C. Horwood 1, L. Haskins 1, K. Vermaak 1, S. Phakathi 1, R. Subbaye 1 and T. Doherty 2

Transcription

1 Tropical Medicine and International Health doi: /j x volume 15 no 9 pp september 2010 Prevention of mother to child transmission of HIV (PMTCT) programme in KwaZulu-Natal, South Africa: an evaluation of PMTCT implementation and integration into routine maternal, child and women s health services C. Horwood 1, L. Haskins 1, K. Vermaak 1, S. Phakathi 1, R. Subbaye 1 and T. Doherty 2 1 Centre for Rural Health, University of KwaZulu-Natal, Durban, South Africa 2 Medical Research Council, Health Systems Research Unit, Cape Town, South Africa Summary objectives To evaluate prevention of mother to child transmission of HIV (PMTCT) implementation and integration of PMTCT with routine maternal and child health services in two districts of KwaZulu- Natal; to report PMTCT coverage, to compare recorded and reported information, and to describe responsibilities of nurses and lay counsellors. methods Interviews were conducted with mothers in post-natal wards (PNW) and immunisation clinics; antenatal and child health records were reviewed. Interviews were conducted with nurses and lay counsellors in primary health care clinics. results Eight hundred and eighty-two interviews were conducted with mothers: 398 in PNWs and 484 immunisation clinics. During their recent pregnancy, 98.6% women attended antenatal care (ANC); 60.8% attended their first ANC in the third trimester, and 97.3% were tested for HIV. Of 312 mothers reporting themselves HIV positive during ANC, 91.3% received nevirapine, 78.2% had a CD4 count carried out, and 33.1% had a CD4 result recorded. In the immunisation clinic, 47.6% HIV-exposed babies had a PCR test, and 47.0% received co-trimoxazole. Of HIV-positive mothers, 42.1% received follow-up care, mainly from lay counsellors. In clinics, there was a dedicated PMTCT nurse, PCR testing was not offered in clinics, and co-trimoxazole was unavailable in immunisation clinics. Nurses and lay counsellors disagreed about their roles and responsibilities, particularly in the post-natal period. conclusions There is high coverage of PMTCT interventions during pregnancy and delivery, but follow-up of mothers and infants is poor. Poor integration of PMTCT services into routine care, lack of clarity about health worker roles and poor record keeping create barriers to accessing services postdelivery. keywords vertical infection transmission, infant, HIV, South Africa, health care systems, programme evaluation Introduction The millennium development goal for child survival requires countries to reduce child mortality by two-thirds before 2015 (UN 2000). South Africa is one of 12 countries globally where child and infant mortality rates have risen, mainly because of the burden of paediatric HIV disease, with more than half of child deaths being caused by HIV AIDS (Chopra et al. 2009). KwaZulu-Natal is the province worst affected by HIV; 38.7% of pregnant women attending antenatal clinics were HIV infected in 2008 (SA National Department of Health 2009). Interventions for prevention of mother to child transmission of HIV (PMTCT) are critical to reduce paediatric HIV infection and therefore child mortality. However, PMTCT implementation, particularly follow-up of HIVexposed infants, is suboptimal (Sherman et al. 2004), and the PMTCT programme has not led to the expected reductions in transmission rates (Rollins et al. 2007b). Early initiation of anti-retroviral therapy in HIV-infected babies can lead to substantial improvements in mortality (Violari et al. 2008), highlighting the critical importance of follow-up and early diagnosis of HIV infection in HIVexposed infants. 992 ª 2010 Blackwell Publishing Ltd

2 The introduction of a PMTCT programme in South Africa in 2002 created an additional workload for already overworked staff in a weak health system. In response, a new cadre of non-professional health workers, known as lay counsellors, was introduced (Chopra et al. 2009). Lay counsellors, who were trained in HIV counselling and whose principal role is voluntary counselling and testing (VCT), are employed by the Department of Health and based in health facilities. Although lay counsellors contributed to the early success in increasing the coverage of PMTCT in KZN (Doherty et al. 2005), lack of integration into maternal, child and neonatal health services has persistently undermined the programme (Chopra et al. 2009). Professional nurses (PNs) continued with their routine practice during antenatal care (ANC) and child health consultations, while PMTCT implementation was seen as the lay counsellors responsibility. The introduction of PMTCT as a largely vertical programme, with budgets and decision-making controlled by a national HIV department, resulted in conflicts that have been difficult to resolve, and which have further exacerbated poor PMTCT integration into routine maternal and child health services (Chopra et al. 2009). The complexity of the PMTCT programme requires that PMTCT activities be carried out by several health workers, at different times and in different settings (Newell 2001). To achieve comprehensive implementation, it is critical that HIV-infected women and their babies are identified at every contact with the health service and that all steps in the PMTCT process are fully implemented. The overall effectiveness of the programme also depends on the identification of pregnant women with low CD4 counts and initiation of highly active antiretroviral treatment (HAART) before delivery in this high risk group (Tonwe- Gold et al. 2007). Accurate record keeping and continuity of care is required, and any break in the chain of activities will lead to a reduction in the expected benefits of the programme for both mother and infant. Although district-wide implementation of PMTCT programmes is feasible in different South African settings (Coetzee et al. 2005; Jackson et al. 2007), very high rates of loss to follow-up have been observed especially for HIVexposed children (Doherty et al. 2005; Jones et al. 2005), and a study to determine the effectiveness of PMTCT in KwaZulu-Natal in 2005 suggested that the PMTCT programme had little impact on transmission rates (Rollins et al. 2007a). Fragmentation of services, with PMTCT care being delivered in particular areas of the clinic, and by different cadres of health workers, has been named as a major barrier to accessing services (Greeff & Phetlhu 2007). Mothers are afraid of being stigmatised by community members if receiving PMTCT services identifies them as HIV infected (Horwood et al. 2009). Fear of such unintended disclosure may be an important reason why mothers fail to seek care for HIV-exposed infants (Varga et al. 2006). Thus, improving integration of PMTCT services into routine care is critical to improve access to these services. No formal quantitative evaluation of the routine implementation of all aspects of the PMTCT programme at district level in KwaZulu-Natal has been reported previously. PMTCT evaluations in South Africa have frequently focussed on large urban centres or on pilot sites. In this article, we present an evaluation of routine implementation of the PMTCT programme in two districts of KwaZulu- Natal and the integration of PMTCT with routine maternal and child health services. Methods A quantitative, cross-sectional descriptive study was undertaken in Amajuba and Uthukela districts, KwaZulu-Natal. These districts are largely peri-urban or rural. Data were collected using structured data collection questionnaires, developed and piloted before the start of data collection, and data collectors were trained in their use. Interviews were conducted with mothers in the postnatal ward (PNW) of all six district hospitals in the two districts and in the immunisation clinics of 27 primary health care (PHC) facilities. Antenatal and labour ward records were reviewed on the PNW, and the child health record (road to health card) was reviewed in the immunisation clinic so that information reported by mothers could be compared with that recorded on the health records. At each clinic visited, interviews were conducted with the sister-in-charge of antenatal and immunisation services and with a randomly selected lay counsellor. The sample size was calculated using Rightsize (CDC 2002) to provide estimates with a 10% level of precision for two key indicators: the proportion of HIV-positive pregnant women who take nevirapine before delivery, and the proportion of HIV-exposed infants whose blood was taken for PCR. The number of interviews on the PNW was calculated as 144 HIV-positive mothers in total, this was allocated to each hospital proportionately to the average number of births at that facility. All mothers on the PNW who were well enough to be interviewed and gave informed consent were included in the study. Interviews continued in each hospital until the required number of mothers reporting themselves HIV positive was reached. The sample size for interviews with mothers exiting immunisation clinics was calculated as 108 HIV-positive mothers; four mothers from each of 27 PHC clinics. Clinics were selected probability proportionate to size (PPS) with ª 2010 Blackwell Publishing Ltd 993

