But just as elsewhere two or three decades ago, the changes are occurring at various paces according to the country, depending on the (numerous) cultural systems, the diversified economies, the political systems, but also on the crises of various kinds that some countries, or even entire sub-regions of the continent, have been experiencing over the last 30 years : conflicts and civil wars, food shortages that sometimes attain the famine stage, a deterioration of living standards, and of course aids. Vulnerability, insecurity and poverty are the fate of many countries, and of large strata of the population within each of them.

In the 1990s the demography of Africa and its sub-regions and countries became the subject of several general surveys. These were mostly collective works with a focus either on all the components of demographic change, or more often on one or the other issue, such as fertility or AIDS [2]. The present chronicle, which is by definition less ambitious, has two objectives. The first is to present data that are as comparable, reliable and recent as possible on the evolution since 1950, and the present characteristics of the population of each country. These data are gathered in a statistical appendix in 14 large tables that correspond to each of the issues addressed in the text. The second objective is to achieve a synthetic view of the major components of change and of the growing diversification of the situations in the region. The information is summarized in graphs and smaller tables. This chronicle focuses on sub-Saharan Africa, and excludes the five countries of North Africa, which have a different history, culture and demography. Together with the Middle East, North Africa will be the subject of a future chronicle.

We begin with a few words comparing the African social and economic context with that of other regions, a review of the progress achieved in the area of information systems and an overview of the evolution of the population over two millennia. Next we consider successively : 1) the size and growth of the populations since 1950 ; 2) the sub-regional patterns of demographic transition ; 3) nuptiality (age at first marriage, polygyny, marriage dissolution) ; 4) fertility (levels, trends) and its proximate determinants (e.g. breastfeeding, abstinence, contraception) ; 5) overall mortality, maternal mortality and AIDS ; 6) child mortality (from age 0 to 5) and health (vaccination, malnutrition, medical services) ; 7) the age structure ; 8) urbanization and the rural exodus ; 9) international migration within Africa as well as to the wealthy countries ; and 10) inequality between men and women in education.

Our approach is essentially descriptive, and involves the study of levels and trends in all countries and sub-regions, the discussion of inequality (in terms of education and type of residence) in countries at different stages in the transition or the process of development (particularly Benin, Nigeria, Mali, Cameroon, Kenya and Zimbabwe). Occasionally we will study the relation between the countries’ demographic indices and various indicators of social, economic, human and health development, which are regrouped by country in Tables A.13 and A.14 of the Appendix.

The preferred sources of data include, on the one hand, the data banks of various United Nations agencies (the Population Division, UNDP, WHO, UNESCO, etc.), which are essential for reconstructing the major changes since 1950 or for identifying specific indicators, and on the other hand, the Demographic and Health Surveys (DHS) or similar surveys. A total of 81 surveys have been conducted since 1984, including 47 since 1994 in 30 African countries. We will also use, with some additions, a few recent studies such as those of V. Hertrich (2001) on age at marriage or of X. Thierry (2001) on immigration in France.

Two of the 50 countries included in sub-Saharan Africa — the Seychelles (453 sq. km. and 80,000 people) and St. Helena (410 sq. km. and 7,500 people) — have been left out of this chronicle because they are very small and poorly documented. The remaining 48 countries are regrouped in the statistical appendix into the commonly accepted four sub-regions : West Africa (16 countries), Central Africa (9 countries), East Africa (18 countries) and Southern Africa (5 countries). The map below shows the geographical location of the countries and their grouping into sub-regions.

Table 1
Some socio-demographic characteristics of sub-Saharan Africa and of the other major world regions in 2001

Overall, the region is also — and by far — the most disadvantaged region in the world from an economic and social point of view. Its income per inhabitant, in terms of purchasing power parity, in 2001 ($1,830) is lower by one third than in South Asia, and about three times lower than that of North Africa and the Middle East. In terms of human development (HDI), the region is clearly behind South Asia and ranks very far back behind Latin America or East Asia. In the UNDP world ranking (2003), 34 out of the 40 poorest countries belong to sub-Saharan Africa [3]. Similarly, in the ranking by degree of poverty, most African countries are among the poorest. Only adult illiteracy (38%) is not worse on the whole than in North Africa (39%) or South Asia (44%). Behind all this, however, are substantial inequalities between countries within the continent (Table A.13 of the Appendix).

