Bulletin of the World Health Organization

Implementing early infant diagnosis of HIV infection at the primary care level: experiences and challenges in Malawi

Queen Dube a, Anna Dow b, Chawanangwa Chirambo a, Jill Lebov b, Lyson Tenthani c, Michael Moore a, Robert S Heyderman a, Annelies Van Rie b & for the CHIDEV study team

a. Malawi–Liverpool–Wellcome Trust Clinical Research Programme, Queen Elizabeth Central Hospital, Blantyre, Malawi.
b. Department of Epidemiology, University of North Carolina Gillings School of Global Public Health, 135 Dauer Drive, Campus Box 7435, Chapel Hill, NC 27599, United States of America.
c. HIV Unit, Malawi Ministry of Health, Lilongwe, Malawi.

Correspondence to Annelies Van Rie (e-mail: vanrie@email.unc.edu).

(Submitted: 31 January 2012 – Revised version received: 05 May 2012 – Accepted: 11 May 2012 – Published online: 18 June 2012.)

Bulletin of the World Health Organization 2012;90:699-704. doi: 10.2471/BLT.11.100776


According to the recommendations of the World Health Organization (WHO), infants known to have been exposed to the human immunodeficiency virus (HIV) should undergo a virological test for infection at 4 to 6 weeks of age.1,2 Antiretroviral therapy (ART) should be initiated upon diagnosis of HIV infection in children aged less than 24 months.3 However, implementing programmes for such early infant diagnosis and treatment has proved challenging.47

In Malawi, 13.8% of the children born to HIV-positive mothers in 2009 were themselves HIV-positive as infants,8 but only 29% of those in need of ART received such treatment.9 National guidelines in Malawi recommend that infants exposed to HIV be tested by polymerase chain reaction (PCR) for the detection of viral deoxyribonucleic acid (DNA) at 6 weeks of age wherever the facilities and resources for these assays are available.10 In Malawi, as in several other countries, most early infant diagnosis is hospital-based and few infants receive ART after testing.8,11

We report here the experiences and challenges encountered during implementation of early infant diagnosis in two community health centres in Malawi.

Setting and approach

As part of recruitment procedures for a study assessing the impact of HIV infection on child neurodevelopmental processes, early infant diagnosis services were established at two health-care facilities in Blantyre. One of these facilities, the Zingwangwa Health Centre, is an urban primary care centre that does not initiate or administer ART but that refers those in need of ART to a hospital. The other study facility, the Mlambe Mission Hospital, is located in a semi-rural area and serves as a primary and secondary care centre with on-site ART services. Both study facilities run programmes for the prevention of mother-to-child transmission (PMTCT) of HIV and both recommend breastfeeding and cotrimoxazole prophylaxis for all HIV-exposed infants.

As pregnant women and mothers with infants attended the study centres for PMTCT and postnatal visits, study staff explained to them the importance of early infant HIV diagnosis and the benefits of early ART for infants. To further increase awareness of early infant diagnosis, posters and brochures were distributed and community sensitization was performed. Each HIV-positive mother was given an appointment card. At 6 weeks of age (or at the earliest visit thereafter), each infant of an HIV-positive mother was referred for cotrimoxazole prophylaxis and tested for HIV DNA. Permission for home visits was requested from the infant’s mother, who was asked to return with the infant, for a follow-up, 4 weeks later.

A sample of blood was collected from each HIV-exposed infant via a heel prick. These samples were transferred to Protein Saver 903® cards (Whatman Ltd, Piscataway, United States of America), which were then dried and individually packaged with desiccant sachets before being transported to the Malawi–Liverpool–Wellcome Trust research laboratory in Blantyre. At the laboratory, the dried blood spots on the cards were tested for HIV DNA using version 1.5 of the Amplicor® HIV-1 DNA test kit (Roche, Basel, Switzerland), which is based on a PCR. The laboratory included internal quality control procedures for each 4-weekly testing batch and participated in an external quality control programme run by the United States Centers for Disease Control and Prevention.

At follow-up, mothers of the children found to be PCR-negative for HIV DNA were counselled on how to minimize transmission risk and the importance of repeat testing while breastfeeding. The mothers of the PCR-positive infants were counselled on the importance of early infant ART and their infants were referred to the nearest ART clinic. Community health staff, who were reimbursed by study funds, were asked to trace the mothers of PCR-positive infants who did not return for the scheduled follow-up.

