidoff, F., and P. Batalden, “Toward Stronger Evidenceon Quality Improvement: Draft Publication Guidelines.The Beginning of a Consensus Project,” Quality & Safety inHealth Care 14 (2005): 319-25.

Diamond, L.H., et al., “Quality Improvement Projects:How Do We Protect Patients’ Rights?” American Journal ofMedical Quality 19 (2004): 25-27.

Diaz, M., and D. Neuhauser, “Pasteur and Parachutes:When Statistical Process Control Is Better Than a Random-ized Controlled Trial,” Quality & Safety in Health Care 14(2005): 140-43.

Doezema, D., and M. Hauswald, “Quality Improvementor Research: A Distinction without a Difference?” IRB:Ethics & Human Research 24 (2002): 9-12.

Doyal, L., “Preserving Moral Quality in Research, Audit,and Quality Improvement,” Quality & Safety in Health Care13 (2004): 11-12.

Emanuel, E.J., D. Wendler, and C. Grady, “What MakesClinical Research Ethical?” Journal of the American MedicalAssociation 283 (2000): 2701-11.

Fox, E., and J.A. Tulsky, “Recommendations for the Eth-ical Conduct of Quality Improvement,” Journal of ClinicalEthics 16 (2005): 61-71.

Gauthier, A., and M. Serber, A Need to Transform theU.S. Health Care System: Improving Access, Quality, and Effi-ciency (New York: The Commonwealth Fund, 2005).

Institute of Medicine, To Err Is Human (Washington,D.C.: National Academy Press, 2000).

Institute of Medicine, Crossing the Quality Chasm: A NewHealth System for the 21st Century (Washington, D.C.: Na-tional Academy Press, 2001).

Institute of Medicine, Envisioning the National HealthCare Quality Report (Washington, D.C.: National AcademyPress, 2001).


Bibliography

Institute of Medicine, Responsible Research: A Systems Ap-proach to Protecting Research Participants (Washington, D.C.:Institute of Medicine, 2002).

Institute of Medicine, Health Professionals Education: ABridge to Quality (Washington D.C.: National AcademyPress, 2003.

Jarvis, H., “Differentiating among Research, Evaluationand Measures to Assure Quality,” Healthcare ManagementForum 13 (2000): 34-36.

Kofke, W.A., and M.A. Rie, “Research Ethics and Law ofHealthcare System Quality Improvement: The Conflict ofCost Containment and Quality,” Critical Care Medicine 31(2003): S143-52.

Larson, E., et al., “A Survey of IRB Process in 68 U.S.Hospitals,” Journal of Nursing Scholarship 36 (2004): 260-64.

Leape, L.L., D.M. Berwick, and D.W. Bates, “WhatPractices Will Most Improve Safety? Evidence-Based Medi-cine Meets Patient Safety,” Journal of the American MedicalAssociation 288 (2002): 501-7.

Levine, R.J., Ethics and Regulation of Clinical Research,second edition (New Haven, Conn.: Yale University Press,1988).

Levine, R.J., “Institutional Review Boards: A Crisis inConfidence,” Annals of Internal Medicine 134 (2001): 161-63.

Lindenauer, P.K., et al., “The Role of the InstitutionalReview Board in Quality Improvement: A Survey of QualityOfficers, Institutional Review Board Chairs, and JournalEditors,” The American Journal of Medicine 113 (2002):575-79.

Lo, B., and M. Groman, “Oversight of Quality Improve-ment: Focusing on Benefits and Risks,” Archives of InternalMedicine 163 (2003):1481-86.

Lohr, K.N., Medicare: A Strategy for Quality Assurance(Washington D.C.: National Academy Press, 1994).

Lynn, J., “When Does Quality Improvement Count asResearch? Human Subject Protection and Theories ofKnowledge,” Quality & Safety in Health Care 13 (2004): 67-70.

McGlynn, E.A., et al., “The Quality of Health Care De-livered to Adults in the United States,” New England Journalof Medicine 348 (2003): 6-26.

McGlynn, E.A., et al., “The Quality of Health Care De-livered to Adults in the United States,” New England Journalof Medicine 348 (2003): 2635-45.

McGlynn, E.A., “Six Challenges in Measuring the Quali-ty of Health Care,” Health Affairs 16 (1997): 7-21.

National Bioethics Advisory Commission, Ethical andPolicy Issues in Research Involving Human Participants(Bethesda, Md.: National Bioethics Advisory Commission,2000).

