Recent changes to the WHO treatment guidelines for HIV-infected infants  are based on the evidence that early initiation of antiretroviral therapy (ART) prolongs their lives . It is difficult, however, to reconcile this recommendation with the experience reported by many prevention of mother-to-child transmission (PMTCT) of HIV programmes that postnatal HIV testing of infants is relatively uncommon . Most HIV-infected children in Africa still present with severe symptoms and die before they have the chance to start ART or even cotrimoxazole prophylaxis. Infants who are HIV infected perinatally are about 17 times more likely to die by 18 months than infants who are uninfected . As much as this advocates for better PMTCT of HIV services, it also calls for better approaches for identifying infected children before they become symptomatic or die from acute opportunistic infections.
Various approaches have been developed to enhance the rates of testing at 6 weeks but with little success. WHO devised an algorithm within the Integrated Management of Childhood Illnesses to identify suspected HIV infected children at primary healthcare clinics ; this has only partial sensitivity and identifies only a proportion of infected children . WHO also simplified guidelines on early infant diagnosis and criteria for ART initiation in an attempt to remove barriers to initiate treatment .
We previously reported a surveillance approach that involved anonymous HIV testing of all infants at immunization clinics, irrespective of whether the mother had been part of a PMTCT programme . High rates of infant HIV infection were identified. We questioned whether the same approach, but not anonymous, would allow for early identification of infected infants. Screening for diseases is justified if a condition is common, and a diagnostic test and treatment are available to treat that condition. Recent reports have called for routine universal HIV testing of high-risk adults attending sexually transmitted infection clinics in high HIV prevalence settings in Africa  and modelled the potential benefit of lifelong antiretroviral treatment for primary prevention . In the United States, Europe and private healthcare facilities in South Africa, screening of pregnant women is routinely offered for neural tube defect (prevalence in South Africa in the ratio of 1: 1000–1: 10 000), Down's syndrome (1: 600–1: 1000) and congenital hypothyroidism (1: 3000–1: 3500). Here, we report the acceptability and feasibility of routinely offering HIV testing of all infants attending immunization clinics at three primary healthcare facilities in KwaZulu Natal, South Africa.
Trained counsellors invited mothers bringing their infants for immunization at three primary healthcare facilities in KwaZulu Natal to give permission for their infants to be tested for HIV. KwaZulu Natal is the largest Province in South Africa with an HIV prevalence rate of approximately 39% amongst women attending antenatal clinics (ANCs) . Mothers who agreed were also asked about their own prior experiences of HIV testing and PMTCT of HIV services. Inclusion for testing was limited to infants attending any of the first three immunizations, that is, at 6, 10 or 14 weeks of age. The testing method, return appointment and provision of results were explained to groups of mothers while waiting for the immunization.
If a mother gave signed consent for her infant to be tested then blood was collected from the infant by heel prick onto filter paper. Filter paper samples were dried and transported to the laboratory once per week. Dried blood spot (DBS) samples were labelled with a sample identification number that was linked to infant details on a separate record sheet. Infant DBS samples were first tested for HIV antibodies (Vironostika HIV Uni-Form II plus O; Biomerieux, the Netherlands) . Samples that were found to have HIV antibodies were then tested for HIV by DNA PCR (AMPLICOR HIV-1 DNA test version 1.5; Roche Diagnostics, Indianapolis, Indiana, USA) [13,14].
Prior to requesting consent, the implications of potential results were explained:
1. The infant's DBS has HIV antibody present. At this age, HIV antibodies will be maternal antibodies and will therefore indicate maternal infection. In this situation:
a. the mother may already know that she is positive (ANC testing or testing prior to the pregnancy);
b. she may have tested negative at the ANC and believed that she was HIV uninfected, she might have been in the window period or became HIV infected after ANC testing;
c. she may not have known her HIV status prior to this test.
2. If HIV antibody is present, the infant's DBS will be tested for HIV by DNA PCR. This test may be positive or negative and will indicate the HIV status of the infant.