Another characteristic of the region is a result of its history, with a colonial period that was much longer than elsewhere. Except for Liberia (created by the United States in 1847) and Ethiopia, most African countries were occupied, and colonized under one form or another from the 1880s onward. Four European countries partitioned sub-Saharan Africa among themselves. France, besides Madagascar, the Comoros and Djibouti, was present in West and Central Africa (18 of the countries existing today). Great Britain (17 countries) occupied a few countries in West Africa (including Nigeria), but mostly in East Africa (Kenya, Malawi, Uganda) and Southern Africa. Belgium colonized three countries (the present-day Democratic Republic of the Congo, Rwanda and Burundi), and Portugal five countries including Mozambique and Angola. Let us not forget Germany which occupied Togo, parts of Cameroon, Tanganyika and Namibia until 1918, and Spain with Equatorial Guinea.

Except for South Africa, which became independent in 1910, all those countries experienced colonial occupation for 70 to 90 years before becoming independent at various dates (Table A.2 of the Appendix). In 1960 France withdrew from most of its colonies, as Belgium did from the DR Congo, and Great Britain from Nigeria. By 1967 most of the countries were liberated, except for the Portuguese colonies that gained independence only in 1974 and 1975, Zimbabwe that was occupied by the British until 1980, the Comoros and Djibouti that got their independence in 1975 and 1977, and Namibia that remained under South African mandate until 1990. Reunion Island became a French overseas département (DOM) in 1946.

On the other hand, the situation improved considerably from the 1970s onward in the area of censuses and surveys, with the development of statistical institutes, the training of statisticians and demographers, substantial external financial support for data collection operations and the launching of large international projects such as the World Fertility Survey (WFS) programme between 1974 and 1982, or the Demographic and Health Surveys (DHS) sponsored by the United States since 1984 [5].

As a result (Table 2) more and more national demographic operations have been conducted in the 48 countries of the region since 1970, and most of all since 1980 : on the average, practically one census and one survey every ten years. National situations, however, vary considerably (Table A.1 of the Appendix), ranging from well documented countries with 3 to 5 censuses and 5 to 8 surveys (Kenya, Ghana, Burkina Faso, Senegal, Cameroon, Zambia) to countries that are statistically very neglected (DR Congo, Congo, Guinea, Gabon, Chad, Somalia, Swaziland). Among the 48 countries, 45 have conducted at least one census since 1994, and 30 at least one DHS. It is mostly thanks to the latter that we know the characteristics of fertility, mortality, and maternal and child mortality and health in many African countries.

Table 2
Number of censuses and national demographic surveys since 1960 in all 48 countries

Table 3
Growth of the population of sub-Saharan Africa since the beginning of our era

Between 1700 to 1900, unlike other regions in the world where the population grew two- or threefold, the population of sub-Saharan Africa stagnated according to J.-N. Biraben (2003), or increased only slightly according to D. Cordell (2001). Its proportion of the world’s population, which had been increasing since Antiquity, declined steadily from the sixteenth century (17%) to the end of the nineteenth century (6%). The demographic growth of the continent generally remained slow until the twentieth century (between 0.13 and 0.21% a year, depending on the period). Historians are unanimous that the period of intense colonization (1880-1920) was even a phase when demographic growth slowed down or sometimes regressed sharply in certain regions (forced labour, displacement of population, importation of diseases, etc.). Only in the 1920s did Africa witness a sharp resumption of its population dynamics.

Since 1990, the whole of sub-Saharan Africa has entered a phase of demographic slowdown, with the growth rate declining from 2.9% a year circa 1985 to 2.3% in 2000. But the slowdown is occurring at various or even divergent paces among the countries (Figure 1). At one extreme, in close to twenty countries from different sub-regions, growth rates have not changed, or have even slightly increased since 1985 (e.g. Niger, Mali, Mozambique, Somalia). At the other extreme, in five Southern African countries plus Zimbabwe and Zambia, growth rates have practically collapsed from 1995 onward, as a result of excess mortality due to AIDS (the topic will be discussed again later). For example, South Africa and Botswana’s growth rates declined from 2.0 and 2.8% respectively in 1990-94, to 0.6 and 0.9% ten years later. This is a unique event in history.

Between those extremes, some 25 countries undergo a decline of growth rates, but at different paces (Figure 1) : rather slow in West and Central Africa, slightly more rapid (on average) in East Africa. And there is no clear relation between the pace of decline and the initial level in 1980-84. Finally, between 1990 and 2005, the populations of the various sub-regions and of most of the countries increased by approximately 50%, except in Southern Africa (24%).

Southern Africa presents a different picture. Still according to the United Nations projections, the growth rates, which have recently collapsed, will soon turn negative and will remain so until 2040 (approximately – 0.30% between 2010 and 2040). The population will significantly decline : between 2005 and 2040, the region will decrease from 52 million people to 47 million, thus recovering its size of 1995. South Africa and Swaziland will lose 10% of their population, and Botswana 19%. The effects of AIDS, as we shall see below, will also cause age structures to be drastically destabilized.

