INTRODUCTION AND AIMS
Comparative effectiveness research (CER) is designed to understand the relative benefits and harms of interventions in “real world” settings1 and thus must be integrated closely with clinical practice. The increased digitization of health care has opened new ways to integrate research with clinical practice, increasingly blurring the traditional divide between research and clinical care.2–6 Considerable attention has been devoted to the ethics of retrospective research relying on electronic clinical data (ECD) and to the relevance of regulatory requirements through the Common Rule, the Health Insurance Portability and Accountability Act, and the Health Information Technology for Economic and Clinical Health Act,7–11 but less attention has been given to ethics and ECD for prospective CER designs, including randomized control trials (RCTs).12,13
In this paper we use an ethics framework developed for a learning health care system14 to analyze 2 case examples of prospective CER studies using ECD to compare different antihypertensive drugs. The first describes a prospective observational CER study and the second describes a pragmatic RCT; for each, we consider scenarios in which outcome data are drawn exclusively from ECD and scenarios where these outcome measures are supplemented with data collected specifically for the study.
NORMATIVE FRAMEWORK FOR EVALUATING CER-ECD RESEARCH
The ethics framework for a learning health care system15 builds upon traditional principles of clinical and research ethics, including the Belmont Report, but is specific to neither clinical research nor clinical practice. Instead, it is designed to provide guidance for activities in which research and practice are intertwined to enable systematic learning from ongoing clinical care that will constantly improve clinical practice.
The framework is comprised of 7 moral obligations, to: (1) respect the rights and dignity of patients; (2) respect the clinical judgments of clinicians; (3) provide optimal care to each patient; (4) avoid imposing nonclinical risks and burdens on patients; (5) reduce health inequalities among populations; (6) conduct activities that foster learning from clinical care and clinical information; and (7) contribute to the common purpose of improving the quality and value of clinical care and health care systems.14 Obligations 1–6 fall on researchers, clinicians, health care administrators, institutions, payors, and insurers. The seventh falls on patients to participate in learning activities integrated with their clinical care.
The framework is intended to guide the actions of researchers, clinicians, patients, administrators, and payors who work in or with, or receive their medical care through, health care organizations and systems that adopt the framework’s obligations as institutional values. In such organizations, patients would be regularly informed about these commitments. Key messages to patients would include why the integration of learning with clinical services is needed to continuously improve the quality of patient care; how patients’ and clinicians’ rights and interests are protected, including that the quality of care will never knowingly be compromised for the sake of learning; and an assurance that any learning activity that might meaningfully change what care is provided or how it is delivered will always include specific informed consent requirements, proceeding only with patients’ express individual written consent.
We use this framework to examine disclosure and consent requirements in 2 hypothetical CER studies of antihypertensive medication. These cases are hypothetical only in that they are not based on any specific current or completed research studies; they are realistic, however, both in their characterization of current hypertension drug management and in mimicking some design features of studies currently being planned16–18 or previously conducted19,20 to compare the effectiveness of antihypertensive medications.
We assume that both studies will receive ethics oversight through institutional review board (IRB) review. We focus exclusively on questions of consent and disclosure.
ANTIHYPERTENSION THERAPY CASE STUDIES
Case A is a prospective observational CER study comparing 2 medications for hypertension that are both widely used and acceptable clinical treatments and for which there are no defined clinical characteristics that would favor one drug over another for individual patients. The study is prospective, with patients identified when physicians prescribe either of the 2 antihypertensive medications. Clinical outcomes are abstracted prospectively from EMRs (Table 1). Case B is a pragmatic RCT of these same 2 antihypertensive medications. In Case B, ECD are used prospectively to select patients eligible for either medication. Eligible patients are randomized to receive one of the medications, with physicians and patients always able to override the randomized choice (Table 1).21 For both cases, we also consider variations in which study outcomes (1) rely only on routinely collected ECD; or (2) supplement ECD with additional data collected specifically for the study (Table 1).
RESULTS AND DISCUSSION: APPLICATION OF FRAMEWORK TO THE CASE STUDIES
We begin with obligations 2–6, as obligations 1 and 7 are heavily contingent on how these other obligations fare. Case A-1 does not interfere with the obligation to respect clinician judgments (obligation 2). It has no impact on the process of delivering medical care or on patients’ clinical outcomes and thus does not negatively affect the obligation to provide optimal clinical care to each patient (obligation 3). The study does not impose nonclinical burdens on patients; assuming robust confidentiality protections, and given the kinds of data to be extracted from ECDs, any nonclinical risks should be minimal (obligation 4). The study as designed neither undermines nor advances the obligation to reduce health disparities (obligation 5). Assuming it is well designed, will be properly conducted, the ECD are of sufficient quality for sound research purposes, and the health care organization has the capacity to translate findings into practice, the study is consonant with professional duties to conduct and contribute to learning activities (obligation 6).
