OBSERVATION
Oversight CLINICAL of Quality Improvement
Focusing on Benefits and Risks
Bernard Lo, MD; Michelle Groman, AB
uality improvement (QI) may substantially improve patient outcomes while posing little risk to subjects. However, the term quality improvement is used to refer to a broad range of projects, which vary widely in the potential benefits and risks to participants. Some projects raise ethical concerns. An explicit protocol for the ethical review of QI would benefit both patients and leaders of QI projects. If a project is considered research rather than QI, review by an institutional review board and informed consent from subjects may be required. In contrast, QI projects may require little oversight beyond what is already in place for clinical care. However, a monolithic approach to oversight of QI is inappropriate in light of the variation in benefits and risks of QI projects and their overlap with research. The key ethical issue is not the classification of a project as QI or research, but the balance of anticipated benefits and harms in the project. We propose a protocol for independent review of QI projects and patient consent that will protect subjects from serious harm while encouraging QI projects that will substantially benefit participants and pose only minimal risk. Arch Intern Med. 2003;163:1481-1486
Quality improvement (QI) projects are intended to improve the outcomes of patient care.1 For instance, they may correct the underutilization of interventions proven to be effective, such as cancer screening tests and drugs to prevent the recurrence of myocardial infarctions. Quality improvement projects are likely to increase because of requirements by the Joint Commission for the Accreditation of Health Care Organizations for QI activities. Generally, QI poses little risk to subjects. Close oversight of such projects may be counterproductive, serving only to delay them or make them prohibitively expensive. However, the term quality improvement is used to refer to a broad range of projects, which vary widely in the potential benefits and risks to participants and which overlap with research (Table 1). A study of fetal ultrasound for the detection of Down syndrome, characterized by the project leaders as QI, was criticized as research on a controversial intervention that should have received institutional review board (IRB) examinaFrom the Program in Medical Ethics and the Division of General Internal Medicine, University of California, San Francisco. The authors have no relevant financial interest in this article.
(REPRINTED) ARCH INTERN MED/ VOL 163, JUNE 23, 2003 1481
Q
tion.2,3 Other projects may aim to reduce health care costs by providing less than standard care. Such projects raise ethical concerns, and their participants may require protection through independent review of protocols or requirements for informed consent. Clearer policies regarding the ethical review of QI would benefit both patients and leaders of QI projects, who are often internists in clinical leadership positions. In the United States, the level of review and regulation is based on the category to which a project is assigned. If a project is classified as research, it may be subject to the Common Rule,4 which usually requires IRB review and patient consent. However, if a project is classified as QI, it requires very little oversight under the Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule.5 Yet research and QI inevitably overlap. Quality improvement projects may have similar objectives, methods, and impacts on patients as projects that are categorized as research. The key ethical issue is not the classification of a project as QI rather than as research, but the balance of benefits and harms in the project. This ar-
WWW.ARCHINTERNMED.COM
Downloaded from www.archinternmed.com at Saint Louis University, on July 26, 2005 2003 American Medical Association. All rights reserved.
Table 1. Examples of QI Projects, Classified According to Benefit and Risk
Anticipated Benefit Substantial Minimal Risk Project 1. Increase use of aspirin, -blockers, and statins after myocardial infarctions. Although these drugs reduce the risk of future coronary events, they are underutilized. Project will identify myocardial infarction survivors who are not receiving these drugs and send letters to physicians recommending their use. Project 2. Retrospective review of patients undergoing colectomy to try to reduce perioperative infections. Project 3. Postdischarge satisfaction survey. More Than Minimal Risk Project 4. Use of rapid HIV testing (without confirmatory test) to decrease percentage of tested clients who fail to return for test results. Project 5. Follow-up study on the morning-after pill to see whether patients also receive STD testing and treatment and advice on contraception.
Not substantial
Project 6. Encourage physicians to decrease the length of stay for patients with pneumonia. The average length of stay at the hospital is already shorter than recommendations in peer-reviewed practice guidelines. Project 7. Evaluate the cost-effectiveness of switching patients with high cholesterol to a less costly drug. Patients will be randomly chosen to be encouraged to switch to the lower-cost drug. Project 8. Retrospective review by a hospice to decrease costs of care for injection drug abusers who are high-cost patients. Unless the hospice can control costs, it will end enrollment of such patients.
