|The IRB Guidebook was last updated in 1993. Developments over the intervening years have made portions of the Guidebook information obsolete, while portions of the information remain valid. There is no errata document to indicate which information has been superseded. OHRP cautions users to verify the current validity of any Guidebook information before relying on the information in a program of human subjects protection.||
Return to Index Page
Institutional Review Board
* CHAPTER IV *
CONSIDERATIONS OF RESEARCH DESIGN
|A. Introduction||E. Epidemiologic Studies|
|Exemption from IRB Review||F. Case-Control Studies|
|Research Methodology in Science||G. Prospective Studies|
|Definitions||H. Clinical Trials|
|B. Observation||I. Identification and Recruitment of Subjects|
|C. Record Reviews and Historical Studies||J. Assignment of Subjects to Experimental and Control Groups|
|D. Surveys, Questionnaires, and Interviews|
The value of research depends upon the integrity of study results. One of the ethical justifications for research involving human subjects is the social value of advancing scientific understanding and promoting human welfare by improving health care. But if a research study is so methodologically flawed that little or no reliable information will result, it is unethical to put subjects at risk or even to inconvenience them through participation in such a study. One question that every IRB member asks is "To what degree is it our responsibility to review the underlying science of the proposed research?" Clearly, if it is not good science, it is not ethical. The federal regulations under which IRBs operate, however, do not clearly call for IRB review of the scientific validity of the research design. Nonetheless, they do require that IRBs determine whether "[r]isks to subjects are reasonable in relation to...the importance of the knowledge that may reasonably be expected to result" [Federal Policy §___.111(a)(2)]. If the underlying science is no good, then surely no important knowledge may reasonably be expected to result.
Left without clear direction on this point, most IRBs appear to take the following approach, which has been described approvingly by Robert Levine (1986, p. 21): Where the investigator conducting the research under review is seeking funding from the federal government or other extramural funding agency, rigorous review of the science is left to the agency's peer review process. The IRB provides a less detailed examination to satisfy itself that there are no obvious flaws that would place subjects at unnecessary risk. Where the protocol will not receive such detailed scientific review, IRBs review the research design with much more care, perhaps with the assistance of consultants, if the IRB itself does not possess sufficient expertise to perform such a review. Levine suggests that IRBs should establish their authority to criticize the scientific merits of protocols and to exercise that authority to require that investigators correct design flaws identified by the IRB before receiving IRB approval, but that IRBs should recognize their limits in this regard as well. [See also Commentary by Levine following McLarty (1987), p. 3.] Benjamin Freedman suggests that research must be both valid and of value [Freedman (1987b)]. Although IRB members do not need to be experts in scientific methodology or statistics, they should understand the basic features of experimental design, and they should not hesitate to consult experts when aspects of research design seem to pose a significant problem.
The purpose of this chapter of the Guidebook is to provide some basic background information on scientific research design, some of the research techniques used by scientists, and some ethical considerations raised by these designs and techniques.
The federal regulations provide for exemption from review for certain kinds of research described in this chapter (e.g., reviews of records or surveys) if certain conditions are met, unless: (1) information will be recorded by investigators in such a manner that subjects can be identified directly or through identifiers; and (2) disclosure of subjects' responses could reasonably place the subjects at risk of criminal or civil liability, or be damaging to the subjects' financial standing, employability, or reputation). The exemptions appear at Federal Policy §___.101(b). In fulfilling the provisions of their institution's Assurance, however, individual IRBs may have policies that require the review of all research involving human subjects, whether or not the research is subject to federal regulation, including research that is exempt from review under the regulations. [See Federal Policy §___.103(b)(1).] Some Sections in this chapter will describe certain kinds of research as "exempt from IRB review." This exemption refers to research subject to the federal regulations. IRBs should follow the written policies established by their institutions. [See Guidebook Chapter 1, Section A, "Jurisdiction of the Institutional Review Board."]
The pursuit of science is an attempt to understand the physical world; that is, to describe the phenomena that characterize physical reality, and, when possible, to define, predict, and even control the conditions under which these phenomena occur. Basic to scientific inquiry is an acceptance of the philosophical perspectives known as empiricism and determinism. Scientists take for granted that knowledge results from experience and is based on observations of physical events. Moreover, these physical events are assumed to follow physical laws in that they depend upon causal factors that can be discovered.
Scientific understanding, then, must be based on objective, systematic observation of physical events and on analytical reasoning, or inference, that is truly logical. The two adjectives used here, objective and systematic, describe critical characteristics of the observations upon which science is based. Objective observations can be experienced directly and are repeatable, making it possible for scientists to verify each others' work. Systematic observations are obtained under clearly specified, and, where possible, controlled conditions that can be measured and evaluated. Research methodology provides the tools needed to produce objective and systematic observations, called empirical data, and to ensure that inferences based on these observations are grounded in logic.
Scientists develop theories to organize their empirical observations. A theory is a set of principles that attempts to explain the causal factors underlying related scientific observations. The usefulness of any theory depends upon its internal consistency, its ability to account for existing data, and its precision in prediction. Scientists use hypotheses to generate predictions that can be tested empirically. It is important to understand that scientific theories and hypotheses can never be "proven true" but can only be supported (confirmed) or not supported (disconfirmed) by currently available data.
Biomedical investigations can be broadly categorized into two types: experimental studies and descriptive studies. A true experimental study is one in which subjects are randomly assigned to groups that experience carefully controlled treatments manipulated by the experimenter according to a strict logic allowing causal inference about the effects of the treatments under investigation. Descriptive studies, although objective and systematic, lack the rigid control achieved through random assignment of subjects and precise manipulation of treatment conditions. As a result, causal inferences cannot logically be derived from descriptive studies.
