I'm supposed to analyse the effect of a governmental health care program on mortality (due to cancer ,cardiovascular disease or any other cause of death). Health care program started in 1975 and is still going on. Citizens are invited regularly (every 2 years) to use this health care program. To evaluate the effect of this intervention we've repeatedly (1986, 1991 and 1998) chosen a randomly selected sample of 2400 citizens. I know that I'm dealing with a competing risks problem but I'm not sure about how to deal with repeated measurements as the research question is "does the health care program have an effect on mortality ?". I've compared the group of people who used the health care program at least once with the ones who never used it but in that way overestimated the effect as we have to assume that people using the program have in general a healthier lifestyle (and therefore lower mortality) as the ones not using health care. Another problem is left-censoring: the program started in 1975 but the first records are from 1985.