Attempts were made to retest each child who was initially found PCR-positive. In most cases, the same PCR-based assay as used initially was employed to test a second blood spot on the child’s 6-week sample card. However, if the relevant sample card could not be located, the HIV ribonucleic acid (RNA) in a second blood sample collected from the child was quantified, at the University of North Carolina’s project laboratory in Lilongwe, using version 1.5 of the Amplicor® HIV-1 Monitor (Roche). If the retest sample gave a positive result, the child was considered truly positive. If the retest sample appeared negative, another blood sample was collected from the child and either tested for HIV DNA or assayed for HIV RNA. If this sample was found negative, the child’s result was considered a false positive. The final result for any child who was found first positive and then negative and then was unavailable for a second retest was recorded as inconclusive.

Both the University of Malawi College of Medicine Research and Ethics Committee and the University of North Carolina at Chapel Hill Institutional Review Board approved the study protocol.


Between January 2008 and June 2010, 7234 women, of whom 1214 (16.8%; 95% confidence interval, CI: 15.9%–17.6%) were HIV-infected, participated in PMTCT activities at either of the two study centres. Although 920 (75.8%) of the infants of the HIV-positive mothers presented for early infant diagnosis, consent for the necessary testing was only obtained from the mothers of 869 (94.5%; 95% CI: 93.0%–95.9%) of these children. In consequence, only 71.6% (95% CI: 69.0%–74.1%) of the HIV-exposed infants seen at the two study centres during the study period were tested for HIV DNA. Although all but one of the 50 women who declined to give consent for the infant testing said that they wanted to get permission from their spouses, none of these women ever returned for early infant diagnosis. Self-reported PMTCT coverage, for both mothers and infants at the two study sites, was 92.4%.

Overall, 126 (14.5%) of the 869 infants tested for HIV DNA gave a positive initial result. The original sample cards for 61 of the PCR-positive children were relocated and another blood spot from each of these cards was checked for HIV DNA in the PCR-based assay. Of the 61 children who were retested in this way, 41 gave a positive result on retesting and 20 gave a negative result. Although fresh blood samples were collected from 17 of the 20 children found PCR-negative on retesting, only two of these 17 fresh samples gave a positive result when checked for HIV DNA (5 samples) or assayed for HIV RNA (12 samples). The original sample cards for 65 of the children who were initially found positive for HIV DNA had been thrown away after the initial testing. Fresh blood samples were obtained from 45 of these 65 children and all 45 of these samples gave a positive result when assayed for HIV RNA. Confirmatory testing was not possible for the 20 children who were lost to follow-up. Among the 106 infants with any confirmatory testing, 88 were confirmed as HIV-infected and 15 were considered HIV-negative; the results for the remaining three children, who could not be retested fully, were inconclusive. Overall, 14.6% of the 103 children who were retested fully were found to have been falsely positive in the initial round of testing. The positive predictive value of the assay used in this initial round, which was based on the detection, by PCR, of HIV DNA in dried spots of blood from infants aged about 6 weeks, was therefore 85.4%.

Only 521 (60%) of the mothers of tested infants returned to the study centres to receive the results of the initial testing. However, compared with the other mothers of tested infants, a mother of a child found HIV-positive when first tested by PCR was significantly more likely to have received the results of her infant’s test (87.3% versus 55.3%; P < 0.0001). The initiation of ART was recorded for 51 children, who represented 58% of all of the children with confirmed HIV infection.


Important lessons can be drawn from our experience (Box 1). Routine early infant diagnosis at the primary care level in a resource-poor setting is feasible but challenging, even when well supported by research funds. A strength of our approach was the integration of the early infant diagnosis programme into the existing PMTCT services, which facilitated delivery of messages about the importance and availability of early infant diagnosis and early ART to the target population of HIV-infected pregnant women. The scheduling of appointments for early infant diagnosis so that they coincided with routine visits for the vaccination of infants at 6 weeks of age eliminated the need for additional visits for the sole purpose of testing infants for early diagnosis. These approaches resulted in the HIV testing of more than two thirds of the HIV-exposed infants seen at the study centres.