Some behavioral research involves only observation of people in public places (e.g., observing shopping or eating habits). Where the subjects are adults, research of this type is exempt from IRB review unless the information obtained is recorded in such a manner that the subjects can be identified, and the information obtained could reasonably place the subjects at risk of criminal or civil liability or be damaging to the subjects' financial standing, employability, or reputation [Federal Policy §___.101(b)(2)]. For research to which the DHHS regulations are applicable, observational research involving the public behavior of children is also exempt, as long as, in addition to the above criteria, the investigator does not participate in the activities being observed [Federal Policy §___.101(b)(2); 45 CFR 401(b)].
Observational studies that involve intervention in or manipulation of the subjects' environment do require IRB review. For example, an investigator studying reactions to emergencies may want to modify the environment and then observe people's reactions in public places. Some researchers studying this phenomenon have contrived "emergencies" with the help of confederates who pretend to have a heart attack on the subway or to be victims of an assault in a public park. Responses of passersby are recorded. Because there is a risk of inducing a real medical emergency or causing psychological distress in a bystander, such research must be reviewed by an IRB. Similarly, if people are to be observed in places or circumstances in which they have a reasonable expectation of privacy, the research must be reviewed by an IRB.
Sometimes a study involves only the use of existing public or privately held records. In such a case, an IRB could exempt the study from review, give it expedited review, or subject it to full board review, depending on the nature of the study and the policy of the institution's IRB [Federal Policy §§___.101(b)(4), ___.110, and ___.111]. For example, if a researcher wanted to know whether the conviction rates for various violent crimes vary from one part of the country to another he or she could examine public records (e.g., court or police records) in different parts of the country. Variations related to sex, race, age, and so forth could also be studied. Such research, utilizing only information available in public documents, would be exempt from IRB review [Federal Policy §___.101(b)(4)].
On other hand, if a researcher wanted to learn about risk factors (e.g., smoking habits, industrial employment, or family history) related to cancer, he or she might start with medical records. This research would be exempt from review under the federal regulations if the records preexist the start of the research project and if the investigator records the information in such a manner that subjects cannot be identified directly or through identifiers [Federal Policy §___.101(b)(4)]. If, however, such identifiers are to be recorded, the research would require IRB review to ensure that, among other things, procedures for protecting privacy and confidentiality are adequate. Furthermore, the investigator studying cancer risk factors may propose to go on to contact the subjects (if still living) or family members (if the subject is deceased) to gather additional information, which may or may not be subject to the federal regulations. Note, however, that some IRBs review all research involving human subjects, even where the research is exempt under the federal regulations, and that records research that is conducted without the prior consent of the subjects raises privacy concerns, which are discussed in the Section on epidemiologic studies, below. [See Chapter 4, Section D, "Surveys, Questionnaires and Interviews," and Chapter 4, Section E, "Epidemiologic Studies."]
Surveys, questionnaires, and interviews are commonly used in social science disciplines such as anthropology, economics, political science, psychology, and sociology. Statistical procedures are used to ensure that the sample interviewed or questioned properly represents the subject population and to estimate measurement and sampling error so that valid and reliable inferences may be drawn about the population surveyed.
Some surveys use interview methods to obtain information directly from individuals. Unlike informal interviews often used in clinical settings or informal surveys, in standardized interviews those interviewed respond to a predetermined set of questions asked by a trained interviewer. The interview instrument (questionnaire) is systematically developed and pretested on a small number of people drawn from the subject population so that any ambiguities or biases in the way the questions are stated can be identified and corrected.
Research involving survey or interview procedures with adult subjects is exempt from the federal regulations unless the information obtained is recorded in such a manner that the subjects can be identified, and the information obtained could reasonably place the subjects at risk of criminal or civil liability or be damaging to the subjects' financial standing, employability, or reputation [Federal Policy §___.101(b)(2)]. Survey and interview research involving children is not exempt, but rather requires full IRB review [Federal Policy §___.101(b)(2); 45 CFR 401(b)]. Furthermore, some IRBs review all research involving human subjects, even where the research is exempt under the federal regulations. [See Guidebook Chapter 4, Section E, "Epidemiologic Studies."]
Epidemiologic studies present several unique problems because they often use sensitive private documents, such as medical records, and link them with other data, such as employment, insurance, or police records. They also often combine historical research with survey and interview techniques. There is some debate in the literature on the question of whether epidemiologic research is exempt from IRB review. [See Guidebook Chapter 1, Section A, "Jurisdiction of the Institutional Review Board."] Nevertheless, epidemiologic studies do present significant problems regarding privacy and confidentiality, issues that IRBs that do review such research must address. A set of ethical guidelines for epidemiologists has been developed, which IRBs may wish to consult. [See Beauchamp, et al. (1991).]
In epidemiologic studies, the investigator is attempting to identify risk factors for particular diseases, conditions, or behaviors, or risks that result from particular causes, such as environmental or industrial agents. The research techniques usually employed involve record reviews to identify potential subjects, followed by telephone or in-person surveys or interviews, or mailed questionnaires. Epidemiologic studies may also be limited to reviews of records from various sources (e.g., medical, employment, and police records), which the investigator links together. The validity of epidemiologic studies requires a very high degree of participation (as much as 90 percent) by potential subjects. The behavioral component of the factors often studied in epidemiologic research means that significant rates of nonparticipation are likely to produce biased findings. IRBs need to balance the need for high participation rates against the ethical concerns raised by epidemiologic research.
The role usually played by IRBs reviewing epidemiologic research is to ensure that epidemiologists:
take adequate steps to preserve the confidentiality of the data they collect, requiring that they specify who will have access to the data, how and at what point in the research personal information will be separated from other data, and whether the data will be retained at the conclusion of the study. IRB reviewers also require a thorough description of interview instruments and questionnaires, and they make sure that the informed consent of subjects will be obtained before interviews are conducted [Wallace (1982), p. 287].
