Mother-to-child transmission of HIV remains the primary route of HIV infection in children. Transmission risk can be dramatically reduced through the use of antepartum, intrapartum, and postpartum antiretroviral (ARV) regimens.1 Recent reports of declining infant mortality in South Africa may be due in part to the effectiveness of the prevention of mother-to-child transmission of HIV (PMTCT) program and survival of infected mothers on antiretroviral therapy.2,3 Yet, antenatal seroprevalence remains high, having increased to 30.2% in 2010,4 from 29.3% in 2008,when this study was conducted.5 New national PMTCT guidelines based on 2010 WHO recommendations6,7 are being implemented throughout public health services, and preliminary results from a recent national survey suggest that early mother-to-child transmission of HIV may currently be as low as 3.5%.8 The elimination of pediatric HIV, however, still requires committed and sustained efforts to achieve high PMTCT program coverage and a continuum of care that also identifies women at risk of incident infection in pregnancy and postpartum vertical transmission.9–12
The effectiveness of PMTCT programs is contingent on a cascade of critical steps which include HIV testing in pregnancy and the provision of and adherence to antepartum, intrapartum, and postpartum ARV regimens.13 Epidemiological modelling of program effectiveness has shown that the use of more effective ARV drug combinations could have limited impact on transmission if adequate service coverage of all the steps of the PMTCT program is not achieved.14 This highlights the importance of identifying and reducing the missed opportunities in program coverage.
Effective monitoring and evaluation of PMTCT programs is hence crucial for the identification of missed opportunities in service delivery along the PMTCT cascade. Since 2001, PMTCT program coverage in South Africa has been reported through indicators routinely collected and aggregated from service data. However, there is uncertainty about the quality of data due to incomplete and inaccurate reporting.15,16 In addition, these data reflect coverage in women attending antenatal care and further document dispensing of PMTCT prophylaxis rather than actual ingestion.17
In 2007, the PMTCT Effectiveness in Africa: Research and Linkages to Care (PEARL) study consortium explored a new approach to the evaluation of PMTCT program coverage through cord blood surveillance of HIV status and nevirapine (NVP) concentration in 4 African countries.18 The study was expanded in the Western Cape Province of South Africa to document the effective administration of azidothymidine (AZT) and combined antiretroviral therapy (cART), which were at the time available as standard of care to HIV-infected women with CD4 cell counts above and below 200 cells per microliter, respectively. Our aim was to describe true standard of care PMTCT coverage using cord blood surveillance of HIV infection and the presence of ARVs. We further sought to compare these outcomes with program reporting and examine missed opportunities for preventing vertical transmission in the Western Cape PMTCT programme,19–21 which started in 2002 and was considered at the time to be effective based on routine reporting.21
HIV prevalence in the Western Cape among pregnant women 15–40 years was 16.1% in 2008, ranging between subdistricts from 9.3% to 17.9%.5 Two delivery services in the metropolitan area and 1 rural area were randomly selected from 32 subdistricts in the province.
At the time of this study, PMTCT services were integrated into primary health care antenatal services, with the exception of cART provision to eligible women, which was provided by a separate program within the same facility at the antenatal service at all 3 sites. At the first antenatal visit, all women received group HIV counseling followed by individual counseling if they agreed to HIV testing. At the same time a CD4 cell count was requested. In 2007–2008, the standard of care for HIV-infected pregnant women with a CD4 cell count >200 cells per microliter comprised AZT from 28 weeks gestation in combination with single-dose NVP to the mother and exposed infant at time of delivery, and 3 hourly AZT during labour to the mother. If a woman had a CD4 cell count ≤200 cells per microliter, she was referred to a separate service to initiate cART.22 Vertical transmission proportions according to these treatment guidelines were reported to be 3.5% among infants who were tested around 6 weeks in immunization services in 2009 in the Western Cape.23
Detailed study methodology has been described elsewhere.18 To ascertain cord blood HIV seroprevalence and ARV coverage, consecutive umbilical cord blood specimens were drawn at the study sites between December 2007 and May 2008. A sample size calculation was based upon country estimates of HIV seroprevalence and an assumed program coverage and is reported elsewhere.18 In some sites, there were more deliveries, and the study was completed within 6 weeks, whereas in others completion took up to 6 months to complete. Anonymous HIV prevalence testing was performed using a rapid immunoassay (Determine HIV1/2, Abbott Laboratories, Chicago, IL). High-performance liquid chromatography and tandem mass spectrometry were used to extract and qualitatively detect the presence of maternal NVP, AZT, and lamivudine (3TC) on all seropositive specimens.24 To draw comparisons between cord blood outcomes and PMTCT program reporting, linked anonymous surveillance data were extracted from patient-held antenatal records and service registers at the time of specimen collection. Information collected included demographic details, antenatal and obstetric data, and recorded maternal and infant PMTCT regimen administration.
