**2. Data**

238 Social and Psychological Aspects of HIV/AIDS and Their Ramifications

stability (or crises) and with economic growth (or recessions), and with the local development of public health, so that the whole picture might appear confusing at first

HIV stroke Africa in the mid-1970's, and spread rapidly throughout the continent, and extensively in Eastern and Southern Africa, with some pockets in West and Central Africa. [Buve et al., 2002; UNAIDS, 2010] By the mid-1990's HIV prevalence was already high in about half of African countries and increasing rapidly, with values ranging from 5% to 15%, well beyond the 1% threshold considered necessary for a rapid spread in the general population. By the mid-2000's the epidemics had stabilized, and HIV prevalence was declining in most countries. Data on HIV prevalence deal primarily with adults aged 15-49 years, and often ignore the children. However, mother to child transmission of HIV is common, either before birth, during delivery or after delivery through breastfeeding, so that a significant proportion of newborns are infected with the virus, and likely to die shortly afterwards. Until recently, HIV infection to children born to HIV positive mothers was common and resulted in high mortality. Since then, efforts were made to limit the mother to child transmission by various means, and to treat infected children with newly available

The dynamics of HIV epidemics in Africa vary widely, with some countries heavily infected (as in Southern Africa) and some other hardly touched by the disease (as in Sahelian West Africa). As a result, the net effect of HIV/AIDS on child mortality is likely to be contrasted,

Several studies have tried to estimate the net effect of HIV/AIDS on child mortality in Africa. [Houweling et al. 2006; Korenromp et al., 2004, Mahy, 2003]. Adetunji [2000] provided an overview by comparing point estimates of under-five mortality in the late 1980's and early 1990's with the late 1990's, using published estimates from Demographic and Health Surveys (DHS). He showed an increase in mortality in countries with high HIV seroprevalence, but a decline in others. He found that in Africa HIV mortality accounted from 13% to 61% of under-five mortality depending on the country. Newell et al. [2004] conducted a similar exercise by using parameters of survival after HIV infection drawn from empirical evidence, and concluded that by year 2002 some 10% of deaths of children were caused by HIV/AIDS. Zaba et al. [2003] compared several countries, and found that HIV/AIDS could account from 10% of deaths of under-five children (in Malawi) to 60% (in Botswana). Walker et al. [2002] found 7.7% of under-five children due to HIV in 1999 in 39 African countries, with a range from 0.4% to 42% (in Botswana). Several authors have reproduced the figures recently issued by UNAIDS, and quote a value of 4.4% of under-five

The aim of this paper is to provide a synthesis on the probable impact of HIV/AIDS on child mortality trends in Africa, in a broad historical context since 1950. We will stop in year 2005, the time when HIV/AIDS mortality was the highest among children. The situation changed after this date with respect to mother to child transmission and treatment with antiretroviral therapy. Furthermore, data were lacking after 2005 for many of countries selected for the study. We will focus on long term trends, and on the heterogeneity between countries, summarized in large areas or groups of countries. This study is an extension of earlier work which presented a full scale reconstruction of under-five mortality trends in countries of sub-Saharan Africa since 1950. [Garenne, 1996; Garenne & Gakusi, 2004 &

deaths due to HIV/AIDS in Africa [Black et al., 2010; Stanecki et al., 2010].

glance.

drugs.

2006a]

depending on the country.

Data on under-five mortality were drawn from the maternity histories recorded in Demographic and Health Surveys (DHS). These surveys provide data that allow one to compute age specific death rates by period, and therefore the under-five death rate by calendar year. All 38 countries for which data were available were kept for the final analysis, covering most of continental sub-Saharan Africa, and several islands (see Figure 1). Data on HIV seroprevalence among pregnant women were taken primarily from the UNAIDS database. [UNAIDS, 2008] When necessary, they were completed with data from DHS surveys in a few countries. Data on Gross Domestic Product (GDP) were drawn from the data base built by Angus Maddison and colleagues in its latest edition [2010]. These income data are given in Purchasing Power Parity (PPP), and in constant 1990 dollars.1