They should also, as they do with other research protocols, require epidemiologists "to justify particular projects according to their anticipated risks and benefits" [id]. With respect to data retention, IRBs should note that other institutional or regulatory policies (e.g., those concerning scientific integrity) may require that data be retained for some period of years.
The primary ethical concerns presented by epidemiologic studies are protection of subjects' privacy (i.e., the right "to determine what will be known about oneself") and the confidentiality of data (i.e., the determination that information will not be disclosed without permission) [Wallace (1982), pp. 277, 278]. Privacy concerns in turn raise questions about the role of informed consent. Even where subjects are not at risk of harm from epidemiologic research, access to records for which individuals have not consented clearly constitutes an invasion of privacy, a moral wrong [Capron (1991)]. The multitude of records that are now kept as a routine part of our daily lives (e.g., medical, employment, insurance, and school records) constitutes a wealth of information, but information in which we have an expectation of privacy. In particular, we expect that the privacy of those records will be maintained, and their contents will be kept confidential. Access to those records without prior consent of the subject raises concerns about the violation of the ethical principle of respect for persons (sometimes referred to as autonomy).
When a study involves reviews of records without any contact with individuals, it can be argued that the subjects of the research are at no risk of harm, beyond the "wrong" of invasion of privacy, unless their identity is or can be linked to the research records. Such linkage is often used in epidemiological research, in which case IRBs must ensure that subjects' privacy interests will be adequately protected. Some commentators have suggested that those interests should be balanced against the importance of the research; others argue that the right of privacy cannot or should not be overridden by the value of the research. However, the obtaining of prior consent as a means of eliminating the problem of invasion of privacy may, as a practical matter, be impossible. The issue of consent is discussed below and in Chapter 4, Section I, "Identification and Recruitment of Subjects."
Where the investigator will have personal contact with subjects, however, a potential for harm does exist. Since they are identified as potential subjects because they either have or are at risk of developing a disease or condition, simple contact with subjects may present a risk of harm, either because of sensitivity to discussing a disease or condition they know they have, or because they may not be aware of their condition. Where the person with the disease or condition is deceased, investigators may want to contact relatives who may not have been aware of the deceased's condition. The potential for harm is greatest in the early stages of the research, when the investigator is identifying appropriate subjects for study. Once potential subjects are identified, the investigator can obtain their consent to participation in the study.
Consider the case of epidemiologic research into risk factors for HIV infection [human immunodeficiency virus (HIV) is the virus that causes acquired immune deficiency syndrome (AIDS)]. Potential subjects are, by definition, under investigation because of an anticipated relationship to HIV (except for control subjects, who may or may not know which group they are in). Members of known risk groups (e.g., injecting drug users, homosexual males, individuals with hemophilia) may face considerable emotional disturbance by being contacted for an HIV study. In cancer studies as well, potential subjects (or their relatives) may be disturbed by the prospect of discussing their medical condition or experience.
With respect to confidentiality, disclosure of information such as that usually collected in epidemiologic studies also presents an ethical concern that IRBs should address. All information collected as part of a study is confidential: Data must be stored in a secure manner and must not be shared inappropriately. The threat of disclosure of data that can be linked to individuals represents another risk of harm to individuals. In properly designed studies, this risk is insignificant. To maintain this confidentiality, researchers must be prepared to resist subpoenas seeking to obtain research data. [See Guidebook Chapter 3, Section D, "Privacy and Confidentiality." The section on confidentiality of research data discusses §301(d) of the Public Health Service Act, which provides for protection of research data.] Using HIV as an example again, subjects included in an HIV-related study would be understandably concerned about the confidentiality of the data, since breaches in confidentiality could have severe adverse consequences such as loss of employment or insurance coverage, or criminal charges. OPRR guidance on HIV studies states that:
where identifiers are not required by the design of the study, they are not to be recorded. If identifiers are recorded, they should be separated, if possible, from data and stored securely, with linkage restored only when necessary to conduct the research. No lists should be retained identifying those who elected not to participate. Participants must be given a fair, clear explanation of how information about them will be handled.
As a general principle, information is not to be disclosed without the subject's consent. The protocol must clearly state who is entitled to see records with identifiers, both within and outside the project. This statement must take account of the possibility of review of records by the funding agency.... [OPRR (1984).]
[See also Guidebook Chapter 5, Section F, "AIDS/HIV-Related Research."]
Another question is whether and at what point subjects must consent to epidemiologic research: prior to selection but after first contact, before the first contact, or before gaining access to records (through the custodian of the records). In general, wherever possible, potentially eligible subjects should be contacted either by the person to whom they originally gave the information, by a person with whom they have a trust relationship [McCarthy and Porter (1991), p. 239]. Guidebook Chapter 4, Section I, "Identification and Recruitment of Subjects," describes the various approaches commonly used for obtaining consent at the subject identification stage.
Where identifiers that can be linked to individuals will be used, each subject must provide informed consent prior to participation, except in certain limited circumstances. The federal regulations allow for waiver or alteration of consent requirements under the following conditions: (1) the research involves no more than minimal risk; (2) the waiver or alteration will not adversely affect the rights and welfare of the subjects; (3) the research could not practicably be carried out without the waiver or alteration; and (4) whenever appropriate, the subjects will be provided with additional pertinent information after participation [Federal Policy §___.116]. Further, the when the study involves the collection of information of a sensitive nature (e.g., sexual or criminal activity), an investigator may request that the requirement to obtain written consent be waived. IRBs may waive the requirement for the investigator to obtain a signed consent form for some or all subjects if it finds either: (1) that the only record linking the subject and the research would be the consent document, and the principal risk would be potential harm resulting from a breach of confidentiality; or (2) that the research presents no more than minimal risk of harm to subjects and involves no procedures for which written consent is normally required outside of the research context [Federal Policy §___.117].
