Mortality effects of average education in current and earlier municipality of residence among internal migrants, net of their own education

Many researchers have shown an association between a person's health or mortality and the socio-economic resources in the community, net of the person's own resources. The focus has typically been on the current community of residence or one where the person lived a few years earlier. Only a few authors have tried to establish whether there is an additional effect of earlier community exposures, and they have not made a distinction between migrants and non-migrants, which there are good reasons to do. The results from these earlier investigations have been rather mixed. In the present study, a discrete-time hazard model for mortality between ages of 60-89 over the period 1991-2002 was estimated from register data that included the entire Norwegian population. For each person, and for each of these years, municipalities of residence during the previous 20 years were known. There was also information on each person's educational level. For the relevant years and each of the 433 municipalities, measures of average education in the population were constructed by aggregating over the individual data. The analysis was focused on the subgroup who had moved across a municipality border only once during the previous 20 years, among whom there were 29,843 deaths during 839,113 person-years of follow-up. The population sizes of the municipalities were included as control variables. Among those who had moved within the last 10 years, the current socio-economic context was not important for mortality. The earlier context had an effect, but this was restricted to men. Those who had lived longer in the current municipality of residence were influenced only by the average education in that municipality. The findings support the idea that neighbourhood socio-economic effects need some time to build up, and that they do not dissipate soon after the person has moved to a new environment.