Staff members were trained to clearly explain each of these scenarios and how to respond to questions. Mothers were also given the opportunity to individually talk more about testing and possible results. Counsellors were trained on how to counsel mothers after receiving the results of the HIV test in each of these scenarios. Infants who were not accompanied by the mother were not included even if the caregiver was the legal guardian.
Mothers were given an appointment date about 2 weeks later to return for results. If mothers were unable to attend the appointment, she could also speak to counsellors at the next scheduled immunization date. Three counsellors were present at each clinic, each day. A simple system using a password chosen by the mother at the time of testing was used to ensure confidentiality when releasing results in case the child was brought to the clinic by another caregiver at a later date.
At post-test counselling, if antibodies were present on the infant DBS, the mother was advised that she herself was probably HIV infected and was taken across to the routine HIV services in the same clinic for confirmation. If HIV DNA testing of the DBS was also positive then counsellors advised mothers that their infant was probably infected and similarly taken across for confirmatory testing; prophylactic cotrimoxazole was also started. The participating clinics were not ART initiation sites but rather ‘stepdown’ facilities that were able to perform CD4 cell testing and dispense antiretroviral drugs once adults or children were initiated. Mothers and infants who were confirmed to be HIV infected were immediately referred for assessment at the nearby ART initiation sites.
All mothers who had brought their infants for immunizations, irrespective of whether they had agreed to HIV testing of their infants, were also invited to participate in exit interviews following the immunization. Information collected through the exit interviews was not linked with earlier interviews or the status of mother or infant. Questions included how they felt when given the option of their infants being tested and what were the advantages or disadvantages of infant testing.
In a prior surveillance project, there was 87% acceptance by mothers for anonymized screening of their infants that adopted the same laboratory-testing algorithm . In this study, to detect a similar 85% uptake, ±5% at the 95% confidence interval (CI) level, we needed to approach 195 mothers with the offer of infant HIV testing. About 200 mothers were therefore approached at each clinic.
The primary outcome was acceptability of the offer of infant testing. Acceptability was assessed first by the proportion of women who agreed to testing of their infants and secondly by the number of mothers who returned for results. Estimates of acceptability and feasibility parameters are reported as percentages (95% CIs). Additional aspects explored included the time interval until mothers returned for results if not on the scheduled appointment date, and whether return rates varied by HIV status of the mother (either according to antibody status of the infant or by maternal self-reported HIV status). Simple frequency distributions were calculated for responses collected from the exit interviews. Associations between categorical variables were assessed using χ2 tests. Although HIV prevalence rates would be available, this was not the primary outcome, and sample size was not based on this estimation. SPSS version 15.0 (SPSS Inc., Chicago, Illinois, USA) was used to analyse the data. The criterion for statistical significance was set at α is equal to 0.05.
The study was approved by the Biomedical Research Ethics Committee of the University of KwaZulu-Natal and also by the KwaZulu Natal Department of Health.
Between November 2007 and February 2008, 646 mothers of infants brought to three periurban primary healthcare facilities in KwaZulu-Natal were approached by study counsellors regarding HIV testing of their infants. The average age of the mothers was 24.8 years (SD 8.4 years) and median age of the infants was 7.7 weeks (Q1, Q3; 6.4, 12.3). Of these, 584 (90.4%, 95% CI 87.8–92.5) mothers agreed to testing (giving signed written consent), and DBS samples were collected from their infants on the same day. Mothers' details and self-reported experiences of HIV testing are included in Table 1. About 98% women reported having ever tested for HIV and almost all stated that they had received results. Two hundred and sixty-six women reported having swallowed nevirapine in the labour ward even though only 233 self-reported being HIV infected. Amongst mothers who reported being HIV infected, about 70% said their infants received single-dose nevirapine, whereas almost a quarter did not know whether their child had received anything.
Of the 584 mothers who agreed to infant testing, 332 (56.8%, 95% CI 52.7–60.9) subsequently returned for the results of their infants. Of those who returned, 160 (48.2%) came back on the scheduled appointment date. Of those mothers who came at a time other than the scheduled date, 80% came about 4 weeks after the first attendance. Data on both HIV antibody status of DBS and date of scheduled appointment for results and post-test counselling were available on 520 infants and mothers. There was no significant difference in return rates (P = 0.092) according to antibody status of the DBS (Table 2). However, women who self-reported being HIV positive were more likely to return than women who self-reported being HIV negative (P = 0.001). Combining the rates of initial uptake and returning for results, 51% (332/646) of infants and mothers were informed of, or had their HIV status confirmed through this approach.