Primary study outcomes were maternal HIV seroprevalence and ARV coverage, defined as the proportion of women who had at least NVP and AZT or cART (ascertained through the presence of 3TC) detected in cord blood specimens at delivery. Cord blood HIV seroprevalence and ARV coverage were then compared with HIV seroprevalence and ARV coverage data extracted from service records. In the service data, standard of care was assumed to be any documentation of AZT in combination with single-dose NVP, or cART, regardless of duration. A series of proportions for each step of the PMTCT cascade was constructed from the extracted services data and cord blood specimen results. Kappa statistics were calculated to measure the agreement beyond chance between the presence of ARVs in the cord blood and reported ARV provision. Maternal–infant coverage was also determined through combined cord blood ARV coverage and documented administration of infant NVP and/or AZT from service records. Twin deliveries were treated as singleton deliveries for the purposes of analysis, as each set of HIV-exposed twins (n = 4) received the same infant prophylaxis. Factors associated with PMTCT coverage were determined using a multivariable logistic regression model. The modeled impact of missed opportunities for maternal prophylaxis was assessed by assuming differential vertical transmission based on standard of care interventions,22 suboptimal prophylaxis, no prophylaxis, and seroconversion during pregnancy. Model estimates were based on assumed probabilities of 0.04 (AZT and NVP), 0.02 (cART), 0.12 (NVP only), 0.07 (AZT only), 0.20 (no intervention), and 0.35 (seroconversion) reported by Johnson et al in a model for pediatric HIV in this setting.25
Ethical approval was granted by the institutional review boards of the United States Centers for Disease Control and Prevention; the Health Sciences Faculty Human Research Ethics Committee, University of Cape Town; the Health Department of the Provincial Government of the Western Cape and the University of Alabama at Birmingham.
Between December 2007 and May 2008, 3109 women delivered at the 3 sites and 3065 (98.6%) cord blood specimens were collected. Specimens were not collected from women who delivered before arrival, or in instances where it was not possible to retrieve blood from the umbilical cord (cord snapped or unhealthy), or accidental specimen spillage. In total, 3034 specimens (97.6%) were tested for HIV. Of the HIV-infected specimens (n = 507), 470 specimens (92.7%) were tested for the presence of ARV regimens (see Figure, Supplemental Digital Content 1, https://links.lww.com/QAI/A281).
The median maternal age of women was 24 years [interquartile range (IQR): 21–28 years], and 37.4% of women were primigravid. The median number of antenatal visits was 4 (IQR: 3–5 visits), with 21.2% of women having 2 or fewer visits and 8.2% of women having had no antenatal visits. Overall, 10.5% of women had cesarian sections. Four infants were stillborn and were excluded from the maternal–infant coverage analyses. The median birth weight of infants was 3100 g (IQR: 2800–3400 g), and 10.8% weighed 2500 g or less (see Table, Supplemental Digital Content 2, https://links.lww.com/QAI/A282).
Documented Antenatal HIV Testing and HIV Seroprevalence Versus Maternal Cord Blood HIV Seroprevalence at Delivery
Documented antenatal HIV testing coverage was 79.4% (n = 2408, Fig. 1), and antenatal HIV prevalence among these women was 17.8% (n = 429). Cord blood HIV seroprevalence among all women who delivered was similar at 16.7% (n = 507, P = 0.53). Among women who did not have an antenatal HIV test, 12.3% (n = 77 of 626) had an HIV-seropositive cord blood result. Based on documented antenatal HIV status versus cord blood status at delivery, there were 13 seroconversions (0.7%) and 12 specimens (2.8%) that were reported as HIV seropositive in pregnancy were cord blood HIV seronegative (Fig. 1).
Documented Dispensed ARV Prophylaxis and Cord Blood ARV Coverage
Table 1 shows that among the women who were documented to be HIV seropositive in service records, 65.8% (n = 275) received AZT and NVP; 16.9% (n = 71) AZT only; 2.4% (n = 10) NVP only, and 7 women were documented to have received no intervention. Although CD4 cell counts were not extracted from the antenatal data and eligibility for cART could not be ascertained, a further 13.2% (n = 55) of women were documented by patient record to be on cART. Hence according to service records, documented ARV coverage for receiving the standard of care including cART was 78.9% (n = 330 of 418), and documented ARV coverage for all women who received any form of ARV drug was 98.3% (n = 411 of 418).
