Clinical Ethics in Anesthesiology. A Case-Based Textbook

4. Research and publication

33. Quality improvement initiatives: when is quality improvement actually a form of human subjects research?

Michael A. Rie and W. Andrew Kofke

The Case

Suarez et al. published an article in the Journal of Neurosurgery examining outcomes after human albumin (HA) administration in patients with subarachnoid hemorrhage (SAH).1 Because some believed that HA increased mortality in critically ill patients, the authors’ hospital developed an administrative policy restricting its use. The authors examined patient outcomes and costs before and after the introduction of the restriction. They concluded that administration of human albumin after SAH probably improved clinical outcomes and reduced hospital costs.

Critics argued that the decision to change the “standard of care” with regard to human albumin administration failed to take into account animal data suggesting that albumin might be neuroprotective in patients with SAH. They charged that patients had been subjected to a change in the “standard of care” because of a corporate cost containment activity that could be “more accurately described as a system level ‘experiment’ with the dual hypothesis of no harm and less cost.” Furthermore, it was an experiment “in which no disclosure occurred, nor was informed consent obtained in an institution receiving federal funds for both patient care and clinical research. The critics questioned who would voluntarily submit to an experiment with “the main dependent variables being money saved and increased morbidity and death rate.”2,3

Quality assurance and quality improvement

In 1980, Donabedian defined the basic elements of healthcare quality to be structure, process and outcome.*4 Early medical quality assurance (QA) was handled in sentinel event case discussions as part of anesthesia departmental conferences and possible prospective prevention practice suggestions. The process of quality improvement (QI) was informal, collegial and educational. Accountability to assure that professional process review would translate into prospective preventive improvements for future patients within the department’s practice of anesthesia was systematically lacking.

QI/QA has evolved from an informal retrospective review of individual cases involving critical incidents, to large retrospective analyses and evaluation of physicians, departments, hospitals, and healthcare systems. Increasingly, the impetus for projects and data collection has come from extramural demands to document accountability for care, safety, quality, efficiency, and cost in all of medical care.

The root cause of nonaccountability was evident. The victim of a prior sentinel event had been possibly unintentionally injured or suffered a “near miss,” and a negligence action in malpractice law was a possible consequence. Systemic analysis of quality is now recognized as a management necessity to address and distinguish system error from individual professional errors and/or patient idiosyncratic events. Thus, the incorporation of such study into “research subject protection ethics” became unavoidable.

The regulatory structure in the CFR and the powers invested in the Office for Human Subjects Research Protections (OHRP) inevitably came into operational inquiry for QI projects. The realization that system errors could produce as many as 100 000 errors of preventable nature per year came to public attention.*5 Shortly thereafter the Joint Commission for Accreditation of Health Care Organizations (JCAHO) published its first Sentinel Event Chapter for the accreditation standards for American hospitals.

The stage was then set to revisit what QA and QI patient subject protections might be. The world had moved before clinicians had from Phase I to Phases II and III of quality improvement activities.6These three phases of QI are:

(1) Reactive or descriptive QI: react to sentinel events.

(2) Outcome improvement QI: improve an already satisfactory outcome.

(3) Cost containment QI: improve efficiency leading to a satisfactory/better outcome.

A decade of conflicting OHRP interpretations of quality improvement research leaves QI project directors and institutional review boards (IRBs) predictably confused. When is patient consent prospectively necessary for a QI process? Can patients be potentially injured by delayed reporting of important data if medical publishers reject reports for lack of IRB approvals? The combined federal definitions of generalizable knowledge and publication cover virtually all conceivable prospective activities that a QI project might encompass while leaving uncertain the nature of exempt “minimal risk” projects.

The practice of population-based cost containment QI initiatives is widespread in healthcare. It is apparent that lessons from the business model (e.g., widget factory) are also commonly applied to patient care. But, unlike patients, widgets are not invested with contractual healthcare rights.

Is patient safety and quality being compromised by cost containment?

In the situation described by Suarez, the hospital imposed clinical research on a population of patients presenting with SAH, with the primary goal of reducing costs and the belief, but no proof, that no adverse change in clinical outcome would occur. Unfortunately, they have demonstrated the very nature of research itself: a hypothesis cannot a priori state the outcome of the research in a specific group of patients until the data are analyzed after the experiment is over.