Commentators on the subject of consent in epidemiologic studies agree that some relaxation of the usual informed consent requirements may be necessary. [See, e.g., Beauchamp, et al. (1991), pp. 159s-161s; McCarthy and Porter (1991), pp. 238-39]. Where prior consent to participation in a survey or record review is precluded by the research design, the investigator might use a veto approach, in which the subject can elect not to have his or her data included in the study. [See Capron (1991), p. 86s] It has also been suggested that epidemiologic research conducted without consent should meet four requirements: "(1) the invasion of privacy involved must be necessary to the conduct of the research; (2) the invasion of privacy must involve only a minimal intrusion; (3) the research must additionally present only insignificant risk of specifiable harms to the interests of subjects; and (4) the results of the research must be likely to bring social benefits of a significant nature" [Wallace (1982), p. 280 and ff].
In those cases where informed consent will be obtained, the specific information that the investigator will give a potential subject, both at the time of first contact and in the consent negotiations, should be considered by the IRB. McCarthy and Porter (1991) provide some useful guidance on the information that should be communicated to subjects. They suggest that the information provided to prospective subjects should include descriptions of: the kind of data that will be collected, the identity of the persons who will have access to the data, the safeguards that will be used to protect the data from inappropriate disclosure, and the risks that could result from disclosure of the data. If identifiers will be collected and retained, subjects should be so informed, and should also be told whether they will be contacted again in the future. The investigator should also provide subjects with a written assurance that any publications that result from the research will present the data only in aggregate form, and in such a manner that individuals cannot be identified. Investigators should inform subjects of what information gained from the study will be passed along to them (e.g., the presence of diseases or conditions they may not have known about) [McCarthy and Porter (1991), p. 239].
When epidemiologic research involves particularly vulnerable populations, an IRB should consider seeking the advice of persons sensitive to their concerns [Federal Policy §107(a)]. Such consultation may help the IRB identify and resolve sensitive ethical concerns.
Various writers have also suggested that consent to epidemiological research should be sought from the communities in which the research will be conducted. [See, e.g., McCarthy and Porter (1991), p. 239.] Further, "the size, composition, mixture, and origin of the study population should be chosen with great care to avoid or minimize community or group harms" [p. 240]. [See also Council for International Organizations of Medical Sciences (1991).]
One popular type of descriptive study is the case-control study, in which persons with a specific condition (the cases) and persons without the condition (the controls) are selected to participate in the study. The proportions of cases and controls with certain characteristics (e.g., exposure to a particular drug) are then compared.
In the usual case-control study, there is no risk of physical injury since no interventions are performed. Such studies may, however, entail legal risks, where, for instance, a study may reveal illegal drug use; or psychological risks, where the investigation reviews traumatic experiences, such as the loss of a child. IRBs should make certain that the investigator has made adequate provisions to protect privacy, assure confidentiality of data, and respect the subject's rights (including refusal to participate). Each case-control study should be considered individually by the IRB, since different levels of protection are needed for different studies.
Most case-control studies require investigators to review medical records and interview subjects, or, when subjects are deceased, their next-of-kin. This type of study may require review by the full IRB, which should assure that adequate informed consent will be obtained and that the investigator will use a suitable system for contacting subjects. Case-control studies that are limited solely to the review of existing records may be appropriate for expedited review or for exemption from review, depending on the nature of the study. [See Federal Policy §§___.101(b)(4) and ___.110; see also Guidebook Chapter 4, Section D, "Surveys, Questionnaires and Interviews," and Chapter 4, Section I, "Identification and Recruitment of Subjects."]
A prospective study is designed to observe events (e.g., diseases, behavioral or physiological responses) that may occur after the subjects have been identified. All concurrently controlled clinical trials are prospective.
Longitudinal studies follow one or more subject cohorts over an extended period of time. The duration of the study may or may not be specified at the outset of the investigation. For example, several large-scale longitudinal studies whose original purpose was to study children eventually followed their subjects into adulthood, and, later, into old age.
Cohorts in longitudinal studies are sometimes divided into those who have and those who have not been exposed to some risk factor prior to the initiation of the investigation [e.g., a study that follows the occurrence of diseases in workers in a particular industry compared to a group of persons not in that industry (the controls) but matched for age, sex, smoking and drinking habits, and other relevant factors]. The well-known Framingham Study, begun in the 1950s, was designed to monitor the incidence of coronary artery disease in over 5,000 residents who were examined every two years for a period of twenty years. This study has yielded important data demonstrating the relationship between various factors (smoking, obesity, diet, and high blood pressure) and the development of heart disease.
The clinical trial is an important research design used to assess the safety and efficacy of new drugs, devices, treatments, or preventive measures in humans by comparing two or more interventions or regimens. Clinical trials are frequently "multicentered," that is, a number of research institutions may cooperate in a common study protocol.
Clinical trials can be used to evaluate most treatments or preventive measures for almost any condition or disease. Clinical trials (sometimes called "randomized clinical trials," or RCTs) are controlled (one subject group receives the treatment under investigation while a control subject group receives either another treatment or no treatment), with participants being randomly assigned to either the subject or control group. They are also either single- or double-masked, so that either the investigator, the subjects or both do not know who is in the treatment or control group until the conclusion of the study.
Randomized clinical trials present numerous ethical issues, some of which are discussed in other Sections of this Guidebook as well as in this Section (e.g., Chapter 4, Section J, "Assignment of Subjects to Experimental and Control Groups.")
