Vertically acquired HIV infection directly contributes to child mortality in high HIV prevalence areas [1,2], and maternal HIV infection and survival indirectly [3–6]. HIV impact is most pronounced in sub-Saharan Africa, with over 90% of all HIV-infected children, and HIV prevalence among women of childbearing age reaching 35% and above in Southern Africa [7–9]. Recent evidence suggests the achievement of Millenium Development Goal (MDG) 4 regarding reducing under-5 mortality by two-thirds is currently off-track in South Africa with a continued rise in child mortality rates, a trend attributed to the burden of HIV .
The effect of paediatric HIV infection on overall child mortality has been poorly quantified here, but without early identification of infection and appropriate treatment, about a third of vertically infected children will die before age 1 year, and half by 2 years [6,11]. Antiretroviral therapy (ART) substantially delays disease progression , but the impact of HIV treatment programmes on overall child mortality at population level remains unquantified.
Prevention of mother-to-child transmission (PMTCT), currently mostly through single-dose nevirapine (sdNVP) to HIV-infected mothers and their children,  can potentially affect child mortality through reducing numbers of vertically infected children. In South Africa, sdNVP reduced the risk of perinatal HIV transmission from an estimated 19%  to about 9% .
When HIV-infected women of childbearing age  proceed to have children, all their children, irrespective of their own HIV infection status, are at increased risk of dying, particularly in the first 2 years of life, both before and after maternal death [4,6,17]. Whether, on a population basis, deaths in infected mothers are prevented by ART, and whether this affects population child mortality rates remains unclear.
We investigated early life mortality in a largely rural population of KwaZulu-Natal, South Africa, where HIV prevalence in adults reaches 23% overall, and 50% among 25–29-year females [16,18,19]. We analysed temporal and spatial associations with roll-out of a PMTCT and HIV treatment programme, and maternal HIV to estimate the direct effect on mortality of a reduction in the number of vertically infected children, and an indirect treatment effect through the survival of their HIV-infected mothers.
The Africa Centre Demographic Surveillance Area (DSA), in the Umkhanyakude district of northern KwaZulu-Natal, covers 438 km2 and approximately 90 000 Zulu-speaking people in 11 000 households. Predominantly rural, with an urban township and informal peri-urban settlements, the principal sources of income are waged employment and state pensions .
Data came from the Africa Centre Demographic Information System (ACDIS), with, since 2000, twice yearly household visits. Questionnaires were administered to key household informants recording events including births, deaths, migrations and pregnancies. Fieldworkers have mapped all homesteads and facilities using differential global position systems . Since 2000, overall fertility was steady in ACDIS at about three children per woman , with an average of 2200 live-births per year . Further, since 2003, an annual population-based HIV surveillance, nested within ACDIS, documented the HIV status of consenting resident women aged 15–49 years and men aged 15–54 years . The consent rate for mothers included in this surveillance was about 55% . Written informed consent allowing use of data was obtained from all participants, and ethical approval from the Biomedical Research Ethics Committee of the University of KwaZulu-Natal.
The Hlabisa HIV Treatment and Care Programme, initiated in late 2004, is a partnership between the Department of Health and Africa Centre, providing free services at 17 local government primary healthcare clinics (PHCs) . The number of women on treatment by the end of December 2008 was 4688 (68% of all adults on treatment), of whom 40% live in the DSA . A PMTCT programme was introduced in 2001 and established at all clinics by the end of 2002 , sdNVP was provided to all HIV-infected women and their infants  with counselling on infant feeding options: HIV-uninfected women were counselled to exclusively breastfeed for the first 6 months of life, with sustained breastfeeding to at least 2 years. HIV-infected women were given the option of either replacement feeding or exclusive breast feeding for the first 6 months of life . The sdNVP uptake among babies born to HIV-positive women in the district was estimated at 45% in 2007 . At about the same time an intervention mother-to-child transmission cohort study, the Vertical Transmission Study (VTS), began enrolment in the five largest clinics in the area . VTS women were supported to optimally feed their infants in line with international policies as described above [14,30].
