• ethics committees/consultation
• policy guidelines/inst. review boards/review cttes.
• public policy
• regulation
• research ethics

What job should authorities give to review boards? We are grateful to Soren Holm, Rosamond Rhodes, Julian Savulescu and G Owen Schaefer for their thoughtful commentaries on our answer.1–4 Here we add to the discussion.

Let us summarise the claims for which we argued.5 Relevant authorities can task boards with review for consistency with duly established code (‘code-consistency review’ or ‘CCR’), thereby making code-consistent activities apt for approval and code-inconsistent activities apt for rejection. They can instead task boards with review for ethical acceptability (‘ethics-consistency review’ or ‘ECR’), making ethically acceptable activities apt for approval and ethically unacceptable activities apt for rejection. For every proposal a board might consider, these two different jobs establish different review bases, and their approvals and rejections also sometimes conflict. Some international and national statements require ECR, others instead require CCR, and others again seem either to require both or just to run the two together. Those responsible for these statements should make them clearer and better aligned here. For reasons of practicality, publicity and separation of powers, authorities do better to task boards with CCR and not ECR. These arguments also count against establishing any code with content that in effect collapses CCR into ECR. If our arguments withstand robust scrutiny, authorities should also remove ‘ethics’ and cognate terms from the names of these boards and their review activities and emphasise code expertise not ethics expertise in the required skill sets of boards.

Our article noted that ‘ethical considerations informed the genesis of these boards and express their aspirations’. Rhodes similarly notes: ‘the authors and endorsers of research ethics codes, declarations and regulations … (are) articulating the ethical standards for conducting human subject research’ and ‘This framework embeds the ethical parameters of human subject research into the moral missions of institutions’. Schaefer too writes: ‘the whole point of research ethics codes … is to ensure that research is conducted ethically’. As we observed, design of many institutions is grounded in fundamentally ethical factors. To add a further example, the basic ethical desideratum that justice be done arguably underpins the design of judicial institutions. It is still a further question which job these institutions should be assigned to serve best their ethical mission. The wide-scope point here is well recognised in many fields. If authorities make the ethical principles for the design of a basic institutional structure, simply flow through into the principles that a particular institution within that structure is tasked with applying case by case, those authorities often fail to do what is best. These two sets of principles might come apart because that is what the best division of the needed ethical labour requires. They might also or instead come apart because pursuit of the main institution-specific ethical consideration (eg, that research be ethically acceptable) must itself be constrained by wider ethical requirements, such as treating with due respect and fair process those who make application to those institutions (eg, by adequately addressing publicity and separation of powers desiderata).

Roughly speaking, the core CCR task is to approve code-consistent activities and reject code-inconsistent activities. To apply a typical research review code well, a CCR board needs to interpret and apply the considerations of harm, benefit, consent, vulnerability, equity, exploitation, rights and so on, that it specifies. This requires from it sensitive and nuanced attention to study context and to many particulars of study protocol, subject population and subject experience. It might need to exercise multidisciplinary deliberation to get itself sufficiently informed on such matters. Some uncertainty, incompleteness, indeterminacy and conflict, within even the best-informed state it can achieve, is the everyday reality of any CCR board and of administrative bodies in general. As we noted, administrative law addresses such situations through many meta-principles that give administrative bodies code-based paths of resolution. Through appeal to these, and to purposive code provisions, administrative bodies can often generate more than one reasonable code basis for proposal acceptance or rejection. As we explicitly noted, CCR can ‘allow that two or more divergent but reasonable decisions are possible in some cases’. Ethical thought and judgement may be exercised within CCR, never in a free-range manner but always instead in the service of best expressing and applying the applicable code. Sensitivity, discretion, interdisciplinary deliberation and ethical thought cannot here legitimately contribute to approval of any proposal that is code inconsistent but ethically acceptable or rejection of any proposal that is code consistent but ethically unacceptable. Unlike ECR, CCR is code driven and code constrained.

Most of the commentators attribute to CCR features that clash with the above account. Schaefer attributes to it ‘blanket rules’ that are ‘too insensitive to contextual considerations’, that are ‘without the intervention of ethical judgements’ and that cannot account for ‘nuance and contextual variations’. He also takes CCR to imply that ‘discretion given to ethics committees … would have to be stamped out’. Holm attributes to CCR ‘mechanical application of rules’. Rhodes implies that CCR bodies cannot be ‘interdisciplinary deliberative bodies’. In general, the commentators say little to justify these attributions. Perhaps they think something unique to review of research does so, or perhaps instead something widely shared by the CCR review of administrative tribunals, licensing boards and other kindred bodies.

