2 April 2014
The framework on the feedback of health related findings (HRFs) published by the MRC and Wellcome Trust on 31 March will be a welcome guide for researchers. The recommendations are pragmatic, seemingly deliberately non-prescriptive, sensible and measured. The reluctance to advocate a single approach, preferring each research project to be considered on a study-by-study basis, seems absolutely correct, given that this framework is intended to apply to all types of research. Unlike physicians who owe a duty of care to their patients in a clinical situation, the responsibilities owed by researchers are less clear. In part, this is because there is a lack of relevant case law clarifying the legal responsibilities. The framework acknowledges that these responsibilities cannot be stated definitively, and addresses the key ethical considerations, in particular, the balance of benefits and harms inherent in different approaches.
Since many research projects are utilizing technologies such as genomics or imaging that have greater potential to generate health related findings, it seems good practice to recommend that researchers consider these matters at the stage when they consider the design of their study, and that they should have available a policy on the feedback of HRFs which can be discussed with the research participant when consent is taken. Many other aspects of the framework are equally useful.
Nevertheless, from our perspective, when focusing upon genome-based research studies, there are three areas where the framework could have provided greater clarity and more nuanced discussion. The first is the failure to discuss adequately the concept of a HRF. The statement defining the concept is that it is a finding “that has potential health or reproductive importance”. In most areas of research, a finding such as an unanticipated radiological feature, or an abnormal result on biochemistry, will need to be investigated because of its potential clinical importance, even though after investigation the abnormality is shown to be a false positive, or benign and of little clinical significance. In genome based studies, the situation is much less clear cut. These findings, unlike those in radiology, do not signify the possible presence of disease, only the possibility of disease risk. There are also many variants that are of uncertain significance, and cannot without further investigation be deemed to be health related or otherwise. The guidance is silent on the question of the criteria by which such findings should be deemed to have potential health or reproductive importance.
A second area where there could have been greater elaboration, concerns the taxonomy of HRFs. The document only choses to distinguish between incidental findings and pertinent findings. Other sources of guidance have advocated a more nuanced (and prescriptive) taxonomy, such as the advice from the Presidential Commission report, Anticipate and Communicate, whose taxonomy distinguishes 1. primary, 2. incidental (anticipatable), 3. incidental (un-anticipatable), 4. secondary and 5. discovery findings. In our view the failure to distinguish incidental (to discover A but learns B) from secondary findings (to discover A but also actively seeks D per expert recommendation) is a potential omission. In genomic research, incidental findings are not truly incidental (as they are in radiology where one might see on the image something that one did not expect to see). Areas of the genome have to be explicitly and intentionally identified before analysis. If the intention is to search only in an area of the genome for a finding related to the research in question, then something else found in that area of the genome would in our view be truly incidental. However, if the research seeks to look at areas of the genome for some other reason, a region where there is no prior expectation that a variant connected with the research would be seen, then the finding is either a secondary finding or a discovery finding. Whilst the framework is clearly intended to be a generic document, applicable to all types of research, if would have been helpful if it had explicitly included a justification for rejecting more comprehensive taxonomies such as that advocated by the PCSBI report.
The third concerns the consent process and ‘opt-out’. In a research context, we believe that it is desirable to allow research participants to opt out of receiving HRFs, unless there are other public interest considerations, for example infectious disease outbreaks. However, it could be legitimate to mandate that all HRF are disclosed, provided that there is transparency at the outset that no opt-out will be provided. (This should be distinguished from the disclosure of HRFs in a clinical context in which we have previously argued that there should always be provision for opt-out. We are therefore very pleased to see that the ACMG have now acknowledged that an opt-out should be offered to patients who are candidates for clinical genome-scale sequencing). We therefore welcome the proposal that in studies where feedback will be provided, researchers should “clearly communicate to potential participants whether and how the conditions of the study allow them to opt out of receiving feedback.”
Despite having some concerns, which are made primarily from the perspective of genome-based research studies, we very much welcome the framework, the fundamentals of which we are in total agreement, and regard it as a valuable contribution to developing good practice in this area.