These national densities, however, conceal huge inequalities within the countries. In Benin, for example, the Atlantic coastal district where the capital and most economic activities are located accounts for 25% of the population and 3% of the national territory ; rural densities, moreover, range between 10 and 300 inhabitants per sq. km. depending on the region. In Kenya, 90% of the people live on 20% of the land. In Burundi, certain rural communes exceed over 350 inhabitants per sq. km., and some hills exceed 700.

On the other hand, if calculated in the classic manner (number of inhabitants divided by the total land area), those densities are poor indicators of the relation between population pressure and agricultural resources (Mathieu and Tabutin, 1996). By recalculating the densities, as we did, based on the surfaces of arable land and permanent crops (Table A.2 of the Appendix), the perspective changes completely. For example, densities rise in Mauritania from 3 to 529 inhabitants per sq. km., in Senegal from 48 to 391, in Kenya from 53 to 676, in Somalia from 14 to 817. Africa thus appears to be far from under-populated, in terms of farmed or cultivable land.

In Africa, four main situations or patterns currently coexist. This appears clearly by taking four countries that have been through diverse economic, political and health experiences in the last 20 years — Mali, Ghana, Liberia and Zimbabwe (Figure 2) :

Economic development, poverty, demographic and social policies, pandemics such as AIDS will determine each society’s transition pattern in the next two or three decades, with each country following very different paths.

We shall now examine the levels and trends of nuptiality, fertility, mortality and migration, on which those past or recent demographic dynamics depend directly.

Traditional African nuptiality regimes were generally characterized by very early entry into a union for women and at a much later age for men, by the widest age differences between the sexes in the world, by the rejection of celibacy, by frequent remarriage of widowed and divorced women, and by the practice of polygyny that was ubiquitous and often frequent (Hertrich, 2001) [9]. What is the current situation ? Table A.4 of the Appendix presents male and female median ages at marriage in 48 countries in the late 1970s and 1990s, the age differences between spouses at those two dates, and the current situation of the proportions never married and of polygyny. The data are summarized by sub-region in Table 4.

Table 4
Women’s median age at first marriage, age difference between spouses and proportion of polygynous women aged 35-44, by sub-region

Among men, the ages of entry into union, which were relatively late already in the 1970s, are often also rising, but less so than among women. In a majority of countries, they are now around 25 or 26, except in Southern Africa and Reunion, where they attain 31 (Table A.4).

As a consequence, age differences at first marriage between men and women, which were very wide in the 1960s and 1970s (usually 6 to 7 years), are now narrowing slightly, though they remain important in a majority of countries (Figure 3).

However, its prevalence, which is high on average, was and is variable between sub-regions of the continent, between countries within sub-regions, or between ethnic groups and living environments within countries (Sala Diakanda, 1980 ; Klissou, 1995). Polygyny has always been most widespread, and by far, in West Africa and the Sahel (Table 4), with currently 30% to 60% of women aged 35 to 44 living in polygynous marriage, with the exception of Mauritania (Table A.4 of the Appendix). It is only slightly less widespread in Central Africa, but markedly less important and more variable in the 17 East African countries (ranging from 4% in Madagascar to 39% in Uganda), and most of all in Southern Africa (14%).

However, the beginning of a trend towards widespread decline of polygyny has been observed almost everywhere in recent times, though it mainly affects the cities and the most privileged or educated social classes. For the time being, one out of every three African women aged 40 is in a polygynous marriage (Table 4) ; almost one out of two may still be involved in that regime at one time or another.

The few studies of divorce conducted in the 1970s and the 1980s have all demonstrated the importance of the phenomenon as well as its diversity over space. According to R. Lesthaeghe et al. (1989), divorce, generally followed soon after by remarriage, was more frequent in West Africa (45% to 53% of unions dissolved in the different countries) than in East Africa (29% to 34%). Although there are few recent studies, those that are available on West Africa have confirmed the phenomenon’s importance. After 30 years of marriage, 35% of women’s first marriages in Ghana, 38% in Togo, 29% in Mauritania are dissolved by divorce ; in Dakar, one out of every three marriages ends in divorce (Antoine and Dial, 2003). In Burkina Faso (Thiombiano, 2004), a mostly rural country, divorce at the national level is less frequent than elsewhere (16% after 30 years of marriage), but it affects one out of every five women in the cities, with increased risks for early marriages, educated women and the younger generations.

