The AIDS epidemic has caused a substantial increase in mortality among adults of reproductive ages [1,2]. One of the consequences of the rapid increase in adult mortality is an increase in the number and proportion of children who are orphaned. According to UNAIDS estimates, sub-Saharan Africa accounts for more than two-thirds of the world's HIV-infected individuals  and for as much as 80% of the world's orphans as a result of AIDS .
This paper uses data from 40 recent national household surveys carried out in sub-Saharan Africa since 1999 to analyse the levels and trends in the prevalence of orphanhood. The cause of parental death is generally not known in population-based surveys, and it is therefore difficult to ascertain directly the impact of AIDS on orphanhood. An attempt is made to assess the impact indirectly, by examining the current prevalence of orphanhood and care patterns, and comparing the results of recent surveys with survey data from the early and mid-1990s in selected countries. Finally, the paper presents the increasing prevalence of orphanhood in the context of general childcare patterns in the different societies.
The descriptive analysis is based on data from two major international survey programmes: the Multiple Indicator Cluster Surveys (MICS) supported by the United Nations Children's Fund (UNICEF), and the Demographic and Health Surveys (DHS) sponsored primarily by the United States Agency for International Development. Both programmes support standardized nationally representative population-based household surveys in many developing countries.
Both MICS and DHS select nationally representative samples using a two-stage design. In the first stage, sampling clusters (usually census enumeration areas) are selected proportional to size. In the second stage, a sample of households is selected from a complete household listing for the cluster. A household questionnaire is administered to all households selected and found, followed by interviews of eligible individuals, usually women (and men) in the 15–49 year age range. The information collected on the household and its members includes the survival status of the biological parents, the residence status of surviving parents for all children residing in the household at the time of the survey, and the schooling status of all children. The household questionnaire may also include questions on child labour.
An additional questionnaire is completed for each child under 5 years of age in the household. In the DHS the questionnaires on child health are administered only to mothers with living children under 5 years of age, so orphans are excluded; the DHS data on orphans are obtained only from the overall household questionnaire, although some DHS include orphans when under-fives are measured and weighed. The MICS include orphans in the questionnaires on child health. Although the DHS and MICS questions are comparable, the MICS interviews are generally shorter. In nearly all instances the National Bureau of Statistics was the main implementing agency.
The analysis focused on data collected in 37 countries during 1999–2002, including 23 MICS and 14 DHS surveys. Three countries with DHS surveys in 1997 or 1998 (Mozambique 1997, Ghana 1998 and South Africa 1998) were also included. These 40 countries cover 97% of all children under 15 years of age in sub-Saharan Africa. For 19 countries, earlier surveys were available for the early 1990s, permitting a trend analysis. Five countries were excluded from part of the analysis because the original datasets of the surveys could not be obtained.
An orphan is defined here as a child under the age of 15 years whose mother or father or both parents have died. A maternal or paternal orphan is defined as a child of whom one parent (mother or father, respectively) has died; a double orphan has no living parent. The Convention on the Rights of the Child defines a child as a person under the age of 18 years, but the household surveys included in this analysis collected data only for children under 15 years of age. Orphans aged 15–17 years are therefore not included. Foster children are defined as children with both parents alive but who do not reside with either of them.
An indication of the severity of the AIDS epidemic during the 1990s is presented by the mean estimated HIV prevalence in the general population aged 15–49 years for the decade preceding the survey. The estimates of HIV prevalence were obtained from a model that uses data generated by HIV surveillance among pregnant women attending antenatal clinics and has been described elsewhere .
Prevalence of orphanhood
Table 1 presents the prevalence of orphanhood in 40 national surveys in sub-Saharan Africa. The survey sample size ranged from 6200 children in Sao Tome and Principe to over 47 000 in the Central African Republic, with an overall median size of 14 524 children under 15 years. In most surveys, the proportion of households with children for whom the survival status of the parents was unknown was fairly small. In five surveys, however, the survival status of at least one parent was unknown for more than 5% of children. The Nigeria 1999 survey had the highest proportion of unknowns (9.2%).
