RESUMEN
PIP: The method of surviving sisters for indirectly estimating maternal mortality is still under development but shows promise for countries lacking alternative sources of data and good statistics. This work uses census or survey data to apply the method to rural villages in Gambia; Mapuche settlements in Cautin, Chile; marginal populations on the outskirts of Lima, Peru; and rural villages of Avaroa, Bolivia. The method is explained in detail following presentation of the results. The necessary basic information is outlined, and the particularities of its application to each Latin American case are discussed. The surviving sisters method was developed by Graham and Brass to derive indicators of maternal mortality based on the proportion of sisters who arrive at fertile age and die during pregnancy, delivery, or the postpartum period. The method transforms the proportions of sisters who died of maternal causes obtained from a census or survey into conventional probabilities of death. The basic information required concerns the numbers of sisters entering the reproductive period (excluding the respondent is she is a woman), the number surviving and decreased at the survey data, and the number who died during pregnancy, delivery, or the postpartum period. The probabilities of dying from a maternal cause were estimated on the basis of the sister survival method at 1/98 in Lima, 1/53 in Cautin, 1/17 in Gambia, and 1/10 in Bolivia. These probabilities correspond to ratios of maternal mortality per 100,000 live births of 286 in Lima, 414 in Cautin, 1005 in Gambia, and 1379 in Bolivia. The results demonstrate great variability in maternal mortality rates. In the cases of Lima and Cautin there were significant differences between estimates derived from the sister survival method and those derived from vital statistics. The 4 cases demonstrated the familiar association between maternal and infant mortality, fertility, and overall female mortality expressed in life expectancy at birth. The more detailed presentation of the 3 Latin American cases stresses the need to formulate survey questions that will enable the population at risk to be correctly identified in each case. A minimum of 3 questions is usually required, and their formulation may vary depending in part on the importance of extramarital fertility.^ieng