Abbreviations: HIV, human immunodeficiency virus; QI, quality improvement; STD, sexually transmitted disease.
ticle presents a protocol for the ethical review of QI projects. Our approach differs from published articles on this topic in encouraging minimal risk QI projects and by not subjecting large numbers of QI projects to the level of review required for research.6,7 CHARACTERISTICS OF QI Potential Benefits of QI In many QI projects, the benefits can be considered substantial in the sense that they apply directly to patients in the project, are of great clinical importance, and are highly likely. Three questions can help clarify the benefits of a project. Who Benefits From the QI Project? Patients who participate in the QI project may benefit personally. Project 1 (Table 1) will identify and notify patients who have not received treatments and preventive measures that have been shown to improve important clinical outcomes in rigorously conducted clinical trials. Arguably, physicians and health care organizations have an ethical responsibility to close the well-documented gap between actual care and evidence-based practice standards. Quality improvement projects like project 1 fulfill this obligation by reducing under-
utilization of beneficial interventions. In other projects, benefits may accrue to other patients with the same condition, not to those who participate in the QI project. For instance, in project 2, a review of postoperative infections in colectomy cases will benefit future surgical patients. Still other QI projects, however, are designed to benefit primarily the health care provider rather than the patients. Project 6 aims to decrease the length of stay for patients with pneumonia. With health care costs sharply increasing, it is important to determine whether care can be provided more efficiently without compromising outcomes. It is unlikely that an even shorter length of stay will improve patient outcomes in project 6. To be sure, patients benefit indirectly from reduced costs of care and from avoiding unnecessarily long hospital stays. The primary goal, however, is to reduce expenses for the health care organization. What Is the Magnitude of the Anticipated Benefit? Two types of QI projects offer the prospect of considerable clinical benefit. First, as noted, some QI projects aim to reduce underuse of such interventions known to be effective, as in project 1. Second, a QI project may aim to improve the organizations performance on an important clini-
cal outcome, such as postoperative infections in project 2. How Likely Is the Anticipated Benefit? In some QI projects, the benefits are highly likely. The leaders of QI projects generally have authority for the clinical unit where the project is carried out and thus can readily implement the findings of the project. Furthermore, the means used in the project may be known to be effective in other similar situations. This high likelihood of implementation distinguishes many QI projects from health services research projects that pose similar questions and use similar methodologies. In other QI projects, however, the likelihood of benefits may be uncertain or even contested. Project 8 reviews the impact of injection drugabuse on length of stay in hospice. However, the characterization of benefits may be contestedin what sense is it a benefit to patients with injection drug use to single them out as high-cost patients who may need to be excluded from hospice? Critics might argue that the outcomes are actually risks to subjects, not potential benefits. Risks of QI Minimal Risk. Many QI projects pose only minimal risks. Federal
(REPRINTED) ARCH INTERN MED/ VOL 163, JUNE 23, 2003 1482
WWW.ARCHINTERNMED.COM
Downloaded from www.archinternmed.com at Saint Louis University, on July 26, 2005 2003 American Medical Association. All rights reserved.
regulations for research on human subjects define minimal risk as comparable with the risks of everyday life or standard medical care.8 In QI projects, to better implement evidence-based practice guidelines, the risks are simply those inherent in good clinical care. In many other QI projects, the interventions are small-scale changes in the process of care, which are known to have only minimal risks and whose effects are closely measured.1 For example, project 2 may attempt to reduce postoperative infections through explicit critical pathways, standardized orders for prophylactic antibiotics, and signs on patient doors to encourage handwashing. Furthermore, QI projects that carefully monitor outcomes of interventions may pose less risk compared with standard clinical care, in which the outcomes of changes in practice are not systematically measured. Risks to Privacy and Confidentiality. For QI projects that review existing medical records, the main risk is a breach of confidentiality. The risks may be minimal if adequate safeguards are in place in the organization to protect confidentiality, as are required under federal privacy regulations. Privacy may be violated even if confidentiality is maintained and the patient experiences no tangible harms. Project 5 intends to ensure that women who receive the morning-after pill receive counseling about sexually transmitted diseases and contraception. Some women, however, may object to others looking at highly sensitive personal information without their permission. Other QI projects, however, may pose more serious risks, as we next discuss. Deviation From Standard of Care Some QI projects may pose considerable risks because they provide considerably less than standard care. Project 6 aims to reduce the length of study for patients with pneumonia below what peer-reviewed, evidence-based guidelines recommend. Patients who are discharged too early may experience physical