All 13 583 children in ACDIS born between 1 January 2000 and 31 December 2006 (2 years before end of observation to allow at least 2 years of observation) were included.
Improved water and sanitation includes borehole and pipes for water, and ventilated-improved-pit latrines and flush for sanitation; all else were classified as unimproved. The household assets included items used for production or consumption, beds, bicycles, tables, telephones, television sets, sewing machines, block makers, wheelbarrows, tractors, cattle and other livestock. Birth season was classified as dry winter (April–September) and wet summer (October–March) . History of foetal death was defined as a stillbirth or miscarriage prior to birth of index child, classified as yes or no ; previous child death was history of child death (age <5 years) prior to birth of index child, classified as yes or no. Maternal HIV status from surveillance was classified into known negative, known positive, never known, and ambiguous (unknown but HIV-negative before pregnancy, or unknown but positive after delivery) . PMTCT availability, based on infant sdNVP, was classified as none (children born in a health facility pre-PMTCT availability), PMTCT only (children born in hospital with sdNVP), PMTCT and VTS (children born in clinics with intensive feeding support and sdNVP) and children born at home (unlikely to have received sdNVP). ART availability was classified as: ‘No ART’ for children born before roll-out of the treatment programme, ‘less than 2 years of ART’ for those born within 2 years from roll-out of the HIV treatment programme and ‘at least 2 years of ART’ for those born 2 or more years later. By end 2007 less than 8% of people in the treatment programme were children; few were infants below age 2 years .
We used a geographical information system (GIS) to allocate each child to one of six PHC catchments within the study area . The clinic catchments were derived using a GIS methodology described in detail elsewhere , which showed that 91% of households regularly used the most geographically accessible clinic. We then constructed thematic maps to investigate the spatiotemporal changes in under-two child mortality rates (U2MR; by clinic catchment).
U2MR were based on deaths per 1000 live-births per year; survival time was estimated by the Kaplan–Meier method; Weibull regression was used to investigate simultaneously the effect of demographic, environmental and maternal factors on childhood survival rates [34,35]. The Cox proportional hazard model was inappropriate because the proportional-hazards assumption, based on Schoenfeld residuals, was violated when including the variables relating to PMTCT and ART . For variables with missing data, we included ‘unknown’ as a category in the model to retain overall denominators. For each child, the observation time t was taken as the time from birth until death, the date of the second birthday, the end of observation period (31 December 2008) or date last seen, whichever came first. The 2-year age cut-off was chosen because mortality is highest in the first 2 years of life, reaching over 50% among HIV-infected, untreated children , and because effect of maternal survival is most acute in the early years . Results for separate models for neonates, postneonates, and children aged 1–2 years are also reported. For neonatal mortality modelling, the observation time t was truncated at the 28th day of observation. For postneonates modelling, all children were included from their 28th day of observation to their first birthday; for the early childhood modelling, children were included from 1 year of age until the time of their second birthday. Multivariate models included all variables with a statistically significant association univariately. The goodness of fit of the final models was assessed using the likelihood-based Akaike information criteria (AIC) . The estimated mortality rates by ART period were obtained from the predicted survivor function of the Weibull regression model. Predictions were calculated stratifying by the levels of the covariate (ART period) with respect to the reference value of the other covariates included in the model.
Further, taking 2005 as example, we modelled the potential contribution of PMTCT and the HIV treatment and care programmes to U2MR (Fig. 1). This model is based on the documented association between maternal survival and child mortality and accounts for the differentials in child survival by maternal HIV status [3–6]. Child mortality rates are then adjusted based on the potential effects that PMTCT  and provision of ART treatment for both mother and child  would be expected to have had. Assumptions in this modelling: antenatal HIV prevalence of 40% , maternal mortality rates among HIV-infected and uninfected mothers of 45.3 and 5.9 per 1000, respectively , three-quarters reduction in early infant mortality due to early ART , and a quarter of children infected with HIV in the absence of PMTCT with sdNVP reducing peripartum transmission from about 20 to 14% (an approximately 5% reduction) [38,39].