In comparison, of the 470 HIV-seropositive women for whom cord blood specimens were available, 46.2% (n = 217 of 470) had AZT and NVP in their cord blood; 15.5% (n = 73 of 470) had either AZT or NVP, and 11.9% (n = 56 of 470) had 3TC, indicating that they were on cART. A further 26.4% (n = 124 of 470) of HIV-seropositive cord blood specimens had no trace of any ARV regimen. In summary, 58.1% (n = 273 of 470) of women confirmed to be HIV seropositive by cord blood had evidence of a standard of care regimen (dual therapy or cART), whereas 73.6% (n = 346 of 470) of women were found to have received some form of ARV prophylaxis (Table 1).
Accuracy of Routinely Recorded Coverage Data
When restricting the analysis to women who were documented by the services to be HIV seropositive, and cord blood HIV seropositive and in whom documented coverage could therefore be verified on cord blood (complete results on n = 388), estimates of coverage were lower for dual therapy and AZT monotherapy but higher for NVP on its own (55.4%% vs. 65.8%, 8.2% vs. 16.9%, and 9.7% vs. 2.4%, respectively, Table 1). Combined estimates for standard of care regimens or any regimen were consequently lower on cord blood analysis than reported by the services (69.6% vs. 78.9% and 87.9% vs. 98.3%, respectively, Table 1).
Kappa statistics measuring agreement beyond chance among the mothers for whom data was available from both sources for cART, AZT, and NVP were 0.60, 0.14, and 0.32, respectively, indicating reasonable agreement for cART, but limited agreement for the individual drugs, especially AZT. Of 242 mothers with AZT detected in their cord blood and for whom clinical record data were available, 236 (97.5%) had this recorded in their charts, whereas there was no evidence of AZT in the cord blood of 61 of 297 (20.5%) of mothers who were reported to have received AZT antenatally. Discrepancies between those reported to have received NVP peripartum and the detection of NVP in cord blood were comparable in both directions, with 15.3% (39 of 255) of those recorded to have received NVP not having it detected in their cord blood, and similarly 14.6% (37 of 253) of those with evidence of NVP in their cord blood not having any record of having received this (see Table, Supplemental Digital Content 3, https://links.lww.com/QAI/A283). Five of 72 women who were documented as not having tested and who were cord blood HIV seropositive had evidence of ARVs in their cord blood, whereas none of the 12 women who were documented as being HIV seropositive but were HIV seronegative on cord blood testing were evaluated for ARVs in their cord blood.
Documented and Cord Blood Comparison of Combined Maternal–Infant ARV Coverage
Two infants born to HIV-seropositive women were stillborn, and these mother–infant pairs were removed from this subanalysis. According to service records, 68.6% (n = 293) of infants born to HIV-seropositive mothers (n = 427) received ARV prophylaxis and 58.3% (n = 249) mother–infant pairs received combined maternal standard of care and infant prophylaxis. When examining documented infant prophylaxis coverage among women with an HIV-seropositive cord blood specimen (n = 468), 62.6% (n = 293) infants were reported to have received ARV prophylaxis, and combined mother–infant coverage (seropositive cord blood and documented infant coverage) was 54.7% (n = 256). More than two-thirds of the exposed newborns who did not get ARV prophylaxis were missed because their mothers had not tested for HIV in pregnancy (72 of 175, 41.1%).
Factors Associated With Maternal Coverage
In multivariable analysis that included maternal age, facility location as metropolitan or periurban, and mode of delivery as possible confounders based on a priori assumptions, 5 or more antenatal clinic visits was the only measured association with maternal coverage (defined as receiving either dual prophylaxis or cART, adjusted odds ratio: 2.47; 95% confidence interval: 1.19 to 5.15; see Table, Supplemental Digital Content 4, https://links.lww.com/QAI/A284).
