Precision public health: mapping child mortality in Africa

Spurred by Millennium Development Goal (MDG) 4, substantial progress in reducing child deaths was made between 1990 and 2015 around the world. The global under-5 mortality rate decreased from 90·6 deaths per 1000 livebirths in 1990 to 42·5 in 2015, a reduction of 53%.1You D Hug L Ejdemyr S et al.for the UN Inter-agency Group for Child Mortality Estimation (UN IGME)Global, regional, and national levels and trends in under-5 mortality between 1990 and 2015, with scenario-based projections to 2030: a systematic analysis by the UN Inter-agency Group for Child Mortality Estimation.Lancet. 2015; 386: 2275-2286Summary Full Text Full Text PDF PubMed Scopus (451) Google Scholar This equated to a worldwide reduction in annual number of under-5 deaths from 12·7 million to 5·9 million.1You D Hug L Ejdemyr S et al.for the UN Inter-agency Group for Child Mortality Estimation (UN IGME)Global, regional, and national levels and trends in under-5 mortality between 1990 and 2015, with scenario-based projections to 2030: a systematic analysis by the UN Inter-agency Group for Child Mortality Estimation.Lancet. 2015; 386: 2275-2286Summary Full Text Full Text PDF PubMed Scopus (451) Google Scholar Although this is an enormous accomplishment, and has been celebrated, these and other analyses have made clear that achievements varied considerably by cause of death,2Liu L Oza S Hogan D et al.Global, regional, and national causes of under-5 mortality in 2000–15: an updated systematic analysis with implications for the Sustainable Development Goals.Lancet. 2016; 388: 3027-3035Summary Full Text Full Text PDF PubMed Scopus (1757) Google Scholar region,3UN Inter-Agency Group for Child Mortality EstimationLevels and trends in child mortality: report 2013. UNICEF, New York, NY2013Google Scholar, 4UNThe Sustainable Development Goals Report 2017. United Nations, New York, NY2017Google Scholar and country.5Burke M Heft-Neal S Bendavid E Sources of variation in under-5 mortality across sub-Saharan Africa: a spatial analysis.Lancet Glob Health. 2016; 4: e936-e945Summary Full Text Full Text PDF PubMed Scopus (64) Google Scholar, 6GBD 2015 Child Mortality CollaboratorsGlobal, regional, national, and selected subnational levels of stillbirths, neonatal, infant, and under-5 mortality, 1980–2015: a systematic analysis for the Global Burden of Disease Study 2015.Lancet. 2016; 388: 1725-1774Summary Full Text Full Text PDF PubMed Scopus (471) Google Scholar As the global community directs its energies towards the Sustainable Development Goals (SDGs),7UNTransforming our world: the 2030 agenda for sustainable development. United Nations, New York, NY2015Google Scholar it is essential to examine these and even more granular variations to target interventions and resources for optimal impact. Several recent analyses will be foundational in global efforts towards the SDG target for child mortality of no more than 25 under-5 deaths and 12 neonatal deaths per 1000 livebirths in every country of the world by 2030.7UNTransforming our world: the 2030 agenda for sustainable development. United Nations, New York, NY2015Google Scholar For example, Burke and colleagues5Burke M Heft-Neal S Bendavid E Sources of variation in under-5 mortality across sub-Saharan Africa: a spatial analysis.Lancet Glob Health. 2016; 4: e936-e945Summary Full Text Full Text PDF PubMed Scopus (64) Google Scholar used Demographic and Health Survey data across sub-Saharan Africa to show the importance of subnational factors in variations in child mortality in the 1980s, 1990s, and 2000s. Liu and colleagues2Liu L Oza S Hogan D et al.Global, regional, and national causes of under-5 mortality in 2000–15: an updated systematic analysis with implications for the Sustainable Development Goals.Lancet. 2016; 388: 3027-3035Summary Full Text Full Text PDF PubMed Scopus (1757) Google Scholar modelled vital registration and verbal autopsy data to identify leading causes of under-5 death in countries with very high mortality rates. Is there more to be added to these child mortality explorations? Now, in The Lancet, Nick Golding, Roy Burstein, and colleagues8Golding N Burstein R Longbottom J et al.Mapping under-5 and neonatal mortality in Africa, 2000–15: a baseline analysis for the Sustainable Development Goals.Lancet. 2017; (published online Sept 25.)http://dx.doi.org/10.1016/S0140-6736(17)31758-0Summary Full Text Full Text PDF Scopus (157) Google Scholar provide estimates of under-5 and neonatal mortality rates that aim to enable “precision public health”9Dowell SF Blazes D Desmond-Hellmann S Four steps to precision public health.Nature. 