Widowhood is obviously another major cause of marriage dissolution in Africa, in contexts where mortality is still high, particularly among men, and age differences between spouses are often important. In Burkina Faso for instance (Thiombiano, 2004), one out of every ten first marriages is dissolved by the man’s death before 20 years have elapsed, and one out of every five before 30 years have. The cumulative impact of divorce and widowhood results in particularly high probabilities of dissolution of their union for women, even before the end of their fecund life.

To the extent that marriage remains a quasi-universal social norm in Africa, these marriage dissolutions by divorce or widowhood are often followed by remarriage, particularly when the women are still comparatively young and fecund. Again, the phenomenon varies throughout the continent, and remarriage is on the whole more frequent and rapid in West Africa than in the eastern or southern parts of the continent, with relatively marked differences between cities and countryside and among social groups.

For the 1970s and the 1980s, R. Lesthaeghe et al. (1989) and V. Hertrich (2001) have distinguished between several main regional patterns of nuptiality. These include the Sahel and the western and central savannah regions, where marriage occurs at a very early age, polygyny is widespread and the age differences between spouses are large ; East Africa, where polygyny is less frequent and marriage occurs at a later age ; and Southern Africa with little polygyny, a fairly late age at marriage, and large proportions of never married individuals, in part because of labour migration.

Approximately the same regional patterns still prevail in the late 1990s, twenty years later, with slightly different results (Table 4), but still with two extreme situations. On the one hand, West Africa, the most polygynous region in the world, has significant age differences between spouses ; on the other hand, Southern Africa has a very different pattern, with late marriage, a significant proportion of never married, little polygyny, and small age differences between spouses.

Those significant spatial and cultural disparities in marriage patterns should be kept in mind as we examine the levels and trends of overall fertility.

The median age at first childbirth, which is partly related to age at first union [14], is 19.7 (i.e. every other African woman has her first child before 20) for sub-Saharan Africa as a whole, with regional averages ranging from 18.9 in Central Africa to 20.9 in Southern Africa. In a majority of countries, women have their first birth in their nineteenth or twentieth year (Table 5) and only in four countries above age 21 : the Comoros, Eritrea, Namibia and Rwanda where, along with Burundi, age at first marriage has always occurred later, and extramarital conceptions are rare. Excluding a few small countries such as Mauritius, or Reunion, fertility still begins at an early age in Africa, and no important changes have occurred.

Table 5
Number of countries by sub-region according to the median age at first birth and the proportion of adolescent girls (15 to 19) who have started childbearing

The high level of adolescent fertility confirms this clearly. The proportion of girls aged 15-19 (with an average age of 17.5) who either have had a child or are pregnant is 25% for the whole of the region, ranging from 19% in Southern Africa to 34% in Central Africa. In a majority of countries it is over 25% (Table 5), with some Sahelian countries (Mali, Niger, Chad), and also Madagascar and Mozambique close to 40%. Countries like Ghana (14%), the Comoros (9%) and Rwanda (7%) occupy the other extreme.

As for the proportion of total fertility achieved before 25 years [15], it is around 35% in a large majority of countries (Table A.6 in the Appendix), with Namibia (20%) at one extreme and Mozambique (40%) at the other.

Early timing of fertility (whether in or out of wedlock) and high levels of adolescent fertility remain two major characteristics of African fertility.

The demand for children naturally varies from one sub-region to the other (ranging from 3.9 in Southern Africa to 5.6 in West Africa) and from one country to the other. In 12 out of 30 countries (Table A.6 of the Appendix), desired fertility is still higher than actual fertility (TFR) ; in the remaining 18 countries it is either the same, or markedly lower.

Changes are undoubtedly occurring. The demand for children is lower than it used to be, the proportion of women of a given age or parity who declare that they do not want any more children is increasing, and social and regional differentials (for example, between educational levels) are well established, in conformity with the geography of fertility. As we wrote as early as 1997 (Tabutin, 1997) :

Table 6
Total fertility rates by type of residence and women’s educational level in six countries at various stages of transition

Today, the city and the countryside are distinct worlds. On the whole fertility rates in the different types of environment are no longer similar, as they used to be in the 1970s. In every country, urban fertility is noticeably lower (Shapiro and Tambashe, 2003a) [17], with between 3 and 4 children, as against 5 to 6 children in rural environments (Table 6). In the capitals and large cities in particular, fertility is markedly lower than elsewhere, the ideal family size is smaller, marriage is later and polygyny less common ; as we shall see below, breastfeeding and postpartum abstinence periods are shorter, and contraception of course is more widespread. Even in countries well advanced in the fertility transition (Ghana, Kenya and Zimbabwe in Table 6), the capital and the rural world still differ by 1 to 2 children.

