No published studies from sub-Saharan Africa have recorded child mortality and morbidity by maternal HIV status beyond age 3 years [1,2]. More information on the impact of orphanhood on child physical well-being is required, especially as the number of children orphaned by AIDS is projected to increase substantially in the period to 2010 . Survival and well-being of children of HIV-positive mothers will depend on their own HIV status and also on their parents’ health status.
If there is no intervention, approximately 35% of offspring of HIV-positive mothers become HIV positive themselves . Parental death may affect child well-being in various ways. The loss of an economically active adult, and costs associated with medical care and funerals, will reduce household income  and can reduce agricultural production . The emotional loss can have a negative psychological impact . Loss of a parent through AIDS may have a greater impact than death from other causes because the death of the other parent is likely to follow if they are also infected; in addition, HIV-positive individuals tend to have longer terminal illnesses than HIV-negative individuals . Death of a parent from AIDS may also stigmatize the child .
Mortality estimates in African studies for infants with HIV-positive mothers have ranged from 13 to 24%, and for children to age 2 years from 20 to 32% [1,2,8,10–14]. The longevity and cause of death of HIV-positive children vary by setting . In Rwanda 57% and in Uganda 68% of HIV-positive children died by age 5 years .
Findings on the impact of maternal mortality on child survival in relation to HIV are mixed. Numbers of maternal deaths are small in most studies, and while some studies have suggested an increase in mortality following the mother's death [12,16], others have found no effect [17,18].
As part of the Karonga Prevention Study in Malawi, the survival and physical well-being of children by maternal HIV status and parental mortality has been examined in a retrospective cohort study with a follow-up of over 10 years.
Karonga district is a rural area in northern Malawi. Most of the population are subsistence farmers. HIV prevalence in adults was 0.1% in the early 1980s, 2.0% in the late 1980s and around 10% in the late 1990s .
Two total population surveys were carried out in the 1980s as part of epidemiological studies of leprosy and tuberculosis [20,21]. In two areas of the district, dried blood spots were collected on filter papers and stored for serological studies of leprosy . Subsequently, with permission from the Malawi Health Sciences Research Committee, these specimens were tested for HIV .
A group of 197 HIV-positive adults were identified from the filter paper blood spots. As a comparison group, 396 HIV-negative individuals were chosen to match the HIV-positive individuals as far as possible for age, sex, area of residence at the time of the survey and family structure.
All 593 ‘index’ individuals, together with their spouses and offspring, were followed up in 1998–2000. Field staff, who were blind to the HIV status of the index individuals, first sought the families at their last known residence and traced them to new homes if necessary. Follow-up of the index individuals and spouses enabled identification of live-born offspring. All offspring were followed up, regardless of age, and the households of the offspring found living within the district were visited. An interview was conducted with the individual themselves, a guardian or an appropriate proxy. Interviewing was done by experienced staff, who were well known and trusted within the district. Adults were tested for HIV, after counselling and if consent was given, and post-test counselling was given to those who wanted to know their results. Informants were used for individuals reported to be alive but living outside the district, or those known to have died.
Data were collected on mortality and recent morbidity (including symptoms included in the World Health Organization (WHO) paediatric AIDS case definition for those aged < 15 years ). Anthropometric measurements were also recorded: height, weight, and mid-upper-arm-circumference. A verbal postmortem was filled for deaths, using different questionnaires for deaths in the neonatal period, in childhood or in adulthood. The verbal postmortems were independently reviewed by three physicians. If they could not give a cause of death, or if the reports were inconsistent, the cause was recorded as undetermined.
Data were double-entered using Foxpro and analysed using STATA 7.0 (Stata Corporation, College Station, Texas, USA). Poisson regression was used to estimate mortality rates, and Cox regression to estimate hazard ratios (HR) for death, according to HIV status of the mother, parental mortality and terminal illness (the latter defined as the 1-year period prior to death). Children were assessed as to whether they were at increased risk over the period beginning 1 year before and ending 1 year after the parent's death. Time was measured from birth. Explanatory variables that changed over the follow-up period were treated as time-varying covariates (e.g., parental death). EPINUT (version 2, EpiInfo version 6.04b) was used to calculate Z-scores of height-for-age and weight-for-height. By this convention, a child of median height will have a Z-score of 0, and a child below the 2.5% centile of the reference population will have a Z-score of less than −2. Anthropometric data were analysed using linear regression. All analyses took into account that outcomes among children born to the same mother are not independent.