Overall, 247 of 584 (42.3%) infant DBS samples had HIV antibodies present which is comparable to estimates of antenatal maternal HIV prevalence. Of these, 54 of 247 (21.9%) samples were positive for HIV DNA by PCR. This equates to 9.2% (54/584) of all infants tested. Results of antibody testing were equivocal in 20 DBSs. Amongst women who self-reported being HIV uninfected, 7.2% of their infants had HIV antibodies indicating that these mothers were in fact HIV infected. These mothers may have been in the window period when tested antenatally or had become HIV-infected since then or had opted to tell the team that they were uninfected. The vertical HIV transmission rate from these mothers to their infants was 38%.
Women's perspectives on being offered testing of their infants are summarized in Table 3. The vast majority stated that they were generally comfortable with being offered testing of their infant (Table 3). A small number were surprised by the presence of counsellors at the clinic and the offer of testing. Reasons given for not testing (Table 3) included anxiety, fear and indecision. A small number commented that their infant had been tested elsewhere and the result was pending. When asked about the advantages of testing (Table 3), confirmation of HIV status was widely acknowledged and about half of women recognized that it would enable opportunities for ART. Only about one-quarter of mothers realized that confirming the infant's and her own HIV status would directly inform what would be the best infant feeding practice; even fewer mothers knew about cotrimoxazole prophylaxis. Women described a number of perceived disadvantages. About a quarter of women commented that the prospect of testing and the implications were ‘frightening’, or the process was ‘too quick’, or that they were concerned that it revealed HIV status. Five hundred and eighty of 646 (89.8%) mothers who were interviewed said they would recommend testing to others.
Counsellors were consistently positive about the testing strategy stating that it had several benefits for infants and mothers. They commented that, in general, mothers were ‘excited’ by the opportunity to learn their infant's status and that the approach was ‘important for all of South Africa’. However, they commented that there was little space and privacy in the clinics and that it was hard to look after an infant if the mother started crying. Other difficulties reported included mothers who refused subsequent care or who were angry or confrontational after being post-test counselled; in particular, the complexities when HIV positive mothers had not disclosed their status to their partner or family.
In this study, routine HIV testing of infants attending primary healthcare clinics for immunization was acceptable and feasible. If implemented as the standard of care at primary health care clinics, more than half of infants and mothers would have known their status at about 6–10 weeks of age after which they could gain access to a continuum of care. With better awareness within communities and a more ‘routine’ approach being offered by staff, this number could increase. In this study, more than half of the peripartum-infected infants were thereby identified and referred for HIV treatment. This contrasts sharply with the experience of PMTCT of HIV programmes in which routine HIV testing of infants is achieved in only 8% HIV-exposed infants . Routine implementation would also serve to identify, and refer for treatment and care, women who previously thought they were HIV uninfected but who became HIV infected during pregnancy. In high HIV prevalence settings, this approach has the potential for significantly increasing the number of mothers and children gaining access to HIV treatment and care.
We gauged acceptability by the number of women who accepted testing of their infants and, perhaps more importantly, the number of women who returned for the results. About 90% of women agreed to testing and 57% returned for the results; there was no significant difference in return rates (P = 0.092) according to DBS antibody status, suggesting that mothers decisions were not significantly influenced by their actual HIV status (Table 2).
Mothers not returning on the scheduled date returned about 4 weeks later, which may have coincided with the next immunization visit. Considerations such as money or other commitments might have precluded them from coming for the ‘extra’ post-test counselling visit. Other mothers may have chosen to come back on the next immunization visit to disguise their return for results.
We implemented an opt-out HIV pretest counselling approach, but with the option of one-on-one counselling, and individualized post-test counselling. Although immunizations are offered every day of the week, mothers still informally adhered to certain patterns of attendance, resulting in certain days of the week being inundated. Three counsellors were allocated to each clinic who together were able to cope with the workload.