We used Stata (Version 11.0, Stata Corporation, College Station, Texas, USA), which allows the analysis of continuous and time-dependent covariates.
Eight hundred and forty-eight children died before 2 years of age from 13 583 live-births (Table 1). A substantial number of infants (10.6%) were born at home and therefore unlikely to have had sdNVP. Median age of mothers was 24 years (interquartile range 20–30); most had at least some primary education; the higher proportion of women with some primary education among mothers of children who died is likely to reflect the association between education and HIV infection . HIV status was recorded within the surveillance for 7479 (55%) of mothers, of whom 702 (9.4%) were tested HIV-positive (note this is not an antenatal prevalence) before or during pregnancy and 4381 (58.6%) were tested HIV-negative during or after pregnancy. Of the 2396 (32%) women with ambiguous HIV status a substantial number are likely to have been HIV-positive during the index pregnancy in this high incidence setting.
Numbers of deaths under 2 years of age declined from a peak 152, in children born between 2001 and 2006, to 72 in 2006; the U2MR halved from a peak of 86.3 to 44.1 deaths per thousand live-births (Fig. 2). Crude mortality rates in neonatal and early childhood ages remained relatively stable; most of the decline in U2MR was in the postneonatal period. Between roll-out of the PMTCT programme (2001) and the HIV treatment programme (2004) U2MR declined by 36% from a peak of 86.3 to 55.3 deaths per thousand live-births, followed by a further 20% decline after HIV treatment roll-out (Fig. 2).
An estimated 5.4% of children would have died before age 2 years during the post-ART (2004–2007) compared to 8.6% during the pre-ART period (2000–2003) (P ≤ 0.001) (Fig. 3).
Factors associated with under-2 mortality
Total exposure time for the 13 583 children was 22 288 years; mean follow-up was 1.6 years; there were 848 deaths (Table 1). Multiple births infants were more likely to have died, as were infants born in poorer households, those born at home, those with a maternal previous history of child death and those born when PMTCT and ART were not available (Table 1). In Weibull regression analysis the inclusion of PMTCT and ART variables did not affect the hazard for the other variables (data not shown). In multivariable analysis, first born children had a 27% significantly increased risk of dying, children born to mothers with at least primary education had a 21% significantly increased risk of dying, and a history of previous child death increased more than three-fold the risk of dying for the index child (Table 2). Further, infants born at home had a 49% increased risk of dying before 2 years; those born in nonurban areas had a 67% increased risk of death than those born in urban areas, whereas twins had an 85% increased risk of dying.
Access to water, sanitation and electricity, household assets, and distance to the nearest health facility were not significantly associated with under-2 mortality. Children born in the wet summer months were 16% more likely to die than those born in the dry winter months. Children born to a mother with a positive HIV test result in surveillance were more than four times as likely to have died by age 2 years as children of mothers tested HIV-negative; the risk was also increased three-fold for infants born to mothers with an ambiguous or unknown HIV status.
The maternal death variable was not introduced in the models because its effect was captured by the ART period variable. We fitted models with ART period variable alone, maternal vital status alone and with both variables present. The AIC for the model with ART period alone was 8603.72, with maternal vital status alone 8626.36, and for both ART period and maternal vital status present 8693.18. The first model was marginally better with smaller value of AIC. We fitted interaction terms in the Weibull regression models, none were statistically significant.
Availability of PMTCT and ART in public health programmes and U2MR
In 2004, pre-ART, the effect of PMTCT on U2MR was evident in catchment areas of five clinics providing PMTCT and infant feeding counselling; the decline in mortality averaged 33% (range 19–51%) relative to baseline (2000) (Fig. 4). In 2006, 2 years into the maternal HIV treatment programme, a further decline in U2MR was observed in catchment areas of four clinics providing both ART and PMTCT, the decline rose to an average of 52% (range 26–70%) relative to 2000. The largest decline in absolute mortality was observed in areas with higher HIV prevalence.