Based on these results, PMTCT coverage seems near-complete in program reporting, although 19% of women received only single therapy and 1.7% received nothing (Table 1). However, Figure 2 shows that according to cord blood surveillance, a substantial portion of mothers (16.4%) identified as cord blood HIV seropositive did not receive prophylaxis because they had not tested for HIV (Fig. 2, black section). A further portion of known HIV-seropositive mothers had no ARVs detectable in their cord blood (7.9%, Fig. 2, diagonal line section). Assuming differential vertical transmission for mothers seroconverting in pregnancy compared with those already HIV seropositive receiving no intervention (0.35 vs. 0.20)25 and for mothers receiving standard of care (0.04 for dual therapy and 0.02 for cART) versus other maternal ARV prophylaxis: 0.12 for NVP and 0.07 for AZT,25 the most frequent missed opportunities for preventing transmission were the result of incomplete antenatal HIV testing and failure to administer any prophylaxis or to ensure adherence in women known to be HIV seropositive.
As the international focus shifts toward the global elimination of pediatric HIV, effectively monitoring the coverage of PMTCT programs is imperative to determine their strengths and to address the missed opportunities for identifying HIV infection and for delivering treatment during pregnancy. Our results show that just over half of women identified as cord blood HIV seropositive received the standard of care (dual therapy or cART) and about three-quarters of women testing cord blood HIV seropositive had received some form of ARV. Maternal cord blood ARV coverage was significantly lower than that suggested by maternal coverage in routine data. Identifiable program gaps included incomplete HIV testing and a failure to administer ARVs or ensure adherence.
Concordance of Reported Versus Cord Blood Maternal HIV Seropositivity
Compared with other studies, seroconversion in this study was low (0.7%), assuming that all were true seroconversions and not documenting or reporting errors. Another study in the Cape Town metropole found no seroconversions among 532 pregnant women in 2004.26 With an estimated transmission probability of 0.35,25 seroconversion in pregnancy may account for a small yet significant proportion of transmission in many settings. As program coverage improves, so the relative contribution of seroconversion will increase, supporting the strategy of repeat testing close to delivery to identify acute infection.27
Concordance of Routine Versus Cord Blood Maternal ARV Prophylaxis Coverage
Results from the cord blood analysis suggested appreciably lower maternal ARV coverage than the near-complete coverage suggested by service data, even when restricted to patients who were definitively known by the services to be HIV seropositive. The largest discrepancy was found between reported and cord blood AZT coverage. Due to the short plasma half life of AZT and reported variability in concentrations between subjects elsewhere,28–31 we cannot assume that all women without AZT in their cord blood did not receive any AZT before delivery. However, guidelines at the time of this study recommended AZT from 28 weeks gestation and 3 hourly during labour. Evidence from other settings has demonstrated the presence of therapeutic concentrations of AZT in cord blood when oral doses were administered within 3 hours of delivery.32 Hence in our setting, AZT concentrations should have been reflected in those women who received PMTCT according to protocol during labour. Furthermore, cord blood results of AZT coverage at delivery cannot be extrapolated to infer adequate antenatal coverage.33 There is also a possibility that regimens may not have reflected in women who delivered too quickly after presenting in labour.
Our results show that 1 in 5 women did not test for HIV in pregnancy, and 12% of these women were HIV seropositive at delivery. Several studies have shown regional variation in antenatal testing coverage in South Africa,34,35 with the national antenatal testing average estimated to be 92.7%.36 This is thought to be an overestimate, as testing coverage is reported to be more than 100% in some health districts.37 Routinely collected data from the sites at the time of this study report testing coverage to range between 77%–100%, suggesting that HIV screening in pregnancy could be improved. Modeling exercises have shown that failure to achieve near universal testing has a greater impact on vertical transmission than attrition further along the PMTCT cascade.14 This analysis supports the importance of initial antenatal testing coverage as the single largest contributor to missed opportunities for preventing transmission, but further highlights substantial missed opportunities, as a result of no prophylaxis being taken by women known to be HIV seropositive, suboptimal prophylaxis and seroconversion during pregnancy.
A further limitation of this study is the absence of CD4 cell count data, which would have assisted in the further disaggregation of the standard of care. There is also concern about record keeping as follows: clerical errors, antenatal false-positive tests, or cord blood false-negative tests could have given rise to the 12 women who tested HIV seropositive in pregnancy and had a seronegative cord blood specimen at delivery. All antenatal HIV-seropositive tests are routinely confirmed by a second rapid test in South African service settings, and in our laboratory, the same protocol was performed, with an additional 10% of tests being tested blind for quality assurance. Our specimen collection rate was high at the sites (98.5%), however, logistical challenges resulted in the random loss of 7% of cord blood seropositive specimens which could not be tested for regimen coverage.
