What do international ethics guidelines say in terms of the scope of medical research ethics?

Abstract

Background

In research ethics, the most basic question would always be, “which is an ethical issue, which is not?” Interestingly, depending on which ethics guideline we consult, we may have various answers to this question. Though we already have several international ethics guidelines for biomedical research involving human participants, ironically, we do not have a harmonized document which tells us what these various guidelines say and shows us the areas of consensus (or lack thereof). In this manuscript, we attempted to do just that.

Methods

We extracted the imperatives from five internationally-known ethics guidelines and took note where the imperatives came from. In doing so, we gathered data on how many guidelines support a specific imperative.

Results

We found that there is no consensus on the majority of the imperatives and that in only 8.2 % of the imperatives were there at least moderate consensus (i.e., consensus of at least 3 of the 5 ethics guidelines). Of the 12 clusters (Basic Principles; Research Collaboration; Social Value; Scientific Validity; Participant Selection; Favorable Benefit/Risk Ratio; Independent Review; Informed Consent; Respect for Participants; Publication and Registration; Regulatory Sanctions; and Justified Research on the Vulnerable Population), Informed Consent has the highest level of consensus and Research Collaboration and Regulatory Sanctions have the least.

Conclusion

There was a lack of consensus in the majority of imperatives from the five internationally-known ethics guidelines. This may be partly explained by the differences among the guidelines in terms of their levels of specification as well as conceptual/ideological differences.

Background

When doing research ethics, the basic question is always, “which is an ethical issue and which is not?” It is interesting that though existing internationally-known ethics guidelines on human research somehow help, we cannot help but share Hussein and Upshur’s observations, “There are many structural similarities to these varied guidelines, but, more importantly, there is considerable variability and lack of harmonization across the globe” [1]. Up to the present, we know of no literature that attempts to look at the imperatives (i.e., the “oughts”) in the major internationally-known ethics guidelines to allow us some glimpse at what these “structural similarities” are and what areas are lacking in harmonization. The most commonly cited document is Emanuel, Wendler, and Grady’s “An ethical framework for biomedical ethics” [2]. Some sources mistakenly perceive this literature as a “coalescence of current standards in a single source” [3]. If we look at Emanuel, Wendler, and Grady’s article, there are clear indications that though the various guidelines were consulted, the aim of the article is to produce a “broader, systematic, and comprehensive framework” precisely because for them, existing guidelines are neither systematic nor comprehensive enough [2]. Hence, the Emanuel, Wendler, and Grady framework is just that, another framework, i.e., a proposed ethical structure or system that takes into consideration the various imperatives of the various guidelines. The eight principles, i.e., the fundamental truths, of their framework, “are conceptually included in most of the previously mentioned guidance, although existing guidelines do not necessarily include all of them” [2]. As such, there is still the need to comprehensively look into what exactly these international ethical guidelines say on what constitutes ethical issues in human research. Doing so would provide us a glimpse of areas that are widely similar, and hence areas where consensus is present among the international guidelines, and areas where dissimilarities are present, i.e., areas where consensus is lacking. Hence, we not only gain a comprehensive picture, we also get a glimpse of the work that needs to be done in reaching a consensus in research ethics.

Methods

When choosing the research ethics guidelines to include in this manuscript, we consulted the list of ethics guidelines discussed in section 2, “Codes, declarations, and other ethical guidance for research with humans,” of the Oxford Textbook of Clinical Research Ethics [4]. Documents that may provide some ethical guidance but are not necessarily ethical guidelines were excluded, such as the International Conference on Harmonisation Guideline for Good Clinical Practice [5]. This meant that we included five landmark internationally-known ethics guidelines on biomedical research namely, the Nuremberg Code [6], the Declaration of Helsinki [7], the CIOMS International Ethical Guidelines for Biomedical Research Involving Human Subjects (including the explanatory notes) [8], the Council of Europe’s Additional Protocol to the Convention on Human Rights and Biomedicine concerning Biomedical Research (including the explanatory notes) [9], and the Belmont Report. [10] From each guideline, we extracted the imperatives. To cluster these imperatives, we used the ethical framework for biomedical research of Emanuel, Wendler, and Grady. Their framework is made up of principles and benchmarks. Benchmarks refer to the imperatives they used as points of reference or measurement of the principles. We used their framework to either check if an imperative is already stated in their benchmarks, to add the imperative under one of their principles, or to add the imperative in a new cluster. It is important to note that since we used Emanuel et al.’s framework as a heuristic tool, we used their principles not as our assumed fundamental truths but as umbrella headings (which we shall call “clusters”) for the imperatives. We took the liberty to add new clusters as necessary and to minimally rephrase a principle to allow for wider coverage. For example, instead of “collaborative partnerships” we used “research collaboration” and instead of “fair participant selection,” we used “participant selection.” Apart from categorizing the imperatives, we also took note where the imperatives came from, and thus we also gathered the data on how many guidelines support a specific imperative, i.e., if there is a strong consensus among the guidelines (SC; 5/5, 4/5), moderate consensus (MC; 3/5), weak consensus (WC; 2/5), no consensus (NC; 1/5), or if it is a non-guideline-based imperative (NGI; 0/5 [in the case of benchmarks in the Emanuel et al. framework that do not correspond to any of the imperatives from any of the guidelines]).