In Weibull regression analysis, the effect of the PMTCT and ART programme on under-2 mortality was clear. Allowing all other variables, children born at home and unlikely to have received sdNVP had a 35% increased risk of dying than children born in a health facility where sdNVP was available. Children who accessed clinics where sdNVP with or without VTS operated were at lower mortality risk compared to the reference group, although this did not reach statistical significance. Children born within a 2-year period after implementation of the HIV treatment programme had an independent 34% reduced risk of dying before 2 years of age, whereas those born 2 years or more after the programme had a 54% reduced risk of dying.
Further analyses, with separate models for neonates, postneonates, and children aged 1–2 years, suggested that the PMTCT availability association applied mainly to the neonatal and postneonatal periods, whereas the ART availability refers mainly to the postneonatal period only (data not shown).
Taking the situation in 2005 only, as an example of the impact the PMTCT and ART programme could have had on U2MR, we estimated that the availability of PMTCT and ART in public health programmes would have explained 8 and 31% of the decline in U2MR from 2000, respectively (Fig. 1).
We present U2MR among children born in a rural South African surveillance site between 2000 and 2006, and show relatively stable crude neonatal and 1–2 year mortality but significantly decreasing postneonatal mortality. Overall U2MR substantially declined from about 2001 onwards, despite continuing high HIV incidence and prevalence [18–20]. It is known that the contribution of HIV infection to child mortality, estimated from national data, is greater in countries with a lower pre-HIV mortality . Thus, in a country like Botswana, with stable, relatively low, child mortality pre-HIV (60 per thousand in the late 1980s), but with antenatal HIV prevalence reaching over 35%, child mortality doubled in the 1990s. By contrast, in Malawi, where child mortality was nearly 250 per thousand before the epidemic, estimated child mortality levels continued to fall throughout the 1990s, although HIV prevalence reached nearly 15%. In Botswana, HIV-attributable mortality in 2000 was over 60%, in Malawi approximately 10% .
We show a 33% decline in U2MR following the introduction of a PMTCT programme, offering sdNVP with infant feeding counselling, and a further 52% decline after implementation of a maternal HIV treatment programme. Modelling the effect of these programmes, with informed, locally appropriate assumptions, we showed the decline in U2MR would partly be attributed to PMTCT (8%) but largely to the HIV treatment and care programme (31%). We previously estimated that the number of infants needing treatment, in 2007, in the Hlabisa sub-district would be 49 . However, only two infants were on ART treatment, suggesting that less than 5% of infants requiring ART had initiated treatment; hence any treatment-associated U2MR decline in this population would largely be attributable to maternal rather than paediatric ART treatment. In Weibull regression analysis, assuming a constant risk which could increase or decrease monotonically over time , we find that availability of PMTCT reduced mortality risk by 15% compared to ART: less than 2 years and at least 2 after roll-out by 34 and 55%, respectively.
Vertically acquired HIV infection directly affects the risk of mortality [3,17]. The limited effect of PMTCT here needs to be considered in the light of available prophylaxis (sdNVP only), although with successful interventions in place, including antiretroviral prophylaxis, elective caesarean section and formula feeding, the rate of mother-to-child HIV transmission can be reduced to under 2% ; in most developing countries, successful interventions are still not widely accessible nor feasible , and the number of vertically infected children is unlikely to decrease substantially .
Risk factors for U2MR included those previously associated with mortality: multipregnancy , birth order , delivery location, area of birth [26,43] history of child death, maternal age , education  and HIV status [4,6]. A higher incidence of diarrhoea has been observed in South Africa during the wet summer, which was associated with increased mortality for infants in our study .