The median prevalence of orphanhood among children under 15 years for the 40 national surveys was 9.2%. Paternal orphans were much more common than maternal orphans in all countries (country median 5.9 and 1.9%, respectively), and 0.9% of children are double orphans. There is considerable variation between countries. Countries currently or recently involved in conflict (Rwanda and Burundi) and countries with high HIV rates (Botswana, Lesotho, Zambia and Zimbabwe) show the highest prevalence of orphans. Orphanhood was more common in the southern Africa subregion than in other parts of sub-Saharan Africa.
The age distribution of orphans is fairly consistent across countries (data not shown). Overall, approximately 15% of the orphans are 0–4 years old, 35% are 5–9 years old, and 50% are 10–14 years old.
Table 2 presents results for countries with data from two or more surveys in the 1990s, classified into three groups using the 2001 UNAIDS estimate of HIV prevalence. The results of the MICS and DHS surveys are generally consistent. In the three countries with HIV prevalence levels below 1% and relatively little epidemic growth during the 1990s, Madagascar, Niger and Senegal, the proportion of children orphaned declined (Fig. 1). Most of the six countries with HIV prevalences between 1 and 5% experienced a modest decrease. Ten countries currently have national HIV prevalence estimates exceeding 5%, and with the exception of Uganda have observed epidemic growth during the 1990s. The orphanhood prevalence increased in eight of the 10 countries. The only exceptions were Uganda and the Central Africa Republic. In Namibia, Zambia and Zimbabwe orphan prevalence increased by more than 50%.
AIDS is also likely to affect the rate of the increase of double orphans disproportionally compared with the increase of single-parent orphans because HIV is sexually transmitted and is more likely to cause the death of both parents than most other conditions. Indeed, there is a clear increase in the proportion of double orphans out of all orphans, especially in countries hit hard by the AIDS epidemic, such as Zambia and Zimbabwe (Table 2). The survey data show no clear patterns and trends in the prevalence of orphanhood by urban rural residence. Orphans are more common in urban areas of 14 countries (difference greater than one-fifth), including four high HIV countries: the Central African Republic, Malawi, Uganda and Zambia. Orphans are overrepresented in the rural areas of seven countries, also including four high HIV countries: Botswana, Kenya, Namibia and Zimbabwe. In the 19 countries with multiple surveys over time, the trends have run both ways (Table 2). In some high HIV countries the prevalence of orphans has shifted significantly from the cities to the rural areas (Kenya, Namibia, Zimbabwe), whereas in others the shift has gone towards the cities (Central African Republic, Malawi, Zambia).
There are important differences in residence patterns for children between countries and regions (Table 1). Children, orphans and non-orphans, are most likely to live with their parent(s) in eastern Africa and least likely to live with their parent(s) in southern Africa. For example, in Botswana, Namibia and South Africa approximately a quarter of non-orphans live with neither mother nor father, whereas in Burundi 91% of such children live with both parents.
The pattern of parent–child co-residence for non-orphans is a fairly good predictor of the pattern for orphans. In countries in which a large proportion of non-orphans are co-residing with their mothers, children who have lost their father are also more likely to be with their mother than in countries where a smaller proportion of non-orphans co-reside with their mother (Fig. 2 and Table 1). The same relationship applies for the chances that a child lives with its father. Based on the country medians, almost nine out of 10 non-orphans live with their mother and nearly eight out of 10 non-orphans live with their father. Children who lost one parent are less likely to live with their surviving parent: three out of four paternal orphans live with their mother and just over half of maternal orphans live with their father. In the countries of southern Africa, single-parent orphans are especially likely to be ‘virtual double orphans', as it is common for the remaining parent to live elsewhere.