A primary ethical concern is one of fairness: If the trial therapy is known to be superior to currently available alternative therapies (i.e., prior research indicates that it is superior), it is unethical to assign subjects to the inferior treatment. Furthermore, it would not be ethical to perform a clinical trial comparing two treatments when there is a third therapy that is known to be superior to either or both, unless there is some reason why that therapy is not useful for the study population. Researchers must therefore honestly be able to state a null hypothesis (also called "theoretical equipoise": the assumption that subjects treated with therapy A - the trial therapy - will not differ in outcome from subjects treated with therapy B - the control therapy) before beginning a randomized clinical trial [Freedman (1987)]. According to a somewhat broader concept called "clinical equipoise," a randomized controlled design may be justified where there is a "current or likely dispute among expert members of the clinical community as to which of two or more therapies is superior in all relevant respects" [Levine, Dubler, and Levine (1991), p. 3, restating Freedman (1987)]. Furthermore, the trial must be designed such that its "successful completion will show which [of the therapies] is superior" [Freedman (1990), p. 5]. The "results of a successful clinical trial should be convincing enough to resolve the dispute among physicians" [Freedman (1987), p. 144]. The control treatment must be the best standard therapy currently available for the condition being treated [Freedman (1990); Levine (1986), (1985)].
Placebos. Placebos may be used in clinical trials where there is no known or available (i.e, FDA-approved) alternative therapy that can be tolerated by subjects. IRBs should scrutinize studies that propose to use placebos to ensure that subjects are not deceived into believing that they have received an active agent.
Where the disease is lethal or seriously debilitating, however (as in the case of HIV), the use of a placebo control in place of an active control may be, and indeed has been, questioned. The onslaught of HIV has led to considerable discussion of clinical trial design and the need to maximize benefits in every arm of the trial. [See, e.g., Levine, Dubler, and Levine (1991) and Freedman (1990).] A design involving a placebo control should not be used where there is a standard treatment that has been shown to be superior to placebo by convincing evidence [Freedman (1990)]. It has been argued that placebo controls must be used, however, when the experimental treatment is of "dubious efficacy" or when there are known serious side effects [Freedman (1990); Levine (1985), 1986)]. The use of placebos in controlled clinical trials must be justified by a positive risk-benefit analysis, and subjects must be fully informed of the risks involved in assignment to the placebo group. There is a consensus that continued assignment of subjects to placebo is unethical once there is good evidence to support the efficacy of the trial therapy. Clinical trials should be stopped or their protocols modified when there is sufficient evidence of either a beneficial therapeutic effect or unacceptable side effects. Monitoring for such information during the course of the trial is discussed in Guidebook Chapter 3, Section E, "Monitoring and Observation."
Some drug trials involve a period during which all subjects receive only a placebo prior to the initiation of the study. This period is called a "placebo washout." The purposes of a washout period include: (1) terminating the effects of any drug the subject may have been taking before entering the clinical trial, so that the effects of the trial drug - and only the trial drug - may be observed; (2) learning whether subjects cooperate with instructions to take drugs ("compliance"); and (3) learning which subjects are "placebo responders," in that they experience a high degree of placebo effect. In some protocols, the investigators plan to exclude those subjects they find either poorly compliant or highly responsive to the placebo. The risks entailed in withdrawing subjects from therapy during a placebo washout period should be carefully evaluated by the IRB; great care must be taken to exclude subjects who are vulnerable to injury if they are withdrawn from effective therapy. In studies involving a placebo washout, subjects should be told that at some point during the study all subjects will receive placebo treatment; investigators but not subjects will know when subjects are receiving placebos for washout purposes, so that during the washout, the study is single-masked.
See also Chapter 4, Section J, "Assignment of Subjects to Experimental and Control Groups."
Informed consent. Informed consent is of particular importance in randomized clinical trials as in all prospective studies. The IRB should assure that initial, and, where necessary, continuing consent is obtained from the subjects at critical intervals (e.g., at points where the study protocol is materially altered, new procedures are introduced, new information - about risks or benefits, for instance - becomes available, or toxicity becomes manifest). In such instances, blanket consent at the beginning of the study does not suffice. Because subjects may forget crucial aspects of the trial, it may also be advisable periodically to ascertain continuing consent in long-term studies.
Despite the fact that subjects may be kept unaware of their treatment assignments in "masked" studies and research involving placebos, the information provided to prospective subjects should clearly communicate the nature of the study design, method of treatment assignment (including the probability of assignment to the various groups), possible interventions, and the implications of the possible interventions. Ethical considerations demand that subjects be informed when their assignment will be random, and that one of the possible consequences of participation is that the group to which they are assigned will turn out to have received the less effective intervention.
Subjects must be fully informed of the likelihood of receiving the experimental treatment. For example, if there are two subject groups, experimental and control, and the assignment of subjects to groups is random, subjects must be informed that they have a 50 percent chance of receiving the experimental therapy and a 50 percent chance of receiving the alternative treatment. If the alternative "treatment" is placebo, subjects must be so informed. Informing subjects that the study involves the use of placebos and the probability of their being assigned to the placebo group eliminates the ethically objectionable element of deception from the study. Further, subjects should be told who will know whether they are receiving the placebo or the active agent. In a double-masked trial, for example, subjects should be told that neither they nor the investigator will know whether they are receiving the placebo or the experimental therapy.
Since assignment to one or another of the interventions should take place after informed consent has been given, the subject must be made aware of all the possible alternative interventions and what is known about the efficacy and safety of each. Prospective subjects should also be told whether participation in the study precludes participation in other programs or therapeutic regimens that may be beneficial to them and the extent to which this restriction presents a risk.