The proportion of deaths attributable to HIV in children before their fifth birthday (the population attributable fraction, PAF) was calculated. The adjusted HRs from the analysis (for ages 1–30 days, 31–365 days, and 13–60 months) were used to estimate the excess mortality for children born to HIV-positive mothers. A 95% confidence interval (95% CI) for the population attributable fraction was calculated using bootstrapping .
Four groups of children were defined according to the HIV status of the mother at the time of the child's birth and evidence of subsequent seroconversion in the parents (Table 1). Group 1 had an HIV-positive mother. In group 2, the mother may have been HIV-positive. Group 3 were children where the mother was HIV-negative at the time of the child's birth but there was evidence of subsequent seroconversion in the mother or father. Group 4 were children of HIV-negative mothers. Group 4 also includes a subset of the children whose mother's HIV status was unknown at the time of the birth. A restriction to mothers who were certainly HIV-negative at the time of the child's birth might introduce bias: it would mean that, of the children born after the baseline filter paper collection, only those whose mothers were alive, still resident in the district, and willing to be HIV tested at the end of follow-up could be included. The separation of groups 3 and 4 allows assessment of whether HIV has an impact on child mortality and morbidity other than through the vertical transmission of HIV.
For the anthropometric analyses, children aged < 5 years and children aged 5–9 years were studied separately, for two reasons. First, a considerable proportion of the children born to HIV-positive mothers who were < 5 years of age when seen may themselves be HIV-positive, whereas few of the children aged > 5 years of age when seen will be HIV-positive. Second, stunting reflects the cumulative impact of malnutrition, so differences among groups of children might increase with age.
Ethical approval for the study was granted from the Malawi Health Sciences Research Committee and the Ethics Committee of the London School of Hygiene & Tropical Medicine.
Of the 593 index individuals, 582 were traced, identifying 2520 offspring, of whom 1141 could be assigned to groups 1–4. All but one of the 1379 excluded were born before filter paper collection; the one exception was a child whose mother was unknown. These 1141 children were born to 407 mothers (160 HIV-negative and 45 HIV-positive index mothers, and 202 spouses of index fathers). Eighteen children were born to parents who were both HIV-positive index individuals.
Information on these 1141 offspring was obtained from the index case or their spouse for 891, from the offspring themselves for 13, from a relative for 192, and from another or unspecified informant for 45. The mean age of the children when seen or last known about was 6.1 years (SD 4.3). Table 2 shows the numbers of offspring who were alive and seen, who were known to have died or who had left the district. For further analyses, 34 children said to be alive or to have died outside the district, but for whom the informant was not a relative, were excluded. One child was excluded as the date last known alive was unknown.
Table 3 shows mortality rates by HIV status of the mother and the child's age, and Table 4 shows neonatal, infant, under 5 and under 10 survival and the HRs for death. Children of HIV-positive mothers (group 1) had the highest mortality rates at all ages. Compared with those with HIV-negative parents (group 4), the HR for death was increased 3.3 times for children aged 31–365 days and by five times for those aged 1–5 years. The HR values for groups 2 and 3 were slightly increased compared with group 4. There was only one death in the group with an HIV-positive mother at ages 5–10 years.
Under-5 mortality decreased with increasing age of the mother, was lower in those whose fathers were traders or salaried workers compared with subsistence farmers or casual labourers, decreased with increased levels of maternal education, was slightly lower in those born in later years and was highest in the far north of the district. After adjustment for these factors, the impact on survival of having an HIV-positive mother increased slightly (Table 4). There was no evidence that the effect of HIV status of the mother varied depending on these risk factors.
Restriction of group 4 to children born after, or ≤ 6 months before, the baseline filter paper collection (the restriction made for children born to HIV-positive mothers), and restriction of group 4 to children of mothers who were certainly HIV-negative at the time of the child's birth, made little difference to the estimates of the HRs (not shown). Inclusion of the 34 children said to be alive or to have died outside the district, but for whom the informant was not a relative, also made little difference to the results.
