Community-level determinants of infant and child mortality in peru

Abstract

A technique is proposed for calculating community-level sex-specific infant and child mortality on the basis of the household data collected by the World Fertiligy Survey. These estimates then serve as dependent variables for a multivariate analysis of 84 Peruvian communities of less than 25 000 population. This analysis is guided by a quasi-theoretical strategy that puts three classes of variables in competition: physical ecology, program interventions, and social structure. The representative of the first category, altitude, was significantly associated with male and female child mortality when the other independent variables were controlled. However this result is probably better interpreted as an indirect effect of social organization in the mountainous areas. The representative variable of the second category, number of local medical institutions, was unrelated to any of the four dependent variables. All three of the indicators of the social organization-community population size, proportion of educated women, and proportion speaking Spanish-were negatively correlated with the dependent variables as expected, but in the multivariate analysis only female education continued to be a consistent negative predictor. However, there is reason to believe that population size and capacity to speak the national language would be predictors with a larger sample. The paper concludes with a preliminary analysis of those communities having significantly higher or lower mortality rates than would have been expected on the basis of a knowledge of Spanish, education, community size and local medical facilities. Such deviant case analysis may pinpoint “problem communities” or, alternatively, communities with special advantages.