Because the study will have no adverse impact on patients’ clinical outcomes or experience, and imposes no nonclinical burdens and no more than minimal nonclinical risks, the study is presumptively respectful of patients’ rights and dignity (obligation 1) and compatible with the view that patients should contribute to learning (obligation 7). Also important to assessing obligations 1 and 7 is assuring there is no reasonable potential for the study to result in stigma or reputational harms to any specific patient groups and assuring that patients are unlikely to find study goals objectionable. Assuming case A-1 will be conducted in an organization that instantiates the values of the framework, has appropriate ethics oversight procedures and practices, and commits to respecting patients by regularly informing them that learning routinely takes place and how their interests and rights are being protected, it would seem consonant with both obligations 1 and 7 not to require express informed consent for this study.
Assuming confidentiality protections are appropriately robust, it may also be justifiable not to require express informed consent for case A-1 even in institutions that have not instantiated the values and practices called for by the framework. Whether an IRB would today approve case A-1 without informed consent is not predictable. To grant a waiver of consent, an IRB must determine not only that the risks are minimal and protections adequate but also that the research could not practically be carried out without the waiver, a judgment that could vary among IRBs. When such waivers are granted, it is generally without a requirement for broad disclosures (as we advocate) to patients, about the kinds of research activities routinely conducted in their health care setting.
Case A-2 differs from A-1 only in that patients are asked to provide additional self-reported data through a survey and to make 2 additional visits for blood pressure readings. Information collected through the survey and additional blood pressure measurements may have clinical value and will be available to treating physicians. Thus, the research may incidentally improve clinical outcomes.
Neither the survey nor extra clinical visits entail appreciable risk of harm. The survey burden is small; surveys are short, and can be completed within the time frame of patients’ regularly scheduled clinic visits. Conducting case A-2 without express consent is arguably ethically justifiable if the survey was the only research-specific activity added and if this study was being conducted in an institution whose patients were regularly informed about research of this type being integrated with practice and about protections in place to provide oversight of learning activities. However, if patients were not regularly informed about the institution’s commitment to the integration of research with practice nor that research of this type is routinely conducted with appropriate oversight, then failing to provide an express disclosure to patients about the purpose of the survey and failing to obtain consent would be harder to justify. A relevant consideration is whether it would be viewed as best clinical practice to obtain the additional information from patients, even if it is not routine in the particular setting to do so.
The 2 clinic visits impose more significant burdens and opportunity costs on patients including transportation costs and the potential loss of income or sick days from work. Here, it would be disrespectful (obligation 1) not to disclose to patients that extra visits are for the separate research purpose of comparing outcomes among patients prescribed different antihypertension medications. Because the visits entail only extra burden and no appreciable additional risk, and patients are being asked to do something that is straightforward and easy to explain, it may be acceptable to rely on oral disclosure rather than written consent. However, it should be made clear to patients that they are being asked to take on the burdens of 2 additional clinic visits solely to help answer a research question, and not to improve their personal clinical care, and they are under no obligation to do so.
Whether an IRB would permit oral disclosure in case A-2, again, is unpredictable. Because blood pressure readings and clinic visits are minimal risk interventions that do not require written consent in clinical care, some IRBs may allow a waiver of usual written consent requirements. Nonetheless, it remains the practice of many IRBs to include written consent when permission is requested from research participants, whether or not a waiver is technically allowable.
Case B-1 differs from case A-1 in that which drug patients are initially prescribed will be determined by random assignment rather than physician preference. Even with this change, several obligations of the framework remain unaffected. Most notably, because patient eligibility is limited to those for whom either of the 2 antihypertensive medications would be clinically appropriate, the act of randomizing patients does not negatively impact the obligation to provide patients with optimal clinical care (obligation 3). That is, given strict eligibility requirements, using either drug is consistent with best medical practice. Assuming the trial is well designed, the ECD are of high quality, the trial is properly conducted, and the health system has the capacity to translate findings into practice, the study is consistent with the professional obligation to contribute to activities that can improve quality of care (obligation 6).