Similarly, education is playing an increasingly important role everywhere, as in other regions of the world. Except for a very few countries that are in a pre-transition phase (Niger, Chad or Mali), primary education is sufficient to change norms and behaviour significantly, though for the time being it is mostly access to secondary (and higher) education that induces a sharp fall in fertility (to around 3 children in many countries). Education, as we shall see, fundamentally changes the role of intermediate variables such as breastfeeding, abstinence and contraception.

Similarly, almost everywhere regional diversity in fertility is increasing in the countries, as a consequence of frequent serious inequalities in cultural, economic, educational or health development. For example, excluding the capital regions, fertility currently ranges, in the different regions, from 5.0 to 6.9 children per woman in Benin, from 4.4 to 6.6 in Cameroon and from 3.7 to 5.7 in Kenya. The comparative spatial homogeneity observed during the 1970s and the 1980s is definitely a thing of the past.

There have been no major changes in the last ten years, and the correlations are of the same order, and significant (R²= 0.62 for human development, R²= 0.70 for infant mortality). In general, the higher a country’s human development level, the lower its fertility ; the more infant mortality is under control, the lower the fertility level. In other words, no fertility transition took place without a drop of infant mortality or without socioeconomic and health progress. In that sense, Africa seems to be conforming to the classic path of the demographic transition theory. But only the extreme situations stand out clearly. The ten poorest countries in the region all have more than 6 children per woman, whereas the six or seven most “developed” ones have between 2.2 and 3.9 children. In between, there is uncertainty and diversity [21]. In Africa as elsewhere, there will probably be multiple paths leading from high to low levels of fertility.

Birth intervals, which traditionally lasted approximately three years in most African cultures, have not changed in a noticeable manner (Table A.6 of the Appendix) : except for Zimbabwe and South Africa where they are now longer (over 40 months, due to contraception for spacing), they are still about 34 months in most countries [22], ranging from 29 months in the Comoros to 38 months in Ghana.

Table 7
Breastfeeding, postpartum sexual abstinence, contraception and sterility by major sub-region around 2002

Comparing the results of the 1980s, 1990s or 2000s DHS, it appears that breastfeeding periods have remained virtually unchanged at the national level. However, they are usually shorter in the cities than in the countryside, and among literate than among illiterate women. This may foreshadow future changes, though not abrupt ones. Meanwhile breastfeeding still contributes largely to lengthening birth intervals.

Whereas breastfeeding is keeping up, abstinence is undergoing more fluctuations, with the duration decreasing in many countries during the last 15 years : it went down from 14 to 8 months in Ghana, from 7 to 2 in Mali, from 6 to 3 in Kenya, from 4 to 3 in Zimbabwe. This is probably the behaviour that is most susceptible to contextual changes. Everywhere, its duration is curtailed most in the large cities, among literate couples or in the more privileged social classes. For example, in Benin in 2001 (Table 8), though the country is poor and at the very beginning of the transition, abstinence now lasts only 5 months in Cotonou (the largest city) and 4 months among well-educated women, as against 10 months in the rural area and 13 months among the illiterate.

Table 8
Age at first marriage, breastfeeding, postpartum abstinence, modern contraception and fertility in benin (2001) by type of residence and women’s educational level

While they are evolving, these postpartum breastfeeding and sexual abstinence practices remain markedly more intense and more widespread in Africa than in the rest of the world. They still induce long periods of postpartum non-susceptibility [24] (12 months in Kenya, between 14 and 16 months in Ghana, Mali and Benin, 19 months in Côte d’Ivoire). In a vast majority of countries, they are still the main fertility regulator in the absence of widespread contraception [25].

Around 2000, modern contraceptive prevalence among married women aged 15-49 is still under 8% in 15 countries (out of the 30 in Table A.7 of the Appendix), between 8 and 15% in six countries, between 15 and 30% in five countries. Only three countries (Cape Verde, South Africa and Zimbabwe) [26] come close to or exceed 50%. Changes are occurring almost everywhere, though usually on a small scale. In 10 years, for example, contraceptive prevalence in Mali rose from 1 to 6%, in Benin from 2 to 7%, in Ghana from 5 to 13%, in Cameroon and Côte d’Ivoire from 4 to 7%.

Except for seven or eight countries, the contraceptive revolution is still a long way off in Africa, though it has started in the large cities and the capitals. Some have only reached a contraceptive prevalence around 15% (e.g. Cotonou, Yaoundé, Abidjan, Libreville), but others verge on 25% (Ouagadougou, Dakar) or have even exceded 35% (e.g. Addis Ababa, Nairobi). Highly literate couples, who are a minority in every country, are everywhere between 30 and 50%. Although health centres do not yet diffuse contraception effectively among unmarried women, it is also gradually reaching that population, but at a different pace from one country to another.