There is no difference in the proportion of non-orphaned boys and girls who live with their father (Table not shown). However, girls who have lost their mother are somewhat less likely to live with their surviving father (median of all countries 48%) than boys who have lost their mother (58%). This difference between boys and girls is not found for those children who have lost their father.
For 13 countries, information is available on the relationship to the head of the household of double orphans and single-parent orphans not living with a surviving parent (Table 3). The (extended) family takes care of nine out of 10 of these children. The main caretakers are ‘grandparents’ and ‘other relatives'. Grandparents are the main caretaker for approximately half of the orphans, ranging from 24% (Cameroon) to 64% (South Africa). The second most important group of caretakers is ‘other relatives'. In all countries, very few households had no adults at all to take care of the orphans (child-headed households).
The extent to which the burden of orphan care is shifting over time is of particular interest in the context of the AIDS epidemic. For five countries, information is available from two surveys over time. In three countries, Tanzania, Namibia, and Zimbabwe, the data are suggestive of a shift in the caretaker pattern from other relatives to grandparents in recent years. Kenya, and to a lesser extent Uganda, however, show the opposite trend.
Table 4 summarizes selected household characteristics for 35 recent national surveys with such information. The proportion of households with children that include one or more orphans differs widely within sub-Saharan Africa. Overall, approximately one in six households with children is caring for orphans, ranging from 7 to 37%. In 13 countries at least 20% of all households with children have at least one orphan. These are all countries severely affected by HIV or recently involved in armed conflict. The proportion of households caring for orphans is invariably greater than the proportion of children orphaned.
Women head roughly one in five households with children but no orphans. In southern Africa one in three households are headed by women. Slightly over two-thirds of double orphans live in male-headed households. Double orphans are more likely to live in households that are headed by women than children with surviving parents. In four countries with a high HIV prevalence, Botswana, Namibia, South Africa and Zimbabwe, women are as likely as men to head a household with double orphans. Households with children that are caring for orphans average 1.8 orphans per household. Female-headed households generally assume the care of more orphans (two per household) than male-headed households (1.6 per household).
Households with orphans have a higher dependency ratio (defined as the sum of children under 18 years and individuals 60 years or older divided by the number of individuals aged 18–59 years), as indicated by the median values for each region in Table 4. The median dependency ratio for all households with children but no orphans was 1.5, and for households with orphans 1.8. Households in rural areas have a slightly higher dependency ratio than in urban areas. In 30 of the 34 countries with data the dependency ratio was less favourable for female-headed households than male-headed households.
Heads of household are approximately 4 years older in households that care for orphans than in households with children who are not orphans (median age 47.1 and 43.3 years, respectively). This age difference rises to 5.8 years for the heads of households with double orphans as against households with children who are not orphans, and in five countries, Niger, Sao Tome and Principe, Tanzania, Uganda and Zimbabwe, the heads of double-orphan households are on average at least 10 years older than the heads of households with children who are not orphans (Table not shown).
Impact on education and health
To assess school attendance among orphans and non-orphans, the school attendance of orphans (aged 10–14 years) is expressed in a ratio with the proportion of non-orphans who are attending school in the same age group (Table 5). If orphans and non-orphans have the same likelihood of attending school the ratio is 1. The analysis is limited to the 31 surveys with 50 or more double orphans in the age group 10–14 years.
Overall, school attendance levels differ across the region. In west, central and eastern Africa two-thirds of all children aged 10–14 years attend school. In southern Africa higher overall levels of school attendance are found (country median 87%). In 30 of the 31 countries the orphan to non-orphan attendance ratio is below 1. Double orphans are more likely to be disadvantaged compared with single orphans. The median school attendance ratio for double orphans is 0.87 for the entire region. Double orphans in eastern Africa have the least favourable ratio (0.72). Orphans have lower levels of education, irrespective of the country's attendance levels (Fig. 3). The countries with highest HIV prevalence levels include the countries with the highest school attendance rates. Therefore, for the continent as a whole, attendance levels among orphans are still fairly high in spite of disadvantages faced by orphans in each country. There is no clear pattern by type (maternal or paternal) or the sex of the orphan (male or female child).