3 the measure of size being the average number of children aged 6 16 weeks immunised at that clinic. All mothers bringing babies aged 6 16 weeks to the immunisation clinic who gave informed consent were included in the study, other carers bringing babies were excluded. Interviews were conducted after the consultation was complete and continued in each clinic until four mothers had been interviewed who reported testing HIV positive. To prevent data collectors from preferentially selecting HIV-positive mothers for interview, data collectors were not informed of the number of interviews required or that this was based on numbers of HIV-positive mothers. Data collectors reported the number of interviews conducted each day and were informed when data collection was complete for a particular facility. Permission to undertake the study was obtained from KZN Department of Health. Written informed consent was obtained from all participants in the local language. Ethical approval was obtained from the Biomedical Research Ethics Committees of the University of the Western Cape and the University of KwaZulu-Natal. Pre-coded data were double entered, cleaned and validated using Epi-info (version 6.04; Centers for Disease Control and Prevention, Atlanta, GA). Analysis was conducted using spss (version 15.0; SPSS Inc., Chicago, IL). Frequencies were calculated, and confidence intervals were calculated for comparison of reported and recorded information. Results Eight hundred and eighty-two interviews were conducted between October 2007 and February Ten mothers who did not report themselves as HIV positive but could be identified as HIV positive from their records were excluded from the analysis as they could not be expected to report accurately on the services they received. The analysis therefore includes 872 interviews, comprising 392 interviews with mothers in the PNWs and 480 with mothers exiting immunisation clinics (Figure 1). Antenatal care Most women ( ; 98.6%) attended the antenatal clinic during their recent pregnancy; (8.4%) women reported attending in the first 3 months of Total number of mothers interviewed/records reviewed n = 872 Postnatal ward (n = 392) Immunization clinics (n = 480) Interviews (392) ANC care (387) a LW care (363) b PNW care (392) Record review (392) ANC card (379) a LW records (365) b PNW care (392) Interviews (480) ANC care (480) LW care (480) PNW care (480) (480) Record review (480) Child health record (480) Reported HIV positive = 155 Tested positive prior to this pregnancy = 43 Tested positive during ANC = 108 Tested positive on PNW = 4 Recorded HIV positive = 161 before delivery = 157 in PNW = 4 Reported HIV positive = 164 prior to this pregnancy = 41 during ANC = 120 on LW = 2 Recorded HIV positive = 81 a Excludes women who did not attend ANC. b Excludes those who delivered outside the health facility. Figure 1 Outcomes of data collection at different data collection sites. 994 ª 2010 Blackwell Publishing Ltd

4 pregnancy, and (70.6%) reported their first attendance between 4 and 6 months. Records showed that the midwife assessed 60.8% of women as being in the third trimester at the time of booking. Of 776 women who attended the ANC and who did not know that they were HIV positive prior to the recent pregnancy, 755 (97.3%) reported being tested for HIV during their recent pregnancy. The results reported were the following: (30.2%) tested HIV positive, (69%) tested HIV negative, and (0.8%) did not receive their results. There were 312 women who reported on the PMTCT services they received during their recent pregnancy; comprising 84 women who tested HIV positive prior to this pregnancy and 228 who tested positive during this pregnancy (Table 1). Although (78.2%) reported having had blood taken for a CD4 count, (54.8%) had received the CD4 result at the time of the interview. Reported coverage of PMTCT interventions during ANC is given in Table 2. Intrapartum post-natal care There were 318 women who reported on PMTCT care received in the labour and PNWs; comprising the 312 mothers mentioned earlier and six additional mothers who tested HIV positive while on the labour ward or PNW (Table 1). Reported coverage of PMTCT interventions during delivery and on the PNW is given in Table 2. Although (69.8%) of mothers reported that the baby had been given nevirapine, 26.7% did not know if their baby had received this intervention. Of 480 mothers interviewed exiting immunisation clinics, 282 (58.8%) reported that the health worker either asked about or knew their HIV status when they attended the first immunisation visit. Three hundred and twenty-four of 480 (67.5%) mothers reported that they had seen only an enrolled nurse during their visit, (3.5%) saw a counsellor in addition to the enrolled nurse, and only (16.0%) reported seeing a PN during their visit to the immunisation clinic. Of 164 mothers interviewed in the immunisation clinic who reported themselves to be HIV positive, 128 (78.0%) reported that they were advised to have the baby tested for HIV during the first immunisation visit, but the heel prick for PCR was only carried out on 78 (47.6%) HIV-exposed babies (Table 2). Few HIV-positive women (69 164; 42.1%) reported receiving monthly follow-up for their own care; and of these, 56.5% reported receiving followup care from a lay counsellor. Coverage of PMTCT interventions: comparing reported and recorded information A comparison between reported PMTCT interventions and those recorded on the maternity records and on the child health record showed that in many cases interventions reported by mothers were not