Results

In total, we extracted 560 imperatives. We included the imperatives for vulnerable population in general, for persons unable to consent, and for emergency research. After double checking for redundancies and merging similar imperatives, the imperatives were reduced to 386. In total, we have 12 clusters: basic principles¹; research collaboration; social value; scientific validity; fair participant selection; favorable benefit/risk ratio; independent review; informed consent; respect for participants; publication and registration; regulatory sanctions; and justified research on the vulnerable population. Table 1 shows the imperatives within these clusters and their consensus levels. The latter three clusters were added from the original eight principles of Emanuel, Wendler, and Grady. In the following, we describe the clusters and the imperatives. Note that in a few instances, we reclustered some of the Emanuel et al. benchmarks. Additional file 1 gives an overview of the reclustered benchmarks of Emanuel et al.

Table 1 Ethical imperatives and their levels of consensus (Strong Consensus [SC] 4 or 5/5; Moderate Consensus [MC] 3/5; Weak Consensus [WC] 2/5; No Consensus [NC] 1/5; Non Guideline-based Imperative [NGI])

Of the 386 imperatives, we see in Table 2 that there is the predominance of the lack of consensus at 72.8 %. SC and MC imperatives comprise 8.2 % of all the imperatives. Among all the clusters, Informed Consent has the most number of SC and MC imperatives put together, comprising 14 of the 32 (43.8 %). The following clusters have no SC or MC imperatives: the basic principles; research collaboration; social value; and publication and registration; and regulatory sanctions.

Table 2 Summary of clusters and the level of consensus per cluster

Clusters with at least MC

Table 3 provides a snapshot of the clusters and the corresponding 11 imperatives with SC. In detail, the clusters with at least MC are the following:

• 1. Scientific validity

  The imperatives may be subclustered as follows: scientific design, protocol, professionalism, use of comparator, issues with placebo, reexamination.

  There is SC that scientific validity entails that the scientific design and the statistical methods of a study must satisfy generally accepted standards and that using these methods and design, the research objectives would likely be met. It also entails that the study must be justified by previous studies and current knowledge (SC); and that the research is conducted only by scientifically qualified persons in terms of both skill and care for the participants (MC). Lastly, in terms of the acceptability of the use of placebo, there is MC that it may be used when no proven intervention exists.

• 2. Participant selection

  There is MC that the research population must be selected to ensure the study’s compliance with scientific norms and the generation of valid and reliable data.

• 3. Favorable benefit/risk ratio

  The imperatives may be subclustered as follows: risks, benefits, and assessment.

  That risks must be justified in the sense that the risks to participants are outweighed by the anticipated benefits to them and the anticipated benefits to society is a SC imperative that acts as the overarching principle of this cluster. This may mean several things such as the following: there is SC that risks and burdens must be minimized. Also, to ensure proper assessment, there is MC that the various risks (whether physical, psychological, social or legal) must be delineated and the probability and magnitude of such risks must be quantified as much as possible.

• 4. Independent review

  By independent review, we refer to the ethical assessment by research ethics committees (RECs; otherwise known as institutional review boards). This cluster may be divided into five subclusters: REC composition and requirements, rights and responsibilities of RECs, responsibilities of the investigator/sponsor, externally sponsored studies, and when deception/incomplete disclosure is involved.

  There is MC that an REC must be independent and competent and that it must examine scientific merit by looking at study design and safety provisions. There is also MC that every interventional study must be submitted by the sponsor/investigator to a REC for evaluation and approval.