National Commission for the Protection of Human Sub-jects of Biomedical and Behavioral Research, Ethical Guide-lines for the Delivery of Health Services by DHEW: Report andRecommendations (Washington, D.C.: U.S. GovernmentPrinting Office, 1978).

National Ethics Committee of the Veterans Health Ad-ministration, Recommendations for the Ethical Conduct ofQuality Improvement (Washington, D.C.: National Centerfor Ethics in Health Care, Veterans Health Administration,2002).

Nerenz, D.R., P.K. Stoltz, and J. Jordan, “Quality Im-provement and the Need for IRB Review,” Quality Manage-ment in Health Care 12 (2003): 159-70.

Palevsky, P.M., et al., “Improving Compliance with theDialysis Prescription as a Strategy to Increase the DeliveredDose of Hemodialysis: An ESRD Network 4 Quality Im-provement Project,” Advances in Renal Replacement Therapy7 (2000): 21-30.

President’s Commission for the Study of Ethical Prob-lems in Medicine, Securing Access to Health Care, VolumeOne: Report (Washington, D.C.: U.S. Government PrintingOffice, 1983).

Pritchard, I.A., “Searching for ‘Research InvolvingHuman Subjects’—What Is Examined? What Is Exempt?What Is Exasperating?” IRB: Ethics & Human Research 23(2001): 5-13.

Protection of Human Subjects, Code of Federal Regula-tions, 1991, 45 Code of Federal Regulations Part 46: Re-vised.

Reinhardt, A.C., and L.N. Ray, “Differentiating QualityImprovement from Research,” Applied Nursing Research 16(2003): 2-8.

Sandrick, K., “Ethics: A Unifying Force. The TavistockPrinciples Provide a Decisionmaking Framework for Every-one,” Trustee 54 (2001): 24-28.

Schoen, C., et al., “Taking the Pulse of Health Care Sys-tems: Experiences of Patients with Health Problems in SixCountries,” Health Affairs Web Exclusives 5 (2005): W509-525.

Schuster, M.A., E.A. McGlynn, and R.H. Brook, “HowGood Is the Quality of Health Care in the United States?”Milbank Quarterly 76 (1998): 509, 517-63.


Sharpe, V.A., “Promoting Patient Safety. An Ethical Basisfor Policy Deliberation,” Hastings Center Report 33 (2003):S3-18.

Smedley, B.D., et al., Unequal Treatment: ConfrontingRacial and Ethnic Disparities in Health Care (Washington,D.C.: National Academy Press, 2003).

Smith, G.C., and J.P. Pell, “Parachute Use to PreventDeath and Major Trauma Related to Gravitational Chal-

lenge: Systematic Review of Randomised Controlled Trials,”British Medical Journal 327 (2003):1459-61.

Stevens, D.P., “Why New Guidelines for Reporting Im-provement Research? And Why Now?” Quality & Safety inHealth Care 14 (2005): 314.

Stimler, C., “Quality Improvement Projects: InactionPresents the Greatest Risk,” Archives of Internal Medicine163 (2003): 2648-49.


This project had its origins in an exploratory confer-ence initiated by Joanne Lynn and Melissa Bottrell. WithBruce Jennings, they developed a proposal to commis-sion four background papers, convene an interdiscipli-nary group of experts, and use the papers as the basisfor group discussion of the ethical and policy issuesraised by the use of quality improvement methods inhealth care. The Agency for Healthcare Research andQuality (AHRQ) funded the proposal, and the conferencetook place in December 2001 in conjunction with the In-stitute for Healthcare Improvement (IHI) Annual QualityForum. Intense discussions over two days, in breakoutsessions and plenary sessions, clarified the issues butmade it clear that further work was needed. This gaverise to The Ethics of Improving Health Care Quality andSafety Project, which was also funded by AHRQ and byin-kind contributions by The Hastings Center.

We deeply appreciate the hard work of the projectparticipants (listed by name at the front of this report)and the authors of the background papers (referred to infootnotes to this report). The papers, the meeting dis-cussions, the comments on drafts of the report, and themany one-on-one conversations we had with partici-pants and authors were critical to the development ofthis document. We would also like to acknowledge thecontributions of several people not listed who sharedtheir expertise at individual project meetings: AndreaKabcenell, Jenny Kitsen, Charles Bosk, and Jason Kar-lawish. In addition, we had many stimulating discussionsof the issues with others interested in the project, in-cluding Louis Diamond, Paul Palevsky, Michael Rie, andAndrew Kofke.

