How evidence based are public health policies for prevention of mother to child transmission of HIV?

BMJ 2013; 346 doi: (Published 20 June 2013) Cite this as: BMJ 2013;346:f3763
  1. Philippe Van de Perre, professor 123,
  2. Thorkild Tylleskär, professor4,
  3. Jean-François Delfraissy, professor5,
  4. Nicolas Nagot, researcher123
  1. 1INSERM U 1058 (Infection by HIV and by agents with mucocutaneous tropism: from pathogenesis to prevention), 34394 Montpellier, France
  2. 2Université Montpellier 1, 34090 Montpellier, France
  3. 3CHU Montpellier, Département de Bactériologie-Virologie et Département d’Information Médicale, 34295 Montpellier, France
  4. 4Centre for International Health, University of Bergen, 5020 Bergen, Norway
  5. 5Agence Nationale de Recherches sur le Sida et les Hépatites Virales, Paris, France
  1. Correspondence to: P Van de Perre p-van_de_perre{at}

WHO is due to issue new guidelines on preventing transmission of HIV from mother to child at the end of June. Philippe Van de Perre and colleagues argue that recommendations have gone ahead of the evidence and call for a more cautious approach

Most public health authorities use recommendations formulated by United Nations agencies when they renew or update national public health guidelines. UN recommendations thus affect the lives of millions of people worldwide and should be based on robust scientific evidence, but that is not always the case. The World Health Organization’s recommendations for the prevention of mother to child transmission of HIV have long been a source of debate, and this has intensified with the forthcoming guidelines expected to advocate lifelong antiretroviral treatment for all HIV infected women, regardless of CD4 count.1 In this article, we discuss WHO’s arguments for adopting this strategy and the concerns we have.

Background of WHO recommendations

WHO has been issuing guidelines for the prevention of mother to child transmission of HIV since the mid-90s, as good evidence has emerged from trials. In 1994, a trial showed that giving zidovudine to selected HIV-1 infected women from the earliest stages of pregnancy with intravenous infusion during labour and delivery, and to their babies, reduced mother to child transmission by nearly two thirds.2 The US rapidly translated this good news into public health recommendations and WHO followed suit shortly after.3 4 However, most health professionals in low resource settings thought the strategy was not feasible and wanted to know whether a short course of oral zidovudine given to mothers during pregnancy and delivery and to exposed newborns would be effective.

By the end of the decade, randomised trials provided compelling evidence that perinatal antiretroviral prophylaxis in mothers and their neonates with zidovudine, nevirapine, or a combination of zidovudine and lamivudine was effective.5 As a result, WHO in 2000 recommended several prophylactic schemes for preventing perinatal transmission of HIV.6 These guidelines were further updated in 2004 and included the use of combination prophylaxis.7

Breastfeeding and HIV

The early 2000s were also dominated by a renewed scientific interest in breastfeeding and its effect on the transmission of HIV. Many HIV infected women in the developing world have no alternative to breastfeeding. Based on the observation that perinatal prophylaxis has only a minimal and short term effect on postnatal transmission of HIV by breastfeeding, and on the demonstration that mixed feeding confers a higher risk of HIV transmission than exclusive breastfeeding, the updated 2006 guidelines recommended that HIV infected women should breastfeed exclusively for six months and abruptly stop thereafter.8 The abrupt weaning was advocated to reduce the duration of mixed feeding and the baby’s exposure to HIV.

Although the recommendation did not precisely determine how “abrupt” the weaning should be, it was interpreted in the field as a plea for complete cessation of breastfeeding within a week of introducing weaning food at 6 months. However, the benefit of abrupt weaning for HIV exposed infants and their mothers was based only on “best guess” opinions of experts and on extrapolations of the risks and benefits of various durations of exposure. Abrupt weaning had been associated with a high rate of morbidity in cattle and with an unacceptably high mortality in calves and piglets,9 but this was ignored at the time.

Adoption of abrupt stopping of breastfeeding at 6 months was consistently followed by the observation of adverse events in both mothers (engorgement) and babies (babies crying, refusing to feed, experiencing diarrhoea, increased mortality), including in exposed uninfected babies.10 11 It was also suggested that a spill over effect of this strategy and its application in babies unexposed to HIV contributed to the serious outbreak of gastroenteritis in Botswana in 2006 with 35 000 cases and 532 infant deaths.12 Abrupt weaning at six months postpartum was perhaps not such a good idea after all.

Two potentially good options

The next wave of recommendations on preventing mother to child transmission of HIV infection and infant feeding had to take into account two bodies of evidence. Firstly, randomised trials showed that both prophylactic maternal antiretroviral therapy13 and infant peri-exposure prophylaxis for up to six months14 15 16 reduced the risk of breastfeeding transmission. In 2010 WHO recommended that countries choose between two options to prevent postnatal transmission when the mother is not eligible for treatment on her own account (women with a CD4 count >350×106/L):

Option A—After initial perinatal prophylactic regimen has ended, infants are given daily nevirapine until one week after all exposure to breast milk has ended.

Option B—Mothers receive antiretroviral drugs from 14 weeks of gestation until one week after all infant exposure to breast milk has ended.17

Secondly, another WHO expert committee on infant feeding, independent from that on mother to child transmission, argued that HIV infected women should continue to breastfeed for a maximum of 12 months.18

The result of the two sets of recommendations resulted in a choice between option A or B for a maximum of 12 months. Again, these public health strategies were not evidence based, as so far there is no evidence on the efficacy and safety of either option used for 12 months. Potential concerns include adherence fatigue after six months, particularly in a context of mixed feeding, and increased harm to infants from prolonged antiretroviral treatment. Of course, a trial would still be possible to determine whether 12 months’ prophylaxis is effective and well tolerated but, as the Botswana diarrhoeal epidemic showed in 2006, it may carry an unexpected societal cost.