Nineteen children of HIV-positive mothers and 15 children of HIV-negative mothers experienced the death or terminal illness of their mother before age 10 years. The figure of 19 appears low and reflects the fact that 18/88 (20%) of children born to HIV-positive mothers died more than 1 year before their mother's death. Among infants born to HIV-positive mothers, mortality was higher among those whose mother had died or who was terminally ill (three deaths among eight children) compared with those whose mother was alive and not terminally ill (HR 2.7, 95% CI 0.8–9.2). Mortality between 1 and 5 years was also increased slightly for those who experienced their mother's death or terminal illness (two deaths among 12 children; HR 1.7, 95% CI 0.4–7.6). There were no deaths among children born to HIV-negative mothers who died or were terminally ill, but only one such terminal illness occurred during the first year of the child's life.
Eight children of HIV-positive mothers and 27 children of HIV-negative mothers experienced the death or terminal illness of their father before age 10 years. Among infants born to HIV-positive mothers, mortality was 25% for children whose father died or was terminally ill (one death among four children) compared with 27% for children whose father was alive and not terminally ill. The corresponding percentages for those born to HIV-negative mothers were 0% (no deaths in four children) and 11%. Five children born to an HIV-positive mother and 19 children born to an HIV-negative mother were aged 1–5 years during the period from 1 year prior to 1 year after their father's death, and none died during that period. Only one child with an HIV-positive mother experienced the death of both parents.
Of the childhood deaths, 54% (114/212) were caused by acute febrile illnesses; this proportion did not vary by maternal HIV status (P = 0.8). A verbal postmortem result classified as AIDS/TB–AIDS was found for 16% (6/38) of dead children in group 1, 15% (4/26) in group 2, 0% (0/28) in group 3 and 4% (5/118) in group 4 (P = 0.03 comparing group 1 and group 4).
Given a prevalence of HIV infection in pregnant women of 10%, 18% (95% CI, 10–26) of deaths in children before their fifth birthday can be attributed to HIV in this population.
None of the children in the study who were seen in the follow-up met the WHO Pediatric AIDS Case Definition  and the symptom review did not reveal a high prevalence of morbidity among surviving children. Children of HIV-positive mothers were more likely to be reported as ‘not well’ at the time of interview [group 1, 3/32 (9%); group 2, 9/78 (12%); group 3, 3/89 (3%); group 4, 24/554 (4%); P = 0.03], but this was not reinforced by specific symptoms or reported health care usage.
Children aged < 10 years were generally stunted [Z-scores were less than −2 for 46% (238/512) for height-for-age]. They were, on average, wasted, although only 3% (15/511) had a Z-score of less than −2 for weight-for-height. The proportion of children aged 10–14 years whose body mass index was below the 15th centile was 41% (109/263) . All indices were normally distributed.
Differences in anthropometric measurements between children in groups 1 and 4, and differences by orphanhood, were small and not statistically significant (Table 5; results for body mass index not shown), both on crude analysis and after adjustment for other risk factors. There was no evidence that the impact of the mother's HIV status or orphanhood varied depending on other examined risk factors, although the power of these analyses was low.
It was not possible to compare double, maternal and paternal orphans because of sample size constraints. Most orphanhood was from paternal mortality: of orphans aged < 5 years, one was a double orphan, three were maternal and eight were paternal orphans. The corresponding figures for children aged 5–9 years were four, four and 34.
Under-5 child mortality is much higher in children born to HIV-positive mothers than in those born to HIV-negative mothers. Assuming around 35% vertical transmission , and given the degree of uncertainty in our mortality estimates, our findings are consistent with the increase in mortality being largely confined to children who were themselves HIV infected. Other studies have found that HIV-negative children of HIV-positive mothers have similar mortality rates to children of HIV-negative mothers [1,16,26]. However, the borderline evidence for an increased mortality rate in those born to HIV-negative mothers for whom there was later evidence of HIV positivity in a parent (group 3) suggests that parental HIV infection may have an indirect effect on child mortality.
The lack of evidence for higher mortality rates after 5 years among children born to HIV-positive mothers compared with those born to HIV-negative mothers probably reflects the low power of the analysis in this age group, since some HIV-positive children are expected to survive beyond 5 years .
The mortality rates observed at 1–2 years were higher than those reported in most other studies [1,2,10–14,27]. This reflects the relatively high child mortality rates reported in Malawi compared with other countries (State of the World's Children, UNICEF, http://www.unicef.org).
