Childhood leukaemia and population movements in France, 1990-2003.

About 20 years ago, a cluster of cases of childhood leukaemia (CL) in the village of Seascale (England) revived interest in an infective basis by leading to the population mixing hypothesis (Kinlen, 1988). This proposed that large population influxes into rural or isolated areas are conducive to an epidemic of the underlying infection(s) of which CL is postulated to be a rare consequence. Childhood leukaemia mortality or incidence has been examined in rural areas after receiving such influxes (Kinlen, 1988, 2006; Kinlen et al, 1990, 1993, 2002; Kinlen and Hudson, 1991; Langford, 1991; Kinlen and John, 1994; Kinlen and Petridou, 1995; Koushik et al, 2001; Boutou et al, 2002). These studies, mainly focusing on historically documented and specific rural population increases (with consequent increases in population density), showed significant increased relative risks, in the range of 1.5–4.7, in the places with the highest population increases, compared to the reference group. Elsewhere, population mixing has mainly been defined not as population increases but as increases in the proportion of residents who changed address over a specified period, or in the year before a census, and not specifically in rural areas (Stiller and Boyle, 1996; Dickinson et al, 2002; Parslow et al, 2002; Law et al, 2003; Labar et al, 2004; Nyari et al, 2006; Rudant et al, 2006). Overall, the studies seem to favour a positive association between the incidence of, or mortality associated with, CL and the highest proportions of migrants, although a few have reported a negative association (Parslow et al, 2002; Law et al, 2003). A few of these also found that the further the migrants came, the higher was the incidence of CL (Stiller and Boyle, 1996; Dickinson et al, 2002; Rudant et al, 2006). In the past, a positive association was found with childhood acute lymphoblastic leukaemia (ALL), particularly in isolated areas with a population density greater than a given threshold at the time of birth. In the present study, we investigated CL incidence on a national scale in relation to the proportion of individuals who changed address over a 14-year period, focusing on the isolation status of residence at diagnosis.