Two �field trips� taken by the project�s principal in-vestigator, Mary Ann Baily, helped to deepen our under-standing of the issues. One was a trip to Dartmouth-Hitchcock Medical Center, made at the invitation of PaulBatalden, who thought it would be a good way to learnmore about the range of QI projects. We would like tothank Polly Campion for organizing a day packed full offascinating give-and-take with people engaged in QI;there were too many to name individually, but we aregrateful to all of them. The second was a trip to M.D. An-derson Cancer Center, where Mary Ann Baily made pre-sentations about the project to several different groups.The trip was a wonderful opportunity to engage in in-tense discussion with people about the issues they faceddoing QI activities in the context of specialized cancertreatment. We are grateful to all the participants andwould especially like to thank Mano Selvan and Mar-garet Holm for organizing the visit.

We are grateful to AHRQ for providing funds for theproject, and we appreciate the assistance of our projectofficers: Elinor Walker, Marge Keyes, and finally, RobertBorotkanics. We thank the RAND Corporation for provid-ing us with meeting space for our final project meeting,and the RAND staff for administrative support for themeeting. Finally, we appreciate the efforts of the variousmembers of The Hastings Center staff who have provid-ed support throughout the project. We thank Mary AnnHasbrouck, Ann Mellor, Jay Camp, and Vicki Peyton fortheir work on meeting arrangements and project admin-istration, Michael Khair and Stacy Sanders for researchassistance, and Greg Kaebnick, Joyce Griffin, and NoraPorter for editorial and design assistance for this SpecialReport.

Acknowledgments

The Hastings Center addresses fundamental ethical is-sues in the areas of health, medicine, and the

environment as they affect individuals, communities, andsocieties. With a small staff of senior researchers at theCenter and drawing upon an internationally renownedgroup of over one hundred elected Fellows for their ex-pertise, The Hastings Center pursues interdisciplinary re-search and education that includes both theory and prac-tice. Founded in 1969 by philosopher Daniel Callahanand psychoanalyst Willard Gaylin, The Hastings Centeris the oldest independent, nonpartisan, interdisciplinaryresearch institute of its kind in the world. From its earli-est days The Hastings Center has understood that themoral problems arising from rapid advances in medicineand biology are set within a broad intellectual and socialcontext. The Center’s collaborations with policy-makers,in the private as well as the public sphere, assist them inanalyzing the ethical dimensions of their work.

ABOUT

THE HASTINGS CENTER

On the cover: Arrow Up & Down, by Alexander Calder, 1972, gouache on paper, 74.9 cm x 110 cm.© Artists Rights Society, New York. Photo courtesy Art Gallery of Ontario.

ORDER INFORMATION

For copies of this or other Hastings Center ReportSpecial Reports, write or call:

Subscription DepartmentThe Hastings Center21 Malcolm Gordon RoadGarrison, NY 10524(845) 424-4040(845) 424-4545 fax; [email protected] www.thehastingscenter.org

Mary Ann Baily, PhD, is associate for ethics and healthpolicy at The Hastings Center. She is an economist bytraining and has written on ethical issues in health carerationing, access to care, managed care, newborn genet-ic screening, and the financing of care for HIV infec-tion, organ transplantation, and Alzheimer disease.

Melissa Bottrell, MPH, PhD, works on practical prob-lems at the intersection of bioethics, health system qual-ity, and policy. Currently, she is a health services re-searcher at the Veteran’s Health Administration Nation-al Center for Ethics in Health Care. She has conductedprojects on informed consent, facility-level ethics pro-grams, end of life care, and quality of geriatric nursingcare in health care facilities such as hospitals and nurs-ing homes. She has authored numerous publications,including three edited books.

Bruce Jennings, MA, is director of the Center for Hu-mans and Nature and senior consultant at The HastingsCenter. He also teaches at the Yale University School ofPublic Health. He has authored numerous books andarticles on ethical and social issues in health care andpublic policy.

Joanne Lynn, MD, MA, MS, is a geriatrician workingas Senior Natural Scientist at the RAND Corporation.She has contributed to bioethics literature, including asstaff to the President’s Commission on Ethical Issues inMedicine; she has led conventional and health servicesresearch; and she has now coordinated the work of morethan four hundred teams doing quality improvement.

ABOUT

THE AUTHORS

Documents

QI M I H CARE - Ethics Research Institute · tice of quality improvement in health care. The project, ti-tled “The Ethics of Improving Health Care Quality and Safety,” was funded