We describe the procedures and show some results of the "longitudinal mortality study of Costa Rican adults 1984-2007" (ELMAC), which consists of a sample of about 20,000 Costa Ricans aged 30 years or more from the 1984 population census and a follow up of their survival status to December 2007. The National Institute of Censuses and Statistics (INEC) digitized the names from the census questionnaire for the individuals in the sample. Computer programs cleaned up name's errors data and linked them to the civil registration system to have the unique id-number that all Costa Ricans have since their birth-cédula number. The linkage was possible for 87% of the sample. Those individuals linked to cédula number were followed up in the computer databases of the Civil Register to establish survival or to identify deaths. More than 5,000 deaths were found between the 1984 census and December 2007. Additional computer programs linked 92% of these deaths with those in the INEC's Vital Statistics, which are the only one containing information on the cause of death. The mortality pattern in the sample, which comprises 373,000 person-years of observation, replicated nicely the rates in preexisting life tables for Costa Rica. The SES gradients in mortality in the sample are not sensitive to the imputations and assumptions made in the linkage and follow up processes. The sample confirms the exceptionally low mortality of Costa Rican adults, especially males. This sample opens up the door for further analyses of the SES determinants of adult mortality in Costa Rica, something rarely done in a developing country.
Keywords: Adult mortality, mortality determinants, longitudinal analysis