recorded on the health record; for example, 77.7% of mothers reported having blood taken for a CD4 count but this was recorded on maternity records in only 47.0% of cases (Table 3). Roles and responsibilities of health workers Interviews were conducted with PNs in 26 clinics (one clinic did not have a PN on duty at the time of the study) and with lay counsellors in 27 clinics. To compare responses from different cadres of health workers in the same clinic, analysis was limited to the 26 clinics where both health workers were interviewed. Table 1 Interviews completed and sites of reported HIV testing Interviews PNW Immunisation clinic Total Cumulative reported HIV positive Interviews completed Number not reporting positive* Number interviews included in analysis Reported testing positive prior to this pregnancy Reported testing positive in ANC Reported testing positive in LW Reported testing positive in PNW Reported testing positive in immunisation clinic n a Total reporting HIV positive ANC, antenatal care; PNW, post-natal ward. *Interviews with women who did not report themselves positive but were recorded HIV positive on the records were excluded. ª 2010 Blackwell Publishing Ltd 995

5 Table 2 Reported coverage of PMTCT interventions Antenatal clinic Reported HIV positive 312 (%) Blood for CD4 taken ( ) Infant feeding counselling given ( ) Able to talk privately with a HW ( ) Issued with NVP ( ) Knew when to take NVP ( ) CD4 results available ( ) CD4 count 200 and below (18 171) CD4 count 201 and above ( ) CD4 count not known (7 171) Attended an ARV clinic (40 312) ARVs taken during pregnancy (27 40) Post-natal ward Reported HIV positive 318 HW talked to mother about feeding plans HW talked to mother about having baby tested Baby given NVP Baby not given NVP Mother did not know if baby was given NVP Reported HIV positive 164 Advised to have baby tested for HIV Health worker was asked about infant feeding Able to talk privately with a HW Baby tested for HIV Individual counselling given for PCR Blood sample taken by nurse Blood sample taken by doctor Blood sample taken by counsellor CTX given for baby CTX given by nurse CTX given by counsellor PMTCT, prevention of mother to child transmission of HIV. In clinics, there was a dedicated PN specifically charged with providing PMTCT services in the antenatal clinic. In many clinics, PMTCT services were not provided at the immunisation clinic; clinics provided co-trimoxazole to HIV-exposed babies during the immunisation clinic, in clinics PCR tests were carried out in the immunisation clinic, formula milk was provided to mothers in the immunisation clinic of 18 clinics. Lay counsellors and PNs were asked who is currently undertaking each PMTCT-related activity at their clinic (Table 4). There was general agreement about the primary role of lay counsellors in HIV pre-test counselling, infant feeding counselling and doing initial rapid HIV tests, and of PNs in doing confirmatory rapid HIV tests, taking blood for CD4 counts, and dispensing nevirapine to the mother. However, the cadre of health worker responsible for CD4 testing, including obtaining results and giving them to the mother, varied from clinic to clinic, and there was frequently disagreement between the PN and lay counsellor at the same clinic about who undertakes these tasks. In the immunisation clinic, there was widespread disagreement about who was currently undertaking PMTCT activities (Table 4). In some clinics, lay counsellors were undertaking tasks outside their scope of practice according to PMTCT guidelines; in two clinics, both cadres of health workers agreed that lay counsellors do confirmatory rapid HIV tests; and in nine clinics, health workers agreed that lay counsellors dispense co-trimoxazole for the baby. 996 ª 2010 Blackwell Publishing Ltd

6 Table 3 Reported vs. recorded coverage of PMTCT interventions Event Source (n) Number (%) 95%CI on (%) P Antenatal clinic Booked in the last trimester Reported Recorded Had HIV test during ANC Reported 342* Recorded HIV test result Reported Recorded Issued with NVP Reported Recorded Blood for CD4 taken Reported Recorded CD4 result available Reported Recorded NVP given to baby Reported Recorded Blood taken for a PCR Reported Recorded CTX given for baby Reported Recorded ANC, antenatal care; PMTCT, prevention of mother to child transmission of HIV. *Excludes the 43 women who tested HIV positive before this pregnancy. Discussion The PMTCT programme is functioning well in the antenatal period, achieving almost universal coverage of HIV testing among pregnant women and good coverage of other interventions, although a small but significant proportion of women did not receive nevirapine (NVP). This study highlights that the major gap in provision of antenatal PMTCT care is the failure to obtain CD4 results, refer women for ART and start HAART where appropriate. It is important for the overall effectiveness of the PMTCT programme that women