We thank Clarke, Robertson and Kerruish, Elger, Lucassen and Fenwick for their commentaries on our recent publication in J Med Ethics, ‘Predictive genetic testing in minors for late-onset conditions: a chronological and analytical review of the ethical arguments’.1^–5

Elger, Robertson and Kerruish generally support the stance we have proposed in our paper, that being to seek empirical evidence to support or refute the ethical arguments for or against predictive testing of minors.2 ,3 Clarke, Lucassen and Fenwick, by contrast, mount a number of arguments against this approach and it is these commentaries that we wish to respond to in detail.

In his commentary, Clarke raises concerns about our assertion that there are questions in relation to predictive testing of minors that may be answered through empirical research.1 Clarke's concerns include the fact that a very long follow-up period is required to study predictive testing in young children, that some tested individuals and their families will be lost to follow-up and that the study would be biased as only some clinicians would enrol their patients into such a study. Clarke suggests instead that clinicians should report the outcomes of testing of minors they have been involved in. He refers to a situation where a pregnancy has been tested for Huntington disease and the fetus is found to have the causative mutation but the parents elect to continue the pregnancy as an instance of the type of situation where outcomes could be reported.1

We acknowledge the challenges raised by Clarke but are firmly of the view that a systematic study is preferable to pooled clinical experience, although clinical reports are helpful and should be encouraged. We would like to clarify some specific details about the type of research we are promoting. First, we suggest that the first group to study are mature minors and not immature minors. Only if testing in this initial group was shown to have similar or better outcomes to predictive testing in adults would we advocate a systematic study of predictive testing in immature minors. We have in the past provided a detailed outline of how such research could proceed.6 Such a study would require far shorter follow-up, and issues of consent would be less fraught than studying the outcome of predictive testing in immature minors. Key outcomes to assess within such a study would include the experiences of clinicians and parents, as Clarke suggests, and primarily the experiences of young people tested both at the time of testing and also as they move into adulthood. Second, we are not advocating that in undertaking research, clinicians would provide testing in circumstances that they deem inadvisable. Rather, we are suggesting that when clinicians do believe testing is appropriate, and thus testing is provided to mature minors, they have an opportunity to invite young people and their families into a longitudinal, international study.

We believe that the opportunistic data reporting suggested by Clarke, and supported by Lucassen and Fenwick's reference to two case reports, is very unlikely to enable substantive conclusions about the appropriateness or otherwise of predictive testing in minors.1 ,4 First, we have investigated the outcomes of predictive testing in minors through an international survey of clinical geneticists and found that there was very limited follow-up of those tested.7 Therefore, we believe that lack of information about outcomes would be a problem. In addition, the issue of clinician bias (in relation to who is followed up and what outcomes are noted) is likely to be far greater in the proposed solution of Clarke than in a systematic study. We think far fewer clinicians would undertake predictive testing in minors with formal follow-up in routine practice than if it were to be part of a formal research study that has the approval of an ethics committee. In addition, negative outcomes may be under-reported by clinicians who fear professional criticism for having performed a test that is strongly discouraged in clinical guidelines.

Clarke questions the feasibility of the long term follow-up required to examine outcomes of predictive testing in minors.1 There are many examples of important research findings that have required decades of follow-up of children into adulthood. Examples include the 1958 British Cohort study8 and the US National Collaborative Perinatal Project.9 This difficulty is not something that of itself precludes this research.

Lucassen and Fenwick rightly questions whether empirical evidence will answer all questions about individual requests for predictive testing in minors.4 We agree that this will not be the case. Nevertheless, some of the ethical concerns that have been raised many times can be studied and the outcomes of such research can assist clinicians when making decisions in individual cases. This is simply how evidence-based medicine works. The most recent study of practice in this area indicates that many more requests are refused than acceded to, and that many refusals are based on the guidelines rather than on any consideration of empirical evidence and the circumstances of the individual case.7

Finally, in relation to Lucassen and Fenwick's commentary, we believe the analogy of the debate about abortion is a poor one. Unlike the abortion debate which is dominated by empirically untestable value claims as to the moral status of embryo and fetus, many of the major concerns about predictive testing of minors can be tested by systematic research.

As Clarke, Lucassen and Fenwick note, professional reflection upon cases will allow us to improve practice. However, it is our belief that such professional reflection is at a high risk of being biased if it is not supported by rigorously collected evidence about long term outcomes of predictive testing of minors in terms of psychological well-being and perceived autonomy of those tested.