Our results show that even in a well-resourced setting, improving HIV screening coverage and ensuring compliance with guidelines for antenatal prophylaxis were the most immediate priorities for reducing vertical transmission. As increasing resources are invested in the rapid expansion of HIV prevention and treatment programs, there is a need for the strengthening of monitoring and evaluation of the effectiveness of these programs.38 Accurate collection and reporting of PMTCT process and outcome indicators remains a challenge because the interventions at each step of the PMTCT cascade occur sequentially through pregnancy and involve more than 1 service and provider, and crucial delivery of treatment at different points in pregnancy. This study, along with evidence from other settings39 shows the value of cord blood surveillance as a feasible tool for confirming both ARV provision and regimen adherence. Cord blood surveillance could be used to assist in the interpretation of program data to improve program effectiveness.
1. Cooper ER, Charurat M, Mofenson L, et al.. Combination antiretroviral strategies for the treatment of pregnant HIV-1-infected women and prevention of perinatal HIV-1 transmission. J Acquir Immune Defic Syndr. 2002;29:484–494.
2. Boulle A, Thompson ML, Laubscher R, et al.. Provincial differences in infant deaths in South Africa—an effect of antiretroviral interventions? South Afr J HIV Med. 2011;12:20–23.
3. Ndirangu J, Newell ML, Tanser F, et al.. Decline in early life mortality in a high HIV prevalence rural area of South Africa: evidence of HIV prevention or treamtent impact? AIDS. 2010;24:593–602.
4. National Department of Health. The 2010 National Antenatal Sentinel HIV & Syphilis Prevalence Survey in South Africa. Pretoria, South Africa: Epidemiology & Surveillance Directorate; 2011.
5. National Department of Health. National HIV and Syphilis Antenatal Sero-Prevalence Survey in South Africa 2008. Pretoria, South Africa:Epidemiology and Surveillance Directorate; 2009.
6. National Department of Health, South African National AIDS Council. Clinical Guidelines: Prevention of Mother-to-Child Transmission. Pretoria, South Africa: National Department of Health; 2010.
7. World Health Organization. Antiretroviral Drugs for Treating Pregnant Women and Preventing HIV Infection in Infants: Recommendations for a Public Health Approach. Geneva, Switzerland: World Health Organization; 2010.
8. Goga A, Dinh TH, Jackson D. Results of an evaluation of effectiveness of the national PMTCT programme at six weeks postpartum, South Africa. Paper presented at: South African AIDS Conference; June 7–10, 2011; Durban, South Africa.
9. Karim QA, Banegura A, Cahn P, et al.. Asking the right questions: developing evidence-based stratgies for treating HIV in women and children. BMC Public Health. 2011;11:388. Available at: http://www.biomedcentral.com/1471-2458/11/388
. Accessed October 18, 2011.
10. McNairy ML, Melaku Z, Barker PM, et al.. Leveraging progress in prevention of mother-to-child transmission of HIV for improved maternal, neonatal, and child health services. J Acquir Immune Defic Syndr. 2011;57:S83–S86.
11. Mofenson LM. Prevention in neglected subpopulations: prevention of mother-to-child transmission of HIV infection. Clin Infect Dis. 2010;50(suppl 3):S130–S148.
12. World Health Organization, UNAIDS, UNICEF. Towards universal access: Scaling up priority HIV/AIDS interventions in the health sector. Progress Report. Geneva, Switzerland: World Health Organization; 2010.
13. Stringer EM, Chi BH, Chintu N, et al.. Monitoring effectiveness of programmes to prevent mother-to-child HIV transmission in lower-income countries. Bull World Health Organ. 2008;86:57–62.
14. Barker PM, Mphatswe W, Rollins N. Antiretroviral drugs in the cupboard are not enough: the impact of health systems' performance on mother-to-child transmission of HIV. J Acquir Immune Defic Syndr. 2011;56:E45–E48.
15. Garrib A, Stoops N, McKenzie A, et al.. An evaluation of the District Health Information System in rural South Africa. S Afr Med J. 2008;98:549–552.
16. Mate KS, Bennett B, Mphatswe W, et al.. Challenges for routine health system data management in a large public programme to prevent mother-to-child HIV transmission in South Africa. PLoS ONE. 2009;4:e5483. Available at: http://www.plosone.org/article/info:doi%2F10.1371%2Fjournal.pone.0005483
. Accessed September 17, 2009.
17. Stringer JSA, Sinkala M, Maclean CC, et al.. Effectiveness of a city-wide program to prevent mother-to-child HIV transmission in Lusaka, Zambia. AIDS. 2005;19:1309–1315.