Nowadays it is no longer only a question of knowledge of contraceptive methods (there has been great progress in that area) [27], but at the same time a problem of demand on the part of a large majority of the population (as stated previously, the demand for children is still high among women, and still more so among men) and of availability of services (in terms of geographical or cultural accessibility as well as of quality).

In the area of contraception, the situation in Africa differs from what is happening or has happened in Asia or Latin America in two major ways : 1) when contraception is practised in marriage, it is still mainly with a view to spacing births, more or less as a substitute for abstinence ; 2) therefore it relies on reversible methods (e.g. pill, IUD, rhythm), with very little female sterilization and no vasectomy at all. Outside of Southern Africa or Kenya, few countries practice stopping behaviour. The vast majority of African cultures still abhor sterility, sub-fecundity and early infertility. Everywhere the proportion of women who say that they do not want any more children increases substantially with age, but a wide gap still persists between stated desires, real expectations and behavioural change.

Overall, as expected, modern contraceptive practice is negatively associated with fertility [28], even though the general level of the former is still low (13%). But the relationship is neither as clear nor as linear as in other parts of the world (Figure 7).

A great variety of situations are observed. For a total fertility of 6 children, prevalence ranges from 5 to 25%. With low prevalence, national fertility rates range from 4 to 6. Countries like Ghana, Mauritania or Eritrea achieved a first fertility transition (from 6 to 4.5 children) without great diffusion of contraception. Even better, cities like Lomé, Yaoundé-Douala, Accra or Abidjan went down to 3 children with little contraception (10 to 15% only among married women), whereas others have achieved the same fertility rates with more common prevalence rates of 30 to 40%. Outside of Southern Africa or countries like Kenya, Cape Verde or Reunion, this confirms the fact that contraception plays an extremely variable and sometimes very small role as a fertility regulator.

On the basis of a special survey conducted in 1998 on 2,400 women in Abidjan, A. Guillaume (2003b) was able to measure the impact of abortion on fertility. In 1998 it reduced the total fertility rate (for women 15-44) by 12%, or by 0.6 children per woman. The effect on age-specific fertility rates was variable (22% at 15-19, 16% at 20-24, and 9% at 25-39). Abortion had virtually the same weight in the decline of fertility in that city between 1994 and 1999 as the postponement of first marriage and the rise of contraception.

Resort to abortion, which is related to various economic, cultural and social reasons in the different countries, is probably on the rise in many African cities. It has — and will have — consequences not only on fertility, but also on maternal morbidity and mortality, in the absence of legalization and of adequate health structures.

Table 9
Life expectancy levels in 1952, 1972, 1992 and 2002, maternal mortality and HIV prevalence around 2001 by sub-region

Obviously, the progress achieved up to the late 1980s was not uniform between countries, and this led to growing inequalities in health and mortality in the region. Around 1950, life expectancy (Table A.8 in the Appendix) ranged from 30 (Angola, Sierra Leone, the Gambia) to 51 years in Mauritius and almost 53 in Reunion. Around 1990, it ranged from less than 40 years (in war-torn countries such as Sierra Leone, Burundi, Angola) to 66 years (Cape Verde, Sao Tome and Principe), and even 70 years in Mauritius and 73.5 in Reunion, while it remained at the time between 45 and 53 years in many countries.

Because of AIDS, wars and economic recession, these health transitions, which on the whole were already less rapid than elsewhere, slowed down or stopped abruptly during the 1990s.

In sum, the African countries have experienced diversified health and mortality evolutions in the last 15 years, as illustrated in Figure 8, where we distinguish between five types depending on the recent evolution of life expectancy : a) steadily progressing countries, b) recently stagnating countries, c) countries in marked regression, d) countries with complete reversal and e) war-torn countries [31].

The irruption of AIDS in the 1980s and its rapid expansion in some countries during the 1990s is of course the cause of a number of these reversals, which were often spectacular and almost without precedent in the history of mankind. As mentioned above, about 15 countries are most severely affected. But in all the countries where mortality has stagnated in recent times, the effects of the economic crisis, of structural adjustment programmes that have disrupted the health and education sectors, and of poverty, have combined to cause a fresh upsurge of infectious diseases that had been “under control” until then (e.g. tuberculosis). There is also a group of countries (Figure 8) that in the last 15 years have been through bloody and sometimes protracted civil wars, culminating in genocide in Rwanda. All those conflicts obviously have direct, but also indirect, effects on the health and mortality of the population, through the dismantling of land tenure and agricultural systems, the slowing down or interruption of international assistance, population displacements, etc.