Maternal survival is critical for the health of young children. Previous studies have shown that allowing for PMTCT and other variables, children born to HIV-infected, compared to HIV-uninfected, mothers were three times more likely to die; the year before and following a mother's death were particularly high risk [6,7,45]. sdNVP is estimated to reduce the transmission rates from about 20 to 14%, thus assuming a 20% HIV prevalence, sdNVP would potentially avoid infection in 1 out of 20 infants born to HIV-infected mothers, and could thus also prevent death associated with vertically acquired infection (however, these uninfected children would still be at risk of dying if their mother died). This would thus be unlikely to result in a dramatic reduction in mortality compared to maternal ART, which, by keeping infected mothers alive, has an effect on both the infected and uninfected children . Therefore our finding of a decline in early child mortality is likely due to the impact of ART on adult health rather than of PMTCT.
We have no data directly linking receipt of PMTCT and ART to women and children; temporal associations were based on known availability of PMTCT and ART at PHC clinics. The lack of direct link to PMTCT  may have under-estimated the coverage of sdNVP, but our assumption that infants born at home would have been unlikely to have received sdNVP seems reasonable. We also have no HIV data on children. However, all children born to HIV-infected women, irrespective of their own HIV infections status, are at increased risk of dying particularly in the first 2 years of life , and few children under 2 years of age accessed the HIV treatment programme.
Even with a modestly functioning PMTCT and HIV treatment programme, benefits for survival of young children become quickly evident. With more comprehensive treatment coverage, and more effective PMTCT including zidovudine in addition to sdNVP (since March 2008), further significant reductions in early life mortality are expected. These findings are important for policy makers and service delivery and highlight the urgency in scaling-up HIV treatment and PMTCT programmes if MDG4 is to be realised.
We are grateful to the fieldworkers and supervisors for their excellent work in the Demographic Information System, Colin Newell for assisting in data management, the Africa Centre community for their participation in the survey and the Department of Health for their partnership in the HIV treatment and care programme.
The Wellcome Trust supported this study through the Africa Centre Demographic Information System grant (GR082384/Z/07/Z).
Authors' contributions: M.-L.N. originally suggested the idea of investigating temporal associations of PMTCT and ART with early life mortality and provided oversight of analysis and writing. J.N. was responsible for overall statistical analysis and writing of the paper. F.T. was responsible for spatial analysis using GIS. A.J.H. was responsible for data extraction. R. Bland gave constructive comments during the analysis and writing of the paper and all authors substantially contributed to, and approved, the final manuscript.
1. UNAIDS/WHO (2007). AIDS epidemic update. Geneva: UNAIDS.
2. Walker N, Schwartlander B, Bryce J. Meeting international goals in child survival and HIV/AIDS. Lancet 2002; 360:284–289.
3. Brahmbhatt H, Kigozi G, Wabwire-Mangen F, Wawer M, Gray R. Two year survival of HIV-positive and HIV-negative children born to HIV-positive and HIV-negative mothers; evidence from Rakai, Uganda
. Conference: empirical evidence for the demographic and socio-economic impact of AIDS, Durban, S. Africa; 2003.
4. Nakiyingi JS, Bracher M, Whitworth JA, Ruberantwari A, Busingye J, Mbulaiteye SM, et al
. Child survival in relation to mother's HIV infection and survival: evidence from a Ugandan cohort study. AIDS 2003; 17:1827–1834.
5. Ng'weshemi J, Urassa M, Isingo R, Mwaluko G, Ngalula J, Boerma T, et al
. HIV impact on mother and child mortality in rural Tanzania. J Acquir Immune Defic Syndr 2003; 33:393–404.
6. Newell ML, Coovadia H, Cortina-Borja M, Rollins N, Gaillard P, Dabis F. Mortality of infected and uninfected infants born to HIV-infected mothers in Africa: a pooled analysis. Lancet 2004; 364:1236–1243.
