How much progress has Sub-Saharan Africa made in reducing mortality?

Yesterday I read Social Determinants of Health Inequalities by Michael Marmot, which was published in the prestigous medical journal The Lancet. One of Marmot's arguments is that "Differences in adult mortality among countries are large and growing." In particular, he claims that between 1970 and 2002, "mortality rose in Africa [...] whereas it declined in the world as a whole." What evidence does he provide for this claim? He begins by dividing African countries into two groups, a high-mortality group and a low-mortality group. He then plots the probability of death between ages 15-60 over time, separately for the two groups of African countries. His charts show that mortality was more-or-less flat for the low-mortality group and had risen for the high-mortality group. Is this a fair description of mortality trends in Sub-Saharan Africa?

Not at all. First, there is no reason to arbitrarily partition African countries into a small number of groups on the basis of their mortality levels. And second, Marmot's measure of mortality--the probability of death between ages 15-60--is highly sensitive to a population's age-structure, and so is not very useful for comparative purposes.

The charts below present a more accurate picture of how mortality in Sub-Saharan Africa has changed since the 1950s. With the unfortunate exception of Zimbabwe, every country has made progress. Moreover, a large number have enjoyed substantial gains. For instance, the typical Gambian now lives a full 25 years longer than she did in 1955. Indeed, the average increase in life expectancy over the time period is >16 years. And the average increase in life expectancy over the shorter time period that Marmot considered is nearly 8 years.

Incidentally, the large dip in the light blue line corresponds to the tragic Rwandan genocide. And the regional dip during the mid-1990s corresponds to the HIV/AIDS epidemic. Data are from the UN.