Narrowing geographic inequality in life expectancy in Brazil: a multilevel analysis between 1991 and 2010

To analyze the geographic inequalities in life expectancy (LE) and the probability of survival up to 40 and 60 years in Brazil between 1991 and 2010, to partition the variance of these indicators by including municipalities, states, and macroregions in the analysis, and to test the association between municipal socio-economic and health services indicators with life expectancy. Multilevel analysis. Census data from 1991, 2000, and 2010 were used to calculate the outcomes and the socio-economic variables. Municipalities were separated into centiles according to their values in each outcome. Absolute and relative differences were calculated for each year. Multilevel linear regression models were performed, taking into account three levels: regions, states, and municipalities. Municipal socio-economic and health services variables were included in the model with the 2010 data. All 5545 Brazilian counties showed improvement in the three indicators, but the magnitude varied significantly across the country. The highest gains in LE were observed in the North and Northeast regions. The gap in LE between the 1st and 99th percentile decreased from 19.6 years to 12.2 years. The relative difference also fell, from 1.37 to 1.18. Most counties’ socio-economic and health services indicators were associated with the outcomes and explained 86.7%, 31.2%, and 32.4% of the variation in LE attributable to regions, states, and counties, respectively. The average life expectancy increased between 1991 and 2010. Concomitantly, a reduction in geographic disparities was observed. The counties’ socio-economic and health services variables explained much of the variation of the outcomes in 2010. •All Brazilian municipalities showed an increase in life expectancy between 1990 and 2010.•There was a remarkable reduction in geographic inequality in the country.•Counties and regions accounted for the major part of the total variability of life expectancy.•County’s variables explained an expressive proportion of the life expectancy variation.