• 5. Informed consent

  There is a wide range of consensus levels in this cluster, the highest being SC. This cluster includes imperatives that relate to an ethical informed consent form and procedure. These imperatives may be subclustered into the following: culture-appropriate consent, securing consent (information, comprehension and voluntariness), renewing consent, withdrawal of consent, use of data/specimen, waiver of IC, and supplementary consent.

  There is SC that the disclosure forms and procedures must be culturally sensitive. This relates to an MC that consent must preferably be secured via writing, i.e., through a signature. Otherwise, consent must at least be formally documented and witnessed. In terms of what the participant must be informed about, there is SC that the participant must be informed about the anticipated benefits of the study and the right to withdraw without reprisal. On the other hand, there is MC that the participant must be informed of the aims; methods/procedures; sources of funding; researchers and their institutional affiliations; all risks and discomforts that a reasonable person would consider material; any current alternative interventions; and that the subjects must at least be given the option to be informed of the findings of the research. In terms of comprehension and voluntariness, there is SC that the investigator must ensure that the participants can in fact comprehend the information that is being provided; and that the participants’ consent is truly voluntary.

• 6. Respect for participants in general

  The imperatives may be subclustered as follows: participant safety, privacy/confidentiality, participant care, dissemination of research results to participants/community, compensation, physicians and participants, sociological/epidemiological studies.

  There is MC that the health of the participants must be monitored for purposes of minimizing harm. When injuries are incurred due to research participation, according to a MC imperative, participants have the right to receive free and appropriate medical treatment. Related to this is another MC imperative which states that participants who have been harmed due to research participation must be appropriately compensated and in case of death, the compensation must go to the dependents of the participant.

• 7. Respect for vulnerable populations

  The imperatives may be subdivided into the following subclusters: vulnerable population in general, persons not able to consent (including minors), third party authorization, emergency research, and pregnant women.

  Research on persons who are not able to consent may be ethically justifiable when the following conditions are met: the authorization is provided in writing by the legal representative who ideally takes into account the participant’s prior wishes and objections (SC); assent is secured and dissent respected (SC); and research of comparable effectiveness cannot be carried out on individuals who are capable of giving consent (MC).

  The following must be in place for the involvement of vulnerable individuals in research to be justified: safeguards must be in place to protect the vulnerable from manipulation or from being involved in a study due solely for administrative convenience (MC).

Table 3 Clusters and imperatives with strong consensus

Clusters with NC

After looking at the imperatives with a relatively good level of consensus, it is also worthwhile to look at the other end of the spectrum, i.e., clusters with at most NC. These are Research Collaboration and Regulatory Sanctions. The imperatives of these clusters are as follows:

Research collaboration

Research must respect the values, circumstances, culture, and social practices of the community. This can be expressed in various ways such as the following: the involvement of community representatives in the planning and conducting research, as well as in the dissemination of results. Also, the details of the health care services information that is distributed to the participants must be the result of an agreement between the sponsor, the host country officials, and when appropriate, the community involved in the research. Discussions on the responsiveness of a research must include stakeholder representatives from the host country. Capacity building plans in the host country must be the result of a bilateral dialogue between the host country and the external sponsor. Lastly, authorship and intellectual property rights must be fairly distributed.

Regulatory sanctions

The preferred methods for regulatory action must be those that cultivate an atmosphere of mutual trust, education, and support. As such, disciplinary sanctions must be used as the last regulatory resort. Also, drug regulatory authorities should consider refusing to accept unethically obtained data, but not without considering the effect of such a refusal.

Discussion

In the introduction, we stated that we wanted a comprehensive list of imperatives to determine consensus areas and areas where consensus is lacking. In the results section and in Table 1, as a response to the former, we extensively presented the imperatives according to clusters and subclusters. We also showed the level of consensus present in each cluster. We are now able to determine the areas where consensus is high or at least moderate, and hence the areas of similarities between the guidelines, and the areas where consensus is lacking. From Table 2, we see that of the 386 imperatives, only 32 are at least with MC and only 11 have SC. This means that currently, based on at least MC imperatives, consensus is at 8.2 %; and, based on SC imperatives, consensus is at 2.8 %.

We also saw that among the imperatives with at least MC consensus, 14 of the 32 come from the Informed Consent cluster. This is also the cluster with the most SC. This may be an indication that there is most agreement within research ethics on informed consent processes and procedures.

