Spatial distribution of proportional infant mortality and certain socioeconomic variables in Salvador, Bahia, Brazil.

The spatial distribution of infant deaths in each zone of the city of Salvador, Brazil, in 1980 was examined and related to selected socioeconomic and health care variables. The spatial distribution of infant mortality (defined as the ratio of infant deaths to total deaths in Salvador) was uneven, ranging from 3.9% in a zone in the center of the old city near the water to 52.5% in an outlying area. There was a direct correlation between proportional infant mortality and low income. The proportion of low-income families was 23.1% in the quartile containing the lowest rates of infant mortality compared with 97.7% in the quartile containing the highest such rates. Also noted was an inverse relationship between proportional infant mortality and water consumption. Similarly, the percentage of substandard housing (shacks) was directly associated with infant mortality, rising from 2% in the 1st quartile (low infant mortality) to 43% in the 4th quartile. There was additionally an inverse correlation between the average number of physicians per inhabitant and proportional infant mortality, with a rate of 5.7/10,000 inhabitants in the 1st quartile compared with 1.8 in the 4th. Overall, analysis of the spatial distribution of infant deaths in Salvador reveals that an especially high proportion of those dying were infants residing in certain zones of the city, most notably outlying neighborhoods and slums. The results further point to an important association between certain socioeconomic and sanitation variables and infant mortality. The fact that the highest correlation coefficient found in the present study was between proportional infant mortality and the percentage of low-income families underscores the significant role that economic structure plays in infant mortality. There is a need both to redefine the urban development model prevailing in Latin America and to reorient the aims of the health sector.