The core ECR task is to approve ethics-consistent activities and reject ethics-inconsistent activities. Its approvals and rejections collide with those of CCR only when proposed activity is code consistent but ethically unacceptable or code inconsistent but ethically acceptable. ECR must then pursue ethical improvement at the expense of code-consistency. Here Holm attributes to us the flaw of ‘trading on a tendentious account of ethics committee review’ through an ‘implausible idealising premise’ treating ethics committees like ‘philosophy seminars … to shape research projects so that they are ethically optimal’. Our view is instead just that ECR needs to be empowered to breach code consistency. Holm actually agrees: ‘What it must be empowered to do is to make a non-code-compliant decision if the research would not be ethically acceptable otherwise’.

On grounds of practicality, publicity and separation of powers, our article argued that authorities do better to task boards with code consistency and not ethics-consistency review. As we read Holm, Savulescu and Rhodes, they do not directly contest these arguments. Holm does seem to reject our conclusion. In some places, Savulescu seems to accept and add to our conclusion, while in others he argues for suprainstitutional specialist ethics-consistency review committees.

Rhodes seems to reject CCR, yet also to be committed to it. She approvingly notes that in the USA ‘a federal agency assigns to institutions the moral and juridical responsibility for assuring that research conducted by its employees conforms to federal regulations’. If an institution’s review board approves or rejects proposed research on this federally assigned regulatory basis, it does CCR. If its approvals and rejections are instead based on ethics consistency, it fails to do its federally given job and instead does a different job. Rhodes also agrees with us that these two jobs can collide: ‘regulatory requirements and ethical requirements can conflict’.

Schaefer reads our article as a valuable challenge to conventional wisdom. He does not contest our arguments for superiority of CCR over ECR. Instead he argues that authorities should task boards with a third sort of review. He writes ‘when they conflict, codes should take primacy over committees’ ethical judgements’. But this is just CCR. Schaefer acknowledges that it implies rejection of ethically acceptable but code-inconsistent proposals, but he does not note that it also implies acceptance of code-consistent but ethically unacceptable proposals. Importantly, Schaefer also requires that codes ‘delineate when a protocol must be rejected, not when it must be approved’. Schaefer-style codes thus leave an approvals vacuum, which he fills with committee ethical judgements. Given a perfect Schaefer-style code, no code consistent but ethically unacceptable proposal can ever arise, so here the code’s formal primacy never has practical effect. Schaefer’s fundamental CCR thereby generates a novel hybrid: reject code-inconsistent proposals, and accept ethically acceptable proposals. This is ingenious. Schaefer is right that we did not consider it. We do so now.

Schaefer argues that his sort of review ‘does not fall prey’ to our critiques and also ‘offers significant advantages over the code-consistency approach’ because it ‘does not require major revisions’. We disagree. Schaefer-style reform still requires search and removal for every code statement that can imply proposal approval. This might require adding a code provision that states it is an exception to the presumption of freedom under which code failure to rule out an activity is sufficient to allow it. A second point is that Schaefer-style reform of proposal rejection is exactly as revisionary as code-consistency reform is, because on that issue, it just is CCR. A third point is that Schaefer’s hybrid does ECR for proposal approval, and consequently has all of its problems of practicality, publicity and separation of powers. It also does CCR for proposal rejection and consequently shares any troubles this might have. A final point, returning to the opening theme, is that the core asymmetry within Schaefer-style review is troubling. A Schaefer-style code specifies when it will disallow its applicants to do as they propose, but never specifies when it will allow them to do as they propose. Though Schaefer might be right that this requires research to meet a ‘higher ethical standard’, it does so by requiring boards to treat their applicants by a lower ethical standard.

We close with one further connection between codes and ethics. A CCR board should generally do its assigned job of approving code-consistent proposals and rejecting code-inconsistent proposals. But if faced with a code-consistent proposal that no board could reasonably regard as ethically acceptable, it should pursue certain resolutions other than approving this unconscionable proposal. Relatedly, if faced with a code-inconsistent proposal that no board could reasonably regard as ethically unacceptable, it should pursue certain resolutions other than proposal rejection. This clarifies our article’s account of unconscionability, highlights a further difference between CCR and Schaefer-style review and further places CCR in its larger ethical context. Our wider hope is that code-consistency review will attract further discussion.