At the other end are the clusters Research Collaboration and Regulatory Sanctions where, at most, NC is present. This means not only that there is least agreement within research ethics on these clusters; it also means that since for each entry, only one guideline provides a contribution, these clusters are also the most neglected.

These findings are important for several reasons: first, having a comprehensive picture of the imperatives from the five international ethics guidelines on researches involving humans provides guidance on the question, “which are ethical issues?” In the field of pharmaceutical clinical trial regulation, for example, drug assessors frequently rely on the general statement in the dossier that the clinical trial has been conducted ethically. Given that there is the current trend of increasingly expecting drug regulators to identify and act upon ethical issues [11], having a comprehensive list such as this would help in identifying and categorizing clinical trial issues as ethical in nature. Second, our findings on the level of consensus provide an idea of the magnitude of work to be done in research ethics in terms of international harmonization and agreement. Third, the findings on the clusters where consensus is most present and those where there is no consensus provide some information of which areas are promising in terms of international harmonization and consensus and which areas need more work either because there is no agreement present or because there is relatively little attention on them.

It is important to note that the limited amount of consensus may have been brought about by the fact that some ethics guidelines are more expounded than others. The CIOMS guideline, for example, provides explanatory notes, while the Helsinki Declaration does not. The level of consensus depends on what is written down; however, it may be reasonable to think that some of the low consensus imperatives may in fact be thought of as specifications of the principles stated in shorter guidelines, i.e., that the amount of consensus may increase if only we see some of the low consensus imperatives as specifications of the more general imperatives. There were also differences in terms of the original intended audience of a guideline. The Belmont Report, for example, was originally meant for scientists, members of IRBs, and Federal employees in the USA [10] while the Declaration of Helsinki is addressed primarily to physicians [7]. The latter is the reason why there is NC, for example, on the imperative, “Physician involved in research must protect the life, health, dignity, integrity, right to self-determination, privacy, and confidentiality of personal info of research subject“.

At the same time, actual disagreements among the guidelines may also partly explain the limited level of consensus. For our purposes, we shall demonstrate using imperatives from Table 1 that weakness in consensus is in fact reflective of the theoretical differences we see in the literature.

First, we see that on the level of principles, there is at most WC. There is WC that “the interests and welfare of the human being participating in research shall prevail over the sole interest of society or science” and there is also WC that principles such as beneficence and justice must be upheld. This means that, two guidelines subscribe to the former principle, and two other guidelines subscribe to the latter principle. After back checking for sources, suffice it to say that there is no overlap between guidelines on these two camps. The WC is in fact reflective of the wider disagreement within biomedical ethics of what justification entails, i.e., must it be a justification from a moral theory of human rights (i.e., the top-down approach) or must it come from a balancing of basic principles (such as the four principles of autonomy, beneficence, justice, and nonmaleficence of Beauchamp and Childress) [12]?

Second, within the cluster Social Value, it is noticeable that both “fair benefits” (i.e., fair distribution of benefits must be determined by the parties involved on a case-by-case basis [13]) and reasonable availability” (i.e., a prior agreement between the trial sponsor and host that the medical product will be made ‘reasonably available’ to the host country or community once the product has been approved [14]) are present. These two concepts refer to the two differing proposals in the literature on how to make research socially valuable to the host country or community [14, 15]. Again, this conceptual or ideological difference explains the lack of consensus, with fair benefits being a NGI, and reasonable available as a NC imperative.

Third, the use of the term “therapeutic research” by imperatives within Table 1 at most garners NC. This is reflective of the discussions in the literature on the acceptability of conflating therapy with research [16–19], i.e., the lack of consensus on any of the imperatives that are hinged on this term is reflective of the disagreement within the literature on the acceptability of therapeutic orientation in research.

The stated differences expounded above in terms of level of detail, intended audience, and actual conceptual/ideological differences among the guidelines are also unavoidable limitations of this manuscript.

Conclusion

In this manuscript, we have provided a comprehensive view of what the various international ethics guidelines for research involving human participants say. We also identified the level of consensus among the imperatives from these guidelines and we discovered that in the majority of the imperatives, there is no consensus. Of the 12 clusters, Informed Consent has the highest level of consensus and Research Collaboration and Regulatory Sanctions have the least. This somehow provides a preview of which areas need more work in terms of achieving international harmonization. The lack of consensus may at least be partly explained by the differences among the guidelines in terms of their levels of specification as well as conceptual/ideological differences on which the imperatives are hinged on.