Box 1. Summary of main lessons learnt

  • Integration of early infant diagnosis services with existing PMTCT services is feasible and facilitates delivery of messages about the importance of early infant diagnosis to HIV-infected women.
  • Despite active tracing of HIV-infected children, there are high rates of loss to follow-up at every stage of the early infant diagnosis programme.
  • Improved diagnostic tools with point-of-care capacity are necessary to allow for more streamlined testing programmes with the potential for better linkage to infant antiretroviral treatment.

Our experiences with the early infant diagnosis programme also highlighted many challenges. Unfortunately, since HIV exposure was not documented on the standard infant health passports and there was limited privacy in the clinics, it proved impossible to integrate the services for early infant diagnosis with vaccination clinics. The high percentage of false positive results in the initial round of HIV tests (14.6%) was unexpected, given the previously reported high specificity of the PCR-based assay that was used.12,13 Further investigations indicated that laboratory contamination, resulting from the manual manipulation of the sample cards, was the most likely cause of the low specificity. This observation underscores the need for stringent quality control and confirmatory testing. The long delay between the collection of the initial blood samples and the availability of the results of confirmatory testing complicated the communication of an initial positive result and the management of the children with such a result. Knowing the overwhelming benefits of early ART, we did not withhold such treatment until the confirmatory results were available. We carefully counselled the mothers of infants who were initially found PCR-positive while realizing that a positive result on an infant’s HIV test is devastating news and that the occurrence of false positive results can undermine community trust in early infant diagnosis.

The use of dried blood spots, which can be collected at any clinic and transported without refrigeration to a laboratory equipped for PCR-based analysis,12 has been an important step towards universal access to early infant diagnosis. However, by eliminating the need to transport samples or to return to a clinic to retrieve the test results, a rapid point-of-care test for infant HIV diagnosis14 could still greatly enhance the operational feasibility of early infant diagnosis programmes in resource-poor settings.

A unique aspect of our approach was the active tracing of HIV-positive infants, which resulted in the receipt of the infants’ HIV test results by the families of 87% of these infants. Tracing is unlikely to be possible without dedicated funds. Sustainable strategies to improve the communication of test results to the caregivers of infants in routine settings need to be explored. Infant treatment programmes also need to be scaled up to maximize the number of HIV-infected infants who initiate ART – the ultimate goal of early infant diagnosis programmes.

While promising, the 71.6% coverage with early infant diagnostic services that was recorded in the present study indicates that one in four HIV-exposed infants did not access such services. Additionally, the coverage recorded here may be an underestimate of the true value, since the 28% of women in Malawi who deliver at home, with the help of traditional birth attendants,15 may never be offered early infant diagnostic testing (Fig. 1). The involvement of traditional birth attendants in early infant diagnosis programmes and the integration of early infant diagnosis into vaccination clinics with community outreach could complement any clinic-based activities. The routine seeking of permission for infant testing from both the mother and her partner should be explored, as permission from the partner was the primary reason that women who presented for early infant diagnosis gave for not proceeding with the testing.

Fig. 1. Opportunities to improve early infant diagnosis of HIV infection, based on experiences at two primary care centres, Malawi, 2008–2010
Fig. 1. Opportunities to improve early infant diagnosis of HIV infection, based on experiences at two primary care centres, Malawi, 2008–2010
ART, antiretroviral therapy; EID, early infant diagnosis; HIV, human immunodeficiency virus; PMTCT, prevention of mother-to-child transmission.

Although careful documentation and a large sample size were important strengths of this evaluation, the research setting limits the generalizability of our findings. The substantial human resources and financial support that were available in the present study, through linkage with the research team, probably created a “best-case scenario” for a setting that is usually resource-poor. Even under these conditions, however, the implementation of early infant diagnosis at the primary-care level was challenging, with the dropouts that occurred at every step diminishing the number of HIV-infected children who gained access to ART. Our experience suggests that, to maximize the benefits of early infant HIV diagnosis programmes, a simple, affordable and highly specific point-of-care test for infant HIV diagnosis and better linkage to care are both needed.


The Malawi–Liverpool–Wellcome Trust Research Programme receives funding from the Wellcome Trust. The authors are grateful to all study and clinical staff at Mlambe Mission Hospital, Zingwangwa Health Centre and Queen Elizabeth Central Hospital, and to all participating women and children.


Funding was provided by the United States National Institutes of Health/National Institute of Child Health and Human Development, via grant R01 HD053216.

Competing interests:

None declared.