Criticism of study of childhood leukaemia near French nuclear reprocessing plant is unfounded

Editor–Two weeks ago Dominique Pobel and I reported our case-control study of leukaemia among young people near La Hague reprocessing plant.1 Dr Jacqueline Clavel, of Unit 170 of Inserm, which is the French national institute of health and medical research, has been quoted in the French lay press as criticising our results.2 3 She argued that the geographical location of the homes of subjects who participated in the study could explain the significant results (use of local beaches and residence in a house built of granite materials) by acting as a confounder.

We were obviously aware that living in the immediate vicinity of one of the beaches might increase the likelihood of a child visiting a beach to play, but, as we stated in our discussion, controls were matched on the general practitioner's catchment area. In Dr Clavel's opinion, however, this was insufficient. She postulated that the higher proportion of cases reported to have used the local beaches simply reflected the geographical distribution of their homes. Furthermore, she said that houses built of granite materials were not unlikely to be located on the coastline and therefore reflected proximity to the sea. Unfortunately, she did not provide any figures to support these statements, which contradict what is known about local traditional dwellings: that they are mainly located in the country and not on the coast, to avoid the effects of wind.

To address this issue I have estimated three distances as the crow flies: the distance between the home of each case and control and the closest coast (in whatever direction), and the distance between the home and the two most popular beaches in this area (Vauville and Urville-Nacqueville). Vauville beach is located near the pipeline used by the nuclear reprocessing plant to release liquid nuclear effluents, and Urville-Nacqueville beach is situated not far from the mouths of the rivers coming from the reprocessing plant and the low level radioactive waste depository. There was no significant difference between the groups of cases and controls in each of the three distances (P>0.36, Mann-Whitney test; table 1).

The potential geographical bias suggested by Dr Clavel can therefore be dismissed and does not affect our study's results. I hope that this additional information will clarify the debate surrounding the conclusions of our study.