HIV seropositivity in this population at the time of the filter paper collection was associated with higher levels of education and with occupations other than subsistence farming . After adjustment for these and other sociodemographic factors, the HR for the association of maternal HIV status and child mortality was slightly increased, indicating that the true impact of HIV was larger than the crude mortality rates suggest. Most other studies have not adjusted for socioeconomic status and so may have underestimated the impact of HIV [8,10,12,16,28,29].
Maternal mortality was associated with child mortality among HIV-positive mothers but not among HIV-negative mothers. The association among HIV-positive mothers may reflect the fact that mothers with more advanced disease are more likely to transmit HIV to their children . Furthermore, among infants with vertically acquired HIV infection, the rate of progression depends on the severity of disease in the mother at the time of delivery [27,30]. Both severe illness and death of the mother could be expected to increase infant and child mortality through problems in child care , but this needs to be distinguished from the increase in risk of HIV, and of rapid progression if HIV positive. The extended family has almost certainly helped to mitigate the impact of parental death [32,33].
It has been suggested that paternal mortality could have greater socioeconomic impact than maternal mortality . This may be influenced by the traditional practice of widow-inheritance by the deceased husband's brother in this population. Neither in our study, in a patrilineal society, nor in another Malawian study in a matrilineal society  was there evidence that the father's death influenced child survival, but both analyses had low power.
Our findings suggest that, in this part of Africa, verbal postmortems are neither sensitive nor specific in identifying whether a childhood death is caused by HIV. This is consistent with two other studies in Malawi [16,28].
We found little evidence of differences in child morbidity according to either mother's HIV status or orphanhood. This is surprising. Children of parents who have been ill or who have died from AIDS have been found to be at high risk of malnutrition in several African countries [5,6,34]. However, these studies also show that the risks are setting dependent.
Our findings are consistent with a study in Lusaka, Zambia that compared children living in households known to be affected by HIV/AIDS with a control group living in households that were not known to be affected by HIV/AIDS; this study found no difference in stunting or weight . They are also consistent with a study of children aged < 3 years in Blantyre, Malawi, which found no differences in morbidity between HIV-negative children of HIV-positive mothers and children of HIV-negative mothers . A study of children aged < 5 years old in Kagera, Tanzania found no evidence that adult mortality affected child wasting  but suggested that orphanhood was associated with stunting, and that the effect of paternal orphanhood on stunting was greater in poorer households.
The study design has several important strengths: identification of ‘index’ individuals was based on a total population survey, HIV-positive and HIV-negative index individuals were closely matched, and the follow-up period is long. However, the retrospective nature of the study has drawbacks. Maternal HIV status at the time of delivery was unknown for a considerable proportion of mothers, and the study relied on recall for children's birth and death dates. Recall was probably less accurate when the child and/or the mother had died or was absent. After controlling for mother's age when last ‘known about', the total number of children reported for mothers living outside Karonga was lower than for mothers living in Karonga. However, the proportion of children who had died was similar in the two groups, suggesting there was no selective under-reporting of dead children for mothers no longer resident in the district.
We were unable to assess the impact of parental morbidity on child mortality and morbidity, since individuals were seen only at baseline and at the 10-year follow-up. However, the lack of evidence for an impact of parental death on child mortality and physical well-being suggests that parental morbidity is unlikely to have had a substantial effect.
The correlation between weight-for-height and mid-upper-arm-circumference in these data (0.57 and 0.68 for children aged < 5 years and 5–9 years, respectively) was moderately strong, suggesting that the anthropometric data are reliable and that our findings have internal validity. However, children who are no longer resident in the district may have been more severely affected by HIV/AIDS, in which case the findings will have underestimated the impact on morbidity.
The HIV status of the children in our study was not known, so the direct (vertical transmission of HIV infection) and indirect impacts of HIV in the mother could not be unambiguously distinguished. However, knowledge of child mortality according to maternal HIV status, rather than child HIV status, is valuable for projections of child mortality in Africa. This is because maternal HIV prevalence is widely available through antenatal clinic surveillance, whereas children's HIV status is usually unknown. Estimates of child mortality based on maternal HIV status now will provide a baseline against which the impact of interventions can be judged.
Mortality among children under 5 years of age is much higher in children born to HIV-positive mothers than in those born to HIV-negative mothers, and approximately 18% of mortality in this population is attributable to HIV. Most of the excess is attributable to vertical transmission of HIV.