with low CD4 counts, whose high viral loads put them at high risk of MTCT, are identified early and started on HAART prior to delivery (Tonwe-Gold et al. 2007). Despite interviews being conducted post-delivery, only 54.8% of HIV-positive women had a CD4 result available and only 8.7% were on HAART. There is a lack of clarity about whose role it is to arrange CD4 testing. However, it is difficult to obtain CD4 results and initiate HAART before delivery when most women book in the third trimester and, even when available, CD4 results are frequently not recorded on the antenatal card, so health workers have to rely on reported results. At the time of this study, single dose NVP was provided for PMTCT, but dual therapy has since been introduced, with zidovudine to be taken from 28 weeks of pregnancy. It is therefore urgent that women are encouraged to book earlier for ANC, so that all PMTCT services can be accessed in time. This data also shows the breakdown in PMTCT coverage and continuity of care after delivery of the baby. Follow-up of HIV-exposed infants remains poor, with fewer than half of these babies having a PCR test and starting on co-trimoxazole at the first immunisation visit. Coverage of first immunisation is high in KZN (Chopra et al. 2009), and the first immunisation visit provides an excellent opportunity for follow-up and testing of HIVexposed children within routine services. However, PMTCT services are frequently not available at the immunisation clinic, and poor recording of PMTCT interventions on the child health record means that followup care depends on the mother reporting that the baby is ª 2010 Blackwell Publishing Ltd 997

7 Table 4 Roles and responsibilities of professional nurses and lay counsellors Response of PN and lay counsellor working in the same clinic N = 26 Both agree lay counsellor does this Both agree PN does this PN and lay counsellor disagree Activity Number of clinics Number of clinics Number of clinics ANC Group counselling Initial rapid test Confirmatory rapid test Manage discordant results Arrange CD4 testing for mother Obtaining the CD4 results Giving mother her CD4 result Refer to ARV clinic Dispense NVP to mother Counselling about infant feeding Give co-trimoxazole to the baby Identify when a baby needs a PCR Take blood for PCR Obtain PCR result Give PCR result to the mother ANC, antenatal care; PN, professional nurse. HIV exposed. There is a lack of clarity around roles and responsibilities of PNs and counsellors in provision of care for HIV-exposed babies, and most immunisation clinics are entirely conducted by enrolled nurses. These are barriers to follow-up and continuity of care for HIV-exposed children. Training of enrolled nurses in PMTCT could facilitate integration of PMTCT into routine well-child services and improve follow-up of exposed children. Enrolled nurses could also be trained to undertake PCR testing and provide co-trimoxazole, which could improve efficiency of PMTCT services and the capacity of the health system to provide such services (Lehmann et al. 2009). In many clinics, PMTCT services are rendered by a dedicated PMTCT nurse, suggesting that these services are not integrated into routine care, and mothers may be reluctant to attend specialised HIV clinics or services because they fear being identified as HIV infected by other community members attending the clinic (Horwood et al. 2009), and this has been identified as an underlying cause of poor infant follow-up (Varga et al. 2006). Follow-up care for HIV-positive mothers is also poor; few women reported attending regular follow-up for their own care, and where this was being carried out it was mostly by lay counsellors, who are inadequately trained. The link between the care of the HIV-infected mother during pregnancy and delivery, and the follow-up care of the mother and baby needs to be strengthened, to avoid the drop off in PMTCT service provision that occurs after delivery. Respect for confidentiality must be balanced with the responsibility of the health service to provide optimum care for the baby, clear written policies are urgently required to guide health workers in recording of PMTCT information on medical records. Roles and responsibilities of different health workers need to be clearly defined and enforced; shifting of tasks to enrolled nurses or lay counsellors may be an important strategy to improve coverage of PMTCT care, this should be carried out within the appropriate policy framework and accompanied by the necessary training (Lehmann et al. 2009). Limitations to the study include that the data rely on the accuracy of information reported by mothers. However, all women were interviewed either at the end of their pregnancy or within a few weeks, so that the period of recall was short. We also relied on the women s willingness to disclose their HIV status, although we were able to confirm information using the record review, and it appeared that very few women did not report their HIV status correctly. We only interviewed one PN and counsellor in each clinic, who may have not been able to report the full picture of their colleagues roles and responsibilities. Conclusion Prevention of mother to child transmission of HIV implementation has made huge strides and achieved high 998 ª 2010 Blackwell Publishing Ltd

