How representative is a sample of relatives of deceased people recruited through physicians?
Muller, MT, VU medical center, Amsterdam, Netherlands, Rurup, ML, VU medical center, Amsterdam, Netherlands, Georges, JJ, VU medical center, Amsterdam, Netherlands, van der Heide, A, Erasmus MC, Rotterdam, Netherlands, Onwuteaka-Philipsen, BD, VU medical center, Amsterdam, Netherlands

Objective: Ethics committees require that recruitment of relatives of deceased people takes place by the doctor that treated the patient. Question is whether there are factors influencing the willingness of doctors to approach relatives to participate in research on end-of-life care.
Methods: A random sample of doctors (n=167) who had reported a case of euthanasia or assisted suicide (EAS) in the period November 1999 and January 2002 to one of the regional review committees (RRC=committees that judge if a doctor who performed EAS have acted in accordance with the requirements for prudent practice) were interviewed about the reported case and were asked to contact the relative most involved to ask him/her if he/she was willing to give an interview to one of the researchers. Eight doctors refused to contact the relatives, since they did not want to burden the relatives, 16 doctors could not reach the relatives, and 46 did not approach the relatives without giving a specific reason. 10 relatives were approached but refused to participate. Results of the doctor interview were compared for cases in which a relative was (n=87) or was not (n=80) interviewed.
Results: No significant differences were found between cases in which the doctor did or did not recruit the relatives for the following characteristics of the patient: sex, age, being depressed, reasons for requesting EAS and the extent of unbearable and hopeless suffering. There were also no differences in doctor characteristics: specialty, feeling pressured into granting the request and experiencing the review of the case of EAS as incriminating or burdensome.
Conclusion: Although recruitment through doctors did not lead to a high participation rate among relatives, there are no indications for selection bias.