The PEARL study
Polly Clayden, HIV i-Base
David Coetzee from the University of Cape Town presented data from the PEARL study. This study was a four-country evaluation of the effectiveness of PMTCT programmes in Africa. 
PEARL evaluated 43 sites in Zambia, Cote DIvoire, South Africa and Cameroon. The primary outcome was coverage ie the proportion of mother-child pairs with HIV antibody-positive cord blood with confirmed receipt of maternal (detectable NVP in cord blood) and infant (documented) NVP. The study also evaluated AZT and 3TC where appropriate.
All PMTCT sites used a minimum intervention of single dose NVP. Some also used short course AZT plus single dose NVP and/or HAART.
Cord blood was collected anonymously from every delivery between April 2007 and October 2008 and tested for HIV (there was an excellent collection rate of about 98% of consecutive deliveries). If the result was positive, the presence of NVP (AZT and 3TC if applicable) was determined by high performance liquid chromatography.
The investigators also used documented information collected anonymously from patients notes including infant ingestion of NVP.
The investigators collected 28,060 cord blood samples of which 3250 were HIV-positive.
Evaluating both sets of information together they described the coverage cascade (see Table 1), where coverage is defined as maternal and infant ingestion of nevirapine.
Table 1: Coverage cascade
|Positive cord blood||100|
|Information in folder||92|
|Offered HIV test||84|
|Result in folder||74|
|Maternal receipt NVP||71|
|NVP in cord blood||57|
They reported wide variation in coverage across different sites (17%-69%).
Reasons for failed coverage included: no offer of HIV test; HIV test declined; testing declined; test result not given; NVP not dispensed; mother did not adhere and infant not dosed.
The risk of failed coverage increased with younger age: <20 years adjusted OR 1.58 (1.23-2.02) vs >30 years. The association also increased with fewer antenatal visits: 0-1 visits AOR 2.92 (2.22-3.84) vs >6 visits.
Risk of poor maternal adherence was significantly higher in the 26-30 years age group, adjusted OR 1.42 (1.04-1.93) vs >30 years; with greater gravidity, OR 1.62 (1.12-2.34), 4 vs 1; fewer antenatal clinic visits OR 2.98 (2.07-4.48), 0-1 vs >6 visits; vaginal delivery OR 1.51(1.11-2.05) vaginal vs caesarean; and AZT plus NVP prophylaxis, OR 1.42(1.04-1.93) vs NVP alone (HAART was not significantly associated with poorer maternal adherence). All rates are adjusted.
The study included an analysis of PMTCT in the Western Cape, where guidelines have recommended HAART for women with CD4 <200 cells/mm3 and short course AZT plus single dose NVP with CD4 >200 cells/mm3 since 2007.
In this province 12% women received HAART (the investigators used detectable 3TC as a surrogate for HAART) and 47% AZT plus NVP, so overall 59% received standard of care. However, 6% received only NVP, so 65% received at least NVP; 8% received AZT alone and 27% of pregnant women received nothing at all. The investigators noted that CD4 data was not collected in this study (so they would not be able to estimate how many women should have received HAART).
The investigators concluded that failures in the cascade of interventions occur at every step of the way, giving only 50% coverage overall across sites.
Even in settings with two-drug prophylaxis and HAART over a quarter of women are not covered by PMTCT prophylaxis (so also not receiving HAART for their own HIV if indicated).
Interventions must systematically target better performance at each step to maximise their benefits, they wrote.
These data provided an important reality check and show that despite interventions that benefit maternal health and/or prevent transmission to their infants, inadequate roll out and coverage is the norm.
The PEARL Studys primary outcome variable is uptake of PMTCT interventions. As such, PEARL measures programme coverage, but does not assess programme effectiveness at preventing or treating HIV infection. During the conference, reports on national roll out of PMTCT protocols were notable for the absence of infrastructure to detect infant HIV infection, and therefore the effectiveness of PMTCT interventions in practice. Early detection of paediatric HIV is belatedly but rapidly becoming a priority in many countries. We will report on research presented on infant diagnostics in the next issue of HTB South.
WHO PMTCT consultations have also used PMTCT cascades and Lynne Mofenson showed estimations from these in her plenary at the paediatric meeting preceding the IAS conference.  In a typical scenario they estimate that of 1000 mothers, 90% (900) will attend ANC; of these 70% (630) will be counselled and tested for HIV and 50% of this group (315) receive ARVs. If 685 women receive no ARVs, that will mean 25% (172) of their infants will be infected.
In this estimation, overall transmission rates, if all women with CD4 <200 cells/mm3 receive HAART and the remainder receive either single dose NVP, short course AZT plus single dose NVP or HAART, would be 19.7%, 18.1% and 17.6% respectively.
With improved uptake, if 96% (960) mothers attended ANC, 99% (950) were counselled and tested and 98% (931) received ARVs, that would mean 69 mothers receive no ARVs and 17 infants would be infected. In this estimation the overall transmission rates would be 9.1%, 4.5% and 3.1% for single dose NVP, AZT plus single dose NVP and HAART respectively.
Coceka Mnyani and coworkers showed data from January to December 2008 from the Soweto programme on which the improved figures were based.  In March 2008 short course AZT was introduced in addition to single dose NVP for women with CD4 >200 cells/mm3. During this period, 30180 women attended ANC of which 99% (29968) accepted HIV testing and 29% (8774) tested HIV-positive. Of these 5704 mothers and 6641 infants received AZT plus single dose NVP and 96% (8137/8514) elected to formula feed. A total of 324/5572 infants were HIV-infected giving a transmission rate of 5.8%. The transmission rate was 7.2% during the period when single dose NVP was used alone and 4.4% using AZT plus NVP, OR 1.67 (95%CI, 1.24-2.3), p=0.0004.
As Lynne Mofenson explained, Programme efficacy is as much related to the PMTCT cascade as the specific regimen.
- Coetzee D et al. Evaluation of PMTCT coverage in four African countries: the PEARL study. 5th IAS Conference, Cape Town.19-22 July 2009. Oral abstract WELBD101.
- Mofensen L et al. Update and controversies on prevention of mother-to-child transmission. 1st International Workshop on prevention of mother-to-child transmission. 1st International workshop on HIV paediatrics Cape Town, 17-18 July, 2009. Session 5.
- Mnyani CN et al. Reducing PMTCT in practice: results from a successful model of a PMTCT program in a high prevalence HIV urban African setting. 5th IAS Conference, Cape Town.19-22 July 2009. Poster abstract TUPEC048.