The publication and responses to the Suarez scenario command a unique position in academic medical publication morality. If negative outcomes occur by corporate actions, then how are critical care practitioners able to speak and professionally publish such data? How do they assure that lessons learned from negative outcomes in one institution are promulgated to prevent the same mistakes in other hospitals? One could argue that any information derived from a negative outcome QI study should be mandated for generalization as a matter of organizational ethics. A similar public safety nondisclosure issue in publication occurred in aviation safety, resulting in an airliner crash into a mountain.7

Critical care professionals work in complex organizational environments in which documentation of negative outcome data for safety and quality purposes pose risk for public discovery of the cases, negligence malpractice litigation, and conflict with hospital corporate management. In many academic hospitals, there is a formal organizational structure in which the medical faculty is employed by corporate entities directly linked with hospital management and budget directors, creating intractable conflicts of interest.

Thus, there exists a culture of silence in advancing QI initiatives driven by cost containment. These professionals are in an untenable master–servant relationship when they want to report negative (as opposed to positive) data necessary for the advancement of medicine, science, and the public welfare.

Is patient safety and professional integrity threatened by state interests?

In the Suarez matter more than 25 years of animal experiments supported the use of human serum albumin to diminish intracranial ischemia and stroke propagation in patients with ruptured cerebral aneurysms. A fundamental principle of the Nuremberg Code is that human experimentation should be based on previous animal experiments, and that experiments that pose risk of serious injury to human subjects cannot be ethically conducted. The publication by Suarez led to the awarding of an NIH research grant regarding the use of human albumin in patients with ischemic stroke, and a national multi-center trial was initiated.

The Suarez matter further reveals “the moral abuse of evidence-based medicine” (EBM) in the pursuit of cost containment. The vast majority of treatments utilized in critical care medicine lack level I and II evidence to support them. To suggest (via a pharmacy and therapeutics committee) that such a lack of evidence may mandate banning of a pharmaceutical in contradiction of what is thought to be an existent standard of care without evidence-based support is to use EBM in a fashion antithetical to the welfare of patients and their abilities to participate or decline cost-driven corporate experimentation. Conversely, because there was insufficient evidence to warrant the use of serum albumin, policy makers might argue for albumin’s elimination and say that the patients were not given an appropriate consent option to decline serum albumin because it was an unproven therapy. Generalization of this approach throughout the healthcare system might drive a cost-containment-driven wedge into the fiduciary trustrelationships that are the legal and moral bases for the delivery of individual healthcare.

Should we simply eliminate therapies that do not have sufficient evidence-based support?

It is quite a different matter to delay FDA approval for new therapeutics that are of unproven benefit than to proscribe historically based treatments until better ones are found. Were this a reality, digoxin might never have entered the pharmacopeia and both the FDA and physicians everywhere have been negligent in continuing to use it.

The Suarez matter was referred to the OHRP for review. Their response was:

“… it appears that the restriction on the administration of the human albumin by the University Hospitals in Cleveland was not research as defined by federal regulations at 45CFR 46.102 … from the information submitted to OHRP it appears that the hospital’s restrictions on the administration on human albumin was not a systematic investigation and was designed to reduce hospital costs and perhaps reduce morbidity.”

We interpreted this OHRP response in the Suarez matter to indicate that the federal government presently has neither legal interest nor congressional mandate to ensure that a return to pre–Nuremberg Code practices does not develop in the practice of QI in the United States.

As rationing becomes reality, the IRB will increasingly be perceived to be under corporate economic pressure, although the converse may be as likely. This would place upon the QI-IRB of the future an apparent moral and legal conflict of interest concerning human subject protection. It seems that conflict for professionals, IRBs, and corporate officials will be unavoidable if negative outcomes derived from cost containment quality improvement activities of the institution were never subject to human subject protections by independent review.

The OHRP has reduced the definition of research for QI in such a fashion as to deputize itself as final arbiter of what systematic clinical observations may be lawfully permitted publication. A former editor of the Annals of Internal Medicine, Dr. Frank Davidoff, has stated publically, “you can be punished if you try to publish the results of your QI work but you will avoid punishment if that same work goes unpublished.”

The coming of healthcare reform and heightened concerns for patient privacy with the explosion of clinical information technology will now awaken concerns for governmental control of professional free speech in publication by healthcare investigators and QI data analysts. As Bosk et al.*8 point out, the art of medicine will be more complex and costly in achieving true improvement than Gawande’s sensational exposé on checklists.9

What should be the responsibility of IRBs with regard to QI initiatives?

