Monitoring progress towards Millennium Development Goal 4 in generalised HIV epidemics: measurement and correction for bias in child mortality statistics

Simon Gregson, Imperial College
Timothy Hallett, Imperial College London
Felicia Kurwa, Imperial College London
Geoff P Garnett, Imperial College London
Sabada Dube, Imperial College
Godwin Chawira, Imperial College London
Peter Mason
Constance Nyamukapa, Imperial College London

Standard statistics may misrepresent trends in child mortality rates in countries with HIV epidemics due to the correlation between maternal and early childhood AIDS mortality. Empirical estimates of bias were derived from fertility history and child mortality data collected in 3 rounds of a prospective household survey in Zimbabwe, 1998-2005, including data on births to deceased women from verbal autopsy interviews. Determinants and temporal dynamics of bias were investigated using an individual-based stochastic model. The standard empirical estimates understated true levels of infant and under-5 mortality by 6.7% and 9.8%, respectively. In the model, bias varied by size and stage of HIV epidemic and level of background mortality. Bias typically caused overestimation of reductions in non-HIV-related child mortality but underestimation of the effect of PMTCT programmes. Analyses of trends in child mortality and assessments of progress towards MDG4 need to correct for correlation between maternal and early childhood AIDS mortality.