18. Stringer EM, Ekouevi DK, Coetzee D, et al.. Coverage
of nevirapine-based services to prevent mother-to-child transmission in 4 African countries. JAMA. 2010;304:293–302.
19. Abdullah MF, Young T, Bitalo L, et al.. Public health lessons from a pilot programme to reduce mother-to-child transmission of HIV-1 in Khayelitsha. S Afr Med J. 2001;81:579–583.
20. Médicines Sans Frontières, Infectious Disease Epidemiology Unit (University of Cape Town). Providing HIV Services Including Antiretroviral Therapy at Primary Health Care Clinics in Resource-Poor Settings: The Experience From Khayelitsha. Cape Town, South Africa: Médicines Sans Frontières; 2003.
21. Draper B, Abdullah F. A review of the prevention of mother-to-child transmission programme of the Western Cape Provincial Government, 2003–2004. S Afr Med J. 2008;98:431–434.
22. National Department of Health. Policy and Guidelines for the Implementation of the PMTCT Programme. Pretoria, South Africa: National Department of Health; February 11, 2008.
23. Provincial Government of the Western Cape Department of Health. Annual Report 2009/2010. Cape Town, South Africa: Provincial Government of the Western Cape; 2010.
24. Koal T, Burhenne H, Römling R, et al.. Quantification of antiretroviral drugs in dried blood spot samples by means of liquid chromatography/tandem mass spectrometry. Rapid Commun Mass Spectrom. 2005;19:2995–3001.
25. Johnson L. A Model of Paediatric HIV in South Africa. Cape Town, South Africa: University of Cape Town; 2010.
26. Theron GB, Schoeman J, Carolus E. HIV seroconversion during pregnancy in the Tygerberg region of Cape Town. S Afr Med J. 2006;96:204.
27. Lockman S, Creek T. Acute maternal HIV infection during pregnancy and breast-feeding: substantial risk to infants. J Infect Dis. 2009;200:667–669.
28. Bazzoli C, Bénech H, Rey E, et al.. Joint population pharmacokinetic analysis of zidovudine, lamivudine, and their active intracellular metabolites in HIV patients. Antimicrob Agents Chemother. 2011;55:3423–3431.
29. Dumond JB, Sunila Reddy Y, Troiani L, et al.. Differential extracellular and intracellular concentrations of zidovudine and lamivudine in semen and plasma of HIV-1–Infected men. J Acquir Immune Defic Syndr. 2008;48:156–162.
30. Mirochnick M, Rodman JH, Robbins BL, et al.. Pharmacokinetics of oral zidovudine administered during labour: a preliminary study. HIV Med. 2007;8:451–456.
31. Stretcher BN, Pesce AJ, Frame PT, et al.. Correlates of zidovudine phosphorylation with markers of HIV disease progression and drug toxicity. AIDS. 1994;8:763–769.
32. Bhadrakom C, Simonds RJ, Mei J, et al.. Oral zidovudine during labor to prevent perinatal HIV transmission, Bangkok: tolerance and zidovudine concentration in cord blood. AIDS. 2000;14:509–516.
33. Bennetto-Hood C, Bryson YJ, Stek A, et al.. Zidovudine, lamivudine, and nelfinavir concentrations in amniotic fluid and maternal serum. HIV Clin Trials. 2009;10:41–47.
34. Horwood C, Haskins L, Vermaak K, et al.. Prevention of mother-to-child transmission of HIV (PMTCT) programme in Kwa-Zulu Natal, South Africa: an evaluation of PMTCT implementation and integration into routine maternal, child and women's health services. Trop Med Int Health. 2010;15:992–999.
35. Rispel LC, Peltzer K, Phaswana-Mafuya N, et al.. Assessing missed opportunities for the prevention of mother-to-child HIV transmission in an Eastern Cape local service area. S Afr Med J. 2009;99:174–179.
36. National Department of Health. Annual Report 2009/2010. Pretoria, South Africa: National Department of Health; 2010.
37. Day C, Barron P, Monticelli F, et al.. The District Health Barometer 2007/08. Durban, South Africa: Health Systems Trust; 2009.
38. Diaz T, De Kock KM, Brown T, et al.. New strategies for HIV surveillance in resource-constrained settings: an overview. AIDS. 2005;19(suppl 2):S1–S8.
39. Stringer EM, Chintu N, Levy J, et al.. Declining HIV prevalence among young pregnant women in Lusaka, Zambia. Bull World Health Organ. 2008;86:697–702.