Although this study suggests that this would be a feasible approach to achieve early infant diagnosis, the demands on the counsellor teams were more complex than immediately evident. Considerable planning and training of counsellors was undertaken to ensure that each were able to manage the various anticipated scenarios. These included mothers believing that they were HIV uninfected and therefore not taking any intervention, only to learn that they and perhaps their infants were actually infected. Such scenarios are not included in routine HIV counsellor training and hence our supervisory team carefully monitored how counsellors were coping. If such a strategy was scaled up, it would be helpful to assess whether a lesser intensity of pretest and post-test counselling would result in lower uptake of testing or return for results or increased stress for the mother. In the study, we implemented a simple password system that protected the confidentiality of the infant's results and therefore the mother's HIV status. If taken to scale, similar consideration would be needed in case the infant was brought for a later immunization by another caregiver.
The testing approach described can only be used in the first 1–2 months of an infant's life when maternal antibodies are present and a high level of sensitivity could be expected. At this age, the approach is efficient and could bridge the gap between antenatal services and opportunities for postnatal care. In older infants and in those who were breastfed, it would be necessary to detect HIV DNA by PCR without any consideration of antibody status; this would have significant cost implications and would limit its utility in determining maternal HIV status.
An ethical question we faced was whether our team had an obligation to seek out the infants and mothers who were found to be HIV infected but who did not return for results. We did not do so, principally because in our explanation to mothers we indicated that we would provide the results only at the clinics, and that we would not solicit addresses or phone numbers. Had we done so and included this in our pretest information, mothers may have felt vulnerable and not agreed to testing. Our approach was consistent with current HIV testing at public health facilities in South Africa namely, patients return for results, as and when they choose.
Mothers also expressed their acceptance of the strategy in the exit interviews. Most mothers felt comfortable with the idea of testing, understood the value of early diagnosis to gain access to treatment and would recommend it to others. Some may have felt that testing of their infant was less intimidating than testing again themselves, yet knowing that it would infer their own HIV status. Only a small proportion expressed anxiety and this was reflected in the small proportion that refused testing. Of concern, few mothers acknowledged the value of knowing their own HIV status in guiding decisions on infant feeding practices.
There are several limitations to the study. The cost effectiveness of the approach if implemented at scale was not estimated. The approach capitalized on a strength of a health system and utilized a modest number of additional staff. Scaling up such an approach would result in many more HIV tests being processed with their associated costs. However, opting not to screen young infants only defers the need for testing until critically ill children present to health services and need to be diagnosed. The approach could be improved by using a HIV antibody rapid test on infants to first determine exposure; mothers could be advised immediately and there would be no need for return appointments or coding systems of results. However, this was a study to establish acceptability and feasibility of a principle. If considered a useful approach and investment of resources, there would be ways to adapt and simplify the process. The study did not follow HIV-positive mothers and infants to determine whether they ultimately received ART or track those who did not come back. The study also did not consider the many ways that HIV-related services might be more user friendly to promote better long-term follow-up and continuity of care, issues that are important but were beyond the scope of this study. Rather, as with the preliminary studies exploring how to increase HIV counselling and testing coverage such as the concept of opt-out testing, proof of acceptability is the first step. If, as at present, only a small number of young infants are being detected through PMTCT of HIV programmes then only a small number of infants can ever receive early treatment and care. In high prevalence settings, and with the awareness of considerable infant infections that might be occurring due to undetected incident infection of women during pregnancy , it begs the question whether routine testing of all infants is preferable to focussed testing of infants through PMTCT of HIV programmes. Multiple and recurring opportunities are needed to enable mothers to learn about their status and gain access to care. It is about making HIV testing a routine practice and removing barriers to early testing of HIV exposed infants.
Testing of all infants attending immunization clinics in this study was acceptable and feasible. In high prevalence settings, using immunization clinics as an opportunity to offer HIV testing of young infants and mothers could enable both to receive early treatment and avert early morbidity and mortality. Mothers demonstrated their understanding of these potential gains and showed that they are not the ones hesitating to make this a reality.