A more fundamental consent question centers on communicating the uncertainties and risks involved in clinical trials to prospective subjects. Particularly in Phase 1 trials, where the safety of therapies in humans is first being tested, IRBs must assure themselves that those risks and uncertainties will be clearly communicated to prospective subjects, and that the process of communication with subjects will continue throughout the study.
A related issue is the selection of subjects. Subjects who are particularly vulnerable, such as persons who are desperately ill, are, perhaps, more likely than others to be willing to accept great risks in the hope that they will benefit from an experimental treatment. IRBs need to ensure that their welfare is protected by requiring full and open disclosure of risks and benefits, while at the same time avoiding paternalism. Vulnerable subjects should not be precluded from studies solely on the basis of their vulnerability. To do so would preclude them from the opportunity to benefit from the availability of investigational therapies through research studies. [See also Guidebook Chapter 3, Section A, "Risk/Benefit Analysis," Chapter 3, Section C, "Selection of Subjects," Chapter 5, "Biomedical and Behavioral Research," and Chapter 6, "Special Classes of Subjects."]
This Section deals specifically with practical aspects of how investigators go about identifying and recruiting individual subjects, and IRB considerations related to these activities. These considerations are especially important in epidemiologic research.
See also Guidebook Chapter 3, Section C, "Selection of Subjects," and Chapter 3, Section G, "Incentives for Participation." "Selection of Subjects" deals with the question of equitable selection of subjects in terms of defining the appropriate group of subjects for a research project. "Incentives for Participation" deals in greater depth than the present Section with incentives offered to encourage participation and the ethical concerns of coercion and undue influence.
Using Records to Identify Subjects. IRBs are responsible for ensuring the equitable selection of research subjects [Federal Policy §___.111(a)(3)]. In fulfilling this responsibility, IRBs should review the methods that investigators use to recruit subjects.
Subjects with specific diseases or conditions are often identified as potential subjects through some type of record (e.g., registries for cancer cases, surgical or X-ray log books, employment or school records). Controls may come from the same population as the subjects (which is always the case in a randomized clinical trial), be persons with unrelated conditions or be volunteers from the general population. Potential subjects may be identified through records maintained at hospitals or physicians' private offices. If potential subjects are identified through medical records, log books, physicians' records, or other records that are not public documents, the IRB should make certain that the following conditions have been met: (1) the investigator is allowed access to such records by the institution or the physician; and (2) responsibility for confidentiality and protection of privacy is clearly accepted by the investigator.
Sometimes, as in epidemiologic research, it is necessary for an investigator to review thousands of medical records to identify a very small number of subjects who are suitable for a study. At present, there is no agreement among commentators as to whether the investigator needs consent from all patients whose records will be searched, or only from the few who are selected for the study. An alternative in some circumstances may be the use of a "data broker," that is, an intermediary who already has access to the data. The broker can review records to identify appropriate subjects, whose consent to participate in the study can then be sought. With automated record keeping systems, it may be easier to identify appropriate subjects without reviewing all the records. Where the records are not computerized, however, IRBs will have to decide under what conditions a scientist may scan thousands of medical or other private records while searching for a small number of appropriate subjects. One factor to consider would be the sensitivity of the information likely to be contained in the records. For example, did the patients have broken ankles or abortions? Were they treated for strep throat or venereal disease? Another factor to consider is the type of information the investigator wishes to obtain from those who are selected as suitable subjects for the study.
Some institutions notify patients at the time of admission or initial treatment that: (1) their records may be used for research purposes; and (2) precautions will be taken to ensure that if the records are used, the researchers will respect and protect the confidentiality of the records. Some institutions go further and provide an opportunity for patients to consent or refuse consent to such use of their records at the time of admission.
In the event that names are sought from physicians' private offices, the patient's physician should request permission from the patient to release his or her name.
For a hospital-based study, most IRBs require that a potential subject's physician give approval before the subject is contacted, particularly when there may be medical or emotional contraindications to participation. If the subject is in the hospital, someone on the hospital staff may inform the patient that he or she is going to be invited to participate in a study, or, more often, an interviewer may approach the subject directly after consultation with his or her physician.
If the subject has left the hospital, various options may be considered. (For each option, most IRBs require that the potential subject's physician give approval before the subject is contacted.) For instance, the investigator may send a letter describing the purpose of the study and requesting that the subject return a postcard indicating whether he or she would like to participate. The effectiveness of this method depends on how many of the postcards are returned.
A second option is to invite participation by letter, and for the subject to send back a postcard (or to telephone) only if he or she does not wish to participate. If no postcard is returned, the subject may then be contacted by an interviewer. This method is less preferable, as it requires that potential subjects take positive action to avoid being made part of a study rather than the other way around. Subjects may become unwitting participants if, for example, they never receive the letter, don't read English, or are simply confused by the instructions. This approach also raises privacy concerns for certain types of research (e.g., research involving sexually transmitted diseases or psychiatric illness, or drug or alcohol abuse).
A third approach that is often used is for the patient's physician to send a letter informing the subject about the study and inviting the patient to participate. This method may work well if the study is being undertaken by a relatively small number of physicians who are willing to cooperate with the investigator. Response rates are likely to be high, since the subject often considers it significant that the letter has come from his or her own physician. IRBs should consider whether use of this method will subject potential participants to coercion or undue influence. Finally, the investigator can send a letter to the potential subject explaining the purpose of the study, and then an interviewer can call to invite the potential subject to participate. By permitting interchange between the subject and interviewer, this method allows the subject to make an informed decision about participation. Although there is the risk of coercion by the interviewer, in general this method helps the subject better understand what the purposes of the research are, why his or her participation is important, what procedures are used to protect confidentiality, and what would be asked of him or her as a participant. This approach usually secures the highest response rate; however, people may be offended, especially in research on sensitive topics, by the investigator's having direct access to their name, address, and phone number. IRBs should be sensitive to this concern.