Many cost-containment activities occasioned either by professionals or hospital corporations constitute human experimentation by accidental but predictable intent.*10 Such activities demand vigorous scientific peer review and some form of independent extramural funded ethical oversight beyond hospital control under the existing rule of law. Unfortunately, the most comprehensive review on this subject proposes only an extension of the existent IRB structure called a QI-IRB to perform the determination of exempt or non-exempt study for consent procedures.11 Keeping such activities within hospital management that has a cost containment integrity conflict related to the care of individuals and its public mission is not likely to prove legally or morally acceptable in the long run. In Sabin’s “Accountability for Reasonableness” illustration, there is diminishing fidelity to individuals as decision-making occurs increasingly in boardrooms rather than exam rooms, and diminishing stewardship of resources as the focus on individuals increases (Fig. 33.1).

The Electronic Medical Record (EMR), QI initiatives and privacy issues

As the US prepares for healthcare reform, the 2009 Emergency Federal Relief Budget and The Health Information Technology for Economic and Clinical Health Act allocated 19 billion dollars for electronic medical records (EMRs) for patients in every physician’s office and acute care hospital by 2014.12 Already, 1.4 billion dollars have been spent to develop regional federal extension centers to assist providers in creating the EMR and its local integrated development.

Defining a “recompensible EMR” and requirements for integrated connectivity upon manufacturers of hardware and software for local and national databases is a first step to future QI evaluation projects using the EMR. QI analysts in the “reform era” and health system leaders await these definitions to produce the “data-driven cultural change” of healthcare delivery.

Figure 33.1.

Figure 33.1. Figure 33.1. Reproduced by permission from J. Sabin. Cost-containment ethics requires fiduciary trust of both physicians and institutions to disclose their summative integrity for patient care under rationing policies.

QI projects in this new world require new ethical/legal regulatory architecture. If there is to be privacy and confidentiality of shared information between patients and providers, then the Office of the National Coordinator for Health Care Information Technology (ONC) should publish the specifics of lawfully maintaining that privacy. QI activities requiring both collection and transparent data dissemination will be in conflict with existent regulations of the OHRP placed on present-day IRBs.

Plans to define efficiency and quality value performance will be morally challenged if quality measures defining payment of institutions and physicians are recognized as social policy experimentation that violates the existent Code of Federal Regulations (CFR). Such policy questions require review and revision of the Belmont Report structure for QI organizational ethics and law interpretations, since such issues were not existent when the Belmont Report was issued in 1979. Defining “meaningful usage” of the EMR in national health policy awaits the necessary regulatory revision. These conceptual issues are the major unknowns for people conducting QI activities in anesthesiology and acute care hospital medicine. Resolving these unknowns are prerequisite to integrating primary care and acute care databases under healthcare reform.

At the operational level, QI project design needs concrete definitions of “meaningful usage.”12 Meaningful QI projects need meaningful end-point definitions of quality in order to define enhanced payments to providers to accomplish those endpoints and to determine what performance data will be deemed of lower payment priority. Detailed definitional output from ONC is an urgent emerging national QI priority.

At the most basic clinical level, QI projects must ask “Will EMRs truly cut costs and find efficiencies without decrements in the quality and accessibility of care?” Contemporary EMRs provide major cost savings and revenue enhancements for billing and coding systems integrated into provider payments.13 Physicians clearly see computers as helpful when they facilitate rapid access to laboratory data and imaging diagnostic results. What has proved burdensome to patients and professionals is the time needed to properly enter computerized physician order entry (CPOE) and the nonmonetarized costs of added professional time.

Users of hospital computer information systems need ongoing real-time transparent flowcharts for ICU, emergency room, operating suites, and procedure area patients on hospital-based computers. The records must fulfill the seamless availability of ongoing minute-to-minute flow sheets to facilitate information transfer across shifts of individuals and nursing personnel to physicians. If these flow records cannot meet these standards of clinical necessity, they should be accounted for as quality and safety decrements in performance evaluation.

Future QI ethics projects will be more easily understood and rewarded when efficiency is properly measured and held financially accountable to maximize professional quality/time inputs and outputs.

A primary care physician in a remote office will need the same enhanced EMR time maximizing efficiency. Leveraging system productivity requires funded connectivity telemedicine applications in offices and patient residential sites that will permit physician extenders to leverage and maximize higher complexity decision-making activity by primary care physicians liberated from present-day documentation inefficiencies. Primary care physicians will need swift telemedicine consultation with specialists when necessary to efficiently solve chronic disease management issues in the home setting to prevent additional hospitalizations. Meaningful usage criteria here should carry the ability of the EMR to produce demonstrable cost savings and efficiency, on the one hand, while providing negative efficiency documentation of redundant functions not captured in deficient EMR hardware and software on the other.

