To the Editor: I note with interest the case study of experimental Ancylostoma caninum infection in a 22-year-old student.1 In his accompanying editorial, Van Der Weyden highlights the courage of these researchers, as well as some of the risks and discomforts associated with their participation,2 including in two studies in which he was a co-author.

However, it is also worth highlighting some of the ethical issues associated with such experimentation. Larry Altman, who provided many of the quoted examples of self-experimentation, also reflects on Walter Reed's experiments with yellow fever vectors in Cuba. Although later credited with the use of written consent forms, on an earlier occasion Altman alleges that Reed withdrew at the last moment from inoculation experiments in which one of his colleagues died.3

Although not mentioned in the published work, the A. caninum experiment was initiated and undertaken by Landmann under the supervision of Prociv, who has himself self-experimented with both A. caninum and Necator americanus (human hookworm) in previous work (Juergen Landmann, Student; Paul Prociv, Senior Lecturer, Department of Microbiology and Parasitology, University of Queensland, personal communication). In this case, the study was wholly initiated by the student so consent was not an issue, but other such studies raise the potential problem of consent in situations of an unequal power relationship.

Students under supervision constitute a "vulnerable" group in that consent may be given under a form of duress.4 Just as special protection is needed for populations for whom research is combined with care, protection is required for students who may feel obliged to participate in such research.

There have been recent calls for a fuller discussion of ethical issues in published experimental studies,5 where ethical justification should be accorded the same weight as statistical considerations. The unusual study by Landmann and Prociv highlights the need for such discussion in potentially controversial research protocols.