A key health indicator to assess children's wellbeing is whether they are underweight (weight-for-age), here measured by comparing the proportion of orphans aged 12–59 months old and other children who fall below two minus standard deviations from the median weight-for-age of the NCHS/WHO reference population. As orphanhood increases with age there are relatively small numbers of children under 5 years of age who are orphaned, and all orphans (double, paternal and maternal) were combined for the analysis. There is little difference in the prevalence of being underweight between orphans and non-orphans under 5 years of age.
A third indicator of wellbeing is related to ‘child protection': the proportion of children working 20 and 40 h or more per week. Overall, levels differ per country. The countries with the lowest levels of school attendance have the highest levels of child labour. Furthermore, child labour levels are much higher in rural areas than in urban areas. There is, however, little difference in labour activities between orphans and non-orphans within these areas.
Overall, the population-based surveys, which cover 97% of the population in sub-Saharan Africa, indicate that on or about the year 2000, 9% of children under 15 years have lost at least one parent, including 1% who have lost both parents (weighted average for all countries). High levels of adult mortality resulted in high levels of orphanhood in the era before AIDS, but there is clear evidence of an increase in orphanhood in countries more severely affected by HIV/AIDS. The most dramatic rise is in the proportion of children who are double orphans. This contrasts sharply with most countries with low HIV prevalence, where a decrease in orphanhood prevalence is observed during the 1990s.
The estimates of the levels of orphanhood in household surveys have a number of limitations. Children outside of family care (i.e. street children and children in institutions) are not included in household surveys . In general, however, it appears that in most countries the proportion of children living outside of family care appears to be small. In addition, orphans aged 15–17 years are currently not included in household surveys, even though these children are also in need of care and support. In five countries, the survival status of at least one parent was unknown for more than 5% of children in the survey. Southern Africa has higher proportions of children for which the survival status of one or both parents was unknown by the head of the household in which the child resides (country median 3.1%). As children are least likely to live with their parent(s) in this sub-region, these higher levels of unknowns might be because absent parent(s) have been out of touch with their children, and current caretakers of the children are not sure about their survival status. If orphans are overrepresented in this group, findings in this assessment might be biased. Respondents may also misreport the survival status of the biological parent . This bias may go in either direction. Parents may report a non-biological child as their own, which would lead to an underestimation of the prevalence of orphanhood. On the other hand, parents, notably fathers, who have not been living with the family for a prolonged period of time may be reported as no longer alive if the survival status is unknown.
In almost all 40 countries, child fostering, caring for non-biological children, is common, although there is considerable regional variation. The prevailing childcare patterns also determine the orphan living and childcare arrangements. There are, however, a number of differences. Orphans more frequently live in female-headed households, in larger households with a less favourable dependency ratio, and in households in which the head of the household is considerably older Among orphans who have lost both parents or who are not living with surviving parents, grandparents are the most common caretakers. Few orphans were living without someone from their family and relatively very few child-headed households were found in the surveys. Households headed by children may be under-represented in households surveys because normally an adult is required to complete the household questionnaire. However, cohort studies in sub-Saharan Africa confirm the relatively low prevalence of child-headed households [8–10].
Recent epidemic trends based on the surveillance of pregnant women attending antenatal clinics clearly show that striking sub-regional patterns can be observed within sub-Saharan Africa and that the gap between the sub-regions appears to be widening . Southern Africa is much more severely affected by the epidemic than countries in eastern Africa, where prevalence is less than half of that in southern Africa and there is evidence of a modest decline, and western Africa where prevalence is roughly one-fifth of that in southern Africa and no rapid growth is occurring.