But HIV prevalence varies considerably through the continent (Table 9), ranging at the sub-regional level from 4.5% in West Africa to almost 30% in Southern Africa. It is less than 2% in only six countries out of the 38 for which data are available (Table A.10 of the Appendix), between 2 and 10% in 20 countries, and more than 20% in seven countries, of which all the Southern African ones. A large proportion of African countries and societies are thus fully concerned by this unprecedented epidemic, which in general is far from being under control. Although HIV prevalence is decreasing in a few countries such as Uganda, where the fight against AIDS became a national priority, or in a few capitals (e.g. Addis Ababa in Ethiopia), the worst is yet to come in countries where prevalence shows no sign of levelling off (e.g. South Africa, Botswana, Lesotho) (UNAIDS, 2003).

The major characteristics and the present or foreseeable consequences of AIDS in Africa are quite well documented. A few words will suffice for a simple summary. Unlike in developed countries, in Africa HIV transmission is mostly heterosexual, and the majority of seropositive women were in fact infected by their husband (Cohen and Reid, 1999). Mother-to-child HIV transmission during pregnancy, delivery or breastfeeding represents the other major mode of transmission of the virus in the region. Breastfeeding is responsible for about one third of the cases of vertical HIV transmission, and this risk constitutes a genuine public health issue. Prevention is quite complex, and there is no simple solution. The use of substitutes for mother’s milk also presents high nutritional and infectious risks, which may be even higher than those from HIV. The dilemma is all the more acute as the risk of transmission from an HIV-positive mother to her child is not absolute, and amounts to 20 to 40% (Desclaux and Taverne, 2000). Another characteristic of the epidemic in sub-Saharan Africa is that women are in proportion substantially more affected than men, by at least 20% (UNAIDS, 2003), because they become sexually active at a younger age, with older sexual partners, and because the risk of HIV transmission from men to women is higher than the other way around.

The demographic consequences of AIDS have been discussed on several occasions. The most obvious consequence is the rise of mortality at adult ages (Timæus, 1999), and AIDS today is sometimes considered as the first cause of adult mortality in sub-Saharan Africa. For example, a study on Zimbabwe has shown that the probability of dying between ages 15 and 50 has increased almost threefold in 15 years, rising from close to 150‰ in the early 1980s to more than 400‰ in the late 1990s (Feeney, 2001). Child mortality, as we shall see below, rose also significantly in the more seriously concerned countries.

The slowdown in population increase, and even the emergence of negative growth in the most severely affected countries that we mentioned, is a consequence of the increases in mortality. The slowdown, which is obvious but moderate at the level of the continent, is extremely serious in the Southern African countries. For example, the natural growth rate of Botswana is likely to be slightly negative during the 2000-2015 period, whereas it would have been about 2.2% but for AIDS. The population of Lesotho too will be decreasing slightly, whereas a 1.6% growth rate would have been observed if there had been no AIDS (United Nations, 2003b). Age structures, as we shall see in the instance of Zimbabwe, will also be disturbed by the sudden increase in mortality [33].

Beyond its demographic effects, the AIDS epidemic naturally also has huge economic, social and health consequences. The population of working age is the most seriously affected by HIV/AIDS, and this results in a loss of human capital because of the rising mortality and a decrease in productivity among people suffering from AIDS. Health systems are also being greatly weakened by the epidemic, which absorbs a growing part of the available resources. Educational systems too are affected, particularly through the rise of mortality among teachers. Another important consequence of the epidemic is the rapid growth of the number of orphans, both in absolute and in relative terms. In Zimbabwe in 2001 there were already about 18% orphans among the under-15s, three-fourths of whom were orphaned as a result of AIDS.

When this is said, maternal mortality varies in Africa as do other phenomena (Table A.10 in the Appendix). It is very high on average (around 950 for 100,000 births) in West, Central and East Africa. At 310 in Southern Africa, it is still three times higher than in North Africa. Excepting Mauritius and Reunion where it is low, it ranges from 100 (Botswana) to 2,000 (Sierra Leone), with many countries between 700 and 1,200. Maternal mortality is clearly a major public health issue for the entire region. Without going into further details, maternal mortality is related both to the supply of health care (the quantity and quality of services) and to the demand of the population for antenatal care, conditions of delivery (place and type of assistance) and post-natal follow-up of mothers. It is also related to fertility (e.g. number of children, age at delivery) and to mothers’ health condition. Some of these aspects will be examined in the next section.