7. Dabis F, Ekpini ER. HIV-1/AIDS and maternal and child health in Africa. Lancet 2002; 359:2097–2104.
9. UNAIDS (2008). Report on the global AIDS epidemic. UNAIDS, the Joint United Nations Programme on HIV/AIDS.
10. Chopra M, Daviaud E, Pattinson R, Fonn S, Lawn JE. Saving the lives of South Africa's mothers, babies, and children: can the health system deliver? Lancet 2009; 374:835–846.
12. Violari A, Cotton MF, Gibb DM, Babiker AG, Steyn J, Madhi SA, et al
. Early antiretroviral therapy and mortality among HIV-infected infants. N Engl J Med 2008; 359:2233–2244.
14. Coutsoudis A, Pillay K, Kuhn L, Spooner E, Tsai WY, Coovadia HM. Method of feeding and transmission of HIV-1 from mothers to children by 15 months of age: prospective cohort study from Durban, South Africa. AIDS 2001; 15:379–387.
15. Coetzee D, Hilderbrand K, Boulle A, Draper B, Abdullah F, Goemaere E. Effectiveness of the first district-wide programme for the prevention of mother-to-child transmission of HIV in South Africa. Bull World Health Organ 2005; 83:489–494.
16. Rice BD, Batzing-Feigenbaum J, Hosegood V, Tanser F, Hill C, Barnighausen T, et al
. Population and antenatal-based HIV prevalence estimates in a high contracepting female population in rural South Africa. BMC Public Health 2007; 7:160.
17. Dabis F, Elenga N, Meda N, Leroy V, Viho I, Manigart O, et al
. 18-Month mortality and perinatal exposure to zidovudine in West Africa. AIDS 2001; 15:771–779.
18. Barnighausen T, Tanser F, Gqwede Z, Mbizana C, Herbst K, Newell ML. High HIV incidence in a community with high HIV prevalence in rural South Africa: findings from a prospective population-based study. AIDS 2008; 22:139–144.
19. Welz T, Hosegood V, Jaffar S, Batzing-Feigenbaum J, Herbst K, Newell ML. Continued very high prevalence of HIV infection in rural KwaZulu-Natal, South Africa: a population-based longitudinal study. AIDS 2007; 21:1467–1472.
20. Tanser F, Hosegood V, Barnighausen T, Herbst K, Nyirenda M, Muhwava W, et al
. Cohort Profile: Africa Centre Demographic Information System (ACDIS) and population-based HIV survey. Int J Epidemiol 2008; 37:956–962.
22. Moultrie TA, Hosegood V, McGrath N, Hill C, Herbst K, Newell ML. Refining the criteria for stalled fertility declines: an application to rural KwaZulu-Natal, South Africa, 1990–2005. Stud Fam Plann 2008; 39:39–48.
23. Muhwava W, Nyirenda M. Demographic and Socio-Economic trends in the ACDIS, Monograph No 2
. Africa Centre for Health and Population Studies, Mtubatuba, South Africa; 2007.
24. Janssen N, Ndirangu J, Newell ML, Bland RM. Successful paediatric HIV treatment in rural primary care in Africa.Arch Dis Child
25. Houlihan CF, Bland RM, Mutevedzi PC, Lessells RJ, Ndirangu J, Thulare H, et al.Cohort profile: Hlabisa HIV treatment and care programme.IJE
26. Rollins NC, Coovadia HM, Bland RM, Coutsoudis A, Bennish ML, Patel D, et al
. Pregnancy outcomes in HIV-infected and uninfected women in rural and urban South Africa. J Acquir Immune Defic Syndr 2007; 44:321–328.
27. KwaZulu Natal Provincial Department of Health. Protocol for the phased implementation of a comprehensive package of care for the prevention of mother to child transmission in KwaZulu Natal
. 4th ed. 2003.