Advertising for Subjects. One method of recruiting subjects is through advertisements (e.g., posted notices and newspaper or magazine ads). Advertising for research subjects is not, in and of itself, an objectionable practice. When advertising is to be used, however, the FDA requests that IRBs review the information contained in the advertisement, as well as the mode of its communication, to determine whether the procedure for recruiting subjects affords adequate protection. IRB review is necessary to ensure that the information is not misleading to subjects, especially when a study will involve persons with acute or severe physical or mental illness, or persons who are economically or educationally disadvantaged.
The FDA believes that any advertisement to recruit subjects should be limited to: (1) the name and address of the clinical investigator; (2) the purpose of the research, and, in summary form, the eligibility criteria that will be used to admit subjects into the study; (3) a straightforward and truthful description of the incentives to the subject for participation in the study (e.g., payments or free treatment); and (4) the location of the research and the person to contact for further information [FDA IRB Information Sheet: "Advertising For Study Subjects" (1989)].
If a study involves investigational drugs or devices, no claims should be made, either explicitly or implicitly, that the drug or device is safe or effective for the purposes under investigation, or that the drug or device is in any way equivalent or superior to any other drug or device. Such representation would not only be misleading to subjects, but would also violate FDA regulations concerning the promotion of investigational drugs [21 CFR 312.7(a)] and investigational devices [21 CFR 812.7(d)].
Paying Subjects to Participate. Another method of recruiting research subjects is to pay them for their participation. It is not uncommon for subjects to be paid for their participation in research, especially in the early phases of investigational drug or device development. In such cases, the IRB should review both the amount of payment and the proposed method of disbursement to assure that neither entails problems of coercion or undue influence. Such problems might occur, for example, if the entire payment were to be contingent upon completion of the study or if the payment were unusually large. Payments should reflect the degree of risk, inconvenience, or discomfort associated with participation. See Guidebook Chapter 3, Section G, "Incentives for Participation."
The choice of study design depends largely upon the nature and goals of the research. Good methodology requires that studies be designed to minimize bias both in assignment to treatment groups (e.g., by randomizing) and in assessment of outcome. Bias may enter into a study in several ways. The investigator may have strong beliefs or hopes regarding the success of a particular intervention or the truth of a particular hypothesis; these expectations may unconsciously influence his or her evaluation of the outcome of the research. To avoid this possibility, it is now accepted and preferred practice to conduct controlled investigations by dividing subjects into at least two groups: those who receive the experimental intervention (the experimental or treatment group) and those who do not (the control group).
See also Guidebook Chapter 4, Section H, "Clinical Trials," for additional discussion of the issues raised by random assignment of subjects to treatment groups and the use of placebos.
Random Assignment. To minimize the possibility that investigators may consciously or unconsciously select one sort of subject (e.g., the most intelligent or the least sick) for the experimental group, it has become accepted and preferred practice to devise methods of randomly assigning subjects to experimental and control groups, unless there are important scientific or ethical reasons to do otherwise. The justifying pre-condition for ethical use of randomization is that a null hypothesis (i.e., the stated experimental hypothesis that the experimental and control conditions have equally beneficial effects) can be reasonably entertained.
Sometimes experimental subjects and control subjects are assigned to groups upon admission to the study and remain in those groups for the duration of the study. This design is called "parallel control." On the other hand, it is sometimes advisable to let each subject be his or her own control by having the subject be on both regimens - first the experimental treatment and then the control, or vice versa. This "cross-over design" can be useful because it reduces variability (since every subject receives both the experimental intervention and the control treatment) and it requires fewer subjects.
Several conditions are required for the cross-over design to be appropriate. First, the condition or disease must be stable, and, if relieved, not permanently cured by either the intervention or control. Second, there must be no "carry-over" effect from the first to the second treatment assignment (e.g., in a drug study, sufficient time must elapse between treatments to ensure that all traces of the first treatment have been eliminated). Unfortunately, it is often difficult to demonstrate that these conditions exist.
Regardless of the type of design, random assignment of subjects to treatment groups is generally the preferred method in most controlled studies. Random assignment is preferred because other techniques hold the opportunity for bias in the selection of subjects for particular treatments. Assigning every other consenting subject to a given treatment or assigning subjects to a treatment group based on the day of hospital admission are not truly random methods of assignment.
In situations where the course of the disease (under currently available standard treatment) is so clear or well-known that randomization is not ethically possible (see Guidebook Chapter 4, Section H, "Clinical Trials"), a historical control design may be an alternative design choice. In historically controlled studies, the condition of subjects is compared with their own past condition on a prior regimen or treatment. For example, if a disease is known to be fatal in 80 percent of the cases and no conventional therapy exists, an experimental treatment with a good chance of success (e.g., based on the results of animal studies) might be offered to all eligible subjects identified by the investigators. It may not be ethically acceptable to ask subjects to accept the possibility of assignment to a control (untreated) group if the null hypothesis cannot reasonably be entertained.
In some circumstances, however, the use of historical controls can yield erroneous conclusions. Because changing standards of hygiene, lifestyle patterns, and medical care can markedly alter the course of a disease, the use of historical controls to demonstrate the effectiveness of a new treatment could be misleading. Small effects are particularly difficult to detect with such designs.