The high HIV prevalence countries in southern Africa are expected to have the largest increase in the proportion of children who are orphans. This is not only because HIV prevalence is much higher than elsewhere, but also because adult mortality was much lower before the AIDS epidemic than elsewhere in sub-Saharan Africa. Coping strategies may also differ considerably from elsewhere in the region. It may be that the high mobility, primarily caused by male outmigration to work in more industrialized areas of the southern African countries (Botswana, Lesotho, Namibia, and Swaziland) not only contributed to the high HIV prevalence rates, but also to high rates of child fostering and high levels of female-headed households. Coping mechanisms that rely on the extended family may be less resilient than elsewhere in the region.
Urban populations in all countries have been exposed to HIV for a longer period of time and their HIV prevalence has also been shown to be considerably higher than in rural populations. Rural populations, however, generally have higher fertility, which increases the numbers of orphans, but not the proportion. Migration may obscure the patterns as children may move after parental death, especially from urban to rural areas. Recognizing the increasing burden in the southern African countries, it is important to note that between and within countries in this region there are also significant differences in the way families and communities cope with orphans. In Malawi and Zambia orphans are more likely to be found in urban than in rural areas. In Botswana, Zimbabwe and Namibia they are more likely to reside in rural areas. A recent shift in moving orphans to rural areas has been reported by other studies in eastern and southern African countries [7,12]. Similar trends have been found in this review for Namibia and Zimbabwe. However, in neighboring countries such as Malawi and Zambia the opposite trend appears to be taking place. The death of the male head of household in a rural area could force the mother and children to the city in search of work or other forms of support.
Only a few aspects of the education and health status of orphans could be compared with those of other children. Orphans, especially double orphans are less likely to attend school. The differences between orphans and non-orphans are highest for countries with the lowest overall school attendance levels. Little impact was found on nutritional status, as has been reported in recent community studies [13,14]. Nutritional data on orphans aged 5–14 years would have been of greater interest, but no anthropometric data have been collected for this age group.
The participation of children in work to assist families in daily survival is not essentially a negative finding. Therefore, the review only looked at ‘extreme’ child labour (for this review defined as 40+ hours a week). In a number of countries, high proportions of children are involved in ‘extreme’ child labour, especially in rural areas. However, in the household surveys no significant differences in child work were found between orphans and non-orphans. A number of recent rapid assessments implemented by the International Labour Organization (ILO) found disproportionately high levels of orphans involved in extreme forms of child labour in sub-Saharan Africa (Zambia, Tanzania, Ethiopia, South Africa and Zimbabwe) [15–19]. However, there are limitations in measuring child labour through household surveys. A significant proportion of children involved in the worst forms of child labour probably do not live in family/household settings . Second, the questions in the surveys were addressed to caretakers of children and these may underreport the worst forms of child labour. Rapid assessments on the other hand do not work with probabilistic samples of children, and therefore more research on the relationship of child labour and orphanhood is needed.
In conclusion, this analysis shows that orphanhood is common in sub-Saharan Africa, irrespective of the AIDS epidemic. The epidemic has, however, caused rapid recent increases in the prevalence of orphanhood in many countries. Prevailing childcare patterns in many countries have dealt with large numbers of orphans in the past and to date there is no convincing evidence that this system is not absorbing the increase in orphans on a large scale. The burden of care for orphans (17% of households care for orphans) is spread wider than the actual prevalence of orphanhood (9% of children are orphaned). However, there is some evidence that children who lost one or both parents are as a group especially vulnerable as they live in households with less favourable demographic characteristics and have lower school attendance. As the brunt of HIV-associated adult mortality is likely to occur during the first decade of the new millennium, with or without treatment, careful monitoring through a multitude of methods is needed to be able to respond rapidly and effectively to the needs of orphans and families and communities caring for them. Increased access to antiretroviral treatment may help avert a rapid increase in orphanhood in the near future, especially if fertility among those on treatment is low. National surveys can play an important role in documenting short and long-term trends in orphanhood and care patterns.
© 2004 Lippincott Williams & Wilkins, Inc.