February 2014, Durban, South Africa.
The workshop was hosted by the Africa Centre for Health and Population Studies.
This study will be conducted with the joint participation of specialist birth cohorts in which children’s HIV status has been determined more accurately and frequently than has been possible in community-based cohorts.
We will seek to:
- assess levels and trends in age-specific mortality among children in the study populations, broken down by HIV status, survival and co-residence of parents
- determine levels and trends in age-specific HIV prevalence in infants and children
- use models developed in specialist birth cohort studies to interpret prevalence data and obtain estimates of likely levels of HIV incidence at birth, in infancy and early childhood
- estimate (directly or indirectly) age-specific mortality patterns and trends in HIV infected children, especially at older ages (5-15) not covered by specialist birth cohort data
- investigate use of care and prevention services (ANC, VCT, PMTCT and ART) by pregnant women and those who recently gave birth in the study communities, and describe the impact of these services on child mortality and infection
- analyse patterns of subsequent contraceptive use by women who received PMTCT
- estimate (directly or indirectly) levels and trends in vertical transmission of HIV from infected mothers to their children, analysed by maternal factors (age, parity and duration of infection); and by child’s age at infection in sites that do infant PCR testing
- verify models developed to adjust DHS data to allow for under-estimation of child mortality from retrospectively reported birth history data
Prior to the ALPHA network start-up, three cohort studies took part in a UNICEF-funded study to estimate population level impacts of HIV on child mortality [159-162]. Most cohorts used antibody detecting ELISA tests rather than virus-detecting PCR analysis, so could not measure infection in children directly, but could generate data on children’s mortality patterns classified by mother’s infection status. Mathematical methods were used to infer age-specific mortality in infected children from estimated levels of vertical transmission of HIV and mortality in children of uninfected mothers .
These patterns were used by UNAIDS to estimate the extent of paediatric HIV, deaths of children from AIDS and treatment needs . This study will update the findings of the earlier work, and will examine access to PMTCT services achieved in ALPHA study sites Detailed data on patterns of infection and survival in children are available from specialist birth cohorts set up to evaluate PMTCT interventions and to study the role of breastfeeding in HIV transmission [165-167].
However, these clinic-based cohorts and trials do not allow us to estimate population-level effects on child mortality. We will develop new methods for combining information from these two types of cohort, and use data on uptake of PMTCT services [168,169] to make robust estimates of population-level impacts on child health, verified using data from those cohort studies that can ascertain infection in children . By making separate estimates for infant mortality and for mortality of children aged 1-4 this will allows sites that measure HIV status of children using antibody tests to contribute data in the older age range, whilst sites that conduct DNA PCR testing of infants of HIV positive mothers can also contribute data on infants. Collaboration between the ALPHA network and the KIDS ART-LINC consortium that links studies following children receiving PMTCT has already begun, with funding from UNAIDS enabling preparatory work with the University of Bordeaux which heads this consortium [164,171].
UNAIDS are particularly interested in bringing together the ALPHA cohort studies and the clinic-based cohorts from the KIDS ART-LINC consortium to establish empirically-based representations of survival curves for children from birth up to teenage, with and without treatment, similar to the adult survival functions described in the first phase of ALPHA. Since numbers observed in any one site are small, this can only be done by pooling and comparing data across different studies. Researchers working in the Manicaland site developed mathematical models for adjusting child mortality estimates obtained in DHS and other cross-sectional studies to allow for the biases due to the under-representation of the experience of HIV infected mothers in retrospectively reported birth histories .
The model has been tested against directly observed data in Manicaland, and this ALPHA workshop will give us the opportunity to validate it in other study settings. The Kisesa site in Tanzania has undertaken a WHO-funded study of family planning use following delivery by woman who received PMTCT . This revealed very low rates of use of contraception, even in HIV infected women who had received special counselling stressing the effectiveness of modern contraceptives in preventing further births of infected children.
This ALPHA network study will give us the opportunity to ascertain whether similarly low levels of family planning use are seen in women who have received PMTCT in other countries. Community-based measures of mortality of children of HIV infected women will inform policy makers about the effectiveness of PMTCT programmes outside of specially constructed trial settings. Since the reduction of child mortality is a Millennium Development Goal (#4) accurate measurements of trends, and an assessment of the contribution of PMTCT programmes to reversing stalled child mortality improvement due to the HIV epidemic will be an important tool for measuring progress in health improvement. Modellers have shown that the use of modern contraception by HIV infected mothers would be a very cost-effective way of avoiding the births of HIV infected children and reducing the orphanhood impact of HIV  so it would be useful for the ALPHA network to record the extent to which contraception is being used by HIV infected women who have already received PMTCT, and to identify the obstacles to contraceptive use.