We are, first and foremost, grateful to the all mothers of infants who attended the participating clinics regardless of whether they agreed to HIV testing or not. We also wish to thank the staff in the immunization/child health clinics and also the HIV teams at referral sites. We appreciate the support of the Provincial and District Departments of Health and their willingness to evaluate innovative approaches to improve care of the mothers and children of KwaZulu Natal. We are grateful to the staff at Global Clinical Virology Laboratories, Durban for the quality of their work and flexibility in working with our team.
The study was funded by the Department for International Development, United Kingdom whom we thank for their support and ongoing interest.
Disclaimer: N.R. is a staff member of the WHO. The author alone is responsible for the views expressed in this publication and they do not necessarily represent the decisions, policy, or views of the WHO.
1. World Health Organization. Report of the WHO Technical Reference Group. Paediatric HIV/ART Care Guidelines Group Meeting; 10–11 April 2008; WHO Headquarters, Geneva
. Geneva: WHO; 2008. p. 4.
2. 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.
3. UNICEF. Children and AIDS. Third Stocktaking Report 2008
. New York, USA: UNICEF; 2008.
4. Rollins NC, Becquet R, Bland RM, Coutsoudis A, Coovadia HM, Newell ML. Infant feeding, HIV transmission and mortality at 18 months: the need for appropriate choices by mothers and prioritization within programmes. AIDS 2008; 22:2349–2357.
5. Qazi SA, Muhe LM. Integrating HIV management for children into the Integrated Management of Childhood Illness guidelines. Trans R Soc Trop Med Hyg 2006; 100:10–13.
6. Horwood C, Liebeschuetz S, Blaauw D, Cassol S, Qazi S. Diagnosis of paediatric HIV infection in a primary healthcare setting with a clinical algorithm. Bull World Health Organ 2003; 81:858–866.
7. World Health Organization. Antiretroviral therapy of HIV infection in infants and children in resource-limited settings: towards universal access. Recommendations for a public health approach
. Geneva: WHO; 2006.
8. 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.
9. Bunnell R, Cherutich P. Universal HIV testing and counselling in Africa. Lancet 2008; 371:2148–2150.
10. Granich RM, Gilks CF, Dye C, De Cock KM, Williams BG. Universal voluntary HIV testing with immediate antiretroviral therapy as a strategy for elimination of HIV transmission: a mathematical model. Lancet 2009; 373:48–57.
11. South Africa National Department of Health. National HIV and syphilis antenatal prevalence survey, South Africa 2006
. Pretoria, South Africa: Department of Health, South Africa; 2007.
12. Biggar RJ, Miley W, Miotti P, Taha TE, Butcher A, Spadoro J, Waters D. Blood collection on filter paper: a practical approach to sample collection for studies of perinatal HIV transmission. J Acquir Immune Defic Syndr Hum Retrovirol 1997; 14:368–373.
13. Sherman GG, Cooper PA, Coovadia AH, Puren AJ, Jones SA, Mokhachan M, Boltan KD. Polymerase chain reaction for diagnosis of human immunodeficiency virus infection in infancy in low resource settings. Pediatr Infect Dis J 2005; 24:993–997.
14. Patton JC, Akkers E, Coovadia AH, Meyers TM, Stevens WS, Sherman GG. Evaluation of dried whole blood spots obtained by heel or finger stick as an alternative to venous blood for diagnosis of human immunodeficiency virus type 1 infection in vertically exposed infants in the routine diagnostic laboratory. Clin Vaccine Immunol 2007; 14:201–203.
15. UNICEF. United Nations Children's Fund, Joint United Nations Programme on HIV/AIDS and the World Health Organization. Towards universal access: scaling up priority HIV services for women and children in the health sector – progress report 2008
. New York, USA: UNICEF; 2008. p.24.
16. Lu L, Legwaila K, Motswere C, Smit M, Jimbo W, Creek T. HIV incidence in pregnancy and the first postpartum year and implications for PMTCT programs
[abstract #91]. 16th Conference on Retroviruses and Opportunistic Infections
; 8–11 February 2009; Montreal, Canada; 2009.
© 2009 Lippincott Williams & Wilkins, Inc.