A key feature of this design relevant to obligations 2 and 3 is that physicians have the authority to override enrollment, randomization, and continued participation at anytime. Randomization alters the expected dynamic between patients and clinicians in ways that have implications for obligations 1 and 2. The significance of these implications depends on the specifics of any particular case. In our example, both therapies are FDA approved and are considered acceptable clinical alternatives; only patients for whom there is no evidence favoring one treatment over the other would be enrolled. There are few differences between the drugs in how they are administered, frequency of administration, or side-effect profiles. Both drugs are well tolerated by patients and adverse events are rare. Thus, patients’ interests in exercising personal preferences, and the role clinicians have in advocating for those preferences, are limited.22 The impact of randomization on obligations 1 and 2 would be greater if the interests at stake were more significant or more sensitive to variations in patients’ preferences and values.
As already noted, case B also incorporates several specific design features that increase its ethical acceptability. Case B has strictly defined eligibility criteria to ensure that only patients clinically appropriate for either medicine are included (obligation 3); also, physicians are allowed to change (with documentation, also relevant for learning) the treatment assignment for any patient they believe would be better served, for any reason, by a different medication or make dose adjustments or add additional therapies (obligations 2 and 3).
We believe it is ethically acceptable to implement case B-1 without requiring express informed consent from participants,22,23 assuming the study will be implemented in a health care context where patients are regularly informed that learning activities involving randomization are occasionally permitted without express consent, after careful review to ensure patients’ interests and rights are protected. That review should involve patients, whose perspectives are necessary to assess whether alternative interventions engage values or preferences meaningful to them.
Although there is increasing discussion about the ethical acceptability of conducting minimal risk, randomized CER research without first obtaining informed consent,12,21,22,24 doing so is a significant departure from current practice. Federal regulations today would require an express waiver of informed consent requirements.
We are not arguing that policy or practice should change immediately to permit the conduct of trials like case B-1 without express informed consent based solely on this brief moral analysis. The structure of argument outlined here must be vetted through sustained engagement with patients and clinicians about the proper role for randomization and consent in a learning health care system. Changes in practice and policy are only possible if a widely shared consensus among relevant stakeholders emerges that some kinds of CER-ECD designs involving randomization to widely used therapies should be allowed to proceed without express consent, a consensus that may require evidence that such studies result in improved quality of care and no important affronts to patient dignity from the perspectives of patients.
The only difference between case B-2 and case B-1 is that patients will complete a survey not routinely administered in clinical care and make 2 additional clinic trips for blood pressure checks. The ethical implications of these additional burdens were discussed in case A-2, where we concluded it was important to explain to patients that the additional visits are for the exclusive purpose of research and wholly discretionary. Case B-2 would require the same disclosure; but then explaining to patients that they are being asked to make additional visits for research purposes without also explaining that the research involves randomization is disrespectful and maybe perceived as deceptive. Once patients are asked permission for additional study-specific clinic visits, it is more straightforward and more respectful to also solicit their consent for randomization. In contrast, although interaction with patients about the extra research visits also occurs in case A-2, it can be explained orally and simply to patients that the visits will provide information to help compare how patients prescribed different antihypertension medications by their physicians do.
CER-ECD studies of the sort illustrated by our cases challenge the traditional research ethics paradigm. Rather than testing experimental, unproven interventions, the research question asks how interventions that are widely used compare, and deploys these interventions in ways consistent with good clinical practice. Some of the usual concerns against which IRB review and informed consent are intended to guard, such as exposure without consent to experimental, substandard, or potentially riskier treatments, are not relevant. Nevertheless, ethical duties to respect patients’ rights and dignity and to provide them with optimal clinical care remain, as do obligations to respect clinician judgment. To these we add the general view that both health care professionals and patients should contribute, under appropriate circumstances, to learning in health care systems in which a commitment to learning is a shared and public value of the system.
Drawing on these obligations, our analysis provides support for conducting projects like case A-1 without informed consent, and supports adding only oral disclosure for some study-specific additional burdens.
Our analysis also suggests that under carefully circumscribed conditions, and only with appropriate prospective ethical review and oversight, some randomized CER-ECD may appropriately proceed ethically without express informed consent. These conditions include addition of no or only minimal risks or burdens to patients, the exclusion of patients for whom study interventions are clinically inappropriate, liberal change and stoppage rules for physicians who believe their patients would be better off or would prefer a clinical alternative, and no plausible reason to conclude that patients are likely to have meaningful preferences for one intervention over another and/or that patients would object to the overall purpose of the study.