harm if they fail to respond to oral antibiotics, experience recurrence of symptoms, or develop complications. Randomization Some QI projects assign patients a certain type of care, as in project 7. Generally, patients assume that their care is selected by their physician on the basis of what is best for them personally. They may feel wronged if their care is determined by the QI protocol without their knowledge or consent. CURRENT REGULATION OF QI Oversight of QI Under Common Rule Federalregulationsgoverningresearch with human subjects, known as the CommonRule,generallyrequirewritten informed consent from subjects and review by an independent IRB. Consent and IRB review are not required for projects that are not categorized as research. The Common Rule defines research by its intention of obtaining generalizable knowledge, which will benefit society and future patients.9 In contrast, the intention in clinical care is to benefit the individual patient. The assumption is that research subjects need protection because a project is not intended to benefit them directly, yet places them at risk. Thus, for QI projects that intend to benefit the patients in the project, such as project 1, the protections required in the Common Rule may not be appropriate. Defining research by intention, however, is problematic. 10 Many QI projects have multiple intentions because they are designed both to improve clinical care within a specific clinical unit and to disseminate findings through presentations and publications. Pragmatically, IRBs vary in how they interpret the definition of research in the context of QI projects. Some IRBs consider a project to be research if the results are to be published or disseminated beyond the clinical unit.10 However, if a QI project that poses only minimal risks is found to substantially improve care, it makes little
sense not to disseminate the findings merely because IRB approval was not obtained. Oversight of QI Under HIPAA Regulations Under federal privacy regulations that went into effect in April 2003, health care providers may use individually identifiable health information for QI without obtaining patient consent.11,12 Quality improvement is not explicitly defined in the regulations. Under HIPAA, no additional confidentiality protections are needed for QI beyond those already required for clinical care and health care operations. These protections include organizational policies and procedures, a privacy officer, data security, and staff training. Recommendations About QI by the NBAC The National Bioethics Advisory Commission (NBAC) recommended that QI be excluded from federal research regulations and specifically from IRB review.13 It reasoned that QI directly benefited patients and thus was more akin to clinical practice than research. Furthermore, NBAC called for less oversight of minimal risk research13; as noted, most QI projects are minimal risk. The approaches to oversight of QI discussed in this section draw a sharp line between QI and research, treat all QI projects alike, and do not define QI carefully. However, as we have seen, the term quality improvement is used to refer to a wide variety of projects, and there is unavoidable overlap between research and QI. Rather than approach all QI projects in the same manner, it is preferable to distinguish different levels of potential benefit and risk to subjects. RECOMMENDATIONS TO PROTECT PATIENTS IN QI PROJECTS
Table 2 presents our recommendations for ethical review of QI projects. The key issues are notice, independent review, and permission from patients.
(REPRINTED) ARCH INTERN MED/ VOL 163, JUNE 23, 2003 1483
WWW.ARCHINTERNMED.COM
Downloaded from www.archinternmed.com at Saint Louis University, on July 26, 2005 2003 American Medical Association. All rights reserved.
Table 2. Recommendations for Ethical Review in QI Projects
Notice care Health organizations should add a brief explanation of QI to information on health privacy that HIPAA requires they give to patients. For most projects, independent review is not necessary. Independent review is warranted for certain projects when (1) The risk is greater than minimal. (2) A project provides untested interventions or less care than the current standard of care. (3) The characterization of outcomes as benefits is contested. For certain projects, permission from individual patients should be obtained (1) If the risk is greater than minimal and (a) A project provides untested interventions or less care than the current standard. (b) Anticipated benefits to subjects are not substantial. (c) Treatment or care is assigned by the protocol, not by a physicians judgment. (2) When recommended by independent review committee. All projects for which permission is required will also receive independent review, but not all projects receiving independent review need to obtain permission from patients.
may view the QI project differently than the project team, who may overestimate the benefits and underestimate the risks.15,16 Options for Independent Review. Institutional Review Board. Institutional review boards are experienced in assessing the risks and benefits of projects and suggesting how to minimize the risks in such projects. Quality improvement directors might work with the IRBs at their institutions to benefit from their experience to clarify local criteria for which projects need IRB review. Because many journals will not publish studies that have not been reviewed by an IRB or exempted from IRB review,17 QI leaders should also consider submitting to IRBs projects whose results they might publish. However, IRBs are already overburdened and lack the resources to review numerous QI projects. In health care organizations that carry out a great deal of QI, it may be desirable to establish an IRB specifically to review selected QI projects. Other alternatives also need to be considered. Privacy Board. Under the HIPAA regulations, research using identifiable health information may be carried out without patient authorization if approved by a privacy board.18 These boards are meant to be more flexible than IRBs. The board may be a single individual in a small organization, such as the organizations privacy officer. A privacy board would be suitable to review QI projects using existing data, whose major risk is a breach of confidentiality. However, it may not be skilled at evaluating other risks to subjects or suggesting how they might be reduced. Community Consultation. Representatives of or advocates for persons who would be affected by QI projects could serve as members of an IRB or privacy board.13 Such persons could help assess benefits and risks and suggest how to reduce risks. There are numerous problems with community consultation, including how to identify and select persons to be consulted and what power the consultants should have to modify the project or reject it. Nonetheless, some input from those directly affected by
Independent review
Permission
Abbreviations: HIPAA, Health Insurance Portability and Accountability Act; QI, quality improvement.