The lack of evidence for excess morbidity in surviving children born to HIV-positive mothers suggests that, at least in terms of physical well-being and for children who have remained in the district, the extended family has not discriminated against children whose parents have been ill or have died from HIV/AIDS. However, it would be wrong to infer that the HIV epidemic has had little impact on surviving children's physical well-being. Traditional coping mechanisms based on the extended family appear to have diluted the impact on individuals, but the impact at the level of the community could still be considerable. It is also likely that pressures will increase in the near future, as the HIV epidemic matures and adult mortality rises .
We thank the Government of the Republic of Malawi for their interest in and support of the project and the Malawi Health Sciences Research Committee of the Malawi Ministry of Health and Population for permission to conduct the study and present these data. We thank Dr Jörg Pönnighaus for organizing and implementing the original field surveys that made this study possible. We thank Drs Stephen Graham, Frank Mwaungulu and Nicola Hargreaves for reviewing the verbal postmortem data. We thank Prof Andrew Prentice for comments on an earlier version.
Sponsorship: Until 1996 the Karonga Prevention Study was funded primarily by LEPRA (the British Leprosy Relief Association) and ILEP (the International Federation of Anti-Leprosy Organizations) with contributions from the WHO/UNDP/World Bank Special Programme for Research and Training in Tropical Diseases. Since 1996, the Wellcome Trust has been the principal funder. BN was supported by the British High Commission in Malawi. JRG was partially supported by the British Department for International Development.
1.Taha TE, Canner JK, Chiphangwi JD, Dallabetta GA, Yang LP, Mtimavalye LA, et al.The effect of human immunodeficiency virus infection on birthweight, and infant and child mortality in urban Malawi.Int J Epidemiol
2.Ryder RW, Nsa W, Hassig SE, Behets F, Rayfield M, Ekungola B, et al.Perinatal transmission of the human immunodeficiency virus type 1 to infants of seropositive women in Zaire.N Engl J Med
3.USAID. Children on the Brink.
Washington DC: USAID; 2002.
4.Schwartlander B, Stanecki KA, Brown T, Way PO, Monasch R, Chin J, et al.Country-specific estimates and models of HIV and AIDS: methods and limitations.AIDS
4.UNAIDS. A Review of Household and Community Responses to the HIV/AIDS Epidemic in the Rural Areas of Sub-Saharan Africa.
Geneva: UNAIDS; 1999:99.39E.
6.FAO. The Effects of HIV/AIDS on Farming Systems in Eastern Africa.
Rome: FAO Farm Management and Production Economics Service; 1995.
4.Poulter C. Psychological and Physical Needs Profile of Families Living with HIV/AIDS in Lusaka, Zambia.
Lusaka: UNICEF; 1997:Research Brief No. 2.
8.Urassa M, Boerma JT, Isingo R, Ngalula J, Ng'weshemi J, Mwaluko G, et al.The impact of HIV/AIDS on mortality and household morbidity in rural Tanzania.AIDS
8.Topouzis D, Hemrich G. The Socio-economic Impact of HIV and AIDS on Rural Families in Uganda: An Emphasis on Youth.
New York: UNDP HIV
and Development Programme, 1994: study paper 2.