Placebos. To minimize the possibility that the investigators' beliefs or hopes regarding the outcome of the research will bias their evaluation of the subjects' responses, investigators may be kept unaware of the identity of subjects who are assigned to each treatment group. Similarly, the subjects' hopes for a "cure" for a disease or their fears of side effects may cause them to experience improvement or adverse effects unrelated to the experimental treatment. Furthermore, it is generally agreed that a substantial number of patients will experience improvement in their symptoms regardless of treatment. To reduce the possibility that subjects' responses will result from their expectations rather than the interventions under study, it has become accepted and preferred practice to have subjects be unaware whether the "treatment" they are given is the experimental intervention. In clinical trials, control subjects may be given either a conventional treatment, or, if none is available or appropriate, a placebo - an inert substance prepared to resemble an experimental drug in size, shape, color, taste, etc. The use of placebos is generally unacceptable if there is an effective therapy that the subjects could be receiving for relief of severe symptoms or amelioration of a serious condition. [See also Guidebook Chapter 3, Section B, "Informed Consent," and Chapter 4, Section H, "Clinical Trials."]
Some drug trials involve a period during which all subjects receive only a placebo prior to the initiation of the study. This period is called a "placebo washout." The purposes of a washout period include: (1) terminating the effects of any drug the subject may have been taking before entering the clinical trial, so that the effects of the trial drug - and only the trial drug - may be observed; (2) learning whether subjects cooperate with instructions to take drugs ("compliance"); and (3) learning which subjects are "placebo responders," in that they experience a high degree of placebo effect. In some protocols, the investigators plan to exclude those subjects they find either poorly compliant or highly responsive to the placebo. The risks entailed in withdrawing subjects from therapy during a placebo washout period should be carefully evaluated by the IRB; great care must be taken to exclude subjects who are vulnerable to injury if they are withdrawn from effective therapy. In studies involving a placebo washout, subjects should be told that at some point during the study all subjects will receive placebo treatment; investigators but not subjects will know when subjects are receiving placebos for washout purposes, so that during the washout, the study is single-masked (see below).
When both the investigator and the subjects are unaware of the treatment assignments, the design is called "double-masked;" when one or the other (but not both) know, it is called "single-masked." Whenever the investigator remains unaware of the treatment that subjects are receiving, it is important that someone be able to find out, in case it becomes necessary to protect the health and well-being of a subject (i.e., in case of serious adverse effect or deterioration of a patient's condition). Therefore, investigators usually arrange for an independent person to have access to a code indicating the identity of subjects assigned to each treatment. This independent person is given the authority to break the code for individual subjects in case of emergency. This arrangement permits treatment to be provided, as necessary, to a particular subject without breaking the "masked" aspect of the experimental design. The protocol should describe how these arrangements will be made. [See also Guidebook Chapter 4, Section H, "Clinical Trials."]
1. Does the study involve reviews of records, observation, surveys, or interviews? If so, does it qualify for exemption or expedited review under the federal regulations and institutional policy?
2. Is the scientific design adequate to answer the questions posed? Is the sample size (number of subjects) adequate? Is the method proposed for selecting and assigning subjects to treatment groups unbiased?
3. Does the investigator serve a dual role that may pose a conflict of interest?
4. Is any of the information to be collected sensitive (e.g., related to sexual practices, substance abuse, or illegal behavior)?
5. Are there adequate plans to protect participants from the risks of breach of confidentiality and invasion of privacy?
6. Are there plans for approaching subjects in a way that will respect their privacy and their right to refuse? If the protocol involves an epidemiologic study, will subjects or their relatives be protected from learning inappropriate information?
7. Does the recruitment process protect subjects from being coerced or unduly influenced to participate? Are any payments to subjects reasonable in relation to the risks, discomfort, or inconvenience to which subjects will be exposed?
8. Are there adequate plans to exclude subjects who are vulnerable to injury during the period of withdrawal of active and effective therapy, if that is part of the research design?
9. Have the rights and interests of vulnerable subjects (e.g., desperately ill persons) been adequately considered?
10. Are all appropriate elements of informed consent clearly provided for [Federal Policy §___.116], including:
a. Do the consent documents describe the study design (including plans for randomization, use of placebos, and the probability that the subject will receive a given treatment) and conditions for breaking the code (if the study is masked)?
b. Do the consent documents describe the risks and benefits of each of the proposed interventions and of alternative courses or actions available to the participants?
c. Do the consent documents clearly describe the extent to which participation in the study precludes other therapeutic interventions?
d. Are provisions made for supplying new information to subjects during the course of the study and for obtaining continuing consent, where appropriate?
e. Must investigators obtain consent before reviewing records?
11. Will the consent process take place under conditions most likely to provide potential subjects an opportunity to make a decision about participation without undue pressure?
12. If the study is a clinical trial, how will the trial be monitored? What will be done with preliminary data? Should an independent data and safety monitoring board be established? How will decisions about stopping the trial be made? By whom? On what basis?
13. At what interval should the IRB perform continuing review of this project?
Federal Policy §___.101 [To what does this policy apply?]Federal Policy §___.101(b)(2) [Exemption for surveys, interviews, and observation of public behavior]
Federal Policy §___.101(b)(4) [Exemption for existing data, documents, records, pathological specimens, or diagnostic specimens]
Federal Policy §___.102 [Definitions]
Federal Policy §___.116(d) [General requirements for informed consent: Alteration or waiver of consent requirements]
Federal Policy §___.116(e) [General requirements for informed consent: No preemption of applicable state and local laws]
Federal Policy §___.117 [Documentation of informed consent]
45 CFR 46.401(b) [DHHS: Protection of human research subjects subpart D - additional protections for children involved as subjects in research.
21 CFR 312.7(a) [FDA: Promotion of investigational drugs]
21 CFR 314.111 [FDA: Statement concerning adequate and well-controlled clinical investigations
21 CFR 812.7(d) [FDA: Promotion of investigational devices]
Applicable state and local laws regarding release of information to persons other than a treating physician
Return to Index Page
Chapter IV: Considerations of Research Design