Notice About QI Projects Many QI projects can be carried out without patients knowing about them. However, notification about QI projects shows respect for those persons who are studied. Even if patients have no meaningful choice about participating in QI projects, it is respectful to tell them about such projects. Such notice may also allow persons to take seek further information about QI projects. The major concern about notice is feasibility. It would be prohibitively expensive to notify every patient of each specific QI project. It would be feasible, however, to notify patients about QI projects in general terms when they present for care or enroll in a plan and tell them how to obtain additional information. Under HIPAA, health care providers and plans must provide notice to patients that their identifiable health information will be used or disclosed for treatment, payment, and health care operations and provide a means to obtain more information about the organizations privacy practices.14 Health care organizations can add a specific brief explanation of QI to this general notice of uses of identifiable health information, with little additional effort or cost. Quality improvement warrants specific mention because patients may be less familiar with QI
than with other types of health care business operations. Independent Review of QI Projects Review of a QI project by persons independent of the QI team might be useful for several reasons. First, such review might raise important overlooked issues. In the project on hospice care for injection drug users (project 8), independent reviewers might ask why the project does not address the problem of poor pain control or analyze other high-cost patients. Second, reviewers might suggest how to reduce the risks of the QI project and thereby make it more acceptable ethically. In project 5, which aims to reduce the length of stay below what practice guidelines recommend, reviewers might ask whether an adequate plan is in place to prevent medical harms to patients. For example, such plans might include enhanced discharge education about warning signs and symptoms and telephone follow-up to check that the patient is improving. Third, determining whether the balance of benefits to risks is acceptable requires judgments that are best made by someone other than the project leaders. It is impossible to define exactly what risks are minimal or considerable. Independent reviewers
(REPRINTED) ARCH INTERN MED/ VOL 163, JUNE 23, 2003 1484
WWW.ARCHINTERNMED.COM
Downloaded from www.archinternmed.com at Saint Louis University, on July 26, 2005 2003 American Medical Association. All rights reserved.
controversial projects is better than none. Recommendations for Independent Review. Subjecting all QI projects to independent review would be costly and deter valuable QI projects (Table 2). There is little need to review QI projects that truly pose only minimal risk. However, it would be appropriate to review projects that present greater than minimal risk. In addition, review is warranted if the project provides untested interventions (project 4) or less care than the current standard of care (project 6), or if the characterization of outcomes as benefits is contested (project 8). Because the protocol might be modified in ways that make it more acceptable, the threshold for independent review should be lower than the threshold for requiring permission from subjects. In project 7, a health maintenance organization evaluates the costeffectiveness of switching patients with high cholesterol to a less costly drug. This project presents a close call regarding permission. The medical risks are the inconvenience of additional visits or blood tests to adjust the dose of the new drug and to check for adverse effects, which are comparable with the risks of standard medical care and thus could be characterized as minimal. Patients may benefit directly from lower copayments for the new drug. The issue is how to balance the dignitary harm of not obtaining permission for randomization against the importance of the information to be gained from the project and the difficulties of carrying out the project if consent were required. Reasonable people could disagree on how to strike this balance. Some would argue that permission is not necessary because the health care organization can simply take the older drug off the formulary without any permission from enrollees. 19 Because of such disagreements, independent review makes more sense than leaving the judgment of benefits and risks entirely to those running the QI project. Permission for QI Requiring permission from patients for QI projects shows respect
by giving them some control over their care and private information. Permission is particularly important if the balance of benefits to risks of the QI project is contested and reasonable patients might not wish to participate in the project. We use the term permission rather than consent because detailed written consent forms, which are often required in research, may not be appropriate in QI. On a practical level, the permission and cooperation of the subjects is necessary if patients in QI projects are asked to complete questionnaires about, for example, satisfaction, functional outcomes, or quality of life. It would be costly, cumbersome, and potentially counterproductive, however, to require permission from patients in all QI projects. People may have no objection to participating in the study, yet may not return a permission form. They may be too busy, misplace the form, or simply give it low priority. If nonparticipation is significant, differences between responders and nonresponders may bias the findings and thereby undermine the validity of the project.10 We recommend that specific permission for QI projects is usually not necessary if the risks a...