Other nations have been implementing EMR integration with direct caregivers for several years. Successful smaller nations have required local integration. Vast data composites to assess the national productivity or net aggregate costs are a political imperative but will not offer a delivery cultural transformation. Smaller QI projects undertaken by QI analysts going forward will offer the necessary guidance to effect real change in individual services, hospitals and communities as a whole. Thus, the data-driven EMR, defined as “meaningful usage” will be fundamental in illuminating the legislative revision of CFRs and OHRP’s legal accountability in advancing efficiency, safety, quality, and privacy in clinical medicine.

Key points

• QI/QA has evolved from an informal retrospective review of critical incidents to large retrospective analyses and evaluation of physicians, departments, hospitals and healthcare systems.

• These three phases of QI are:

• Reactive or descriptive QI: react to sentinel events.

• Outcome improvement QI: improve an already satisfactory outcome.

• Cost containment QI: Improve efficiency leading to a satisfactory/better outcome.

• Some forms of QI activity actually represent human subjects research, with potential for patient harm. As such, similar ethical imperatives should apply in protecting patients from unconsented research when QI initiatives involve significant changes in patient care.

• As rationing and cost-containment efforts increase, IRBs may come under increasing pressure, and therefore experience conflicts of interest, with regard to evaluation of cost-containment QI activities.

• The Electronic Medical Record (EMR) carries powerful potential as a tool in QI and cost-containment efforts. Use of patient records carries risk for future privacy conflicts necessitating revision of the CFR and OHRP’s regulatory authority in federal policy creation.


1 Suarez, J.I.Shannon, L.Zaidat, al. (2004). Effect of human albumin administration on clinical outcome and hospital cost in patients with subarachnoid hemorrhage. J Neurosurg100, 585–90.

2 Heros, R. (2004). Fluid management. J Neurosurg100, 581–2.

Rie, M.A.Fahy, B.G., and Kofke, W.A. (2005). Human serum albumin. J Neurosurg101, 564–6; author reply 566.

4* Donabedian, A. (1980). Definition of Quality and Approaches to Its Assessment. Ann Arbor, Michigan: Health Administration Press.

5* To Err Is Human. (1999). A Report of the Institute of Medicine. Washington, DC: National Academy Press.

6 Bellin, E. and Dubler, N.N. (2001). The quality improvement–research divide and the need for external oversight. Am J Public Health91(9), 1512–17.

7 Aviation Safety Network.–0.

8* Bosk, C.L.Goeschel, C.A., and Pronovost, P.J. (2009). The art of medicine: reality for checklists. Lancet374, 444–5.

9 Gawande, A. (2007). A life saving checklist. Opinion, New York Times December 30. Available at

10* Kofke, W.A. and Rie, M.A. (2003). Research ethics and law of healthcare system quality improvement: The conflict of cost containment and quality. Crit Care Med31(Suppl), S143–52.

11* Baily, M.A.Bottrell, M.Lynn, J., and Jennings, B. (2006). The ethics of using QI methods to improve quality and safety. Hastings Cent Rep36, S1–40.


13 Haig, S. Electronic records: will they really cut costs?.,8599,1883002,00.html/.

Further reading

Childress, J.F.Meslin, E.M., and Shapiro, H.T. (2005). Belmont Revisited. Washington, DC: Georgetown University Press (p.vii).

Grady, C. (2007). Quality improvement and ethical oversight. Ann Int Med146(9), 680–1.

Miller, F. and Emanuel, E. (2008). Quality improvement research and informed consent. NEJM358(8), 765–7.

Rie, M.A. and Kofke, W.A. (2007). Non-therapeutic quality improvement: The conflict of organizational ethics and societal rule of law. Crit Care Med35 (Suppl), S66–84.

Rodwin, M.A. (1993). Medicine, Money and Morals: Physicians’ Conflicts of Interest. New York: Oxford University Press.

Tapp, L.Edwards, A.Elwyn, al. (2010). Quality improvement in general practice: enabling general practitioners to judge ethical dilemmas. J Med Ethics36(3), 184–8.

US Department of Health Education and Welfare. (1979). Protection of human subjects: Belmont Report. Ethical principles and guidelines for the protection of human subjects of research. Fed Reg44,23192–7.