8 coverage of interventions during pregnancy and delivery, but coverage of critical PMTCT interventions for the baby remains poor. Many clinics have designated health workers for PMTCT, PMTCT services are frequently not provided as part of routine care, and record keeping is poor. This creates barriers for mothers and babies access to care after delivery and has lead to fragmentation of services, placing the responsibility for follow-up on the mother. Clear policies are required in regard to record keeping and roles and responsibilities of health workers; tasks should be shifted to improve integration of PMTCT into routine services; and mothers need to be encouraged to attend ANC earlier in pregnancy. Acknowledgements All funding was provided by the Medical Research Council (MRC) with a grant from CDC-PEPFAR. The authors would like to thank Mickey Chopra from University of the Western Cape and MRC, Duduzile Nsibande from the MRC and Steve Reid and Catherine Cosser from the Centre for Rural Health, UKZN, for their support in developing this study. We are grateful for the support of Amajuba and uthukela district management teams. We thank the data collection teams who worked so hard, the staff of the participating clinics for their support, and all the mothers and infants who agreed to participate. References CDC (2002) RightSize China-Uganda-Zimbabwe version Developed by Centres for Disease Control. Chopra M, Daviaud E, Pattinson R, Fonn S & Lawn JE (2009) Saving the lives of South Africa s mothers, babies, and children: can the health system deliver? Lancet 374, Coetzee D, Hilderbrand K, Boulle A, Draper B, Abdullah F & Goemaere E (2005) Effectiveness of the first district-wide programme for the prevention of mother-to-child transmission of HIV in South Africa. Bulletin of the World Health Organization 83, Doherty TM, McCoy D & Donohue S (2005) Health system constraints to optimal coverage of the prevention of mother-tochild HIV transmission programme in South Africa: lessons from the implementation of the national pilot programme. African Health Sciences 5, Greeff M & Phetlhu R (2007) The meaning and effect of HIV AIDS stigma for people living with AIDS and nurses involved in their care in the North West Province, South Africa. Curationis 30, Horwood C, Voce A, Vermaak K, Rollins N & Qazi S (2009) Routine checks for HIV in children attending primary health care facilities in South Africa: attitudes of nurses and child caregivers. Social Science and Medicine 70, Jackson DJ, Chopra M, Doherty TM et al. (2007) Operational effectiveness and 36 week HIV-free survival in the South African programme to prevent mother-to-child transmission of HIV-1. AIDS 21, Jones SA, Sherman GG & Varga CA (2005) Exploring socioeconomic conditions and poor follow-up rates of HIV-exposed infants in Johannesburg, South Africa. AIDS Care 17, Lehmann U, Van Damme W, Barten F & Sanders D (2009) Task shifting: the answer to the human resources crisis in Africa? Human Resources for Health 7, 49. Newell ML (2001) Prevention of mother-to-child transmission of HIV: challenges for the current decade. Bulletin of the World Health Organization 79, Rollins N, Little K, Mzolo S, Horwood C & Newell ML (2007a) Surveillance of mother-to-child transmission prevention programmes at immunization clinics: the case for universal screening. AIDS 21, Rollins NC, Coovadia HM, Bland RM et al. (2007b) Pregnancy outcomes in HIV-infected and uninfected women in rural and urban South Africa. Journal of Acquired Immune Deficiency Syndromes 44, SA National Department of Health (2009) National Antenatal Sentinel HIV and Syphillis Seroprevalence Survey. Department of Health, Pretoria. Sherman GG, Jones SA, Coovadia AH, Urban MF & Bolton KD (2004) PMTCT from research to reality results from a routine service. South African Medical Journal 94, Tonwe-Gold B, Ekouevi DK, Viho I et al. (2007) Antiretroviral treatment and prevention of peripartum and postnatal HIV transmission in West Africa: evaluation of a two-tiered approach. PLoS Medicine 4, e25. United Nations (2000) United Nations Millenium Declaration. Varga CA, Sherman GG & Jones SA (2006) HIV-disclosure in the context of vertical transmission: HIV-positive mothers in Johannesburg, South Africa. AIDS Care 18, Violari A, Cotton MF, Gibb DM et al. (2008) Early antiretroviral therapy and mortality among HIV-infected infants. New England Journal of Medicine 359, Corresponding Author Christine Horwood, Centre for Rural Health, University of KwaZulu-Natal, Umbilo Road, Durban 4013, South Africa. horwoodc@ukzn.ac.za ª 2010 Blackwell Publishing Ltd 999