Using the UN data, aggregated into six continents (see bmj.com
), we calculated the slope index of inequality for life expectancy from 1950 to 2005. Data for 2001-5 form the central projection used and published by various UN statistical agencies. We calculated the index by continent rather than country for several reasons: we do not have a consistent measure of poverty or wealth to rank all countries at all points in time; data for countries in war or crisis (or both) are unreliable; the ranking of continents by gross domestic product per capita is constant over time; and a measure of intercontinental inequality in mortality captures most international inequality. Data on gross domestic product before 1950 are unreliable, so our study starts at 1950.
The index was calculated by ranking the six continents in order of increasing life expectancy: Africa, Asia, Latin America and the Caribbean, Oceania, Europe, and North America. The order did not change over the 50 year study period and reflects the average gross domestic product per capita of the inhabitants of each continent. The population of each continent was calculated as a proportion of the world population in each time period. At the start of the study 8.8% of the world's population lived in Africa (the poorest continent) and the median person in Africa was 4.4% along the world population ranking. For the period 1950-5 these cumulative proportions for the ranked continents were: 8.8%, 64.3%, 70.9%, 71.5%, 93.2%, and 100%. The median person in each ranked continent thus stood at 4%, 37%, 68%, 71%, 82%, and 97%. By the end of the study period the median person in Africa reached 7% along the world population ranking as the population grew. In the years 1950-5 the life expectancies for the ranked continents were 38, 41, 51, 60, 66, and 69 years. The slope index of inequality is the slope coefficient in a simple regression analysis of life expectancy in years against the ranking of the continents (where ranking is expressed as the cumulative proportions of the world population—for example, 0.37 for Asia in 1950-5). Because grouped data are used and heteroskedasticity (unequal variance in regression errors) exists we estimated coefficients by using the transformation proposed by Low and Low.6
In 1950-5 the coefficient was 41.9 years of life expectancy. This means that the hypothetical poorest person in Africa had a life expectancy 41.9 years shorter than the richest person in North America. The slope index of inequality uses the information for all continents rather than simply comparing the extreme life expectancies.
and the show that global inequality in life expectancy fell between 1950 and 1990, but since then it has risen and is now at the same level as in the late 1970s. For the two most extreme continents (see )—North America and Africa—the gap in life expectancy fell from 30.6 years in 1950-5 to less than 24 in 1985-90 but has since risen to 28.6; it is now almost at the same level as in the 1950s. The slope index of inequality correlates with the dispersion measure of mortality used by Moser et al: both methods show that global inequality in mortality is rising.5
A plot of life expectancy in the six continents over the study period () shows that Africa has been most affected by the widening global inequality in mortality.
Global life expectancy slope index of inequality (in years). Black triangle shows estimated index in 2000-5 with impact of AIDS removed
Mean life expectancies across the globe
Life expectancy (in years)