A CER study comparing 2 hypertension drugs is not the only comparative study likely to fit this approach. Consider, for example, a multicenter pragmatic CER trial currently under development that proposes to randomize patients to taking a medication either at bedtime or in the morning,25 or one that propose to randomly assign patients to 2, commonly used statins. If the movement to learning health care systems progresses, and if more health care organizations adopt ethical commitments similar to the framework we rely on here, including responsiveness to patient values and concerns, then express written consent may not be ethically required for some CER-ECD studies, including a circumscribed subset with randomized designs. The researcher will remain critical to maintaining the ethics of such research, both to ensure methodological rigor and to ensure research is conducted for the socially valued goals of improving the quality and safety of clinical care, and institutions will remain critical in ensuring the accountability and ethical integrity of oversight of research within their systems.
1. .Report to the President and to Congress.2009.Washington, DC:Department of Health and Human Services.
2. Grossman C, Powers B, McGinnis JM.Digital Infrastructure for the e Learning Health System: The Foundation for Continuous Improvement in Health and Health Care: Workshop Series Summary.2011.Washington, DC:National Academies Press.
3. Friedman CP, Wong A, Blumenthal D.Achieving a nationwide learning health system.Sci Transl Med.2010;2:57cm29.
4. Wallace PJ.Reshaping cancer learning through the use of health information technology.Health Aff (Millwood).2007;26:w169–w177.
5. Abernethy AP, Etheredge LM, Ganz PA, et al..Rapid-learning system for cancer care.J Clin Oncol.2010;28:4268–4274.
6. D’Avolio L, Ferguson R, Goryachev S, et al..Implementation of the Department of Veterans Affairs’ first point-of-care clinical trial.J Am Med Inform Assoc.2012;19:e170–e176.
7. Gostin LO, Turek-Brezina J, Powers M, et al..Privacy and security of personal information in a new health care system.JAMA.1993;270:2487–2493.
8. Melton LJ III.The threat to medical-records research.N Engl J Med.1997;337:1466–1470.
9. Annas GJ.Medical privacy and medical research—judging the new federal regulations.N Engl J Med.2002;346:216–220.
10. Rothstein MA.Is deidentification sufficient to protect health privacy in research?Am J Bioeth.2010;10:3–11.
11. Miller FG.Research on medical records without informed consent.J Law Med Ethics.2008;36:560–566.
12. Staa TP, Goldacre B, Gulliford M, et al..Pragmatic randomised trials using routine electronic health records.BMJ.2012;344:22–25.
13. Sabin JE, Mazor K, Meterko V, et al..Comparing drug effectiveness at health plans: the ethics of cluster randomized trials.Hastings Cent Rep.2008;35:39–48.
14. Faden R, Kass N, Goodman S, et al..An ethics framework for a learning health care system: a departure from traditional research ethics and clinical ethics.Hastings Cent Rep.2013;43:S16–S27.
15. Kass N, Faden R, Goodman S, et al..The research-treatment distinction: a problematic approach for determining which activities should have ethical oversight.Hastings Cent Rep.2013;43:S4–S15.
16. Lederle FA.Chlorthalidone versus Hydrochlorothiazide: Impact on Cardiovascular Outcomes (CHICO) Trial: A VA Cooperative Study Request for Planning [Draft Proposal]. August 8, 2012. Unpublished; on file with authors.
17. Personal Communication with Frank Lederle, MD; Director, Minneapolis VA Center for Epidemiological and Clinical Research (CECR). February 15, 2013.
18. Personal Communication with Gary Gerstenblith, MD; Professor of Cardiology, Johns Hopkins School of Medicine. May 1, 2013.
19. Davis BR, Cutler JA, Gordon DJ, et al..Rationale and design for the antihypertensive and lipid lowering treatment to prevent health attack trial (ALLHAT).Am J Hypertens.1996;9:342–360.
20. Ernst ME, Carter BL, Goerdt CJ, et al..Comparative antihypertensive effects of hydrochlorothiazide and chlorthalidone on ambulatory and office blood pressure.Hypertension.2006;47:352–358.
21. Morris MC, Nelson RM.Randomized, controlled trials as minimal risk: an ethical analysis.Crit Care Med.2007;35:940–944.
22. Truog RD, Robinson W, Randolph A, et al..Is informed consent always necessary for randomized, controlled trials?N Engl J Med.1999;340:804–807.
23. Faden RR, Beauchamp TL, Kass NE.Learning health care systems and justice.Hastings Cent Rep.2011;41:3.
24. Largent EA, Joffe S, Miller FG.Can research and care be ethically integrated?Hastings Cent Rep.2011;41:37–46.