28. WHO/UNICEF/UNAIDS. HIV and Infant Feeding Counselling: a training course
. WHO/SCH/CAH/00.4; UNICEF/PD/NUT/(J)00-3; UNAIDS/99.57Eed. Geneva, Switzerland: World Health Organization; 2000.
29. Peter B, Day C, Monticelli F. Technical report: the District Health Barometer – Year 2006/07
. Health Systems Trust. Available at http://www.hst.org.za/publications/717
. [Accessed 4 November 2009].
30. Coovadia HM, Rollins NC, Bland RM, Little K, Coutsoudis A, Bennish ML, et al
. Mother-to-child transmission of HIV-1 infection during exclusive breastfeeding in the first 6 months of life: an intervention cohort study. Lancet 2007; 369:1107–1116.
31. Steele AD, Peenze I, de Beer MC, Pager CT, Yeats J, Potgieter N, et al
. Anticipating rotavirus vaccines: epidemiology and surveillance of rotavirus in South Africa. Vaccine 2003; 21(5–6):354–360.
32. Janssen N, Ndirangu J, Newell ML, Bland RM. Successful paediatric HIV treatment in rural primary care in Africa.Arch Dis Child
33. Tanser F, Gijsbertsen B, Herbst K. Modelling and understanding primary healthcare accessibility and utilization in rural South Africa: an exploration using a geographical information system. Soc Sci Med 2006; 63:691–705.
34. Mosley W, Chen L. An analytical framework for the study of child survival in developing countries. Populat Develop Rev 1984; 10(Suppl):24–45.
35. Becher H, Muller O, Jahn A, Gbangou A, Kynast-Wolf G, Kouyate B. Risk factors of infant and child mortality in rural Burkina Faso. Bull World Health Organ 2004; 82:265–273.
36. Gutierrez R. Parametric frailty and shared frailty survival models.Stata J
38. Newell ML. Prevention of mother-to-child transmission of HIV: challenges for the current decade. Bull World Health Organ 2001; 79:1138–1144.
39. Newell ML, Little KE, Bunders M. Preliminary deliberations towards estimating the burden of HIV disease in children, in the light of need for ART: Ghent-IAS Group on HIV infection in women and children
. [Accessed 13 April 2009]
40. Barnighausen T, Hosegood V, Timaeus IM, Newell ML. The socioeconomic determinants of HIV incidence: evidence from a longitudinal, population-based study in rural South Africa. AIDS 2007; 21(Suppl 7):S29–S38.
41. Cooke GS, Little KE, Bland RM, Thulare H, Newell ML. Need for timely paediatric HIV treatment within primary healthcare in rural South Africa. PLoS One 2009; 4:e7101.
42. Department of Health Medical Research Council. South Africa demographic and health survey 2003
. Pretoria: Department of Health; 2007.
43. HaR NA, Lie RT, Oneko O, Shao J, Bergsjo P, Daltveit AK. Sociodemographic characteristics and perinatal mortality among singletons in North East Tanzania: a registry-based study. J Epidemiol Community Health 2008; 62:960–965.
44. Norrgren H, Fonseca A, Andersson S, Dias F, Naucler A, Biberfeld G. Child survival in children born to HIV-2 infected women in Guinea-Bissau, West Africa. Acta Trop 1999; 72:309–315.
45. Zaba B, Whitworth J, Marston M, Nakiyingi J, Ruberantwari A, Urassa M, et al
. HIV and mortality of mothers and children: evidence from cohort studies in Uganda, Tanzania, and Malawi. Epidemiology 2005; 16:275–280.
46. Herbst AJ, Cooke GS, Barnighausen T, Kanykany A, Tanser F, Newell ML. Adult mortality and antiretroviral treatment roll-out in rural KwaZulu-Natal, South Africa. Bull World Health Organ 2009; 87:754–762.
47. Rollins N, Little K, Mzolo S, Horwood C, Newell ML. Surveillance of mother-to-child transmission prevention programmes at immunization clinics: the case for universal screening. AIDS 2007; 21:1341–1347.