Letters

Prevention of vertical transmission of HIV in South Africa

BMJ 1999; 319 doi: http://dx.doi.org/10.1136/bmj.319.7222.1431 (Published 27 November 1999) Cite this as: BMJ 1999;319:1431

Findings probably do not apply to rest of sub-Saharan Africa

  1. Christian Gunneberg, Save the Children Fund UK, health adviser
  1. Salima, Malawi
  2. The Africa Centre, PO Box 198 Mtubatuba 3935, South Africa
  3. Division of Community Paediatrics, Department of Paediatrics, University of the Witwatersrand, PO Box 1038, Johannesburg 2000, South Africa
  4. Centre for Health Policy, University of the Witwatersrand

    EDITOR—Söderlund et al report the cost effectiveness of options available in South Africa to prevent the vertical transmission of HIV.1 I have concerns about the applicability of their findings (from an urban South African hospital setting) to the rest of sub-Saharan Africa.1 The paper's conclusion that the treatment is cost effective assumes that each child in whom seroconversion is prevented would otherwise have consumed substantial hospital resources.

    From the figures in the paper, I calculate that at least £0.33-1.73 ($0.50-2.60) per capita is spent just on HIV positive children aged under 1 in an average sub-Saharan country. This clearly does not apply to a country like Malawi, whose health expenditure is around £2.70 ($4) per capita. This cost-benefit conclusion is not applicable to most of sub-Saharan Africa, where children with HIV infection are generally treated at home and at health centres Altogether 70-80% of public health costs in most health systems are staff costs and the salaries of health workers; thus even in South Africa the postulated hospital health cost savings in the analysis could not be recovered to fund the additional drug costs.

    Using the authors' figures, I calculate that the antiretroviral drugs alone (£59 ($89) (PETRA regimen) to £267 ($400) (ACTGO76 regimen) per person treated) would cost an average sub-Saharan African country between £0.59 ($0.89) and £2.10 ($3.20) per capita to implement. This added cost does not include the cost of setting up and running testing and counselling facilities. This intervention represents more than the entire healthcare budget in many sub-Saharan countries. Its benefit would be on average 15 seroconversions prevented for every 500 births.

    The 66 year average lifespan used for the calculation of cost per life year saved will not be realistic in South Africa as the epidemic spreads, nor is it correct to assume that orphaned HIV negative children have an average lifespan. Evidence suggests that the loss of parents (in this case due to HIV infection) affects the survival of newborn children.

    My concern is that this analysis, which is possibly more relevant for middle income countries, may be used by donors to push sub-Saharan governments into piloting these costly and unsustainable interventions. This will lead to yet another high cost vertical programme putting further pressure on the fragile health infrastructure, which is not coping even with current demands. The danger is that the overall effect is to the detriment of the health systems and that the only real beneficiaries will be the drug companies receiving the donor funds, which would be better spent on the health system infrastructure.

    References

    1. 1.

    Paper did not include as a factor suboptimal effects that arise

    1. K Rotchford, senior scientist, Centre for Epidemiological Research South Africa (CERSA) (rotchfok{at}mrc.ac.za),
    2. S S Abdool Karim, director of CERSA,
    3. N Rollins, project leader of the perinatal project
    1. Salima, Malawi
    2. The Africa Centre, PO Box 198 Mtubatuba 3935, South Africa
    3. Division of Community Paediatrics, Department of Paediatrics, University of the Witwatersrand, PO Box 1038, Johannesburg 2000, South Africa
    4. Centre for Health Policy, University of the Witwatersrand

      EDITOR—Söderlund et al conclude that antiretroviral interventions are cost effective in preventing the vertical transmission of HIV in South Africa.1 Our experiences concerning the operational aspects of implementing antiretroviral therapy in a rural district of South Africa suggest that there are additional practical considerations in estimating the costs and benefits of such a programme.

      We followed up a cohort of 369 antenatal women in Hlabisa district. The outcome of pregnancy was known for 327 women. Their estimated date of delivery, based on a history of their last menstrual period, was a poor predictor of their actual delivery date, with 26% delivering before 38 weeks' gestation and 18% over two weeks after term (SD 27 days) (figure). Most (81%) did not register before 18 weeks, which precluded accurate ultrasound confirmation of dates. Altogether 93%, however, registered for antenatal care before 34 weeks' gestation, which allowed sufficient time for the screening and counselling necessary to receive an antiretroviral intervention. Finally, 82% of deliveries were in the health facility but only 2% of the women delivering at home intended to do so, which precluded early identification of this group.

      Figure1

      Distribution of gestational age at delivery based on history of last menstrual period

      Estimates of benefits of antiretroviral intervention need to include as a factor the suboptimal effects of a shortened regimen caused by preterm or home delivery and extended costs in those delivering after term. The implementation of an antiretroviral intervention to prevent vertical HIV transmission is a desirable policy but must be accompanied by the necessary resources and not just become one more burden on the rural primary healthcare service.

      References

      1. 1.

      Authors' reply

      1. Karen Zwi, senior lecturer (092zwi{at}chiron.wits.ac.za),
      2. Neil Söderlund, health economist (soderlun{at}icon.co.za)
      1. Salima, Malawi
      2. The Africa Centre, PO Box 198 Mtubatuba 3935, South Africa
      3. Division of Community Paediatrics, Department of Paediatrics, University of the Witwatersrand, PO Box 1038, Johannesburg 2000, South Africa
      4. Centre for Health Policy, University of the Witwatersrand

        EDITOR—Gunneberg is concerned about applicability of the findings of our study to other countries in sub-Saharan Africa. We do not recommend that the results are applied to settings different from the one studied. The sensitivity analysis in the original article deals with this issue to some extent by simulating circumstances different from the urban South African setting to which the model was applied. This cannot, however, replicate all possible scenarios (it is likely that many factors, not just healthcare spending, differ considerably between South Africa and Malawi). As we stated in the paper, we thus recommend that policymakers apply the model to their local circumstances, with a full set of local data to make a properly informed decision.

        The model uses a 66 year average lifespan (current life expectancy at birth) to calculate the life years saved. We are aware that the life expectancy of children orphaned by AIDS may not be the same as that of those growing up in ordinary family settings but do not have quantitative data on this issue. Even if life expectancy dropped to 40 years, the overall model results would remain largely unchanged as a result of the effect of discounting more “distant” life years saved (PETRA regimen changes from £9.33 ($14) to £10.66 ($16) per life year saved). Clearly, the issue of raising children who have been orphaned because of AIDS in environments similar to those of other children—for example, by encouraging fostering and adoption with state incentives—requires further study in its own right.

        Gunneberg assumes that the interventions assessed would be implemented as a vertical programme. Our model costs the intervention as an add-on to existing antenatal care, and not as a vertical programme.

        We agree with Rotchford et al that local practical constraints need to be considered before any intervention is implemented. However, with the newer short course regimens (such as the PETRA arm B and HIVNET 012 Nevirapine regimens), in which drugs are given during labour and after birth, late booking becomes a less important constraint Non-eligibility for the intervention because of failure to use antenatal services does not fundamentally affect cost effectiveness, because the non-attending group consumes no resources related to the screening or the intervention.

        View Abstract