Evidence is still not the priority

WHO’s latest revision of guidelines for preventing mother to child transmission is due to be launched at the International Aids Society conference in Kuala Lumpur on 30 June, despite the fact that relevant trials will not release their results until later this year.1 These include the National Institutes of Health Promise programme study of maternal and infant prophylaxis and the ANRS 12174 trial on infant prophylaxis, which assess the benefit-risk balance over 12 months.19 20

There is considerable pressure from some transmission experts and some institutions to prioritise a so called option B+, consisting of lifelong antiretroviral treatment for all HIV infected pregnant women, regardless of their clinical or immunological status.21 Some even propose to exclude the other two options (A and B) from the new guidelines. It is assumed that lifelong treatment would protect infants in future pregnancies and prevent transmission of HIV to the sexual partner.22 In addition, the experts argue for this strategy on the basis of the principles of equity (access to care), simplicity (in particular, no need for resources to measure CD4 counts, which may represent an obstacle to access to antiretroviral treatment in some countries), and cost effectiveness.23 24 Malawi has already committed itself to lifelong treatment.

Problems of lifelong treatment

However, other experts have challenged the principle of lifelong treatment on programmatic and scientific grounds. 25 Not a single comparative trial or observational cohort has been conducted on its ability to improve long term maternal health, to reach an acceptable adherence rate, to prevent HIV transmission to children (including in subsequent pregnancies), or to offer measurable wellbeing to the community. On the contrary, some preliminary observations based on field evaluation or on clinical trials of maternal treatment throughout breastfeeding strongly suggest that efficacy is suboptimal (29-52% efficacy in preventing postnatal transmission in the Kesho Bora trial, depending on the maternal CD4 count26), that post-partum adherence may be alarmingly low (52% at 12 months postpartum in a recent meta-analysis),27 and that most of the babies that become infected have multiclass resistant viruses, further compromising care.28 29 30

The simplicity of treating all HIV infected pregnant women for life (because it does away with the need for CD4 counts) is also challengeable. The strategy will be effective in preventing transmission only if it produces an undetectable viral load. Obviously, monitoring of treatment cannot rely on clinical examination alone since HIV related symptoms will take years to appear if the initial CD4 count was high. The CD4 count itself will be of little value for similar reasons. Eventually, HIV viral load, which is more difficult and expensive to measure than CD4 count, will have to be monitored. Missing mothers with viral escape is serious because they will falsely feel protected from transmitting HIV to their partners or babies and will not use additional protection measures.

Adherence to long term treatment will be a key concern, but we still don’t know how to sustain reasonable adherence throughout even the first pregnancy, let alone between pregnancies. And we still don’t know how to improve breastfeeding practices at a population level, although there is some evidence that peer counselling is helpful for the promotion of six month exclusive breastfeeding among non-HIV infected women.31 Finally, if lifelong treatment becomes a standard of care, subsequent evaluation of risks and benefits becomes complicated.

Lifelong treatment is also likely to be hard to fund in resource constrained countries (that is, most of those where HIV is endemic). Many countries already find it difficult to provide antiretroviral drugs to women with CD4 counts below the current threshold of 350×106/L, and shortages of drugs are common. The likelihood of reduced funding for HIV from major international bodies means that giving drugs to women with CD4 counts above 500×106/L, for whom there is no proved individual benefit and cheaper options are available to tackle transmission, may deprive those in real need of the drugs. Finally, the principle of equity for access to care is questionable when within a family the mother may get treatment but the father and other infected family members will not.

Evidence based policy

Since the early 1990s guidelines on preventing mother to child transmission of HIV have dangerously shifted from recommendations supported by strong scientific evidence, to recommendations based on experts’ best guesses and extrapolations followed by field evaluation, and then to recommendations proposed on the ground of theoretical modelling and ideological principles with limited possibilities for validation or refutation. Although nobody could argue against a quest for equitable access to care, simple interventions, conciliation of prevention and care, or efficiency of public health policies, the priority of medical intervention is to make sure that we first do no harm.

Emergency has not been a good adviser in the recent past, and this is unlikely to change. In order to avoid subjecting millions of citizens worldwide to the unanticipated adverse consequences of incorrect public health choices, revisions of public health recommendations should be carefully planned according to the international research agenda and the expected availability of trial results. In addition, we believe the guidelines should emphasise the need to support women’s’ adherence not only during pregnancy and breastfeeding but, as importantly, between pregnancies. Initiating lifelong antiretroviral treatment for all pregnant women irrespective of CD4 count is a promising strategy. However, many scientific and programmatic gaps have to be addressed before it can be recommended as health policy.


Cite this as: BMJ 2013;346:f3763


  • Contributors and sources: The publicly available WHO documents were the principal sources of this manuscript. All authors conceptualised this commentary. PV wrote the first draft of the manuscript and coordinated the revised versions. All authors (PV, TT, JFD, and NN) reviewed and approved the final version and are responsible for the final content of the manuscript.

  • Competing interests: We have read and understood the BMJ Group policy on declaration of interests and have no relevant interests to declare.

  • Provenance and peer review: Not commissioned; externally peer reviewed.