As in the case of fertility and life expectancy, the data on the evolution by country since 1950 were drawn from United Nations estimates [35](Table A.8 of the Appendix), and the recent data on mortality by age (under 5), the conditions of delivery, vaccination coverage and malnutrition rates, are drawn from the latest DHS in 30 countries (Table A.9 of the Appendix).

Table 10
Infant mortality from 1952 to 2002 by sub-region

From 1952 to 1972 infant mortality in Africa declined almost as fast as elsewhere. Between 1972 and 1992 the decline was already not as rapid, but it is mostly since the early 1990s that the signs of its slowing down or stalling have multiplied. The decline amounted to 10% between 1992 and 2002 for the whole of sub-Saharan Africa, as against approximately 25% for all other regions (Table 10). As we shall see below, there has even been a recent upsurge of mortality in some countries.

Today, with a current infant mortality rate of the order of 95‰, Africa is far behind the other regions, where rates range from 30 to 50‰. Since mortality between ages 1 and 5 is also very high at around 75‰ (Table 11), currently one African child out of every seven dies before age 5. And yet there has been some progress, though slower than elsewhere and slower than was expected in the 1980s, and variable from one sub-region and from one country to another. Southern Africa is still well ahead (with an infant mortality rate of 52‰), but has been making no further gains since AIDS (which involves mother-to-child transmission) began to spread. At the other extreme, Central Africa, riddled with conflicts and poverty, has the highest infant mortality rates (116‰), and here too no progress is taking place. East and West Africa are still making progress, but rather slowly. Now, let us to take a closer look at the changes.

Table 11
Mortality under 5 by age, conditions of delivery, vaccination and malnutrition by sub-region around 2000 (29 countries)

Such growing heterogeneity is a result of the diversity in the pace of progress, which incidentally has little relation to initial levels. It is related to social policies (health, education, family planning), political history, the environment, and the degree of economic development in each country. In Africa, the mortality of children is closely related to the human development index [38]. And there are neighbouring countries that on the whole have the same climatic and cultural environment, and yet experience different situations. For example, the probability of dying between 0 and 5 is 108‰ in Ghana, while it is 146‰, 181‰ and 160‰ respectively for its neighbours, Togo, Côte d’Ivoire and Benin. A set of factors contributes to this lower mortality in Ghana. Though not much richer than its three neighbours (Table A.13 of the Appendix), Ghana is one of the sub-Saharan countries ranking highest in terms of human development (8th out of 47 countries), with a comparatively high level of adult literacy (72%), a slightly more effective infant and child health policy (Table A.9 of the Appendix), a higher prevalence of modern contraception (13% as against 7% in the neighbouring countries) and a somewhat lower fertility level (4.6 children per woman in 1998, as against 5.2 to 5.6 among the others).

In all the countries, there is great spatial and social inequality in the mortality from 0 to 1 as well as from 1 to 5, often with threefold differences between the cities and the countryside or between social groups. We shall come back to this point later.

Along with Malawi, Guinea, Eritrea or Ethiopia, Tanzania and Ghana (which are both included in Figure 10) have on the whole made progress in the last 25 years, even if there was a slowdown in Ghana recently at ages 1 to 5. In Burkina Faso, where there seems to be more inconsistency between the two surveys, progress has halted and mortality is stagnating, as it is also in Mali and Gabon among others. Worse, mortality is rising again in Cameroon and Côte d’Ivoire, with consistent results between the two surveys ; as in Togo, Zambia, Nigeria or Kenya, the rise affects children aged 1 to 5 most of all, who are more susceptible to infectious diseases. Finally, Zimbabwe exemplifies the negative impact of AIDS on the mortality of children, which is also observed in Zambia, Botswana and, to a lesser degree, South Africa. In sum, as of 2000 many countries find themselves once again with the infant and child mortality rates of the 1980s.

Those different or even diverging trends between countries, which are not related to the level of health development reached in 1990, must be attributed to a set of intertwined factors whose weight varies from one country to another : AIDS, of course, particularly in those countries most heavily affected, but also the deteriorating living standards and purchasing power in many countries, the dismantling of public health systems, the universal rise in health costs, which in Cameroon (Beninguisse, 2003) for example has caused notable changes in the behaviour and practices of the population (among the poor as well as among the middle classes) in such areas as antenatal care, delivery care or vaccination.

Table A.9 of the Appendix presents, for 30 countries with recent DHS, the proportion of births occurring without antenatal visits, the proportion of deliveries unassisted by medical or trained personnel, the proportion of children who receive the complete set of vaccinations (BCG, DPT, polio and measles), and the proportion of children in a state of