2016; 540: 189-191Crossref Scopus (97) Google Scholar across Africa. Their analyses benefit from a large range of sources (235 surveys), more robust modelling, more contemporary estimates (to 2015), high spatial resolution of estimates (5 × 5 km), and more rigorous tests of the predictive ability of the model than previously used. Novel developments in this analysis include the use of summary birth histories in addition to complete birth histories.10Rajaratnam J Tran L Lopez A Murray C Measuring under-five mortality: validation of new low-cost methods.PLoS Med. 2010; 7: e1000253Crossref PubMed Scopus (84) Google Scholar The authors applied age-specific and period-specific adjustments, generated from surveys in which both complete and summary birth histories were available, to surveys in which only summary birth histories were available, thereby greatly increasing the number of usable datasets. Another novel approach is the use of ensemble methods, which were originally developed to improve predictive power in infectious disease mapping,11Bhatt S Cameron E Flaxman S Weiss D Smith D Gething P Improved prediction accuracy for disease risk mapping using Gaussian Process stacked generalisation.https://arxiv.org/abs/1612.03278Date: 2016Google Scholar to increase rigour in selection of covariates and identification of interaction. Four sub-models were fitted (generalised additive models, boosted regression trees, lasso regression, and ridge regression) for each age group and timepoint, and a Bayesian geostatistical model was then fitted using the cross-validated predictions. Given the distinction of under-5 from neonatal targets in SDG 3.2, the authors distinguished four age groups during modelling and report neonatal and under-5 results separately. Model validation was extensive and indicated generally good model fits. The overall declines in child mortality previously documented1You D Hug L Ejdemyr S et al.for the UN Inter-agency Group for Child Mortality Estimation (UN IGME)Global, regional, and national levels and trends in under-5 mortality between 1990 and 2015, with scenario-based projections to 2030: a systematic analysis by the UN Inter-agency Group for Child Mortality Estimation.Lancet. 2015; 386: 2275-2286Summary Full Text Full Text PDF PubMed Scopus (451) Google Scholar are substantiated and displayed by Golding, Burstein, and colleagues using intuitive figures that include representation of uncertainty intervals. Crucially, rates are presented at four levels of resolution: national, administrative 1, administrative 2, and 5 × 5 km. This level of detail allows striking subnational variations to be highlighted in under-5 and neonatal mortality rates, and in rates of decline from 2000 to 2015. For example, in 2015, even in countries that achieved MDG 4, two-fold variations in under-5 mortality rates can be seen across administrative 2 levels (eg, per 1000 livebirths, the lowest estimate by district in Liberia was 54·2 vs the highest estimate of 120; the lowest estimate by wilaya in Tanzania was 34·7 vs the highest estimate of 89·3). This level of precision is important for equitable policy making and efficient targeting of resources, and sets a high standard for future analyses of child mortality. Some limitations remain, although most are acknowledged by the authors. Scarcity of summary birth history data from some regions poses continuing challenges and increases uncertainty: stronger vital registration systems must remain the ultimate goal. Migration was not accounted for, and potentially important covariates such as access to water and sanitation and immunisations were not included in the analyses. Future work will need to address these gaps. For even greater precision in directing resources for child health, we urge scaling up the approach globally and disaggregating data to the third administrative level, the level at which resourcing decisions are made in many countries. We declare no competing interests. KD is supported by a Wellcome Trust Intermediate Fellowship in Public Health and Tropical Medicine (grant number 201900). Mapping under-5 and neonatal mortality in Africa, 2000–15: a baseline analysis for the Sustainable Development GoalsIn the absence of unprecedented political commitment, financial support, and medical advances, the viability of SDG 3.2 achievement in Africa is precarious at best. By producing under-5 and neonatal mortality rates at multiple levels of geospatial resolution over time, this study provides key information for decision makers to target interventions at populations in the greatest need. In an era when precision public health increasingly has the potential to transform the design, implementation, and impact of health programmes, our 5 × 5 km estimates of child mortality in Africa provide a baseline against which local, national, and global stakeholders can map the pathways for ending preventable child deaths by 2030.