10.Sewankambo NK, Wawer MJ, Gray RH, Serwadda D, Li C, Stallings RY, et al.Demographic impact of HIV infection in rural Rakai district, Uganda: results of a population-based cohort study.AIDS
11.Nduati R, John G, Mbori-ngacha D, Richardson B, Overbaugh J, Mwatha A, et al.Effect of breastfeeding and formula feeding on transmission of HIV-1: a randomized clinical trial.J Am Med Assoc
12.Ota MO, O'Donovan D, Alabi AS, Milligan P, Yamuah LK, N'Gom PT, et al.Maternal HIV-1 and HIV-2 infection and child survival in The Gambia.AIDS
13.Zijenah L, Mbizvo MT, Kasule J, Nathoo K, Munjoma M, Mahomed K, et al.Mortality in the first 2 years among infants born to human immunodeficiency virus-infected women in Harare, Zimbabwe.J Infect Dis
14.Sewankambo NK, Gray RH, Ahmad S, Serwadda D, Wabwire-Mangen F, Nalugoda F, et al.Mortality associated with HIV infection in rural Rakai District, Uganda.Aids
15.Lepage P, Spira R, Kalibala S, Pillay K, Giaquinto C, Castetbon K. Care of human immunodeficiency virus-infected children in developing countries. International Working Group on Mother-to-Child Transmission of HIV.Pediatr Infect Dis J
16.Taha TE, Kumwenda NI, Broadhead RL, Hoover DR, Graham SM, van der Hoven L, et al.Mortality after the first year of life among human immunodeficiency virus type 1-infected and uninfected children.Pediatr Infect Dis J
17.Ryder RW, Kamenga M, Nkusu M, Batter V, Heyward WL. AIDS orphans in Kinshasa, Zaire: incidence and socioeconomic consequences.AIDS
18.Taha TE, Kumwenda NI, Hoover DR, Biggar RJ, Broadhead RL, Cassol S, et al.Association of HIV-1 load and CD4 lymphocyte count with mortality among untreated African children over one year of age.AIDS
19.Glynn JR, Ponnighaus JM, Crampin AC, Sibande F, Sichali L, Nkhosa P, et al.The development of the HIV epidemic in Karonga District, Malawi.AIDS
20.Pönnighaus JM, Fine PEM, Bliss L, Sliney IJ, Bradley DJ, Rees RJW. The Lepra Evaluation Project (LEP) an epidemiological study of leprosy in Northern Malawi. I: Methods.Lepr Rev
21.Fine PEM, Pönnighaus JM. Leprosy in Malawi. 2. Background, design and prospects of the Karonga Prevention Trial, a leprosy vaccine trial in northern Malawi.Trans R Soc Trop Med Hyg
22.Fine PEM, Pönnighaus JM, Burgess P, Clarkson JA, Draper CC. Seroepidemiological studies of leprosy in northern Malawi based on an enzyme-linked immunosorbent assay using synthetic glycoconjugate antigen.Int J Lepr
23.World Health Organization. Acquired immunodeficiency syndrome (AIDS). WHO/CDC case definition for AIDS.Weekly Epidemiolog Rec
24.Efron B, Tibshirani R. An Introduction to the Bootstrap.
New York: Chapman & Hall; 1993.
25.World Health Organization Expert Committee on Physical Status. The Use and Interpretation of Anthropometry
. Technical Report Series,
No. 854. Geneva: WHO; 1995.
26.Spira R, Lepage P, Msellati P, van de Perre P, Leroy V, Simonon A, et al.Natural history of human immunodeficiency virus type 1 infection in children: a five-year prospective study in Rwanda. Mother-to-Child HIV-1 Transmission Study Group.Pediatrics
27.Dabis F, Elenga N, Meda N, et al.18-month mortality and perinatal exposure to zidovudine in West Africa.AIDS
28.Bloland PB, Wirima JJ, Steketee RW, Chilima B, Hightower A, Breman JG. Maternal HIV infection and infant mortality in Malawi: evidence for increased mortality due to placental malaria infection.AIDS
29.Ryder RW, Nsuami M, Nsa W, Kamenga M, Badi N, Utshudi M, et al.Mortality in HIV-1-seropositive women, their spouses and their newly born children during 36 months of follow-up in Kinshasa, Zaire.AIDS
30.Blanche S, Mayaux MJ, Rouzioux C, Teglas JP, Firtion G, Monpoux F, et al.Relation of the course of HIV infection in children to the severity of the disease in their mothers at delivery.N Engl J Med
31.Taha TE, Miotti P, Liomba G, Dallabetta G, Chiphangwi J. HIV, maternal death and child survival in Africa.AIDS
32.Foster G, Williamson J. A review of current literature on the impact of HIV/AIDS on children in sub-Saharan Africa.AIDS
33.Foster G, Makufa C, Drew R, Kralovec E. Factors leading to the establishment of child-headed households: the case of Zimbabwe.Health Transit Rev
33.Nduati R, Muita JW, Olenja J, Muiva E, Myuthani N, Manguyu F. A survey of orphaned children in Kibera Urban slum, Nairobi.IX International Conference on AIDS
. Berlin, June 1993 [abstract WS-D26-4].
35.Taha TE, Graham SM, Kumwenda NI, Broadhead RL, Hoover DR, Markakis D, et al.Morbidity among human immunodeficiency virus-1-infected and -uninfected African children.Pediatrics
36.Ainsworth M, Semali I. The Impact of Adult Deaths on Children's Health in Northwestern Tanzania. Working Paper
No. 2266. Washington DC: World Bank; 2000.